Case Mix Tools FOR DECISION MAKING IN HEALTH CARE

Case Mix Tools FOR DECISION MAKING IN HEALTH CARE
Case
Mix
Tools
FOR DECISION MAKING IN HEALTH CARE
EDITORS:
Lina M. Johnson, MBA
Julie Richards, MHSc
George H. Pink, PhD
Lindsay Campbell, MHSc
Case
Mix
Tools
FOR DECISION MAKING IN HEALTH CARE
Lina M. Johnson, MBA
Research Associate
Hospital Management Research Unit
Department of Health Administration
Faculty of Medicine
University of Toronto
Julie Richards, MHSc
Consultant
Canadian Institute for Health Information
George H. Pink, PhD
Investigator
Hospital Management Research Unit
Associate Professor
Department of Health Administration
Faculty of Medicine
University of Toronto
Lindsay Campbell, MHSc
Manager, Coding Classification & Data Quality
Canadian Institute for Health Information
HMRU
All rights reserved.
No part of this publication may be reproduced or transmitted in
any form or by any means, electronic or mechanical, including
photocopying, recording, or by any information storage and
retrieval system now known or to be invented, without the prior
permission in writing from the owner of the copyright, except by
a reviewer who wishes to quote brief passages in connection
with a review written for inclusion in a magazine, newspaper or
broadcast.
Requests for permission should be addressed to:
Canadian Institute for Health Information
377 Dalhousie Street
Suite 200
Ottawa, Ontario
K1N 9N8
Telephone: (613) 241-7860
Fax: (613) 241-8120
http://www.cihi.ca
ISBN 1-896389-78-3
' 1998 Canadian Institute for Health Information
Registered Trade-mark of the Canadian Institute for Health Information
Acknowledgements
ABOUT THE CANADIAN INSTITUTE
FOR HEALTH INFORMATION (CIHI)
The Canadian Institute for Health Information (CIHI) plays a critical role in the
development of Canada's health information system. Incorporated in December
10th 1993, CIHI is a federally chartered but independent, not-for-profit organization. It brings programs, functions and activities from The Hospital Medical
Records Institute (HMRI), the MIS Group, Health Canada (Health Information Division)
and Statistics Canada (Health Statistics Division) together under one roof. Its primary functions are:
• collecting, processing and maintaining a comprehensive and growing number of health
databases and registries, covering health human resources, health services and health
expenditures;
• setting national standards for financial, statistical and clinical data as well as standards
for health informatics technology; and
• producing value-added analysis from its information holdings.
Through the pursuit of these primary functions, CIHI enables its many clients to make
sound health decisions based on quality health information. Stakeholders include ministries
of health, health care facilities, health-related organizations and associations, the research
community, private sector and the general public.
ABOUT THE HOSPITAL MANAGEMENT
RESEARCH UNIT (HMRU)
The Hospital Management Research Unit, in partnership with Sunnybrook Health
Science Centre since 1989, carries out a wide range of theoretical and practical
research activities as part of its mandate:
• to bring about improvements in the organization and management of hospitals in
Ontario;
• to enhance the hospital system’s ability to deliver effective health services to the
public; and
HMRU
• to help Ontario maintain its place among the best health services systems in
the world.
Areas of research include: health quality; performance measurement and benchmarking;
primary care reform; integrated health systems; health information systems; human
resource management; and development and monitoring of new organizational forms
including patient-focused care, hospital realignment and restructuring, strategic alliances,
integrated health systems, disease management, and virtual organizations.
The Unit is supported by a grant from the Health system-Linked Research Program of the
Ontario Ministry of Health.
ACKNOWLEDGEMENTS
ii
ABOUT THE TORONTO ACADEMIC
HEALTH SCIENCE COUNCIL (TAHSC)
Health Services are undergoing dramatic changes and more reforms within the system can be expected into the next millennium. These changes are affecting health
care delivery, research, and the education of health care professionals. In 1992 the
Toronto Academic Health Science Council was formed by the University of
Toronto and the 11 fully affiliated teaching institutions in order to work more collaboratively
on service, research and educational issues that were of concern to the Academic Health
Science Centre as a whole. The members of TAHSC are committed to providing high quality health care while embracing necessary changes to the health care system. In addition there
is a strong commitment to working together to advance scientific research and to improve
training and education for the professions in the health care system.
The goals of the Council are to:
1.
2.
3.
Provide a forum for discussion, exchange of information and development of shared
policy directions on issues of concern to its members;
Facilitate and support changes within and between TAHSC institutions, thereby
moving TAHSC towards a common vision for the future; and
Advocate for the special needs of TAHSC institutions (bearing in mind unique local,
provincial, national and international responsibilities), thereby enabling these
institutions to fulfill their clinical and academic missions.
Contents
Preface
1. CIHI Case Mix Tools
Lindsay Campbell
i
1
2. Day Surgery Incentive Model:
Funding Hospitals
Hy Eliasoph and Natalie Rashkovan
11
3. Program Information:
The Application of Information to Clinical Decision Making
Barbara Willis, Eck Hoffman, Jolyn Lawrenson, and Valerie Smith
29
4. Optimizing Clinical Utilization:
Structure and Strategies
Lynn M. Nagle, Judith Shamian, Margaret Catt, Martin Stein, and Joseph Mapa
41
5. Using Case Mix Tools with Case Costing Data
for Utilization Management
Cathy Davis
53
6. Utilization Management at
St. Boniface General Hospital, Winnipeg, Manitoba
Deborah Nowicki, Diane French, and Katherine Choptain
67
7. The Effect of Complexity and Age Adjustment
on Measures of Length of Stay Performance
Brenda Tipper and Darren Arndt
79
8. Impact of the Complexity Methodology
on an Ontario Teaching Hospital
Robert Fox, Jianli Li, and Robert Bear
87
9. Maintenance of Case Mix Tools:
The CIHI Revision Process for MCC 25 (Trauma)
Julie Richards, Greg Fallon, Karen Horne, and Margaret Catt
103
Conclusion
Lina M. Johnson
127
Glossary
131
Preface
As the year 2000 draws near, Canadian health care managers, clinicians and other health
professionals face increasingly difficult challenges of reducing costs while maintaining or
improving quality of care and access. In this environment, the importance of information
that decision makers use to manage complex hospitals and regional health authorities has
never been greater. Our purpose in producing this casebook is to provide a timely opportunity for health care managers, clinicians and other health professionals to share their experiences with the use of health information for decision-making.
On April 1 1997, CIHI implemented the Complexity Overlay called Complexity (PlxTM)
which improves the Case Mix Group method of classifying inpatients by adjusting for various complications and patient age. CIHI also produces information about ambulatory activity called Day Procedure Groups which continues to increase in importance as hospitals
move more care from an inpatient to an ambulatory setting. We hope that this casebook
provides an opportunity for health care managers, clinicians and other health professionals
to describe in depth how they have used CIHI information to analyze various clinical and
management problems and to identify the benefits and limitations of the information.
The first section contains an introductory chapter that describes Case Mix Groups,
Complexity and Day Procedure Groups that are produced and reported by CIHI. The first
chapter ends with a general discussion of how this information can be used for decision
making. The main body of the casebook consists of eight cases that describe how CIHI
information is being used by a variety of different Canadian health care organizations. These
cases were received in response to a request distributed by the Hospital Management
Research Unit at the University of Toronto to the Toronto Academic Health Science Council
(a council of the fully affiliated teaching hospitals and the University of Toronto) and by
CIHI to its clients in the Spring of 1997. The last section of the casebook contains some
reflections and lessons learned from the cases and the described use of CIHI information.
This casebook is a collaboration among CIHI, the Hospital Management Research Unit,
and the Toronto Academic Health Science Council. In coordinating a book of this nature,
there are many individuals to whom we owe a debt. First, we thank the authors who wrote
the cases. They spent much time and expended much effort in producing cases that are
clear, well written and directly relevant to those health care managers, clinicians and other
health professionals who must use information to make difficult decisions every day. As the
quality of the cases shows, the authors made a special effort to write their cases in a way
that would arouse interest and be useful in practice.
Others to be thanked include Dick Alvarez, President and CEO of CIHI, who supported
the use of CIHI resources in this initiative. Individuals from the CIHI Publications
Department helped bring the casebook to fruition, in particular, Lise Poirier, Manager
Publications, Scott Young, Multi-Media Specialist and Annie Desjardins, Publications
PREFACE
ii
Assistant for design, format and desktop publishing of the casebook. John Blackmore,
Communications Manager for CIHI, provided editorial assistance. Warren Skea, Senior
Analyst for CIHI provided statistical advice and assistance. Catharine Aird of the Hospital
Management Research Unit provided editorial assistance. David Shedden and Lin Grist,
the executives that support the Toronto Academic Health Science Council, provided
advice and assistance in facilitating the generation of cases by their member hospitals.
Peggy Leatt, Principal Investigator of the Hospital Management Research Unit, and Scott
Rowand, previously CEO of Wellesley Central Hospital and currently CEO of the
Hamilton Health Sciences Corporation, initiated the collaborative research relationship
between the Hospital Management Research Unit with the Toronto Academic Health
Science Council, with the support of Dr. Arnold Aberman, Dean of the Faculty of
Medicine at the University of Toronto.
We acknowledge the following members of the Toronto Academic Health Science Council
for their financial support of this casebook:
Ms. Pat Campbell, CEO, Womens College Hospital
Mr. Tom Closson, CEO, Sunnybrook Health Science Centre
Mr. Ted Freedman, CEO, Mt. Sinai Hospital and Chair of the Toronto Academic
Health Science Council
Mr. Stephen Herbert, CEO, Baycrest Centre for Geriatric Care
Dr. Alan Hudson, CEO, The Toronto Hospital (General, Western, and Ontario Cancer
Institute / Princess Margaret Hospital Divisions)
Dr. Sandra Jelenich, Acting CEO, Wellesley Central Hospital
Dr. Perry Kendall, CEO, Addiction Research Foundation
Mr. Jeffrey Lozon, CEO, St. Michael's Hospital
Ms. Jean Simpson, COO, Clarke Institute of Psychiatry (for Dr. Paul Garfinkle, CEO)
Mr. Michael Strofolino, CEO, Hospital for Sick Children
The Hospital Management Research Unit acknowledges the financial support provided by
the Ontario Ministry of Health and Sunnybrook Health Science Centre.
Lina Johnson, MBA
Julie Richards, MHSc
George Pink, PhD
Lindsay Campbell, MHSc
C H A P T E R
1
LINDSAY CAMPBELL
CIHI Case Mix Tools
INTRODUCTION
Grouping methodologies such as CMGTM and DPGTM are de facto standards for grouping
hospital patients with similar diagnoses and similar treatment requirements. Over the years,
through their application, these methodologies and their accompanying indicators have
established a track record of assisting health care facilities to effectively plan, monitor and
manage the services they provide.
To be broadly applicable, standards for classifying or grouping patients should meet the following criteria:
²
²
²
²
the data elements required for grouping should be limited to routinely collected data;
the number of possible groups or categories should be manageable;
cases within a group should be clinically similar; and
cases within a group should be statistically similar specifically in terms of length of
stay and/or total resource use.
With these criteria in mind, and using the extensive store of data in its Discharge Abstract
Database (DAD), CIHI and its forerunner, HMRI, have developed and maintained case
mix tools for use in Canada since 1983.
The DAD was originally developed in 1963 to collect data on hospital discharges in
Ontario. Its present format dates back to 1979. Currently, it captures about 3.6 million
records annually or 85% of all inpatient discharges in Canada. In addition, since 1993, at
least 1.5 million outpatient records a year, mostly related to day surgery, have also been collected in the DAD. Each record captures a standard clinical, demographic and administrative data set on a patient specific basis. Clinical information used in the grouping methodologies is further standardized by coding diagnosis and procedures according to established
classification guidelines. The two systems currently supported by CIHI are the International
Classification of Diseases, 9th revision (ICD-9), in combination with the Canadian
Classification of Procedures (CCP) and the American Clinical Modification of ICD-9,
ICD-9-CM, which incorporates procedures.
This chapter will provide an overview of Case Mix Groups, Day Procedure Groups,
Complexity Methodology, and Resource Intensity Weights.
CASE MIX GROUP METHODOLOGY
The CMG methodology was designed for use with hospital inpatients. Introduced in 1983,
the original CMG were an adaptation of American Diagnosis Related Groups (DRG) built
using ICD-9-CM to the Canadian environment of ICD-9 and CCP. By 1987, however, the
CMG adaptation was mapped back to ICD-9-CM and the methodology has been applied
to both coding systems since. Although CIHI introduced Complexity in 1997, the CMG
methodology remains at the foundation of this new system and the way in which a CMG is
assigned has not changed.
2
CIHI CASE MIX TOOLS
Besides the coding system used in construction, there are
two major differences between CMG and DRG. The driver for DRG assignment is principal diagnosis while the driver for CMG assignment is most responsible diagnosis
(MRDx). The principal diagnosis is the diagnosis which,
after investigation, is found to have been responsible for
the admission of the patient to hospital. The MRDx is
the diagnosis which is determined, at discharge, to have
been responsible for the greatest portion of the patient’s
length of stay. While principal and most responsible diagnoses are often the same, a significant post admission comorbidity (complication) may not be acknowledged in the
DRG assignment.
The DRG methodology uses pre-defined tables to determine whether an additional ICD-9-CM diagnosis is a
complication. The CMG methodology uses additional
diagnoses which are specifically identified as complicating
or cormobid conditions having an impact on length of
stay. The identification of these additional diagnoses is
accomplished using a standard coding guideline for
Canada called Diagnosis Typing. This Canadian coding convention has been used to improve the CMG methodology, most recently as the basis for the development of the
Complexity Overlay.
Since 1991 an expert team has existed at CIHI with a
mandate to continually improve the CMG methodology.
Clinical consultants work with CIHI methodologists, coding experts and systems analysts to ensure that the CMG
patient classification accurately reflects Canadian requirements and patterns of practice in hospitals.
Case Mix Groups are ordered within Major Clinical
Categories (MCC) which identify either a body system (e.g.
Respiratory System), or other specific types of clinical
problems (e.g. Mental Disorders, Neonates, Burns). There
are 25 Major Clinical Categories, 15 of which have been
revised by the CMG expert team between 1992 and 1997.
MCC assignment, which represents the first step in the
grouping methodology, is almost always determined by the
MRDx. Usually, the Most Responsible Diagnosis is a
unique assignment to one MCC known as the 'home' MCC.
There are some exceptions to this rule, such as diagnoses
with gender edits and the assignment of cases to MCC 15.
MCC 15, Newborns and Neonates, is based on age < 29
days or an entry code of newborn. A further division within this MCC is based on the weight of the baby.
Although the most responsible diagnosis is defined by CIHI
as 'the one diagnosis which describes the most significant
condition causing a patient’s stay in hospital,’ this may not
always be the condition for which the patient is admitted. If
the diagnosis recorded as most responsible is invalid, the
case is assigned to MCC 999, Ungroupable Data.
The Surgical Partition
Each MCC is divided into medical and surgical partitions.
The assignment of a case to a CMG within the surgical
partition is determined by the presence of a procedure.
The grouping methodology reads through all procedures
recorded to find one that is in the same MCC as the
MRDx. If it finds more than one procedure in this category the case is assigned to the CMG highest on the hierarchy. The surgical hierarchy is a decision rule that generally orders CMG within the surgical partition of each
MCC from most to least resource intensive. Embedded in
the surgical hierarchy is a grading of procedures used for
CMG assignment according to the following categories:
Extensive, Non-Extensive, May Not Require
Hospitalization and Other Procedures used for
CMG assignment.
In some cases, the MCC of the only procedure recorded
is not a match with the MCC of the MRDx. When the
procedure and the MRDx are not associated with the
same MCC, the case is assigned to a series of CMG
labelled Unrelated Operating Room (OR) Procedures.
The Medical Partition
If there are no procedures used for CMG assignment
recorded on the abstract, the case is assigned to the medical partition of the MCC. The medical partition consists
of groupings of similar diagnoses defined clinically
and/or by homogeneity of length of stay. Generally, the
methodology uses only the MRDx to assign a medical
CMG. One exception, however, is MCC 14 (Pregnancy
and Childbirth) where all delivered diagnoses are taken
into account.
CIHI CASE MIX TOOLS
3
Figure 1:
CMG Methodology
Flowchart
CMG GROUPER
METHODOLOGY FLOWCHART
Yes
Age < 29
Days
MCC 15
No
Assign MCC
Based on Most
Responsible
Diagnosis
Yes
O.R.
Procedure
Yes
Procedure Found
in MCC
No
Medical
Partition
Surgical
Partition
No
Unrelated O.R.
Procedure
COMPLEXITY METHODOLOGY
Over the past several years, the CMG development team at
CIHI has been able to effect improvements in the clinical
precision and statistical homogeneity of the CMG methodology. However, substantial variation in resource use could
still be demonstrated. Complexity was intentionally
designed to enhance the prediction of resource utilization
in acute care. It builds on the strengths of the CMG classification and Canadian morbidity coding practice through
the application of clinical judgment and CIHI abstracting
guidelines while still relying on data elements routinely captured in the CIHI Discharge Abstract Database.
The perception of a need for complexity classification
usually arises from concerns about the relevance or accuracy of case-mix estimates. Some hospitals, for instance,
may feel that particular CMG assignments may include a
much broader range of patients or treatments than fits
with the more specialized care they provide. Consequently
they may be uncertain about the accuracy or relevance of
CMG-based resource estimates in the context of their
programs. With a more case specific estimate, case-mix
and LOS comparisons among programs can be evaluated
with greater precision.
Complexity is applied to acute inpatient cases and makes
use of diagnoses over and above the MRDx employed in
CMG assignment. In general, Complexity will differentiate cases with:
APR97chronic disease conditions outside of
² one or more
the primary focus of the acute care episode;
² cases with multi-system failure; and
² cases with iatrogenic or other complications.
The patient’s Most Responsible Diagnosis continues to
be used to assign the case to one of 25 Major Clinical
Categories (MCC) and the case continues to be directed
to the medical or surgical partition based on the presence
or absence of an operative procedure. However, the step
of further directing some cases to a specific CMG
according to the presence or absence of a compliction/co-morbidity (CC) or the patient’s age range no
longer occurs. Complexity addresses these two significant
variables in a different and enhanced way. CMG with age
and CC splits have been collapsed back to a single CMG
to which the new Complexity Overlay can be applied.
The Complexity Overlay identifies diagnoses, over and
above the MRDx used for CMG assignment, for which
prolonged length of stay and more costly treatment might
reasonably be expected. The application of the overlay to
the base CMG divides the cases assigned to the CMG
into four new groups or cells for analytical purposes.
4
CIHI CASE MIX TOOLS
These new groups are called Plx groups and each represents a more homogeneous aggregation of patients on
which to predict length of stay and resource use. Table 1
depicts a possible range of co-morbidities related to a case
with a history of breast cancer presenting with brain
metastases as the MRDx.
Table 1: Possible Range of Co-morbidities in Craniotomy CMG
CMG
Plx Level with possible co-morbidity
Plx Group
001 - Craniotomy
Level 1
(history of cancer)
001 - Level 1
001 - Craniotomy
Level 2
(bone metastases)
001 - Level 2
001 - Craniotomy
Level 3
(hemiplegia)
001 - Level 3
001 - Craniotomy
Level 4
(obstructive hydrocephalus)
001 - Level 4
Where age is found to be predictive of length of stay or
resource use, the age category of the patient will be used
to further refine the Plx group estimate of length of stay
and resource use.
The intent of the age adjustment is not to discourage
documentation of the clinical characteristics of cases.
Rather, the adjustment functions as a rough proxy for
severity of illness within (and in addition to) the outlines
of a clinical population defined by CMG and Complexity.
It may soon be possible to match age effects seen in the
age adjustment methodology with those documented in
the clinical record. Because LOS effects are generally
divided between variables of age and complexity, age
tends to limit the impact of subjectivity in coding practice. In the development of Complexity grade lists, the
age adjustment was calculated before determining the
diagnoses that contributed to Complexity. This approach
eliminated certain age-related diagnoses that, in themselves, generally did not prolong LOS but were often
associated with long-stay cases.
The age and complexity components remain distinct and
hence can be used independently or together to predict
resource requirements. For example, age adjustments may
be helpful in tracking global resource needs in relation to
populations with differing proportions of elderly.
Complexity may assist in looking at seasonal variation in
resource requirements when weather or environmental
conditions precipitate a need for acute care for patients
with chronic respiratory conditions. Health-care managers
may be able to link particular demographic or environmental characteristics by examining the age or complexity components of acute-care treatment patterns. Whatever the
aim of the analysis, the new methodology makes it possible
to specify the effects of age group and Complexity.
Diagnosis Grade Lists
The treatment context is crucial to interpreting length of
stay (LOS) statistics for any given co-morbidity and,
therefore, to assessing the probability of extra resource
requirements. The abstracting guidelines for the DAD
result in the differentiation or typing of other significant
diagnoses (co-morbidities) into conditions that do influence LOS and conditions that do not. In addition, these
diagnoses are qualified as to whether they are present on
admission or arise after admission.
Initially, the 1993 Reference LOS database was used to
identify diagnoses on the MCC-specific lists where, when
the diagnosis was a co-morbid diagnosis, at least 20 cases
were associated with a length of stay greater than the
CMG average. In addition, the prolonged stay diagnoses
on this comprehensive list were specified as pre- or postadmission and belonging to either the surgical or medical
partition. The lists were then further developed through
an iterative process involving both statistical analysis and
clinical review.
For a given patient, other significant diagnoses on the
abstract are matched to the appropriate (medical or surgical) Complexity lists. A match is made with both the specific diagnosis and the diagnosis type. Given the possibility of recording as many as 15 primary diagnoses on the
CIHI abstract, a complex case may be matched to more
than one list or may exhibit multiple matches to one or
more lists. This information helps determine the case's
Complexity level.
As a result of the combined statistical and clinical analysis,
five Plx grades were established to which each diagnosis
could be assigned. An individual diagnosis could be assigned
to more than one grade based on its differential impact
when found in either the pre- or post-admission context.
Complexity Levels
CIHI’s "Surgical" and "Medical" partitions accommodate
four Complexity levels. These levels are monotonic in
nature, so that moving from Level 1 to Level 4 entails
progressively more resource-intensive treatment or care.
CIHI CASE MIX TOOLS
The following represents the Plx Levels:
1. No Complexity;
2. Complexity related to chronic condition(s);
3. Complexity related to serious/important condition(s);
and
4. Complexity related to potentially life-threatening condition(s).
Both grade and number of matches help determine Plx
level. Only Grade A or life threatening diagnoses show no
level difference between a single match and combinations
of A with itself or other grades. A grade A diagnosis will
assign a case to Level 4 regardless of additional diagnoses
with other grades. Where no grade A diagnosis is recorded a combination of diagnoses from other grade lists
could still elevate a case to level 4.
For C diagnoses, which represent chronic conditions,
Complexity distinguishes between those associated with
the "Home" or MRDx MCC–say, Pneumonia with
Diseases of the Respiratory System (MCC 4)—and those
outside the MCC. Statistical analyses have shown that C
list diagnoses from outside the home MCC are a more
powerful predictor of LOS than those within the same
MCC. That is consistent with the experience that multiple-system involvement often entails longer LOS than
single-system involvement.
Although in many cases, there may be a correlation to
severity, it is important to remember that Complexity is
intended to reflect the interaction of a patient’s multiple
diagnoses with the length of hospital stay. Severity primarily qualifies how seriously ill a patient is within a given
illness. For example, patients with chronic obstructive
pulmonary disease can expect to stay longer in hospital if
they also develop pneumonia. This "complexity" is evident from standardized currently available discharge
abstract data. On the other hand, the characterization of
the pneumonia as mild, moderate or severe requires the
collection of additional data according to one of a number of available severity scoring systems.
In certain clinical categories (e.g. pregnancy and childbirth,
newborns and neonates, mental diseases and disorders and
HIV infections), the Complexity Overlay fails to demonstrate an improvement in the estimate of length of stay or
resource use and, consequently, is not applied. In some
cases, some other data element such as birth weight is a
more powerful predictor. In other cases, such as HIV, complexity has been addressed already in the logic of CMG
assignment. The exclusions represent a fundamental design
principle of Complexity, namely that it is only applied where
it can reasonably be expected to improve homogeneity.
Abstract with CMG and
MCC Assigned
Plx Level
Complexity
Applicable to MCC
or CMG
Figure 2:
Complexity
Assignment 1997
No
9
Yes
Mechanical
Vent >96 Hours
Yes
4
No
Presence of Service
Transfer or CC
Diagnosis
*
No
1
Yes
Eliminate Duplicate
Diagnoses
*
Assign Grades to each
Dx. using appropriate
Grade Finder
**
Assign Plx level 1, 2, 3,
or 4 according to the
combination of Plx
Grades per abstract
*
* *
*
* *
5
Include Diagnosis Types 1, 2, W, X, Y Only
Consider Service Transfer Dx Type W, X, Y as Type 1 for Plx Grade List
Include Diagnosis Types 1, 2, W, X, Y Only
Consider Service Transfer Dx Type W, X, Y as Type 1 for Plx Grade List
6
CIHI CASE MIX TOOLS
Length of Stay (LOS) Indicators
Every year, effective April 1st, updated length of stay
(LOS) indicators are introduced which have been calculated using the most recent 12 months of data available
from the discharge database. For example, indicators published for use with the 1997 grouping methodology are
calculated from a reference database of cases collected
between October 1995 and September 1996 and
regrouped to the 1997 version . The LOS indicators represent an estimation or prediction of "typical" length of
stay for all cases within the CMG based on actual cases
occurring during the reference period. Before the introduction of Complexity, there was only a single value for
all "Typical" cases within each CMG. The predictive value
used was simply the average length of stay (ALOS) for
typical cases in that CMG. As a result, the ALOS was also
the ELOS or expected length of stay.
It is important to remember that about 35% of CMG
were split when LOS was found to vary systematically
with age group (age split) or with complicating or comorbid secondary diagnoses (CC splits) within a clinical
CMG category. A LOS indicator was assigned to each
adjacent CMG resulting from these splits. In this way, two
different ALOS indicators were available to describe different populations with the same MRDx.
In the pre-Complexity estimates, length of stay expectations corresponding to the different age range or CC
splits functioned roughly as those now calculated more
sensitively with the age adjustments and Complexity.
Because Complexity represents an advance over CCs, CC
splits no longer appear in areas where Complexity levels
are assigned. Similarly, where age adjustments for the
ELOS are used, age-splits have been removed.
In the Complexity methodology, where age is found to be
predictive of length of stay, the patient's age group (0–17,
18–69, 70+) will be used to further refine the Plx group
estimate of ELOS. Where a CMG is refined for both age
and complexity, as many as 12 ‘analytical’ groups or cells
will result. These analytical cells are referred to as APlx
cells. For example, CMG 001, Craniotomy, is refined for
both age and complexity. Therefore, for each of the 4 Plx
groups, there will be ELOS calculated at each of 3 age
categories, as follows:
Four levels x 3 age categories = 12 Aplx cells and
12 ELOS values.
The relationship between age, complexity and ELOS
varies considerably among CMG. A statistical test (F-test)
is applied to each CMG included in Complexity to determine whether or not there is a statistically significant difference between the Plx levels. If the difference is shown
to be statistically insignificant, the model used to predict
ELOS will not incorporate Complexity. Decision rules are
also used to determine when age categories are used to
further refine the estimate of ELOS. Consequently, the
number of APlx cells and the resulting number of ELOS
values that result will vary by CMG.
This variation in the effects of age and complexity on
ELOS makes it difficult to define a a single, efficient or
parsimonious model to capture the relationship appropriately for each CMG. The pilot version of Complexity,
which used age as a continuous variable, is an example of
a model where the number of variables produced differences whose significance was often difficult to assess. In
order to avoid using a model with excessive parameters in
the version introduced for 1997, regression analysis is
used to compute a set of LOS expectations for each of
the predictors using 5 different models. The number of
cases at the CMG and Plx level is combined with a determination of the statistical significance of each of the
parameters, age and complexity, and an examination of
whether there is interaction between them determines
which model is applied. Each model has a set of decision
rules governing its use. For example, for Plx groups that
have more than 200 cases in each Plx level and where the
parameters are found to be statistically significant at
0.001, the formula used models the age interaction at
each of the Plx levels within the CMG. On the other
hand, where the complexity parameter is not found to be
statistically significant at 0.005 in a CMG with more than
200 cases, the formula used models age at the CMG level.
The application of these models results in an ELOS value
that is case-specific, within age groups, and expectations
will vary for cases within a CMG. A trim point is calculated for each APlx cell, and is used to identify cases as outliers if the patient's actual LOS is greater than this trim
point. For atypical cases such as outliers, deaths, acutecare transfers and signouts, Complexity levels are
assigned, but ELOS is not.
Quarterly reports are available to each hospital which
summarize the experience of each case with respect to
length of stay. The actual length of stay is compared to
similar cases in the national database and comparisons are
provided at various levels of aggregation such as by doctor or patient service. Some examples of how length of
stay reports are used are:
² review of bed utilization and length of stay patterns;
² monitoring the allocation of beds to service/program
areas;
² research and planning for future service requirements;
and
² assigning expected date of discharge.
CIHI CASE MIX TOOLS
DAY PROCEDURE GROUP (DPG)
METHODOLOGY
DPG was developed to address the information needs
related to the growing use and importance of ambulatory
care as a service delivery setting. Introduced in 1993, ten
years after the CMG methodology, the DPG methodology
is modelled on an American system, the Products of
Ambulatory Surgery. The adaptation for use in Canada also
required evaluation and mapping between the two different
classifications of procedures, CCP and ICD-9-CM.
The DPG methodology assigns cases to one of 69 mutually
exclusive groups according to the principal or most significant procedure recorded as part of a standard outpatient
record or abstract. Cases assigned to the same DPG category represent similar clinical episodes and are intended to be
homogeneous with respect to resource consumption.
The DPG methodology has often been used only to
monitor day surgery procedures. In recent years, advances
in medical technology, changes in anaesthesia use and
techniques and the drive for lower cost service delivery
have enabled the shift of certain procedures from the
inpatient to the outpatient setting. In response to requests
from hospitals and to assist the process of monitoring
the transfer of cases from inpatient to outpatient surgery,
a clinical review was conducted which resulted in certain
CMG being designated as "May Not Require
Hospitalization" (MNRH). This group of CMG represented a target population for a possible shift to outpatient service. The growing shift to ambulatory surgery
could be evaluated by monitoring CMG and DPG in parallel. It should be noted, however, that the CMG and
DPG grouping methodologies behave differently and
remain two separate methodologies. In addition, the designation MNRH was never intended to suggest that all
cases in a CMG with this label did not require hospitalization. In fact, there are many valid clinical reasons for
these cases to be treated as inpatients. This is demonstrated in the new Complexity methodology where cases
grouped to a CMG labelled MNRH can subsequently be
assigned to the higher levels of complexity.
To adequately fulfil the information needs of the ambulatory care service setting, a more comprehensive methodology than DPG is required. A new grouping methodology which encompasses not only day surgery but also
emergency and ambulatory clinic services has been developed and is now being introduced. The integrity of the
day surgery DPG has been maintained, however, and
imported without change as a module of the new
methodology. The increasing demand for a comprehensive ambulatory grouping methodology reflects a change
in the emphasis on information. Information about
ambulatory services is no longer required primarily to
evaluate the rate of transfer to this lower cost environ-
7
ment. Rather, ambulatory care has been acknowledged as
a setting of choice for the delivery of a wide range of
services and it has become increasingly important to have
comprehensive information to manage this setting appropriately. The challenge in ambulatory care remains the
cost of data capture. Monitoring the savings of a shift to
ambulatory surgery offered ample justification for the
cost of submitting these records for processing by a hospital’s Health Records Department. To many, the number
of patient encounters in, for example, Emergency and
Outpatient Clinics, precludes the collection of a standarized data set requiring abstraction by Health Records.
The automation of data collection in these areas is seen
as the key to successful implementation of a comprehensive ambulatory care reporting system.
RESOURCE INTENSITY
WEIGHTS
Resource indicators are also updated with new values calculated based, in part, on the same data set used for
length of stay. The new values are also applied to data
effective April 1st of each year. The Resource Intensity
Weight system is a resource allocation methodology for
estimating the costs for both acute inpatients and day
surgery cases. The RIWTM is used to standardize the
expression of hospital case volumes, recognizing that different patients represent different burdens on health care
resources. After weighting, volume can then be expressed
in terms of "weighted cases." RIW estimates the resource
intensity of particular cases relative to an average inpatient cost. Values for the indicator are assigned to each
analytical cell of the grouping methodologies (CMG,
DPG, Plx) and are defined by a model of how case costs
and ELOS vary by CMG, Plx level and age. The estimation procedures are specific for three different sets of
cases:
² those with a course of treatment "typical" of the
cases in the cell;
² those with "atypical" courses of treatment; and
² day procedures.
Typical cases are those with a length of stay at or less
than the trim point established for the cell. The trim
point is calculated to exclude long stay outliers and is
defined by the third quartile of the database length of
stay for the cell plus twice the inter-quartile range. The
typical cases are used as the basis for RIW estimation.
Atypical cases are defined as, deaths, transfers to or from
other acute care institutions, voluntary signouts and long
stay outliers. The patient LOS used to differentiate inliers
from outliers is based on the patient’s total LOS including
any days identified as alternate level of care (ALC). This
acknowledges that non-acute days of stay generate costs,
and therefore, should be included in the estimation of
RIW values.
8
CIHI CASE MIX TOOLS
Until it is possible to guarantee a sufficient volume of
comprehensive Canadian cost data on a regular basis, an
American database from the state of Maryland is used in
the calibration of RIW values. This U.S. database,
referred to as the ‘Maryland calibration database,’ is a
population database and contains information on all inpatient and same day surgery acute care provided in the
state. These data are released annually and, in part, to
ensure adequate case volumes for each cell in the grouping methodology, the most recent two years are used in
the calculation of RIW. The rationale for using U.S. data
to calibrate RIW values is based upon the assumption
that the relationship of hospital charges to length of stay,
patient age, and other classification variables which exist
in the U.S. data can also be used to estimate relative costs
for Canadian cases.
Maryland’s history of state-wide rate regulation establishes a basic similarity with the experience of Canadian hospitals. As a result of regulation, Maryland remains one of
the few states where the burden of uncompensated care
is shared equally by all payer groups. This means that
patients pay basically the same price for a hospital service
regardless of the type of payment or insurance. This
‘insurance’ system helps to secure similar standards of
care for all patients in Maryland, regardless of insurance
status. The result is a database that is generally applicable
in Canada for RIW estimation.
It is desirable to have weights for inpatient and ambulatory cases on the same relative value scale in order to facilitate comparability of weights and weighted cases across
treatment settings. The Maryland calibration database
provides inpatient and ambulatory costs from the same
population and year which are reported using similar coding standards. Therefore, CIHI is also able to use
Maryland charge data as the RIW calibration database for
DPG weights.
The Calculation of RIW for Complexity
The first step in calculating RIW values is to assign
Maryland inpatient cases in the calibration databases to
CMG and Plx levels and SDS cases to DPG using the
CIHI grouping methodologies. To adapt the CIHI grouping methodology to Maryland inpatient coding practice,
the principal diagnosis is treated as equivalent to the
MRDx on the CIHI abstract. Unlike the CIHI database,
there is no diagnosis typing in the Maryland calibration
database. As a result, it is often difficult to identify
whether another significant condition occurred pre- (type
1) or post- (type 2) hospital admission. To enable the
grouping of Maryland data into Complexity CMG, lists of
complexity diagnoses were created for the U.S. database.
These lists are based on the differential impact of typing on
Complexity by diagnosis, and the distribution of type 1
and type 2 cases by diagnosis.
The next step consists of an adjustment to remove organ
retrieval costs from Maryland charges for heart, liver and
kidney transplants. This reflects differences in the way
these programs are funded in the U.S. and Canada.
Statistical regression analysis is applied to the Maryland
database to estimate the average organ costs for the
CMG. This estimate is then subtracted from the charges
attached to the corresponding CMG.
Total charges and routine/ancillary per diem (RAPD)
charges are adjusted for age and complexity where sample
size permits and there exists a statistically significant
effect on ELOS. Where these are not significant or a
small sample size exists, pooling of adjacent levels, or age
categories occurs. In addition, estimates of RIW are constrained to monotonicity in Complexity so that total RIW
is the same or increases with the level of complexity.
Adjustments are made for hospital specific factors which
affect patient charges. Hospital specific factors include
teaching status, size, factor input prices, and the demographics of referral populations. These adjustments are
made using a Hospital Specific Relative Value (HSRV)
method. The HSRV method removes these hospital specific effects through iterative computations of population
relative values from hospital specific estimates. In the
past, these factors had to be directly specified in the
regression equations to estimate and remove the bias. The
advantage of the HSRV method is that the direct specification of these types of variables is not required, allowing
us to control for any unspecified differences that may not
have been specified or identified as variables—differences
that are essentially unknown.
Finally, the weights are adjusted for differences between
Maryland average LOS and CIHI ELOS. This length of
stay adjustment relies on the separation of routine and
ancillary (RA) charges that vary with length of stay from
fixed charges that do not. To the extent that CIHI ELOS
is greater than (or less than) Maryland ALOS, an unadjusted weight would underestimate (or overestimate) the
Canadian resource intensity.
The CIHI RIW for a typical case, therefore, is estimated
as the ratio of the ELOS adjusted mean charge for each
analytical cell to the average case weighted expected
charge or standardization factor. The standardization
ensures that the average typical RIW equals one for the
CIHI typical database.
There are no LOS adjustments with the DPG calibration
databases. With this database the first step is to compute
the average charge by procedure, which is then weighted
by the day procedure volumes from the most current
CIHI fiscal year database. The weighted average of the
mean charges by procedure are calculated for each DPG.
CIHI CASE MIX TOOLS
9
Each type of atypical case is calculated differently. For
Deaths, surgical and medical cost curves are computed
which define how much more expensive per day a death
case is expected to be than a typical case. The average cost
per day will be higher, the shorter the length of stay prior
to death. In addition, surgical cases will have higher relative
costs than medical cases because fixed costs represent a
higher proportion of the per diem. The death cost curves
are used to assign a weight for all deaths up to the length
of stay trim point. Deaths occurring after the trim point
are assigned a weight using the outlier formula.
average inpatient cost per weighted case can be applied to
the total weighted case volume for any service or program to determine the expected approximate cost to the
hospital of that program.
Similar curves are calculated and used to assign weights to
Transfer cases. There are four curves for transfers, surgical cases transferred in and surgical cases transferred out,
medical cases transferred in and medical cases transferred
out. Transfer cases staying beyond the trim point are
assigned a weight using the outlier formula.
$2,000 x 575 = $1,150,000
Voluntary signouts represent less than 1% of the total
cases. A signout case receives credit for each day of stay
up to their ELOS but any patient days beyond the ELOS
are not considered.
Outlier cases are those which stay beyond the LOS trim
point. The per diem weight is adjusted using a formula
which assumes that some days beyond the trim point are
"low severity" or less acute than days within the trim
point. Each outlier case receives an RIW equal to the
weight of a typical case plus an adjusted per diem weight
for each day from the ELOS until discharge.
Application of RIW
The primary application of RIW information is to support the translation of case mix data into costs. This is
done by allocating the resulting weighted cases to clinical
categories or programs, services, physicians or other
groups and then assuming that dollars are likely consumed on the same basis.
Unit Cost Determination
To determine the approximate average cost of hospital
programs and services using RIW weighted cases, you
must first determine the overall average inpatient cost per
weighted case for the hospital. The bottom line of any of
the CIHI RIW Basic Summary of Activity reports shows
the total weighted cases for a hospital. By calculating the
total net inpatient cost for the same period and dividing
this by the total weighted cases, you can calculate an average inpatient cost per weighted case. Net inpatient costs
only must be identified because they were included in the
cost data used to create the RIW values. To find the Net
Total Inpatient Cost for a period, costs associated with
ambulatory care activity, chronic patient activity and nonpatient activity must be removed. Once calculated, the
For example, if a hospital has 4,000 RIW weighted cases
during a period and a net total inpatient cost of $8 million, the average inpatient cost per weighted case is $2,000.
If this same hospital had 575 weighted cases in the
Orthopaedic Surgery service, the expected approximate
cost to the hospital of the Orthopaedic Surgery service is:
If the same hospital has 224 weighted cases in Major
Clinical Category number 1, Diseases of the Nervous
System, the approximate cost to the hospital of treating
inpatients with diseases of the nervous system is:
$2,000 x 224 = $448,000
Use of the RIW as a measure of resource use cannot
replace full implementation of patient costing or Global
Dimension Reporting as described in the Guidelines for
Management Information Systems in Canadian Health
Care Facilities (MIS Guidelines), but it can provide a simple approach to estimating patient and program costs.
Cost information can then be communicated to physicians and other health care providers to help ensure that
they are aware of how the resources of the hospital are
being used. This cost information can be used in conjunction with length of stay comparisons to target services which have the greatest apparent opportunity for
improvement. Weighted case information can help you
assess the financial "materiality" of the services provided
by your hospital.
Targeting CMG Assignments for
Utilization Review
The CIHI Basic Summary of Activity—Top 40 Case Mix
Groups report will show which 40 CMG assignments
account for approximately one half of a hospital’s inpatient costs. Examples of questions to be asked when
examining this report are:
² Are the majority of the weighted cases produced by
typical or by atypical cases? Is your caseload typical or
atypical and, therefore, can you use typical "benchmarks" such as ELOS to plan and monitor?
² If typical cases contribute most of the weighted cases
for the CMG, how does the actual LOS compare to
the expected LOS? If the actual LOS is greater than
the expected LOS, the actual cost of these patients to
your hospital could be even greater than the report
suggests.
10
CIHI CASE MIX TOOLS
² If atypical cases contribute most of the weighted
cases for the CMG, you should look at the atypical
Case Summary—Top 40 Case Mix Groups report.
Which category of atypical cases is producing the
weighted cases? If the majority of your atypical cases
are from transfers, this requires a different strategy
than if the majority of your atypical cases are long
stay outliers.
² Is the allocation of weighted cases what you would
expect? Are your resources really being allocated to the
types of cases or programs that you intend to emphasize at your hospital? A hospital may designate certain
programs as strategically important and resource allocation may not actually reflect that decision.
Strategic Planning
Comparing the costs of programs or services may produce some unexpected results. Is the allocation of your
inpatient budget (as measured by the RIW values) consistent with the planned areas of emphasis for your hospital? Are you spending appropriately on those services that
you have established as your priorities? Is what you
planned to do actually what is occurring?
The RIW reports can help to assess the potential financial
impact of strategic decisions. As long as decisions can be
modelled using inpatient volumes or case mix, the RIW
information will provide a measure of the impact on
resource use. For example, the RIW reports show the
case volume, actual length of stay and weighted cases by
MCC and physician/patient service. If a hospital is considering the addition of a new program or physician, the
RIW report can assist in estimating the impact of the
change using the hospital’s own cost per RIW weighted
case figure.
SUMMARY
This chapter has briefly explained the case mix tools currently available and their use and application for individual hospitals. Each methodology has additional documentation which can be obtained from CIHI and which
describes the methodology in more depth and provides
the information updated annually. In recognition of the
growing need for health information for evidence based
decision making, CIHI continues to work to develop and
expand its case mix tools beyond the hospital setting and
across the continuum of care.
Additional Material Available from the
Canadian Institute for Health Information:
²
²
²
CMG Directory for Use with Complexity 1997
DAD Length of Stay Indicators for Use with
Complexity 1997
DAD Resource Indicators for Use with
Complexity 1997
²
DAD Procedure Codes and Resource Indicators
for Use with DPG 1997
²
Report: Evaluation of the Complexity
Methodology
About the Author
Lindsay Campbell, MHSc is Manager, Coding, Classifications
and Data Quality, Canadian Institute for Health Information,
Toronto, Ontario.
C H A P T E R
2
HY ELIASOPH, NATALIE RASHKOVAN
Day Surgery Incentive
Model: Funding Hospitals
CHAPTER OVERVIEW
A Day Surgery Incentive Model was developed and incorporated into the acute hospital
funding formula in Ontario. The model has been used by the Ontario Ministry of Health
(MoH) since April, 1996. It provides explicit incentives for hospitals to perform day
surgery and transfer caseload from inpatient to outpatient where clinically appropriate. The
methodology is based on the percentage of age-adjusted cases on a procedure-specific
basis that a hospital completes on an outpatient care basis relative to the provincial average
for each procedure. Using current clinical practice in Ontario hospitals as a benchmark for
funding adjustments is viewed as an objective way to set realistic expected rates for outpatient surgery. The model is based on a clear definition of day surgery that incorporates an
exclusion list.
The methodology is based on Day Procedure Groups (DPG), developed by the CIHI.
DPG is a case mix, grouping methodology for day procedures that, when combined with
resource intensity weights (RIW), is used as a basis for determining the relative cost and
commensurate level of funding. The Day Surgery Incentive Model is an overlay on the
DPG RIW for the express purpose of providing an incentive for day surgery, where clinically appropriate. The model is derived through a five-step process that identifies qualifying
day surgery procedures, determines expected day surgery rates, and calculates hospital-specific weighted case adjustments and outpatient surgery indices.
Hospitals with relatively high proportions of day procedures and low average cost per
weighted case experienced increased funding, whereas those with relatively low proportions
of day surgery and high average cost per weighted case experienced reduced funding levels.
It was anticipated that these funding changes would lead to greater funding equity and
address a persistent shortcoming of the existing Ontario Equity Funding Formula.
INTRODUCTION
Technological advances, progressive changes in medical practice, better comparative information, shifting public expectations and fiscal pressures have all combined to give impetus
to an unprecedented shift of hospital caseload from inpatient to ambulatory or outpatient
care. Nowhere has this shift been more pronounced than in day surgery. Between 1991/92
and 1995/96, the number of day surgery procedures in Ontario grew from 602,600 to
760,800, an increase of 26%, compared to a decrease of 8% in the number of hospital
separations and 19% in the number of patient days during the same period. During this
time, day surgery, as a percentage of all surgery, increased from 42% to 58% (Ontario
Ministry of Health, 1997).
12
DAY SURGERY INCENTIVE MODEL: FUNDING HOSPITALS
Despite this tremendous growth in day surgery, there
were no commensurate mechanisms developed and
implemented to reimburse hospitals for performing these
procedures. Rather, the global budget, which was used to
fund the majority of hospital activity, was increasingly
viewed as unable to provide equitable funding to meet
changing clinical practice and hospital resource requirements. This recognition led Ontario to consider new
approaches and methods for funding hospital activity that
were more equitable and responsive to changes in hospital activity - both inpatient and outpatient. A chronological summary of these events and developments can be
found in The History of Ontario’s Hospital Funding System
(JPPC, 1993b).
This chapter describes the development of a methodology
used to implement an incentive model for funding hospital-based day surgery in Ontario. Although the model was
designed to be used with a case-costing approach for funding day surgery, it can be used in conjunction with other
funding formulae. The milestones surrounding the development of the model are chronicled, the definition of day
surgery is described, the methodology is detailed, the implications and applications are outlined, and the benefits and
limitations are identified.
CHRONOLOGY OF EVENTS
In the fall of 1992, the Board of the Ontario Hospital
Association (OHA) passed a resolution in which they
urged the Ministry of Health (MoH) to "take immediate
action to rectify the disincentives in the Equity Formula
respecting ambulatory care and day surgery to support hospitals to use such delivery methods to provide care and allocate resources on the most appropriate and cost effective
basis" (OHA, 1992). The MoH, which had introduced the
mandatory collection and reporting of day surgery
abstracting through the CIHI a year earlier, agreed to the
need to resolve funding disincentives for day surgery. The
task of developing a funding approach and methodology
that was equitable and that encouraged the clinically driven delivery of day surgery was undertaken by the
Ontario Joint Policy and Planning Committee (JPPC).
The JPPC is a partnership between the MoH and the
OHA to recommend and facilitate the implementation of
hospital reform within the context of the health care
reform agenda in Ontario (JPPC, 1996a). The JPPC
strives to meet the following objectives through its work:
² to define the role of hospitals in a reformed health
care system;
² to recommend hospital funding that promotes effectiveness, efficiency and equity among hospitals;
² to provide input with respect to policies and methods
to implement health care reform;
² to assist in the development and implementation of
provincial standards for hospital information; and
² to assist hospitals to effectively use information for
internal management purposes.
