Round Table Health information system reform in South Africa: developing

Round Table Health information system reform in South Africa: developing
Round Table
Health information system reform in South Africa: developing
an essential data set
Vincent Shaw1
Abstract Health services are increasingly under pressure to develop information systems that are responsive to changing health
needs and appropriate to service objectives. Developing an essential data set provides managers with a clearly defined set of
indicators for monitoring and evaluating services. This article describes a process that resulted in the creation of an essential data
set at district level. This had a significant impact on neighbouring districts and resulted in the development of a regional essential
data set, which in turn helped to influence the creation of a provincial and then national essential data set. Four key lessons may
be drawn from the process. The development of an essential data set both requires and can contribute to a process that allows the
reporting requirements to be adjusted over time in response to changing circumstances. In addition, it contributes to (and requires)
the integration of programme reporting requirements into a coherent information system. While the case study describes a bottom-up
approach, a top-down consultative process is advocated because it establishes a framework within which information needs can be
reviewed. Lastly, the use of surveys can aid efforts to keep the essential elements to a minimum. In conclusion, the development of
an essential data set contributes to strengthening health services because it necessitates dialogue between programme managers
and defines indicators to be monitored by them.
Keywords Information systems/organization and administration; Health status indicators; Data collection/methods; Community health
services; Delivery of health care; Program evaluation/methods; South Africa (source: MeSH, NLM).
Mots clés Système information/organisation et administration; Indicateur état sanitaire; Collecte données/méthodes; Service public
santé; Délivrance soins; Evaluation programme/méthodes; Afrique du Sud (source: MeSH, INSERM).
Palabras clave Sistemas de información/organización y administración; Indicadores de salud; Recolección de datos/métodos; Servicios
de salud comunitarios; Prestación de atención de salud; Evaluación de programas/métodos; Sudáfrica (fuente: DeCS, BIREME).
Bulletin of the World Health Organization 2005;83:632-639.
Voir page 635 le résumé en français. En la página 636 figura un resumen en español.
Increasingly, information systems for monitoring health services
are being scrutinized for their appropriateness and ability to
provide meaningful information to managers (1–4). The vision
of the District Health Information System (DHIS) developed
in South Africa is “to support the development of an excellent
and sustainable health information system that enables all health
workers to use their own information to improve coverage and
quality of health care within our communities” (5). According to
the basic principles of the DHIS, it supports the district-based
primary health care approach, collects essential data used to
calculate indicators, encourages decentralized use of information by health workers, includes all service providers at all levels,
and integrates with and supports other information systems.
Over the past 10 years, a comprehensive primary health
care information system has been developed in South Africa.
One of its key elements is an essential data set, which may be
defined as a set of the most important data elements, selected
from all primary health care vertical programmes, that should
be reported by health service providers on a routine basis,
with the aim of being able to generate indicators that monitor
the provision of health services in an integrated manner. An
essential data set is thus important in that it contributes to
the principles listed above and facilitates decentralized use of
information by health workers because their monitoring needs
are clearly defined.
There are two key messages in this definition, contained
in the linked concepts of integration and an essential data set.
Programme managers (e.g. coordinators of the Expanded Programme on Immunization (EPI) and programmes for women’s
health, HIV/AIDS and sexually transmitted infections, and
tuberculosis), in an effort to ensure that all angles of service
delivery are taken into consideration, often require a very large
amount of information for their specific programmes. Their
primary concerns are their programme needs, and little attention is given to the means of collecting the information or
Manager, Health Information Systems Programme, School of Public Health, University of the Western Cape, South Africa (email: [email protected]).
Ref. No. 04-015032
Bulletin of the World Health Organization | August 2005, 83 (8)
Special Theme – Health Information Systems
Vincent Shaw
the needs of other programmes. The requirements of various
programmes may duplicate each other, and the vertical reporting of this information often requires separate data collection
tools. As a result, the health worker is faced with a myriad of
books and forms, all used to collect information for specific
managers, but with little integration and no vision of its use at
the local level. Experience has shown that the larger the number
of data elements to be reported upon, the poorer the quality of
the data (6, 7).
The creation of an essential data set is based upon two key
principles: limiting the routine reporting requirements for primary health care and hospital services to a set of 100–150 data
elements, enabling the calculation of 80–120 indicators; and
integrating the reporting requirements of various programme
managers, so that their needs are contained within the set of
essential data elements and indicators.
This article describes a process that resulted in the creation
of an essential data set at district level. This had a significant
impact on neighbouring districts and resulted in the development of a regional essential data set, which in turn helped to
influence the creation of a provincial data set.
Developing an essential data set at district
In a remote district in the northern part of the Eastern Cape
Province, the district management team found that the data collected by the clinic staff seemed inappropriate for the adequate
management of services. Until then, data had been submitted to
the head office on a routine (quarterly) basis, but no feedback
was ever received. The requirements for data submission had
been determined by head office staff many years earlier and had
not been revised to accommodate recent changes in the priorities for health service delivery.
The decision to review the data collected at facility level
took place in 1994, when the “new” South Africa was requiring an increasingly decentralized management structure and
greater transparency in terms of access to information and
health statistics. In addition, the focus had shifted significantly
from a hospicentric health service to a health service orientated
towards primary health care.
The district team evaluated all the services that they were
providing, and identified data elements or indicators that
would accurately monitor these services. This exercise included
a process of evaluating existing data elements being collected.
For each data element that health workers wanted to be included, they had to answer the questions: Why do we want
to collect this information? How will we use it? The answers
should underpin the need to monitor the integrated strategic
plan for the district. If the health workers could not come up
with a reasonable answer to either of the questions, the data
element was discarded or reformulated so that the questions
could be answered.
A long list of data elements and their associated indicators was developed. Then began a process of whittling away
at the list until about 70 elements remained, which were considered the most essential data elements required to calculate
about 75 indicators. This set of indicators was small enough
to allow management to focus on the key aspects of service
provision, yet was large enough to monitor services across all
programmes. It provided management with an integrated system for assessing services. Along with the data elements and
indicators, a set of data collection tools was developed.
Bulletin of the World Health Organization | August 2005, 83 (8)
Developing an essential data set for health information
Box 1. An essential data set must be able to accommodate
changes over time
There were a number of important modifications that came about
as a result of the process of establishing an essential data set. One
example that highlights how experience influenced the development
of the data set hinges on the data elements collected for antenatal
services. Two of the antenatal indicators were:
• antenatal coverage (first antenatal visit divided by the expected
number of pregnancies in the women of childbearing age);
• average antenatal visits per antenatal client (first antenatal visit plus
follow-up antenatal visit divided by the first antenatal visits).
The district management team found a very high antenatal coverage
rate in the district (112% for January–December 2000); with an
average of three antenatal visits per antenatal client, it was apparent
that the pregnant population was able to access the required services.
The next step was to seek to improve the quality of services by first
determining the percentage that accessed the services within the first
20 weeks of pregnancy and, if necessary, increase it. The team’s gut
feeling was that a very low percentage actually accessed services in
the first trimester. Hence a new, more specific indicator was introduced
and the existing data element “First antenatal visit” to be split into:
“First antenatal visit within first 20 weeks of pregnancy” and “First
antenatal visit after 20 weeks of pregnancy”.
Managing upwards: the district influences
the region
As the district implemented the new system, adjoining districts
came to learn about the new data set and its efficiency. Gradually, pressure from other areas within the region to implement
the same data set mounted; as a result a regionwide consultative meeting was convened, at which the district data set was
assessed and adapted to accommodate the needs of the region.
This, in turn, resulted in the region approaching the province
to reduce the number of indicators being reported upon; finally, after some time, the province approached the national
administration, and in June 2002 a national workshop adopted
an essential data set for the country.