In August, 1993, the JPPC’s Activity Measurement
Working Group released two reports on Day
Procedures—Report on Day Procedures (JPPC, 1993c) and
Equity Formula Adjustments for Day Procedures (JPPC, 1993a).
The former report contained recommendations that
advocated the adoption of the CIHI DPG grouping
methodology (see related chapter) and the use of weights,
derived from the Maryland charge database, for the funding of day surgery/procedures. DPG are a procedure-driven grouping methodology which classifies approximately
1,650 surgical and medical procedures.
The second report, Equity Formula Adjustments for Day
Procedures, outlined a methodology to incorporate funding
adjustments into the Equity Formula for the proportion
of outpatient surgery a hospital performs, relative to the
Ontario experience. The development of this methodology was particularly timely because it appeared likely that
the Equity Formula would be used for reallocating funding among hospitals in 1994/95. This methodology, however, was dependent on the acceptance and use of the
DPG grouping methodology and the use of weights to
develop the DPG equivalent of inpatient weights (JPPC,
1994b, p. 4).
Before accepting the recommendations of the Activity
Measurement Working Group respecting DPG, the
Hospital Funding Committee of the JPPC requested that
an independent review be conducted. The purposes of
this review were: (1) to evaluate the proposed modified
DPG system as a grouping methodology and weighting
scheme for Day Procedures; (2) to evaluate the appropriateness of using the Maryland charge database to develop
DPG weights; and (3) to recommend the scope of procedures (definition and/or list) to be incorporated into
Ontario funding formulae.
The independent study, Review of Day Procedure Groups
and Weights (JPPC, 1995d), concluded that the DPG
grouping methodology was "as robust as any other system currently in use" and that DPG could serve well as a
grouping system around which to structure a weighting
system". The report cautioned, however, that more investigation was needed before the Maryland charge data
could be used to develop a weighting system for funding
day procedures.
DAY SURGERY INCENTIVE MODEL: FUNDING HOSPITALS
13
Following the independent review, an Ad Hoc Day
Procedure Group was convened to coordinate the necessary follow-up analyses that had been identified in the
report. This included an assessment of the stability of
charge-based weights year over year, the potential impact of
implementing the weights on overall hospital funding, and a
comparison with the Ontario Case Cost Project day procedure costs. The report also recommended that a decision be
made concerning a (revised) definition of day procedures/
outpatient surgery that could be applied on a consistent
basis by/across all hospitals, particularly as it relates to a
specified period of time (JPPC, 1994bs).
Although this new definition does not provide an incentive for moving 2– or 3–day cases to 24 hour observations, it does provide an incentive for hospitals to discharge patients by midnight. It also makes cost profiles
for day surgery patients more homogeneous. It was
thought that this definition would ensure more consistent
reporting, particularly in relation to the time parameters
for a day procedure, and the specific identification of
procedures that should be excluded from the definition.
Much of the concern expressed over the adoption of
DPG pertained to the use of the Maryland charge database to determine relative weights. CIHI outlined its
rationale for moving from RIW, based on 1985 New York
data, to RIW, based on 1991 and 1992 Maryland charge
data, in a report entitled, Moving to the Maryland Database
(CIHI, 1994).
² Visits within the definitions of a General and
Special Outpatient Clinic or a Medical Day/Night
Care Program.
² Attendances to private physicians’ offices located
within the hospital (such as rented space or GFT
offices).
² Attendances solely for the purpose of undergoing a
diagnostic procedure for which a professional and
technical component is billable (including interventional radiography procedures).
² Endoscopic procedures such as anoscopy, colposcopy, otoscopy, proctosigmoidoscopy, rhinoscopy,
and vaginoscopy are normally performed as a simple
outpatient procedure and may not be counted as a
Same Day Surgery Program except in the unusual circumstance where use of hospital Same Day surgery
facilities is required because of clinical indications.
² Minor procedures such as removal of cysts, warts, toe
nails, and simple nasal polyps may not be counted as
Same Day Surgery Program cases except in unusual circumstances where use of anaesthesia and post-anaesthesia recovery, or an operating suite and a post-recovery room is required because of clinical indications.
Day Procedure funding was first implemented in 1995/96
(see Appendix III). At that time, the Maryland charge database was used to calculate the relative weights for the DPG
grouping methodology for outpatients as well as for inpatients, via Resource Intense Weights (RIW). While
Ontario’s own Ontario Case Cost Project (OCCP) data
were considered as an alternative to Maryland, the outpatient case costs were just beginning to be collected and,
therefore, weights derived from these costs would not be
available for several years. Even then, the sample size
would not be sufficient to generate a set of weights for all
day procedures. The Maryland database provided an
acceptable alternative while "made in Ontario" weights
continued to undergo development, testing and evaluation.
REVISING THE DEFINITION FOR DAY
SURGERY/PROCEDURES
Both the independent review and the Report on Day
Procedures stressed the need to review the definition of
day surgery/procedures (as used in the Ontario funding
formulae) to clarify its interpretation and use. Clarification
would ensure more consistent reporting, particularly in
relation to the time parameters for a day procedure. In
the fall of 1994, the Day Surgery Definition was revised,
retroactive to the Spring of 1994. The definition read as
follows:
All procedures performed on patients whose
hospital stay/visit (from time of registration
to discharge), occurs on the same calendar day
or (if over the midnight hour) is less than 12
hours, irrespective of the site/location within
the hospital, anaesthetic route/administration,
or whether the procedure was scheduled/
unscheduled (JPPC, 1994b, p. 5).
The following exclusions were applied to the definition of
Day Surgery:
With CIHI’s assistance, the funding formulae were applied
retroactively using the revised definition because admission
hour and discharge hour were already mandatory fields in
the day surgery database. With minor exceptions hospitals
were not required to re-code or re-abstract.
METHODOLOGY
In developing the Day Surgery Incentive Model, several
options were considered, including:
1. Identify inpatient procedures from the ICD-9
(International Classification of Diseases—9th
Revision) procedure list which "should" be completed
as day surgery. Then identify these inpatient procedures at individual hospitals and give them credit for
the day surgery resource intensity weight (RIW).
14
DAY SURGERY INCENTIVE MODEL: FUNDING HOSPITALS
2. Based on the ICD-9 procedure list, map day surgery
procedures to the corresponding inpatient procedure at
the provincial level. Then identify these mapped day
surgery procedures at individual hospitals and give
them the inpatient RIW for the day surgery procedure.
3. Similar to option (1) and (2) except calculate an acceptable percentage of cases that will be expected to be
performed on an inpatient basis taking patient age, comorbidities, and other factors into consideration. Then
calculate a "blended" rate which would be a weighted
average of the inpatient and outpatient weights.
These options were discussed but dismissed for several
reasons. Options (1) and (3) required the identification of
surgical procedures that are suitable for surgical day care.
Though option (2) does not require this identification,
this option would have resulted in hospitals potentially
being over-compensated for day surgery procedures
because there is no consideration of the expected percentage of day surgery cases.
In reviewing the strengths and weaknesses of these different options, it was decided that clinical practice in
Ontario at the time should be used to determine the
expected percentage of day surgery cases. This reduced
the need to make arbitrary decisions concerning eligible
outpatient procedures and allowed for automatic inclusion of future advances in day surgery. The expected percentage of outpatient cases is also a moving target which
will remain responsive over time to changes in clinical
practice in Ontario hospitals (JPPC, 1995b, pp. 1–2).
The Day Surgery Incentive Model was, therefore, selected
based on clinical utilization. With the application of the
Day Surgery Incentive Model,
hospitals with a higher proportion of same
day surgery cases than the provincial average by procedure [were] rewarded. They
received the inpatient weight for the percentage of day surgery cases greater than
the provincial average by procedure.
Hospitals with a lower proportion of day
surgery cases than the provincial average by
procedure were penalized and received the
outpatient weight for the percentage of
inpatient cases above the provincial average
(JPPC, 1994b, p. 7).
1
DAY SURGERY INCENTIVE MODEL—
METHODOLOGY: A FIVE-STEP
PROCESS
Step 1: Application of Exclusions to the
Outpatient Database to Identify Appropriate
Day Surgery Cases
Exclusions (listed in Appendix I) were applied to CIHI’s
1993/94 statistical database for Ontario hospitals to identify surgical procedures performed on a day surgery basis.
The original model was developed using 1993/94 data.
More recent data were used in subsequent refinements
and applications (JPPC, 1996b, 1996c). Initially, no consideration was given to those outpatient surgery cases
where the procedure was performed (e.g. pacemaker
implants) and the patient returned to the referring hospital on the same day. This was addressed in the first refinement to the incentive model.
To assist hospitals in reconciling their CIHI reports with
the Day Surgery Incentive Model results, a special report
was prepared. Outpatient cases that were coded by individual hospitals as a sub-service other than sub-service 11 (i.e.
qualifying day surgery cases) are listed in a separate companion report entitled Comparison of Hospitals’ Weighted Case
Adjustment and Overall Performance Relative to the Ontario
Experience (JPPC, 1995a). Those qualifying day surgery
cases that were excluded due to the aforementioned exclusion criteria are also listed in this report. Outpatient cases
excluded in step 1 were also excluded from both the Day
Surgery Incentive Model and Equity Formula calculations.
Step 2: Application of Exclusions to the
Inpatient Database to Identify Candidates for
Day Surgery
To operationalize the incentive model for day surgery it
was necessary to estimate the volume of inpatient cases
(by procedure) that might be eligible for shifting to day
surgery. Exclusions listed in Appendix II were applied to
CIHI’s acute inpatient statistical database for hospitals in
Ontario to identify surgical procedures performed on an
inpatient basis which are also performed on a surgical day
care basis (surgical procedures were automatically excluded from the Day Surgery Incentive Model if no Ontario
hospital completed them on a day surgery basis).
In addition to the listed exclusions, there were a few hospitals where the inpatient RIW value was lower than the
comparable outpatient RIW value for a small number of
procedures. This occurred because of differences in the
Sub-service 1 or patient sub-service 1 is a category of the patient service reporting process, created to identify Same
Day Surgery cases. The sub-service 1 captures all procedures which meet requirements for Surgical Day Care, also known
as Qualifying Day Surgery abstracts. Patient sub-service 2, 3 and 4 are also used to abstract other visits not considered
to be qualifying Surgical Day Care procedures. These categories are used by hospitals primarily for internal utilization management purposes (HMRI, 1991).
DAY SURGERY INCENTIVE MODEL: FUNDING HOSPITALS
methodology used to assign cases to DPG and CMG.
Day surgery cases are assigned to DPG by procedure.
CMG, however, are assigned by most responsible diagnosis, then by principal procedure, patient age, and the
absence or presence of complications and co-morbidities.
If these cases were considered in the Day Surgery
Incentive Model, a hospital would be penalized for completing them on a day surgery basis. To remove this disincentive, in those cases where the inpatient RIW value was
lower than the outpatient RIW value, the inpatient RIW
value was changed to the outpatient RIW value plus an
additional 25%. A 25% adjustment was applied because it
was believed this would provide an appropriate incentive
to complete these cases on a day surgery basis. This
adjustment was implemented to compensate for a small
number of procedures in which the weighting systems
used in conjunction with CMG and DPG are neither
15
aligned nor reflective of true costs. This incentive measure was used to encourage hospitals to continue to perform procedures on an outpatient basis where clinically
appropriate by removing any financial barriers.
Step 3: Calculation of the Expected
Inpatient Rate for ICD-9 Procedures
After applying the aforementioned exclusions to the inpatient and outpatient data sets, each ICD-9 procedure was
subdivided into six categories based on patient age and
the presence of complications or co-morbid conditions
(CC): 1) 0 to 15 no CC; 2) 0 to 15 with CC; 3) 16 to 69
no CC; 4) 16 to 69 with CC; 5) 70 plus no CC; and 6) 70
plus with CC. For each of the six categories within each
ICD-9 procedure, the percentage of cases performed on
an inpatient basis in Ontario, as reported to CIHI, was
calculated as shown in Table 1.
Table 1: Calculation of the Expected Inpatient Rate
ICD-9 Procedure
DPG
Description
Average & CC
Categories
ONT.
Out-patient
Cases
ONT.
In-patient
Cases
ONT
In & Outpatient Cases
ONT.
% Inpatient
01.03 Direct Laryngoscopy
17
0 to 15 no cc
0 to 15 w/ cc
16 to 69 no cc
16 to 69 w/ cc
70 plus no cc
70 plus w/ cc
17
1
128
26
18
3
10
2
13
8
5
0
27
3
141
34
23
3
37.0
66.7
9.2
23.5
21.7
0
01.04 Other Non-operative
17
0 to 15 no cc
0 to 15 w/ cc
16 to 69 no cc
16 to 69 w/ cc
70 plus no cc
70 plus w/ cc
16
4
230
33
36
9
3
7
22
3
5
1
19
11
252
36
41
10
15.8
63.6
8.7
8.3
12.2
11.1
01.05 Pharyngoscopy
17
... etc.
These calculated inpatient percentages by ICD-9 procedure code split into 6 categories, were then used as proxies for the expected percentage of procedures that should
have been performed on an inpatient basis.
Step 4: Calculation of the Hospital-Specific
Weighted Case Adjustment
The weighted case adjustment by procedure is calculated
by multiplying the difference between a hospital’s actual
and expected number of inpatient cases, by the difference
between the inpatient and outpatient RIW. A hospital’s
weighted case adjustments are then summed for all surgical procedures, with the resultant total being the net
weighted case adjustment to be applied to a hospital’s
total weighted cases using the equity formula.
... etc.
For each Ontario hospital, the expected number of inpatient cases by ICD-9 procedure was calculated, as outlined
in a hospital-specific report entitled, Calculation of the
Hospital Specific Weighted Case Adjustment, distributed
to all Ontario hospitals. The expected number of inpatient cases were compared with a hospital's actual number
of inpatient cases and used to calculate a hospital's
weighted case adjustment as shown in Table 2.
16
DAY SURGERY INCENTIVE MODEL: FUNDING HOSPITALS
Table 2: Calculation of the Hospital-Specific Weighted Case Adjustment
Col 1
Col 2
Col 3
ICD-9
Procedure
Hosp In &
Outpatient
Cases
Ont %
Inpatient
Cases
Col 4
Col 5
Hosp Expected
Inpatient Cases
Col 2 & 3
Col 6
Hosp Actual Expect minus
Inpatient
Actual
Cases
Col 4 5
Col 7
Col 8
Col 9
Col 10
Average
Inpatient
RIW
Aver age.
Outpatient
RIW
RIW Diff.
Col 7 8
Weighted
Case Adjust
Col 9 & 6
Carpel Tunnel
90
11%
10
10
0
0.9
0.7
0.2
0
D&C
50
0%
0
5
-5
0.8
0.6
0.2
-1.0
100
14%
14
10
4
1.1
0.6
0.5
24
25
Hernia Repair
Total
This table calculates an "expected" number of inpatient
cases for this hospital based on the Ontario wide experience (col 2 & col 3). The actual number of inpatient procedures performed is subtracted from the expected values
(col 4—col 5). The resultant number is multiplied by the
difference between the inpatient RIW and the corresponding outpatient RIW (col 9 & col 6). This calculation
is made for all procedures and then summed. The resultant total is the weighted case adjustment to be made to a
hospital's total weighted cases used in the Equity Funding
Formula calculations.
2.0
1.0
Step 5: Calculation of the Hospital-Specific
Outpatient Surgery Index and Weighted
Case Index
In order for individual hospitals to compare their performance relative to the Ontario experience, a hospital specific index was calculated. This was completed to help
hospitals better understand their weighted case adjustment (see Table 3).
Table 3: Calculation of the Hospital-Specific Outpatient Surgery and Weighted Case Index
1
2
3
Col 1
Col 2
Col 3
Col 4
Hospital
Name
Hospital
Expected
Inpatient
Cases
Hospital
Actual
Inpatient
Cases
Actual/
Expected
Case Index
(Col 3/Col 2)
100
300
200
200
300
100
2.000
1.000
0.500
The hospital specific Actual/Expected Case Index is calculated by dividing the actual number of inpatient cases by
the expected number of inpatient cases. Hospitals with a
lower percentage of inpatient cases than expected will
have an index less than one, while hospitals with a higher
percentage of inpatient cases than expected will have an
index greater than one. Hospitals with a weighted case
index less than one will have positive weighted case
adjustments, and hospitals with a weighted case index
greater than one will have negative weighted case adjustments. Since the expected inpatient cases are based on the
provincial average, the total expected inpatient cases for
the province divided by the total actual inpatient cases for
the province approximates 1.000.
IMPLICATIONS & IMPACT OF THE
WEIGHTED CASE ADJUSTMENT
A hospital receives a positive weighted case adjustment if
its expected number of inpatient cases is higher than its
actual number of inpatient cases (i.e. the actual number
of outpatient cases is higher than expected). This adjustment means that a hospital receives the inpatient weight
for the percentage of outpatient cases completed in
Col 5
Actual/Expected Weighted
Case Index
(Hospital Actual Inpatient
weighted cases divided by
Hospital Expected I/P
weighted cases)
Col 6
Weighted
Case
Adjustment
(see Table 2 in
step 4)
excess of the provincial average. In the Equity Funding
Formula, this adjustment lowers a hospital's average cost
per weighted case because a hospital receives the inpatient
weight but incurs the outpatient cost.
Conversely, a hospital receives a negative weighted case
adjustment if its expected number is lower than its actual
number of inpatient cases (i.e. the actual number of outpatient cases is lower than expected). This adjustment
means that a hospital receives the outpatient weight for
the percentage of inpatient cases completed in excess of
the provincial average. In the Equity Funding Formula
this adjustment increases a hospital's average cost per
weighted case because a hospital receives the outpatient
weight but incurs the inpatient cost.
The percentage change in hospitals' total weighted cases
as a result of this weighted case adjustment ranges
between plus or minus 2%. This translates into a similar
change in a hospital's average cost per weighted case.
Appendix IV provides a detailed example of how the Day
Surgery Incentive Model works, and the impact of this
weighted case adjustment on a hospital's average cost per
weighted case.
DAY SURGERY INCENTIVE MODEL: FUNDING HOSPITALS
PRACTICAL APPLICATION FOR
UTILIZATION MANAGEMENT
In developing the Day Surgery Incentive Model, it was
argued that if such incentives were to be incorporated
into the equity funding formula, tools should be developed to assist hospitals in developing better utilization
management practices. Toward this end, the JPPC
Utilization Management Committee was asked to develop
tools that could assist hospitals in the management of day
surgery utilization. CIHI outpatient surgery data and the
outpatient grouping methodology were used by the JPPC
to produce comparative reports to assist hospitals to
improve their performance and make shifts to outpatient
surgery. As part of the JPPC’s "How Do You Compare?"
series, Moving to Outpatient Surgery (JPPC, 1994a, 1995c)
provides hospitals with comparative information to determine where to focus their attention for change, and to
identify opportunities and strategies to improve utilization
management practices and processes. Moving to Outpatient
Surgery (JPPC, 1994a, 1995c) illustrates several strategies
that have been successfully used in hospitals to achieve
high outpatient surgery rates. These strategies include:
² fostering progressive attitudes about utilization management;
² emphasizing patient education and guiding patient
expectations;
² making the care process convenient for patients, surgeons and anaesthetists;
² identification and involvement of physician champions;
² post-discharge services;
² ensuring teamwork and accountability; and
² investing in technology, new anaesthetic agents and
pain/nausea management.
BENEFITS AND LIMITATIONS
The Day Surgery Incentive Model was developed in
response to the demand by hospitals to not only fund day
surgery, but to also encourage the move away from inpatient surgery. While the Day Surgery Incentive Model had
some technical and data-based limitations, some of which
were resolved in subsequent refinements to the original
model, it clearly resulted in several benefits, including:
² It addresses the day surgery part of the OHA resolution passed in the fall of 1992 (Proceedings from
OHA Annual Convention, November, 1992) by providing a financial incentive for hospitals to perform
more cases on a day surgery basis. It is also consistent
with the Ministry of Health directive for hospitals to
increase the percentage of procedures performed on
an outpatient basis.
17
² The weighted case adjustment is simple to calculate
and understand. Hospitals with a high proportion of
day surgery cases are rewarded, as weighted cases will
be added to the Equity formula "total weighted cases"
calculation. Hospitals with a low proportion of day
surgery cases will be penalized, as weighted cases will
be removed from the Equity Formula "total weighted
cases" calculation. The case and weighted case indices
also provide an easy to understand summary measure
of a hospital’s overall performance.
² The methodology accounts for differences in age mix
and complications and co-morbid conditions, making
it more acceptable to clinicians.
² No additional data collection or definitions are
required.
² The methodology is based on current clinical practices. It does not make arbitrary decisions concerning
eligible outpatient procedures. The expected percentage of outpatient cases is also a moving target which
will remain responsive over time to changes in clinical
practice in Ontario hospitals.
² The methodology allows for automatic inclusion of
new advances in day surgery as they occur from year
to year.
² The Model also has the flexibility to be applied to
future funding methodologies under development.
The limitations of the model are as follows:
² The percentage of procedures that a hospital is
expected to perform on a day surgery basis is a
"moving target" based on historical data. Hospitals
will not know the current year’s targets until
September/October of the following year, when
CIHI year end data are available. Some view this as an
advantage, however, because hospitals would not
know if they have achieved or surpassed the targets
and, therefore, would continue to strive for improvements.
² Hospitals with a low percentage of day surgery cases
are penalized equally, whether their reasons are appropriate or not. For example, a hospital may have a low
percentage of day surgery cases due to factors
beyond its control (e.g. geography, demographics of
the referral population, availability of community services, etc.). There are also no provisions for peer
group differences when calculating the expected percentage of inpatient cases by procedure. Inpatient
and outpatient data for all Ontario hospitals were
combined to compute these percentages.
18
DAY SURGERY INCENTIVE MODEL: FUNDING HOSPITALS
SUMMARY
For the first time, a significant and growing segment of
hospital activity in Ontario is being funded on a case-mix
basis. Hospitals which have a low proportion of day
surgery cases relative to the Ontario average receive a
negative weighted case adjustment, resulting in a higher
average cost per weighted case in the Equity Formula.
Hospitals with a higher proportion of day surgery cases
relative to the Ontario average receive a positive weighted
case adjustment, resulting in a lower average cost per
weighted case.
The incorporation of funding and incentives for day
surgery/procedures into the hospital funding formula in
Ontario was an important milestone that would not have
been possible without the development of DPG and
RIW. The success of this undertaking in Ontario has
fueled the current search for case-mix and case-weight
based funding for other areas of hospital activity, including chronic/continuing complex care, rehabilitation, mental health and ambulatory care.
The impetus for undertaking these developmental efforts
derives from the hospitals who have argued for equitable
approaches to hospital funding. Implementation of these
efforts has been supported by the Ontario Ministry of
Health and the Ontario Hospital Association, and has
been made possible by the pioneering work and cooperation of the CIHI.
DAY SURGERY INCENTIVE MODEL: FUNDING HOSPITALS
REFERENCES
CIHI (1994). Moving to the Maryland Database: What are the
Implications for Ontario? Presentation by C. Fitzgerald, S.
Halpine, S. Maloney, at a special meeting organized by the
JPPC. Toronto, ON: Canadian Institute for Health
Information, April.
HMRI (1991). HMRI Surgical Day Care Abstracting
Requirements for Fiscal 1991/92 Memorandum. Mailed in conjunction with Ontario Growth Committee Mailing,
January. Toronto, ON: Hospital Medical Record Institute.
JPPC (1993a). Equity Formula Adjustments for Day Procedures.
Working Paper of the Activity Measurement Working
Group. Toronto, ON: Ontario Joint Policy and Planning
Committee, August.
JPPC (1993b). The History of Ontario’s Hospital Funding
System (Reference Document DP1-3). Toronto, ON:
Ontario Joint Policy and Planning Committee, October.
19
JPPC (1996b). Revised 1994/95 Day Surgery Incentive Model
(Reference Document #RD3-8). Toronto, ON: Ontario
Joint Policy and Planning Committee, March.
JPPC (1996c). 1996/97 Day Surgery Procedure Exclusion List
(Reference Document # RD3-7). Toronto, ON: Ontario
Joint Policy and Planning Committee, March.
JPPC (1997). Ontario Hospital Cost Distribution Methodology
By Patient Activity (Reference Document # RD4-6A).
Toronto, ON: Ontario Joint Policy and Planning
Committee, March 20.
OHA (1992). Unpublished Minutes of the Board
Meeting, November 1. Toronto, ON: Ontario Hospital
Association.
Ontario Ministry of Health (1997). Unpublished Statistics
derived from the Planning and Decision Support Tool.
JPPC (1993c). Report on Day Procedures to the Methodology
Sub-Committee (Reference Document # DP1-6). Toronto,
ON: Ontario Joint Policy and Planning Committee,
August.
JPPC (1994a). Moving to Outpatient Surgery: How Do You
Compare? A Resource Manual for Ontario Hospitals (Reference
Document RD1-3). Toronto, ON: Ontario Joint Policy
and Planning Committee, January.
JPPC (1994b). Report & Recommendations for Day
Surgery/Procedures Funding (Reference Document #DP2-2).
Toronto, ON: Ontario Joint Policy and Planning
Committee, November.
JPPC (1995a). Comparison of Hospitals' Weighted Case
Adjustment and Overall Performance Relative to the Ontario
Experience (Reference Document #RD3-1). Toronto, ON:
Ontario Joint Policy and Planning Committee, April.
JPPC (1995b). Day Surgery Incentive Model (Reference
Document #RD3-1). Toronto, ON: Ontario Joint Policy
and Planning Committee, April.
JPPC (1995c). Moving to Outpatient Surgery: Data Report
Update (Reference Document RD2-8). Toronto, ON:
Ontario Joint Policy and Planning Committee.
JPPC (1995d). Review of Day Procedure Groups and Weights
(Reference Document #RD2-4). Toronto, ON: Ontario
Joint Policy and Planning Committee, March.
JPPC (1996a). Orientation Package. Toronto, ON: Ontario
Joint Policy and Planning Committee, August.
About the Authors
Hy Eliasoph, MA, CHE is Director Hospital Relations and Health Policy,
Ontario Hospital Association and Former Executive Director, Ontario
Joint Policy and Planning Committee; and
Natalie Rashkovan, MHSC, CHE is Technical Planning Consultant,
Ontario Joint Policy and Planning Committee.
Acknowledgement
This chapter was published with the kind permission of the Ontario
Ministry of Health and the Ontario Hospital Association.
APPENDIX I
21
The following exclusions were applied to the 1993/94 statistical outpatient database for
Ontario Hospitals:
1. Cases not assigned to patient sub-service 01 (i.e. non-qualifying same-day surgery cases).
Note: Cases that are re-assigned to sub-service 09 are excluded.
2. Cases where the principle procedure suffix was:
- 0—procedure performed out of hospital
- 8—cancelled surgery, or
- 9—previous surgery prior to admission
3. Cases where the principle procedure code was blank or invalid.
4. Cases not assigned to a DPG category.
5. Stillborns
6. Cases assigned to patient service 51—obstetrics delivered, patient service 54—newborn, Dilation and Curetage (D & Cs) following delivery or abortion (ICD-9 81.01);
are not excluded.
7. Cases assigned to the following DPG:
DPG 02 Spinal Procedures
DPG 03 Nerve Injections
DPG 20 Angiography
DPG 59 Skin Procedures (no complications or comorbid conditions recorded)
DPG 62 Haemodialysis
DPG 63 Transfusions
DPG 64 Cardioversion
DPG 65 Chemotherapy
DPG 66 Myelogram
DPG 99 Ungroupable
DPG 02, 03, and 62 to 66 are medical day care cases and do not qualify as day surgery
according to the old day surgery definition. DPG 20, angiography, is excluded because
it has a technical component which is billable. Minor skin procedures under DPG 59
(i.e."lumps and bumps") are also excluded because they do not qualify as day surgery
according to the old day surgery exclusion list.
8. Minor endoscopic procedures (ICD-9 (CCP1) procedure codes):
01.01 Rhinoscopy
01.23 Sigmoidoscopy
01.24 Proctosigmoidoscopy
01.25 Anoscopy
01.32 Otoscopy
01.36 Vaginoscopy
82.81 Culdoscopy/Colposcopy
9. Cardiac Catherizations (i.e. ICD-9 codes: 49.95 right, 9.96 left and 49.97 combined
right and left catherizations).
10. Outpatient surgery Cases where the procedure was performed and the patient returned
to the referring hospital on the same day.
(from JPPC reference document #RD4-6A, Ontario Hospital Cost Distribution
Methodology By Patient Activity, March 20, 1997, pp. 28–29.)
1
CCP is the Canadian Classification of diagnostic, therapeutic, and surgical Procedures.
APPENDIX II
23
The following exclusions were applied to the 1993/94 statistical inpatient database for
Ontario Hospitals:
Atypical cases (i.e. deaths, transfers, signouts, and outliers)
1. Cases where inpatient length of stay was greater than 3 days (i.e. 4 days or greater)
2. Cases where the principal procedure suffix is 0 (i.e. procedures performed out of hospital), 8 (i.e. cancelled surgery), or 9 (i.e. previous surgery).
3. Cases where the principal procedure code was blank or invalid (i.e. zzzz).
4. Cases where the DPG is not assigned (i.e. DPG = **)
5. Stillborns
6. Cases assigned to patient service 51—obstetrics delivered, including ICD-9 procedure
code 81.01, D & C following delivery or abortion.
7. Cases assigned to patient service 54—newborn.
8. Cases where the principal procedure maps to DPG 02 (Spinal Procedures), DPG 03
(Nerve Injections), DPG 20 (Angiography), DPG 59 no CC (minor Skin Procedures),
DPG 62 (Haemodialysis), DPG 63 (Transfusions), DPG 64 (Cardioversion), DPG 65
(Chemotherapy), DPG 66 (Myelogram), and DPG 99 (Ungroupable).
9. Cases with the following ICD-9 (CCP) procedure codes as principal procedure:
01.01 Rhinoscopy
01.23 Sigmoidoscopy
01.24 Proctosigmoidoscopy
01.25 Anoscopy
01.36 Vaginoscopy
82.81 Culdoscopy/Colposcopy
APPENDIX III
25
At the September 16, 1994 meeting of the Joint Policy and Planning Committee (JPPC) the
following recommendations were approved in principal:
1. Outpatient Grouping Methodology: That the Canadian Institute for Health
Information’s (CIHI’s) Day Procedure Group (DPG) grouping methodology be used to
group day procedures.
2. Outpatient Weights: That the Maryland charge database be used to calculate a set of
relative weights for the DPG classification scheme.
3. Linking Inpatients and Outpatients: That the Maryland charge data base be used to
calculate a set of relative weights for both inpatients RIW and outpatients DPG.
4. 1995/96 Funding: That Equity Funding Formula (i.e. Case Cost Formula) calculations
for 1995/96 incorporate day procedures using 1993/94 data (or the most current data
available).
(from JPPC Discussion Paper: Report & Recommendations for Day Surgery/Procedure
Funding, Draft 1.7, November 21, 1994, p. 1.)
27
APPENDIX IV
Hospital Example of Calculations in the Day Surgery Incentive Model
A hospital example is given to help clarify the different components of the Day Surgery
Incentive Model and to show how it will be applied to the Equity Funding Formula.
Verification of Exclusions Applied to the Outpatient Database in Step 1
H ospital
Total
Outpt
Cases
x
Total
Outpt
RIW
9000
Total
Subs1
Cases
2000
Total
Subs1
RIW
6000
N on
Subs1
Cases
1500
N on
Subs1
RIW
3000
Subs1
Excl
Cases
500
Subs1
Excl
RIW
1000
N et
Equity
Cases
200
N et
Equity
RIW
5000
1300
Total Outpt (Oupatient) Cases refers to those cases for which hospitals submitted a Same Day
Surgery abstract to CIHI.
Total Subs1(Sub-Service) Cases refers to those outpatient cases coded by the hospital as subservice 1 (i.e. day surgery).
Non Subs1 Cases refers to those outpatient cases coded by the hospital as a sub-service
other than subservice 1.
Sub1 Excl (Excluded) Cases refers to those day surgery cases that were excluded due to the
exclusion criteria outlined in step 1.
Net Equity Cases refers to those outpatient cases that are eligible for the Day Surgery
Incentive Model and the Equity Formula calculation because they passed all of the exclusion criteria for qualifying as day surgery cases.
Calculation Weighted Case Adjustment for Hospital X (Step 4)
Col 1
Col 2
Col 3
Col 4
Col 5
Col 6
Col 7
Col 8
Col 9
Col 10
ICD -9
Procedure
H osp Inpt
& Outpt
Cases
% Ont
Inpt
Cases
H osp
Expected
Inpt
H osp
Actual
Inpt
Expect
m inus
Actual
Avg Inpt
RIW
Avg
Outpt
RIW
RIW
D iff.
Wtd Case
Adjust
Col 4
Col 2 & 3
Carpel
Tunnel
90
11%
Col 8
7
Col 9 & 6
5
10
10
0
0.9
0.7
0.2
0
D & C
50
0%
0
5
-5
0.8
0.6
0.2
-1.0
H ernia
Repair
100
14%
14
7
7
1.1
0.6
0.5
3.5
24
22
Total
+2.5
Calculation of the Hospital X’s Case and Weighted Case Indices (Step 5)
Col 1
Col 2
Col 3
Col 4
Col 5
Col 6
Col 6
Hospital
Name
Hospital
Expected
Inpatient
Cases
Hospital
Actual
Inpatient
Cases
Actual /
Expected
Case Index
Hospital
Expected
Inpatient
Wtd Cases
Hospital
Actual
Inpatient
Wtd. Cases
Actual/
Expected
Weighted
Case Index
X
24
22
(Col 3/ Col 2)
.9167
24.4
20.7
As reported on Hospital X's Equity Worksheets for 1993/94:
Total Acute & Newborn Costs + Day Surgery Costs = $200,000
Total Equity Weighted Cases = 100
Average Cost Per Weighted Case = $200,000/100 = $2,000/ wtd. case
Weighted Case Adjustment (as per Day Surgery Incentive Model) = +2.5
Revised Total Equity Weighted Cases = 102.5
Revised Average Cost Per Weighted Case = $200,000/102.5 = $1951.21/ wtd. case
Net Impact of Day Surgery Incentive Model:
% Change in Total Weighted Cases = 2.5%
% Change in Average Cost Per Weighted Case = 2.5%
.8483
C H A P T E R
3
BARBARA WILLIS, ECK HOFFMAN, JOLYN LAWRENSON, VALERIE SMITH
Program Information: The
Application of Information
to Clinical Decision Making
CHAPTER OVERVIEW
This chapter describes the experience of the London Health Sciences Centre (LHSC) in the
application of information to planning and decision making. The Program Information
Department was established at the LHSC to provide leadership in facilitating the application of
information to management of both clinical and administrative processes. Team members,
known as Program Information Specialists, are aligned to specific clinical programs as the
information resource and expert analyst for the clinical team. The chapter describes how this
collaborative effort has enabled teams to translate health care data into information that
enables planned change in practice patterns and approaches to care.
Three case examples illustrate how internal and external information resources have been
effectively applied to planning and decision making within the LHSC. Case 1 describes the
development of utilization targets in operational planning; case 2 describes how changes from
inpatient to ambulatory care were facilitated; and case 3 deals with planning for consolidation
of clinical services. These cases illustrate the effective transformation of health care data to
information and the facilitative role of the Program Information Specialist in this process,
working directly with the clinical teams as an expert resource and support.
As health care is evolving toward integrated delivery systems and more regionalized planning
models, the information requirements are also changing. These developments require access to
and meaningful application of information. Information remains a critical success factor in
achieving a preferred future. At the LHSC, the Program Information Department is evolving
to meet these challenges posed by the new dynamic health care environment.
INTRODUCTION
In recent years, the health care industry has experienced unprecedented reductions in funding from government sources and increased demand to maximize operational efficiency.
This has resulted in enormous pressure within the industry to maintain the standards of
care Canadians have come to expect while responding to the economic imperative. Many
jurisdictions have looked to new organizational models and partnerships to facilitate adaptation to this challenge.
30
PROGRAM INFORMATION: THE APPLICATION OF INFORMATION TO CLINICAL DECISION MAKING
In London, Ontario, the community envisioned a future
that included exemplary health care for the residents of
Southwestern Ontario and a leadership role in the provision of health services education and research. They
responded by supporting the realignment of institutional
health care services within London. The London Health
Sciences Centre (LHSC) was established in September
1995 through the merger of two acute care teaching hospitals affiliated with the University of Western Ontario,
University Hospital and the two sites of Victoria
Hospital, South Street Campus and Westminster Campus.
The priorities for the LHSC were established within the
context of the corporate vision, the provincial health services restructuring process, and the economic imperative
created by Ministry of Health funding levels. Among
these priorities were:
² safeguarding and maintaining our commitment to
high quality patient care, teaching and research;
² fulfilling the LHSC mission effectively and
efficiently within available resources, including the
pursuit of innovative and appropriate business
opportunities; and
² consolidating LHSC from three to two sites.
It was recognized that one of the keys to successfully
achieving these outcomes was the application of information to planning and decision making. The need for leadership in managing information resulted in the establishment of a new department, Program Information, within
the division of Finance and Information Management.
The Program Information Department was developed in
April 1996 to facilitate the application of information to
management of both clinical and administrative processes. As Backer (1995) describes, a "diverse and creative
palette of strategies [is] needed to change clinical practice
at the individual and organizational levels." Thus, the
department was designed to bring together a team of
individuals from throughout the organization, with a
range of skills and experience in utilization management,
clinical pathway planning and evaluation, operational
reviews, program planning, case costing analysis, and education in the application of health care information. Each
of these individuals, titled Program Information
Specialists, was aligned to specific clinical programs as the
information resource and expert analyst for the clinical
team. As Shortell and his colleagues have stated:
“personnel directly involved in the caregiving
function typically lack the time and all of the
needed expertise to do what is required …
Caregivers can identify the problem, and of
course, be held responsible for implementation,
but they must be given resources and support in
order to transform clinical processes and better
coordinate such processes across the continuum. (Shortell et. al., 1996, p. 174)”
The Program Information Specialist role was designed to
be dynamic and responsive to the needs of each clinical
team. The Specialists collaborate with the team to determine the information needs associated with addressing
issues specific to their area, and develop mechanisms for
disseminating that information in meaningful ways.
"Striking examples where the intelligence of the team
exceeds the intelligence of the individuals in the team and
where teams develop extraordinary capacities for coordinated action" are described by Senge (1990). It is this collaborative effort and the facilitative role of the Specialist
that have enabled teams to translate health care data into
information that enables planned change in practice patterns and approaches to care. Senge further suggests that:
“the fundamental "information problem" faced
by managers is not too little information but too
much information. What we most need are ways
to know what is important and what is not
important, what variables to focus on and which
to pay less attention to—and we need ways to
do this which can help groups or teams develop
shared understanding. (Senge, 1990, p. 128)”
Based on this, it was concluded that a proliferation of
health care data within the clinical team was clearly not the
answer. A focused approach was taken to develop standard
reports that would meet generic information needs and
then to develop customized information products consistent with the needs of the specific clinical situation.
Reports consisting of internal utilization and costing data
were distributed to the clinical teams. In addition, information provided by the Canadian Institute for Health
Information (CIHI), the Joint Policy and Planning
Committee (JPPC) and the Ministry of Health (Ontario)
including the Planning Decision Support Tool (PDST)
and the Day Surgery Incentive Model was used extensively. These resources provided data for Case Mix Group
(CMG) and Day Procedure Group (DPG) analysis,
benchmarking to best practices, and case costing analysis.
The information was applied to a range of processes:
operational planning, clinical pathway development and
variance tracking, program planning, and peer review. The
restructuring initiatives within the Province of Ontario
rely heavily on these same data resources.
31
PROGRAM INFORMATION: THE APPLICATION OF INFORMATION TO CLINICAL DECISION MAKING
Three case examples are described in this chapter to illustrate how internal and external information resources
have been effectively applied to planning and decision
making within the LHSC. As Backer (1995) states, we
need to "think globally and act locally ... micro-change is
represented by practice change for physicians and hospitals; macro-change is represented by health care reform
and other changes in the whole health care system." The
impact of these types of changes at LHSC is illustrated in
the three cases presented below.
CASE 1: OPERATIONAL PLANNING—
DEVELOPMENT OF UTILIZATION
TARGETS
To facilitate restructuring of health services within the
province of Ontario, the Ministry of Health has established the Health Services Restructuring Commission
(HSRC). The role of the Commission is to make decisions about health system restructuring that will assist
local, district and regional processes to move forward logically and sensibly. This process is to occur while accounting for quality of care, management and administrative
efficiency, broader health system integration, availability
of capital and operating resources, accessibility to health
services in the community, and other relevant factors
(HSRC, 1997).
In July of 1996, the Health Services Restructuring
Commission undertook a review and analysis of the
London region. With the Commission report and directives pending, the LHSC proceeded in November 1996
with budget planning for fiscal 1997/98. Critical organizational decisions needed to be made that would address an
expected 8% reduction in Ministry of Health funding, yet
ensure alignment with the anticipated recommendations
of the Commission. In order to forecast utilization patterns, define inpatient bed requirements, and develop
operational plans, it was decided that the HSRC utilization analysis methodology be adopted based on the two
previously published reports in Thunder Bay and
Sudbury, Ontario.
The primary utilization tool employed for the analysis was
the Ontario Ministry of Health Planning Decision
Support Tool (PDST). The PDST is designed to assess
hospital utilization patterns against provincial targets and
benchmark performance levels in key categories of hospital activity (PDST, 1996). It was decided that the standard
targets would be applied to LHSC utilization data according to the following assumptions:
² average length of stay (ALOS) would be reduced to
achieve the 50th percentile relative to comparable hospitals using benchmark data available through CIHI;
² preoperative days would be reduced by 100% for
elective admissions;
² alternative level of care (ALC) days would be
reduced by 50%;
² patient days related to CMG 851(social admissions)
and CMG 910 (diagnosis not generally admitted)
would be reduced by 100%;
² may not require hospitalization (MNRH) days would
be reduced by 25%; and
² day surgery would meet the 75th percentile benchmark
according to the JPPC Day Surgery Incentive Model.
The targets established for the organization were developed with a view to modifying the HSRC approach to
meet the specific needs and interests of the LHSC and
the one year time horizon for the budget process. The
assumptions were generally consistent with the PDST
except for the following:
² ALC days were reduced by 50% as opposed to
100% given the one year time frame;
² occupancy rates were established at 85% versus
90%; and
² benchmark ALOS was set at the 50th percentile in
recognition that the 75th percentile level was a multiyear target and would be difficult to achieve in one year.
The Program Information team accessed Health Records
data to provide CMG specific information for the two
campuses (Victoria and University) and determine the
number of cases and days currently associated with each
of the variables. The assumptions were then applied to
the data to establish utilization targets for 1997/98, as
shown in Table 1.
Table 1: Conservable Patient Days Fiscal
1997/98
Strategy
1. Reduce 50% ALC days
Total Days
to be Saved
1,601
2. Eliminate 100% CMG 851
307
3. Eliminate 100% CMG 910
109
4. Reduce 25% MNRH
5. Adjust to 75th percentile day surgery
1,966
2,600
6. Adjust to 50th percentile ALOS
21,938
Total
28,521
Elimination elective pre-operative days
5,004
Adapted from the Planning Decision Support Tool (1996)
Ontario Ministry of Health.
32
PROGRAM INFORMATION: THE APPLICATION OF INFORMATION TO CLINICAL DECISION MAKING
While this analysis established a goal for the entire hospital, micro analysis needed to occur that was specific to
doctor service and would facilitate planning of operational changes to address these targets. Therefore, a further step was undertaken by the Program Information
Specialist to apply the amended PDST tool to each doctor service for the two campuses. The data were further
categorized based on elective, urgent or emergent status.
These were further subdivided to capture whether or not
pre-admit workup had occurred; whether or not it was a
same-day admit for surgery; the number of pre-operative
days; whether or not the case was transferred in from
another institution; whether or not the case came through
the emergency department; the total number of cases;
and the total number of days.
While the target reduction was established at approximately 28,500 days for the organization, this additional
detailed analysis indicated that 5,000 of these days could
be achieved by eliminating all pre-operative days for elective cases only (see Table 1). Furthermore, the detailed
analysis by doctor service provided the Program
Information Specialists with insight into the clinical areas
where savings could be achieved for specific categories.
For example, the majority of ALC days were attributable
to the Neurology service at the University Campus and
Family Practice at the Victoria Campus. Differences in
patient populations at the two campuses were clearly illustrated and, as a result, strategic decisions were more carefully
considered. Areas where improvements could be made in
terms of operational efficiencies became apparent.
The experience of the Musculoskeletal program illustrates
how the application of information to decision making
and operational planning was achieved. Analysis of the
program specific information indicated a high number of
pre-operative days for urgent cases. An analysis by CMG
indicated that the issue was predominantly associated with
patients with hip fracture where the average number of
pre-operative days per case was 2.3 (see Table 2). The
Program Information Specialist facilitated discussion
among the Chief of Orthopaedics, Manager and
Coordinator of the program to identify the system issues
that would account for this number. A review of operational processes indicated that Operating Room (OR)
scheduling was a contributing factor. Through a collaborative effort with the Operating Room leadership, a
change in OR scheduling is being considered that will
improve access for urgent cases during the week. This will
not only reduce unnecessary days for the service, but will
achieve a more patient centred approach to care. In addition, the Musculoskeletal program planned to reduce 25%
of MNRH cases and shift inpatient cases to day surgery
where appropriate. In cases where the surgeons felt it
imperative to admit patients with designated MNRH procedures, the patient population was aggregated with other
CMG with a length of stay of less than 4 days. This
allowed a critical mass to be identified that could sustain
the implementation of a 5-bed short-stay unit that would
only be operational Monday to Friday and consequently
achieve a reduction in paid hours.
Table 2: Pre-Op Days for Orthopaedic Admissions
Admission Category
Elective
CMG™
356 Fractured Femur Proc
with CC
Cases
10
Average
357 Fractured Femur Proc
w/o CC
Average
12
Urgent
Emergent
Grand Total
Total Pre-Op Cases Total Pre-Op Cases Total Pre-Op Cases Total Pre-Op
Days
LOS
Days
LOS
Days
LOS
Days
LOS
(Days)
(Days)
(Days)
(Days)
125
12
12.5
1.2
75
10
6.3
0.8
73
88
1219
183
16.7
2.5
871
128
9.9
1.5
4
0
45
4
11.3
1
0
0
87
100
1389
199
16
2.3
946
138
9.5
1.4
PROGRAM INFORMATION: THE APPLICATION OF INFORMATION TO CLINICAL DECISION MAKING
In order to achieve the 75th percentile length of stay
benchmarks for the peer group, the Musculoskeletal program implemented clinical pathways and established target lengths of stay for all high volume cases. To facilitate
this, extensive education for staff, physicians, and physicians’
secretaries was undertaken to ensure a uniform communication strategy to patients and their families around expected
hospital stay. As a result of these combined utilization strategies, a reduction of five inpatient beds, an increase in day
surgery activity, and the implementation of a short-stay unit
were achieved for the Musculoskeletal program without
affecting patient throughput.
Overall, clinical services throughout the organization
determined what utilization efficiencies they could
achieve, and these were subsequently matched to bed
requirements and outpatient capacity. This approach
resulted in expansion of preadmission and one-day stay
programs as well as the closure of 85 beds for fiscal
1997/98, all being achieved while sustaining the corporate
commitment to meeting service demand. Subsequent to
this planning process, the LHSC has been compelled to
review utilization at the 90th percentile for fiscal 1997/98
as a result of continued financial constraint. The result of
this review has not been concluded at this time.
CASE 2: FACILITATING CHANGES IN
CARE MODALITY—INPATIENT TO
AMBULATORY CARE
In February 1996, the Funding Integration Sub-Committee
of the Joint Policy and Planning Committee released a
report on day surgery activity comparing activity within specific institutions to the provincial average. This information
related directly to a Day Surgery Incentive/Disincentive
funding methodology that had been adopted by the
Ministry of Health and applied to funding allocation decisions within the province. The incentive was designed to
encourage hospitals to achieve a benchmark level of performance for day surgical procedures. The report stated that
"hospitals completing more procedures on a day surgery
basis relative to the provincial average would receive a financial incentive, while hospitals completing fewer procedures
on a day surgery basis relative to the provincial average
would receive a financial disincentive" (JPPC, 1996). The
LHSC proceeded to fully examine practice patterns in comparison to the provincial experience.