Important lessons learned from this process
Reporting requirements must be able to change
over time
Changing needs of patients require changing reporting requirements, as evidenced by the emergence of the HIV/AIDS
epidemic and increasing access to antiretroviral therapy. In the
case of the HIV/AIDS epidemic, managers at the central level
have generally been responsive to the needs of patients and the
reporting requirements of donors. The reality is, however, that
these managers operate from positions of power and are able to
impose reporting requirements even if they have not been well
thought through. Managers at facility level may have a better
grasp of the needs of patients, but they have less influence in
making these needs known. The result is that once reporting
needs have been defined, they remain cast in stone.
Reviewing reporting requirements requires a broadly consultative process between managers of different programmes
in order to ensure that duplication is avoided. This is not easy
to achieve.
As managers use information, their understanding of its
meaning improves and their demands become more sophisticated. Bodart & Shrestha (8) describe four types of indicator
Special Theme – Health Information Systems
Developing an essential data set for health information
(count, proportion, rate and ratio). Our experience has shown
that managers are initially most comfortable identifying count
indicators (e.g. number of patients with hypertension). They
should be encouraged to look beyond this, however, to identify
how they would use this count indicator to improve service delivery. This often leads to the development of indicators which
bring in another data element as the denominator, creating one
of the other three indicator types. Box 1 provides an example
from which it can be seen that an atmosphere needs to be created where the review of reporting requirements is acceptable,
even the norm. This needs to be the case both at the central
and the peripheral levels.
Programme reporting requirements must be
integrated in order to ensure the development of
coherent information
There should be agreement among programme managers to
adhere to the principle of developing an integrated data set. In
the absence of this, health workers at the facility level are likely
to have to cope with uncoordinated and often duplicate demands for information that will result in their being distracted
from their primary function — providing health services. An
example would be where the nutrition programme requires
reporting on the number of children under five years of age
attending the clinic, number of children weighed, and number
of children with malnutrition. In order to be sure that facilities
report on these elements, the programme develops a data collection form specific to its needs. At the same time, EPI requires
reporting on children attending under two years of age, and
Vincent Shaw
immunizations given to them (BCG, DPT1-2-3, OPV1-2-3,
etc.). It also develops a report format specific to its needs. Both
these programmes are targeting the same population group,
without considering that, from a health care worker’s point of
view, the child who needs to be immunized also needs to be
weighed and given vitamin A, and in the clinic it is the same
health worker who provides all these services. Adopting an integrated approach would ensure that systems are developed which
complement each other and are appropriate to the manner in
which services are delivered.
An integrated data set also provides managers at facility
and district alike a clearly defined target to work towards, both
in terms of collecting and using the information.
A top-down consultative process establishes a
framework for review
The creation of an essential data set for South Africa began as
a bottom-up process. Decentralized districts are often less bureaucratic in nature than central systems, and more responsive
to the changing needs of patients. As a result, they are able to
see the need to revise reporting requirements and are able to
effect a change reasonably easily. It is also easier to bring different programme managers together at district level, to ensure
integrated reporting.
A more strategic position to adopt, however, would be
for a national ministry to take it upon itself to facilitate the
development of an essential data set for the country. A concept in support of this process is a hierarchy of information
needs (5, 9).
Fig. 1. The hierarchy of information needs
Information used by national programme managers
Information collected by
and used at facility level
Information required by national authority
Additional information required by district or municipal department
Additional information required by facility management
WHO 05.95
Bulletin of the World Health Organization | August 2005, 83 (8)
Special Theme – Health Information Systems
Vincent Shaw
A hierarchy of information needs operates at the following levels.
1. A national ministry determines an essential data set — this
is the minimum reporting requirement for all facilities and
health service providers in the country.
2. The next level of management (a region or province) adds
indicators that they believe they should collect in order to be
able to manage their services efficiently (e.g. in the example
in Box 1 the original indicator was split in order to obtain
more detailed information).
3. A district management team adds to the regional essential
data set indicators they believe are important to manage
their services (e.g. a district wants to improve the supervision
process in clinics, so monitors the percentage of planned
supervisory visits undertaken).
4. A facility develops an essential data set which includes indicators from the ministry, region and district, as well as their
own indicators (e.g. if facility staff want to confirm that
they carry an increasingly large workload of patients from
outside their official catchment area, they develop an indicator “Percentage head count attendances from outside the
catchment area”). The type of information important for a
facility management committee, and possibly for a district, is
not necessarily relevant at the national or regional levels.