The JPPC report provided DPG data as well more specific information at the ICD-9-CM procedural code level.
The DPG data enabled the Program Information
Specialist to identify where the most significant opportunities for improvement existed in the percentage of day
surgery procedures compared to the provincial average.
33
Where activity was adversely affecting funding levels, further analysis was conducted at the ICD-9-CM code level
of detail. One of the limitations in the analysis was the
historical nature of the data. A number of clinical areas
had realized improvements in practice during the current
fiscal year that would not be evident in the data. In an
effort to overcome these limitations, information for the
most recent fiscal year was extracted from the Health
Record database and the methodology used by the JPPC
was applied. The analysis focused on typical inpatient
cases with a length of stay less than four days. These
cases provided the highest probability of conversion to
day surgery.
The data were developed into a table that illustrated cases
attributable to day surgery and typical inpatient activity
during the past two fiscal years. This approach enabled a
high level analysis which could then be used to set corporate objectives for achieving changes in care modality.
Within the clinical teams, the Program Information
Specialists provided information on the JPPC methodology and the comparison of specific practice patterns at the
LHSC. Where areas of opportunity for improvement
were identified, further analysis was provided to identify
variability in surgeon practice patterns.
A number of clinical programs were able to benefit from
this process. For example, the General Surgical program
identified laparoscopic cholecystectomy and hernia repair
as procedures which could potentially shift to day surgery.
Table 3 provides an illustration of these procedures over
a two year period. The information provided to the team
served to validate the direction in which the team had
been proceeding in development of clinical pathways for
ambulatory care. The clinical team had identified effective
pain control as a major obstacle which prevented physicians from moving their patients to day surgery. As part
of the clinical pathway process, a literature search was
undertaken and the principles of evidence-based research
were applied to the resulting articles. The physicians were
asked to review the articles that addressed the options for
pain control. They identified proven pain control techniques that were integrated into the clinical pathways.
This provided a level of confidence that facilitated moving to the ambulatory care model.
34
PROGRAM INFORMATION: THE APPLICATION OF INFORMATION TO CLINICAL DECISION MAKING
Table 3: Inpatients with <4 Days Length of Stay and Day Surgery (University Campus)
(Procedures with at least 1 Inpatient and 1 Day Surgery occurrence)
1995 96
Procedure No & Description
LAPAROSCOPIC CHOLECYSTECTOMY
HERNIA REPAIR
1996 97
Total
Total
%
Total
Total
%
Inpatient
Day Surgery
Day Surgery
Inpatient
Day Surgery
Day Surgery
152
2
1%
112
58
34%
44
9
17%
21
45
68%
Another focal area was Ophthalmology. Analysis based
solely on length of stay benchmarks led to the conclusion
that all inpatient cases had an average length of stay
below that of the provincial standard. The Day Surgery
Incentive model, on the other hand, highlighted that over
half of these cases could be managed on a day basis. This
provided the impetus for change. Understanding the day
surgery benchmarks for retinal procedure, phakofragments, and lens procedures led to the development of
admission criteria for the program. Together, the Chief of
Ophthalmology, Manager, Coordinator and Program
Information Specialist established guidelines and admission criteria that were utilized by the ophthalmologist in
determining appropriateness of admission versus day
surgery. Within several months, a shift in practice pattern
was clearly evident.
These examples illustrate how utilization information was
successfully applied in planning practice changes in a
number of clinical areas. The data were also useful in
projecting activity levels for fiscal 1997/98 as part of the
operational planning process described in the previous
section of the chapter. By using the tools available and
adapting these for the specific needs of the organization,
effective change has been realized.
CASE 3: PLANNING FOR CONSOLIDATION OF SERVICES—CLINICAL
NEURO SCIENCES
One of the many reasons why hospitals merge is to be
able to provide improved service and reduced costs without compromising patient volume. These were significant
objectives of the merger for the LHSC.
Early in the process, the Clinical Neuro Sciences (CNS) service was identified as a clinical service ready and willing to
consolidate onto one site, namely, the University Campus.
Historically, like most tertiary care hospitals, each had its
own neurology/neurosurgery service. The merger provided
the opportunity for consolidating the service in order to
provide improved patient care and physician training, and to
achieve operational efficiencies. A pre- and post-consolidation look at the CNS experience illustrates the application
of information to program planning.
The "old" model (pre-consolidation) offered adult CNS services at both campuses from specialist physicians, with the
exception of epilepsy diagnosis and treatment which was
centred at University Campus for both adult and paediatric
cases. In the newly consolidated model, all adult CNS services are located at University Campus. The use of clinical
and financial information was central in the process of
planning for the consolidation of CNS services.
Clinical Information
In order to plan or corroborate the number of beds to be
transferred from one Campus to the other, a review of
the number of cases by doctor service was carried out.
This information provided a snapshot of the current
activity at each campus. Since Family Practice and
Medicine would continue to see certain neurological cases
at Victoria Campus, only cases in which the Doctor
Service was either Neurology or Neurosurgery were evaluated. The data were categorized into those patients who
were classified as elective, urgent or emergent admissions.
The data were collected with an age split which allowed
the Program Information Specialist to determine the
number of paediatric patients who would no longer be
serviced at University Campus. Since the emergent category of patients would continue to be treated at the
Emergency Room in which they presented, only urgent
and elective patients were reviewed in detail.
The number of Emergency Room visits in which the
CNS service played an important role was determined for
both campuses as a basis for predicting the volume of
patients which would be redirected. In general, patient
volume data provided information that was essential in
defining the resources required at the "recipient" campus—from the number of beds required to the staffing
and other operating budget requirements. CMG data were
also utilized to determine the number of trauma patients
PROGRAM INFORMATION: THE APPLICATION OF INFORMATION TO CLINICAL DECISION MAKING
who would be redirected to Victoria from University
Campus. In addition, CMG data were useful in determining the number of surgical patients who would require
the Operating Room and ICU resources.
Data were collected with respect to the diagnostic services
utilized, specifically laboratory and imaging services. This
was important in order to allow the diagnostic services to
reassign staff depending on the volume of work which was
to be transferred from one campus to the other. For example, it might be necessary to relocate an imaging technologist if the capacity of the imaging department at the recipient campus did not exist. The information also provided
the imaging department with an opportunity to redistribute
resources within supply budgets.
Financial Information
Financial information was essential in redeploying
resources from one budget centre to another. Knowing
the patient volumes within the different components of
CNS along with their associated costs provided an opportunity to calculate the total dollars involved in the care of
the patient population. Total costs included nursing costs
and associated unit expenses, as well as costs incurred in
diagnostic and support areas including laboratory, radiology, pharmacy, operating room, health records, food services, and others. An example of the total costs associated with a particular population of CNS patients is illustrated in Table 4.
Table 4: CNS Consolidation
Neurosurgery
6040
6045
6050
6052
6085
6105
6120
6235
6325
6330
6335
6340
6345
6354
6358
6452
6456
6611
6615
6616
6617
6625
6652
6662
6680
6710
6720
6732
6742
7405
7500
7556
7558
7564
7805
7825
35
Actual Variable
Direct Cost ($)
CCU
DAY SURGERY
CRITICAL CARE TRAUMA CENTRE
CCTC TECHNICAL SUPPORT
CNS 8 MDX
BURNS/PLASTICS/SURGERY
LEVEL 7,TOWER 2,(ONCOLOGY)
7 E&M PSYCHIATRY
6C UROLOGY
3W (VASC SURG)
2 W (CARDIOVASCULAR)
3M (ORTHOPAEDIC)
3E (ORTHOPAEDIC)
4B FAMILY MEDICINE
5E MEDICINE SSC
OPERATING ROOM
POST ANAESTHETIC CARE UNIT
ANATOMICAL PATHOLOGY
INTEGRATED LAB SSC
INTEGRATED LAB WC
BLOOD BANK S S
CLINICAL MICROBIOLOGY
HEMATOLOGY S S
NUCLEAR MEDICINE
DIAGNOSTIC CYTO PATHOLOGY
PHARMACY
RADIOLOGY
PHYSIOTHERAPY
OCCUPATIONAL THERAPY
PATIENT REGISTRATION
HEALTH RECORDS DEPARTMENT
PATIENT FOOD SERVICES
NUTRITIONAL SUPPLEMENTS
CLINICAL NUTRITION SERVICES
PATHOLOGISTS
PROFESSIONAL FEES DIAG RAD
Total Neurosurgery
22
4,803
28,248
416
225,978
3,311
3,101
36
1,533
812
1,495
828
612
1,005
1,151
231,645
23,011
14,526
7,417
256
4,741
2,257
10
896
309
50,418
23,656
11,027
6,450
5,024
7,131
6,461
50
2,700
0
18,623
$689,959
36
PROGRAM INFORMATION: THE APPLICATION OF INFORMATION TO CLINICAL DECISION MAKING
Costs associated with Operating and Post Anaesthetic
Recovery Rooms form a significant component of the
total costs for surgical CNS patients. An example showing
the utilization of the Operating Room by a population of
CNS patients is shown in the top section of Table 5. For
the most part, direct variable costs were used throughout
the consolidation planning process. Table 5 illustrates one
approach in determining the supply costs which may be
transferred to the "recipient" Operating Room.
Collecting and interpreting the patient volumes, as
described above, also provided information necessary to
determine the staffing needs at both the "donor" and
"recipient" campuses. This exercise required insight into
work patterns because it was not always possible to move
staff and patients in the same proportion. Indeed, in
some situations in the CNS consolidation, additional staff
were required. For example, the need for EMG and EEG
services at the Victoria Campus would continue despite
the consolidation of the majority of services to the
University Campus. This required additional paid hours.
However, economies of scale were achieved in other
areas such as inpatient units which resulted in reduced
total staffing.
Table 5: OR Expenses
Operating Room Costs
Admit Category
Doctor Service
NEUROLOGY
Data
Elective
Sum of Var Dir Cost
Urgent
524
Sum of Cases
NEUROSURG
1
Sum of Var Dir Cost
252,053
Sum of Cases
189
Total Sum of Var Dir Cost
252,577
Total Sum of Cases
190
Grand Total
10,674
15
173,532
139
184,206
154
11,198
16
425,585
328
436,783
344
Operating Room Budget
Total
Materials
12,361,639
6,135,040
Materials of Total Budget (%)
49.6
Labour of Total Budget (%)
50.4
Distribution of O.R. Direct Variable Costs for Neuro and Neurosurgery
T1 Direct Variable Costs (from Table)($)
436,783
Materials ($)
216,774
Labour ($)
220,009
"T1 Target"
The dollars actually transferred between sites were based
on analysis of the data and on negotiation between the
key stakeholders. Once the amounts were determined, the
transfer of both staffing and supplies dollars between
Victoria and University Campuses was tracked on a
spreadsheet. The sheet clearly illustrated the changes in
staff mix as well as the total financial impact on LHSC
resulting from the consolidation. The template used in
the consolidation of CNS is shown in Appendix I. This
template has subsequently been used in the consolidation
of other LHSC programs and for planning service moves
with other hospitals in the area.
analysis of both clinical and financial information allowed
the planning team to make objective and informed decisions and to understand the potential economic implications of subsequent decisions.
The cornerstone of the consolidation process was a clear
vision of the new model and its objectives for the future.
Implementation of the new model required a significant
amount of clinical and financial information, information
which is inextricably linked together. The collection and
As health care is evolving toward integrated delivery systems and more regionalized planning models, the information requirements are also changing. The increasing
emphasis on evidence-based practice, clinical pathway
development and monitoring, and more aggressive stan-
CONCLUSION
These cases illustrate the effective transformation of
health care data to information and the resulting application of this information to a range of planning and decision-making processes. The Program Information
Specialist has facilitated this process by working directly
with the clinical teams as an expert resource and support.
PROGRAM INFORMATION: THE APPLICATION OF INFORMATION TO CLINICAL DECISION MAKING
dards for best practice benchmarking are demanding current and accessible information. As program management
structures evolve, more members of the health care team
participate in planning and decision making; they need to
understand the information and its application to management. The new partnerships and alliances that are
developing in response to a more integrated care system
have needs for sharing information and creating comparability across organizations and into the community. The
need to better understand the needs of the population and
the referral patterns of care is gaining greater relevance.
All of these developments require access to and meaningful
application of information. Information remains a critical
success factor in achieving our preferred future. At the
LHSC, Program Information is evolving to meet these challenges posed by the new dynamic health care environment.
37
38
PROGRAM INFORMATION: THE APPLICATION OF INFORMATION TO CLINICAL DECISION MAKING
REFERENCES
Backer, T.E. (1995). Integrating behavioural and systems
strategies to change clinical practice. The Joint Commission
Journal on Quality Improvement 21 (7), 351-353.
JPPC (1996). Revised 1994/95 day surgery incentive model.
Day Surgery Incentive Model Adhoc Group, the Funding
Integration Sub-Committee of the JPPC. Toronto, ON:
Joint Policy and Planning Committee.
HSRC (1997). London Health Services Restructuring Report.
Toronto, ON: The Health Services Restructuring
Commission.
PDST (1996). Planning Decision Support Tool. Toronto, ON:
Ontario Ministry of Health.
Senge, P.M. (1990). The fifth discipline: The Art and Practice of
the Learning Organization. New York, NY:
Doubleday/Currency.
Shortell, S.M., Gillies, R.R., Anderson, D.A., MorganErikson, K., and Mitchell, B. (1996). Remaking health care in
America: Building Organized Delivery Systems. San Francisco,
CA: Jossey-Bass Publishers.
About the Authors
Barbara Willis, MBA, CHE is Manager, Program Information, London
Health Sciences Centre, London, Ontario;
Eck Hoffman, MSc, MBA and Jolyn Lawrenson, CCHRA(C) are
Program Information Specialits, London Health Sciences Centre; and
Valerie Smith, MSc, MBA, CHE was a former Program Information
Specialist, London Health Sciences Centre and is currently
Administrative Officer, Children’s Hospital of Western Ontario, University
of Western Ontario, London, Ontario.
Acknowledgments
The authors would like to thank Doris Hotz, Departmental Secretary, London
Health Science Centre, for her assistance with preparing this chapter.
APPENDIX I
39
Program Transfer Template
CURRENT
PROPOSED
Victoria
University
8 Middlesex
CNS program
Consol Ortho
FTE
FTE
FTE
Dollars
Dollars
Victoria
Dollars
Paediatrics
FTE
Dollars
London Health Science Centre
Trauma
FTE
Dollars
Rehab
FTE
Dollars
Corporation
FTE
Summary
Dollars
FTE
1 Inpatient Unit
Reg Nurse
Reg Practical Nurse
Unit Clerk
Orderly
Coordinator
Total
Supplies
CNS/NP
Drugs (from Pharmacy)
In-Patient Summary
2 Applied Health Services
Physiotherapist
Occ Therapist
Social Worker
Neuropsychologist
Dietician
Psychometrist
Sp Lang Path
Applied Health Summary
3 Critical Care
Reg Nurse
Supplies
Drugs (from Pharmacy)
Critical Care Summary
4 Operating Room/PACU
PSA
Staff :CN/RN/ORT
Supplies
Drugs (from Pharmacy)
Operating Room Summary
5 Emergency Transfer
Dir Var Costs - evaluate over time
Emergency Summary
6 Outpatients
7 Radiology
8 Laboratory
9 Current - Proposed
-
-
-
Dollars
C H A P T E R
4
LYNN M. NAGLE, JUDITH SHAMIAN, MARGARET CATT, MARTIN STEIN, JOSEPH MAPA
Optimizing Clinical
Utilization: Structure and
Strategies
CHAPTER OVERVIEW
Achieving a balance between cost-savings and sustaining the quality of patient care is a
challenge for all health care organizations in Canada today (Leatt, et al., 1996). Many organizations have adopted innovative approaches to support effective and efficient management of limited hospital resources (Sheps, Anderson, and Cardiff, 1991). These approaches
have included new organizational structures, models of care delivery, and strategies for
resource utilization management. In this chapter, we describe the development and evolution of one hospital's structures, strategies, and processes to optimize resource utilization.
Specific clinical examples are used to demonstrate the application of data from the
Canadian Institute for Health Information (CIHI). A case study describes the successful
application of utilization tools and strategies by one of the Clinical Teams in order to optimize resource utilization.
BACKGROUND
The Organization
A number of external factors have provided the impetus for health administrators to analyze the focus and mechanisms of care delivery within health care organizations.
Specifically, funding reductions and hospital restructuring have brought issues of appropriate and effective utilization of resources to the fore. Moreover, changes to the hospital
accreditation standards and processes necessitated a rethinking of the existing clinical management infrastructure.
The development of Mount Sinai Hospital's (MSH) utilization management program
reflects the transformation of the Canadian health system and the focus on cost reduction.
In this context, MSH initiated a multi-dimensional, multi-disciplinary strategy aimed at
improving efficiency while maintaining quality of service rendered. Three fundamental
tenets provided the basis for this strategy:
1.
2.
3.
excellence, which not only pertains to quality care but also includes using resources
effectively to provide cost-efficient care without compromising quality;
continuous improvement, which recognizes the need for continual reassessment of the way
in which we use our limited resources; and
responsiveness, through the creation of an infrastructure that facilitates excellence and
continuous improvement in an environment of cost constraints.
42
OPTIMIZING CLINICAL UTILIZATION: STRUCTURE AND STRATEGIES
In 1995, the Hospital established a new organizational
structure to manage the affairs of the institution and support the principal foci of quality and utilization management in accordance with these tenets. Essentially, the new
infrastructure consists of three multi-disciplinary
Planning Councils with overall responsibility for a number of related patient care teams. While the Councils have
an overarching macro-perspective, the patient care teams
focus on micro-level activities associated with the care of
specific patient populations (e.g. Surgical Council,
Muskuloskeletal Team). This model was selected to facilitate more decentralization of more decision-making,
engage the largest number of constituents, and operationalize the implementation of necessary changes, while
respecting the needs of the patients being served. In
addition, the utilization program needed to develop initiatives that would effect changes across councils and teams.
Given our organizational culture, it was recognized early
on that the involvement of clinicians from all disciplines
would be critical to the success of a utilization program.
Thus, we have built the organizational structure with multiple teams and multi-disciplinary membership.
In sum, the model was designed to support the operationalization and optimization of utilization management
strategies. This structure provides the platform for distributing the accountability for clinical utilization to interdisciplinary teams. Moreover, a convergence of clinicians
and administrators provides an opportunity to optimize the
decision-making associated with clinical cost reductions.
More detail about the specific work of the structure's constituents follows in a discussion of roles and responsibilities.
The new structure was implemented in the summer of
1995 and in May of 1996, the hospital was scheduled for
accreditation. Many team members believed learning the
new accreditation standards and preparing to respond to
the accreditors as a team rapidly greatly facilitated team
development. They also perceived the multiple team discussions pertaining to the standards helped them learn
more about each others’ role and system wide issues.
Teams were aggressively reviewing their performance in
relation to accreditation and simultaneously discussing
methods to reduce length of stay for their target populations. This ground work set the stage for each team to
develop and implement clinical utilization strategies.
Each team is co-chaired by a physician and another professional (e.g. nurse, social worker, respiratory therapist).
The co-chairs of the teams were asked to appoint a
Resource Utilization (RU) liaison who would be the designated change agent for this effort. In addition, each team
identified a Quality Improvement Liaison and Patient
Care Process representative (see Figure 1). Each of these
individuals is accountable to the team for certain responsibilities including attendance at educational sessions provided by the hospital and standard meetings. The liaison
for each corporate initiative may be the expert for the
team or ensure that individuals with specific expertise
meet with the team on a regular basis.
Figure 1:
Organizational structure
for resource utilization
management.
OPTIMIZING CLINICAL UTILIZATION: STRUCTURE AND STRATEGIES
43
A Resource Utilization Management Committee (RUMC)
was created to develop the utilization strategies and to
support the Councils and Teams in their implementation.
In order to bridge the financial and clinical perspectives,
RUMC is co-chaired by the Vice-President Finance and
the Vice-President Nursing, since in our opinion, these
perspectives are inseparable when it comes to cost-control and utilization management. RUMC includes a broad
multi-disciplinary membership and a close working relationship with the Councils and Teams through the teambased Resource Utilization Liaisons.
agement and practice patterns (Payne, 1987). To this end,
many of the concepts found in total quality management
(TQM) are employed in these processes (Cyr, Secord, and
Prendergast, 1995). The work of the Resource Utilization
Management Committee is therefore encapsulated by the
following three questions:
RUMC established an educational subcommittee which
was charged with the responsibility of developing a multifaceted educational strategy. The subcommittee's educational strategies were of varied breadth and depth for different groups. Each team RU liaison, at least one other
team member, and the Directors and Management team
attended a two day in-house educational session offered
in December, 1995. This workshop provided:
Roles and Responsibilities
The Resource Utilization Management Committee
(RUMC) was given the mandate to develop, implement,
and evaluate a comprehensive, integrated resource utilization program including components of utilization review
and utilization management. In this regard, RUMC provides support to the councils and patient care teams to
develop and implement utilization management initiatives.
This committee is also responsible for the conduct of
external scanning and impact analysis related to pending
changes (e.g. funding recruitment, program shifts, etc.).
Educating the members of all teams and councils about
issues of resource utilization is also the responsibility of
this group.
² a comprehensive overview of utilization management
and review processes;
² an introduction of CIHI data and terminology including
CMG, average length of stay, case mix and case weights;
² an overview of case costing;
² an overview of reports available from Health
Records and Decision Support and processes to
request data;
² an introduction to benchmarking; and
² a review of Patient Care Process strategies such as
care plan development.
Each RU liaison was provided with a reference binder containing the workshop materials. Members of the councils
and teams attended an abbreviated educational session
which introduced them to the concepts of the utilization
program. These sessions were conducted over a period of
six weeks and served to raise awareness and enhance a
broader understanding of utilization management.
Resource utilization management is the application of
evidence-based practice to resource consumption.
Moreover, resource utilization management represents a
shift away from process management based upon tradition and personal preference, towards process management based upon verifiable data (Sheps, Anderson, and
Cardiff, 1991). However, this shift does not suggest that
there exists a single best method of practice or management. Instead, it encompasses the idea that there are
many possible solutions to the complex problems
encountered in healthcare. In combination, the processes
of utilization management and review allow for the identification, analysis, and evaluation of variations in man-
1.
2.
3.
Are we doing the right things? (Effectiveness)
Are we doing things right?
(Efficiency)
How can we improve the things we do?
(Continuous Improvement)
The three Planning Councils have the responsibility for
determining the volume and case mix for their clinical
populations. They receive quarterly reports from both
RUMC and their respective teams in relation to utilization
performance. The councils participate in utilization management and review, reporting to the Management
Executive Committee and the Medical Advisory
Committee on the utilization performance of their teams.
The Patient Care Teams have responsibility for monitoring volume, case mix, length of stay, and cost per case
and must report to the Planning Councils on a monthly
basis. Each team identifies areas of variance and is
expected to develop, implement, and evaluate utilization
management initiatives to correct undesired variance.
Strategies are developed in partnership with other relevant
teams, departments, and stakeholders. One of the tools
used by the teams in this process is CMG specific data
for the purposes of reviewing and benchmarking their
performance. These data encompass costing and length
of stay comparative data. The identification of opportunities for utilization improvements are followed by discussions of strategies such as the development of critical
pathways and care protocols. As will be described later in
this chapter, many of the patient care teams have developed expertise in this strategy.
44
OPTIMIZING CLINICAL UTILIZATION: STRUCTURE AND STRATEGIES
Each patient care team has designated an individual to be
the team's Resource Utilization (RU) liaison. All of the
team liaisons have been provided with in depth training
about utilization review and management. More specifically, they have developed a high level of understanding
of various utilization reports including CIHI data. Hence,
they are able to provide the teams with utilization review
and management expertise. They are further supported in
these efforts through regular monthly meetings with the
other liaisons and continued development under the
direction of the RUMC. They collaborate with other utilization liaisons to maximize utilization management
opportunities and provide regular updates about utilization activities to the Team. Through the regular meetings
with other liaisons and links to RUMC, these individuals
stay abreast of new external developments (e.g. CIHI and
hospital funding changes) and strategies (e.g. patient care
processes) and bring this information to team discussions.
The liaisons also submit requests for utilization reports
on behalf of the team and in most cases, oversee the
resource utilization review processes of the team. They
are active participants in the resource utilization management initiatives and prepare and submit utilization review
and management reports to RUMC, Councils and others.
These diagnostic processes helped us to identify the areas
where we had the greatest opportunity to achieve
improvements in utilization. Based on the diagnostic data,
and considering organizational factors, RUMC decided
that the first wave of organizational utilization initiative
would focus on LOS reductions. Our analysis demonstrated the potential to reduce current inpatient stays by
approximately 14,000 patient days.
CLINICAL UTILIZATION STRATEGIES
The Resource Utilization liaison for all teams attend a
regularly scheduled meeting to discuss and share team utilization strategies. At each meeting, the RU liaison
updates the larger group about their team's methods to
reach the LOS target assignment. The meetings have also
included presentations from Health Records, Decision
Support, Finance, and CIHI. The RU liaisons have found
these meetings very helpful, particularly when they were
new to the role. RUMC developed a set of standard
reports to assist councils and teams to become familiar
with their performance generally, and in relation to their
specific targets for their assigned LOS.
“Optimalization”, a term used by Petryshen and colleagues (1995), refers to the balancing of costs against
efficient, efficacious, and effectiveness of care (Petryshen,
O’Brien-Pallas, and Shamian, 1995). In establishing a clinical utilization program, our goal was to reach an optimalization of patient care resources. The strategic direction
of the utilization program was developed by RUMC. In
determining the key elements of the utilization program a
number of clinical data sets were examined. The areas of
utilization that were reviewed included:
² Length of stay (LOS) for different CMG and how it
compared to other institution’s LOS.
² Inpatient—Outpatient ratio. How do we compare to
other institutions and what are the benchmarks?
² A comparison of our utilization performance to others was conducted using Canadian and American
comparator hospitals (CIHI, Health Records Data).
² Internally, we used case costing and LOS data to
compare costs and practice variations among physicians and clinical departments. Are there differences
between like cases being cared for by different teams?
For example, is the care of a pneumonia patient
(CMG 140) under Family Medicine similar to the care
of a pneumonia patient by General Medicine? If not,
what can we learn from it?
In January, 1996 each team responsible for inpatient clinical management received an assignment to reduce length
of stay (LOS) for two or more CMG. The target reductions in the assignments were derived from the CIHI
1995/96 database and benchmarks provided by external
consultants. Some of the target LOS in the assignments
aimed for a shorter LOS than the CIHI database.
Upon receiving a team assignment for LOS reduction, the
teams developed mechanisms to implement the corporate
clinical utilization strategies to meet the LOS targets.
Assistance was available to help the RU liaisons understand and analyze the data. Furthermore, expert staff
were available to provide guidance with utilization strategy development and implementation.
Each RU liaison also met individually with the person
overseeing the implementation of the corporate resource
utilization strategies. During these meetings, time was
allocated to compare LOS data for typical patients from
January 1, 1996 to March 31, 1996 to a second time period, April 1, 1996 to June 30, 1996 and similarly for each
subsequent quarter of the year. Team progression toward
the LOS reduction targets and strategies for meeting the
targets continued to be reviewed and discussed. For some
of the target CMG, LOS was reduced significantly. For
example, for CMG 237 (Arrhythmia >70 with complications) average length of stay decreased from 7.0 days to
4.7 days from January to December 1996). Depending
upon success to date, some teams have been assigned
additional CMG targets for LOS reductions.
OPTIMIZING CLINICAL UTILIZATION: STRUCTURE AND STRATEGIES
In order to achieve length of stay reductions of this magnitude and sustain the provision of quality patient care,
several management strategies and review tools were
developed and implemented. Global strategies which were
developed and implemented to achieve length of stay
reductions included: 1) processes for CMG assignment; 2)
clinical tools to streamline patient care processes and
advance evidence-based practice (e.g. critical pathways
and protocols); 3) case management; and 4) on-line utilization reports. In the next section of this chapter, each
of these initiatives will be described in more detail. The
final section provides a description of one team's efforts
to utilize these strategies in order to achieve their target
reductions.
Processes for CMG Assignment
Prior to the initiation of the clinical utilization program, all
CMG data were available monthly, on a retrospective basis.
These data are based upon the final chart coding procedures conducted within the Health Records department and
are not available until at least 30 days after discharge. The
goal of assigning CMG upon admission was to provide
clinicians, RU liaisons, and Care Managers with a reference
point for expected length of stay (ELOS). The existing
MSH admission/discharge/transfer system was modified to
incorporate a CMG table and ELOS for all CMG.
Because of the unique nature of each clinical population,
the processes of assigning CMG on admission were different across the three planning councils. For example, in
surgery the CMG is assigned at the time of the booking
request for each case, whereas medical patients have a
CMG assigned as soon as possible after admission. It is
the responsibility of the Care Managers in the medical
units to assure that CMG are assigned to patients, with a
focus on the target CMG. The process of assigning a
CMG automatically links to an ELOS (based upon CIHI
database figures). As a result, care managers, nursing unit
administrators, social workers, and physicians are able to
review the patient list daily and identify those at risk for
exceeding the expected length of stay. Subsequently, action
may be taken to expedite interventions and discharge planning to assure timely disposition of each patient. Although
our initial focus has been to have CMGs assigned for target populations, it is intended that this process will be
eventually applied to as many patients as possible.
At this time, the CMG assigned on admission is not necessarily congruent with that which is ultimately assigned
by Health Records coders. However, we are reviewing the
incidence of discrepancy and attempting to understand
why this may be the case. In many instances, we have
identified the need to improve clinical documentation so
that complications and co-morbid conditions (which
45
change the CMG assignment) are captured. It is premature to report on our findings about these differences at
this time. However, it has been a useful mechanism to
enhance the clinicians' recording of relevant information.
The Patient Care Process
The Patient Care Process at MSH is an innovative
approach to co-ordinating patient care. It aims to create
an environment for quality patient care and cost effectiveness by capturing: 1) the multiple processes and phases of
care, 2) multidisciplinary scope of clinical events, 3) comorbid conditions, and 4) outcomes of care. The goals of
the Patient Care Process are to:
² develop clinical care plans, guidelines and protocols
to support the clinical management of patients and
the concept of “best practice”;
² decrease length of stay for target populations;
² decrease the cost per case for target populations;
² enhance patient care outcomes;
² enhance the discharge planning process;
² support efficient and effective use of clinical
resources;
² provide the capacity for clinical variance analysis; and
² support the values of total quality management.
The patient care process is facilitated by individuals with
expertise in the development of patient care protocols
and care plans. As will be illustrated later in this chapter,
these care plans encompass a broad range of clinical
interventions, but primarily focus on activities other than
those found in care plans (i.e. they are not limited to tests,
treatments, consults, and medication) or the typical Care
Map™ as described elsewhere (Zander, 1988).
Additionally, the tools being developed to address the
patient care process have shifted to a focus on discharge
outcomes. Identified outcomes thus have associated care
activities provided by the multidisciplinary team.
To date, several teams have developed and begun to
implement plans of care which primarily focus on the
CMG targeted for LOS reductions. However, there have
also been several initiatives which focus on protocols of
treatment, phases of care, and co-morbid conditions.
Ideally, these plans of care should be computerized for
ease of review, documentation, and analysis. The ongoing
development and implementation of an order entry system incorporates the care elements identified within the
various protocols and care plans for specific CMG.
Although future plans for clinical computerization include
this functionality, paper documentation is maintained at
present and an interim method has been developed for
data analysis. Actual data from the variance tracking
records collected thus far are currently being analyzed.
46
OPTIMIZING CLINICAL UTILIZATION: STRUCTURE AND STRATEGIES
Case Management
As in other organizations (Cohen, 1991; Ethridge, 1991;
Ethridge and Lamb, 1989), our case management program was designed to support the co-ordination of
patient care and actualization of cost-savings. The work
of population-focused Care Managers occurs within the
context of the council, team, and departmental structures.
Some of these individuals fulfil a dual role as the RU liaison for specific teams. In general, the Care Managers utilize a variety of strategies to support the co-ordination of
patient care. They work in close collaboration with all
members of the multi-disciplinary team and have been
extremely effective in identifying process issues which
impede timely and effective patient care.
Overall, the desired outcomes associated with case management include:
decreased length of stay for target population;
decreased cost per case for target population;
improved patient care outcomes;
seamless co-ordination of services to patients;
reduced duplication and delays in service;
processes which enhance/ensure continuity of care;
improved discharge planning process;
efficient and effective use of clinical resources;
strong linkages to community-based services and
supports; and
² strong interdisciplinary approach to patient care.
²
²
²
²
²
²
²
²
²
At the time of this writing, these individuals have been in
their new roles for almost a year. Evaluation of performance to date, demonstrates that in conjunction with
other utilization efforts, the Care Manager provides a valuable supporting role to the work of the clinical teams.
On-line Utilization Reports
The RU liaisons have access to multiple paper reports to
help them monitor CMG specific resource utilization.
The specific details of these reports are described in
Chapter 5 by Davis. In order to reduce the workload and
costs associated with generating multiple copies of multiple reports, a computerized reporting tool was developed.
The RUM Online Access (ROLA) application was
designed by in-house resource personnel using
PowerBuilder™ database software. On a monthly basis,
data extracts are obtained from the case costing system,
validated by Health Records and Decision Support, and
downloaded to the application. These data can be queried
by the end users according to the categories of council,
team, and physician. Searching for CMG specific performance, data can be retrieved in relation to average costs
and lengths of stay, comparing over time, team to team,
or by physician. Unfortunately, the currency of the data is
always at least 90 days retrospective, but still provides a
visual, tabular, and interactive way for teams to review
their performance.
A CASE STUDY: THE FAMILY
MEDICINE TEAM
The Family Medicine team was formalized as a result of
the new organization structure. Although many of the
team members worked together day to day to direct and
provide patient care, they had not routinely met to discuss
quality of care, utilization or operational issues.
This Clinical Team's assignment was to review and manage the length of stay for CMG 13 (Cerebral Vascular
Disorders except Transient Ischemic Attacks), herein
referred to as the CVA patient population, and CMG 14
(Transient Ischemic Attacks and Precerebral Occlusions).
This case study will specifically discuss the team's
approach to the CVA patient population. This CMG was
targeted because the average LOS at MSH for typical
patients (11.4 to 12.5 days) was greater than the CIHI
expected length of stay (11.0 days). Furthermore, according to reports generated by the Health Records
Department, the LOS for CMG 13 varied considerably by
physician and across teams (see Figure 2), as did the cost
per case (see Figure 3). The target for LOS reduction
involved multiple teams and care providers in several geographical locations. Thus, the strategies used to meet the
target LOS required extensive communication with several people across clinical teams (e.g. Emergency, General
and Specialty Medicine, and Family Medicine).
OPTIMIZING CLINICAL UTILIZATION: STRUCTURE AND STRATEGIES
47
Figure 2:
Comparison of length
of stay for CMG 13
across clinical teams
CMG 013 - SPEC CEREBROVASC DISORD(XTIA)
Oct/1996 - Dec/1996
Typical Cases [CMG 1996 Updtd: 06/09/97]
(unsorted) DISPLAY: 1 to 3 of 3
Total
TotalCase
Case Cost
Cost
20
15
10
5
0
General and
Specialty
Cardiology Team
Family Medicine
Team
Team Description
P Expected LOS (Days)
Avg L Avg
LOS (Days)
Total Case Cost
CMG 013 - SPEC CEREBROVASC DISORD(XTIA)
Oct/1996 - Dec/1996
Typical Cases [CMG 1996 Updtd: 06/09/97]
Series: Total Case Cost DISPLAY: 1 to 3 of 3
Figure 3:
Comparison of total cost
per case for CMG 13
across clinical teams
50000
40000
30000
20000
10000
Team Description
The Family Medicine team requested assistance to evaluate the care of the CVA population within the context of
the Patient Care Process. They believed this patient population could be better managed using a consistent plan of
care but were not clear as to how to proceed with the
development. To better understand the current LOS of
the CVA population, the team was encouraged to discuss
processes in current patient care management. They were
then guided to develop a flow diagram (see Figure 4)
which provided them with a map of the multiple processes involved in clinically managing this patient grouping.
During these discussions, some team members identified
a delay of diagnostic tests and consults which could have
contributed to an increased length of stay. The flow diagram highlighted barriers in the processes. It should be
noted the flow diagram reflected not only patient flow,
but also provider flow and information flow.
48
OPTIMIZING CLINICAL UTILIZATION: STRUCTURE AND STRATEGIES
Figure 4:
Flow Diagram of Clinical Care
Events for CVA Patients
Acute Focal
Neurologic
Symptom
Start
History,
Physcial,Labs
Tests*
Likely TIA
LOWRISK
TRANSIENT DEFICIT
HIGH RISK**
* CBC, PT, PTT, glucose,
renal and hepatic function
test, electrolytes, ECG
PERSISTENT DEFICIT
LikelyCVA
** First TIA,
>70 and HT, DM, CAD or
Cresendo TIAs
Y
Symptoms
<3 Hours
Consider Acute
Thrombolytic
Therapy
LysisProtoco l
(tobedeveloped)
N
Outpatient
Management
CT Scan
CT Scan
Abnormal?
Y
Admit PT
(CVA)
N
TestsBooked
-CT, -ECHO
-Doppler
Follow-UP
-Family MD
-Neurology
Admit PT
(TIA)
Elderly Pt?
Cresendo
TIA?
ISCHEMIC
Y
Consider Acute
Heparin
Anticoagulation
-Intracerebral
hemorrhage
-Abscess, -Tumor
-Subdural
Infarct
Normal CT
N
Antiplatelet Therapy
-ECASA 325mg OD
-Alternate
Antiplatelet Agent
Start Antiplatelet
Agent
Discharge Pt
Home
Implement
Long-Term
Plan
End
The Family Medicine team requested assistance to evaluate the care of the CVA population within the context of
the Patient Care Process. They believed this patient population could be better managed using a consistent plan of
care but were not clear as to how to proceed with the
development. To better understand the current LOS of
the CVA population, the team was encouraged to discuss
processes in current patient care management. They were
then guided to develop a flow diagram (see Figure 4)
which provided them with a map of the multiple processes involved in clinically managing this patient grouping.
During these discussions, some team members identified
a delay of diagnostic tests and consults which could have
contributed to an increased length of stay. The flow diagram highlighted barriers in the processes. It should be
noted the flow diagram reflected not only patient flow,
but also provider flow and information flow.
N
Progressive
Deficit? Cardiac
Source? ***
Y
ConsiderAcute
Heparin
Anticoagulation
(v.s. Antiplatelet Tx)
Appropriate
Management
*** If uncertain,
consult Neurology
Doppler
Echo
End
NON-ISCHEMIC
Cardiac Studies
Echo, +/- Holter,
Telemetry
Doppler
Studies
Interpretation with
appropriate LongTerm management
End
End
The purpose of this diagram is to demonstrate the multiple steps of a process within the context of a more
extensive process. Further analysis was required to identify the specific system issues which were barriers to the
timely delivery of care. The discussions which occurred at
team meetings allowed the team to realize that they did
not have a consistent approach to manage this patient
population, nor did they always know the contribution or
priority setting of other team members and departments.
Therefore the team reviewed each subset of the larger
CVA flow diagram. This task was tedious and labour
intensive, but the team then had a much better understanding of delays in diagnostic tests and consults. The
discovery of the steps involved in the multiple processes
added to the difficulty of reviewing current practice to
manage this patient grouping.
OPTIMIZING CLINICAL UTILIZATION: STRUCTURE AND STRATEGIES
Once the flow diagram was complete, a few members of
the team conducted chart reviews of the CVA population. The members were looking for commonalities in
physician practice, referral patterns, patient functionality,
and delays in obtaining diagnostics. The information
obtained from the chart review assisted the team to establish discharge goals for the development of a patient care
plan. The information obtained from the flow diagram
and the chart review prepared the team to develop a
49
patient care plan. The patient care plan at MSH is a tool
which has established events/activities to occur within an
expected time frame to meet predetermined discharge
outcomes. A new template for the patient care plan has
been developed at MSH which incorporates the Patient
Care Plan and Variance Tracking Record on one form.
This format allows clinicians to visualize the multidisciplinary plan of care and to document the variances on the
same form (see Figure 5).
Figure 5:
Sample Page of Care Plan
for CVA Patient Population
50
OPTIMIZING CLINICAL UTILIZATION: STRUCTURE AND STRATEGIES
As mentioned earlier, information gathered from the
chart review assisted the team to develop the discharge
outcomes. However, it also helped the members of the
team establish which events/activities were to be tracked.
Medical guidelines were developed for implementation in
the Emergency Department, thus the focus of events/
activities were not a rewrite of physicians’ orders. The
team was encouraged to establish the discharge outcomes,
phases of care and milestones to meet the outcomes. It
should be noted the discharge outcomes are patient specific, the phases of care may be hours, days or weeks and
the milestones are events/activities which must occur
prior to or within the phase of care. This team reviewed
the literature, talked to people in other organizations and
developed a care plan to reflecting the LOS target rather
than current practice.
It is interesting to note this population of patients may be
admitted to any one of three teams at MSH. However,
the Family Medicine team was given the responsibility to
develop a methodology to manage these patients within
the targeted length of stay and then to share the results
with other teams. Throughout this process, it was a challenge to know when to include members from other
teams as they also had LOS targets to meet and not all
practitioners could attend every meeting during the development process. This became more significant as physicians on teams began to develop medical directives.
Nonetheless, the physicians and other team members
developed a process to communicate the medical directives and care plan progress to all potential users of the
plan including patients. The Care Manager with this team
was also their RU liaison and actively participated in the
development and communication of the CVA patient
care plan. The CVA care plan was being implemented at
the unit level at the time of this writing.
Using the ROLA application, the Care Manager for this
team was able to review utilization performance for CMG
13 according to doctor number and by team. The capability to graph CMG length of stay and costs over time, by
physician and team, was demonstrated to be extremely
valuable in monitoring utilization performance (see
Figures 2 and 3). At the time of this writing the LOS for
CMG 13 has fallen to 6.6 days (CIHI expected LOS 11.0 days) for the last quarter of 1996. In general, the
change in LOS can be attributed to the combined efforts
of care managers and clinicians, and the application of
consistent plans of care. Improved discharge planning
and focused attention on the specific care needs of this
population has contributed to the success of the team’s
work. Although the decrease has been substantial over the
year, we continue to monitor this CMG as well as many
others to ensure that changes to practice persist.
In the future, the clinical events and outcomes being captured on the care plan will be reviewed in conjunction
with these data. This convergence of clinical, health
records, and costing data will thus provide further understanding of care processes and guide future utilization
efforts. In particular, we have begun to monitor several
clinical indicators to ascertain the "optimalization"
(Petryshen, O’Brien-Pallas, & Shamian, 1995) of resource
use, and to make certain that efforts to achieve increased
efficiency in patient care delivery are not compromising
the provision of quality care.
CONCLUSION
At the organizational level there are a number of lessons
to be learned from our experience.
² To have a successful utilization program one needs to
have a strong partnership between clinicians and
management.
² The clinical utilization initiative at the clinical level
has to be driven by clinicians.
² This work needs designated champions (in our case
the RU Liaisons and Care Managers).
² Experts need to be driving the process (e.g. Health
Records, Clinicians).
² The patient population focused approach is an effective strategy to facilitate context based action plans.
² The overall organizational team (RUMC) needs to
have members from both the clinical side of the
house, and the data—financial, health records, and
information services.
² Strategic initiatives need to be identified and developed at the corporate level.
In summary, a utilization program will become successful
when the organization integrates the basic tenets of such
a program as part of the culture of the organization. At
this era of evidence-based clinical practice and shrinking
resources, a strong active utilization program can help
organizations to optimize their resource utilization.
In retrospect, focusing on a single issue (LOS reductions)
at the outset was very important to the successful introduction to the utilization program. Clinicians and administrators were able to focus on something tangible.
Through 1996/97 we have seen a significant drop in LOS
in all teams. Many of the LOS targets have surpassed and
dropped below the benchmarks. In the future, we intend
to further expand the utilization work to include the
review of specific diagnostics and therapeutics and the
development of team-based scorecards to monitor the
"optimalization" of organizational resources.
OPTIMIZING CLINICAL UTILIZATION: STRUCTURE AND STRATEGIES
51
REFERENCES
Cohen, E. (1991). Nursing case management: Does it
pay? Journal of Nursing Administration 21(4), 20-25.
Cyr, J. J., Secord, P., & Prendergast, P. (1995).
Comprehensive utilization management: The Whitby
Psychiatric Hospital model. Leadership 3(2), 26-30.
Ethridge, P. (1991). A nursing HMO: Corondelet St.
Mary's experience. Nursing Management 22(7), 22-27.
Ethridge, P. & Lamb, G. (1989). Professional care nursing
management improves quality access and costs. Nursing
Management 20(3), 30-35.
Leatt, P., Lemieux-Charles, L., Aird, C., & Leggat, S.G.
(Eds) (1996). Preface, Strategic alliances in healthcare: A casebook in management innovation. Ottawa, ON: Canadian
College of Health Service Executives.
Payne, S.M.C. (1987). Identifying and managing inappropriate hospital utilization: A policy synthesis. Health
Services Research 22(5), 709-769.
Petryshen, P., O'Brien-Pallas, L. L., & Shamian, J. (1995).
Outcomes monitoring: Adjusting for risk factors, severity
of illness, and complexity of care. Journal of the American
Medical Informatics Association 2(4), 243-249.
Sheps, S.B., Anderson, G., & Cardiff, K. (1991).
Utilization management: A literature review for Canadian
health care administrators. Healthcare Management Forum
4(1), 34-39.
Zander, K. (1988). Nursing case management: Strategic
management of cost and quality outcomes. Journal of
Nursing Administration 18(5), 23-30.
About the Authors
Dr. Lynn Nagle RN, PhD is Managing Director, Clinical Informatics and
Utilization, Mount Sinai Hospital, Toronto, Ontario;
Judith Shamien, RN, PhD, CHE is Vice-President Nursing, Mount Sinai
Hospital;
Margaret Catt, RN, MHSc was Patient Care Process Co-ordinator,
Mount Sinai Hospital, and is currently Information Manager,
Temiskaming District Hospital;
Martin Stein, BComm, CMA, CA is Vice-President, Finance, Mount
Sinai Hospital; and
Joseph Mapa, FACHE is Executive Vice-President and Chief Operating
Officer
Acknowledgments
The authors of this chapter would like to acknowledge the contributions
of Andrew Asa, Programmer Analyst in Resource Utilization, Mount
Sinai Hospital, for his contribution around resource utilization.
C H A P T E R
5
CATHY DAVIS
Using Case Mix Tools with
Case Costing Data for
Utilization Management
CHAPTER OVERVIEW
During the past two years, Mount Sinai Hospital has changed its organizational structure
from a departmentally-focused structure to a maxtrix-based structure with multi-disciplinary patient care teams. To support this change, a challenge for Information Services was
to provide teams with information that facilitates decision making. This chapter describes
how information was used to meet this challenge. It discusses the use of information to:
1) assign patients to teams; 2) set utilization targets for length of stay and shifting inpatient
care to an ambulatory setting; 3) identify best practice hospitals; 4) develop standard reporting packages; 5) analyze cost variances; 6) enhance reports with complexity data; and 7) isolate case types for patient care planning. Finally, key results and recommendations for
Information Services at Mount Sinai Hospital are identified.
INTRODUCTION
During the past two years, Mount Sinai Hospital has undergone a shift from the traditional,
department-specific organization structure to a matrix-based structure with multi-disciplinary patient care teams. The shift to a team-based structure corresponds to a shift in the
focus of the Canadian accreditation program. In 1994 the Canadian Council of Health
Services Accreditation (CCHSA) published a document advising that "each client/patient
care team serves a particular group of clients/patients with similar needs and patterns of
resource consumption" (CCHSA, 1994, p. 17). This chapter describes how case mix and
case costing information was used to identify these groups at Mount Sinai Hospital and to
support changes in organization structure as well as provide a basis to focus utilization
management efforts.
THE ORGANIZATIONAL STRUCTURE
Mount Sinai Hospital is a teaching hospital in Toronto with a range of programs including
high risk obstetrics, neonatal intensive care, oncology, gastrointestinal diseases, musculoskeletal, sarcoma, diabetes and psychiatry. Mount Sinai Hospital began participating in the
Ontario Case Cost Project in 1991. During the past six years, the use of the data for budgeting, decision-making and utilization review has continued to increase.