Applying the concept of a hierarchy of information necessitates
adherence to the principle of only transmitting the information
that is required to the next level. With the advent of computers,
and patient-based information systems, the temptation to transmit all the information through to all levels must be resisted.
Additional information can be collected through
specific programme surveys
In order to ensure that the essential data set is limited in size, it
is necessary to provide programme managers with an alternative
mechanism for collecting programme data outside the routine
reporting system. This is done by using surveys.
Surveys should be used to gather information that will
complement the routine reporting. As some indicators do
not change much over time, they do not need to be reported
Developing an essential data set for health information
on monthly — they could be collected annually or quarterly
through the use of surveys. Typical contents of a survey questionnaire would be questions about quality of care, availability
of equipment, staffing and budget allocations. Surveys can be
used creatively to strengthen health services. For example, it
may be that in order to reduce the cost of surveys, a three-year
rolling plan is developed, ensuring that each year a third of
all facilities are surveyed: all are surveyed over the three-year
period. A survey data set would contain core information that
is common to all the years, and additional information could
be changed from year to year according to need.
Developing an essential data set provides managers with a clearly
defined set of indicators for monitoring and evaluating services.
The process of developing an essential data set can strengthen
the health services and the health information system because
it requires coordination of reporting requirements among programme managers, and the creation of a framework for reviewing information needs over time. Applying the concept of a
hierarchy of information needs allows each level within the
health service to develop its own data set, while still responding to the needs of the central administration. This encourages
the use of information at a local level because each level has
been involved in determining the indicators and data elements
that are collected. Annual surveys can be used to complement
routine reporting, enabling the essential data set to be kept to
a minimum. O
This article would not have been possible without the innovations developed in the Ukhalhamba District under the guidance
of Arthur Heywood, supported by the vision and leadership of
Jenny Brown, Elsabe Schlebusch and, later, Rudi Thetard. The
nurses who worked in the clinics in the Ukahlamba District
are thanked for their perseverance in testing the hierarchy and
in developing the initial set of indicators.
Competing interests: none declared.
Réforme du système d’information sanitaire en Afrique du Sud : mise au point d’un jeu de données
Les services de santé sont de plus en plus incités à développer
des systèmes d’information réagissant à l’évolution des besoins
sanitaires et répondant aux objectifs de service. La définition d’un
jeu de données essentielles fournit aux gestionnaires une série
d’indicateurs clairement définie permettant de surveiller et d’évaluer
les services. Le présent article décrit un processus aboutissant à
la création d’un jeu de données essentielles au niveau du district.
Cette opération a eu un impact important sur les districts voisins
et a conduit à la mise au point d’un jeu de données essentielles
pour la région, processus qui, à son tour, a contribué à l’élaboration
d’un jeu de données essentielles à l’échelle de la province, puis du
pays. Quatre enseignements peuvent être tirés de cette expérience.
La mise au point d’un jeu de données essentielles nécessite un
processus permettant d’ajuster les exigences de notification
Bulletin of the World Health Organization | August 2005, 83 (8)
au cours du temps en réponse à l’évolution des circonstances,
processus auquel elle peut en même temps contribuer. En outre,
elle facilite (et impose) l’intégration des exigences de notification
des programmes dans un système d’information cohérent. Bien
que l’étude de cas décrive une démarche partant de la base, un
processus consultatif descendant est préconisé car il fixe un cadre
dans lequel les besoins en information peuvent être analysés. Enfin,
la réalisation d’enquêtes peut contribuer aux efforts pour limiter
le plus possible le nombre d’éléments essentiels. En conclusion,
la mise au point d’un jeu de données essentielles participe au
renforcement des services de santé car elle implique un dialogue
entre les directeurs de programmes et la définition d’indicateurs
que ces derniers doivent surveiller.