54
USING CASE MIX TOOLS WITH CASE COSTING DATA FOR UTILIZATION MANAGEMENT
With the shift to a team-based organizational structure at
Mount Sinai Hospital, Planning Councils were established
for each of three major patient populations: Perinatal,
Surgical/Peri-operative, and Medical and Academic
Community Health. Each Council is Co-chaired by a clinical chief and a vice president. The Planning Councils
provide the Medical Advisory Committee and the
Executive Management Committee with advice and guidance on clinical policy issues in relation to the associated
patient populations. Within the three Planning Councils, a
total of 20 patient care teams were established. The
Councils are also responsible for working with the teams
to meet the needs of the patient population, promote
clinical and academic activities, ensure appropriate
resource utilization management including costs associated with volume, case mix and modalities of care, and
determine the need for and impact of new physician
recruits and capital equipment.
1. Accreditation and quality management. This includes
development of quality indicators to enable continuous self-assessment to evaluate specific areas of
patient care such as assessment, care and treatment
planning, and discharge follow-up.
2. Day-to-day operations. Day-to-day operational issues
are addressed using team analysis and resolution to
ensure efficient and effective workflow.
3. Resource utilization management. The monitoring of
resource use by patients is assigned to each team, and
suggesting and implementing processes to improve
management of those resources is a team responsibility.
The interaction of teams, councils and departments is
outlined in the organizational chart in Figure 1.
The patient care teams focused on clinically similar
patient populations and are responsible for three key and
equally important processes as they pertain to the team's
specific patient population. These processes are:
Patients/Community
Figure 1:
Mount Sinai—
Organizational Structure
(October 1997)
Board
Strategic Directions & Clinical Policy
Operational
CEO
Management Committee
Research
Advisory
Committee
Executive
Medical
Advisory
Council
Perinatal Planning
Council
Clinical
Departments
PLANNING COUNCILS
MANDATE:
- STRATEGIC AND RESOURCE/BUDGET PLANNING
(VOLUME AND CASE MIX) BASED ON HOSPITAL MISSION
AND STRATEGIC DIRECTIONS.
COMPOSITION:
- TEAM CO-LEADER, RELATED CHIEFS, NURSING
REPRESENTATIVE, RESEARCH REPRESENTATIVE,
PROFESSIONAL (NON-NURSING) REPRESENTATIVE,
VICE-PRESIDENT FINANCE, CO-CHAIRS, CEO/COO (EX
OFFICIO)
ACCOUNTABILITY:
- TO MEDICAL ADVISORY COUNCIL AND CEO/BOARD
PATIENT CARE TEAMS
MANDATE:
- ACCREDITATION & QUALITY MANAGEMENT,
UTILIZATION/RESOURCE MANAGEMENT, OPERATIONAL
ISSUES
COMPOSITION:
- STAKEHOLDER DEPARTMENTAL AND RESOURCE
REPRESENTATIVES INCLUDING A CONSUMER
REPRESENTATIVE
- TEAMS WILL BE LED BY CO-TEAMS MAY VARY (E.G.
OTHER SUB-TEAMS MAY BE FORMED)
ACCOUNTABILITY:
- TO VICE-PRESIDENTS, COO AND CEO IN TURN FOR
OPERATIONAL ISSUES
- TO COUNCILS FOR PLANNING AND CLINICAL POLICY
ISSUES
Maternal &
Newborn
Team
Gynaecology
Team
Clinical
Chiefs
COO
Regional
Perinatal
Team
VicePresidents
Surgical &
Perioperative
Planning Council
Dentistry
Ent/Eye
Team
Minimal Access/
General &
Specialty
Surgery Team
GI
Team
Department
Heads
Oncology Thoracic Musculoskeletal
Team
Team
team
Medical &
Community Health
Planning Council
Headache &
Facial Pain
Team
Diabetes
Team
Emergency General &
Specialty
Medicine
Medicine
Team
Team
Cardiology
Team
Mental
Health
Care
Adult
Critical
Care
Team
USING CASE MIX TOOLS WITH CASE COSTING DATA FOR UTILIZATION MANAGEMENT
This change in organization structure has had a significant impact on hospital processes for decision making,
budgeting and information management including data
collection, reporting and use. Rather than an absolute
shift as a result of the change in structure, these processes have moved, and continue to move, along a continuum
from a department focus to a patient focus. Each of these
processes changes at its own rate and some are further
along the continuum than others. The extreme points on
this continuum are described for some specific processes
in Figure 2. Almost all decision-making, budgeting and
information management processes within the hospital
can be described in terms of their progress along this
continuum.
Continuum
Degree of Focus
Figure 2:
Hospital Processes—
Department verses
Patient Focus
12
10
8
6
4
Hospital Processes
Department Focus
Patient Focus
WORKLOAD
MEASUREMENT
* workload measured retrospectively
by department function
* workload measured and tracked
to individual patient activity and
episodes of care
MONITORING
* retrospective monitoring of
resource use
* concurrent monitoring of
resource use
CHARTING
* charting often retrospective, incomplete,
low level of specificity
* concurrent charting complete with
high level of specificity
BUDGETING
* budgets based on historical departmental
data and departmental projections
* budgets based on projected case mix,
volume, complexity and funding potential
DECISION MAKING
* isolated decision making
* team decision making, multi-disciplinary
* limited impact analysis
* impact on resources and quality of
care accurately determined
* assignment of patients to teams
retrospectively
* assignment of patients to teams
at registration
PATIENT ASSIGNMENT
TO TEAMS
55
56
USING CASE MIX TOOLS WITH CASE COSTING DATA FOR UTILIZATION MANAGEMENT
MOVING TOWARD PATIENT
FOCUSED PROCESSES
To move continually along this continuum as an organization, many changes were necessary. A significant step
away from the department focus was to move accountability for utilization/resource management to patient
care teams. This part of the teams' mandate was undertaken simultaneously and in conjunction with quality
improvement in patient care. The goal was to provide
teams with the tools they needed to move forward with
their mandate. The challenge was to provide information
so that decisions could be made without data overload.
To begin the process of utilization management, teams
needed information on their specific patient population
including volumes, case mix, practice patterns and length
of stay. By combining these CIHI data elements with
patient specific costing, a "tool kit" of information was
provided to teams to focus their efforts and support their
decision-making process.
CLARIFYING GUIDELINES FOR
ASSIGNING PATIENTS TO TEAMS
named on the organizational chart using CIHI data.
Using one month of data, a process was outlined for
establishing criteria that would be used to assign patients
to teams. This step generated a series of questions for
the teams and councils which would help to more accurately describe the patient population.
Each team was led through a series of questions by the
data managers. Sample questions for the Oncology team
are outlined in Figure 3. To assist the teams in deciding
patient groupings, many data elements in the CIHI
abstract were used. Of key importance were not only
Major Clinical Categories (MCC) and Case Mix Groups
(CMG) but more specifically diagnosis codes, diagnosis
types, procedure codes, doctor service and patient service.
The rationale for using data at a more specific level than
CMG for the assignment of cases to patient groupings is
illustrated with the oncology team. Although there are
some CMG which relate specifically to oncology, many
CMG include cases with both oncology and non-oncology cases. Using CMG only would have assigned many
oncology cases to other care teams.
The first step in providing this information involved
establishing criteria for assigning patients to teams. To
accomplish this, data managers in Health Records
Services were asked to sort cases into one of the teams
Sample questions posed to our oncology team to clarify What is an
oncology patient ?
Do we include patients with malignant neoplasms only or should
neoplasms that are benign, in situ or of uncertain behaviour be
included?
Do we include patients with an oncology workup for breast or
sarcoma, regardless of the outcome?
Do we include patients who return to the hospital with
complications of chemotherapy or radiation therapy such as febrile
neutropenia? how do we determine this from the documentation?
Since the oncology team is part of the Surgical Council, is both
medical and surgical oncology included?
How do we distinguish between medical and surgical oncology?
Is thoracic surgery for malignancies included in the oncology team
or the thoracic team?
Figure 3:
Sample Questions
to Clarify Oncology
Patient
57
USING CASE MIX TOOLS WITH CASE COSTING DATA FOR UTILIZATION MANAGEMENT
SETTING UTILIZATION TARGETS
Currently each patient episode is assigned to one team
only, even though several teams may have been involved
with the care of the patient. In the future, with the development of the costing system, specific portions of the
episode of care will be assigned to the appropriate team.
For example, currently there are no patients assigned
specifically to the Adult Critical Care Team because, using
the criteria, they have all been assigned to one of the other
medical or surgical teams. As the information management
process moves along the continuum toward the patient
focus, information will be collected in such a way as to isolate the resources used during the critical care portion of a
visit from those used on the other nursing units.
Goals for Length of Stay
After patients were sorted into teams, the teams began
setting utilization targets. Length of stay review was the
initial focus. Teams needed information that would help
them highlight specific groups of patients for whom
length of stay could be shortened. All teams were provided with data that sorted patients by assigned team, CMG,
and typical and atypical groups (with the CIHI database
length of stay displayed). The Mount Sinai Hospital
length of stay for typical cases was compared to the CIHI
ALOS and patient groups were highlighted where it
appeared that a significant number of patient days could
potentially be saved. The following formula was used to
target those CMG with a high potential for days saved.
Discussions were held with the teams including both
administrative and clinical team members to confirm data
elements used for team assignment. It was necessary to
ensure agreement on the definition of the patient population being assigned to the team. Without this agreement
at the beginning of the process, teams would have resented being held accountable for patients for whom they did
not control resources. Once agreement was reached, team
members took responsibility for addressing areas where
the data highlighted variances.
999
∑ (MSH
ALOS CMGi
− CIHI
ALOS CMGi ) × typical cases
=
potential
days
i =1
As a result, 40 CMG were targeted for possible length of
stay reduction. Teams were provided with computerized
case summaries of each individual patient in the targeted
CMG (Figure 4 shows a sample).
Figure 4: GI Team—Inpatients—Case Summary (Apr. 1 to Apr. 30, 1996)
msh
no.
Age
Res
alc
los
act
los
CIHI
LOS
Plx
LOS
md
#
1234
29
scrar
2
11
16.3
10.9
xx
5678
68
etob
0
12
16.3
13.5
xx
Diag
type
Diag
code
Diagnosis
Pr
md
xx
MO
5556
univ ulcer colitiis, chro
AD
5569
ulc colitis, unspec
PR
v552
att’n to ileostomy
Proc
code
Procedure
Disc
unit
MOH MOH
Q
RIW
Plx
RIW
Plx
CMG
Plx
Lev
4622
continent illeostomy
14S
2
3.7425
3.0532
251
1
485
abd-perineal rect
16N
2
3.7425
3.1562
251
1
resect’n
MO
7872
dysphagia
PR
438
late eff cerebrovasc dis
xx
PR
7843
aphasia
PR
9993
inf’n compl’n care nec
SE
6823
cellulitis of arm
SE
4111
staph aureus inf’n
E
e8798
abn rxn-procedure nec
AD
436
CVA
4632
percutan jejunostomy
Key to Column Titles
msh no.
hospital chart number
Diagnosis
diagnosis description
Age
patient’s age
Pr md
procedural doctor
Res
patient’s residence description
Proc Code
procedure code (ICD-9-CM)
alc los
days awaiting placement (alternative level of care)
Procedure
procedure description
Act los
acute hospital length of stay
Disc unit
nursing unit patient discharged from
CIHI ALOS
CIHI database length of stay for CMG
MOH Q
Ontario Ministry of Health LOS quartile
Plx LOS
complexity expected length of stay
MOH RIW
Ontario Ministry of Health resource weight
md #
most responsible doctor identification number
Plx RIW
complexity resource intensity weight
Diag type
diagnosis type (defined by CIHI)
Plx CMG
complexity CMG
Diag code
diagnosis code (ICD-9-CM)
Plx
complexity level
Lev
saved
58
USING CASE MIX TOOLS WITH CASE COSTING DATA FOR UTILIZATION MANAGEMENT
Standard pieces of information in the summary included
the most responsible diagnosis, complications and comorbid conditions, procedures, date of surgery, physicians involved in the case, age of patient, acute LOS,
CIHI ALOS, Alternate Level of Care (ALC) days and residence code. Additional pieces of information were added
on request for specific teams. Using this summary, team
members could quickly highlight factors that seemed to
account for the longer length of stay. The information
helped to focus the team’s attention on specific issues to
address, rather than wading through stacks of charts
looking for patterns.
Examples of changes implemented as a result of team
review of this basic information include:
² improvement of the process for admission of elective
surgical patients on the day of surgery;
² reduction of ALC days by changes in the discharge
planning process;
² identification of 'flags' for specific patients to identify
potential problems in advance; and
² heightened awareness by all team members of their
impact on utilization management and cooperative
effort toward the utilization goals.
Shifting from Inpatient to Ambulatory Care
In conjunction with the goals for reducing length of stay,
many teams began to ask for data that would help them
determine whether patients who had an inpatient stay
should have been treated in an ambulatory setting.
Reports were prepared that highlighted any case in a
CMG flagged as May Not Require Hospitalization
(MNRH) and any elective admission that stayed less than
48 hours. Case summaries were prepared for these as
well. This process resulted in improvements in the registration process, better utilization of the pre-admission
unit, and a reduction of MNRH cases from 9% of discharges to less than 4%. Hospital staff in the areas of
patient registration, booking and pre-admission were
educated on possible MNRH cases and a process was put
in place to follow-up with physicians who booked possible
MNRH cases as inpatients. Doctors were asked to identify
the clinical condition of the patient that made an overnight
stay more appropriate than ambulatory treatment.
Physicians were encouraged to document these conditions
on the booking form as well as the hospital chart. This documentation would eliminate delays in booking their patients.
This also highlighted the important link between complete
documentation and data quality.
IDENTIFYING BEST PRACTICE
HOSPITALS
As the teams moved along the continuum, they began to
focus on establishing benchmarks and best practice guidelines for their patient group. Using the Comparative
Hospital Activity Program (CHAP) reports, names of
hospitals with the best results for length of stay and
ambulatory surgery by procedure were highlighted and
communicated to teams. Some teams chose to contact
their counterparts at these hospitals to investigate their
processes to see if these could be implemented at Mount
Sinai Hospital.
DEVELOPING A STANDARD REPORT
PACKAGE
It became clear during the process that much of the same
information was of interest to all teams. The common
information required by each team is summarized in
Figure 5. The identification of these common data elements resulted in development of a set of standard
reports which was made available to teams and councils
on a routine basis. These reports provided a snapshot of
how well the team had performed in the most recent time
period. The Standard Report Package is listed in Figure 6.
In keeping with the quality improvement philosophy at
Mount Sinai Hospital, these reports and the utilization
targets are reviewed and refined on an ongoing basis as
processes are investigated.
USING CASE MIX TOOLS WITH CASE COSTING DATA FOR UTILIZATION MANAGEMENT
Figure 5: Common Information Required by Teams
Common Information
Required by Teams
Data Elements / Calculations
Discussion
• volumes of cases by
CMG
typical
cases
total # by CMG; average LOS for
MSH; CIHI ALOS; total weighted cases
• typical / atypical
split
atypical
cases
total # by CMG
• length of stay
performance
total
cases
total # by CMG; average LOS for MSH
total weighted cases
• % admissions on
day of surgery
total elective inpatient admissions with admission
date equal to surgery date divided by the
total elective inpatient admissions with surgery
High percentages of same day admission and
ambulatory surgery may be indicators of efficient
performance and patient-focused care.
• % ambulatory
surgery
total day surgery cases divided by the
total elective inpatient admissions with surgery
plus total day surgery visits (including cystoscopy
and endoscopy)
Individual physician practice often changes with
regular feedback and reporting on these
indicators for an accepted peer group.
• % pre-admission
work-up
total patients with a pre-admission work-up
divided by the total elective inpatient admissions
with surgery plus total day surgery visits
The percentage of elective patients who have a
preadmission work-up. This information is
collected in a number of categories including
work-up outside of Mount Sinai Hospital,
diagnostic work-up only and diagnostic work-up
with a visit to the preadmission unit at MSH.
Effective use of the preadmission unit can reduce
inpatient costs and length of stay.
• average case weight
(or case mix index)
• weighted case load
• average cost per
case
• average cost per
weighted case
x case weight =
• direct vs. indirect
costs
¦ ocw y ¦ cases
¦ocw
weighted caseload =
x cost per case = ¦ cost y ¦cases
x cost per weighted case =
¦ cost y ¦ocw
Actual direct cost; actual indirect cost; actual
total cost
Direct costs are patient specific costs allocated
from direct patient care departments such as
laboratory, diagnostic imaging, nursing &
pharmacy.
These three pieces of information require analysis
and presentation together since they impact each
other. The report is presented as a matrix with
CMGs listed down the left side. Volume and LOS
information is then presented for typical, atypical
and total groups across the top. This provides
the team with a snapshot of their performance in
these categories for the time period. When
presented for consecutive quarters, trends in
volumes, case mix, length of stay and practice
patterns become useful indicators.
As with volume, LOS and case mix review,
weighted caseload and cost also require analysis
and presentation together since they impact each
other. Here again reports are presented as a
matrix with CMGs listed down the left side.
Average case weights, total weighted caseloads
and average cost per case are presented for
typical, atypical and total groups.
This information allows teams to focus on direct
costs, since teams can have the most impact on
these costs specifically.
Indirect costs are allocated across the entire
patient population and include costs of support
departments such as administration, finance,
human resources and health records.
• variance from
funding potential
based on OCW
Actual cost minus potential funding
Funding potential = ocw x hospital-specific cost
per weighted case assigned by Ministry of Health
The Ministry of Health in Ontario has calculated a
hospital specific funding amount for one
weighted case. This is based on a variety of
variables including the level of care, teaching
activity and age of patients. By determining the
variance between the actual cost of a case and
the potential funding, teams can investigate
possible utilization questions.
59
60
USING CASE MIX TOOLS WITH CASE COSTING DATA FOR UTILIZATION MANAGEMENT
To maximize the use of information, each team identified
one individual as a utilization liaison who would be
responsible for initial data review and presentation of
information to the team. To enhance data analysis and
use, reports were made available to team utilization
liaisons on-line, so that information on specific patient
groups within the team could be sorted, graphed and
reported in a format customized to the team.
Figure 6: Resource Utilization Management—Standard Report Package
REPORT
#
REPORT DESCRIPTION
FREQUENCY
DISTRIBUTION
COUNCIL
CO-CHAIRS
TEAM
COLEADERS
UTILIZATION
LIAISON
1
Analysis of % Weighted Caseload by Council & Team
Acute Inpatients
Semi-annual
x
x
x
2
Analysis of % Weighted Caseload by Council & Team
Day Surgery
Semi-annual
x
x
x
3
Planning Council Cost Variance Summary by Team by CMG
Semi-annual
x
-
-
4
Clinical Team Cost Variance Summary by CMG
Semi-annual
-
x
x
5
Clinical Team Cost Variance Summary by DPG
Semi-annual
-
x
x
6
Analysis of Top 25 CMG by Team
Semi-annual
-
x
x
Acute Inpatient
Acute Inpatient
Day Surgery
Inpatients
7
May Not Require Hospitalization CMG
Semi-annual
x
x
x
8
% of Same Day Admit and Pre-admit Work-up by Team
Semi-annual
-
x
x
9
LOS and Critical Care Unit Days for Typical/Atypical Cases by Team
Quarterly
-
x
x
10
Clinical Team Cost Variance (Top 25 CMG )
For Planning Council
Semi-annual
x
-
-
11
Clinical Team Cost Variance (Top 25 CMG )
For Clinical Team
Semi-annual
-
x
x
12
Clinical Team Feeder Department Costs (Top 25 CMG )
Semi-annual
-
x
x
13
Cost Variance By Physician with Typical/ Atypical Split (Top 25 CMG )
Monthly
-
-
x
14
Feeder Department Costs by Physician with Typical/Atypical Split (Top 25
CMG )
Monthly
-
-
x
15
LOS and Volume by CMG
Monthly
-
-
x
16
LOS Variance by CMG (All CMG ) by Team
Quarterly
-
x
x
17
Cost Variance by Physician (Top 25 CMG )
Monthly
-
-
x
18
Feeder Department Costs by Physician (Top 25 CMG )
Monthly
-
-
x
ANALYSIS OF COST VARIANCE PER
WEIGHTED CASE
Mount Sinai Hospital is fortunate to be part of the
Ontario Case Cost Project (OCCP). This provides the
opportunity to analyze information from both the Health
Records database using CIHI methodologies and tools, as
well as information from the Case Costing database. At
Mount Sinai Hospital, the CIHI data collected in Health
Records, as well as the costing data collected in Decision
Support, are accumulated at the visit specific level by patient.
All data submitted in the abstract to CIHI are transferred
to the costing system to be integrated with cost data. The
data elements are merged to produce an integrated financial/clinical record for every inpatient, day surgery and
emergency visit. In this way, detailed resource utilization
management information can be provided at the episode
specific patient level and aggregated in a variety of ways.
Costing data linked with case-specific diagnostic and procedural information is a powerful tool for utilization management within health care organizations. Cost variances
were calculated for each case by using the case weight to
determine the potential funding and then comparing it to
actual cost. Variances were summarized at the team and
Planning Council levels. From these variance reports,
cases were sorted into two categories:
² Positive variance cases’ if the potential funding
exceeded the actual total cost of treating the case at
Mount Sinai Hospital; and
USING CASE MIX TOOLS WITH CASE COSTING DATA FOR UTILIZATION MANAGEMENT
² Negative variance cases’ if the costs were in excess of
the potential funding for the case.
This report allows the team to concentrate on those areas
of clinical practice which would have the biggest impact
on the bottom line.
These variance reports helped teams identify which practices should be benchmarked at Mount Sinai Hospital "the positive variances." The "negative variances" were
investigated using CIHI information together with the
costing information to determine if cases within that particular CMG fell into clusters which had not been recognized by the CIHI methodology and that seemed to make
a difference in the cost. As a result of this analysis, several areas were identified which highlighted the need for
review of the CMG and RIW methodology. One example
of this review within a CMG identified a significant difference in the cost of joint replacement surgery between
oncology and non-oncology cases. This resulted in a
review by CIHI of all cases in the Canadian database in
this group, and an additional CMG for Joint Replacement
for Malignancy was added in CMG 1995.
USE OF COMPLEXITY DATA
In order to maximize the value of the complexity overlay
to the CMG, this information was incorporated into
reporting to the teams. To do this, a screen was added to
the hospital's abstracting software (Med2020) with eight
basic data elements on complexity for each case. These
data elements are:
²
²
²
²
²
²
²
²
Plx CMG;
Plx level;
Plx expected length of stay (ELOS);
database expected length of stay (ELOS) for
average age;
database average age;
Plx RIW category;
procedure for Plx CMG; and
Plx RIW value.
These pieces of information have been used to produce
reports that illustrate the variance of LOS and cost within
a CMG based on the level of complexity. This is then
compared to the same data without the complexity overlay to illustrate the wide range of differences within one
CMG. This varies greatly by team and requires individual
team attention.
Figure 7 shows that for 60 cases from the Gastrointestinal team, the ELOS changes greatly after complexity is applied. This information allowed the team to focus
utilization efforts for reducing LOS on more clinically
similar groups of patients.
Figure 7: Cases—Without Complexity
Current CMG
251 Gastrostomy &
Colostomy
Mount Sinai Hospital Cases
61
Total Cases
MSH ALOS
Database LOS
60
29.5
16.3
With Complexity
Complexity CMG & Level
Total Cases
MSH ALOS
Plx Exp LOS
CMG 251 - Level 1
27
16.5
11.8
CMG 251 - Level 2
10
12.2
15.0
CMG 251 - Level 3
5
52.4
17.7
CMG 251 - Level 4
11
77.9
35.6
Subtotal
53
31.8
17.9
CMG 253 - Level 1
6
12.0
8.3
CMG 253 - Level 2
1
8.0
11.0
Subtotal
7
11.4
8.7
Total
60
29.5
16
62
USING CASE MIX TOOLS WITH CASE COSTING DATA FOR UTILIZATION MANAGEMENT
Teams were also provided with data which outlined the
distribution of cases among the different levels of complexity for the team (see Figure 8). Teams were advised to
determine, from their clinical knowledge of the patients,
if the proportion of patients in each level of complexity appeared appropriate. In many cases the teams
were surprised by the high proportion of patients in complexity level 1. They used this information to do a spot
check of charts to confirm accurate and complete documentation. In this way, they were able to identify specific
areas of concern for incomplete documentation and/or
recommended changes in coding practices.
Figure 8: Council/Team Complexity Level Analysis (Apr. 1 to Sept. 30, 1996)
Team
Total
Cases
Level One
%
Team
Cases
Regional Perinatal
Maternal / Newborn
Gynaecology
Perinatal Council
1573
3812
282
5667
3
0
189
192
GI
Gen & Spec Surgery
Musculoskeletal
Thoracic Surgery
Minimal Access
Oncology
Dental / Eye / ENT
Surgical Council
725
148
433
130
88
1304
257
3085
Gen & Spec Med
Cardiology
Diabetes
Craniofacial Pain
Family Medicine
Palliative Care
Mental Health
Medical Council
Unassigned
Hospital Total
Level
Level
Level
Level
Level
Level
Level Two
%
Cases
Team
%
Hosp
<1
0
67
3
<1
0
6
6
551
75
303
83
78
832
93
2015
76
51
70
64
89
64
36
65
544
700
31
36
126
2
179
1618
344
580
23
4
79
1
2
1033
17
10387
11
2
37
50
GI
Gen & Spec Surgery
Musculoskeletal
Thoracic Surgery
Minimal Access
Oncology
Dental / Eye / ENT
Surgical Council
Gen & Spec Med
Cardiology
Diabetes
Craniofacial Pain
Family Medicine
Palliative Care
Mental Health
Medical Council
Unassigned
Hospital Total
Level Three
%
%
Cases
Team
Hosp
%
Hosp
0
0
5
5
1
0
1
2
17
2
9
3
2
26
3
62
69
9
43
9
5
119
3
257
10
6
10
7
6
9
1
8
16
2
10
2
1
27
<1
58
41
7
12
7
3
55
5
130
6
5
3
5
3
4
2
4
63
83
74
11
63
50
1
64
11
18
<1
<1
2
<1
<1
32
78
64
5
0
16
0
0
163
14
9
16
0
13
0
0
10
18
14
1
0
4
0
0
37
39
27
0
0
8
1
0
75
7
4
0
0
6
50
0
5
4
24
<1
0
0
0
2
12
3244
31
442
4
209
2
%
Hosp
31% of cases in
the hospital are
at level 1
complexity
Level Nine
%
Team
Cases
<1
<1
13
1
1
<1
4
6
1558
3810
31
5399
36
53
63
23
2
224
155
556
5
36
15
18
2
17
60
18
4
6
7
3
<1
25
18
63
44
11
1
32
10
0
177
275
8
2
3
89
8
0
99
17
2
12
883
level 1 are in the GI team
0
0
8
0
Level Eight
%
Cases
Team
Regional Perinatal
Maternal / Newborn
Gynaecology
Perinatal Council
17% of the 3244 MSH cases in
team are in level 1
0
0
22
22
One - no complexity
Two - related to chronic conditions
Three - related to serious / important conditions
Four - related to potentially life-threatening conditions
Eight - age adjustment applied only
Nine - no complexity or age adjustment (exclusion)
Team
76% of cases in the GI
%
Hosp
Total
Cases
99
100
11
95
29
71
<1
100
1573
3812
282
5667
0
1
0
0
0
2
0
3
0
1
0
0
0
<1
0
<1
0
<1
0
0
0
<1
0
<1
725
148
433
130
88
1304
257
3085
5
1
<1
4
1
0
20
31
0
2
0
0
0
0
0
2
0
<1
0
0
0
0
0
<1
0
<1
0
0
0
0
0
<1
544
700
31
36
126
2
179
1618
<1
0
0
5404
52
9
Level One - no complexity
Level Two - related to chronic conditions
Level Three - related to serious / important conditions
Level Four - related to potentially life-threatening conditions
Level Eight - age adjustment applied only
Level Nine - no complexity or age adjustment (exclusion)
0
17
10387
9% of hospital cases have been
adjusted for age only (level 8) and
the majority of these are in Oncology,
M ental Health & Dental / Eye / ENT teams
<1
0
<1
0
Level Four
%
Cases
Team
<1
0
<1
<1
%
Hosp
0
0
2
2
0
0
<1
<1
0
0
<1
<1
28
3
12
8
0
72
1
124
4
2
3
6
0
6
<1
4
14
1
6
4
0
35
<1
60
19
13
0
0
4
<1
0
36
39
16
2
0
13
0
0
70
7
2
6
0
10
0
0
4
19
8
<1
0
6
0
0
34
<1
9
53
205
2
20
3
6
3
1
26
2
62
Only 4% of hospital
cases are at complexity
levels 3 & 4
4
USING CASE MIX TOOLS WITH CASE COSTIN FOR UTILIZATION MANAGEMENT
The importance of the complexity data will continue to
increase as health care organizations are compared at
local, regional, provincial and national levels. One of the
difficulties with setting benchmarks is the ability to compare like cases. Even with national guidelines for coding
and a standard CMG methodology, there are still hospital
decisions which have a significant impact on coding practices including the typing of diagnoses which plays a significant role in all of the data. The current CIHI guidelines for assigning diagnosis types state that a "pre-admit
co-morbidity" is one which "usually has a significant
influence on the patient's length of stay and/or significantly influences the management/treatment of the
patient while in hospital." (CIHI, 1994, pp. 31–32). But
how does the individual hospital interpret 'significant
influence'? Health records coders require team specific
guidelines for assigning diagnosis types. This point is illus-
63
trated by an example from the Gastrointestinal care team
at Mount Sinai. A review of GI cases showed that specific diagnoses were being missed consistently or typed as a
secondary diagnosis. Secondary diagnoses are not considered in the complexity methodology because by definition
they "did not significantly contribute to the patient's
length of stay in the hospital." Many patients in the
Inflammatory Bowel Disease program suffer from malnutrition. Since this is such a common co-morbidity for
these patients, it was not consistently being recorded on
the patient's chart. Sometimes it was captured by the
coder as a secondary diagnosis since the coder could not
always determine from the chart if the malnutrition significantly affected length of stay or treatment. In discussion with the GI team, guidelines were developed for
Mount Sinai Hospital coders (see Table 1).
Table 1: Coding Guidelines for Malnutrition
Diagnosis
Malnutrition
Diagnosis Type
1 if before surgery
Guidelines
•
always code if patient is on TPN
or TEN at any time during their
course of treatment. Use both
diagnosis and procedure codes.
•
if malnutrition is noted as
severe , use codes 261 or 262.
•
always code if recent weight
loss is >10%.
•
for albumin values <28 g/L
review chart for other indicators
of malnutrition or contact
physician.
2 if post surgery
This variation by hospital is also true for the costing data.
Implementation of the MIS guidelines will not eliminate
all hospital-based decisions on allocation of costs such as
'what is included in department costs?' and 'what is
included in indirect cost?'. In determining what to include
in department costs, consider two hospitals, each with a
Diagnostic Imaging Department. One hospital may have
an administration cost centre for the department and one
hospital may not. For the hospital that does not have the
administration cost centre all the administrative costs will
be allocated to patients as a direct cost. For the hospital
with the administration cost centre, administrative and
support costs can be allocated as indirect and removed
from direct costs, thereby making the information less
comparable.
Standardizing the assignment of direct and indirect costs
is also difficult. Consider a hospital that has introduced a
multi-skilled worker such as a 'Service Assistant'. This
position combines duties previously provided by the
departments of housekeeping, portering and food services. This new position now reports to the Department
of Nursing. For the hospital with the service assistant, the
cost for the position is included in direct costs because
the department of nursing is a direct department. For
hospitals without this position and operating with the
more traditional department divisions, the costs will all be
indirect because housekeeping, portering and food service
departments are indirect departments.
ISOLATING CASE TYPES FOR DEVELOPMENT OF PATIENT CARE PLANS
CIHI data have been useful in the process of development of patient care plans. Initially the data were used to
identify high volume, high cost cases within a CMG with
a wide variation in practice patterns among physicians (for
example LOS). After a CMG was identified, complexity
64
USING CASE MIX TOOLS WITH CASE COSTING DATA FOR UTILIZATION MANAGEMENT
data were used to define a specific case type within the
CMG, for inclusion on the care plan. This data analysis
also helped to identify specific co-morbid conditions
which may be indicators to exclude the case from the
plan or to remove the patient from the plan if one of
these conditions develops in hospital. This initial review
of the cases within a CMG highlighted variances in complexity and allowed the team to refine the definition of
the cases to be considered in development of the patient
care plan.
SUMMARY OF RESULTS
The use of case mix data to support utilization management activities has been reviewed in the preceding discussion. To emphasize the usefulness of the data, key results
for Mount Sinai Hospital are highlighted here:
1. documentation of specific criteria, developed by the
teams to assign all inpatients and day surgery cases to
16 patient care teams;
2. development of a standard report package for utilization management and review for each patient
care team;
3. development of an on-line tool which facilitates
information review by utilization liaisons for
each team;
4. enhanced understanding of the multiple clinical definitions possible for any "program" and an improved
ability to assess the feasibility of integration of clinical programs between facilities as a result of in depth
knowledge of the mix of patients;
5. facilitated organizational move along the continuum
toward preparing clinical budgets for patient care
teams and councils;
6. increased organizational motivation to change from
episode-specific information to date-specific information, i.e., to adopt a system that not only shows that
the patient's stay cost $5,000 but also the distribution
of this cost on a day-to-day basis within the stay to
highlight resource use for specific areas such as critical care and operating room;
7. improved data quality due to heightened awareness of
the importance of timeliness, completeness, accuracy,
specificity and relevance;
8. improved coding practices with clinical input and
education of coding staff on disease processes within
specific patient populations;
9. a heightened awareness and understanding by all
team members, of the interaction among patient
care, costs, charting, funding and budgeting has
improved resource utilization at all levels of the
organization; and
10. significant improvement in determining the impact of
decisions for all departments and at a hospital-wide
level. Before case mix and costing information was
available at a patient specific level, it was more diffi-
cult to do an impact analysis. For example, decisions
about the care and treatment of patients with acute
myocardial infarction were appropriately made by the
cardiologists and nursing unit where most of these
patients were treated. This has not changed, but the
additional information has highlighted the significant
impact on a variety of other areas including family
practice, social work, emergency department, and
pharmacy.
RECOMMENDATIONS
As information to support hospital initiatives is improved,
some basic guidelines and goals for the Mount Sinai
Hospital information service have been identified. These are:
1. Include data collection for as many episodes of care
as possible, e.g. inpatient acute care, day surgery,
emergency, pre-admission visits, clinic visits, etc.
Current mandatory data collection in Ontario
includes only inpatient and day surgery episodes. To
address the full spectrum of health service and related utilization, data must be collected on all visits.
2. Develop criteria to easily link related episodes of care
in the data base and to ensure the ability to report
either separately or together. For example, a preadmission unit visit should link with the related day
surgery visit; the emergency visit should link with the
related inpatient admission; the mother's visit information should link with that of the newborn.
3. Conduct regular data quality checks between systems
and databases. For example, request the same data
from two different departments (health records and
decision support) and compare the results. Different
results may not mean an error, but rather a difference
in the interpretation of the request. Data quality
checks help both the data user and the information
provider. They assist users to improve the quality of
the request and to understand the complexity of the
data and they develop the skill of the information
provider in interpreting the request to ensure the
users' needs are met.
4. Develop careful processes and time frames for making changes, updates and corrections. Changes to the
criteria for assigning patients to teams need to be
documented and authorized by all teams affected.
Removing patients from a team also removes
resources and costs just as adding patients to a team
will add resources and costs. Both teams should have
this information available to make an informed decision about whether or not to add or delete patients.
Changing criteria mid-year may be suggested as teams
become more familiar with the implications of the
criteria they have developed for patient assignment.
USING CASE MIX TOOLS WITH CASE COSTING DATA FOR UTILIZATION MANAGEMENT
However, changing it is not recommended because
this complicates the ability to compare between time
periods and to identify trends.
5. Ensure that data being compared are run on the same
CMG year. Mount Sinai Hospital has a policy to
maintain the last full fiscal year as well as the current
year using the most recent CMG methodology.
Documenting the CMG year on all reports is also
encouraged as previous reports are often compared
with current reports and can create confusion for the
user if this is not clear.
6. Inform teams regularly on significant changes to
CMG and RIW methodology and coding practices.
7. Provide users with a 'Glossary of Terms' which
explains all available data elements which appear
on reports.
8. Minimize the time delay between the patient’s stay
in the hospital and the availability of data related to
that stay.
9. Educate! Educate! Educate! There are few things
that can be stated without question when it comes to
information management, but one thing is certain,
there can never be too many education sessions for
users. The more information sessions are provided
for users, the more the data will be used and the better the data quality. Mount Sinai Hospital has developed a series of education modules that cover many
different topics at a variety of levels, including coding
practice and diagnosis typing, the impact of documentation, CMG and RIW methodology, Case
Costing methodology, report interpretation and
analysis, and funding methodologies.
The Mount Sinai Hospital experience in using the CIHI
data for decision making has been successful. The hospital continues to improve information processes due to
increased and varying demands as it moves away from the
department focus. Users are shifting their focus from the
vast amount of information that is not yet available
towards using and improving the vast amount of information that is available.
65
66
USING CASE MIX TOOLS WITH CASE COSTING DATA FOR UTLIZATION MANAGEMENT
REFERENCES
Canadian Council on Health Services Accreditation
(CCHSA), 1995. Standards for Acute Care Organization: A
Client-centred Approach 6.1.1. Ottawa, Ontario: Author.
Canadian Institute for Health Information (CIHI), 1995.
CIHI Abstracting Manual Ottawa, Ontario: Author.
About the Author
Cathy Davis, CCHRA (C), MEd is Director, Health Record Services,
Mount Sinai Hospital, Toronto, Ontario.
Acknowledgments
The author would like to thank to Michael Stewart, former Director of
Utilization Review & Decision Support, Mount Sinai Hospital, Toronto,
Ontario and Connie Lambert, Health Record Analyst, Mount Sinai
Hospital for their assistance with this chapter.
C H A P T E R
6
DEBORAH NOWICKI, DIANE FRENCH, KATHERINE CHOPTAIN
Utilization Management at
St. Boniface General Hospital,
Winnipeg, Manitoba
CHAPTER OVERVIEW
St. Boniface General Hospital (SBGH) is a 603-bed teaching hospital and is a nationally
recognized centre of excellence in patient care, education and research, under the ownership and guidance of the Grey Nuns. Located in Winnipeg, Manitoba, the hospital offers a
comprehensive range of outpatient, outreach and inpatient services at all levels, including
primary, secondary and tertiary care.
In the fall of 1991, SBGH developed the Utilization Management (UM) program, a central
component of the Total Quality Management initiative within the hospital. The formation
of this program was timely, given the provincial government's health reform protocol
announced in May 1992 which focused on strategies to assure the future of Manitoba's
health services system within decreased funding from the Federal Government. The mandate of the UM program is to provide administrative and clinical staff with access to information to aid in evaluating the appropriateness, effectiveness, and efficiency of health care
delivery at SBGH.
To achieve the objectives of the program, information from various hospital databases is
merged, extracted and presented to clinical team leaders at routine meetings. Length of stay
(LOS) data, a primary indicator of hospital utilization, are regularly reviewed. For this purpose, data from Canadian Institute for Health Information (CIHI) reports are provided by
CMG and doctor service. LOS data by CMG are particularly useful since team leaders are
able to compare their performance with both the CIHI database and select peer groups.
When LOS appears higher at SBGH, questions are raised and explanations regarding practice
patterns and patient characteristics are sought. Evaluation of this type ensures the continuous
monitoring of health care delivery within the hospital. Further, CIHI's introduction of the
complexity overlay and age-adjustment enhancement will provide more precise information
by which to define the patient population within SBGH and peer group facilities.
In addition to inpatient reports, DPG data are regularly presented to surgical department
heads. The inpatient/outpatient comparison by doctor service identifies short-stay inpatients and illustrates the potential transfer of these patients to day surgery procedures.
Continued effort to reduce LOS in Surgery is also measured by examining the percent of
elective patients admitted on the same day of their surgical procedure.
With the availability of CIHI data, the UM program at SBGH is able to provide precise and
timely information to clinicians, the service providers ultimately responsible for the efficient
delivery of health care services to patients.
68
UTILIZATION MANAGEMENT AT ST. BONIFACE GENERAL HOSPITAL, WINNIPEG, MANITOBA
INTRODUCTION
In March 1990, St. Boniface General Hospital undertook
a voluntary and ongoing partnership with the Canadian
Institute for Health Information (CIHI) as part of its
endeavor to establish a comprehensive health care information system within the hospital. This partnership is in
addition to fulfilling the mandatory requirement for submission of hospital information to the provincial Ministry
of Health (i.e. Manitoba Health). Shortly afterwards, St.
Boniface introduced its Utilization Management Program.
The objectives of this chapter are: 1) to describe the
Utilization Management Program and 2) to describe for
client hospitals, the role of CIHI within this program.
For purposes of illustration, length of stay data are presented for the Departments of Obstetrics and Surgery.
Strengths and limitations of CIHI data are discussed.
UTILIZATION MANAGEMENT
PROGRAM
St. Boniface General Hospital (SBGH) is a 603-bed teaching hospital affiliated with the University of Manitoba,
and is a nationally recognized centre of excellence in
patient care, education and research, under the ownership
and guidance of the Grey Nuns of Manitoba (St.
Boniface General Hospital, 1995). Located in Winnipeg,
the hospital serves the City of Winnipeg, Province of
Manitoba and the Northwestern Region of Ontario. A
comprehensive range of outreach, outpatient and inpatient services at all levels, including primary, secondary
and tertiary care is provided.
During the fall of 1991, the Utilization Management
Program was developed, a central component of the total
quality management initiative within the hospital. The formation of this program was timely, given the provincial
government's health reform protocol, announced in May
1992, which focused on strategies to assure the future of
Manitoba's health services system with decreased funding
from the federal government (Manitoba Health, 1992).
The mandate of the Utilization Management Program is
to provide clinical and administrative staff with access to
information to aid in evaluating the appropriateness,
effectiveness, and efficiency of health care delivery. To
achieve these objectives, SBGH maintains a health care
information system comprising several databases (see
Figure 1). Using medical record statistics (MRS) which
comprise all inpatient and day-care encounters as the primary source of information, CIHI (inpatient and day-care
encounters), MediQual (inpatient encounters) and PatientBased Accounting and Budgeting (inpatient and day-care
encounters) databases have been established. Each of
these databases may be utilized individually or linked by
common data elements to provide accurate and comprehensive health care information for patients at SBGH.
Medical Record
Manitoba
Health
Health &
Welfare
Canada
Figure 1:
Flow & Exchange of Health
Information Between Data Bases
Mediqual
Database ASG,
DRG, Morbidity
CIHI
MRS Demographics &
ICD-9-CM Codes
Paper
Reports
Paper
Reports
MCHPE
Key:
CIHI
MCHPE
MRS
PBAB
CMG
MCC
RIW
ASG
DRG
DI
- Canadian Institute for Health
Information
- Man Centre for Health Care
Policy & Evaluation
- Medical Record Statistics
- Patient Based Accounting and
Budgeting
- Case Mix Groups
- Major Clinical Category
- Resource Intensity Weight
- Admission Severity Group
- Diagnostic Related Group
- Diagnostic Imaging
MCC
CMG™
RIW™
Linking
of MRS,
MRS,CIHI,
CIHI,
Linking of
PBAB,
FBAR, MCHPE &
& Mediqual
Mediqual
UTILIZATION
MANAGEMENT
Resource Utlization Information
PBAB
Diagnostic, Therapeutic, Activities
Costing Information
UTILIZATION MANAGEMENT AT ST. BONIFACE GENERAL HOSPITAL, WINNIPEG, MANITOBA
On a monthly basis, SBGH submits one hundred percent
of its monthly discharges to CIHI. Case mix groups
(CMG) and Resource Intensity Weights (RIW) are
assigned to each inpatient record. Day-care records are
grouped into Day Procedure Groups (DPG) and DPG
weights are assigned accordingly. CIHI returns CMGgrouped data on diskette. In addition, reports describing
inpatient and day-care utilization are distributed to the
hospital. Similarly, reports presenting hospital comparisons are included for client peer-group hospitals.
MedisGroups, a Severity of Illness grouping system and
product of MediQual Systems Inc., was implemented at
SBGH in October 1991. In addition to demographic data
and ICD-9-CM diagnosis and procedure codes, MediQual
(MQ) uses Key Clinical Findings abstracted from the
medical record. A computer algorithm subsequently
assigns the patient's severity of illness upon admission, as
measured by the probability of death (Steen et al., 1993).
With the exception of obstetrics, newborns and neonates,
psychiatry and palliative care, all remaining inpatient discharges are included within the MQ system (approximately 55% of 22,000 annual inpatient discharges).
Most recently, the Patient-Based Accounting and
Budgeting (PBAB) database was developed to provide
accurate cost data for SBGH patients. Information from
clinical databases (e.g. radiology, pharmacy, laboratory,
physiotherapy) is integrated with data elements from the
MRS database. By linking patient-specific activity of treatment, this system calculates and assigns the cost of treatment per individual patient, whether patient groups are
defined by diagnosis group, CMG or other criteria.
Currently, the Utilization Management Program includes
three utilization analysts with training in nursing and
health records administration. Each analyst is responsible
for data abstraction (e.g. MQ—Key Clinical Findings),
analysis, presentation and interpretation of health care
data. A data entry clerk supports the MQ System. Further
assistance is provided by an advisory group comprising
three individuals with training in epidemiology, computer
programming, mathematics and health records analysis.
Overall, the Utilization Management Program is managed
by the Director of Medical Information Services.
Meetings are routinely scheduled with clinical staff and
administrators to review health care data. Since its inception, more than 200 meetings have taken place and many
clinical staff have adopted a leadership role, identifying
information to monitor resource utilization within their
1
69
department. In the near future, the health care information system will be available on-line so that clinical team
leaders and hospital administrators will have immediate
access to hospital data to facilitate utilization review.
LENGTH OF STAY
In Manitoba, hospital efficiency has been examined by
comparing length of stay (LOS) across urban hospitals
after adjusting for patient characteristics (e.g. age, casecomplexity, income level, Treaty Indian status) (Brownell
and Roos, 1995). This study estimated the potential for
substantial savings, as measured in patient days and hospital beds, if study hospitals were to adopt similar LOS
practices as those reported at the shortest-stay facility. At
SBGH, LOS practices are similarly reviewed. For this purpose, utilization analysts regularly review various CIHI
reports, such as Length of Stay, RIW and Comparison of
Hospital Activity Program (CHAP) reports, and present
LOS data by CMG and doctor service to clinical team
leaders. CIHI data are particularly useful since hospitalspecific LOS may be compared with the LOS of the CIHI
database or select peer-group hospitals while controlling
for the case-mix of the patient population. With the availability of these data, performance relative to the reference
may be assessed and LOS benchmarks may be considered
as practice targets. Most importantly, the availability of
comparative data facilitates utilization review by generating
questions and plausible explanations to explain variation in
LOS (e.g. practice patterns, patient characteristics).
In the following sections, data from the Departments of
Obstetrics and Surgery are presented to illustrate the use
of CIHI case-mix tools in the management of LOS.
While not exhaustive, these examples demonstrate how
utilization analysts report and interpret CIHI data.
Obstetrics
At SBGH, the Department of Obstetrics and Newborns
accounts for the largest segment of inpatient activity
(approximately 45% of annual inpatient discharges and
18% of inpatient days). Therefore, any consistent reduction in LOS will serve to improve overall hospital efficiency. Table 1 reports the number of deliveries and
average length of stay (ALOS) by case mix group for fiscal year 1995/1996. The higher volume CMG include
CMG 611 (Vaginal Delivery, no VBAC, no CC) and
CMG 609 (Vaginal Delivery, no VBAC, with CC)
accounting for 54.1% and 22.3% of deliveries respectively. Overall, Cesarean sections (C/S) were performed in
17.4% of cases1.
The percent of Cesarean sections is based on the total number of cases for corresponding case mix groups divided by the
total number of deliveries (e.g. 657/3770).
70
UTILIZATION MANAGEMENT AT ST. BONIFACE GENERAL HOSPITAL, WINNIPEG, MANITOBA
Table 1: Number of Deliveries and Average Length of Stay (days) by Case Mix Group,
Department of Obstetrics, St. Boniface General Hospital, April 1, 1995 to
March 31, 1996
CM G
(1)
600 - Major Proc. in Childbirth
No. Total
Cases (%)
(2)
ALOS
1
(3)
No. Typical
2
Cases (%)
Typical
ALOS
CIHI
3
ALOS
(4)
(5)
(6)
3 (0.1)
10.0
0 (0)
-
-
79 (2.1)
9.1
61 (77.2)
5.0
4.8
602 - C/S, with CC
199 (5.3)
7.8
171 (85.9)
6.0
5.6
603 - Repeat C/S, no CC
173 (4.6)
4.2
168 (97.1)
4.1
4.3
604 - C/S, no CC
206 (5.5)
4.9
188 (91.3)
4.6
4.7
1 (0.0)
3.0
1 (100.0)
3.0
2.1
43 (1.1)
4.3
41 (95.3)
4.1
3.4
17 (0.5)
5.0
15 (88.2)
4.1
3.2
55 (1.5)
4.5
48 (87.3)
3.5
3.1
839 (22.3)
4.1
738 (88.0)
3.5
3.2
114 (3.0)
2.7
107 (93.9)
2.6
2.4
2041 (54.1)
2.6
1958 (95.9)
2.5
2.4
3770 (100)
3.6
3496 (92.7)
3.2
3.0
601 - Repeat C/S, with CC
605 - Fetal Surgery
606 - Vaginal Delivery
with Sterilization
607 - Vaginal Delivery
with Minor Procedure
608 - VBAC, with CC
609 - Vaginal Delivery, no
VBAC, with CC
610 - VBAC, no CC
611 - Vaginal Delivery,
no VBAC, no CC
TOTAL
1
2
3
ALOS is average length of stay.
For each CMG, the percent of typical cases is calculated as the number of typical cases (column 4)
divided by the total number of cases (column 2) multiplied by 100.
CIHI ALOS is average length of stay of typical cases discharged from Canadian teaching hospitals
(peer group 5) that are CIHI client hospitals.
Source: CIHI Reports, Length of Stay and Resource Intensity Weights, April 1, 1995 to March 31, 1996.
To compare LOS with the CIHI database, typical cases
are reviewed. These cases represent the completion of a
full course of inpatient treatment. Deaths, sign-outs,
transfers and outliers are excluded. As shown in Table 1,
typical ALOS ranges from 2.5 days for CMG 611 (Vaginal
Delivery, no VBAC, no CC) to 6.0 days for CMG 602
(C/S with CC). With the exception of two case mix
groups, CMG 603 (Repeat C/S, no CC) and CMG 604
(C/S, no CC), the remaining groups demonstrate lengths
of stay longer than would be expected, although average
length of stay is based on a small number of typical cases
for CMG 605 (Fetal Surgery) and CMG 607 (Vaginal
Delivery with Minor Procedure). For CMG 609 (Vaginal
Delivery, no VBAC, with CC), ALOS is 3.5 days while
CIHI ALOS is 3.2 days. Upon initial inspection, this difference of 0.3 days suggests that patients at SBGH
remained in hospital 9% longer than expected (i.e.
observed/expected ALOS). In other words, an additional
221 days were required to care for patients within CMG
609. Before such findings are substantiated, however, several factors require consideration. First, is a difference of
0.3 days statistically significant? If so, is this difference
clinically meaningful? CIHI does not report accompanying statistics such as the standard deviation. Without
information describing the variability of LOS, these questions cannot be addressed. Second, supplementary infor-
mation is required that describes the clinical and demographic characteristics of the patient population, since
differences among patient groups, hospital-specific characteristics, provincial geography and health services systems might explain variation in LOS. In Manitoba, for
example, the province is characterized by a distinct
urban/rural dichotomy with a concentration of residents
in the City of Winnipeg. Women flown in from remote
Northern communities to receive obstetrical care at
SBGH might be expected to remain in hospital slightly
longer than their Winnipeg counterparts.
CIHI does not report comprehensive demographic data;
however, information describing the age distribution of
SBGH patients and of those within the CIHI database or
select peer group hospital is available within the CHAP 2
report. Likewise, utilization analysts may consult the CHAP
1 report for information describing the atypical case load
by CMG. A higher percentage of atypical patients relative
to the CIHI database might indicate differences in patient
characteristics despite case-mix adjustment. Lastly, CIHI
data for SBGH may be integrated with MRS data (see
Figure 1) to identify other important information such as
co-existing disease (e.g. diabetes), parity, aboriginal status
and region of residence. Plausible explanations for higher
length of stay practices may then be inferred.
UTILIZATION MANAGEMENT AT ST. BONIFACE GENERAL HOSPITAL, WINNIPEG, MANITOBA
After reviewing the spectrum of obstetrical CMG (for
deliveries), individual CMG are presented and LOS performance is reviewed relative to peer group facilities. In
Table 2, the number of cases and ALOS for CMG 602
(Cesarean Delivery with Complicating Diagnosis) is
reported for selected teaching hospitals. Referring to the
provincial ALOS, SBGH appears to discharge patients
more efficiently, with a typical ALOS of 6.0 days (0.3
days less than the provincial average). When compared to
the CIHI database average of 5.6 days, SBGH appears to
discharge their patients less efficiently than would be
expected (0.4 days above the CIHI average). When
reporting these values, however, utilization analysts
71
describe the composition of hospitals upon which the
provincial and CIHI ALOS values have been computed.
The provincial average length of stay, for example, is
based upon data reported by acute-care facilities that submit discharge abstracts to CIHI (regardless of peer group
status). In Manitoba, current CIHI hospitals include two
teaching and five community hospitals located within the
City of Winnipeg. With respect to the CIHI database,
teaching hospitals across Canada are included. Given the
population density and complete CIHI participation rate
in several provinces (CIHI, 1994b), CIHI ALOS is likely
influenced by data submitted from provinces such as
Ontario, Alberta and British Columbia.
Table 2: CMG 602—Cesarean Delivery with Complicating Diagnosis: Number of Cases and
Average Length of Stay (days) by Peer Group Facility, April 1, 1995 to March 31, 1996
Hospital
No. Total
Cases
(1)
(2)
ALOS
1
No. Typical
2
Cases (%)
Typical
ALOS
Prov.
3
ALOS
CIHI
4
ALOS
(4)
(5)
(6)
(7)
(3)
Atlantic
Hospital A
357
9.2
285 (79.8)
5.5
6.5
5.6
Ontario
Hospital
Hospital
Hospital
Hospital
208
377
174
332
7.8
8.0
7.0
8.4
126
275
138
233
(60.6)
(72.9)
(79.3)
(70.2)
5.5
5.7
5.5
5.5
5.6
5.6
5.6
5.6
5.6
5.6
5.6
5.6
Prairies
St. Boniface
Hospital G
Hospital H
Hospital I
199
269
251
391
7.8
7.7
7.9
5.5
171
204
206
353
(85.9)
(75.8)
(82.1)
(90.3)
6.0
5.8
6.4
4.0
6.3
6.3
6.2
5.0
5.6
5.6
5.6
5.6
British Columbia
Hospital J
696
8.9
575 (82.6)
5.4
5.5
5.6
1
2
3
4
B
C
D
E
ALOS is average length of stay.
For each CMG, the percent of typical cases is calculated as the number of typical cases (column
4) divided by the total number of cases (column 2) multiplied by 100.
Provincial ALOS is average length of stay of typical cases discharged from provincial acute-care
hospitals that are CIHI client hospitals.
CIHI ALOS is average length of stay of typical cases discharged from Canadian teaching hospitals
(peer group 5) that are CIHI client hospitals.
Source: CIHI CHAP 1 Report: CMG 602, April 1, 1995 to March 31, 1996.
To consider "best practice" benchmarks, the relative
placement of SBGH among select hospitals may be
examined by calculating and ordering the number of days
below and above the CIHI database. Referring to the data
presented in Table 2, Hospital I would be ranked first,
reporting the best performance, with a typical ALOS of
4.0 days (1.6 days or 29% below the CIHI average), while
Hospital H would rank last, with a typical ALOS of 6.4
2
days (0.8 days or 14% above the CIHI average). SBGH
(0.4 days or 7% above the CIHI average) would rank just
ahead of Hospital H and behind Hospital G2. Given the
LOS performance at Hospital I, clinical team leaders may
review their practice to determine whether a benchmark
practice of 4.0 days for CMG 602 is feasible, or whether
the CIHI database average is a more realistic target given
the patient population at SBGH and other related factors.
The relative placement of SBGH is based on the teaching hospitals selected for comparison as outlined in Table 2.
Relative positioning would likely vary if each of the teaching hospitals participating with CIHI were included for review.
72
UTILIZATION MANAGEMENT AT ST. BONIFACE GENERAL HOSPITAL, WINNIPEG, MANITOBA
tic and ENT surgery, with both hospitals reporting ALOS
below the CIHI database average. For general surgery,
patients at Hospital A are discharged more quickly than at
Hospital B. Each of the remaining services demonstrates
better than expected LOS relative to the CIHI average. In
most instances, however, LOS differences between hospitals exceed 1 day, perhaps indicating differences in discharge efficiency. Although additional information
describing clinical, demographic and hospital-specific
information is required before drawing such a conclusion,
the data presented in Table 3 provide an overall review of
surgical services and may serve to generate questions and
direct resources for further study. Also included in Table
3 (column 4) is the percent of elective cases receiving surgical services on the same day as admission. This indicator is particularly useful when identifying opportunities to
reduce pre-operative LOS.
Surgery
Working collaboratively with the University of Manitoba
and Health Sciences Centre, hospital administrators
recently appointed one clinical leader (a surgeon) to manage the Departments of Surgery at each of the teaching
hospitals. To provide an overall review of the surgical services and for consistency in reporting between the two
hospitals, utilization analysts report LOS by doctor service for each institution. Table 3 presents the number of
cases, average length of stay, and days over/under the
CIHI database average for surgical services at each of the
Winnipeg teaching hospitals for fiscal year 1995/96.
Because case-mix groups may vary within doctor services,
it is not appropriate to directly compare typical ALOS at
each hospital. Instead, a case mix adjusted comparison of
LOS variation is more appropriate, as reported in column
8. LOS practices are similar for urology, orthopedic, plas-
Table 3. Number of Cases and Average Length of Stay (days) by Doctor Service, Department
of Surgery, Winnipeg Teaching Hospitals, April 1, 1995 to March 31, 1996
Doctor Service
(1)
General Surgery
Hospital A
Hospital B
Cardiac Surgery
Hospital A
Hospital B
Neurosurgery
Hospital A
Hospital B
Oral Surgery
Hospital A
Hospital B
Orthopedic
Surgery
Hospital A
Hospital B
Plastic Surgery
Hospital A
Hospital B
Thoracic Surgery
Hospital A
Hospital B
Urology
Hospital A
Hospital B
ENT
Hospital A
Hospital B
Vascular Surgery
Hospital A
Hospital B
1
2
No. Total
Cases
No.
Elective
Cases
% Elect.
Op. on
Admit Day
No. Typical
Cases (%)
Typical
ALOS 1
DB
ALOS 2
Days Over/
Under DB
(2)
(3)
(4)
(5)
(6)
(7)
(8)
1790
2176
910
802
64
88
1573 (87.9)
1712 (78.7)
5.4
5.7
5.6
5.4
-0.2
+0.3
640
317
366
177
10
45
433 (67.7)
218 (68.8)
10.1
8.1
10.6
10.1
-0.5
-2.0
140
791
113
308
72
81
132 (94.3)
458 (57.9)
3.6
6.6
6.7
7.9
-3.1
-1.3
62
177
49
72
96
90
60 (96.8)
159 (89.8)
2.3
1.9
2.9
3.3
-0.6
-1.4
636
1337
231
661
88
95
521 (81.9)
1018 (76.1)
7.0
7.4
7.3
7.5
-0.3
-0.1
342
791
187
184
91
84
317 (92.7)
536 (67.8)
2.9
4.5
3.4
4.7
-0.5
-0.2
201
369
136
176
50
79
179 (89.1)
293 (79.4)
6.3
8.0
8.2
8.7
-1.9
-0.7
876
611
534
353
66
83
814 (92.9)
544 (89.0)
4.0
4.6
4.9
5.4
-0.9
-0.8
243
221
132
45
85
84
227 (93.4)
186 (84.2)
2.3
4.2
3.4
5.5
-1.1
-1.3
304
682
193
356
41
64
251 (82.6)
535 (78.4)
6.3
6.5
8.0
9.1
-1.7
-2.6
ALOS is average length of stay.
DB ALOS is average length of stay of typical cases discharged from Canadian teaching hospitals (peer group 5)
that submit data to CIHI.
Source: CIHI CHAP 1 and 2 Report, April 1, to March 31, 1996.
UTILIZATION MANAGEMENT AT ST. BONIFACE GENERAL HOSPITAL, WINNIPEG, MANITOBA
By shifting surgery from inpatient to outpatient activities,
hospital resources might be expended more efficiently.
Included in Appendix I is an example of a comparison of
inpatient to outpatient activity by doctor service at SBGH
during fiscal year 1995/96. The number of one-day cases
(column 6), percent outpatient activity (column 10), database percent outpatient activity (column 11) and potential
inpatient cases to move (column 13) is reviewed for the
following nine doctor services: general surgeon, neurosurgeon, oral surgeon, orthopaedic surgeon, plastic surgeon,
thoracic surgeon, urologist, otolaryngologist, and vascular
surgeon. An overall assessment obtained by reviewing the
service totals indicates that five of the nine doctor services performed less surgery on an outpatient basis than
the CIHI database. For example, 17.7% of procedures
performed by a vascular surgeon were completed on an
outpatient basis, compared with a database value of
44.1%. Further, 77.2% (61/79) of screened inpatients, the
majority of which had a vascular procedure (i.e. DPG 21),
reported a LOS of one day, indicating potential for
greater outpatient activity. If procedures within this doctor service were performed at an outpatient activity level
equal to that of the CIHI database, 25 cases overall (column 13), and more specifically, 19 cases within DPG 21
(column 13), could potentially be moved from inpatient
to outpatient activities (i.e. number of inpatient and outpatient cases x database percent outpatient activity - number of outpatient cases). For DPG 21, 19 hospital days
would be spared if 19 of the 52 one-day cases were
moved from inpatient to outpatient activities. However, if
all of the 14 cases remaining in hospital 2 to 3 days were
moved to outpatient activities, 28 to 42 days would potentially be saved.
DISCUSSION
With the availability of CIHI data, LOS practices may be
examined across Canadian acute-care institutions.
Inpatient records are grouped by CMG, day care encounters are assigned a DPG and inter-hospital and CIHI
database comparisons are presented by peer group.
Limited descriptive information such as patient age and
atypical case load is available to assist users in assessing
the comparability of each hospital relative to each other,
or with the CIHI database. With full implementation of
complexity overlay and age adjustment (CIHI, 1997),
more precise information will become available to
describe the patient population and strengthen estimation
of resource use.
Each year, CIHI revises its CMG methodology to reflect
current practice patterns, coding conventions, and medical technology. In 1994/95, CIHI completely redefined
Major Clinical Category (MCC) 14 (Conditions
Originating during Pregnancy and Childbirth). The num-
73
ber of corresponding case mix groups (for delivered
cases) increased from five to twelve (CIHI, 1994a).
Undoubtedly, revisions to the grouping methodology
more accurately reflect the current health care system.
However, as health reform initiatives continue to be
implemented, SBGH and other client hospitals will
require information to monitor utilization over time.
Unless client hospitals approach CIHI for historical
regrouping of their data, hospitals will be unable to conduct these reviews.
When presenting and interpreting CIHI data, there are several measurement issues to consider. First, the user must be
aware that the average is influenced by extreme observations or outliers when reporting and comparing average
length of stay. Although hospital comparisons are based on
typical cases, the length of stay distribution may remain
skewed at one or more of the various levels of aggregation
(e.g. CMG, doctor service, etc.) regardless of the number
of cases. Without accompanying statistics to describe the
variability of these data, the user must interpret them with
some degree of caution. For this reason, CIHI might consider including additional elements within their reports to
assist client hospitals in interpreting the variation in LOS.
In addition to the minimum and maximum length of stay
values, the 25th, 50th (median) and 75th percentile values
should be included. Furthermore, the standard deviation
should accompany each ALOS. At present, CIHI includes
a LOS quartile comparison which is based upon the LOS
distribution of cases within the CIHI database. For each
client hospital, the LOS distribution is normalized accordingly, revealing short- and long-stay hospitals.
Second, statistical significance is not assessed for ALOS
comparisons using traditional parametric tests. Instead,
CIHI reports statistical significance based upon the binomial distribution—assigning the 75th percentile as the
normative value. For this test, the LOS comparison is
limited to a client hospital and the CIHI database. Similar
results are not presented using the "best practice" hospital
or an alternative hospital as the reference.
Lastly, any variation in LOS might be explained by patient
characteristics as well as hospital-specific characteristics
not currently measured or reported by CIHI. Additional
questions are likely to be raised and explanations sought
when reviewing CIHI data.
Although traditional paper reports will not easily support
the inclusion of additional variables, computerized databases should be able to incorporate these additions with flexibility. Most recently, CIHI included the variables date of separation and postal code within its electronic version of
1996/97 data (CIHI, 1996). With upgrading to the electronic data, users could further manipulate current levels of
74
UTILIZATION MANAGEMENT AT ST. BONIFACE GENERAL HOSPITAL, WINNIPEG, MANITOBA
aggregation. At SBGH, analysts would be interested in
making LOS comparisons with the CIHI database for
select CMG within a given doctor service. In the case of
obstetrical data, LOS measured in hours rather than days
might be more appropriate given continuing trends to
shorten LOS. Given the increased flexibility and timeliness
of electronic reporting, CIHI might consider providing
client hospitals with a complete, computerized version of
their paper reports. In Ontario, the Institute for Clinical
and Evaluative Sciences has developed for users an electronic format of their Practice Atlas (Goel et al., 1996).
From a technical perspective, the addition of a unique
identifying variable would undoubtedly facilitate record
linkage projects. At SBGH, CIHI records have previously
been merged with other databases using variables date of
admission and medical record number. However, the success of
future linkages would be greatly enhanced with the addition of claim number (i.e. Manitoba Health-assigned unique
identifier for each hospital encounter) and the recently
introduced date of separation. A current pilot project at
SBGH involves the linkage of CIHI complexity data
(beta version—fiscal 1995/96) with MQ and PBAB databases to assign severity of illness and cost per case for
records defined as Acute Myocardial Infarction (CMG
194 and CMG 195). These data are to be examined by
doctor services, specifically, medicine, family practice and
geriatrics.
CIHI has made significant progress in developing
improved methodologies (e.g. complexity and age adjustment enhancements) to better understand and utilize the
data. However, both hospital-based expertise in data
analysis and expectations of users within SBGH have
evolved substantially since 1990. As a result, analysts and
users are requesting additional statistics (e.g. median
length of stay, standard deviation of mean length of stay)
and increased flexibility in accessing and manipulating the
data. The challenge for CIHI will be to provide such
information on a timely and ongoing basis in order to
meet the needs of their clients, whether it be at the hospital, regional, provincial or national level.
UTILIZATION MANAGEMENT AT ST. BONIFACE GENERAL HOSPITAL, WINNIPEG, MANITOBA
75
REFERENCES
Brownell, M. D. and Roos, N. P. (1995). Variation in
length of stay as a measure of efficiency in Manitoba
hospitals. Canadian Medical Association Journal 152(5):
675-682.
Canadian Institute for Health Information. (1997). CIHI
Directions 3(4): 1-4. Ottawa, Ontario: Author.
Canadian Institute for Health Information. (1994a).
Bulletin: CMG updates for fiscal 1994/1995. (May 19, pp.
1-10). Ottawa, Ontario: Author.
Canadian Institute for Health Information. (1994b).
Comparative Analysis Report: Hospital Utilization in Canada.
Ottawa, Ontario: Author.
Canadian Institute for Health Information. (1996). Sample
1996/97 Diskette Record Layout and Importing Instructions.
Ottawa, Ontario: Author.
Goel V., Williams, J. I., Anderson, G. M., BlackstienHirsch, P., Fooks, C. and Naylor, C.D., (Eds.). (1996).
Patterns of Health Care in Ontario. The ICES Practice Atlas
(2nd Ed.). Ottawa: Canadian Medical Association, (pp.
347-348).
Manitoba Health. (1992). Quality Health for Manitobans: The
Action Plan. Winnipeg, Manitoba: Author.
St. Boniface General Hospital. (1995). St. Boniface General
Hospital Annual Report 1994/95. Winnipeg, Manitoba:
Author.
Steen, P. M., Brewster, A. C., Bradbury, R. C., Estabrook,
E., & Young, J. A. (1993). Predicted probabilities of hospital death as a measure of admission severity of illness.
Inquiry, Summer, 1993 (30), 128-141.
About the Authors
Deborah L. Nowicki, MSc is Research Associate, University of
Manitoba, Winnipeg, Manitoba;
Diane French, CCHRA (A) is Senior Health Records Analyst, St.
Boniface General Hospital, Winnipeg, Manitoba; and
Katherine Choptain is Manager, Information Services, St. Boniface
Hospital.
APPENDIX I
77
Appendix 1. Comparison of Inpatient to Outpatient Activity by Doctor Service, St. Boniface
General Hospital, Winnipeg, April 1, 1995 to March 31, 1996
DPG
2
3
28
55
59
27
29
30
23
26
01
11
15
17
21
22
25
53
12
14
24
08
09
16
36
61
34
39
40
41
44
57
01
02
20
21
22
66
CASES 1
DAY
CASES
4
ENDOSCOPY - GI
MASTECTOMY
SKIN PROCEDURES
HEPATOBILIARY PROCEDURES
ANO-RECTAL PROCEDURES
MINOR ANAL PROCEDURES
LYMPHATIC PROCEDURES
HERNIA REPAIR
NERVE AND OTHER PROCEDURES
SINUS PROCEDURES
OTHER RESPIRATORY PROCEDURES
ENDOSCOPY - ENT
VASCULAR PROCEDURES
OTHER VASCULAR PROCEDURES
CHOLECYSTECTOMY
SOFT TISSUE PROCEDURES
OTHER SINUS PROCEDURES
NASAL PROCEDURES
MINOR VASCULAR PROCEDURES
EXTERNAL EYE PROCEDURES
RESPIRATORY PROCEDURES
EXTERNAL EAR PROCEDURES
VASECTOMY
BIOPSY
LOWER URINARY & GENITAL PROC
UTERUS AND ADNEXAL PROCEDURES
ENDOSCOPY AND GYN PROCEDURES
MINOR GYN PROCEDURES
CHEST WALL PROCEDURES
BREAST PLASTIC PROCEDURES
NERVE AND OTHER PROCEDURES
SPINAL PROCEDURES
ANGIOGRAPHY
VASCULAR PROCEDURES
OTHER VASCULAR PROCEDURES
MYLEOGRAM
60
59
43
52
11
13
58
DENTAL SURGERY
SKIN PROCEDURES
MAXILLO FACIAL PROCEDURES
REMOVAL INTERNAL FIXATION
SINUS PROCEDURES
TONSIL/ADENOID PROCEDURES
PLASTIC RECONSTRUCTION
50
45
47
52
49
53
48
44
46
51
01
59
54
61
KNEE PROCEDURES
UPPER EXTREMITY PROCEDURES
TENDON AND MUSCLE PROCEDURES
REMOVAL INTERNAL FIXATION
LOWER EXTREMITY PROCEDURES
SOFT TISSUE PROCEDURES
CLOSED REDUCTIONS
CHEST WALL PROCEDURES
OPEN REDUCTION & INTERNAL FIX
ANKLE AND FOOT PROCEDURES
NERVE AND OTHER PROCEDURES
SKIN PROCEDURES
MANIPULATIONS
BIOPSY
59
14
53
01
08
47
11
SKIN PROCEDURES
NASAL PROCEDURES
SOFT TISSUE PROCEDURES
NERVE AND OTHER PORCEDURES
EXTERNAL EYE PROCEDURES
TENDON AND MUSCLE PROCEDURES
SINUS PROCEDURES
CASES
2-3
DAYS
5
25
35
18
7
57
18
4
86
2
24
2
1
6
8
6
10
1
32
6
2
35
1
11
1
1
7
17
29
8
6
25
12
2
51
1
13
1
4
168
1
89
3
79
2
1
1
OUTPT
CASES
8
460
209
173
109
68
44
40
30
9
7
7
7
7
7
6
6
5
5
3
2
2
1
1
1
1
1
1
6
24
1
1
1
1
489
232
257
1,209
22
8
2
1
2
1
7
7
2
15
1
12
36
18
3
3
35
1
2
1
2
9
1
1
44
14
30
77
25
19
5
3
47
1
4
3
3
6
15
1
17
6
1
1
21
8
13
4
2
26
1
3
340
134
56
48
39
26
19
17
11
10
9
6
3
1
132
55
77
719
21
2
7
19
1
15
3
12
1
3
16
1
10
1
9
1
4
3
669
92
76
64
57
45
40
3
21
1
1
1
3
3
1
1
1
2
1
1
3
1
2
1
1
18
12
26
52
10
8
6
1
1
1
2
2
4
14
5
2
INPT +
OUTPT
ACTIVITY
9
485
244
191
116
125
62
44
116
11
31
9
8
7
11
174
6
7
5
3
2
2
1
2
2
1
6
24
1
1
1
1,698
34
8
2
1
2
1
48
55
10
43
7
3
1
2
121
365
153
61
51
86
27
23
20
14
16
24
7
3
1
851
690
94
83
83
58
60
43
% OUTPT
10
94.84%
85.65%
90.57%
93.96%
54.40%
70.96%
90.90%
25.86%
81.81%
22.58%
77.77%
87.50%
100.00%
63.63%
3.44%
100.00%
71.42%
100.00%
100.00%
100.00%
100.00%
100.00%
50.00%
50.00%
71.20%
35.29%
PO
INP
CA
TO
DATABASE
% OUTPT
DATABASE
75 %ILE
11
74.12%
77.86%
88.92%
84.55%
68.93%
79.59%
66.72%
42.10%
81.01%
52.23%
66.55%
82.63%
40.19%
55.36%
5.35%
87.43%
68.15%
85.88%
96.84%
94.39%
26.52%
98.57%
91.81%
79.53%
56.49%
35.98%
91.49%
90.63%
62.27%
75.06%
12
95.99%
87.87%
94.31%
92.30%
80.00%
87.23%
81.63%
54.49%
94.11%
69.35%
82.35%
92.85%
61.58%
78.78%
13.79%
93.63%
90.00%
92.30%
98.64%
97.18%
46.15%
99.24%
96.59%
91.34%
70.50%
46.94%
94.90%
92.85%
73.91%
88.46%
13
10
19
311
18
5
11
19
94.11%
94.06%
88.25%
61.58%
78.78%
95.40%
16
7
1
9
1413
1111
1
1
2
22
1
1
1
67.01%
81.01%
82.27%
57.54%
40.19%
55.36%
69.03%
25.00%
78.07%
94.54%
100.00%
18.60%
85.71%
33.33%
93.97%
88.92%
25.67%
78.56%
52.23%
45.32%
59.20%
63.63%
66.37%
93.15%
87.58%
91.80%
94.11%
45.34%
96.29%
82.60%
85.00%
78.57%
62.50%
37.50%
85.71%
100.00%
100.00%
85.56%
28.96%
64.57%
78.56%
18.56%
87.43%
46.92%
62.27%
47.59%
48.71%
81.01%
88.92%
85.56%
79.53%
84.48%
63.73%
96.95%
97.87%
91.56%
77.10%
98.27%
75.00%
93.02%
88.92%
85.88%
87.43%
81.01%
94.39%
64.57%
52.23%
1
1
25
97.50%
94.31%
41.66%
88.70%
69.35%
72.34%
77.27%
13
11
1
3
94.04%
38.03%
80.95%
88.70%
28.33%
93.63%
75.00%
73.91%
63.79%
62.90%
94.11%
94.31%
93.54%
91.34%
28
90
17
823
2854210
94.31%
92.30%
93.63%
94.11%
97.18%
80.95%
69.35%
55
11
33
2618
78
APPENDIX I
DPG
2
3
59
14
53
01
08
47
11
56
55
46
57
58
52
04
51
45
43
44
10
20
23
26
22
48
60
17
28
61
09
23
35
34
36
37
33
61
26
59
60
23
31
32
41
CASES 1
DAY
CASES
2-3
DAYS
5
21
2
7
19
1
15
3
103
3
18
2
2
6
12
1
3
16
1
10
1
32
1
11
7
9
1
4
3
6
12
3
4
1
3
1
1
3
1
1
220
102
118
1,148
39
2
3
1
3
27
2
2
12
29
1
1
48
31
17
31
79
213
40
34
1
43
2
95
26
31
1
16
1
118
14
3
526
25
21
12
11
7
1
1
1
739
65
55
13
54
9
1
1
1
4
7
1
1
CASES
4
SKIN PROCEDURES
NASAL PROCEDURES
SOFT TISSUE PROCEDURES
NERVE AND OTHER PORCEDURES
EXTERNAL EYE PROCEDURES
TENDON AND MUSCLE PROCEDURES
SINUS PROCEDURES
AUGMENTATION/MAMMOPLASTY
MASTECTOMY
OPEN REDUCTION & INTERNAL FIX
BREAST PLASTIC PROCEDURES
PLASTIC RECONSTRUCTION
REMOVAL INTERNAL FIXATION
ORBITAL & OTHER EYE PROCEDURES
ANKLE AND FOOT PROCEDURES
UPPER EXTREMITY PROCEDURES
MAXILLO-FACIAL PROCEDURES
CHEST WALL PROCEDURES
TYMPANOPLASTY
ANGIOGRAPHY
LYMPAHATIC PROCEDURES
HERNIA REPAIR
OTHER VASCULAR PROCEDURES
CLOSED REDUCTIONS
DENTAL SURGERY
ENDOSCOPY - ENT
ENDOSCOPY - GI
BIOPSY
RESPIRATORY PROCEDURES
LYMPHATIC PROCEDURES
BLADDER & URETHRAL PROCEDURES
LOWER URINARY & GENITAL PROC
VASECTOMY
CIRCUMCISION
UPPER URNIARY PROCEDURES
BIOPSY
HERNIA REPAIR
SKIN PROCEDURES
DENTAL SURGERY
LYMPHATIC PROCEDURES
MECHANICAL IMPLANTS
LITHOTRIPSY
MINOR GYN PROCEDURES
5
2
71
2
7
2
2
3
8
1
1
1
3
27
1
4
7
1
1
1
3
1
1
3
4
346
176
170
OUTPT
CASES
8
669
92
76
64
57
45
40
27
25
12
11
6
5
4
4
3
2
2
1
1
1
1
605
INPT +
OUTPT
ACTIVITY
9
690
94
83
83
58
60
43
130
28
30
13
8
5
4
4
9
14
2
1
1
1
2
3
1
1
1,368
68
3
4
1
3
951
% OUTPT
10
96.95%
97.87%
91.56%
77.10%
98.27%
75.00%
93.02%
20.76%
89.28%
40.00%
84.61%
75.00%
100.00%
100.00%
100.00%
33.33%
14.28%
100.00%
100.00%
100.00%
100.00%
50.00%
83.91%
42.64%
33.33%
25.00%
39.24%
71.17%
38.46%
38.18%
92.30%
20.37%
77.77%
100.00%
100.00%
100.00%
63.61%
11
SINUS PROCEDURES
55
50
5
288
343
83.96%
13
TONSIL/ADENOID PROCEDURES
14
13
1
146
160
91.25%
16
EXTERNAL EAR PROCEDURES
1
1
110
111
99.09%
12
OTHER SINUS PROCEDURES
2
1
1
86
88
97.72%
17
ENDOSCOPY - ENT
23
18
5
68
91
74.72%
10
TYMPANOPLASTY
21
18
3
61
82
74.39%
59
SKIN PROCEDURES
4
3
1
47
51
92.15%
14
NASAL PROCEDURES
4
1
3
12
16
75.00%
23
LYMPHATIC PROCEDURES
6
6
100.00%
61
BIOPSY
1
1
5
6
83.33%
58
PLATIC
RECONSTRUCTION
2 31, 1996
100.00%
Source: CIHI Reports,
Inpatient/Outpatient
Comparison by Doctor Service, April 1, 19952 to March
04
ORBITAL & OTHER EYE PROCEDURES
1
1
100.00%
08
EXTERNAL EYE PROCEDURES
1
1
100.00%
09
RESPIRATORY PROCEDURES
14
6
8
1
15
6.66%
15
OTHER RESPIRATORY PROCEDURES
2
2
1
3
33.33%
28
ENDOSCOPY - GI
2
2
1
3
33.33%
43
MAXILLO-FACIAL PROCEDURES
1
1
1
2
50.00%
DATABASE
% OUTPT
11
88.92%
85.88%
87.43%
81.01%
94.39%
64.57%
52.23%
32.97%
77.86%
47.59%
75.06%
59.20%
78.56%
57.59%
48.71%
28.96%
25.67%
62.27%
54.71%
57.54%
66.72%
42.10%
55.36%
46.92%
93.97%
DATABASE
75 %ILE
PO
INP
CA
TO
12
94.31%
92.30%
93.63%
94.11%
97.18%
80.95%
69.35%
40.00%
87.87%
63.79%
88.46%
77.27%
88.70%
88.23%
62.90%
38.03%
41.66%
73.91%
78.78%
88.25%
81.63%
54.49%
78.78%
75.00%
97.50%
13
55
1
33
261
16
32
11122-
92.85%
95.99%
91.34%
46.15%
81.63%
27
1
2
2
1-
2
1
77.83%
82.63%
74.12%
79.53%
26.52%
66.72%
80.83%
79.29%
56.49%
91.81%
92.60%
37.99%
79.53%
42.10%
88.92%
93.97%
66.72%
9.36%
97.44%
90.63%
3
94.95%
70.50%
96.59%
96.39%
52.63%
91.34%
54.49%
94.31%
97.50%
81.63%
18.18%
99.29%
92.85%
60
12
29
69.35%
72.34%
99.24%
90.00%
92.85%
78.78%
94.31%
92.30%
81.63%
91.34%
77.27%
88.23%
97.18%
46.15%
82.35%
95.99%
41.66%
10
73
126
7
16
22
2-
75.74%
52.23%
45.32%
98.57%
68.15%
82.63%
54.71%
88.92%
85.88%
66.72%
79.53%
59.20%
57.59%
94.39%
26.52%
66.55%
74.12%
25.67%
2
10
13
1
1
1
1
13
1
1
C H A P T E R
7
BRENDA TIPPER, DARREN ARNDT
The Effect of Complexity
and Age Adjustment on
Measures of Length of Stay
Performance
CHAPTER OVERVIEW
The Toronto Hospital (TTH) is a teaching hospital affiliated with the University of
Toronto. It is among the largest hospitals in Canada, and is often the institution of last
resort for many of the difficult cases in Ontario. An objective for improving the utilization
of resources at The Toronto Hospital is to minimize length of stay as appropriate. A key
indicator of length of stay performance used at this hospital is days per weighted case. This
indicator is reported on a regular basis to clinicians to assist them in monitoring and
improving their length of stay performance. The indicator is based on an internal comparison of the current month’s actual length of stay performance against the previous year’s
performance. A concern with using external comparisons of length of stay performance is
that it is difficult to find benchmarks of length of stay based on similar cases. Thus, the
case presented in this chapter seeks to determine whether the new Complexity methodology developed by the Canadian Institute for Health Information (CIHI) is a better predictor
of length of stay performance at TTH than is the CMG methodology. It is shown that the
new prototype, Complexity 96 methodology is a significantly better predictor of length of
stay performance for cases at TTH than the CMG 96 methodology. Thus, the Complexity
methodology may be an appropriate means for making external comparisons. The details
related to this finding are discussed.
INTRODUCTION
Opportunities for improving resource utilization in an institution can often be identified
through comparing indicators of utilization to those of similar cases in other institutions.
Comparisons of indicators through benchmarking can assist not only in determining if
opportunities exist, but also where those opportunities lie, for example, particular groups of
cases, certain programs, etc.
Length of stay (LOS) is a significant predictor of the use of resources such as nursing care,
diagnostic and therapeutic interventions, drugs, etc. Therefore, the focus in utilization
improvements is often on reducing LOS. LOS is one of the key utilization indicators used
by many institutions to assess performance. However, when benchmarking this indicator
against that achieved by other institutions to determine areas where performance might be
improved, a key question is "what are ‘similar’ cases?"
80
THE EFFECT OF COMPLEXITY AND AGE ADJUSTMENT ON MEASURES OF LENGTH OF STAY PERFORMANCE
Until 1997, CIHI’s case mix methodology provided a way
of grouping “similar” cases based on the diagnosis at discharge considered to be most responsible for the patient’s
stay in the institution. Although there is variation in
resource utilization within any particular CMG, the
Resource Intensity Weight (RIW) assigned to a CMG provides an indication of the intensity of resource utilization
relative to other groups. In order to determine the RIW and
also to provide further information with which to assess
resource utilization performance, CIHI calculates a number
of statistics based on the length of stay including average,
median and 25th and 75th percentiles for all typical cases.
An institution can use this data to examine how its length
of stay performance is different from that of CIHI’s
national database. If it is found that cases within related
groups of CMG generally have a length of stay higher
than the CIHI ALOS, then these cases can be further
examined to find practice patterns and systems that might
be changed to decrease LOS. Improvement efforts can
then be focused in those areas identified, without spending effort unnecessarily where LOS performance may be
better than the CIHI ALOS.
However, prior to 1997, case mix groups, based in large
part on the diagnosis most responsible for length of stay,
did not fully reflect factors that may lead to longer LOS
(and higher resource use) for cases with the same most
responsible diagnosis. Complications and comorbid conditions along with the age of the patient (very young or
very old) will significantly affect length of stay and
resources required to provide care for a case. Although a
number of CMG reflect expected differences in resource
use based on broad age groups (e.g. over 70 years, under
16 years) and the presence or absence of complications,
these do not take into account extreme age differences, or
the severity of complications and other conditions.
If all institutions have cases with similar age distributions
and similar patterns of complications and other conditions, comparisons to the CIHI database indicators would
always be valid. However, more complex cases are often
referred to tertiary/quaternary institutions specializing in
care for these cases. This makes it difficult to compare
the length of stay performance in institutions providing
this specialized care to averages calculated over all institutions. Even within the “tertiary institutions,” patterns of
complex cases may be different from one institution to
another depending on the programs of focus. For example, Hospital A may have more complex cardiac cases
than Hospital B, while Hospital B specializes in care for
more complex cancer cases. It would be misleading to target areas for review using comparisons to averages based
on less complex cases.
1
With the addition of complexity and age adjustments for
most CMG, CIHI’s new Complexity methodology offers
the opportunity to compare utilization performance to
indicators that more accurately reflect the range of an
individual institution’s cases. An institution’s average LOS
can be compared to LOS for cases with similar factors
(patient age and comorbid conditions) affecting LOS.
The Situation at The Toronto Hospital
The Toronto Hospital is a teaching hospital affiliated with
the University of Toronto. With nearly 40,000 inpatient
cases and over 260,000 patient days per year, it is among
the largest acute care facilities in Canada. It provides care
for significantly complex cases and is often the institution
of last resort for many of the most difficult cases in
Ontario. The Toronto Hospital has priority programs in
cardiac sciences, oncology, transplantation, neurosciences
and primary care. In a comparison of case complexity
among Toronto Academic Health Science Council
(TAHSC) Hospitals, the “average complexity level” for The
Toronto Hospital was higher than that of the other teaching hospitals associated with the University of Toronto
with the exception of a specialized cancer treatment centre.
In order to optimize resource utilization, an objective at
The Toronto Hospital is to minimize length of stay as
appropriate. To this end, LOS performance is monitored
internally, with monthly indicators that compare LOS to
TTH historical performance. “Days per weighted case” is
used as the indicator. Current performance is compared
to the previous year’s performance by using “comparator
days.” The comparator days reflect the length of stay
given the current month’s case mix that would have
resulted with the previous year’s length of stay performance by CMG1. A sample indicator report is shown in
Figure 1. The LOS indicator can be examined by physician program (e.g. cardiology, neuro-surgery, transplantation). This high-level performance indicator is routinely
provided to clinicians. Ad hoc reports and analysis may
be developed if further investigation is required to identify specific opportunities for LOS reduction, or to determine why LOS may be have been increasing.
As can be seen from the sample chart in Figure 1, “patient
days per weighted case” has been decreasing over the past
two years. Although performance by this measure has been
improving, there has been no indicator developed to compare LOS performance to external standards and identify
where the largest opportunities exist for improving LOS
performance. The impact of complexity on length of stay,
and the issues described above around comparing performance to averages based on the previous CMG methodology, lead to questions regarding the validity of comparing
TTH performance to previous national averages.
The comparator days are obtained by calculating the previous year’s ALOS for each CMG and predicting patient days based on this year’s case
mix and the previous year’s ALOS.
THE EFFECT OF COMPLEXITY AND AGE ADJUSTMENT ON MEASURES OF LENGTH OF STAY PERFORMANCE
Figure 1:
Sample LOS
Indicator Chart
Patient Days per Weighted Case
TTH Average
Patient Days per Weighted Case
81
4,4
4,2
4,0
3,8
3,6
3,4
3,2
3,0
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Month
95/96 Actual
96/97 Actual
The development of CIHI’s Complexity methodology
presents the opportunity to review LOS performance
using a grouping methodology that much more fully
addresses the impact of complexity and age on length of
stay performance. The case study presented in this chapter uses the 1996 pilot version of the Complexity
methodology (Complexity 96) to examine the following
questions:
1. How does TTH LOS performance compare to the
CMG 96 database ALOS; and, how does it compare
to the Complexity 96 ELOS which has been adjusted
for complexity and age?
2. Does the new methodology offer an improvement
in predicting LOS through a decrease in variability
of the difference in actual LOS compared to the
national averages ?
3. For what types of cases is the impact of the age and
complexity adjustment on the CMG most significant?
4. Would performance indicators based on the
Complexity methodology better identify opportunities
for examining LOS performance? What would these
opportunities be?
METHODS
Analysis
To determine if the Complexity methodology better
reflects actual TTH length of stay, we compared actual
TTH length of stay to the CIHI database length of stay
95/96 Compara
for each case included in the review. The length of stays
were identified and compared as follows:
1. For the CIHI CMG 96:
a) the TTH actual LOS was compared with the
CIHI database ALOS for the case, based on
CMG 96
b) the LOS difference was calculated as actual case
LOS minus CIHI database ALOS
2. For the CIHI Complexity 96:
a) the TTH actual LOS was compared with the
CIHI ELOS for the case, based on Complexity 96
b) the LOS difference was calculated as actual case
LOS minus the CIHI ELOS
For both methodologies, a positive difference indicated a
case LOS greater than the CIHI LOS and a negative difference indicated a case LOS less than the CIHI LOS.
The LOS differences calculated using the CMG 96
methodology were compared to the LOS differences calculated using the Complexity 96 methodology.
For each set of differences, an average, median and standard deviation were determined. The average, median and
standard deviation were reviewed for all cases and for
cases broken into the following categories:
² by type of admission (emergent, urgent, elective);
² by medical or surgical partition; and
² by Major Clinical Category.
82
THE EFFECT OF COMPLEXITY AND AGE ADJUSTMENT ON MEASURES OF LENGTH OF STAY PERFORMANCE
Complexity 96 ELOS of 6.72 days (average difference of
0.60 days). These results are shown in Table 1. The CIHI
LOS was 0.23 days closer to TTH’s actual ALOS experience when using the Complexity 96 methodology. The
percent difference improves from 11.3% under the CMG
96 methodology to 8.3% under the Complexity 96
methodology. The variability of the differences in actual
LOS compared to the CIHI LOS (based on the standard
deviation) decreased from 10.15 to 9.25 under the
Complexity 96 methodology (see Table 2). The decreases
in both the average difference and the standard deviation
were significant at a=.01. Figure 2 illustrates the improvement in the mean, median and standard deviation by
showing the distribution of differences in TTH actual
LOS compared to the ELOS on a case by case basis.
Cases Reviewed
The cases reviewed comprised the first quarter of
1995/96 (April, 1995 through June, 1995). Only cases
where complexity and/or age adjustments were applied
under CIHI’s pilot methodology were included. Since
complexity and/or age adjustments were not used for
cases in the following MCC, these were excluded from
the analysis:
²
²
²
²
²
02
03
14
15
19
Eye;
Ear, Nose, Throat, Mouth;
Pregnancy and Childbirth;
Newborns; and
Mental Diseases and Disorders.
In addition, all MNRH (May Not Require
Hospitalization) CMG cases were excluded, as were cases
grouped to CMG > 909. Atypical cases due to signouts,
transfers and deaths were also excluded. However, LOS
outliers were included in the analysis. Total cases included
in the review were 6,627.
Results by Medical/Surgical and
Admission Type
To determine if the improvement was more pronounced
for certain types of cases, the results for all cases were
broken down by medical and surgical partitions and by
admission type. Tables 1 and 2 show the results for these
categories. The overall pattern of results holds for medical and surgical cases, and also for emergent, urgent and
elective cases; however, the degree of improvement
changes by category.
RESULTS/ANALYSIS
Overall Results
The actual TTH ALOS for all cases under review was
7.32 days. This compared to the CMG 96 ALOS of 6.49
days (average difference of 0.83 days) and to the
Table 1: Comparison of TTH Actual ALOS vs. Methodology Difference
TTH
CMG 96 Methodology
Actual
# Cases
Category
All Cases
Medical / Surgical
Admission Type
ALOS
Complexity 96 Methodology
Methdology Difference
Diff v.
TTH
ELOS
% Diff
ELOS
Diff v. TTH
% Diff
6.49
0.83
11.33%
6.72
0.60
8.25%
0.23
3.08%
Medical
Surgical
2,716
3,911
3.48
9.99
4.30
8.01
-0.82
1.98
-23.54%
19.78%
4.19
8.48
-0.70
1.51
-20.20%
15.14%
-0.12
0.46
-3.34%
4.63%
Emergent
6.11%
2,152
10.45
7.07
3.38
32.38%
7.71
2.75
26.27%
0.64
Urgent
704
9.58
7.78
1.80
18.81%
8.13
1.45
15.16%
0.35
3.65%
Elective
3,771
5.12
5.93
-0.81
-15.80%
5.89
-0.78
-15.15%
-0.03
-0.65%
Category
# Cases
All Cases
Admission Type
%
7.32
Table 2: Comparison of Mean, Median and Standard Deviation of Differences
Medical / Surgical
Days
6,627
Medical
Mean Difference
Median Difference
CMG
96
CMG
96
Complexity
96
Std. Dev.
Complexity
96
CMG
96
Complexity
96
6,627
0.83
0.60
-0.70
-0.60
10.15
9.24
2,716
-0.82
-0.70
-0.70
-0.70
3.31
3.04
Surgical
3,911
1.98
1.51
-0.80
-0.60
12.80
11.67
Emergent
2,152
3.38
2.75
0.10
0.10
13.23
12.34
Urgent
704
1.80
1.45
-0.50
-0.40
9.80
8.74
Elective
3,771
-0.81
-0.78
-1.10
-0.80
7.53
6.64
THE EFFECT OF COMPLEXITY AND AGE ADJUSTMENT ON MEASURES OF LENGTH OF STAY PERFORMANCE
For medical cases, TTH actual ALOS was less than the
CIHI LOS for both methodologies. However, with the
Complexity methodology, the absolute difference
becomes smaller, as the CIHI LOS moves closer to the
TTH actual ALOS. For medical cases, the percent difference in LOS compared to the CIHI LOS improves from
-23.5% to -20.2%. The ALOS for TTH surgical cases was
1.98 days higher than the CMG 96 ALOS. Using
Complexity 96 ELOS, the difference drops to 1.51
days and the percent difference improves from 19.8%
to 15.1%. For both medical and surgical cases, the standard deviation of the difference in actual LOS
compared to the CIHI LOS was smaller under the
Complexity methodology.