Special Theme – Health Information Systems
Developing an essential data set for health information
Vincent Shaw
Reforma de los sistemas de información sanitaria en Sudáfrica: desarrollo de un conjunto de datos
Los servicios de salud se encuentran sometidos a una presión cada
vez mayor para desarrollar sistemas de información sensibles a
las nuevas necesidades de salud y apropiados para alcanzar los
objetivos fijados. Un conjunto de datos esenciales puede dotar
a los administradores de un abanico claramente definido de
indicadores para vigilar y evaluar los servicios. En el presente
artículo se describe un proceso que permitió crear un conjunto
de datos esenciales a nivel de distrito. Ello tuvo importantes
repercusiones en los distritos vecinos y condujo a desarrollar un
conjunto de datos esenciales de ámbito regional, lo cual influyó a
su vez en la creación de un conjunto de datos esenciales provincial
y más tarde nacional. Cabe extraer del proceso cuatro importantes
lecciones. El desarrollo de un conjunto de datos esenciales exige y
al mismo tiempo puede favorecer un proceso que permita ajustar
las necesidades de información con el tiempo en respuesta a la
evolución de las circunstancias. Además, propicia (y requiere)
la integración de los requisitos programáticos en materia de
presentación de informes en un sistema de información coherente.
Mientras el estudio de casos describe un enfoque ascendente,
aquí se preconiza un proceso consultivo descendente, pues así se
establece un marco en el que pueden analizarse las necesidades
de información. Por último, la utilización de encuestas puede
ser una ayuda para reducir al mínimo los elementos esenciales.
En conclusión, el desarrollo de un conjunto de datos esenciales
contribuye a fortalecer los servicios de salud porque requiere que
haya diálogo entre los gestores de los programas y define los
indicadores que éstos deberán vigilar.
1. Green A. An introduction to health planning in developing countries. Oxford:
Oxford University Press; 1999.
2. White Paper on the transformation of health services. Durban: South African
Department of Health; 1997. Available from:
3. Information support for new public health action at district level. Report of
a WHO Expert Committee. Geneva: World Health Organization; 1994. WHO
Technical Report Series, No. 845.
4. Campbell B, Adjei S, Heywood A. From data to decision making in health:
the evolution of a health management information system. Amsterdam:
Royal Tropical Institute; 1996.
5. Heywood A, Rohde J. Using information for action: a manual for health
workers at facility level. South Africa: EQUITY Project.
6. Stoops N, Williamson L. Using health information for local action: facilitating
organisational change in South Africa. In: Krishna S, Madon S, editors. The
digital challenge: information technology in the development context.
Gateshead: Athenaeum Press Ltd.; 2003.
7. Williamson L, Stoops N. Using information for health. In: South African Health
Review 2001. Available from:
8. Bodart C, Shresta L. Identifying information needs and indicators. In:
Lippeveld T, Suaerborn R, Bodart C, eds. Design and implementation of
health information systems. Geneva: WHO; 2000:Chapter 4.
9. Braa J, Hedberg C. The struggle for district-based health information systems
in South Africa. The Information Society 2002; 18:113-27.
Bulletin of the World Health Organization | August 2005, 83 (8)
Round Table
Round Table Discussion
Round Table Discussion
Information is not only for managers
Richard E. Cibulskis1
The product described by Shaw — a simplified health information system implemented on a national scale — is not new (1,
2). What is interesting is the bottom-up process, as it is quite
unusual for systems originating in one district to take hold over
a wide geographical area. This may be because some districts
are reluctant to use a system designed by other districts or because they do not have the same resources for implementation.
Whatever the circumstances, bottom-up processes can produce
a variety of incompatible information systems, each competing
to be adopted as a national standard. South Africa does not
appear to have succumbed to this problem, but it is still not
clear if its system really works. It would be instructive to learn
more about its reporting rate from institutions. This is a good
indicator of an information system’s performance as it requires
several basic processes to be fulfilled, such as a complete listing
of reporting units, compliance with reporting requirements and
processes for monitoring compliance. A good reporting rate is
also critical to the eventual interpretation of indicators.
Whether a top-down or bottom-up approach is preferred,
the design of health information systems requires a clear understanding of why data are being collected; ultimately they
should influence the behaviour of those in control of resources
in ways that will enable the health sector to achieve its objectives. Data should certainly be used by health managers to plan
and monitor programmes, enabling them to allocate resources
to priority health problems or populations. The information
required for this task, however, is wider than that provided by
routine health information systems. Some relevant data systems
— for population, finances, and staffing — are managed by
other government departments, while some information is collected through censuses or surveys (particularly if many services
are delivered through non-government providers). An optimal
information strategy needs to consider how the different sources
of information will work together. For example, it should be
clear about the definitions of indicators and the coding systems
used for geographical units. Some form of centrally coordinated
approach seems inevitable. Such coordination should not be
mistaken as being set up for the purpose of supplying central
level managers with information: although they might benefit
from information, they rarely have the capacity or authority to
respond to large quantities of data. Rather, central coordination
is primarily to bring together data from districts so that they can
be summarized in such ways that districts can compare their
performance with that of others.
Health managers are often ineffective users of information, despite efforts to train them, encourage them or provide
them with new information systems. Time and again, inequities or inefficiencies in the use of resources go unheeded. This
may be because health managers have little influence over key
decisions in government or perhaps because they are not motivated to respond. There is a growing awareness that if we are
interested in enhancing the performance of the health sector
then external uses of information can carry greater weight than
internal uses (3). Thus, public disclosure of information can help
“politicians, patients and citizens to scrutinize the operations
which they are financing” (4) and in this role it can encourage
managers to be more responsive to their clients’ needs. Similarly,
information can be used by health managers to lobby external
authorities for greater support. If external uses of information
have greater impact than internal uses, should not the health
sector take this into account when designing its minimal data
sets? It may be that the minimal data set for politicians, citizens
and the treasury are the same as that for health managers, but
not necessarily so. Citizens may be more interested in learning whether basic inputs such as staff and drugs are available,
whereas a ministry of finance may be interested in learning
whether national development priorities are being delivered.
Whatever the final content, minimal data sets need to begin
with the key users and uses of information and they should not
remain the preserve of health managers. O
Competing interests: none declared.
1. Lippeveld T, Sauerborn R. A framework for designing health information
systems. In: Lippeveld T, Sauerborn R, Bodart C, editors. Design and
implementation of health information systems. Geneva: WHO; 2000:15-32.
2. Cibulskis RE, Hiawalyer G. Information systems for health sector monitoring in
Papua New Guinea. Bulletin of the World Health Organization 2002;80:752-8.
3. The World Bank. World development report 2004 – Making services work
for poor people. Washington (DC): the World Bank; 2003.
4. Smith P. The use of performance indicators in the public sector. Journal of
the Royal Statistical Society Series A 1990;153:53-72.
Indicators for a health information data
set in Ghana
Sam Adjei2
A health information system handles the recording, storage,
retrieval and processing of health data. Broadly defined, the
health information system should cover such data sources as
vital registration, censuses, routine service-generated statistics,
population-based surveys and research information, in order to
provide evidence for decision-making in the health system.
Assessments of health information systems have given
rise to several misgivings. Foremost among them is the fact that
multiple data sources are not linked to each other; indeed, different instruments may generate different data on the same person or event. Routine service data are collected with the needs
of higher-level programme managers and donors in mind; in
addition, they may be incomplete or of doubtful quality, and
timeliness can be a problem. Surveys are useful, but they tend
to be expensive and donor driven and are often not linked to
routine service data. Research data are generally available but
are rarely included as part of the health information system
because research is conducted outside the scope of ministries
of health. Dissemination of the information collected is usually
Monitoring and Evaluation Advisor, World Bank Office Jakarta, Jakarta Stock Exchange Building Tower 2, 12th floor, Jl Jenderal Sudirman Kav 52–53, Jakarta 12190,
Indonesia (email: [email protected]).
Deputy Director-General, Ghana Health Service, Private Mail Bag, Ministries, Accra, Ghana (email: [email protected]).