Comparing Actual Length of Stay to Expected Length of Stay
CMG 96 Methodology
1000
Frequency (Cases)
83
900
Figure 2:
Comparing Actual
Length of Stay to
Expected Length of
Stay
a) Using 1996 CMG
Methodology
800
700
600
500
400
20
18
16
14
12
8
10
6
4
2
0
-2
-4
-6
-8
-10
-12
-14
100
-16
-20
200
-18
300
0
LOS m inus Expected Length of Stay
Comparing Actual Length of Stay to Expected Length of Stay
Complexity 96 Methodology
900
b) Using the 1996
pilot Complexity
Methodology
800
700
600
500
400
0
20
18
16
14
12
10
8
6
4
2
0
-2
-4
-6
-8
-10
-12
-14
100
-16
200
-18
300
-20
Frequency (Cases)
1000
Figure 2:
Comparing Actual
Length of Stay to
Expected Length of
Stay
84
THE EFFECT OF COMPLEXITY AND AGE ADJUSTMENT ON MEASURES OF LENGTH OF STAY PERFORMANCE
For emergent admissions there was a difference of 3.38
days (32.4%) compared to the CMG 96 ALOS. This difference improved to 2.75 days (26.3%) compared to the
ELOS under the Complexity 96 methodology. The difference in actual ALOS compared to the CIHI LOS for
urgent admissions also showed some improvement, moving from a difference of 1.80 days (18.8%) under the
CMG 96 methodology to 1.45 days (15.2%) using the
Complexity 96 methodology. The difference for elective
admissions was almost unchanged, moving from -0.81
days under the CMG 96 methodology to -0.78 days using
the Complexity 96 methodology. The standard deviations
of the difference in actual LOS compared to the CIHI
LOS for emergent, urgent and elective admissions were
again smaller under the Complexity 96 methodology.
Results by Major Clinical Category
Results by Major Clinical Category (MCC) are shown in
Tables 3 and 4. These tables do not include MCC that
were listed in the exclusions described in the methodology section. Also, MCC with fewer than 20 cases (Male
Reproductive System, Burns, and Multiple Significant
Traumas) are not shown in the tables.
Of the sixteen MCC shown in Table 3, thirteen had a
CIHI LOS closer to the TTH ALOS under the
Complexity 96 methodology. For three MCC
(“Hepatobiliary System,” “Endocrine, Nutritional,
Metabolic Diseases” and “Lymphoma, Leukemia, or
Unspecified Site Neoplasms”), the Complexity 96 ELOS
was further away from TTH actual ALOS.
Table 3:
Comparison of TTH Actual ALOS vs. Methodology Difference by MCC
TTH
Actual
Major Clinical Category
ALOS
# Cases
CMG 96 Methodology
Diff vs.
CIHI ALOS
TTH
Complexity 96 Methodology
CIHI
Diff vs.
% Diff
ELOS
TTH
Methdology Difference
% Diff
Days
%
01
Nervous System
576
10.33
8.45
1.87
18.16%
8.82
1.51
14.59%
0.37
3.56%
04
Respiratory System
432
8.32
7.71
0.61
7.30%
7.82
0.50
6.02%
0.11
1.28%
05
Cardiovascular System
1,890
6.68
6.13
0.56
8.33%
6.31
0.37
5.61%
0.18
2.73%
06
Digestive System
634
7.03
5.99
1.04
14.82%
6.27
0.76
10.88%
0.28
3.94%
07
Hepatobiliary System
474
6.02
6.12
-0.10
-1.68%
6.43
-0.41
-6.83%
0.31
5.15%
08
Musculoskeletal System
734
7.89
6.72
1.16
14.77%
7.12
0.77
9.80%
0.39
4.96%
207
5.14
4.94
0.20
3.90%
5.15
-0.01
-0.17%
0.21
4.07%
204
5.00
5.26
-0.27
-5.34%
5.35
-0.36
-7.18%
0.09
1.84%
422
8.76
6.96
1.80
20.55%
7.05
1.72
19.59%
0.08
0.96%
19
6.26
6.72
-0.46
-7.31%
7.43
-1.16
-18.57%
0.71
11.26%
09
10
Skin, Subcutaneous
Tissue, Breast
Endocrine, Nutritional,
Metabolic Diseases
11
Kidney and Urinary Tract
12
Male Reproductive
System
13
Female Reproductive
System
246
5.51
4.61
0.90
16.30%
4.89
0.62
11.25%
0.28
5.04%
16
Blood and Immunological
Disorders
90
8.51
4.99
3.52
41.40%
5.01
3.50
41.11%
0.02
0.29%
17
Lymphoma, Leukemia, or
Unspec. Site Neoplasms
217
9.59
9.91
-0.32
-3.35%
10.08
-0.49
-5.13%
0.17
1.78%
18
Multisystemic or Unspec.
Site Infections
74
8.53
6.79
1.74
20.35%
7.02
1.51
17.67%
0.23
2.68%
21
22
23
24
25
Injury, Poisoning, Toxic
Effects of Drugs
Burns
217
7.85
5.31
2.54
32.35%
6.01
1.85
23.51%
0.69
8.84%
4
2.25
7.18
-4.93
-218.89%
6.48
-4.23
-187.78%
-0.70
-31.11%
Other Reasons for
Hospitalization
130
2.21
3.93
-1.72
-77.94%
3.12
-0.91
-41.11%
-0.81
-36.83%
HIV Infections
55
9.05
9.10
-0.05
-0.52%
9.08
-0.03
-0.28%
-0.02
-0.24%
2
12.50
14.00
-1.50
-12.00%
16.75
-4.25
-34.00%
2.75
22.00%
Multiple Significant
Trauma
THE EFFECT OF COMPLEXITY AND AGE ADJUSTMENT ON MEASURES OF LENGTH OF STAY PERFORMANCE
Table 4: Comparison of Mean, Median and Standard Deviation of Differences by MCC
Mean Difference
Major Clinical Category
CMG
96
# Cases
Median Difference
Complexity
96 CMG 96
Std. Dev.
Complexity
96
CMG
Complexity
96
96
01
Nervous System
576
1.87
1.51
-1.70
-1.60
15.80
14.80
04
Respiratory System
432
0.61
0.50
-1.50
-1.55
9.44
8.46
05
Cardiovascular System
1,890
0.56
0.37
-0.50
-0.60
10.44
9.53
06
Digestive System
634
1.04
0.77
-0.45
-0.60
8.06
7.55
07
Hepatobiliary System
474
-0.10
-0.41
-0.45
-0.55
7.78
6.84
08
Musculoskeletal System
734
1.16
0.77
-0.90
-1.00
8.54
7.71
09
Skin, Subcutaneous
Tissue, Breast
207
0.20
-0.01
-0.60
-0.70
8.24
7.52
Metabolic Diseases
204
-0.27
-0.36
-1.10
-0.95
5.13
5.15
Kidney and Urinary Tract
422
1.80
1.72
-1.70
-1.00
11.48
10.16
19
-0.46
-1.16
-1.00
-1.50
3.57
4.90
246
0.90
0.62
0.30
0.00
4.50
3.85
90
3.52
3.50
1.05
0.50
13.00
13.12
217
-0.32
-0.49
-1.60
-1.20
9.59
8.31
74
1.74
1.51
-0.60
-0.30
7.62
7.29
10
11
12
13
16
Endocrine, Nutritional,
Male Reproductive
System
Female Reproductive
System
Blood and Immunological
Disorders
17
Lymphoma, Leukemia, or
Unspec. Site Neoplasms
18
Multisystemic or Unspec.
Site Infections
21
Injury, Poisoning, Toxic
Effects of Drugs
22
Burns
23
24
25
217
2.54
1.85
-0.50
-0.60
14.13
11.42
4
-4.93
-4.23
-3.30
-3.00
5.84
5.14
Other Reasons for
Hospitalization
130
-1.72
-0.91
-2.10
-0.90
2.31
2.44
HIV Infections
55
-0.05
-0.03
-3.10
-3.10
8.31
8.23
2
-1.50
-4.25
-1.50
-4.25
11.74
7.85
Multiple Significant
Trauma
The MCC showing the greatest improvement in the difference between actual ALOS and ELOS were “Other
Reasons for Hospitalization” (MCC 23), “Injury,
Poisoning, and Toxic Effects of Drugs” (MCC 21) and
“Musculoskeletal System” (MCC 08) (see Figure 3).
Figure 3: Comparison of Difference in Actual vs. ELOS by MCC
Comparison of Difference in Actual vs. ELOS by MCC
Difference Actual vs. Expected (Days)
4,00
3,00
2,00
1,00
0,00
-1,00
-2,00
01
04
05
06
07
08
09
10
11
13
16
17
18
85
86
THE EFFECT OF COMPLEXITY AND AGE ADJUSTMENT ON MEASURES OF LENGTH OF STAY PERFORMANCE
CONCLUSION
Based on the cases reviewed, the ELOS calculated using
the Complexity 96 methodology provides a closer estimate of TTH actual LOS experience than the CIHI database ALOS using the CMG 96 methodology. The average
and median of the differences are both closer to zero and
the standard deviation is smaller. However, the degree of
improvement varies based on certain characteristics of
the cases examined (medical or surgical partition, admission type and major clinical category). Surgical cases
showed more improvement than medical cases; emergent
more than urgent; and urgent more than elective. When
examined by MCC, the largest improvements in the LOS
difference were found in “Injury, Poisoning, Toxic Effects
of Drugs”, “Other Reasons for Hospitalization” and
“Nervous System.” For three MCC (“Hepatobiliary
System,” “Endocrine, Nutritional, Metabolic Diseases”
and “Lymphoma, Leukemia, or Unspecified Site
Neoplasms”), the CIHI LOS moved further away from
TTH’s actual LOS under the Complexity 96 methodology.
Given both the results of this analysis and the premise on
which the complexity and age adjustment is based (specifically accounting for factors that would affect a case’s
LOS), the answer to Question 4 in Section 1 (“would performance indicators based on the age and complexity
adjustment better identify opportunities for examining
LOS performance?”) is “yes.” The new methodology
would likely identify both more realistic and better targeted opportunities for improving length of stay performance. Although the opportunities cannot be clearly
identified at the MCC level of disaggregation in this
analysis, the following measures, monitored on an ongoing basis, would provide better information with which to
assess LOS performance:
² average difference in actual LOS compared to the
ELOS for Complexity (total hospital and by physician
program); and
² within each physician program, the top ten or twenty
Complexity CMG with the greatest difference in
actual LOS compared to ELOS for Complexity.
Ultimately, the usefulness of any indicator depends on
how well it is accepted as a legitimate measure of performance by the individuals who have significant control
over it. By specifically using complexity as one of the factors in identifying “similar” cases, comparisons to the
ELOS are not only closer as shown in this analysis, but
the ELOS can be recognized as a measure more reflective
of the reality of the cases than ALOS.
About the Authors
Brenda Tipper, BComm, MHSc is Manager, Activity Reporting, The
Toronto Hospital;
Darren Arndt, BSc is Manager, Utilization Projects, The Toronto
Hospital, Toronto, Ontario.
C H A P T E R
8
ROBERT FOX, JIANLI LI, ROBERT BEAR
Impact of the Complexity
Methodology on an Ontario
Teaching Hospital
CHAPTER OVERVIEW
In April 1997, the CMG methodology was changed to include the addition of adjustments
for case complexity. The resulting adjustment for complexity considers diagnoses that prolong length of stay and increase the cost of treatment. Traditionally, St. Michael’s Hospital
(SMH) has used Resource Intensity Weights (RIW) in strategic planning, re-engineering
projects, program operations and Ministry of Health discussions. This chapter explains the
impact of the complexity adjustment changes on these efforts. It also identifies problems
that result when a standardized grouping methodology is changed. A discussion follows
that identifies the advantages and disadvantages of the methodological changes as perceived by SMH and the changes at SMH that have occurred and need to occur as a result
of these methodological changes.
BACKGROUND ON ST. MICHAEL’S HOSPITAL
St. Michael’s Hospital is a Teaching and Research Institute affiliated with the University of
Toronto. Located in the heart of Toronto’s downtown core, the hospital’s patients and staff
represent a rich diversity of cultures, religions, languages, lifestyles, and economic status. St.
Michael’s Hospital was founded by the Sisters of St. Joseph in 1892. They remain a vital influence, and are the present-day owners and sponsors. Their mission and values shape decisions,
guide day-to-day actions, and inspire the hospital’s future. Based on respect and dignity for
everyone, the mission and values have led to the hospital’s reputation as “Toronto’s Urban
Angel.” In 1995-96, the hospital had approximately 350 beds, employed over 2,000 people,
admitted 16,377 patients and saw 272,396 outpatients. Its average length of stay was 6.3 days.
That same year, St. Michael’s had the highest funding adjustment amongst its peer group for
tertiary and teaching activity, based on the Ontario Case Cost Formula. This formula was created by the Ontario Ministry of Health to evaluate the efficiency of hospital operations in
terms of cost per weighted case (for acute, newborn and day surgery patients) and was the
basis for funding adjustments to hospitals. St. Michael’s Hospital also ranked as number one
among Ontario teaching hospitals when comparing the amount by which actual cost per
weighted case was below the expected cost per weighted case. To achieve this standing, the
hospital restructured the services it provided in response to the discovery of a debt that
exceeded $60 million. A debt recovery plan was the initial driving force for change and it was
the basis for the creation of a programmatic Strategic Plan. Later, the processes used to provide such services were re-engineered using the Patient Care Journey review. Programs continue to use information reports to monitor and continually improve utilization, quality of patient
care and financial performance. The Health Services Restructuring Commission in Ontario
recommended the closure of certain hospitals and the redistribution of their programs and
services, based upon the same type of operational information. The combination of strategic
planning, re-engineering and program reporting processes has been the backbone of the hospital’s current position. In each of these processes, use of the RIW measure was instrumental.
88
IMPACT OF THE COMPLEXITY METHODOLOGY ON AN ONTARIO TEACHING HOSPITAL
CURRENT USE OF THE RIW IN
MANAGEMENT AND OPERATIONAL
DECISIONS
The RIW is used in most hospital information reports
and plays a critical role in the area of decision support.
Since the RIW reflects resource requirements it is important to review this measure in conjunction with other
measures of quality, utilization and finance. Even though
the RIW does not allow for much differentiation of
resource requirements within a CMG, it does provide a
basis for comparison across CMG and is an effective
measure for strategic planning, process re-engineering for
program operations and planning, for advocacy with the
Ministry of Health regarding funding issues, and for planning regarding the Health Services Restructuring
Commission’s mediated hospital mergers.
Strategic Planning
Upon identification of financial issues in 1991, St.
Michael’s Hospital created a Strategic Plan that identified
areas targeted for resource enhancements or reductions.
Recognizing that the hospital could no longer be "all
things to all people," a detailed review of program costs
and activity was conducted so that an appropriate program configuration could be formulated. By grouping
similar physician services together, clinical programs were
conceived. Next, budgeting and operational planning
exercises utilized program CMG volumes, RIW and
lengths of stay to determine the proper financial budgets
and bed allocations for the programs.
An example of such a review process was with the development of the Cardiac Services program. The program
leadership began with the creation of a task force of
diverse representation that reviewed the hospital activities
relating to the treatment and prevention of heart disease.
These activities were measured using key performance
indicators that included RIW. An optimal program structure was formed based on the hospital’s strengths in this
particular area including clinical expertise, cost comparisons and volumes. The number of weighted cases was
reviewed as part of the planning around volumes and
became an essential component in the creation of the
budget for the program. Estimates of resource requirements were obtained by comparing the RIW per planned
case to historical estimates of case costs. The final program plans were submitted to the Board of Directors for
approval.
Core and support programs were identified from the program exercises. The core programs conceived were: Inner
City Health, Cardiovascular Services, and the Treatment/
Rehabilitation of Trauma Victims, Nutrition and
Metabolism, Neuromusculoskeletal and Urology/
Renal/Endocrine/ENT/Ophthalmology (UREEO)
Service (organized around the care of the complex diabetic). The support programs of diagnostic imaging, laboratory medicine and perioperative services were sized
according to projected workload by the core programs.
With the aid of these case mix tools, a streamlined and
program-based organizational structure emerged. Built on
the foundation of the strategic plan, the hospital programs were re-engineered to optimize their operations.
The use of weighted case information was essential to
this re-engineering effort and to the daily information
needs of the program. The value of the weighted case
assignment was fully appreciated when plans regarding
outpatient clinic activity were required for complete
development of the program. Unfortunately, there are no
case weights assigned to outpatient clinic procedures and
visits and therefore, case mix tools could not be used in
resource planning.
Re-Engineering
In mid-1995, St. Michael’s initiated a major process
improvement project including a complete review of all
patient care and support services. Entitled "The Patient
Care Journey," this re-engineering effort was undertaken
for a number of reasons including:
² the announcement by the Ontario government, in
1995, of its intention to decrease hospital funding by
approximately 5%, 6% and 8% in the next three consecutive years, with individual reductions to be determined by efficiency comparisons with other medium
and large acute hospitals;
² the evolution of the internal and external health care
environment required a rethinking of all processes to
ensure improved cost-effectiveness; and
² the increasing pressure to control costs in the
Toronto health care environment would require hospitals to improve productivity and customer satisfaction while keeping key staff.
The goals and objectives of the re-engineering project
included:
² redesign of the patient care process;
² enhancement of the quality of care delivered;
² enhancement of the quality of work life for physicians, staff and volunteers at SMH;
² reduction of total patient care delivery costs by a
minimum of 15% over 2 years;
² maintenance of academic output;
² maintenance of patient care volumes at current levels
with an appropriate shift from inpatient to ambulatory and short stay services; and
² development of an organizational capacity to continuously improve its basic processes.
IMPACT OF THE COMPLEXITY METHODOLOGY ON AN ONTARIO TEACHING HOSPITAL
The project began with an extensive start-up aimed at
educating staff, physicians and other stakeholders about
the planned process; interviewing and selecting consulting
partners for the venture; recruiting internal project team
members; and visiting other facilities that had completed
similar activities. Once all team members were in place,
two to three months were spent collecting benchmark
information about best practices in Canada and the
United States and establishing care delivery, financial, clinical and quality goals to be used in the redesign phase.
This information included length of stay, case mix
groups, budgets and case weight volume detail. It was
important to consider estimates of the resource requirements using historical CMG weights to plan for program
budgets and size.
A Clinical Resource Management group was established
to review internal and external operational indices in
detail. Its goal was to review lengths of stay, case volumes, weighted cases and ancillary resource utilization so
that new targets and operational standards could be created. Once the targets were reached, the savings generated
would constitute the achievement of a major portion of
the overall goal, with the remainder of the savings to be
derived from the review of all hospital processes, reduction in management staff and staff role redesign.
The process began with an assessment of the current
level of operational utilization. Programs and individual
physicians were compared against one another and a best
practice external hospital. Standardization of resource
requirements was created by normalizing based on the
RIW measures. This allowed for accurate and reliable
comparison measures. By totalling the number of RIW
across programs, information on the amount of activity
the program had conducted over a certain time period
was provided. This information was categorized by CMG
and reviewed on a physician by physician, service by service, or program by program basis.
With this information, initiatives for improvement
were established. These included the formulation and
implementation of clinical pathways, case management
and the re-aggregation of beds to reduce costs and performance variability.
To monitor the effects of these initiatives, three new standard reports were developed. Each report provided new
information in a format that is easily understood by the
various users and each includes cost and RIW information. St. Michael’s Hospital operates on the principle that,
since the average cost per weighted case is determined by
dividing the net cost by the total number of weighted
cases, the estimated expected cost should be equal to the
89
individual patient RIW multiplied by the hospital average
cost per weighted case. This expected cost becomes the
"target" cost for the inpatient or day surgery stay. Using
this indicator in conjunction with other standard performance indicators, three new reports were created that
included the Balanced Scorecard, Program Activity Report and
the Physician Profile.
The Balanced Scorecard (Kaplan and Norton, 1992) was
created to monitor the goals and objectives of the Patient
Care Journey. The report is termed the Balanced Scorecard
because it identifies financial, utilization, academic and
quality indicators together. It is a very powerful report
since it shows the total operational picture and variation in
one specific area that may affect other areas of performance. For example, the decrease in costs of a particular
area may trigger a decrease in total weighted cases or
patient satisfaction ratings. The Balanced Scorecard allows
management to obtain a picture of the entire hospital
across all types of activities. The physician profile report
and the program activity reports were developed in the
Patient Care Journey redesign and have been used in
program planning.
The Program Activity Report is a valuable tool for detailing utilization and financial indicators. Since information
is organized by Case Mix Group (CMG), similar activities
of high volume can be reviewed together. Large variations in performance indicators signal opportunities for
improvement. Since there were many indicators and procedures conducted by the programs, it was important to
sort the information by areas of greatest leverage potential. This allowed the programs to utilize the information
in manageable units starting with those that had the greatest impact. Programs used this report to plan for and
monitor their usage of ancillary resources such as of laboratory tests, operating room (OR) time and pharmacy
costs. It was essential that the programs monitor the general ledger expense report against the activity credits (in
terms of case weights) that were received through program activity. Total program costing at SMH was
obtained through the hospital decision support/case costing program. This clinical activity database captures costs
related to activities by linking hospital databases to the
financial budgets and then to the individual patient information. The final product is patient level costing for tests,
procedures and care received by a patient while in the
hospital. Patients can then be grouped and their data
summarized by case mix group, activities, physician, program, service and total hospital population. On a patient
level, the expected costs are compared to the actual cost
so that opportunities for improvement of resource use
can be identified.
90
IMPACT OF THE COMPLEXITY METHODOLOGY ON AN ONTARIO TEACHING HOSPITAL
Even though program costing is important, the majority
of costs are driven by physicians. Therefore, an operational report comparing individual physician practice and
costs was produced on a CMG basis with the RIW totals
grouped into programs. This allowed the programs to
monitor the activity of each physician against every other
and against an industry standard. By comparing individual
physician activity adjusted for case mix and case weighting, judgements can be made on the efficiency of care,
utilization of ancillary resources (laboratory, physical therapy, nursing workload, etc.) and quality of care using
specified indicators. Typical and atypical case adjustments
are made for outliers, so that an accurate profile of physician activity and patient flows can be tracked. A review of
the cost, case weight (RIW) and volume variation per
CMG revealed a number of opportunities for the development of clinical pathways to reduce and standardize
resource consumption. Total hip replacements are an
example of how standardization of care using a clinical
pathway resulted in reduced lengths of stay and more
efficient utilization of ancillary services. The creation of
the pathways included the formation of a multidisciplinary team to review existing pathways at other institutions
and create paths specific to SMH based on the indicators
within the variance data. These data compare the eligibility of patient groups for a clinical pathway. The final
report was the program report which facilitates operational program management and program planning as discussed below.
Program Planning and Operations
In conjunction with the standard program reports created
for the patient care journey, corporate level operational
indicators were reported in both a clinical activity report
and on-line using a graphical display of the data. These
reports were available hospital wide with monthly data
updates. The total RIW and average RIW (RIW per discharge) were reviewed on a corporate, program, and service level to ensure that the volume of work and the acuity of patients was maintained in conjunction with financial, quality and other utilization indicators. Since the RIW
is such a powerful indicator, two efficiency ratios, length
of stay per RIW and total cost per RIW, were also monitored. This information allowed the hospital to see how
efficiently a patient was being treated, standardized by the
weighted case assignment. The ratios allowed both management and caregivers to determine the efficiency of
their operations compared to their internal and external
peers. Additionally, a target value for performance compared to a CIHI percentile benchmark was set. With the
implementation of a cost accounting decision support
system at St. Michael’s Hospital, detailed costs were
assigned to patients based on the activities or products
received. With such a powerful system, the actual cost per
patient, physician, service or program could be identified.
By comparing the actual cost for a group of patients
against Ministry Program funds or against expected costs
(hospital average cost per weighted case x the weighted
cases assigned to the patients), the hospital can identify
areas that may cost more to provide than the credited value
assigned by the RIW. Recently, St. Michael’s Hospital has
used this type of analysis to show what is believed to be
inadequate compensation in the areas of Cardiovascular
Surgery, Haemodialysis and Trauma. This is currently being
reviewed by the Ontario Ministry of Health.
Ministry Funding Discussions
The RIW is an important measure in the calculation of
actual cost per weighted case and expected cost per
weighted case differentials. The expected cost per weighted case is calculated by adding premiums to a base rate
depending on the services that are provided. These premiums may be for tertiary, newborn and teaching activity.
For the past five years, SMH has had a positive expected
minus actual cost per weighted case, and in 1995/96 it
was the highest amongst Ontario teaching hospitals. The
cost per case weight measure is important in that it
enables SMH to highlight to the Ministry of Health that
the proportionate reward for high efficiency provided by
reduced funding cuts is minimal compared to the money
saved by the hospital for operational effectiveness. Over
the past four years, the savings achieved by reducing the
cost per weighted case amount to greater than $35 million, while the savings achieved by a reduced funding cut
are in the hundreds of thousands of dollars range.
Therefore, even though SMH operates at a cost millions
of dollars less than expected, the savings achieved by a
smaller reduction in funding are small. The cost per
weighted case indicator allows for change opportunity
discussions with the Ministry of Health.
Restructuring Commission Directions
As SMH moves forward with the implementation of the
Health Services Restructuring Commission’s recommendations, the availability of appropriate tools such as utilization
indicators is required. Of these, the RIW volume can provide information that is required in planning the movement
of services from the Wellesley Hospital to SMH, to create
appropriate program synergies. Patient acuity reviews and
budget planning for patient volumes are made possible
through RIW reviews of the two institutions. The addition
of the complexity adjustment to the CMG methodology
changes the case weight values at both institutions on a service by service basis. This information is important to consider when programs are negotiated and strategic plans are
built between the two hospitals.
The RIW measures are also important to consider when
physicians from both hospitals practice in the new environment. There are different utilization performance lev-
91
IMPACT OF THE COMPLEXITY METHODOLOGY ON AN ONTARIO TEACHING HOSPITAL
els and standards at the two hospitals, and it is important
that new expectations of the new medical staff are set
regarding lengths of stay, case weights and patient volumes. The impacts of program transfers from the
Wellesley hospital on resource consumption can be minimized by monitoring changes in case weights, lengths of
stay and volumes compared to budget.
per case). Calculated fields included summaries and averages of each indicator as well as a variance analysis comparing the old and new complexity adjusted RIW. The
effects of the complexity levels were then examined from a
cost, weighted case and length of stay perspective. For the
overall hospital cases, the total number of weighted cases
remains unchanged with the addition of complexity level
adjustments to the Resource Intensity Weights (see Table
1). This is congruent with the 0.3% change reported by
HMRU (1997) in the Management Practice Atlas for the
TAHSC hospitals who analysed St. Michael’s Hospital
1995/96 data and compared the total weighted cases calculated with and without the complexity assignment.
The RIW indicator has been used extensively at SMH in
the development of a strategic plan, planning for change
with the re-engineering effort and the operations of its
programs. Now that an adjustment for case complexity
has been incorporated into this indicator, the strategic
decisions that were made with the old measures may no
longer be valid. The assumptions of The Patient Care
Journey that led to re-engineering recommendations may
be inaccurate now that the RIW assignment has changed.
Program planners will see different weighted case assignments for the CMG and may experience an increase or
decrease in total weighted cases simply because the
methodology has changed. SMH has reviewed all of its
inpatient activity between the old and new methodologies
to quantify some of these impacts.
However, a number of significant changes were noted
when the various complexity levels were examined in
detail. The breakdown by individual Complexity CMG
showed some groups with significant RIW increases or
decreases. This is of particular interest to the programs
that do high volumes of this work. By reviewing the significantly affected Complexity CMG by level, the areas of
opportunity and threat could be identified. To accurately
identify the impact of complexity on weighted cases, the
full range of data elements needed to be reviewed concurrently. The data were therefore reviewed by
Complexity Level, Complexity CMG and program. This
type of analysis allowed for the identification of areas
that were significantly affected by the change.
IMPACT ANALYSIS OF A
COMPLEXITY ADJUSTMENT
The data for St. Michael’s Hospital inpatients from April
1996 to December 1996 were reviewed. Each patient case
was summarized according to the length of stay, RIW and
cost indicators (i.e. direct, indirect, total, and average cost
Table 1: Comparison of CMG™ 96 and Complexity 97 Methodologies in LOS and RIW™
(Apr. 1996—Dec. 1996 in SMH)
T OT AL LOS
LEVEL
T OT AL RIW™
CASES
ACT UAL
EXPECT ED
VARIANCE
% VARIANCE
CMG™ 96
Plx™ 97
VARIANCE
% VARIANCE
(1)
(2)
(3)
(4) = (3) - (2)
(5) = (4)/(2)
(6)
(7)
(8) = (7) - (6)
(9) = (8)/(6)
1
6,877
33,987
32,871
-1,115
-3.3
11,330
10,702
-627.11
2
1,120
11,342
10,754
-588
-5.2
3,404
3,399
-5.34
-0.2
3
480
6,977
6,123
-854
-12.2
1,746
1,927
181.59
10.4
4
403
13,669
11,467
-2,202
-16.1
3,020
3,525
504.85
16.7
8
1,860
6,889
6,477
-412
-6.0
2,119
2,122
2.34
0.1
9
2,925
7,314
7,846
532
7.3
1,415
1,419
3.33
0.2
T OT AL
13,665
80,178
75,539
-4,639
-5.8
23,035
23,094
59.58
0.3
LEVEL
CMG™ 96
Plx™ 97
VARIANCE
COST
(10) = (6)*$3688
(11) = (7)*$3688
(12) = (11) - (10)
(13)
EXPECTED COST
ACTUAL
*
COST VARIANCE
-5.5
*
CMG™ 96
% VARIANCE
Plx™ 97
% VARIANCE
(14) = (13) - (10)
(15) = (14)/(10)
(16) = (13) - (11)
(17) = (16)/(11)
1
$41,783,246
$39,470,451
($2,312,795)
$32,089,653
($9,693,593)
-23.2
($7,380,798)
-23.0
2
$12,555,539
$12,535,837
($19,703)
$12,184,560
($370,979)
-3.0
($351,276)
-2.9
3
$6,438,504
$7,108,222
$669,717
$6,798,305
$359,801
5.6
($309,916)
-4.6
4
$11,138,508
$13,000,381
$1,861,873
$17,293,822
$6,155,314
55.3
$4,293,441
24.8
8
$7,816,194
$7,824,837
$8,643
$7,810,260
($5,934)
-0.1
($14,577)
-0.2
9
$5,219,807
$5,232,105
$12,298
$5,249,683
$29,876
0.6
$17,579
0.3
TOTAL
$84,951,799
$85,171,832
$220,033
$81,426,285
($3,525,515)
-4.2
($3,745,548)
-4.6
*
Fictitious costs for the sake of example
92
IMPACT OF THE COMPLEXITY METHODOLOGY ON AN ONTARIO TEACHING HOSPITAL
Level Analysis
There are significant variations in the RIW assignments
by Plx level when the Complexity 96 (pilot version) is
considered. The average RIW is reduced only slightly for
the first level complexity and increased more significantly
for Plx levels three and four. Plx levels 2, 8, and 9 remain
about the same (see Figure 1). When considering the
actual volume of cases and total case weights, the variance across complexity levels changed significantly. The
high volume of Plx level 1 cases with a small RIW reduction was balanced by the relatively low volume of level 3
and 4 cases with a large increase in average RIW.
Interestingly, the percentage decrease for Plx level 1 is
much lower than the percentage increase for either Plx
level 3 or 4, but since the volume of level 1 cases is 3.5
times greater than levels 2, 3 and 4 combined, there was a
balancing effect. When we consider the RIW change in
terms of total weighted cases difference, we notice that
the level one cases drop by 627 weighted cases due to the
relatively large volume of cases. The level 3 and 4
increase in case weights of 686 balances the level 1
decrease (see Figure 2).
Figure 1:
Variance in Average
RIW™ by Level
Variance in Average RIW by Level
- between CMG 96 and Complexity 97 Methodologies Resource Intensity Weights
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
1
2
3
4
8
9
Level
Figure 2:
Variance in Total
RIW™ by Level
Variance in Total RIW by Level
- between CMG 96 and Complexity 97 Methodologies Resource Intensity Weights
600
400
200
0
-200
-400
-600
-800
1
2
3
Level
4
8
9
IMPACT OF THE COMPLEXITY METHODOLOGY ON AN ONTARIO TEACHING HOSPITAL
From a cost estimation perspective, the new RIW is a better estimation of costs on Complexity Levels 1 to 4. To
estimate the expected cost using the old and new RIW, it
was assumed that the dollar value cost credit per weighted
cases was equal to the average cost per weighted case for
the hospital. This value multiplied by the weighted case
assignment gives an estimate of the expected cost per
weighted case. The expected versus actual cost comparison
is a good indicator of operational efficiency, since it compares cost and activity against an institutional average. The
variance between expected and average cost was highest
for Plx level 1 cases at 9.6 versus 7.4 million dollars for the
old and new methodologies respectively. Even though the
RIW value has decreased, the actual costs remain much
lower than the expected costs based on the average cost
per weighted case. There is a higher expected cost to treat
the more complex patients in level 4 than in the other levels. For this level the variation between actual versus
expected cost is reduced from 6.1 million to 4.2 million
with the implementation of the complexity methodology.
The ability of the new methodology to more accurately
reflect resource requirements is enhanced by introducing
complexity levels, however, the cost variance—calculated as
expected cost (average cost per weighted case x RIW)
minus actual cost—is still high. This suggests that the level
adjustment may not be large enough at the extremes. The
variation between expected and actual cost is reduced with
the use of the complexity adjusted RIW when individual
levels are examined. A summation of all levels shows a
93
higher variation with the new methodology from a cost
estimation perspective (see Table 1).
CMG Analysis
On a more detailed level, a list of Complexity CMG with
a variance of greater than 10 total case weights was
reviewed. For the patient volume at St. Michael’s
Hospital, the greatest single CMG change as a result of
the complexity adjustment was for Kidney Transplants
(CMG 500). The 283 case weight increase was significant
when compared to the 46 weighted case decrease in the
top CMG with reduced RIW (craniotomy procedures,
CMG 2). There were 17 CMG that had greater than a 10
weighted case reduction versus 12 CMG that had an
increase of greater than 10 weighted cases. Generally, the
case weight reductions occurred in the craniotomy and
open heart procedures, while the areas of increase were
kidney transplant, tracheostomy and leukemia procedures
(see Table 2).
The CMG with a variation of greater than 10 case
weights resulting from the new methodology were examined. Level one had 21 CMG with a decrease in total
weighted cases while there were 4 CMG with an increase.
Level 2 showed 3 CMG with a decrease in total weighted
cases and 2 CMG with an increase. The pattern reverses
for Plx levels 3 and 4. Level 3 had one CMG decrease
and two increase. All seven CMG that fell into level 4 had
increased weighted cases (see Table 3).
Table 2: Comparison of CMG™ 96 and Complexity 97 Methodologies in LOS, RIW™ and Cost
(Apr. 1996—Dec. 1996 in SMH)
CMG™
CMG™
DESCRIPTION
TOTAL LOS
SERVICE
TOTAL RIW™
CASES
ACTUAL
EXPECTED
VARIANCE
% VARIANCE
CMG™ 96
Plx™ 97
VARIANCE
% VARIANCE
(1)
(2)
(3)
(4) = (3) - (2)
(5) = (4)/(2)
(6)
(7)
(8) = (7) - (6)
(9) = (8)/(6)
2 CRANIOTOMY PROCEDURES WITH CC
NS
108
1,868
1,585
-283
-15.2
555.76
510.12
-45.64
-8.2
PACEMAKER IMPLANT EXCEPT FOR AMI,
188 HEART FAILURE OR SHOCK
CARDIAC VALVE PROCEDURES WITH
178 CARDIAC CATH
PACEMAKER IMPLANT FOR AMI, HEART
189 FAILURE OR SHOCK
CVS
135
906
1,146
240
26.5
388.85
352.35
-36.50
-9.4
CVS
110
1,875
1,981
106
5.6
731.62
695.93
-35.69
-4.9
CVS
118
509
549
40
7.9
266.33
232.91
-33.42
-12.5
503 DIALYSIS PROCEDURES
CARDIAC VALVE PROCEDURES WITHOUT
177 CARDIAC CATH
URO
59
877
514
-363
-41.4
205.37
179.87
-25.50
-12.4
CVS
87
943
995
52
5.5
558.24
533.55
-24.69
-4.4
NS
172
1,501
1,525
24
1.6
581.69
557.15
-24.54
-4.2
ORT
42
260
198
-62
-23.8
92.55
69.00
-23.55
-25.4
ORT
38
156
127
-29
-18.8
71.98
54.73
-17.24
-24.0
1 CRANIOTOMY PROCEDURES WITHOUT CC
MAJOR LOWER EXTREMITY PROCEDURES,
370 AGE > 70 WITH CC
MAJOR LOWER EXTREMITY PROCEDURES
372 WITH CC
280 DIGESTIVE SYSTEM MALIGNANCY WITH CC
LYMPHOMA AND CHRONIC LEUKEMIA AGE
731 < 70 WITH CC OR AGE > 70 WITHOUT CC
GAS
46
675
374
-301
-44.5
132.77
115.91
-16.86
-12.7
HAE
47
530
399
-131
-24.8
126.42
109.57
-16.84
-13.3
131 RESPIROLOGY NEOPLASMS
FRACTURED FEMUR PROCEDURES WITH
356 CC
ACUTE MYOCARDIAL INFARCTION
RES
91
1,425
795
-630
-44.2
288.19
271.77
-16.41
-5.7
ORT
48
744
724
-20
-2.7
177.96
161.81
-16.15
-9.1
CAR
26
58
59
1
2.2
51.99
37.82
-14.17
-27.3
TRA
71
413
474
61
14.7
162.66
150.73
-11.94
-7.3
NS
10
53
39
-14
-25.8
24.20
12.42
-11.77
-48.7
EYE
52
75
73
-2
-2.9
35.30
24.44
-10.86
-30.8
WITHOUT CARDIOVASCULAR
194 OTHER
COMPLICATIONS
PROCEDURES FOR DIAGNOSIS OF
TRAUMA AGE < 70 WITH CC OR AGE > 70
804 PERIPHERAL,
W/O CC
CRANITAL NERVE AND
OTHER NEUROLOGICAL PROCEDURES
8 WITH CC
56 LENS INSERTION WITHOUT CC (MNRH)
94
IMPACT OF THE COMPLEXITY METHODOLOGY ON AN ONTARIO TEACHING HOSPITAL
Table 2 (cont’d): Comparison of CMG™ 96 and Complexity 97 Methodologies in LOS,
RIW™ and Cost (Apr. 1996—Dec. 1996 in SMH)
EXPECTED COST
CMG™
CMG™ 96
(10) = (6)*$3688
2
188
178
189
503
177
1
370
372
280
731
131
356
194
804
8
56
Plx™ 97
(11) = (7)*$3688
ACTUAL
VARIANCE
(12) = (11) - (10)
COST
COST VARIANCE
*
(13)
CMG™ 96
% VARIANCE
(14) = (13) - (10)
(15) = (14)/(10)
*
Plx™ 97
(16) = (13) - (11)
% VARIANCE
(17) = (16)/(11)
$2,049,636.5
$1,881,333.2
($168,303.4)
$1,982,300.9
($67,335.6)
-3.3
$100,967.7
5.1
$1,434,068.7
$1,299,466.8
($134,601.9)
$953,846.6
($480,222.1)
-33.5
($345,620.2)
-36.2
$2,698,206.1
$2,566,591.3
($131,614.8)
$2,619,995.7
($78,210.4)
-2.9
$53,404.4
2.0
$982,227.7
$858,987.9
($123,239.8)
$495,481.8
($486,745.9)
-49.6
($363,506.1)
-73.4
$757,409.5
$663,372.7
($94,036.7)
$835,210.8
$77,801.3
10.3
$171,838.0
20.6
$2,058,801.4
$1,967,736.1
($91,065.3)
$2,285,347.5
$226,546.1
11.0
$317,611.5
13.9
$2,145,273.6
$2,054,766.5
($90,507.1)
$1,601,828.7
($543,444.9)
-25.3
($452,937.8)
-28.3
$341,342.8
$254,472.4
($86,870.3)
$266,048.6
($75,294.2)
-22.1
$11,576.1
4.4
$265,448.2
$201,849.4
($63,598.7)
$196,060.2
($69,388.0)
-26.1
($5,789.2)
-3.0
$489,672.0
$427,487.5
($62,184.5)
$435,728.3
($53,943.7)
-11.0
$8,240.8
1.9
$466,232.2
$404,108.6
($62,123.6)
$456,239.7
($9,992.5)
-2.1
$52,131.1
11.4
$1,062,837.0
$1,002,300.0
($60,537.0)
$986,499.9
($76,337.0)
-7.2
($15,800.1)
-1.6
$656,311.2
$596,747.9
($59,563.3)
$639,968.1
($16,343.1)
-2.5
$43,220.2
6.8
$191,730.0
$139,465.3
($52,264.7)
$70,615.1
($121,114.8)
-63.2
($68,850.1)
-97.5
$599,901.6
$555,874.5
($44,027.1)
$495,785.0
($104,116.6)
-17.4
($60,089.5)
-12.1
$89,241.9
$45,818.2
($43,423.6)
$66,582.5
($22,659.4)
-25.4
$20,764.2
31.2
$130,196.7
$90,148.1
($40,048.6)
$101,443.5
($28,753.2)
-22.1
$11,295.4
11.1
*
Generally, as the RIW are increased under the new
methodology, the expected cost variance increased
because the calculation for expected cost is based on the
RIW multiplied by a constant. However, to obtain the
cost variance the new and old expected costs are subtracted from the actual cost. There is no apparent pattern
between levels and the cost variance since the actual costs
are sometimes greater than or less than the expected cost
(see Table 3).
Program Analysis
When the CMG are grouped by program, the total impact
of the weights at each Plx level can be determined. All
hospital programs experience a change in RIW due to
adjustments for complexity at all levels. Each of the pro-
Fictitious costs for the sake of example
grams except Urology/Renal/Endocrine/ENT/
Ophthalmology (UREEO) show a decrease in weighted
cases for level one and an increase for level four. The
UREEO program has an increase in all levels that was
primarily due to the level one increase in the Kidney
Transplant RIW. All programs experienced an increase in
cases at Plx level 2 and level 3 except for the decrease in
Plx level 2 in the heart program. This is primarily because
the significant weighted case reductions in level 2 are for
valve procedures with cardiac catheterization, coronary
bypass without cardiac catheterization and pacemaker
implant for AMI (heart failure or shock). These are all
CMG belonging to the Heart program (see Table 4).