Bulletin of the World Health Organization | August 2005, 83 (8)
Round Table
Round Table Discussion
Fig. 1. Conceptual framework for health sector reforms and information
Health status
Availability and utilization
of health interventions
Availability and utilization
of health-related interventions
Package of health interventions
WHO 05.98
weak and its use, particularly in policy-making, is infrequent.
Several efforts undertaken to strengthen health information
have not taken into account any general framework for designing the information system.
The paper by Shaw describes efforts to correct the multiplicity of data sets in South Africa, especially at the periphery,
without describing an overall framework for how this is to be
achieved. Even though the process reported has reduced the data
set to 100–150 elements and 80–120 indicators, the numbers
still appear too large to manage effectively. An overall vision
of health information needs in the context of health development is important even at the district level beyond programme
managers. The current wave of health sector reforms and health
system strengthening will require this broader context for the
development and standardization of health information.
In Ghana, a conceptual framework for health sector development (health sector reforms) helped to generate indicators
for health information (see Fig. 1). This framework allows data
required for policy development, priority setting and programme performance measurement, as well as monitoring and
evaluation, to be determined in the sector as a whole. Sectorwide indicators that fell into three main categories were agreed
upon; 20 indicators are collected and used at all levels, which
does not exclude the use of more indicators at any level.
The three categories of indicator concern: health status,
including mortality and morbidity; programme output, covering programme performance in public health and clinical care
interventions as well as health-related indicators such as enrolment at school; and systems development, in which a package
of five cross-cutting areas of access to care, quality of care, efficiency in the use of resources, collaboration with other sectors
(communities, other providers of care, other ministries and
donors ) and financing of care is determined. Indicators in the
first two categories are easily developed but are more difficult to
define in the third category, where methods of data collection
are also difficult as the indicators do not lend themselves to
routine service statistics and surveys may be needed.
The impact of the process adopted in the South African
experience is commendable, as it appeared to influence other
districts, the regions and national levels. Its impact at the global
level is not indicated but, given that global initiatives and donors
have major information requirements, a process that links with
global development is important.
With this in view, the Health Metric Network initiative
is timely. Particularly welcome is the development of a simple
framework to define the scope of the health information
system. The framework should serve as a diagnostic tool for
evaluating the state of a country’s health information system, a
road map for developing plans for improvement, and a process
for monitoring and evaluating progress. Its application at the
country level should build on experiences such as that described
in South Africa. O
Competing interests: None declared.
A data warehouse approach can manage
multiple data sets
Jørn Braa1
Development of essential national indicators and data sets — or
national standards — is regarded as the key issue in country
health information system reform. Most countries, however, fail
to achieve this goal. The reasons are: fragmentation (difficulty
in reaching agreement on standards across health programmes);
focus on reporting rather than on use of data and information;
constantly changing needs (e.g. with regard to HIV/AIDS); and
standards that are “cast in stone” (software and paper tools are
difficult to change).
Department of Informatics, University of Oslo, PO Box 1080, Blindern, Norway (email: [email protected]
Bulletin of the World Health Organization | August 2005, 83 (8)
Round Table
Round Table Discussion
Despite these problems, South Africa has managed to
develop national standards that are flexible enough to “absorb”
local innovations and changes over time. The following points
may help to explain why this is so.
• The “hierarchy of standards” (“information needs” in Shaw’s
figure) has been a powerful tool to negotiate a balance
between the national needs for control with the local (e.g.
province or health programme) needs for flexibility or more
data. While all health units are required to collect and report
the core national data, they are at the same time allowed to
collect their own additional data.
• Use of information is highlighted by linking data sets to
targets and indicators.
• The flexible approach to standards following the hierarchy
makes it easy to absorb and implement changes over time;
there is no “final” data set. Local innovations are allowed for
and may eventually be included in the national data set.
• The flexibility of the South African District Health Information System (DHIS) database application is crucial to
managing the ever changing national and local data sets.
Data elements, indicators and data sets are added, edited and
managed by the health services themselves, thus making it
possible to manage multiple data sets at district level. This
“data repository” or “warehouse” approach may be a key to
how the lessons from South Africa could be applied in other
The data set must focus on service quality
Jens Byskov 1 & Oystein Evjen Olsen2
Over the years, I have been involved in efforts to apply the
South African lessons in many countries. It has not been easy.
National health information system databases and reporting
formats tend to be rigid and unable to respond to changes, thus
leading to fragmentation of the system. The HIV/AIDS programmes are currently aggravating this situation.
So what can be done? Current efforts to establish integrated data sets in contexts as different as Addis Ababa (Ethiopia),
Botswana, Zanzibar (United Republic of Tanzania), Andhra
Pradesh (India) and Ho Chi Minh City (Viet Nam) may provide some answers. Here data sets from all or most programmes
are combined and streamlined by sorting out overlaps, gaps
and inconsistencies. Following the South African district data
warehouse approach, the combined essential data set is then
further improved and reduced by focusing on the need-toknow indicators. Programme-specific software applications are
linked electronically to the DHIS, thus providing a shared data
repository. The objective of integrating all indicators relevant
to the Millennium Development Goals has proven important
in building consensus.
Purists may argue that the data warehouse approach based
on flexible standards advocated here is violating the spirit of the
South African minimum essential data set approach, by taking a
combined — maximum — data set as the point of departure. It
may, however, be the most appropriate way to apply the South
African indicator-driven approach in a situation increasingly
dominated by strong programmes and multiple uncoordinated
data sets and software applications. O
The paper by Vincent Shaw highlights some of the long awaited
practical approaches to ensure relevance and use of health information systems in developing countries. The South African
experience he recounts is very relevant as current “best practice”. It is a very important step forward in the simplification
and integration of programme areas and routine services into
a shared essential set of routine data. The cohesion of health
services and the whole system will be much strengthened by
such a shared data reference and information base.
The essential data are still to be selected by programme
managers, however, even though the definition of the data
set emanated from the district level. It is not shown how the
data will be turned into useful information that will assist in
planning and monitoring at all levels of the health system. The
examples taken from the nutrition programme and the Expanded Programme on Immunization only exemplify health
status (outcome) and service provision (output) data.
In the section on specific programme surveys, resource
and staff availability are mentioned as well as service quality.
It is worrying that these are not shown to be included in an
essential routine data set, as it is extremely important that an
essential data set allows production and sharing of information
on health management. It is also not shown whether quality of
care will be viewed from both provider and user perspectives
and whether user views on service priorities and other qualitative data are included.
In Shaw’s figure, the column of information used by the
national level on a routine basis within the triangle of information should be seen as the core data on facility performance that
must be shared in an accessible database for the whole health
service and other parts of the health system, and be available
for sharing with users and the public.
The information needs triangle could also be depicted
as service quality at the bottom, supervision and coordination
needs at the intermediate level, and policy needs at the top.
The main focus of the data set must be on service quality, with
less emphasis on supervision and coordination and even less
on policy. Service quality must relate to health management,
service output and outcomes as viewed from both the provider
and the user sides.
The number of data elements and indicators still seems
to be high in relation to similar elements in the core part of
the health information system in other African countries and
may indicate a still limited degree of compromise and shared
focus between the levels and programmes. A stronger emphasis
is needed on the iterative nature of the health information
system, and not so much on annual or quarterly “reporting”.
We need to move away from a culture of reporting to a culture
of using the data for ourselves — facilities or districts first
— on a continuous basis. Benchmarking and quality assurance
processes exemplify some useful frameworks for continuous
use of data. O
Competing interests: none declared.
Competing interests: none declared.
DBL-Institute for Health Research and Development, Jaegersborg Allé 1D, Charlottenlund, DK-2920 Denmark (email: [email protected]). Correspondence should be
sent to this author.
DBL-Institute for Health Research and Development, Primary Health Care Institute, Iringa, PO Box 105297, Dar es Salaam, United Republic of Tanzania
([email protected]).
Bulletin of the World Health Organization | August 2005, 83 (8)
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

Related manuals

Download PDF