95
IMPACT OF THE COMPLEXITY METHODOLOGY ON AN ONTARIO TEACHING HOSPITAL
Table 3: Comparison of CMG™ and Complexity Methodologies in LOS, RIW™, and Cost
(Apr. 1996—Dec. 1996 in SMH)
Top CMG™ at Level 1 with RIW™ Decrease
CMG™
CMG™
DESCRIPTION
2 CRANIOTOMY PROCEDURES WITH CC
OTHER PROCEDURES FOR DIAGNOSIS OF TRAUMA AGE
804 < 70 WITH CC OR AGE > 70 W/O CC
PACEMAKER IMPLANT EXCEPT FOR AMI, HEART FAILURE
188 OR SHOCK
MAJOR LOWER EXTREMITY PROCEDURES, AGE > 70
370 WITH CC
TOTAL LOS
SERVICE
TOTAL RIW™
CASES
ACTUAL
EXPECTED
VARIANCE
% VARIANCE
CMG™ 96
Plx™ 97
VARIANCE
% VARIANCE
(1)
(2)
(3)
(4) = (3) - (2)
(5) = (4)/(2)
(6)
(7)
(8) = (7) - (6)
(9) = (8)/(6)
NS
56
818
442
-376
-45.9
274.11
229.49
-44.62
TRA
61
230
304
74
32.2
122.25
95.04
-27.21
-16.3
-22.3
CVS
96
569
737
168
29.4
256.79
230.00
-26.79
-10.4
ORT
36
214
143
-71
-33.4
80.86
55.03
-25.83
-31.9
177 CARDIAC VALVE PROCEDURES WITHOUT CARDIAC CATH
CVS
31
206
273
67
32.4
172.15
147.40
-24.76
-14.4
356 FRACTURED FEMUR PROCEDURES WITH CC
PACEMAKER IMPLANT FOR AMI, HEART FAILURE OR
189 SHOCK
ORT
27
268
258
-10
-3.8
80.54
57.93
-22.61
-28.1
CVS
55
191
140
-51
-26.5
117.64
96.83
-20.81
-17.7
503 DIALYSIS PROCEDURES
URO
42
227
131
-96
-42.3
88.48
67.99
-20.49
-23.2
363 BACK AND NECK PROCEDURES WITH FUSION, AGE > 65
NS
21
171
177
6
3.6
79.45
62.35
-17.11
-21.5
253 MAJOR INTESTINAL AND RECTAL PROCEDURES WITH CC
SUR
53
853
491
-362
-42.4
202.76
185.77
-16.99
-8.4
131 RESPIROLOGY NEOPLASMS
SPECIFIC CEREBROVASCULAR DISORDERS EXCEPT
13 TRANSIENT ISCHEMIC ATTACKS
RES
50
668
340
-328
-49.1
144.15
127.36
-16.79
-11.6
-10.3
1 CRANIOTOMY PROCEDURES WITHOUT CC
372 MAJOR LOWER EXTREMITY PROCEDURES WITH CC
LYMPHOMA AND CHRONIC LEUKEMIA AGE < 70 WITH CC
731 OR AGE > 70 WITHOUT CC
PERCUTANEOUS TRANSLUMINAL CORONARY
222 ANGIOPLASTY (PTCA)
125 MAJOR CHEST PROCEDURES
PERIPHERAL, CRANITAL NERVE AND OTHER
8 NEUROLOGICAL PROCEDURES WITH CC
280 DIGESTIVE SYSTEM MALIGNANCY WITH CC
SIMPLE PNEUMONIA AND PLEURISY AGE 18-69 WITH CC
140 OR AGE > 70 WITHOUT CC
ACUTE MYOCARDIAL INFARCTION WITHOUT
194 CARDIOVASCULAR COMPLICATIONS
NS
87
685
757
72
10.5
155.84
139.78
-16.05
NS
165
1,316
1,304
-13
-1.0
525.35
510.30
-15.04
-2.9
ORT
33
111
81
-30
-26.8
56.03
42.31
-13.72
-24.5
HAE
27
183
173
-10
-5.4
53.40
40.41
-12.99
-24.3
CVS
42
263
237
-26
-9.9
59.04
46.05
-12.98
-22.0
RES
37
349
281
-68
-19.4
111.37
98.77
-12.60
-11.3
NS
9
46
29
-17
-37.6
21.61
10.03
-11.58
-53.6
GAS
27
377
155
-222
-58.9
70.87
60.19
-10.68
-15.1
RES
47
289
269
-20
-6.8
59.73
49.54
-10.19
-17.1
CAR
24
35
46
11
30.3
42.97
32.87
-10.10
-23.5
Top CMG™ at Level 1with RIW™ Increase
CMG™
CMG™
DESCRIPTION
TOTAL LOS
SERVICE
CASES
(1)
ACTUAL
(2)
EXPECTED
(3)
TOTAL RIW™
VARIANCE
(4) = (3) - (2)
% VARIANCE
(5) = (4)/(2)
CMG™ 96
(6)
Plx™ 97
(7)
VARIANCE
(8) = (7) - (6)
% VARIANCE
(9) = (8)/(6)
500 KIDNEY TRANSPLANT
OTHER PROCEDURES FOR DIAGNOSIS OF TRAUMA AGE
803 < 70 WITHOUT CC
ACUTE LEUKEMIA WITHOUT LYMPHOMA PROCEDURES
726 WITHOUT CC
URO
26
216
264
48
22.3
150.05
289.34
139.29
92.8
TRA
35
134
150
16
11.6
48.79
60.20
11.41
23.4
HAE
9
126
68
-58
-46.0
32.85
43.37
10.52
32.0
357 FRACTURED FEMUR PROCEDURES WITHOUT CC
ORT
63
483
701
218
45.1
137.25
149.43
12.18
8.9
Top CMG™ at Level 2 with RIW™ Decrease
CMG™
CMG™
DESCRIPTION
TOTAL LOS
SERVICE
CASES
(1)
ACTUAL
(2)
EXPECTED
(3)
TOTAL RIW™
VARIANCE
(4) = (3) - (2)
% VARIANCE
(5) = (4)/(2)
CMG™ 96
(6)
Plx™ 97
(7)
VARIANCE
(8) = (7) - (6)
% VARIANCE
(9) = (8)/(6)
178 CARDIAC VALVE PROCEDURES WITH CARDIAC CATH
CVS
63
910
1,014
104
11.5
369.15
335.65
-33.50
179 CORONARY BYPASS WITHOUT CARDIAC CATH
PACEMAKER IMPLANT FOR AMI, HEART FAILURE OR
189 SHOCK
CVS
237
1,570
2,038
468
29.8
885.63
857.38
-28.26
-9.1
-3.2
CVS
55
253
341
88
34.8
122.14
111.17
-10.97
-9.0
Top CMG™ at Level 2 with RIW™ Increase
CMG™
CMG™
DESCRIPTION
500 KIDNEY TRANSPLANT
SKIN GRAFT & WOUNDED DEBRIDENENT FOR SKIN
427 ULCER OR CELLULITIS
TOTAL LOS
SERVICE
CASES
(1)
ACTUAL
(2)
EXPECTED
(3)
TOTAL RIW™
VARIANCE
(4) = (3) - (2)
% VARIANCE
(5) = (4)/(2)
CMG™ 96
(6)
Plx™ 97
(7)
VARIANCE
(8) = (7) - (6)
% VARIANCE
(9) = (8)/(6)
URO
4
44
46
2
3.9
23.08
43.33
20.24
87.7
PLA
5
57
146
89
156.7
14.27
24.32
10.05
70.4
Top CMG™ at Level 3 with RIW™ Decrease
CMG™
CMG™
DESCRIPTION
ACUTE LEUKEMIA WITHOUT LYMPHOMA PROCEDURES
727 WITH CC
TOTAL LOS
SERVICE
CASES
(1)
HAE
ACTUAL
(2)
9
EXPECTED
(3)
302
TOTAL RIW™
VARIANCE
(4) = (3) - (2)
148
% VARIANCE
(5) = (4)/(2)
-154
CMG™ 96
(6)
-50.9
92.32
Plx™ 97
(7)
VARIANCE
(8) = (7) - (6)
78.49
% VARIANCE
(9) = (8)/(6)
-13.82
-15.0
Top CMG™ at Level 3 with RIW™ Increase
CMG™
CMG™
DESCRIPTION
TOTAL LOS
SERVICE
TOTAL RIW™
CASES
ACTUAL
EXPECTED
VARIANCE
% VARIANCE
CMG™ 96
Plx™ 97
VARIANCE
% VARIANCE
(1)
(2)
(3)
(4) = (3) - (2)
(5) = (4)/(2)
(6)
(7)
(8) = (7) - (6)
(9) = (8)/(6)
500 KIDNEY TRANSPLANT
URO
14
168
220
52
31.0
80.80
164.56
83.76
103.7
179 CORONARY BYPASS WITHOUT CARDIAC CATH
CVS
48
409
470
61
15.0
188.86
200.91
12.06
6.4
Top CMG™ at Level 4 with RIW™ Increase
C MG™
C MG™
D ESC R IPTION
TOTA L LOS
SER VIC E
C A SES
A C TU A L
EXPEC TED
(1)
(2)
(3)
TOTA L R IW™
VA R IA N C E
(4) = (3) - (2)
% VA R IA N C E
C MG™ 96
(5) = (4)/(2)
(6)
Plx™ 97
(7)
VA R IA N C E
(8) = (7) - (6)
% VA R IA N C E
(9) = (8)/(6)
40 TRACHEOSTOMY & GASTORSTOMY PROCEDURES
ACUTE LEUKEMIA WITHOUT LYMPHOMA PROCEDURES
727 WITH CC
SUR
32
2,425
2,579
154
6.4
489.62
632.08
142.47
29.1
HAE
11
337
344
7
2.2
120.63
164.57
43.94
36.4
500 KIDNEY TRANSPLANT
URO
5
74
120
46
62.0
28.17
68.60
40.43
143.5
179 CORONARY BYPASS WITHOUT CARDIAC CATH
CVS
26
502
406
-96
-19.2
163.38
194.37
30.99
19.0
253 MAJOR INTESTINAL AND RECTAL PROCEDURES WITH CC
SUR
14.00
638
309
-329
-51.5
112.86
141.16
28.30
25.1
127 OTHER RESPAROLOGY PROCEDURES WITH CC
RES
7.00
415
356
-59
-14.2
88.24
110.74
22.50
25.5
NS
4.00
175
107
-68
-39.0
36.04
49.36
13.32
36.9
363 BACK AND NECK PROCEDURES WITH FUSION, AGE > 65
96
IMPACT OF THE COMPLEXITY METHODOLOGY ON AN ONTARIO TEACHING HOSPITAL
Table 3 (cont’d): Comparison of CMG™ and Complexity Methodologies in LOS, RIW™, and
Cost (Apr. 1996—Dec. 1996 in SMH)
Top CMG™ at Level 1 with RIW™ Decrease
EXPECTED COST
CMG™
ACTUAL
CMG™
Plx™
VARIANCE
COST
(10) = (6)*$3688
(11) = (7)*$3688
(12) = (11) - (10)
(13)
COST VARIANCE
*
*
CMG™ 96
% VARIANCE
Plx™ 97
% VARIANCE
(14) = (13) - (10)
(15) = (14)/(10)
(16) = (13) - (11)
(17) = (16)/(11)
2
$1,010,900.9
$846,349.7
($164,551.2)
$822,562.1
($188,338.7)
-18.6
($23,787.6)
-2.9
804
$450,844.9
$350,504.6
($100,340.2)
$278,435.9
($172,408.9)
-38.2
($72,068.7)
-25.9
188
$947,050.1
$848,256.5
($98,793.6)
$626,268.0
($320,782.1)
-33.9
($221,988.6)
-35.4
370
$298,210.2
$202,961.1
($95,249.1)
$221,637.9
($76,572.3)
-25.7
$18,676.8
8.4
177
$634,895.4
$543,595.3
($91,300.1)
$634,315.3
($580.1)
-0.1
$90,719.9
14.3
356
$297,028.2
$213,639.3
($83,388.9)
$260,866.5
($36,161.8)
-12.2
$47,227.2
18.1
189
$433,872.7
$357,110.0
($76,762.8)
$187,800.5
($246,072.2)
-56.7
($169,309.4)
-90.2
503
$326,319.4
$250,745.6
($75,573.8)
$188,771.5
($137,547.9)
-42.2
($61,974.1)
-32.8
363
$293,017.5
$229,929.2
($63,088.4)
$209,854.7
($83,162.8)
-28.4
($20,074.5)
-9.6
253
$747,778.8
$685,130.6
($62,648.2)
$650,192.2
($97,586.6)
-13.1
($34,938.5)
-5.4
131
$531,621.5
$469,707.4
($61,914.1)
$419,033.4
($112,588.1)
-21.2
($50,674.0)
-12.1
13
$574,733.3
$515,525.8
($59,207.6)
$448,855.7
($125,877.6)
-21.9
($66,670.1)
-14.9
1
$1,937,472.9
$1,882,000.2
($55,472.7)
$1,350,574.7
($586,898.2)
-30.3
($531,425.5)
-39.3
372
$206,644.9
$156,028.2
($50,616.7)
$154,790.3
($51,854.6)
-25.1
($1,237.9)
-0.8
731
$196,948.4
$149,024.4
($47,923.9)
$126,570.8
($70,377.6)
-35.7
($22,453.6)
-17.7
222
$217,729.6
$169,845.0
($47,884.5)
$185,877.9
($31,851.6)
-14.6
$16,032.9
8.6
125
$410,732.6
$364,266.6
($46,466.0)
$204,985.6
($205,747.0)
-50.1
($159,281.0)
-77.7
8
$79,689.3
$36,983.2
($42,706.2)
$59,920.1
($19,769.2)
-24.8
$22,936.9
38.3
280
$261,364.1
$221,974.8
($39,389.3)
$226,402.8
($34,961.3)
-13.4
$4,428.0
2.0
140
$220,284.1
$182,706.5
($37,577.5)
$190,461.3
($29,822.7)
-13.5
$7,754.8
4.1
194
$158,455.1
$121,213.6
($37,241.4)
$44,233.6
($114,221.5)
-72.1
($76,980.1)
-174.0
Top CMG™ at Level 1with RIW™ Increase
EXPECTED COST
CMG™
ACTUAL
CMG™ 96
Plx™ 97
VARIANCE
COST
(10) = (6)*$3688
(11) = (7)*$3688
(12) = (11) - (10)
(13)
COST VARIANCE
*
*
CMG™ 96
% VARIANCE
Plx™ 97
% VARIANCE
(14) = (13) - (10)
(15) = (14)/(10)
(16) = (13) - (11)
(17) = (16)/(11)
500
$553,379.2
$1,067,080.0
$513,700.8
$327,520.9
($225,858.4)
-40.8
($739,559.2)
803
$179,951.7
$222,030.5
$42,078.8
$154,832.0
($25,119.7)
-14.0
($67,198.5)
-225.8
-43.4
726
$121,161.1
$159,945.6
$38,784.5
$67,576.0
($53,585.1)
-44.2
($92,369.6)
-136.7
357
$506,166.1
$551,103.7
$44,937.7
$398,772.1
($107,394.0)
-21.2
($152,331.6)
-38.2
Top CMG™ at Level 2 with RIW™ Decrease
EXPECTED COST
CMG™
ACTUAL
CMG™
Plx™
VARIANCE
COST
(10) = (6)*$3688
(11) = (7)*$3688
(12) = (11) - (10)
(13)
COST VARIANCE
*
*
CMG™ 96
% VARIANCE
Plx™ 97
% VARIANCE
(14) = (13) - (10)
(15) = (14)/(10)
(16) = (13) - (11)
(17) = (16)/(11)
178
$1,361,419.7
$1,237,859.1
($123,560.5)
$1,208,653.8
($152,765.9)
-11.2
($29,205.3)
-2.4
179
$3,266,207.4
$3,162,002.5
($104,205.0)
$2,873,374.6
($392,832.8)
-12.0
($288,627.8)
-10.0
189
$450,467.1
$410,000.5
($40,466.6)
$249,646.4
($200,820.6)
-44.6
($160,354.0)
-64.2
Top CMG™ at Level 2 with RIW™ Increase
EXPECTED COST
CMG™
CMG™ 96
(10) = (6)*$3688
Plx™ 97
(11) = (7)*$3688
ACTUAL
VARIANCE
(12) = (11) - (10)
COST
(13)
*
COST VARIANCE
CMG™ 96
(14) = (13) - (10)
% VARIANCE
(15) = (14)/(10)
*
Plx™ 97
(16) = (13) - (11)
% VARIANCE
(17) = (16)/(11)
500
$85,135.3
$159,793.7
$74,658.4
$59,588.4
($25,546.8)
-30.0
($100,205.2)
-168.2
427
$52,624.1
$89,686.6
$37,062.6
$70,278.4
$17,654.3
33.5
($19,408.2)
-27.6
Top CMG™ at Level 3 with RIW™ Decrease
EXPECTED COST
CMG™
CMG™ 96
(10) = (6)*$3688
727
Plx™ 97
(11) = (7)*$3688
$340,460.3
$289,479.7
ACTUAL
VARIANCE
(12) = (11) - (10)
($50,980.6)
COST
(13)
COST VARIANCE
*
$271,528.1
CMG™ 96
(14) = (13) - (10)
% VARIANCE
(15) = (14)/(10)
($68,932.2)
*
Plx™ 97
(16) = (13) - (11)
-20.2
% VARIANCE
(17) = (16)/(11)
($17,951.6)
-6.6
Top CMG™ at Level 3 with RIW™ Increase
EXPECTED COST
CMG™
ACTUAL
CMG™ 96
Plx™ 97
VARIANCE
COST
(10) = (6)*$3688
(11) = (7)*$3688
(12) = (11) - (10)
(13)
COST VARIANCE
*
CMG™ 96
(14) = (13) - (10)
% VARIANCE
*
Plx™ 97
(15) = (14)/(10)
(16) = (13) - (11)
% VARIANCE
(17) = (16)/(11)
500
$297,973.4
$606,882.5
$308,909.1
$228,246.5
($69,726.9)
-23.4
($378,636.0)
-165.9
179
$696,500.9
$740,969.4
$44,468.4
$725,608.8
$29,107.8
4.2
($15,360.6)
-2.1
Top CMG™ at Level 4 with RIW™ Increase
EXPEC TED C OST
C MG™
C MG™ 96
(10) = (6)*$3688
Plx™ 97
(11) = (7)*$3688
A C TU A L
VA R IA N C E
(12) = (11) - (10)
C OST
C OST VA R IA N C E
*
(13)
C MG™ 96
(14) = (13) - (10)
% VA R IA N C E
*
Plx™ 97
(15) = (14)/(10)
% VA R IA N C E
(16) = (13) - (11)
(17) = (16)/(11)
40
$1,805,703.8
$2,331,122.8
$525,419.0
$3,273,785.1
$1,468,081.3
81.3
$942,662.3
28.8
727
$444,884.9
$606,933.8
$162,048.9
$336,084.0
($108,800.9)
-24.5
($270,849.8)
-80.6
500
$103,883.6
$252,978.4
$149,094.8
$130,448.3
$26,564.7
25.6
($122,530.1)
-93.9
179
$602,550.6
$716,839.5
$114,288.9
$1,142,299.3
$539,748.7
89.6
$425,459.8
37.2
253
$416,226.2
$520,595.1
$104,368.9
$644,202.2
$227,976.0
54.8
$123,607.1
19.2
127
$325,439.1
$408,406.5
$82,967.5
$843,070.1
$517,631.0
159.1
$434,663.6
51.6
363
$132,927.3
$182,042.6
$49,115.3
$186,457.1
$53,529.8
40.3
$4,414.5
2.4
*
Fictitious costs for the sake of example
IMPACT OF THE COMPLEXITY METHODOLOGY ON AN ONTARIO TEACHING HOSPITAL
97
Table 4: Comparison of CMG™ and Complexity Methodologies in LOS and RIW™
(Apr. 1996—Dec. 1996 in SMH)
Heart Program
AVG LOS
LEVEL
AVG RIW™
CASES
ACT UAL
EXPECT ED
VARIANCE
(1)
(2)
(3)
(4) = (3) - (2)
T OT AL LOS
CMG™ 96
Plx™ 97
VARIANCE
(5)
(6)
(7) = (6) - (5)
1
2,082
3.36
3.93
0.58
1.55
1.45
-0.10
2
541
8.17
9.68
1.51
3.69
3.54
-0.15
3
143
12.01
12.60
0.59
4.46
4.55
0.08
4
120
25.84
28.33
2.49
7.57
8.58
1.01
8
21
6.15
4.23
-1.92
1.41
1.31
-0.09
5.63
6.44
0.81
2.34
2.28
-0.05
9
TOTAL
2,907
Heart Program (cont'd)
T OT AL LOS
LEVEL
T OT AL RIW™
ACT UAL
EXPECT ED
VARIANCE
% VARIANCE
CMG™ 96
Plx™ 97
VARIANCE
% VARIANCE
(8)
(9)
(10) = (9) - (8)
(11) = (10)/(8)
(12)
(13)
(14) = (13) - (12)
(15) = (14)/(12)
1
6,989
8,190
1,201
17.2
3,217
3,014
-203
-6.3
2
4,420
5,237
817
18.5
1,998
1,917
-81
-4.0
3
1,718
1,802
84
4.9
638
650
12
1.9
4
3,101
3,400
299
9.6
909
1,030
121
13.4
8
129
89
-40
-31.2
30
28
-2
-6.7
2,362
14.4
6,791
6,639
-152
-2.2
9
TOTAL
16,357
18,719
EXPECTED COST
LEVEL
ACTUAL
CMG™ 96
Plx™ 97
VARIANCE
COST
(16) = (12)*$3688
(17) = (13)*$3688
(18) = (17) - (16)
(19)
COST VARIANCE
*
*
CMG™ 96
% VARIANCE
Plx™ 97
% VARIANCE
(20) = (19) - (16)
(21) = (20)/(16)
(22) = (19) - (17)
(23) = (22)/(17)
1
$11,865,456
$11,116,811
-$748,646
$7,634,478
-$4,230,978
-35.7
-$3,482,333
-45.6
2
$7,368,104
$7,070,818
-$297,286
$6,729,484
-$638,620
-8.7
-$341,334
-5.1
3
$2,352,818
$2,397,224
$44,406
$2,230,442
-$122,376
-5.2
-$166,782
-7.5
4
$3,351,551
$3,799,245
$447,694
$5,557,728
$2,206,177
65.8
$1,758,483
31.6
8
$108,916
$101,638
-$7,278
$93,395
-$15,521
-14.3
-$8,243
-8.8
9
TOTAL
$25,046,845
$24,485,735
-$561,110
$22,245,527
-$2,801,318
-$2,240,208
-11.2
*
DISCUSSION
With the help of more precise information, healthcare
providers are able to more accurately plan treatment
strategies that are in line with internal strategies and comparable to external benchmarks. The addition of an overlay to the CMG methodology improves the accuracy in
estimating resource requirements within a CMG.
However, there still remain a number of limitations in the
current system and, therefore, opportunities for improving resource estimation.
Advantages
The old CMG methodology grouped dissimilar patients
because it failed to recognize the secondary diagnoses that
are often present. When we consider the existence of certain co-morbidities, the length of stay rises and the costs
and resources required to treat these secondary diagnoses
also increase. This is the logic underlying the complexity
overlay adjustment. The resulting detail reflected in the
-10.1
Fictitious costs for the sake of example
RIW measure reduces variation across levels and adds
credibility to this important measure. The increased
resource load of patients that require more care for conditions related or unrelated to their primary diagnosis is
reflected in the complexity adjustment. The new
Complexity methodology adds more consistency across
all CMG when complexity is considered. Previously, only
some CMG were separated based on the presence or
absence of a co-morbidity related to the diagnosis. Now
that the complexity methodology is applied, most CMG
have some type of complexity adjustment that can be
from, within, or outside the major clinical category.
With such increased accuracy, the RIW value better calibrates the patient population across hospitals, programs,
services and physicians. More detailed comparisons at the
patient information level allow physicians and programs
to determine their performance against a predetermined
index of RIW per Discharge or LOS per RIW. The ability
98
IMPACT OF THE COMPLEXITY METHODOLOGY ON AN ONTARIO TEACHING HOSPITAL
to discount the data by arguing that it serves a special
population or that some patients are sicker is reduced.
Interestingly, the data show that, in some cases, the RIW
is reduced to better reflect the relative simplicity of
uncomplicated cases. This reduction is balanced by an
increase in resource weighting at the higher levels. Since
teaching hospitals receive cases with higher complexity,
on average, this methodology provides more of an incentive to treat highly complex cases. Based on the data
review, Complexity has no significant impact on total
weighted case volume at St. Michael’s Hospital. However,
opportunities have arisen for individual programs to better align their resources to their caseload and to emphasize specific volumes of cases based on the credits provided by the new RIW. The increased awareness of this
change in RIW has provided information on the weighted
case impacts to different hospitals. This allows hospitals
to turn to their peers to compare performance levels and
identify areas of opportunity.
Disadvantages
Even though the adjustments for case complexity provide
more accurate information about resource requirement, a
number of problems and limitations still exist around the
assignment of case weights to Plx groups in a way that is
an accurate reflection of the resources required to treat
these cases. There are limitations to the methodology
used by CIHI in the area of age, database variability and
an imbalance in the weights across levels.
The basic concept of the RIW assignment is to determine the most appropriate resource weighting for a highly
variable patient population. Unfortunately, the calculation
of RIW is not simply a matter of identifying a ratio or a
general equation. It is a complex interrelationship
between database derived constants and coefficients
aligned with database target lengths of stay and some
American based costing. If the ELOS versus age and
RIW versus age graphs are compared, a similar distribution pattern results across the four complexity levels
because the same variables are used to calculate the RIW
(see Figures 3 and 4). These lines of best fit are not as
smooth as expected when the database is examined in
detail. For instance, in CMG 179, the RIW and ELOS for
a 42 year old and a 46 year old patient are somewhat similar. However, the RIW and ELOS for a 44 year old is
greater. On average, the RIW and ELOS for the 44 year
old should be greater than or equal to the 42 year old and
less than or equal to the 46 year old. However, since the
numbers are drawn from a database set with inherent statistical variability, areas exist where the data are limited.
This causes variation in case weight assignment that could
be reduced if there was a greater data set to draw from or
smoothed data.
In most cases, there is an improved estimate of costs or
resource requirements for the various levels of complexity. The benefit of more accurate weightings is an
improved estimate of expected costs for each patient.
Expected costs can be compared to the actual costs
incurred to determine the financial "benefit" of providing
a service. When the actual cost is greater than the expected cost, it identifies an opportunity for cost containment.
If the actual cost is less than the expected cost, this is of
benefit to the institution because the program is providing a service at an expense level that is lower than the
hospital’s average. The program should be encouraged to
continue and should be examined to provide ideas for
other programs.
Figure 3:
Relationship Between
ELOS and Age
Relationship Between ELOS and Age
12
ELOS (days)
10
8
6
4
2
0
30
40
50
60
70
Age (years)
179
188
353
428
80
IMPACT OF THE COMPLEXITY METHODOLOGY ON AN ONTARIO TEACHING HOSPITAL
99
Figure 4:
Relationship Between
RIW™ and Age
Relationship Between RIW and Age
6
Case Weights
5
4
3
2
1
0
30
40
50
60
70
80
Age (years)
179
188
353
428
If we consider the relationship between cost and RIW for
the different complexity levels, we find that there are
three distinct relationships instead of the defined four
levels. The level two and three patients share very similar
cost profiles. This suggests that there is a requirement for
only 3 levels of complexity and the possibility to have
equations to calculate the expected cost (see Figure 5).
From these 3 equations, a cost per weighted case can be
obtained as a standard indicator of performance. Since
weighted cases are based on costs adjusted for LOS and
age, the internal total cost per RIW provides a relative
indication of performance on all of these levels and can
be estimated by the graph.
Figure 5:
Relationship Between
Cost and RIW™
Relationship Between Cost and RIW
Cost (dollars)
$50,000
$40,000
$30,000
$20,000
$10,000
$0
1
2
3
4
5
6
7
8
9
10
11
12
Case Weights
1
2
3
4
As more changes occur in the way hospitals receive credit
for activity, there will be an inclination to respond with
strategies that emphasize the activities that have had
increases in case weights and de-emphasize those that
have not. George Pink dismisses the value of this "gaming" response, since the incentive changes are frequent
and the patient population is variable (HMRU, 1996).
However, this is an issue worth considering because, in
these times of resource constraint, there are hospitals
fighting to maintain volumes while reducing costs. The
implementation of the complexity adjustment should in
fact help to solve this type of problem. Before the complexity implementation, the incentive was to treat the
"straight forward" cases or those that had only one diagnosis and could be treated and released. Other than the
adjustment for outliers, there was no credit for the
100
IMPACT OF THE COMPLEXITY METHODOLOGY ON AN ONTARIO TEACHING HOSPITAL
resources spent treating the complex cases with multiple
complications. Since many complex cases are usually sent
to teaching hospitals, it is an advantage for St. Michael’s
Hospital to have such an adjustment in place. It will be
interesting to note how transfers from other institutions
change as a result of the complexity adjustment.
Previously, there was an incentive to have complex and
highly resource intensive patients transferred from a community hospital to a teaching hospital. This could allow a
less complex patient to take their position and provide
the community hospital with the same RIW assignment as
the teaching hospital. The community hospital might use
fewer resources to treat the less complex cases.
Since the adjustment impacts can be reviewed on a CMG
basis, there are incentives and disincentives that cannot be
overlooked at this level. A reduction of almost 70 weighted
cases for 280 craniotomy procedures is a significant
decrease that seriously affects the needs of the program
providing this service (see Table 2). The 363 open heart
cases in CMG 188, 178 and 189 are reduced by 106
weighted cases which also seriously affects the performance of the Heart Program compared to the other programs competing for resources. Equally important are the
significant increases in case weight adjustments. Kidney
transplants and tracheostomy/gastrostomy procedures
have their weighted cases increased by 284 and 151 respectively. Such a large increase in weighted cases will also significantly affect the planning of resource allocation.
The greater weighting assigned to the level 4 complexity
cases is important since it is a better reflection of the
resources these patients consume. However, with such
large increases in case weighting, the combined funding
adjustment that is made for tertiary cases must be considered. The percentage tertiary activity based on specifically
identified CMG plays a large part in the expected cost per
weighted case adjustment. If a hospital conducts a high
volume of a level 4 complexity of a CMG that is identified as tertiary, the result may be a significant increase in
the RIW allocation.
One final limitation to consider is that planning and
developments continue to escalate around the inpatient
case weight assignment, but the outpatient component is
completely missed. The outpatient population is a very
large component of a total hospital’s resource allocation.
As patients are shifted away from inpatient procedures
and outpatient volumes rise, a standardized resource
tracking and acuity assignment tool is required. We use
the RIW indicator in many of our operational decisions
but we are limited in the information that is available
around clinic volumes and their patient types. Before the
full story can be determined, resource and acuity assignments for clinic patients are required.
Operational Changes Based on the New
Complexity Adjustment
The new complexity adjustment is an important step
towards improved accuracy and validity in determining
the allocation of weighted cases to the cases of Ontario
Hospitals. A number of steps should be taken by hospitals in response to this change including education programs, charting programs or audits, information reporting
and impact analyses. These responses are essential to the
optimal operation of a practice, program or hospital.
As funding methodologies and performance monitoring
changes, it is important that individuals who play a key
role in the assignment of the RIW understand their new
responsibilities. Since the complexity adjustment is dependent upon the identification and clear documentation of
all secondary diagnoses related to the case, it is imperative
that this be a recognized and agreed upon role. Attached
to this clarified role should be responsibility and accountability mechanisms. Historically, there has been a gap in
comprehension of the coding system between physicians
and hospital medical records or administrative staff. It is
important for the care providers to be familiar with all
that is required for proper documentation. This will allow
the institution to be properly credited for the resources
utilized through accurate case weight assignment. In-service education on the impacts of improper charting is
essential for all persons working with the chart. Similarly,
the medical records personnel responsible for coding and
abstracting must be sensitive to the secondary diagnoses
and capture them to their fullest potential. Missed secondary diagnoses cause an inappropriate reduction in the
actual case weights performed by the hospital and in turn
make the operations look less efficient when compared to
its peers through cost per weighted case analysis.
With such a heavy reliance on accurate documentation
and coding, proper monitoring and information sharing is
required. Patient charts must be monitored for accuracy
while the patient is still in the hospital. Such concurrent
chart reviews will allow for a complete recording of
patient diagnoses and resource consumption. This role
would be best assumed by case managers, part-time
salaried physicians or even medical records staff.
Most medical records departments currently conduct
audits on their coding and abstracting procedures. Due to
the relative importance of accurate coding for all patients
based on the weighted case usage in funding adjustments,
more frequent and rigorous reviews are required. Random
sampling should occur frequently by coders, by patient
service, or by program. Even though standards exist,
much of the coding process remains subjective.
Minimized variation within and across coders should be
the goal of audits and testing.
IMPACT OF THE COMPLEXITY METHODOLOGY ON AN ONTARIO TEACHING HOSPITAL
Since there is variation in the coding practice within individual hospitals, it is likely that the variation gets larger as
we compare the institutions against one another. The
RIW measure is a provincially accepted indicator used for
funding allocation and management decision support.
Consistent coding by all institutions is a factor that, if not
properly maintained, could threaten the validity of this
provincial indicator. To maximize the power of the
resource indicator, comparative reviews of institutional
charting should be strengthened and standardized. It is
important that institutions are compared on a level playing field.
The most important factor in adjusting to the new RIW is
timely, accurate and helpful information. It is important
for the care and management teams of the hospital to
know the impacts of the complexity changes for proper
management of resources. Comparative CIHI reports
allow for benchmarking the operations of other hospitals
as well as having better detail provided around internal
performance. It is equally important for a hospital to be
aware of the key financial and utilization indicators as
well as those of quality and outcomes. The integration of
quality information is useful for effective and complete
decision making. St. Michael’s Hospital is privileged to
have a cost accounting system that can provide easily
accessible information to detail the operating costs ranging from the patient to the corporate level. This complete
information picture is essential for informed decision
making so that changes such as the complexity adjustment or other system wide shifts in policy can be planned
for using appropriate opportunity and threat analysis.
In summary, all hospital programs are affected by the
implementation of the complexity adjustment.
Consequently, strategic planning, re-engineering recommendations and program operations will need to be
revised. When a new hospital strategic plan is developed,
the new expected and actual costs (calculated using the
new RIW ) will need to be compared to identify strengths
and weaknesses. Quality, clinical resources, financial and
process opportunities have changed with the new complexity levels. The impact of this change will need to be
monitored from this perspective. Program planning, statistical utilization evaluation and funding will all need to
be evaluated to determine how practice should change to
reflect the complexity change impacts.
101
102
IMPACT OF THE COMPLEXITY METHODOLOGY ON AN ONTARIO TEACHING HOSPITAL
REFERENCES
Hospital Management Research Unit (1996). Responsive
Global Funding of Ontario Hospitals: Notes on the
Implications for the Health Care System and Hospital
Manager, HMRU Technical Report 96-1. Toronto,
Ontario: Author.
Hospital Management Research Unit (1997). Management
Practice Atlas for the Toronto Academic Health Science Council.
Version 1-July 1997. Toronto, Ontario: Author.
Kaplan, R. & Norton, D. (1992). The Balanced
Scorecard—Measures That Drive Performance. Harvard
Business Review. January-February, 71-79.
About the Authors
Robert Fox, MBA is Director, Quality and Clinical Resources, St.
Michael’s Hospital, Toronto, Ontario;
Jianli Li, MSc, PhD is Utilization Specialist, St. Michael’s Hospital; and
Robert Bear, MD, FRCP(C), FACP was Executive Vice-President, St.
Michael’s Hospital and is now Executive Vice-President, Capital Health
Authority, Edmonton, Alberta.
C H A P T E R
9
JULIE RICHARDS, GREG FALLON, KAREN HORNE, MARGARET CATT
Maintenance of Case Mix
Tools: The CIHI Revision
Process for MCC 25 (Trauma)
CHAPTER OVERVIEW
The revision project addressing the Major Clinical Category Multiple Significant Trauma
(MCC 25) was a result of interest expressed across Canada to change the way cases were
identified as trauma and to improve the percentage of "true" trauma cases assigned to
MCC 25. The trauma field and the identification of trauma patients was not well-defined.
Consequently, the variation in severity and cost of trauma was also not well-defined. This
broad mix of patients decreased the utility of CMG and RIW as management tools for
trauma programs. Most specifically, it was felt that the heterogeneous mix of cases resulted
in RIW that systematically underestimated the resources required to treat patients seen in
trauma units. In jurisdictions where RIW values were used to allocate resources, institutions
treating a disproportionately large number of ‘severe’ trauma cases were perceived to be
negatively affected. In response, CIHI established a Trauma Task Force to review and
redesign MCC 25 in 1994.
The Task Force provided valuable input on the identification of indicators for significant
trauma patients. However, further analysis and redesign of the MCC was limited because
data to analyze the relationship between the indicators and costs was not available. Then,
Ontario Case Cost information on trauma patients from the Sunnybrook Health Science
Centre, available January 1996, provided an opportunity for further development of trauma
patient classification methodology.
This chapter describes the MCC Multiple Significant Trauma, the revision process, and its
outcome. The chapter closes by describing the revised MCC 25 and demonstrating that
there is a considerable improvement in the percentage of "true" trauma cases now assigned
to MCC 25.
BACKGROUND
Enhancing the CMG methodology has been an ongoing process and the primary focus of
both the CMG Team and the Physician Advisory Committee. The CMG Team was composed of CIHI staff representing statistical, coding/data quality and information system
expertise and the CIHI Medical Consultant. The work of the CMG Team was reviewed
and approved by the Physician Advisory Committee—physicians from across Canada with
a range of medical and surgical specialties. This advisory committee’s mandate was to
ensure that CIHI products continued to reflect the requirements and practice patterns of
Canadian health care.
104
MAINTENANCE OF CASE MIX TOOLS: THE CIHI REVISION PROCESS FOR MCC 25 (TRAUMA)
CIHI also relied on client input and feedback to focus on
where revisions and enhancements should occur. The
revision project that addressed the Major Clinical
Category Multiple Significant Trauma (MCC 25) was initiated as a result of interest expressed across Canada to
change the way cases were identified as trauma and
improve the percentage of "true" trauma cases assigned
to MCC 25. In response, CIHI established a Trauma
Task Force to review and redesign MCC 25.
cases that exhibited multiple significant trauma were
assigned to MCC 25. A case assigned to MCC 25 had to
have a MRDx from one trauma site, and at least one
other significant diagnosis from another trauma site. The
following were possible trauma sites:
• Head;
• Chest;
• Abdomen;
• Kidney;
• Genitourinary;
• Pelvis, and Spine;
• Upper Limb; and
• Lower Limb.
MCC 25—Multiple Significant Trauma, the revision
process and its outcome are described in this chapter. The
revision process began by determining whether there was
a standard definition of significant trauma. It was hoped
that a standard definition would identify diagnoses and
procedures that may be used as indicators of significant
trauma. A literature review and the work of the Trauma
Task Force was used to develop a list of indicators for
further analysis. The statistical analyses performed by
CIHI and the clinical investigations of the Trauma Task
Force which resulted in a revised MCC 25 are then
described. The chapter is concluded with a description of
the revised MCC 25 and demonstration that there was
considerable improvement in the percentage of "true"
trauma cases assigned to MCC 25.
The method of assignment to MCC 25 is presented by
Figure 1 and an example of assignment to a CMG within
MCC 25 presented by Figure 2. The methodology
scanned all operative procedures to identify any matches
within the surgical list for MCC 25. If a match was
found, the case was assigned to the surgical partition
based on the hierarchy of procedures. If no match was
found within MCC 25, then the case was assigned to
CMG 900-906, Unrelated O.R. Procedures. Where no
operative procedures were found, the case was assigned
to the medical partition based on the MRDx.
DESCRIPTION OF MCC 25
MULTIPLE SIGNIFICANT TRAUMA
The trauma CMG were based on the original American
DRG system and had not been altered since 1990. Only
Yes
Age < 29
Days
MCC 15
No
Yes
HIV Diagnosis
MCC 24
No
Trauma
Diagnosis from
2 Body Sites
Yes
MCC 25
No
Assign MCC
Based on Most
Responsible
Diagnosis
Yes
O.R.
Procedure
No
Medical
Partition
Yes
Procedure Found
in MCC
No
Unrelated O.R.
Procedure
Surgical
Partition
Figure 1: CMG 1995
Grouper Methodology
Flowchart
MAINTENANCE OF CASE MIX TOOLS: THE CIHI REVISION PROCESS FOR MCC 25 (TRAUMA)
Surgical Partition
CMG
Craniotomy
875
OR
Procedure
105
Figure 2: Example of the
Grouping Logic for MCC
25 – Multiple Significant
Trauma
Limb Reattachment Hip and Femur Procedures
876
Other Trauma Procedures
877
Inrelated O.R. Procedures
1
Head, Chest or Lower Limb
878
Trauma Diagnosis
All Other
879
Medical Partition
This method of classifying trauma cases was problematic
for two reasons. First, it was possible that a ‘severe’ trauma case could have injuries from only one site (i.e. head
trauma). In this situation, the case was not assigned to
MCC 25, but fell into another MCC, based upon the
MRDx. This ‘severe trauma’ case was then included in a
CMG with other less severe, non-trauma patients. As a
result, only approximately 40% of 'true' trauma patients
were being assigned to MCC 25. Second, the range of
injuries across sites was quite broad. The trauma field
and the identification of trauma patients was, and still is,
not well-defined. Consequently, the variation in severity
and cost of trauma was also not well-defined. In other
words, MCC 25 had less homogeneity with respect to
clinical conditions and resource use than it should have.
LITERATURE REVIEW OF THE
DEFINITION OF A TRAUMA PATIENT
This broad mix of patients decreased the utility of CMG
and RIW as management tools for trauma programs.
Most specifically, it was felt that the heterogeneous mix of
cases resulted in RIW that systematically underestimated
the resources required to treat patients seen in trauma
units. In jurisdictions where RIW values were used to
allocate resources, particular institutions treating a disproportionately large number of ‘severe’ trauma cases were
perceived to be negatively affected.
The classification systems and scoring systems used to
identify trauma patients included the International
Classification of Diseases 9th Revision, the Abbreviated
Injury Scale (AIS), the Injury Severity Score (ISS), the
Revised Trauma Score (RTS), the Trauma Score Injury
Severity Score (TRISS) and A Severity Characterization of
Trauma (ASCOT). Two common scoring systems
referred to in the literature and the two systems that were
later considered during the revision of the trauma MCC
25 were the Abbreviated Injury Scale (AIS), and the
Injury Severity Score (ISS).
A review of the literature was conducted to explore
whether classification systems or other means of categorizing trauma patients were available to provide possible
indicators for a revised trauma MCC. The literature
revealed that there were several methods to categorize
injuries and outcomes, allowing interventions to be evaluated and compared (Baxt & Valda, 1990, Bazzoli et al,
1995, Osler, 1993 and Offner et al, 1992). One researcher
suggested "an ideal scoring system should perform two
separate functions. First, it should allow accurate description of clinical injuries in a manner that reflects the
vocabulary of clinicians. Second, a scoring system should
define severity indices for each injury" (Osler, 1993).
106
MAINTENANCE OF CASE MIX TOOLS: THE CIHI REVISION PROCESS FOR MCC 25 (TRAUMA)
The Abbreviated Injury Scale (AIS)
The AIS was originally developed for impact
injury assessment, and has been expanded to
facilitate the coding of penetrating trauma. The
AIS was intended to provide researchers with a
simple numerical method for ranking and comparing injuries by severity, and to standardize the
terminology used to describe injuries.
Conversion tables have been developed which
enable translation of ICD-9-CM coded discharge diagnoses into AIS body regions and
severity codes. The AIS was the foundation for
the Injury Severity Score.
when disability is the outcome measure. A gunshot
wound to the femur might be just the reverse, infrequently resulting in death, but predictably causing prolonged
disability" (Osler, 1993).
The Injury Severity Score (ISS)
The ISS was the sum of the squares of the
highest AIS code in each of the three most
severely injured ISS body regions. ISS values
ranged from 1 to 75, where the increase in ISS
indexed more ‘severe’ trauma. The ISS provided
a much better fit between overall severity, and
probability of survival than AIS alone.
The findings in this literature review were the basis for
initial discussions at CIHI regarding the identification of
possible indicators for a revised trauma MCC.
Although there were systems to classify trauma patients,
there continued to be problems with categorizing injuries
and identifying comparable injuries and their outcomes.
According to Offner, Jurkovich, Gurney and Rivara
(1992), predictors of survival included systolic blood
pressure, best motor response, Injury Severity Score, and
age. It was considered possible to utilize these predictors
of survival as markers to define the trauma patients.
However, using intubated patients was limited as a predictor of survival or as a marker for trauma patients. Patients
intubated for different reasons may had have vastly different prognoses. "Intubated patients are not a homogenous
group. Indications for intubation of trauma patients are
variable, including head injury, combativeness, airway protection and respiratory arrest" (Offner, Jurkovich, Gurney
and Rivara, 1992). Trauma patients who required intubation may recover completely from their injuries within a
few weeks or conversely may never return to their premorbid state of health.
Another reason it was so difficult to categorize trauma
patients was the enormous number of possible injuries.
"The number of possible injuries a patient may sustain in
a motor vehicle accidents is very large. Injury description
is the process of subdividing the continuous landscape of
human injury into individual, well-defined injuries" (Osler,
1993). Osler reported "the severity for an injury may vary
with the outcome that is being measured. Thus, a gunshot wound to the aorta may have a high severity when
mortality is the outcome measure, but a low severity
Finally, the conclusion of a study undertaken in California
by Baxt and Valda (1990) to correlate Injury Severity
Scoring and Resource Needs demonstrated that "there is
a significant mismatch between the ISS and patient
resource requirements" (Baxt and Valda, 1990). This suggested that further investigation was necessary before
considering ISS as an indicator of resource use in MCC
25.
THE TRAUMA TASK FORCE
In response to the concern that MCC 25 was not representative of 'true' trauma patients, CIHI convened a
Trauma Task Force in 1994 (Appendix I). The objective
of the task force was to review and redesign the existing
MCC 25 into more appropriate groups to better reflect
Canadian patterns of practice.
A considerable amount of time during the initial task
force meetings was spent trying to define the scope of a
Trauma MCC. Two definitions were discussed:
² an event resulting from a transfer of energy or force;
this excluded poisoning and drowning, but included
hangings ; and
² patients having an ISS score indicating a severe trauma, who are resource intense and who represent a
large portion of transfers. (Trauma Task Force
Meeting Minutes 1994)
Current ISS software required ICD-9-CM diagnosis
codes. This was problematic considering the large number
of CIHI client hospitals using ICD-9 codes. As well, the
value of the ISS score used to identify significant trauma
patients was not consistent in Canada. Therefore, implementing the collection of ISS scores on the discharge
abstract may be difficult. It was recommended that ISS
and AIS scores be evaluated for their ability to predict
trauma cost and then be investigated for their future use
with ICD-10. ICD-10, the latest revision of the
International Statistical Classification of Diagnosis and
Related Heath Problems, will be implemented in Canada
by 2001.
MAINTENANCE OF CASE MIX TOOLS: THE CIHI REVISION PROCESS FOR MCC 25 (TRAUMA)
The Ontario Trauma Registry, Trauma Patient Definition
defined trauma according to the International
Classification of Disease E codes (External Cause of
107
Injury codes). The E codes recommended for inclusion in
the trauma definition for the Minimal Data Set of the
Ontario Trauma Registry were:
E Codes
Description
E800-807
E810-819
E820-825
E826-829
E830-838
E840-845
E846-848
E880-888
E890-899
E900-902, 906-909
E910, 913-915
E916-928
E950-952
E953-958
E960-961 963-968
E970-976, 978
E983-988
E990-998
railway accidents
motor vehicle traffic accidents
motor vehicle non-traffic accidents
other road vehicle accidents
water transport accidents
air and space transport accidents
vehicle accidents not elsewhere classifiable
accidental falls
accidents caused by fire and flames
accidents due to natural and environmental factors
accidents caused by submersion, suffocation, and foreign bodies
other accidents
suicide & self inflicted injury (poisonings)
suicide and self-inflicted injury
homicide and injury purposely inflicted by other persons
legal intervention
injury undetermined whether accidentally or purposely inflicted
injury resulting from operations of war
(Ontario Trauma
Registry,
1992)1992)
(Ontario
Trauma
Registry,
Identification of Potential
Procedural Indicators
The Task Force identified the types of patients and variables that may be useful in identifying most severe trauma
cases. It was from these variables that various models
were developed to predict total cost. Examples of
patients which should be identified included:
² head injury patients that do not regain consciousness;
² high spinal cord injury patients with incomplete
mobility;
² lower limb fracture in elderly patients with co-morbidities (cardiac diseases, COPD, arthritis);
² unstable pelvic fracture;
² soft tissue defects (crushing injury);
² patients with prolonged stay in ICU; and
² multiple organ failure.
The Task Force then proceeded to discuss possible means
of identifying resource-intense trauma patients. It was
agreed that the following variables may be useful in identifying the most resource intense trauma cases:
1. ventilation for greater than 96 hours;
2. temporary tracheostomy and gastrostomy procedures;
3. social economic factors (use V-code diagnoses);
4. age (significant in trauma patients expected LOS);
5. co-morbid and complicating diagnoses; and
6. patients with a high ISS score.
It was also recommended that entry into the MCC should
not require multiple body system diagnoses but rather
one significant trauma diagnosis. Burn patients were also
recommended to be included in the Trauma MCC, as
these patients consume a considerable amount of
resources.
It was suggested that E-codes be further investigated for
their use as trauma indicators. However, it was found
that E-codes did not give a good picture of trauma.
There were a number of cases with an external cause of
injury but without an injury diagnosis.
The Task Force recommended ICU days not be used as a
marker because of the variability of the collection of this
data. There was also a concern that patients with multiple fractures may not have any ICU days, while other less
extensive cases may be admitted to the ICU.
108
MAINTENANCE OF CASE MIX TOOLS: THE CIHI REVISION PROCESS FOR MCC 25 (TRAUMA)
The work of the Task Force provided valuable input on
the identification of indicators for significant trauma
patients. However, further analysis and redesign of the
MCC was limited because data to analyze the relationship
of the indicators to costs were not available. The final
recommendations of the Task Force were:
² the trauma MCC should include all significant trauma
cases not just those with multiple significant trauma;
² the scope of the trauma MCC should include all trauma cases including burns; and
² the following severity markers should be investigated
as predictors of resource use: ISS score, ventilation
> 96 hours, co-morbid and complicating diagnoses
(e.g. multiple organ failure) and procedure markers
such as gastrostomy and tracheostomy.
DEVELOPMENT OF MCC 25
SIGNIFICANT TRAUMA
The participation of the Sunnybrook Health Science
Centre (SHSC) in the Ontario Case Cost Project (OCCP)
enabled the capture of patient-specific costs. The availability of this cost information in January 1996 provided
an opportunity for further development of trauma patient
classification methodology.
The Sunnybrook Health Science Centre is a fully affiliated
teaching hospital of the University of Toronto. The trauma program operates as a regional trauma centre and represents a strategic focus for the organization.
Sunnybrook was one of the original volunteer sites selected to participate in the Ontario Case Costing Project, a
joint initiative of the Ontario Ministry of Health and the
Ontario Hospital Association. This enabled Sunnybrook
to collect patient-specific costs using a standardized
methodology established and validated by the OCCP.
When Sunnybrook cost data became available, Dr. Barry
McLellan, Chief of the Sunnybrook Trauma Program
and past member of the CIHI Trauma Task Force, supported the use of this data to answer some of the questions left by the Task Force. This data proved to be the
key ingredient in completing the revision process at CIHI.
Analysis of administrative data—including patient-specific procedural, diagnostic, length of stay and cost information for patients admitted to the Regional Trauma Unit—
contributed to an understanding of the statistical relationship between total patient care costs, current CMG and
other indicators that may index higher costs. The objectives of analyzing this data were to identify indicators predicting total patient cost and length of stay, and to rank
these indicators. These indicators could then be incorporated into a new method of assigning cases to MCC 25
and help develop new case mix groups for trauma.
The first step in the development of a grouping methodology began with the identification of diagnoses which,
when determined to be the MRDx, resulted in the assignment of a case to one of 25 MCC. The list of MRDx
for assignment to MCC 25 was developed by a sub-group
of the Trauma Task Force, discussed in the section
Revised MCC 25 Significant Trauma.
The assignment of a case to a CMG within the surgical
partition was determined by the presence of an operative
procedure. A surgical hierarchy ordered the CMG in the
surgical partition from most, to least, resource-intense. If
there was more than one procedure recorded for the case,
then it was assigned to the CMG highest on the hierarchy.
The analyses that follow were designed to confirm which
of the procedures identified by the Task Force were significant indicators of resource use and to determine the
surgical hierarchy of these procedures.
If there were no procedures used for CMG assignment
the case was assigned to the medical partition of the
MCC. The medical partition consisted of groupings of
similar diagnoses defined clinically and/or by homogeneity of resource use.
The best measure for resource use is patient-specific cost
data. However, this information is not commonly collected by health care facilities in Canada. In the absence of
patient-specific cost data, LOS routinely collected on the
CIHI abstract is used instead. The analysis of the
Sunnybrook trauma cost data assessed the validity of
using LOS as a proxy for total cost.
The analysis of the Sunnybrook trauma data set, for the
purposes of developing MCC 25, proceeded in two parts:
1) a preliminary analysis of trauma cases admitted to the
Regional Trauma Unit; and 2) a final analysis of all
Sunnybrook cases.
This section continues by describing the objectives, the
method and the results for both analyses.
MAINTENANCE OF CASE MIX TOOLS: THE CIHI REVISION PROCESS FOR MCC 25 (TRAUMA)
PRELIMINARY ANALYSIS OF
SUNNYBROOK TRAUMA DATA
Preliminary analysis contributed to an understanding of
the statistical relationship between total trauma patient
care costs, current CMG and other indicators that may
index higher cost.
The Analysis
The analysis of this database was undertaken in the following steps:
1.
2.
3.
4.
Summarize numbers in the Trauma database;
Use linear regression techniques to clarify the
relationship between several indicators and the total
cost of patient care;
Identify possible models that may be broadly applied
to the Sunnybrook database to identify ‘trauma’ cases
that may not fall into MCC 25; and
Identify diagnoses or procedures that may be added
to the complexity grade lists to index an increased
need for acute care.
109
The Data
The data used for this analysis contained 611 cases from
the Regional Trauma Unit at the Sunnybrook Health
Sciences Centre, from fiscal 1994/95. All patients admitted to the trauma program in that time period were
reviewed, not just those admitted to the unit. This data
included patient-specific clinical, diagnostic, and cost
information. It had more detail than the CIHI discharge
abstract which was used to develop the patient classification groups.
The 611 patients in the database had diverse clinical histories, mode of injuries, actual injuries, treatment plans
(identified via OR procedures) and lengths of stay. Of
the 611 trauma patients in the database, 329 were
assigned to MCC 25 and 282 were assigned to another
MCC. Table 1 shows that only 54% of the Regional
Trauma Unit patients were being assigned to the trauma
MCC. This was higher than the original claim of 40%
but still unacceptably low.
Table 1: Percent of Trauma Patients Assigned to Trauma MCC 25, 1996 CMG Methodology
Category
MCC 25
Other MCC
Total
Number of Cases
329
282
611
The trauma data had detailed patient-specific information
from the Record of Stay sheet. Suggested elements for
analysis included: ventilator type, tracheostomy, and total
parenteral nutrition (TPN). Number of days, patient ISS,
and AIS values were also recorded.
Ventilator type included bag, mechanical, spontaneous,
unknown, and cases not coded. In order to identify the
patients with mechanical ventilation, a cross-tabulation
was produced which compared the ventilator types with
the number of ventilator days.
In the trauma data there were 40 cases coded as spontaneous, but which also showed ventilator days. For the
study, these cases were included with the mechanical ventilation group.
% of Total
54%
46%
100%
The Variables
The following elements have been modeled in various
combinations to predict total cost:
•
•
•
•
•
•
•
•
•
•
•
age, and age-squared;
ISS score (continuous);
ISS greater than 16 (dummy) (ISS16);
number of ventilator days;
ventilator days > 4 (dummy) (vent4);
number of mechanical ventilator days;
mechanical ventilator days > 4 (dummy) (vent4);
number of TPN days;
TPN days > 0 (dummy) (TPN0);
number of tracheostomy days; and
Tracheostomy days > 0 (dummy) (Trach0).
110
MAINTENANCE OF CASE MIX TOOLS: THE CIHI REVISION PROCESS FOR MCC 25 (TRAUMA)
The Regression Models
The general model used for analysis of total cost was:
Equation 1
Total Cost = (AvCost) + b * ISS + c * ventilator + d *
TPN + e * tracheostomy
where each variable represents one of the potential variable types (continuous or dummy). For example, either
ISS or ISS16.
AvCost was the average cost, by CMG, within the data
set. This was used to control for the different case mix
within each CMG. The AvCost acted as the ‘intercept’ or
constant in these regression models. This is an accepted
technique that allows for the magnitude, rather than the
absolute value, of the independent variables to be accurately determined.
The regression models included every case except deaths
from the Sunnybrook Trauma data—a total of 548 cases.
Variables that measure number of days may have been
skewed if death was included in the analysis.
As well, the following equation was used to model the
AIS value by body area as a predictor of total cost:
Equation 2
Total Cost = (AvCost) + b * AIS_FACE + c *
AIS_CHEST + ……
In the third model, the AIS and other variables were
combined, based upon the strength of the variables in a
more comprehensive model. Finally, the fourth model
modified the significant variables that were derived in the
first three models to reflect data currently captured in the
CIHI standard abstract.
There were two approaches to modeling total costs with
this data. First, elements that appeared to have a significant impact on the total cost of patient care were identified. This general analysis identifies elements to consider
in re-designing the discharge abstract. The second
approach narrowed the focus to those elements that were
reliably collected on the discharge abstract under current
coding practice.
Statistical Regression
Results - A. General Analysis
Model One: ISS Score, Mechanical
Ventilation, TPN and Tracheostomy
The following model used ISS score, ventilator, total parenteral nutrition (TPN), and tracheostomy days as independent and continuous variables to predict total patient
care cost. The Average CMG cost was used to control
for the differing case-mix in the database.
Note that as expected, all coefficients were positive, indicating that the total cost increases as the value of each of
the variables increases (see Table 2). As well, the Average
CMG Cost was the coefficient with the largest relative
effect on predicting total cost (as indicated by the numbers in the beta column), followed by the number of ventilator days.
The coefficient for #TPN days was not significant at the
95% confidence level, and was dropped from the equation. The adjusted R2 coefficient, a summary measure of
the predictive ability of the model, was 0.672.
This analysis demonstrated that ISS, mechanical ventilator, and tracheostomy were associated with increased
patient care cost in the trauma database.
Table 2: Total Cost Regression Model One
Adjusted
Variable
Average CMG Cost
ISS Score
# Ventilator Days
# Trach Days
1
2
R coefficient
Coefficient
0.553
225.80
1819.18
720.93
0.672
SE of Beta
.06
51.75
188.05
215.31
Beta 1
.421
.209
.290
.094
p value
0.0000
0.0000
0.0000
0.0009
Beta Coefficients can be used as indicators of the relative importance of variables, and do not depend on the
units of measurement.
MAINTENANCE OF CASE MIX TOOLS: THE CIHI REVISION PROCESS FOR MCC 25 (TRAUMA)
Model Two: AIS and Total Cost
A second multiple linear regression model was used to
begin assessment of the importance of AIS scores in predicting total patient cost. From this analysis, it was clear
that the Head and Neck AIS score was important and significant in predicting total patient cost (see Table 3). The
coefficient for Extremities AIS was significant and nega-
111
tive, indicating that as this AIS increases, the expected
patient cost decreases, within this database. It was interesting to note that although this model had a greater
number of significant variables included in the model,
the adjusted R2 coefficient was lower than Model One
( 0.580 vs. 0.672).
Table 3: Total Cost Regression Model Two
2
R coefficient
Adjusted
Variable
Average CMG Cost
Limbs
Head and Neck
Chest
Abdomen
Face
Extremities
Coefficient
.745
1414.29
2153.94
1283.74
1425.41
1825.22
-6326.95
Model Three: Combined Model
In Model Three, the AIS indicators were combined with
the Model One indicators—mechanical ventilation, tracheostomy, and TPN days. A forward selection model
was used to enter variables into the model. This type of
model enters variables into the equation one at a time,
starting with the largest correlation with the dependent
variable, until the significance criterion is no longer met.
In Model Three, only the Extremities AIS variable was
significant at the 95% level (see Table 4). The other AIS
0.580
SE of Beta
.072
557.87
501.59
454.65
554.63
807.31
1569.22
Beta
.551
.117
.206
.106
.093
.079
-.249
p value
.0000
.0115
.0000
.0049
.0104
.0242
.0001
variables were removed from this model. While in Model
Two, the Head and Neck AIS score was significant, the
relatively high correlation with the ISS value may have
made Head and Neck AIS redundant in this model. As
well, the number of days TPN was not significant.
Note that although this model had one more variable than
Model One, the adjusted R2 coefficient values were approximately equal (0.672 vs. 0.674). This indicates that although
the Extremities AIS variable is significant, its inclusion adds
relatively little predictive value to Model Three.
Table 4: Total Cost Regression Model Three
Adjusted
Variable
Average CMG Cost
Extremities AIS
ISS
# Trach Days
# Mech Vent Days
2
R coefficient
Coefficient
.594
-2717.09
307.56
636.27
1781.49
0.674
SE of Beta
.064
1311.58
64.96
218.51
188.36
Beta
.452
-.111
.285
.083
.284
p value
.0000
.0388
.0000
.0037
.0000
112
MAINTENANCE OF CASE MIX TOOLS: THE CIHI REVISION PROCESS FOR MCC 25 (TRAUMA)
Statistical Regression Results - B. Using
Information Currently Available on the CIHI
Discharge Abstract
Finally, in Model Four, several of the previously identified
important variables were modified. In this model, the
variables were changed into ‘dummy’ variables, indicating
the presence or absence of the corresponding procedure
or condition. Table 5 describes the changes from continuous to dummy variable. The variable Extremities AIS was
not included since it added relatively little predictive value
to Model Three and because it was not an element on the
discharge abstract. The variables were changed as shown
in Table 5.
Table 5: Dummy Variables for Regression Model Four
Continuous Variable
ISS
# Trach Days
# Mech Vent Days
# TPN Days
Dummy Variable
ISS > 16
Trach Days > 0
Mech Vent Days > 4
Mech Vent Days >0,
<=4
TPN Days > 0
These types of variables are more consistent with the current CIHI abstracting guidelines. Currently, the greater
than 4 mechanical ventilation days variable is coded separately on the abstract. As well, procedure codes may be
used to indicate the presence of tracheostomy and TPN,
although the number of days is not available on the discharge abstract.
Definition
- ISS score of greater than 16
- Presence of tracheostomy days
- Greater than 4 Mechanical vent days
- Greater than 0 and less than/equal to 4
Mechanical vent days
- Presence of TPN days
As in Model Three, a forward-step regression technique
was also used in this modeling process. This allowed variables contributing most to the predictive value of the
model to be entered first. The results from this process
are shown below in Table 6.
Table 6: Total Cost Regression Model Four
Adjusted
Variable
Average CMG Cost
Ventilation > 4
days
TPN
Tracheostomy
2
R coefficient
0.65239
Coefficient
.784
23299.36
SE of Beta
.037
2676.36
Beta
.596
.263
p value
.0000
.0000
7072.45
15368.85
3531.56
3957.12
.055
.112
.0457
.0001
Both ISS > 16, and mechanical ventilation between 0 and
4 days were not significant at the 95% confidence limit.
As expected, ventilation > 4 days was significant, and the
absolute value of the coefficient was large. The presence
of TPN became significant at the 95% level, as did tracheostomy. It is interesting to note that although the
variables in this model were less specific—they were
dummy variables rather than continuous—the adjusted R2
coefficient values were relatively close (0.672 in Model
One vs. 0.652 in Model Four).
Results
Analysis confirmed the importance of mechanical ventilation in this database as an indicator of increased resource
use, above that expected by the CMG average cost alone.
As well, tracheostomy was consistently important in predicting higher costs than the CMG average.
This preliminary analysis used four models to develop an
improved understanding of possible indicators of greater
resource utilization among trauma patients treated at
SHSC. Models One to Three used various indicators,
including ISS, AIS, TPN days and tracheostomy days, that
are routinely collected at Sunnybrook for trauma patients
but are not currently submitted to CIHI by any of the 600
client hospitals. Model Four used the variables identified
in the first three models and redefined these variables so
that they may be identified from the current standard
CIHI discharge abstract. It is notable that the predictive
validity of Model Four, as measured by the adjusted R2
MAINTENANCE OF CASE MIX TOOLS: THE CIHI REVISION PROCESS FOR MCC 25 (TRAUMA)
coefficient value, was comparable to the models that used
the expanded information contained in the Sunnybrook
trauma data.
The Analysis
Analysis of this database was undertaken in the
following steps:
In Models One to Three, the number of total parenteral
nutrition days was not significant at the 95% confidence
level. In contrast, in Model Four, TPN was redefined as
the presence or absence of TPN, rather than the number
of days of TPN. This redefined TPN variable was significant in Model Four.
1.
2.
It is important to note that this analysis was performed
on a database of patients admitted to the trauma centre at
Sunnybrook—a total of 611 patients. The relatively small
size of this database prevented more detailed analysis on
particular diagnoses. For this reason, and to confirm
these findings, the analysis was repeated for all
Sunnybrook cases.
4.
FINAL ANALYSIS OF ALL
SUNNYBROOK CASES
The purpose of this analysis was:
1) to confirm that the findings of the preliminary
analysis of the Sunnybrook Trauma data were more
general to all abstracts submitted by Sunnybrook
to CIHI;
2) to identify additional diagnoses that may indicate
increased patient cost; and
3) to asses the validity of using LOS as a proxy for total
cost to measure these effects in the CIHI database.
3.
113
summarize numbers in the Sunnybrook database;
use regression techniques to clarify the relationship
between several procedural indicators and the total
cost of patient care;
identify possible models that may be broadly applied
to the Sunnybrook database to identify ‘trauma’ cases
that may not fall into MCC 25; and
identify diagnoses or procedures that may be added
to the complexity grade lists to index an increased
need for acute care.
The Data
The data used for this analysis contained 17,426 cases
from Sunnybrook Health Sciences Centre, from fiscal
1994/95. This data included patient-specific cost information collected by the Ontario Case Cost Project, chart and
register number. The chart and register number were
used to match this data to the CIHI Discharge Abstract
Database for SHSC. Cases admitted before April 1, 1994
were excluded from the analysis.
The Variables
Table 7 contains the administrative data elements modeled in various combinations to predict total patient cost
and LOS.
Table 7: Data Elements Modeled for Predicting Total Patient Cost and LOS
Procedure
Gastrostomy
TPN
Tracheostomy
Dialysis
ICP Monitor
Ventilation
Description
Temporary/Permanent
Percutaneous Endoscopic
Temporary
Permanent
Mediastinal
Peritoneal
Hemodialysis
< 96 hours
> 96 hours
Eleven dummy variables were used to indicate the presence or absence of the procedure on the patient’s
abstract. To avoid double-counting, all cases in CMG 40
(Tracheostomy/Gastrostomy) were assigned a dummy
value of 0 for the tracheostomy and gastrostomy proce-
CCP Code
55.2
55.1
13.59
43.1
43.29
43.21
66.98
51.95
14.88
13.62
13.62
ICD-9 CM Code
43.19
43.11
99.15
31.1
31.29
31.21
54.98
39.95
01.18
96.71
96.72
dures. As well, cases in CMG 538 (Dialysis) were
assigned a dummy value of 0 for the dialysis procedures.
The occurrence of each procedure is summarized in the
cross-tabulation table in Appendix II.
114
MAINTENANCE OF CASE MIX TOOLS: THE CIHI REVISION PROCESS FOR MCC 25 (TRAUMA)
The Regression Models
Two general models were used to generate predictions for
patient treatment cost (total cost), and LOS.
Death and Transfer cases were excluded from the analysis, as the costs or LOS of these cases did not represent a
full course of treatment.
Equation 3
Statistical Regression Results
A stepwise regression procedure was used to enter the
variables into the models. In this method, variables were
added to the model, one at a time, based on the strongest
partial correlation. The summary for both models is
shown in Table 8 and Table 9.
Total Cost = (AvCost) + b * Procedure1 + c *
Procedure2 + …
Equation 4
LOS = (AvLOS) + q * Procedure1 + r * Procedure2
+…
AvCost and AvLOS were the average cost or LOS, by
CMG, within the data set. They were used to control for
the different case mix within the database. The AvCost
and AvLOS acted as the ‘intercepts’ or constants in these
regression models.
The stepwise regression technique generated a series of
output results, based on the variables that were in the corresponding step of the model. Table 8 summarizes the
model after all significant variables were in the model.
However, this is not necessarily a recommended model.
While all variables may have been significant, they may
not have necessarily contributed any predictive value to
the model, as measured by the adjusted R2 coefficient.
Table 8: Total Cost Regression Model
adjusted R
Variable
Average CMG Cost
Ventilation >96 hrs
Tracheostomy -Temp
TPN
Gastrostomy - Percutaneous
endoscopic
Ventilation - <96 hrs
ICP Monitor
Gastrostomy - Temp/Perm
Dialysis - Peritoneal
Dialysis - Hemodialysis
2
coefficient=0.662
Coefficient
0.840
36330.1
21187.2
21264.5
12591.5
SE of Beta
.007
711.838
772.438
2022.540
934.818
Beta
.619
.258
.137
.122
.065
p value
.000
.000
.000
.000
.000
2390.652
11184.7
-9460.7
2074.197
2300.823
262.610
1251.534
2080.201
544.065
692.772
.047
.043
-.053
.018
.016
.000
.000
.000
.000
.001
The adjusted R2 coefficient for this model was 0.662.
It is interesting to note that this R2 value remained
unchanged in the last three iterations of the model
(Appendix III). Although indicators for Temp/Perm
Gastrostomy, and Dialysis (Peritoneal and Hemodialysis)
were statistically significant, they did not improve the predictive ability of the model. Recall that TPN and
Gastrostomy (Temp/Perm) were highly correlated, indicating that the results in this model for these variables
may have been confounded. Consideration should be
given to removing one or more of these variables from
the model. As well, it may be noted that the coefficient
for Gastrostomy (Temp/Perm) was negative, indicating
that the presence of a gastrostomy procedure would lead
to lower expected costs.
All of the variables that were significant in the part 1
analysis of the Sunnybrook Trauma database remained
significant. It is interesting to note that the absolute value
of each of the coefficients was greater than in the part 1
analysis. Also, note that Ventilation <96hrs was significant in this model. Consideration should be given to the
consistency and reliability of the coding practice for this
procedure, as well as the others.
MAINTENANCE OF CASE MIX TOOLS: THE CIHI REVISION PROCESS FOR MCC 25 (TRAUMA)
The same model, and process was used to predict the
patient’s LOS (see Table 9). Note that fewer variables
remained significant in this analysis. Temp/Perm
Gastrostomy, Peritoneal Dialysis and Ventilation <96hrs
were not statistically significant. As well, dialysis was
the last variable to enter the model and had the lowest
Beta value.
115
In this model, the adjusted R2 coefficient did not change
after the addition of either ICP Monitor or Hemodialysis
(Appendix III).
Table 9: LOS Regression Model
adjusted R
Variable
Average CMG LOS
Ventilation >96 hrs
Tracheostomy -Temp
Gastrostomy - Percutaneous
endoscopic
TPN
ICP Monitor
Dialysis - Hemodialysis
2
coefficient=0.558
Coefficient
0.940
22.629
20.576
23.029
SE of Beta
.008
1.250
1.371
1.661
Beta
.677
.103
.086
.077
p value
.000
.000
.000
.000
15.620
15.016
2.684
1.469
2.219
1.231
.058
.037
.012
.000
.000
.029
Results
The results of this analysis confirmed the importance of
mechanical ventilation >96 hrs, tracheostomy and TPN in
predicting both patient cost and LOS. Percutaneous
endoscopic gastrostomy was also a significant predictor of
cost and LOS.
Indicators for ICP Monitor and Hemodialysis were also
statistically significant, although they did not add value to
the model, as measured by the adjusted R2 coefficient.
² calculation of average length of stay and standard
deviation for each proposed CMG to ensure homogeneity of the group;
² regression analysis of each proposed CMG to ensure
a single distribution within a group - any group with a
binomial distribution was split into two groups and
re-analyzed; and
² the clinical identification of procedures predictive of
high resource use and their hierarchy were confirmed
by regression analysis.
The analysis confirmed that in the absence of routinely
collected, detailed patient-specific cost information, LOS
may be used as a valid proxy for cost.
The final list of diagnoses and procedures selected for
inclusion in MCC 25 may be found in the 1997 CMG
Directory for Use with ComPlexity (CIHI, 1997).
REVISED MCC 25
SIGNIFICANT TRAUMA
Assignment to the new MCC 25 Multiple Significant
Trauma was now based on the presence of a single significant trauma code as the Most Responsible Diagnosis. A
flow diagram representing this assignment may be found
in the introduction chapter of this case book, Figure One,
CIHI Case Mix Tools. The new list of diagnoses included
ICD (800-999) Injury codes. The list was refined to
demonstrate conditions of "classic trauma" (the results of
accidents). The focus for the new MCC was on defining
significant trauma cases. Exclusions consisted of the following conditions: foreign body, eye trauma, sprains &
strains, late effect codes, some contusions, poisonings, and
burns. Burns remained in MCC 22 and some minor trauma diagnosis were still found in MCC 2, 8 and 21. Many
trauma diagnoses which appeared in the other MCC were
now assigned to the new trauma version.
In conjunction with the Sunnybrook trauma data analyses,
the CIHI CMG Team and Dr. Tony Ashworth and Barry
McLellan from the original Trauma Task Force proceeded
with the identification of most responsible diagnoses and
procedures to include in MCC 25. The ICD-9 or ICD-9CM most responsible diagnosis assigned a case to the
MCC 25 while the CCP or ICD-9-CM procedure assigned
the case to a specific CMG within MCC 25.
The identification of specific diagnoses and procedures,
the hierarchy of procedures (those most predictive of
resource use) and the case mix groups were developed on
the basis of clinical judgment. These indicators and
groups were also confirmed by the following analyses on
the entire CIHI Discharge Abstract Database for fiscal
1994/95:
116
MAINTENANCE OF CASE MIX TOOLS: THE CIHI REVISION PROCESS FOR MCC 25 (TRAUMA)
A portion of the new groupings have been created using
a combination concept to illustrate the significant trauma
cases (cranial, spine, femur, thoraco-abdominal, lower
extremity).
There was a true Surgical Partition in the new trauma
MCC, however, there was no logic for "Unrelated O.R.
Procedures" (a concept similar to what has been incorporated into MCC 14). Therefore, in cases where a procedure was abstracted and the case was not included in the
logic of the Surgical Partition, the case became a medical
CMG based on the MRDx.
The introduction of MCC 25 Significant Trauma in 1997,
coincided with the introduction of Complexity. The Plx
overlay was an enhancement to those CMG methodology.
It allowed diagnoses other than the labeled "most responsible" to be considered in the grouping of cases. The
CIHI Complexity methodology grade ‘lists’ were used to
index greater patient resource requirement. Mechanical
ventilation > 96hr was one procedure on the Complexity
grade lists for all MCC.
The problem with using procedures on these grade lists
was that with the previous CMG methodology, many procedures were used in CMG assignment. If a procedure
was on the CMG assignment list and the Complexity list
as a grade ‘A’, then all cases in that CMG were assigned to
Plx Level 4. Thus, although many procedures had been
found to contribute to greater cost or LOS, most of these
procedures were also used for CMG assignment. In the
analysis of the Sunnybrook data, only two procedures,
mechanical ventilation >96hrs, and TPN were not currently used for CMG assignment. Further analysis indicated that TPN was found to be insignificant in predicting greater cost. Mechanical ventilation > 96hr was then
identified as a condition contributing to Complexity for
MCC 25.
The new trauma MCC consisted of 44 case mix groups
compared to five case mix groups previously. The resulting case mix groups arranged by surgical partition and
medical partition may be found in Table 10a and Table
10b respectively. An example flow diagram of the grouping logic can be found in Figure 3 and Figure 4.
CMG
Surgical Partition
Tracheostomy and Gastrostomy Procedure for Trauma
650
Yes
Yes
OR
Procedure
Intracranial
Procedure
for Trauma
No
651
Spinal
Procedure
Yes
652
No
Femur
Procedure
Yes
4
653
No
Thoraco-Abd
Procedure
Yes
No
654
Wound Db/
Low Extrem
Procedure
No
1
660
Figure 3: Example of
Grouping Logic for the
Surgical Partition of
MCC 25
MAINTENANCE OF CASE MIX TOOLS: THE CIHI REVISION PROCESS FOR MCC 25 (TRAUMA)
CMG
Medical Partition
Figure 4: Example of the
Grouping Logic for the
Medial Partition of
MCC 25
4
Yes
Fractures and
Dislocations of
Femur or Pelvis
Intracranial
Injuries
675
Yes
677
No
Spinal
Injuries
Yes
679
No
Thoraco-Abd
Injuries
680
No
Frostbite
681
5
Table 10a: Comparison of 1996 MCC25 and 1997 MCC 25 by CMG using CIHI 1995/96
Abstract for All Provinces – Surgical Partition
Surgical Partition
MCC 25 Multiple Significant Traum a
CMG
CMG D escription
875
MU LT SG TRAU MA W CRAN IOTOMY
876
877
MCC 25 Significant Traum a
Cases
CMG
CMG D escription
Cases
92
650
CRAN IAL W SPIN AL PROC TRX
9
MU LT SG TRM,LIMB REAT/H IP,FEM
372
651
CRAN IAL W FEMU R PROC TRX
10
MU LT SG TRAU MA W OTH TRAU M PR
765
652
CRAN /FEMU R W TH OR/ABD PR TRX
96
653
CRAN IAL W LE/WOU N D D B TRX
26
654
SPIN AL W FEMU R PROC TRX
19
655
SPIN AL W TH OR/ABD PROC TRX
16
656
SPIN AL W LE PR/WN D D B TRX
35
657
FEMU R W LE PR/WN D D B TRX
480
658
TH OR/ABD W LE PR WN D D B TRX
122
660
CRAN IAL PROC FOR TRX
367
661
SPIN AL PROC FOR TRX
662
MAJOR JOIN T/FEMU R PROC TRX
663
TH ORACO/ABD OM PROC TRX
664
WOU N D D B/SKN GRFT TRX
665
ELEV ATED SKU LL FRACTU RE
666
MAJOR LOW EX T PROC TRX
667
MIN OR LOW EX T PROC TRX
668
MISC MU SCU LOSKEL PROC TRX
U nrelated O.R.
900
EX TEN SIV E U N RELATED O.R. PROC
2
669
V ASCU LAR REPAIR FOR TRX
901
N ON -EX T U N RELATED O.R. PROC
2
670
U PPER EX TREMITY PROC TRX
Total Cases
1233
117
410
13,183
929
2,332
79
12,364
401
1,578
856
8,323
41,635
118
MAINTENANCE OF CASE MIX TOOLS: THE CIHI REVISION PROCESS FOR MCC 25 (TRAUMA)
Table 10b: Comparison of 1996 MCC25 and 1997 MCC 25 by CMG using CIHI 1995/96
Abstract for All Provinces – Medical Partition
Medical Partition
MCC 25 Multiple Significant Traum a
CMG
CMG D escription
MCC 25 Significant Traum a
Cases
CMG
CMG D escription
Cases
878
MU LT SG TRM IN V H D /CH /LW LMB
737
674
IN TERCRAN IN J WITH SPIN AL IN J
38
879
MU LT SG TRM (N OT H D /CH /LW LMB)
140
675
IN TERCRAN IN J WITH FX FEMU R
26
676
IN TERCRAN IN J WITH TH OR/ABD
677
SPIN AL IN JU W FX FEMU R
134
678
SPIN AL IN J W TH ORACO-ABD IN J
260
679
FRACT FEMU R W TH ORACO-ABD IN J
680
FRACT/D ISLOC OF FEMU R/PELV IS
681
FROSTBITE
682
SPIN AL IN JU RIES
3,081
683
IN TERCRAN IAL IN JU RIES
1,232
684
H IP AN D TH IGH IN JU RIES
685
TH ORACO-ABD OMIN AL IN JU RIES
686
MAJOR N ERV E IN JU RIES
687
FRACTU RE OF H U MERU S
688
WEIGH T BEARIN G IN JU RIES
690
GEN ITO-U RIN ARY IN JU RIES
691
CRU SH IN G IN JU RIES/CON TU SION S
692
MIN OR LOWER EX TREM FRACTU RES
693
WOU N D S
694
AMPU TAT/V ASC/OTH N ERV E IN JU RY
695
FACIAL IN JU RIES
1,625
696
OTH ER CRAN IAL IN JU RIES
7,524
697
U PPER EX TREMITY FRACTU RES
6,221
Total Cases
877
Tables 10a and 10b also compare the number of cases
assigned to the 1996 MCC Multiple Significant Trauma
with the number assigned to the 1997 MCC Significant
Trauma using the CIHI fiscal 1995/96 DAD for all
provinces. Both the surgical and medical partition of
MCC 25 Significant Trauma showed a considerable
increase in the number of trauma cases captured: 41,635
cases assigned to the surgical partition of the 1997 MCC
25 compared with 1,233 cases assigned to the surgical
partition of the 1996 MCC 25; and 43,120 cases assigned
to the medical partition of the 1997 MCC 25 compared
with 877 cases assigned to the medical partition of the
1996 MCC 25.
73
166
3,517
115
973
4,155
66
980
3,778
511
2,937
152
4,978
578
43,120
A similar result was found when making the same comparison with the Sunnybrook Trauma data from fiscal
1994/95 and fiscal 1995/96. Table 11 demonstrates that
the revised MCC 25 captures 89% of The Regional
Trauma cases compared with 54% captured by the existing MCC 25. Using the Sunnybrook Trauma Patient
Service 38 data from fiscal 1995/96 the same comparison
resulted in an increase from 51% to 96% (see Table 12 ).
MAINTENANCE OF CASE MIX TOOLS: THE CIHI REVISION PROCESS FOR MCC 25 (TRAUMA)
119
Table 11: Regional Trauma 1994/95 Cases Grouped by Multiple Significant Trauma
and Significant Trauma
Category
MCC 25
Other MCC
Total
# of Cases Multiple
Significant Trauma
329
282
611
% of Total
54%
46%
100%
# of Cases
Significant Trauma
541
68
609 1
% of Total
89%
11%
100%
Table 12: Sunnybrook Patient Service 38 - Trauma 1995/96 Cases Grouped by Multiple
Significant Trauma and Significant Trauma
Category
MCC 25
Other MCC
Total
# of Cases Multiple
Significant Trauma
304
291
595
% of Total
51%
49%
100%
# of Cases
Significant Trauma
574
21
595
% of Total
96%
4%
100%
Both comparisons, the CIHI DAD and the Sunnybrook
Trauma data, suggest that the revised MCC 25 Significant
Trauma is a better representation of significant trauma
cases than the previous MCC 25 Multiple Significant
Trauma.
MCC 25, has been achieved. Ninety-six per cent of
Sunnybrook's Trauma cases are now assigned to MCC 25.
The case mix groups within MCC 25 are now more
homogeneous yielding RIW estimates which are more
reflective of resource use in true trauma.
SUMMARY
The implementation of ICD-10 and CCI (Canadian
Classification of Health Interventions) may present us
with an opportunity to identify additional indicators of
high resource use and, because these standards are more
specific, they may assist with improving the homogeneity
of the case mix groups within MCC 25. Moving to a single national diagnosis classification standard across
Canada may also provide the opportunity to implement
the collection of ISS scores on the discharge abstract and
define a consistent ISS scoring system for the identification of significant trauma patients.
The revision of MCC 25 was a process involving both
clinical input and statistical analysis. The recommendations of the Trauma Task force contributed to the definition of a significant trauma case and the identification of
procedures and diagnosis for inclusion in MCC 25. The
regression analysis of the Sunnybrook trauma cost data
confirmed the importance of tracheostomy, gastrostomy
and mechanical ventilation >96hrs as an indicator of
increased resource use. Finally, the analysis also confirmed that in the absence of routinely collected, detailed
patient-specific cost information, LOS may be used as a
valid proxy for cost.
The revision of the Trauma MCC 25 demonstrated how
client input and feedback played a significant role in the
revision and/or enhancement process of the case mix
grouping methodology. Without the initial interest and
concern expressed across Canada to change the way cases
were identified as trauma and without the availability of
client cost data, the revision of MCC 25 would not have
been possible. The original objective of this project, to
improve the percentage of true trauma cases assigned to
1
2 records could not be matched to the CIHI database and were therefore not included in the total.
120
MAINTENANCE OF CASE MIX TOOLS: THE CIHI REVISION PROCESS FOR MCC 25 (TRAUMA)
REFERENCES
Baxt, W.G. & Valda, U. (1990). The Lack of Full
Correlation Between the Injury Severity Score and the
Resource Needs of Injured Patients. Annals of Emergency
Medicine 19(12),1396-1400.
Bazzoli, G.J., Madura, K.J., Cooper, G.F., MacKenzie, E.J.,
Maier, R.V. (1995). Progress in the Development of
Trauma Systems in the United States. Jama 273(5),
395-401.
CIHI MCC 25 Task Force Meeting Minutes 1994, May 17
CIHI MCC 25 Task Force Meeting Minutes 1994, July 11
Offner, P.J., Jurkovich, G.J., Gurney, J. & Rivara, F.P.
(1992). Revision of TRISS for Intubated Patients. The
Journal of Trauma 31(1)32-35.
Ontario Trauma Registry (1992). October 22
Osler, T. (1993). Injury severity scoring: perspectives in
development and future directions. American Journal of
Surgery 165(2A Suppl) 43.
About the Authors
Julie Richards, MHSc is ICD-10/CCI Plan Coordination Manager,
Canadian Institute for Health Information, Toronto, Ontario.
Greg Fallon, BSc, MBA was Researcher/Sr. Analyst in Standards and
Methodologies at CIHI, and is currently a Consultant at Hay Health Care
Consulting Group.
Margaret Catt, RN, MHSc was the Patient Care Process Co-ordinator at
Mount Sinai Hospital, Toronto and is currently Information/Nursing
Leader at Temiskaming District Hospital.
Karen Horne is Senior Programmer/Analyst, Canadian Institute for
Health Information, Toronto, Ontario.
Acknowledgement
The authors of this chapter would like to thank Dr. Barry McLellan, Chief
of Trauma at Sunnybrook Health Sciences Centre and Dr. Tony
Ashworth, Medical Consultant to CIHI who spent valuable time providing clinical input to the revision of MCC 25. Cost data used in the revision of MCC 25 was generously provided by Sunnybrook Health
Science Centre and used with the kind permission of the Ontario Case
Cost Project.
APPENDIX I
CIHI
•
•
•
•
•
•
Trauma Task Force
Dr. Barry McLellan, SHSC, Toronto
Dr. Judy Vestrup, VGH (now at MOH), Victoria
Dr. Murray Girotti, Victoria Hospital, London
Dr. David Fleiszer, Montreal General Hospital
Ms. Tamara Stefanits, Trauma Coordinator, Victoria Hospital, London
Dr. Tony Ashworth, Hotel Dieu, Kingston
121
APPENDIX II
123
Procedure Frequencies for the Analysis of all Sunnybrook Cases
Procedure * RIW Exclusion Category
Cross-tabulation
Gastrostomy - Temp/Perm
not present
present
Total
Ventilation - <96 hrs
not present
present
Total
Ventilation - > 96 hrs
not present
present
Total
Gastrostomy - Percutaneous
not present
endoscopic
present
Total
TPN
not present
present
Total
Tracheostomy - Temp
not present
present
Total
Tracheostomy - Permanent
not present
present
Total
Tracheostomy - Mediastinal
not present
present
Total
Dialysis - Hemodialysis
not present
present
Total
ICP Monitor
not present
present
Total
Typicals
14514
38
14552
13967
585
14552
14493
59
14552
14519
33
14552
14514
38
14552
14498
54
14552
14551
1
14552
14552
14552
14489
63
14552
14529
23
14552
RIW Exclusion Category
Outliers
Transfers
Sign-outs
859
954
119
19
11
878
965
119
835
833
116
43
132
3
878
965
119
832
943
119
46
22
878
965
119
853
954
119
25
11
878
965
119
853
951
119
25
14
878
965
119
855
956
119
23
9
878
965
119
878
965
119
Deaths
887
25
912
783
129
912
853
59
912
891
21
912
882
30
912
900
12
912
912
Total
878
878
965
965
119
119
912
912
17333
93
17426
16534
892
17426
17240
186
17426
17336
90
17426
17319
107
17426
17328
98
17426
17425
1
17426
17426
878
866
12
878
871
7
878
965
960
5
965
960
5
965
119
119
912
900
12
912
896
16
912
17426
17334
92
17426
17375
51
17426
119
119
119
APPENDIX III
125
Complete Regression Results for Total Cost
Regression - Total Cost, CMG Average Cost and Procedures
Model Summary
Model
Variables Entered
1
2
3
4
5
6
7
8
9
10
11
AVCOST1
Ventilation - >96 hrs
Tracheostomy - Temp
TPN
Gastrostomy - Percutaneous
Ventilation - <96 hrs
ICP Monitor
Gastrostomy - Temp/Perm
Dialysis - Peritoneal
Dialysis - Hemodialysis
Dialysis - Hemodialysis
cde
Adjusted r
coefficient
2
Std. Error of
the Estimate
.542
.628
.648
.654
.658
.660
.661
.662
.662
.662
.662
6967.07
6285.87
6106.44
6061.28
6025.64
6008.37
5993.13
5989.25
5986.38
5984.40
5984.40
Adjusted r 2
coefficient
.519
.538
.547
.553
.556
.558
.558
.558
Std. Error of
the Estimate
11.09
10.88
10.77
10.70
10.66
10.64
10.64
10.64
endoscopic
Complete Regression Results for LOS
Regression - LOS, CMG Average LOS and Procedures
Model Summary
c
d
e
f
g
h
Model
Variables Entered
1
2
3
4
5
6
7
8
AVLOST1
Ventilation - >96 hrs
Tracheostomy - Temp
Gastrostomy - Percutaneous
TPN
ICP Monitor
Dialysis - Hemodialysis
Dialysis - Hemodialysis
fgh
endoscopic
Dependent Variable: Total Cost
Linear Regression through the Origin
Method: Stepwise (Criteria: Probability-of-F-to-enter <=.050, Probability-of F-to-remove
>=.100)
Dependent Variable: LOS
Linear Regression through the Origin
Method: Stepwise (Criteria: Probability-of-F-to-enter <=.050, Probability-of F-to-remove
>=.100)
Conclusion
Lina M. Johnson
The cases presented in this casebook provide examples of how case mix tools have been
used in health care decision making in a wide range of settings and applications by
Canadian health care organizations. It has been shown how case mix tools can be used in
utilization management, budgeting, program planning, restructuring, care planning, case
management, and hospital funding. Although only acute care applications are represented
by these cases, one can expect in the near future, the range of settings for the use of case
mix tools to expand into ambulatory care, chronic care, and long term care, as these tools
continue to be developed and implemented.
What are some of the lessons that can be drawn from the experiences of these organizations with the use of case mix tools in health care decision making?
Lesson 1: Case Mix Tools are Dynamic & Existing Tools
Change and New Tools Will Be Introduced
Existing case mix tools are updated annually by CIHI; thus, the users of these tools can
expect the length of stay, resource intensity weights, and, less frequently, the groups or
methodologies to change. These changes have implications for users as illustrated in some
cases presented in this book. For example, the methodological change from CMG to
Complexity was shown by The Toronto Hospital to produce a more useful tool for their
utilization management efforts as Complexity better predicted their length of stay performance. The changes in RIW associated with the methodology change also had implications
for program budgets, as discussed in the case on St. Michael’s Hospital. It was found that
as the RIW changed, some program budgets required relative decreases, while others
required relative increases in the funds allocated. Although the new Complexity methodology was found to be a better predictor of length of stay, users should remember that the
indicators produce only estimates of resource consumption at individual sites. This limitation must be recognized and taken into account when applying the tools in resource planning activities.
Until case mix tools are developed and implemented for care delivery modes other than
acute care, other and perhaps less precise methods may be required for planning and utilization management. The need and desire for case mix tools that go beyond acute care,
was shown in the St. Michael’s Hospital case, as efforts were made by clinicians and administrators to plan programs spanning modes of care delivery from inpatient, to day surgery
and ambulatory care.
Lesson 2: Users Can Influence the Enhancement of Existing Tools
Users can influence the enhancement of existing case mix tools in two ways: 1) by suggesting that CIHI undertake methodology reviews in specific clinical areas; and 2) by
improving the quality of data submissions to CIHI.
128
CONCLUSION
In Chapter 9, it was shown that the review of MCC 25
was, in part, undertaken as a result of suggestions from
hospitals that a significant percentage of trauma patients
were not being assigned to the trauma clinical category
(MCC 25). Thus, users who have concerns that methodologies are weak in certain areas should approach CIHI
about suggestions for enhancement. CIHI has demonstrated that it is committed to continually improving its
methodologies to better meet the needs of users of case
mix tools and to better reflecting current patterns of
practice in health care.
CIHI works with its clients to define and promote guidelines and standards for data elements submitted to the
database. More accurate reporting by hospitals will in turn
improve CIHI’s ability to identify and introduce robust
enhancements to its methodologies. The predictive value
of the methodologies and associated case weights should
then reflect this quality improvement. Thus, improvements to the case mix tools can begin with the efforts of
Health Records Departments. Efforts aimed at assisting
clinicians to increase their understanding of the tools may
contribute to improving the quality of data submissions
to CIHI. Increased awareness and appreciation for the
significance of documenting co-morbid conditions and
other factors contributing to length of stay may lead to
more useful clinical information on the health record and,
subsequently, on the discharge abstract. As seen in the
cases written by Mt. Sinai Hospital and St. Boniface
Hospital, education of clinicians on case mix tools may
be primarily undertaken for the purpose of improving
patient care processes and the utilization of resources
within the organization. Yet, a beneficial by-product of
these efforts can be improved data quality, and hence,
improved methodologies in subsequent years.
Lesson 3: Computer Systems are Essential
for Database Management, Timely Analysis
and Reporting of Information
The organizations represented in this casebook all manage relatively large amounts of case mix data as compared
to those of small community hospitals. For a large and
complex caseload, computer systems are essential for
data management and case mix analysis. Without computer systems to assist with the collection, analysis and
reporting of data, the efforts described in these cases
would not have been possible. As illustrated in the St.
Boniface and Mt. Sinai cases, not only are computer systems essential for storing data, but systems with the capability to integrate data from several systems are critical for
analyses used in program planning, resource allocation,
and utilization management efforts. However, it is likely
becoming more and more difficult for organizations with
smaller data bases to meet their data and information
requirements without some type of computer system.
With the resource constraints that many health care organizations face, investment in computer systems can assist
organizations to shift the use of their human resources
away from routine data management and collection
toward increased analysis and support for decision making. Such shifts in the use of human resources may assist
organizations to better meet resource constraints by
enabling better and more timely decisions.
Lesson 4: Multi-disciplinary Decision
Support Teams are Required to Transform
Data into Useful Information
Several cases in this book discussed the need for expert
decision support teams to work with clinicians and
administrators to transform data into useful and meaningful information for program management. Although the
composition of these decision support teams and their
positions within the organization varied, it is apparent
that these teams must have a multi-disciplinary set of
skills including, at a minimum, health records, finance,
and patient care processes. More specifically, the analysts
need to understand case mix methodologies, financial and
costing data, budgeting and patient care processes. They
also need to have some level of computer skills, quantitative skills, and the ability to communicate and collaborate
with clinicians and hospital or program administrators.
Since change is ongoing in case mix methodologies,
methods of medical treatment, and patient care processes,
decision support teams will need to continually update
their knowledge to stay current and maintain their expertise in this relatively new field of health care. They will
also need to continually update clinicians and administrators on the implications of case mix tool applications as
they relate to their roles and responsibilities. With the
combination of all these skills, decision support teams
need to understand the information needs of clinicians
and administrators and then transform the data into
information and reports that will be useful in decision
making.
Lesson 5: Decision Makers Need a Culture
of Empirical-Based Decision Making
Each of the organizations presenting their experiences in
this casebook, likely had corporate cultures that support
empirical-based decision making. Otherwise, the organizations would not have supported the efforts described in
these cases. Although most decisions should not be made
on the basis of tools and purely quantitative methods
only, organizational cultures that dismiss the concept of
empirical-based decision making on may be foregoing
opportunities for better decisions and faster adaptation to
the changing health care environment. Organizations that
embrace empirical-based decision making and support the
use of resources for this set themselves apart as leaders in
adapting to the health care environment of the future. An
example of the benefits of a culture that embraces empirical-based decisions is that of the London Health
Sciences Centre. Their response to the Ontario Health
Services Restructuring Commission (HSRC) report was to
CONCLUSION
extend the analyses done by the HSRC and look for specific opportunities to make the necessary changes to meet
the set targets.
Lesson 6: Significant Improvements in
Hospital Performance are Possible with the
Use of Case Mix Tools in Health Care
Decision Making
Some of the cases presented in this book demonstrated
that improvements in hospital performance are possible
with the use of case mix tools and the collaborative
efforts of clinicians, administrators, and decision support
teams. For example, London Health Sciences Centre
demonstrated how it achieved dramatic shifts in the provision of surgical procedures from an inpatient to day
surgery basis with the aid of case mix tools and the day
surgery incentive model. Mt. Sinai Hospital demonstrated
how it made dramatic reductions in its average length of
stay for one CMG related to stroke. St. Michael’s hospital
demonstrated that it had ranked number one among its
peer hospitals for cost per weighted case, a measure of
efficiency in the delivery of acute inpatient, newborn, and
day surgery care. These are a few of the many examples
of improvement in hospital performance illustrated in the
cases presented in this book, and made possible through
the use of case mix tools in decision making.
129
GLOSSARY
Glossary
ALC Days—The number of days assigned to the alternate level
of care (ALC) patient service during the patient’s hospitalization.
ALOS—Average length of stay
APlx Cell—The analytical cell representing CMG, Complexity
Level and Age Category
Atypical Cases—These cases are not used for the calculation of
ELOS.
² Invalid length of stay
² Deaths
² Transfers to or from acute-care institutions
² Sign-outs
² Outliers (cases with lengths of stay beyond the Trim
LOS)
² CMG 910, 912, 997, 998, 999
CCP—Canadian Classification of Procedures
Complexity Overlay—The complexity overlay is an enhancement to the CMG methodology which allows the consideration
of diagnoses other that that labeled "most responsible" to be
considered in the grouping of cases.
Complexity (Plx)—The name and trademark given to the
grouping methodology employing both CMG and the complexity overlay. Versions released by CIHI are:
Complexity 95—preliminary or alpha version
Complexity 96—pilot or beta version
Complexity 97—product version implemented April 1, 1997
131
Most Responsible Diagnoses (MRDx)—"An ICD code identifying the diagnosis, considered by the physician, to be most
responsible for the patient's stay in the institution. The one diagnosis which describes the most significant condition of a patient
which causes his/her stay in hospital. In the case where multiple
diagnoses may be classified as the most responsible, the diagnosis which is responsible for the greatest length of stay is used."
Outlier—Cases where the total LOS is greater than the Trim LOS.
Plx Level—The Complexity Level assigned to the cases by the
grouping methodology.
1. No Complexity
2. Complexity related to chronic condition(s)
3. Complexity related to serious/important condition(s)
4. Complexity related to potentially life-threatening condition(s)
Post-Admission Co-morbid (Type 2) Diagnoses—CIHI
defines a post-admission co-morbid or type 2 diagnosis as:
"An ICD diagnosis describing a condition arising after the
beginning of hospital observation and/or treatment which
influences the patient's hospitalization (i.e. LOS) and/or significantly influences the management or treatment of the
patient."
Pre-Admission Co-morbid (Type 1) Diagnoses—CIHI
defines a pre-admission co-morbid or type 1 diagnosis as:
"An ICD diagnosis describing another important condition
of the patient which has a significant influence on the
patient's hospitalization and/or significantly influences the
management or treatment of the patient."
CMG—Case Mix Groups
RAPD—Routine and Ancillary Per Diem - An estimate of the
variable charges found in 5 of the 8 Maryland charge buckets
(nursing, drugs, lab, diagnostic imaging, physical medicine)
DAD—The CIHI Discharge Abstract Database containing hospital inpatient and day procedure records.
RIW—Resource Intensity Weight—A relative value derived
from case-weighting cost or charge data.
Deaths—Cases where the patient died while in hospital.
SDS—Same Day Surgery—Used for the calculation of DPG
weights
DPG—Day Procedure Groups
ELOS—The expected length of stay for a ‘typical’ case in a
APlx cell calculated by one of five potential models.
HSRV—Hospital Specific Relative Value
ICD-9—International Statistical Classification of Diseases,
Injuries and Causes of Death, Ninth Revision
ICD-9-CM—ICD-9, Clinical Modification - prepared for use in
the United States
LOS—Length of Stay, in days.
May Not Require Hospitalization (MNRH) CMG—”MNRH
discriminates patients whose characteristics often allow ambulatory treatment not requiring admission. MNRH does NOT determine a sample of patient who MUST be treated as ambulatory,
given that these patients may have a justifiable basis for inpatient
admission.”
MCC—Major Clinical Category. The CMG methodology
assigns each case to one of 25 MCC.
Sign-outs—Cases where the patient signs out of the hospital
against medical advice.
Standardization Factor—A ratio of the sum of typical charge
times the case volume to the case volumes. Used to set the average typical RIW = 1.
Typical—Typical cases are those remaining in the CIHI calibration database after the exclusion of deaths, signouts, transfers
and long stay outliers. Only typical cases are used for RIW estimation.
Transfer—A cases that is transferred into or out of an acute
care institution.
Trim LOS—For a given APlx cell, the LOS value at which outliers are excluded from the data.
The Trim LOS is calculated with the long stay cases included.
Using the formula (Q3 + 2 * IQR)
NOTE: Cases whose calculated trim is 1 day have been assigned
a trim of 3 days to avoid inappropriate rewarding of RIW.
(CMG 538, 617,852)
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement