Chronic Diseases in Canada Inside this issue

Chronic Diseases in Canada Inside this issue
Chronic Diseases in Canada
Volume 29 · Number 3 · 2009
Inside this issue
96
Validation of perinatal data in the Discharge
Abstract Database of the Canadian Institute for
Health Information
K. S. Joseph, J. Fahey for the Canadian Perinatal
Surveillance System
102 Validity of autism diagnoses using administrative
health data
L. Dodds, A. Spencer, S. Shea, D. Fell, B. A. Armson, A. C. Allen,
S. Bryson
108 Associations between chronic disease, age and
physical and mental health status
W. M. Hopman, M. B. Harrison, H. Coo, E. Friedberg, M. Buchanan,
E. G. VanDenKerkhof
118 Statistical modelling of mental distress among
rural and urban seniors
C. P. Karunanayake, P. Pahwa
128 Factors associated with the adoption of a smoking
ban in Quebec households
É. Ouedraogo, F. Turcotte, M. J. Ashley, J. M. Brewster, R. Ferrence
136 Myalgic Encephalomyelitis/Chronic Fatigue
Syndrome program
E. Stein, M. MacQuarrie
140 Book review – Dissonant disabilities: women with
chronic illnesses explore their lives
M. Rezai
Chronic Diseases in Canada
a publication of the Public Health Agency
of Canada
Howard Morrison
Principal Scientific Editor
(613) 941-1286
Robert A Spasoff
Associate Scientific Editor
Claire Infante-Rivard
Associate Scientific Editor
Elizabeth Kristjansson
Associate Scientific Editor
Michelle Tracy
Managing Editor
CDIC Editorial Board
Jacques Brisson
Laval University
Neil E Collishaw
Physicians for a Smoke-Free Canada
James A Hanley
McGill University
Clyde Hertzman
University of British Columbia
C Ineke Neutel
University of Ottawa Institute on
Care of the Elderly
Kathryn Wilkins
Health Statistics Division
Statistics Canada
Chronic Diseases in Canada (CDIC) is a quar­
terly scientific journal focussing on cur­rent
evidence relevant to the control and pre­
vention of chronic (i.e. non-communicable)
diseases and injuries in Canada. Since 1980
the journal has published a unique blend of
peer-reviewed feature articles by authors
from the public and private sectors and
which may include research from such fields
as epidemiology, public/community health,
bio­statistics, the behavioural sciences, and
health services or economics. Only feature
articles are peer reviewed. Authors retain
responsibility for the content of their arti­
cles; the opinions expressed are not neces­
sarily those of the CDIC editorial committee
nor of the Public Health Agency of Canada.
Chronic Diseases in Canada
Public Health Agency of Canada
785 Carling Avenue
Address Locator 6805B
Ottawa, Ontario K1A 0K9
Fax: (613) 941-9502
E-mail: [email protected]
Indexed in Index Medicus/MEDLINE
To promote and protect the health of Canadians through leadership, partnership, innovation and action in public health.
— Public Health Agency of Canada
Published by authority of the Minister of Health.
© Her Majesty the Queen in Right of Canada, represented by the Minister of Health, 2009
ISSN 0228-8699
This publication is also available online at www.publichealth.gc.ca/cdic
Également disponible en français sous le titre : Maladies chroniques au Canada
Validation of perinatal data in the Discharge Abstract
Database of the Canadian Institute for Health Information
K. S. Joseph, MD, PhD (1) and J. Fahey, MMath (2) for the Canadian Perinatal Surveillance System
Abstract
We compared perinatal information submitted to the Canadian Institute for Health
Information (CIHI) hospitalization database with information submitted to the Nova
Scotia Atlee Perinatal Database (NSAPD) in order to assess the accuracy of the CIHI
data. Procedures such as Caesarean delivery were coded accurately (i.e. sensitivity of
99.8%; specificity of 98.7%). Postpartum hemorrhage, induction of labour and severe
intraventricular hemorrhage also had sensitivity and specificity rates above 85% and
95%, respectively. Some diagnoses, defined differently in the two databases, were less
accurately coded, e.g. respiratory distress syndrome (RDS) had a sensitivity of 50.9% and
a specificity of 99.8%. Restriction to more severe forms of the disease improved accuracy,
e.g. restriction of RDS to severe RDS in the NSAPD and identification of severe RDS in the
CIHI database, using codes for RDS and intubation, resulted in a sensitivity of 100% and
a specificity of 99.6%. Our study supports the use of CIHI data for national surveillance
of perinatal morbidity, with the caveat that an understanding of clinical practice and
sensitivity analyses to identify robust findings be used to facilitate inference.
Key words: perinatal, surveillance, database, maternal, infant, morbidity, respiratory
distress, discharge abstract data
Introduction
Perinatal health surveillance in Canada
relies on data from various sources,
including vital statistics databases and
hospital discharge databases.1–3 Although
vital statistics data on births and fetal and
infant deaths remain an important source for
documenting temporal trends and regional
variations in perinatal health, the decline
in mortality rates in recent decades has
shifted the focus of perinatal surveillance
increasingly towards monitoring trends
and patterns in serious morbidity. This is
particularly true with regard to serious
maternal morbidity4 and serious neonatal
morbidity.5,6
The quality of hospitalization data
from the Canadian Institute for Health
Information (CIHI) Discharge Abstract
Database (DAD) is an important concern.
Although previous studies have concluded
that this data source is suitable for sur­
veillance purposes,7 a recent medical
chart re-abstraction study, commissioned
by the Canadian Perinatal Surveillance
System (CPSS) and carried out by CIHI,
showed variable quality with regard to
several indicators of perinatal health.8 The
study on hospital discharges in 1999/2000
included hospitals selected randomly after
stratifying by geography and size and type
of hospital. The charts of 385 newborns
and 872 mothers were re-abstracted by
CIHI classification specialists and compared
with information in the DAD. False positive
rates for Caesarean, vacuum and forceps
deliveries, preterm labour and episiotomies
were < 1%; however, the false positive rates
for other indicators were high (e.g. medical
induction of labour, 12.8%; third degree
perineal lacerations, 40.3%). Similarly,
false negative rates were low for some
indicators (e.g. < 1% for Caesarean
deliveries, third degree lacerations, respi­
ratory distress syndrome (RDS), preeclampsia / eclampsia), but high for other
indicators (e.g. fetal asphyxia/fetal distress,
23.6%; medical induction of labour,
38.3%; rare neonatal conditions, 41.3%;
rare congenital conditions, 53.9%).
Two concerns were voiced within the
Canadian Perinatal Surveillance System
(CPSS) Maternal Health and Fetal and
Infant Health Study Groups who routinely
use CIHI data for perinatal surveillance
purposes.9–13 First, detailed analyses of
phenomena such as amniotic fluid embo­
lism,12 postpartum hemorrhage13 and other
conditions9-11 have shown patterns that
are congruent with clinical expectation,
and suggest a higher level of data quality
than that observed in the re-abstraction
study. Another reason for questioning the
high error rates in the re-abstraction study
was related to the technical aspects of
the stratified sampling (and the weighted
calculation of population rates of false
positive and false negative errors). Several
error rates were relatively small in the
study sample, but were substantially
inflated after the population weights
were applied. The most serious inflation
was observed for third degree perineal
Author References
1 Perinatal Epidemiology Research Unit, Departments of Obstetrics & Gynaecology and Pediatrics, Dalhousie University and the IWK Health Centre, Halifax, NS, Canada.
2 Reproductive Care Program of Nova Scotia, Halifax, NS, Canada.
Correspondence: K.S. Joseph, MD, Division of Neonatal-Perinatal Medicine, IWK Health Centre, 5980 University Avenue, Halifax, NS B3K 6R8,
Tel.: 902-470-6652; Fax: 902-470-7190; E-mail: [email protected]
Chronic Diseases in Canada
96
Vol 29, No 3, 2009
lacerations (i.e. false positive rate in the
sample, 5.6%; in the population, 40.3%);
rare neonatal conditions (i.e. false negative
rate in the sample, 1.6%; in the population,
41.3%); and rare congenital conditions (i.e.
false negative rate in the sample, 3.1%; in
the population, 53.9%).
Consequently, we decided to reassess
the accuracy of the DAD information by
comparing the perinatal data in the DAD
with data in the NSAPD. The latter database,
which is smaller and clinically focused, is
believed to have a relatively high degree
of accuracy with regard to diagnoses and
procedures.
Methods
During a brief period in 2002, the perinatal
data pertaining to newborns and mothers
in Nova Scotia were simultaneously coded
for both the DAD and the NSAPD. Although
the coding rules for the two databases were
different, the availability of data under the
two independent systems for 6194 mothers
and 6315 newborns permitted an evaluation
of the DAD data for quality assessment
purposes. The mothers and newborns
included in this duplicate coding were not
selected by gestational age or outcome,
and represented all of the deliveries/ births
that occurred during a specific period.
The NSAPD is a clinically focussed,
population-based database that stores
detailed information from antenatal and
medical charts. The information is
extracted by trained personnel using stan­
dardized forms. An ongoing data quality
assurance program, which carries out
periodic abstraction studies, has shown
the database information is reliable. The
database has been used to validate the vital
statistics, birth-and-infant-death-linked files
at Statistics Canada.14,15 Perinatal infor­
mation for the DAD was also collected by
trained personnel in Nova Scotia under
the CIHI data abstraction rules. For the
period in question, the data were coded
using International Classification of
Diseases Revision 10 (ICD-10-CA) codes for
diagnostic information and the Canadian
Classification for Health Interventions (CCI)
codes for interventions/procedures.
Vol 29, No 3, 2009
We compared diagnoses and interventions/
procedures of interest between the two data­
bases, assuming the NSAPD represents
the gold standard. Rates of sensitivity
(i.e. proportion of true [NSAPD] positives
identified as being positive by the DAD) and
specificity (i.e. proportion of true [NSAPD]
negatives identified as being negative by
the DAD) were calculated along with exact
binomial 95% confidence intervals (CI).
An evaluation of gestational age estimates
from the two sources was also carried out
using agreement statistics (i.e. weighted
kappa and the intraclass correlation coef­
ficient). For this analysis, gestational
age was grouped into clinically relevant
prognostic categories routinely used by the
Canadian Perinatal Surveillance System
(< 20, 20 to 21, 22 to 23, 24 to 25, 26 to
27, 28 to 31, 32 to 33, 34 to 36, 37 to 41, 42
to 45 weeks and unavailable). The specific
diagnoses and interventions/procedures
identified for assessing the accuracy of
the DAD data were based on clinical
and public health relevance and on the
definitional compatibility of the diagnoses
and interventions/procedures in the
two databases.
Results
According to the DAD, the rate of preterm
delivery (i.e. proportion of women with
information on gestational age who
delivered prior to 37 completed weeks)
was 9.1% (95% CI 8.4 to 9.9%), whereas
this rate was 8.8% (95% CI 8.1 to 9.6%)
according to the NSAPD. The rate of
postterm delivery (i.e. delivery at or after
42 completed weeks) was 0.6% (95%
CI 0.5 to 0.9%) and 2.1% (95% CI 1.8
to 2.5%) according to the DAD and the
NSAPD, respectively. No gestational age
information was stated for 54 women in
the NSAPD and for 7 women in the DAD
(i.e. 47 women with missing gestational
age in the NSAPD had a gestational age
between 37 and 41 weeks, according to
the DAD). Of the 543 women who had a
preterm delivery according to the NSAPD,
495 were coded as having delivered preterm
according to the DAD (i.e. sensitivity of
91.2%). Of the 5597 women who delivered
at term or postterm gestation according to
the NSAPD, 5531 were coded as having
97
delivered at term or postterm gestation
according to the DAD (i.e. specificity of
98.8%). A detailed examination of the data
on preterm delivery showed that 64 (i.e.
97%) of the 66 women coded as having
delivered preterm by the DAD (but at term/
postterm by the NSAPD) were at 36 weeks
gestation according to the DAD, i.e. a
large proportion of the false positive errors
were due to a minor one-week difference
in gestational age. Similarly, 31 of 48 (i.e.
65%) of the women who delivered at
preterm gestation according to the NSAPD,
but at term or postterm gestation according
to the DAD, were at 37 weeks of gestational
age at delivery according to the DAD. The
weighted kappa statistic assessing the agree­
ment between gestational age from the DAD
data and gestational age from the NSAPD data
was 0.75 (i.e. 95% CI 0.72 to 0.78), and the
intraclass correlation coefficient was 0.86
(i.e. 95% CI 0.83 to 0.88).
Table 1 shows the sensitivity and specificity
rates for several maternal health indicators.
Most indicators in the DAD showed a high
degree of accuracy. Sensitivity rates of
85% to 90% were noted for blood transfu­
sions, induced labour and any gestational
hypertensive disorders. There was good
agreement between the many specific types
of induction procedures/agents coded in
the DAD and induction of labour coded
in the NSAPD. For example, the 29 women
who, according to the DAD, had labour
induced through a combination of routes
and involving the use of an oxytocic agent,
were also coded as having their labour
induced according to the NSAPD. The main
discrepancy between the two databases
occurred for cases where the DAD coded
labour as having been induced by artificial
rupture of membranes, whereas the NSAPD
coded many such cases (i.e. 45 of 191) as
having artificial rupture of membranes,
but not as having been induced. Since
artificial rupture of membranes after the
onset of labour does not constitute labour
induction, this discrepancy probably reflects
a coding error in the DAD (i.e. assuming
the information in the NSAPD is correct).
Discrepancies were also noted for the
diagnosis of hypertension in pregnancy.
This was expected, because of the varied
subtypes of hypertensive disorders in
Chronic Diseases in Canada
Table 1
Validity of maternal data from the Discharge Abstract Database of the Canadian Institute for Health Information, using data
from the Nova Scotia Atlee Perinatal Database as the gold standard (based on 6194 mothers, Nova Scotia, 2002)
Indicator
Sensitivity (%)
95% CI
Specificity (%)
95% CI
Preterm delivery (< 37 weeks)
91.2
88.5
to
93.4
98.8
98.5
to
99.1
Postpartum hemorrhage
90.2
86.2
to
93.3
98.2
97.8
to
98.5
Blood transfusion
85.7
42.1
to
99.6
99.8
99.6
to
99.9
Induction of labour
89.2
87.7
to
90.6
96.9
96.4
to
97.4
Caesarean delivery
99.8
99.5
to
100.0
98.7
98.3
to
99.0
Perineal laceration to – 1st degree
– 2nd degree
91.7
89.7
to
93.3
97.9
97.4
to
98.4
97.7
96.8
to
98.3
99.1
98.7
to
99.4
– 3rd degree
97.1
92.7
to
99.2
99.9
99.8
to
100.0
– 4th degree
94.7
74.0
to
99.7
99.9
99.8
to
100.0
Hypertension - chronic
83.3
73.6
to
90.6
99.9
99.8
to
100.0
Gestational hypertension with proteinuria
(vs. severe PIH or HELLP)
75.2
67.5
to
81.8
99.5
99.3
to
99.7
Any gestational hypertensive disorder
(vs. mild or severe PIH or HELLP)
87.9
85.0
to
90.4
99.6
99.4
to
99.8
Preterm delivery refers to women delivering before 37 completed weeks of gestation.
PIH denotes pregnancy induced hypertension.
HELLP denotes hemolysis, elevated liver enzymes and low platelets count syndrome.
pregnancy and the different labelling/
classification schemes used by the NSAPD
and the ICD-10-CA system.
Table 2 presents the assessment of infant
health indicators among the 6315 newborns.
The false negative rate for bacterial sepsis
was high (i.e. sensitivity of 38.4%). This
rate improved when cases in the DAD were
also identified using adult codes for sepsis in
addition to newborn codes. The more serious
grades of intraventricular hemorrhage and
fracture of the clavicle were accurately
coded in the DAD, while RDS had the same
discrepancies as hypertensive disorders of
pregnancy (i.e. RDS classification is highly
detailed in the NSAPD and differs from
the diagnostic entities in the ICD-10-CA
system). Nevertheless, combining an ICD10-CA code of RDS with a procedure code
of intubation in the DAD resulted in virtual
agreement with a diagnosis of severe RDS
in the NSAPD. Severe RDS in the NSAPD
refers to RDS requiring assisted ventilation.
Fetal/birth asphyxia was essentially not
coded in the DAD (i.e. sensitivity of 14.3%,
Table 2).
Chronic Diseases in Canada
Discussion
Our study confirmed that major procedures
such as Caesarean delivery were coded
accurately in the DAD of CIHI. It also
showed that the information on more
minor diagnoses (e.g. first to fourth degree
perineal lacerations) and more challenging
diagnoses and procedures (e.g. induction
of labour, which is easily confused
with augmentation of labour) was also
reasonably accurate. Similarly, gestational
age—a difficult entity to capture accurately,
given different methods of ascertaining
gestational age—showed a relatively high
degree of agreement between the two
sources. The overall preterm birth rates
were non-significantly higher in the DAD
compared to the NSAPD, whereas postterm
birth rates were significantly lower. Although
these differences were relatively small,
the direction of the differences suggests a
greater influence of early ultrasound (and,
hence, greater accuracy) on gestational
age estimates in the DAD. In comparison
with menstrual date-based estimation of
gestational age, early ultrasound dating
tends to slightly increase preterm birth
rates and substantially lower postterm
birth rates.16–18 The disagreements in the
two databases arose mainly with regard
to variably defined diagnostic entities, such
98
as bacterial sepsis and RDS. Nevertheless,
identifying bacterial sepsis using both
neonatal and adult codes for sepsis and
confining RDS to “any respiratory distress”
or to a severe form of RDS resulted in rela­
tively accurate information for the DAD.
The serious discrepancy between the
NSAPD and the DAD with regard to fetal/
birth asphyxia is to be expected and is not
a reflection of data inaccuracy in the DAD.
Studies have shown that this diagnostic
label has essentially disappeared19 from the
clinical lexicon due to malpractice concerns,
even though the clinical entity remains
essentially unchanged in its frequency.20
Thus, it is rarely captured in the DAD
system, where coders identify cases only if
the term “asphyxia” is documented in the
medical chart, while the NSAPD continues
to identify the condition based on its
clinical components.
This study was carried out using medical
records from 2002, a time when Nova Scotia
first began to implement the ICD-10-CA
coding in its hospitals. Since the ICD-10-CA
system is substantially different from the
previous version, some coding errors may
have occurred in 2002 that would have been
resolved with continued use. One possible
example of this may be seen with regard
Vol 29, No 3, 2009
Table 2
Validity of neonatal data from the Discharge Abstract Database of the Canadian Institute for Health Information, using data from
the Nova Scotia Atlee Perinatal Database as the gold standard (based on 6315 live births, Nova Scotia, 2002).
Indicator
Sensitivity (%)
95% CI
Specificity (%)
95% CI
99.7
99.5 to 99.8
Bacterial sepsis
38.4
28.1 to 49.5
Bacterial sepsis (adult/neonatal codes)
67.4
56.5 to 77.2
99.6
99.4 to 99.8
Intraventricular hemorrhage, grade 3, 4
88.9
51.8 to 99.7
100.0
99.9 to 100.0
Fracture of clavicle
91.7
61.5 to 99.8
100.0
99.3 to 100.0
Respiratory distress – any (vs. any)*
94.2
90.8 to 96.6
96.6
96.1 to 97.1
– RDS (vs. RDS)
50.9
43.1 to 58.6
99.8
99.7 to 99.9
– RDS (vs. severe RDS)
96.3
89.6 to 99.2
99.6
99.4 to 99.8
– any RDS + intubation
(vs. severe RDS)
100.0
95.5 to 100.0
99.6
99.4 to 99.8
14.3
6.4 to 26.2
99.3
99.1 to 99.5
Fetal/birth asphyxia
* Any code vs. any code refers to an evaluation of any respiratory distress code in CIHI vs. any respiratory distress code in the NSAPD.
RDS vs. RDS refers to a respiratory distress syndrome code in CIHI vs. a respiratory distress syndrome code in the NSAPD.
RDS vs. severe RDS refers to a respiratory distress syndrome code in CIHI vs. a severe respiratory distress syndrome code in the NSAPD.
Any RDS + intubation refers to an evaluation of a respiratory distress syndrome code in CIHI plus an intubation code vs. a severe respiratory distress syndrome code in the NSAPD.
to the bacterial sepsis of the newborn codes,
which had a poor sensitivity rate arising
at least partly because adult sepsis codes
were used. Presumably, such errors would
have been corrected as familiarity with the
ICD-10-CA system increased.
The improved diagnostic accuracy of
diseases, such as RDS in the DAD with
the restriction to more severe forms of the
disease, suggests that researchers using
these large hospitalization databases should
routinely carry out sensitivity analyses to
assess the robustness of their findings.
Familiarity with the clinical culture and
changes in clinician habits (e.g. knowledge
of the declining use of terms, such as
birth asphyxia)19,20 is a critical factor in
appropriately interpreting patterns in large
databases. A multi-disciplinary approach to
research using such databases is probably
the most appropriate approach.
One limitation of our study arose because,
in most instances in Nova Scotia, the same
health-records personnel coded medical
charts for both the DAD and the NSAPD.
This may mean that agreement between the
two systems is higher than would be expected
if the two systems were fully independent.
On the other hand, both systems have
detailed rules regarding coding (including
criteria for diagnosis and specified parts of
Vol 29, No 3, 2009
the medical chart from which information
is to be extracted). These coding rules are
evident in the discrepancies in diagnoses
for which standard diagnostic criteria were
unavailable.
Other potential limitations of our study
include the assumption that the NSAPD, a
smaller, clinically focused database with
a data quality assurance program, is more
accurate than the DAD. This assumption,
though tenable, is unlikely to hold with
regard to all of the information in the
two databases. For example, the CIHI
database collects information on blood
and blood products in much greater
detail than the NSAPD does. Furthermore,
the NSAPD algorithm for determining
gestational age relies solely on menstrual
dates and a pediatric examination of the
newborn infant. Conversely, gestational
age in the DAD could represent a better
estimate, because it also incorporates early
ultrasound information.16–18
Finally, our assessment was limited to one
Canadian province and, hence, may not be
generally applicable to other provinces/
territories, or even to Nova Scotia at a
different time. Nevertheless, the relatively
high level of accuracy observed in our
study is encouraging and more in line
with the expectations of CPSS investigators
99
who have worked with DAD data9–13 than
the results of the abstraction-re-abstraction
study.8
In summary, our study compared infor­
mation in the DAD with that from a smaller
clinically focused database, showing that
the DAD information was accurate for many
of the diagnoses/procedures examined.
Furthermore, less accurate diagnoses, typi­
cally observed in the case of variably
defined clinical entities, can be improved
using combined codes and a restriction
to more severe forms of the disease. This
study therefore supports the use of the
data in the CIHI DAD for national perinatal
surveillance and research, with the caveat
that appropriate inference rests on an
understanding of clinical practice and
the use of sensitivity analyses to identify
robust findings.
Acknowledgements
This study was funded by the Maternal and
Infant Health Section, Health Surveillance
and Epidemiology Division of the Public
Health Agency of Canada. We are grateful
to the Reproductive Care Program of Nova
Scotia for access to the data.
Chronic Diseases in Canada
References
1.
Health Canada. Canadian perinatal health
report. Ottawa: Minister of Public Works
and Government Services Canada; 2000.
2.
Health Canada. Canadian perinatal health
report. Ottawa: Minister of Public Works
and Government Services Canada; 2003.
3.
Public Health Agency of Canada. Canadian
perinatal health report. 2008 ed. Ottawa:
Minister of Public Works and Government
Services Canada; 2008.
11. Liu S, Liston RM, Joseph KS, Heaman M,
Sauve R, Kramer MS. Maternal mortality
and severe morbidity associated with
low-risk planned cesarean delivery versus
planned vaginal delivery at term. CMAJ.
2007;176:455–60.
12. Kramer MS, Rouleau J, Baskett TF,
Joseph KS. Amniotic-fluid embolism and
medical induction of labour: a retrospective,
population-based cohort study. Lancet.
2006;368:1444–8.
13. Joseph KS, Rouleau J, Kramer MS, Young DC,
Liston RM, Baskett TF. Investigation of an
increase in postpartum haemorrhage in
Canada. BJOG. 2007;114:751–9.
4.
Wen SW, Huang L, Liston R, et al. Severe
maternal morbidity in Canada, 1991–2001.
CMAJ. 2005;173(7)759–64.
5.
Wilson-Costello D, Friedman H, Minich N,
Fanaroff AA, Hack M. Improved survival
rates with increased neurodevelopmental
disability for extremely low birth weight
infants in the 1990s. Pediatrics. 2005;
115:997–1003.
6.
Wilson-Costello D, Friedman H, Minich N,
Siner B, Taylor G, Schluchter M, Hack M.
Improved neurodevelopmental outcomes
for extremely low birth weight infants in
2000–2002. Pediatrics. 2007;119:37–45.
7.
Wen SW, Liu S, Marcoux S, Fowler D.
Uses and limitations of routine hospital
admission/separation records for peri­
natal surveillance. Chronic Dis Can. 1997;
18:113–9.
8.
Health Canada. An evaluation of the quality
of obstetric/neonatal discharge abstract data
by reabstraction of medical charts. Ottawa:
Health Canada; 2003.
9.
Liu S, Heaman M, Kramer MS, Demissie K,
Wen SW, Marcoux S. Length of hospital
stay, obstetric conditions at childbirth,
and maternal readmission: a populationbased cohort study. Am J Obstet Gynecol.
2002;187:681–7.
18. Joseph KS, Huang L, Liu S, Ananth CV,
Allen AC, Sauve R, Kramer MS. Reconciling
the high rates of preterm and postterm
birth in the United States. Obstet Gynecol.
2007;109:813–22.
10. Liu S, Heaman M, Joseph KS, et al. Risk
of maternal postpartum readmission
associated with mode of delivery. Obstet
Gynecol. 2005;105:836–42.
19. Wu YW, Backstrand KH, Zhao S, Fullerton HJ,
Johnston SC. Declining diagnosis of birth
asphyxia in California: 1991–2000. Pediatrics.
2004;114:1584–90.
Chronic Diseases in Canada
20. Dzakpasu S, Joseph KS, Huang L, Allen A,
Sauve R, Young D. Decline in birth asphyxia
in Canada: fact or artefact. Pediatrics.
2009;123:e668–72.
14. Fair M, Cyr M, Allen AC, Wen SW,
Guyon G, MacDonald RC. An assessment
of the validity of a computer system for
probabilistic record linkage of birth and
infant death records in Canada. Chronic
Dis Can. 2000;21:8–13.
15. Fair M, Cyr M. Allen AC, Wen SW,
Guyon G, Macdonald RC. Validation study
for a record linkage of births and infant
deaths in Canada. Ottawa: Statistics Canada;
1999.
16. Kramer MS, McLean FH, Boyd ME,
Usher RH. The validity of gestational age
estimation by menstrual dating in term,
preterm, and postterm gestations. JAMA.
1988;260:3306–8.
17. Goldenberg RL, Davis RO, Cutter GR,
Hoffman HJ, Brumfield CG, Foster JM.
Prematurity, postdates, and growth
retardation: the influence of use of ultra­
sonography on reported gestational age.
Am J Obstet Gynecol. 1989;160:462–70.
100
Vol 29, No 3, 2009
Validity of autism diagnoses using administrative health data
L. Dodds, PhD (1); A. Spencer, MSc (1); S. Shea, MD (2); D. Fell, MSc (1); B. A. Armson, MD (3) A. C. Allen, MD (1);
S. Bryson, PhD (2)
Abstract
It is necessary to monitor autism prevalence in order to plan education support and
health services for affected children. This study was conducted to assess the accuracy
of administrative health databases for autism diagnoses. Three administrative health
databases from the province of Nova Scotia were used to identify diagnoses of autism
spectrum disorders (ASD): the Hospital Discharge Abstract Database, the Medical Services
Insurance Physician Billings Database and the Mental Health Outpatient Information
System database. Seven algorithms were derived from combinations of requirements for
single or multiple ASD claims from one or more of the three administrative databases.
Diagnoses made by the Autism Team of the IWK Health Centre, using state-of-the-art
autism diagnostic schedules, were compared with each algorithm, and the sensitivity,
specificity and C-statistic (i.e. a measure of the discrimination ability of the model)
were calculated. The algorithm with the best test characteristics was based on one
ASD code in any of the three databases (sensitivity = 69.3%). Sensitivity based on an ASD
code in either the hospital or the physician billing databases was 62.5%. Administrative
health databases are potentially a cost efficient source for conducting autism surveillance,
especially when compared to methods involving the collection of new data. However,
additional data sources are needed to improve the sensitivity and accuracy of identifying
autism in Canada.
Introduction
The prevalence of autism spectrum
disorders (ASDs) and autism, specifically,
is reported to have been increasing over
time.1-4 If this is in fact correct, there would
be major implications for the education
system and agencies that provide services
for these children: the availability of support
and services will not match the increasing
demands on the education system and
health service providers. To date, there
have been isolated efforts in Canada to
estimate the prevalence of ASDs in some
jurisdictions, but there are currently no
systems in place to routinely monitor and
report autism incidence and prevalence.
Active surveillance of autism, conducted
by population screening, provides excellent
prevalence information, but is expensive
and generally limited to short-term
investigations.1 Passive surveillance using
existing databases provides a relatively
inexpensive method to derive ongoing,
population-based prevalence estimates.
The broad continuum of associated
cognitive and neurobehavioural disorders,
of which autism is the most extreme, are
called pervasive developmental disorders
(PDDs) or autism spectrum disorders
(ASDs).1,5 According to the diagnostic
criteria of the International Classification
of Diseases (ICD-10) by the World Health
Organization (WHO), PDDs include
childhood autism, atypical autism, Rett
syndrome, other childhood disintegrative
disorders, overactive disorders associated
with mental retardation and stereotyped
movements, Asperger’s syndrome, other
pervasive developmental disorders and
unspecified pervasive developmental
disorders. Childhood autism, atypical
autism and Asperger’s syndrome represent
the more common diagnoses. In this
study, we use the term ASD, which is
equivalent to PDD, except that ASD does
not include Rett syndrome and childhood
disintegrative disorder, both of which are
extremely rare.
In 1985, Bryson et al. made the first effort
to estimate autism prevalence in Canada
by screening all children (i.e. n = 20 800)
aged 6 to 14 years in a specific geographic
area of Nova Scotia and conducting followup diagnostic assessments for children
who screened positive (i.e. n = 46).6 Of
the 46 children who screened positive,
21 children fell within the relatively narrow
autism spectrum that was defined at the
time (i.e. most, if not all, of whom would
meet the more stringent criteria for autistic
disorder).7
More recently, researchers in Canada
have used existing data to estimate ASD
prevalence. Ouellette-Kuntz et al. reported
estimates of the prevalence of PDDs
among children 15 years or younger
during 2002 in the provinces of Prince
Edward Island (PEI) and Manitoba.8
In PEI, cases were identified by the
Department of Social Services and Seniors
and the Department of Education; parental
consent was required for the researchers
Author References
1 Perinatal Epidemiology Research Unit, Departments of Obstetrics and Gynaecology and Pediatrics, Dalhousie University, Halifax, NS, Canada.
2 Department of Pediatrics, Dalhousie University, Halifax, NS, Canada.
3 Department of Obstetrics and Gynaecology, Dalhousie University, Halifax, NS, Canada.
Correspondence: Linda Dodds, PhD, Departments of Obstetrics and Gynaecology and Pediatrics, Dalhousie University, 5850/5980 University Avenue,
P.O. Box 9700, Halifax, NS, Canada B3K 6R8, Tel.: 902-470-7191, Fax: 902-470-7190, Email: [email protected]
Chronic Diseases in Canada
102
Vol 29, No 3, 2009
to collect the information. In Manitoba,
cases were identified through referrals to
the Children’s Special Services program
of the Department of Family Services and
Housing. PDD prevalence rates among
1- to 15-year-olds in both provinces were
similar (i.e. 2.84 per 1000 in Manitoba
and 3.52 per 1000 in PEI). Fombonne et
al. reported prevalence (of PDDs) based
on a population of children registered at
a large Anglophone school board in the
Montreal area on October 1, 2003 (i.e. n =
27 749).9 In Quebec, school boards submit
information on children with PDDs and
other disorders to the Ministry of Education
in order to receive supplemental funding. In
this 2003 survey, a total of 180 identified
children had been diagnosed with a PDD
(i.e. rate of 6.5 per 1000), 61 of whom
were specifically diagnosed with autism.9
In summary, surveillance and reports of
autism prevalence in Canada are infrequent
and variable rates have been reported.
To date, administrative health databases
have not been used in Canada to estimate
autism incidence or prevalence, although
they have been used to estimate the incidence
and prevalence of other conditions; e.g.
algorithms have been developed and tested
using administrative data for determining
the incidence and prevalence of childhood
asthma, osteoporosis, diabetes mellitus
and diabetic macular edema.10–13 In a study
to evaluate the validity of ICD codes from
administrative hospital discharge data,
Quan et al.14 compared ICD-9 and ICD-10
coding (i.e. the coding systems used in
the administrative health databases) with
medical chart data for 32 clinical conditions
(ASD was excluded from the conditions
assessed). They found that detection
rates (e.g. sensitivity) varied by condition
from 82% for renal failure to 9% for
weight loss.14
Administrative health databases are a
potential source for determining autism
prevalence, but the validity of ASD
diagnoses from administrative health data
must be determined before these databases
are used to measure the prevalence of
autism in a population. Based on a cohort
of children born in Nova Scotia between
1989 and 2002, we used administrative
health databases linked to a “gold standard”
Vol 29, No 3, 2009
clinical autism database to assess the accu­
racy of autism diagnoses ascertained from
administrative health databases.
Methods
This study was based on data from a
retrospective cohort study designed to
examine prenatal, obstetrical and neonatal
factors related to the development of
autism. A cohort of all children born in
Nova Scotia between 1989 and 2002 was
identified from the Atlee Perinatal Database,
i.e. a population-based database of all
hospital births in Nova Scotia. The cohort
of births was linked to the administrative
health databases at the Population Health
Research Unit at Dalhousie University. Data
linkage was accomplished using encrypted
health card numbers, common to all data
sources. The cohort of children born
between 1989 and 2002 were followed, by
way of the administrative health databases,
until December 2005.
For residents of Nova Scotia, as in the rest
of Canada, access to hospital and physician
services is universal within a system of
publicly funded health care. For this study,
three administrative health databases in
Nova Scotia were used to identify diagnoses
of autism spectrum disorders (ASD), i.e.
the Hospital Discharge Abstract Database
(available since 1989); the Medical
Services Insurance (MSI) Physician Billings
Database (available since 1989); and the
Mental Health Outpatient Information
System (MHOIS) Database (available since
1992). The Hospital Discharge Abstract
Database includes diagnoses, which are
noted in the medical chart and abstracted
upon discharge. The MSI Physician Billing
Database included a physician diagnostic
code(s), which was sent to the provincial
agency that handled payment for these
insured services. The MHOIS Database was
used for all outpatients seen in the mental
health clinics and day patients in
mental health day-treatment programs.
Diagnoses were recorded by psychiatrists
or psychologists, or both. An ASD diagnosis
was defined from these administrative
databases by an ICD-9 code 299 or an
ICD-10 code F84 from any primary or
secondary diagnostic field.
103
Seven algorithms were derived from
combinations of requirements for single
or multiple ASD claims from the three
administrative databases. For example, in
one algorithm, a child was considered to
have an autism diagnosis if there was at
least one autism code from the hospital
discharge database; autism codes from
the other databases were not required. The
algorithm allowing for the most “hits” for
an autism diagnosis was required for at
least one ASD claim from any of the three
aforementioned databases.
“Gold standard” diagnoses were obtained
from a clinical database generated by the
Autism Team of the IWK Health Centre.
Referrals to the Autism Team were made
largely by health care professionals and
some teachers in the Halifax Regional
Municipality to assess children with sus­
pected autism. The IWK Autism Team
consisted of pediatricians, psychologists,
social workers, psychiatrists, speechlanguage pathologists, occupational thera­
pists and nurses. Final determination of
diagnoses was made by psychologists and/
or pediatricians or psychiatrists, who led or
co-led the diagnostic teams and was based
on the Autism Diagnostic Interview –
Revised, the Autism Diagnostic Observation
Schedule and clinical judgment using
DSM-IV-TR.15–17 These instruments and
criteria were consistent with recommended
practice parameters for diagnosing ASDs.18,19
Diagnoses made by the Autism Team,
considered the “gold standard,” were
recorded in a database starting in 2001.
The linkage between the Atlee Perinatal
Database, the administrative health data­
bases and the “gold standard” data was
accomplished using a multi-step procedure
to ensure anonymity. The first step was
the creation of a “cross-walk file,” which
included a unique number assigned to all
individuals in each of the databases, along
with their encrypted health card number.
A third party used a sophisticated algo­
rithm to encrypt health card numbers,
assigned to every individual in the province
and a common field in each data source).
Finally, the requested variables from
each file were linked back to the “cross-walk
file,” using the unique encrypted number
Chronic Diseases in Canada
Table 1
Comparison of algorithms1 using combinations of autism spectrum disorder (ASD) diagnoses from three administrative health databases
compared to a “gold standard” diagnosis
Type of administrative data
Hospital data
(# of times
ASD coded)
Physician
billing data
(# of times
ASD coded)
Comparison of results to “gold standard”
Mental
health
outpatient
data
(# of times
ASD coded)
1
1
1
1
1
1
1
1
2
1
2
Test characteristics of algorithms
# True
positives
# True
negatives
# False
positives
# False
negatives
Sensitivity
Specificity
C-statistic
21
86
2
155
11.9%
97.7%
0.55
105
75
13
71
59.7%
85.2%
0.72
1
29
81
7
147
16.5%
92.0%
0.54
1
122
68
20
54
69.3%
77.3%
0.76
110
73
15
66
62.5%
83.0%
0.74
75
78
10
101
42.6%
88.6%
0.67
65
82
6
111
36.9%
93.2%
0.65
2
Algorithms based on autism code(s) from more than one database indicates that an autism diagnosis was assigned if an autism code was used in either of the databases indicated.
assigned to the individuals in each database,
and a linked, anonymous analysis file con­
taining data elements from each data source
was generated.
Diagnoses of children assessed by the
“gold standard data” (i.e. the IWK Autism
Team) from 2001 to 2005 were compared
to ASD diagnoses from each of the seven
algorithms, based on the administrative
health databases. The accuracy of each
algorithm was evaluated by calculating the
sensitivity, specificity and a C-statistic (i.e. a
nonparametric estimate of the area under
a receiver operating characteristic curve that
provides a measure of a method’s ability
to predict an autism diagnosis). C-statistic
scores range from 1.0 for a “perfect” test
with a sensitivity and specificity of 100%,
to 0.5 for a method that was unable to
discriminate.20
For the true ASD cases that were missed
by the administrative databases (i.e.
false negatives [FN]), codes for other
psychological conditions were examined.
In addition, codes that occurred both
before and after the date of the true (i.e.
“gold standard”) diagnosis were evaluated.
Various factors were evaluated for those
patients who had an autism code in one
of the administrative databases, but who
were not given an ASD diagnosis after
assessment by the Autism Team (i.e.
false positive [FP]). These included the
number of incorrect claims; the years when
Chronic Diseases in Canada
these ASD claims occurred, whether the
incorrect claims occurred after the IWK
negative diagnosis date; and whether there
had been other claims made in relation to
psychological conditions. Sensitivity and
specificity rates were compared for maternal
and infant factors, such as low birth
weight and maternal age (available from
the Atlee Perinatal Database), to determine
if certain characteristics were associated
with the accuracy of autism diagnoses
based on administrative health data.
Approval for this study was obtained by
the Research Ethics Board of the IWK
Health Centre.
Results
The IWK Autism Team evaluated 270 patients
linked to the overall study cohort of
children born in Nova Scotia. According
to the team’s assessment, there were
176 confirmed ASD cases and 88 non-cases
(i.e. 6 had undetermined diagnoses and
were dropped from further analysis). All
remaining 264 children had at least 2 years
of administrative data available following
the date of their birth. When seen by the
Autism Team, 58% of the children were
4 years or younger; only 12% of the chil­
dren were 10 years or older when the team
saw them. The majority of confirmed
cases were coded with a general diagnosis
of ASD, without any specific autism
diagnosis noted.
104
Table 1 shows the definition of each of
the seven algorithms tested, along with
the sensitivity, specificity and C-statistic
associated with each algorithm. The
algorithm with the highest C-statistic (i.e.
0.76), the highest sensitivity (i.e. 69.3%)
and a specificity of 77.3% was the algorithm
that defined an ASD diagnosis by at least
one claim in any of the three administrative
databases. Using this algorithm, 190 of
the 264 children were correctly diagnosed.
There were 20 FPs and 54 FNs, which were
examined in more detail to help explain
the inaccuracies in the administrative
databases.
An examination was made of the 54 FN
children diagnosed with ASD by the
Autism Team, but who did not have an
ASD claim in any of the three databases,
to see if other claims might have been
systematically recorded instead of ASD. Of
the 54 FNs, 46 children had at least one
MSI physician billing claim for neurotic
disorders, personality disorders and other
non-psychotic mental disorders (i.e. ICD-9
codes 300-316). Of these 46 children,
35 (i.e. 76%) had an ICD-9 code of 315
(i.e. “specific delays in development”)
coded at least once. This code occurred in
22 children before the Autism Team diag­
nosis date and in 26 children after; some
children had an ICD-9 code of 315 before
and after the Autism Team diagnosis date.
Vol 29, No 3, 2009
Table 2
Comparison of the number of autism spectrum disorder (ASD)
claims per child among false positives and true positives
been reluctant to use an autism code before
an autism diagnosis was verified.
False positives
True positives
Database
Frequency of ASD
claims per child
# children (%)
(n = 20)
# children (%)
(n = 122)
Hospital
1 or more
2 (10%)
21 (17%)
MSI
MHOIS
Any of 3 databases
2 or more
1 (5%)
7 (6%)
1 or more
13 (65%)
104 (85%)
2 or more
4 (20%)
55 (45%)
1 or more
7 (35%)
29 (24%)
2 or more
7 (35%)
23 (19%)
1 or more
20 (100%)
122 (100%)
2 or more
11 (55%)
74 (61%)
The number of ASD claims from each of the
three databases was compared between
the 20 FP children and the 122 TP children
(see Table 2). For the 20 FPs, 2 chil­
dren (i.e. 10%) had ASD coded from
Hospital Discharge Data, 13 (i.e. 65%) from
MSI Physician Billing Data and 7 (i.e. 35%)
from MHOIS Data (see Table 2). Among the
13 subjects from the FP group with one or
more ASD claims from the Physician Billing
Database, 4 of 13 (i.e. 31%) had more than
one ASD claim in the Physician Billing
Database, compared to 55 of 104 (i.e. 53%)
of the true positives. Among the MHOIS
claims for the FP group, all had more than
one MHOIS claim for ASD. Of the 122 TPs,
21 (i.e. 17%) had hospital claim(s),
104 (i.e. 85%) had MSI claim(s) and 29 (i.e.
24%) had MHOIS claim(s); 27 (i.e. 22%)
had claims from 2 databases and 5 (i.e.
4%) had claims from all 3 databases (data
not shown). While most ASD claims from
the hospitalization and MHOIS databases
occurred after the Autism Team diagnosed
TPs, 55 of 104 (i.e. 53%) of children had
MSI claim(s) before this date. Other than
ASD codes, the most common code used
was ICD-9-CM 315 (“specific delays in
development”), which was recorded
equally before and after the Autism Team
diagnosis date.
Sensitivity and specificity values were
compared according to maternal and
neonatal characteristics (see Table 3). The
sensitivity of the administrative data in
identifying an ASD diagnosis was similar
across most factors, including for males
Vol 29, No 3, 2009
(i.e. 69.7%) and females (i.e. 66.7%).
The sensitivity of the administrative data
in identifying an ASD diagnosis was not
significantly lower for children with a
major congenital anomaly (i.e. 55.6%)
compared to children without an anomaly
(i.e. 69.9%). The sensitivity was not
significantly higher among children outside
of Halifax County compared to residents of
Halifax County (i.e. 75.0% versus 68.1%),
although specificity was lower (i.e. 66.7%
versus 80.0%, respectively).
Discussion
In the current study, we used codes from
three administrative health databases to eva­
luate multiple algorithms for their accuracy
in identifying autism among children in
Nova Scotia. Although the overall study
cohort included all children born in Nova
Scotia between 1989 and 2002, only children
seen by the Autism Team (between 2001
and 2005) who linked to the study cohort
were included in this validation study.
Based on the algorithm defining autism
by at least a single claim in any one of the
hospitalizations, the physician billing or
the outpatient mental health databases, the
ability of administrative health databases
in Nova Scotia to correctly identify children
with autism was moderately successful
(i.e. sensitivity of 69%). Most of the true
ASD cases who were incorrectly identified
within the administrative data (i.e. FNs) had
codes indicating some other non-psychotic
psychological disorder or developmental
delay, suggesting that physicians may have
105
A strength of this study was the quality of
the autism diagnosis in the “gold standard”
population. However, the “gold standard”
diagnosis was limited to children who were
referred to the Autism Team. It should be
noted that children in this validation study
without an ASD diagnosis when assessed
by the Autism Team would have had some
behavioural and/or developmental feature
that warranted referral to the Autism
Team. Therefore, the false positive rate
observed in this study is likely higher, and
the specificity lower, than it would have
been had we been able to establish a “gold
standard” diagnosis for all children in the
administrative databases. Nevertheless,
the specificity we observed was reasonably
high (i.e. 77%), an estimate which is likely
below the true specificity. Other algorithms
tested in this study had more stringent
requirements for defining autism (e.g.
two physician claims required), and there­
fore had better specificity than the oneclaim algorithm, albeit at the expense of
reduced sensitivity.
In order to improve the detection rate
observed in this study, other data sources
would be required. In Canada, information
on ASD diagnoses is available from regional
school boards in some areas or from some
provincial Departments of Social Services
or Family Services, as previously discussed.
The use of education data sources (i.e. alone
or in conjunction with clinical data) and
data from other government-administered
programs have been used to identify autism
cases in the United States. The Centers
for Disease Control and Prevention have
established a multi-source surveillance
network for ASD and other developmental
disabilities.21
Children 8 years of age with ASD who reside
within one of the 16 states comprising part
of the network area were identified in a
two-phase process. First, children suspected
of having an ASD were identified through
screening and abstraction of records
from multiple sources within clinical
and education records. In phase two, the
abstracted behavioural data were scored
Chronic Diseases in Canada
Table 3
Comparison of sensitivity and specificity of autism spectrum disorder (ASD) diagnoses using
administrative data compared to “gold standard” diagnoses, according to maternal and
neonatal factors
Factor
Number
Sensitivity (95% CI)
Specificity (95% CI)
Maternal age
< 35
218
71.2% (63.2 to 78.1)
77.2% (66.8 to 85.2)
35
46
62.2% (46.1 to 76.0)
77.8% (44.3 to 94.7)
Halifax County
214
68.1% (60.0 to 75.1)
80.0% (69.1 to 87.8)
Outside Halifax
50
75.0% (57.7 to 87.0)
66.7% (43.6 to 83.9)
< 2500 g
15
72.7% (42.9 to 90.8)
75.0% (28.9 to 96.6)
2500 g
234
69.1% (61.7 to 75.7)
77.4% (67.3 to 85.1)
County of residence
Birth weight
Major congenital anomaly*
Yes
13
55.6% (26.6 to 81.2)
100% (45.4 to 100)
No
247
69.9% (62.5 to 76.4)
76.5% (66.2 to 84.5)
231
69.7% (62.0 to 76.4)
80.3% (69.8 to 87.8)
33
66.7% (45.2 to 83.0)
58.3% (31.9 to 80.7)
Sex
Male
Female
Birth order
First born
145
69.8% (60.0 to 78.1)
77.6% (64.0 to 87.1)
Second or higher
119
68.8% (57.9 to 77.9)
76.9% (61.5 to 87.6)
* A major anomaly is defined as a defect of structure or function that is present at birth and affects length of life,
impacts quality of life or requires surgery.
by clinicians to determine whether they
met the ASD case definition. The rates
varied somewhat between sites, with an
overall mean prevalence rate of 6.6 per
1000 eight-year-old children.22 Extensive
quality assurance activities were incor­
porated into the network to maximize data
quality and consistency.
Newschaffer et al. used a national source of
administrative data (i.e. the United States
Department of Education, Office of Special
Education Programs) to examine trends
in ASD between 1992 and 2001. However,
limitations of these data were noted, in
particular with the specific classification
of impairment and the likelihood of
underestimating autism prevalence based
on special education data alone.23 In
California, individuals with autism (and
other conditions) are eligible to receive
services through the Department of
Developmental Services. Eligibility is based
on diagnoses provided by qualified health
care professionals. Croen et al.24 used these
data to estimate autism prevalence. They
suspected that their observed prevalence
Chronic Diseases in Canada
of 12.3 per 10 000 children for the years
1987 to 1994 was an underestimation, since
approximately 20% to 25% of the children
who were eligible to receive services were
not enrolled in the program.24
In Canada, all provinces and territories
have administrative data that include
hospitalizations and physician visits. In
Nova Scotia, the addition of an outpatient
mental health database increased the
sensitivity of ASD diagnoses by about 7%,
compared to the sensitivity using only
hospitalization and claim data regarding
physician visits. On the other hand, the
specificity increased by about 6% when
the mental health outpatient data were
excluded. Since relatively few children
were hospitalized for (or with) autism (i.e.
12% of the true cases had an autism code
from the hospitalization data), this source,
by itself, was inadequate to determine
autism diagnoses in a population. However,
an autism diagnosis in the hospitalization
database was very likely correct. Although
we explored ICD codes that were used
other than ASD codes, their use was too
106
inconsistent to suggest an algorithm that
would improve the false positive or false
negative rates.
Research or surveillance of health condi­
tions using administrative health databases
has advantages over other data collection
methods. Administrative health databases
are available in all Canadian provinces and
territories and provide a source for a large
number of population-based cases, likely
at a lower cost than would be possible
with newly collected data. In addition,
diagnoses are entered into the databases
without knowledge of underlying exposureoutcome hypotheses. However, there are
limitations to using administrative data,
particularly with respect to the accuracy
of diagnoses that are being used for billing
purposes (as is the case with the Physician
Billing Database).
Given that we measured maximum sensi­
tivity at 69%, it is likely that administrative
health data alone would underestimate the
true incidence and prevalence, as observed
in this study. This would suggest that
additional data sources are necessary to
enhance the detection rate of ASD diagnoses
from existing databases, since it is unlikely
that a single source of administrative
data will provide a complete accounting
of all autism cases in Canada. Although
challenging, the jurisdictions should work
together toward acquiring standard data
from multiple sources to enable ongoing,
passive surveillance of ASDs in Canada.
Acknowledgements
This project was funded by Cure Autism Now
(i.e. now Autism Speaks). Linda Dodds was
supported by the Clinical Research Scholar
Award from Dalhousie University and the
New Investigator Award from the Canadian
Institutes of Health Research. The authors
thank the Reproductive Care Program of
Nova Scotia and the Population Health
Research Unit of Dalhousie University
for data access. Although this research
is based partially on data obtained from
the Population Health Research Unit, the
observations and opinions expressed
are those of the authors and do not
represent those of the Population Health
Research Unit.
Vol 29, No 3, 2009
References
1.
Bryson SE, Smith IM. Epidemiology of
autism: prevalence, associated character­
istics, and implications for research and
service delivery. Ment Retard Dev Disabil
Res Rev. 1998;4:97–103.
2.
Fombonne E. The prevalence of autism.
JAMA. 2003;289:87–9.
3.
Fombonne E. Epidemiology of autistic
disorder and other pervasive developmental
disorders. J Clin Psychiatry. 2005;66 Suppl
10:3–8.
4.
Williams JG, Higgins JP, Brayne CE.
Systematic review of prevalence studies of
autism spectrum disorders. Arch Dis Child.
2006;91:8–15.
5.
Filipek PA, Accardo PJ, Baranek GT, et al.
The screening and diagnosis of autistic
spectrum disorders. J Autism Dev Disord.
1999;29:439–84.
6.
Bryson SE, Clark BS, Smith IM. First report
of a Canadian epidemiological study of
autistic syndromes. J Child Psychol
Psychiatry. 1988;29:433–45.
7.
American Psychiatric Association. DSM IV
diagnostic and statistical – manual. 4th ed.
Washington (D.C.): American Psychiatric
Association; 1994.
8.
Ouellette-Kuntz H, Coo H, Yu CT, Chudley
AE, Noonan A, Breitenbach M, et al.
Prevalence of pervasive developmental
disorders in two Canadian provinces.
J Appl Res Intellect Disabil. 2006;3:164–72.
9.
Fombonne E, Zakarian R, Bennett A,
Meng L, McLean-Heywood D. Pervasive
developmental disorders in Montreal,
Quebec, Canada: prevalence and links
with immunizations. Pediatrics. 2006;118:
e139–50.
10. To T, Dell S, Dick P, Cicutto L, Harris J,
Tassoudji M, Duong-Hua M. Burden of
childhood asthma. Toronto, Ontario: ICES,
2004.
Vol 29, No 3, 2009
11. Lix LM, Yogendran MS, Leslie WD,
Shaw SY, Baumgartner R, Bowman C, et al.
Using multiple data features improved the
validity of osteoporosis case ascertainment
from administrative databases. J Clin
Epidemiol. 2008;61:1250-60.
12. Hux JE, Ivis F, Flintoft V, Bica A.
Determination of prevalence and incidence
using a validated administrative data algo­
rithm. Diab Care. 2002;25:512–6.
13. Bearelly S, Mruthyunjaya P, Tzeng JP,
Suner IJ, Shea AM, Lee JT, et al.
Identification of patients with diabetic
macular edema from claims data. Arch
Ophthalmol. 2008;126:986–9.
14. Quan H, Li B, Saunders D, Parsons GA,
Nilsson CI, Alibhai A, et al. Assessing validity
of ICD-9CM and ICD-10 administrative data
in recording clinical conditions in a unique
dually coded database. Health Serv Res.
2008;43:1424–41.
15. Lord C, Rutter M, Le Couteur A. Autism
diagnostic interview-revised: a revised
version of a diagnostic interview for
caregivers of individuals with possible
pervasive
developmental
disorders.
J Autism Dev Disord. 1994;24:659–85.
16. Lord C, Risi S, Lambrecht L, Cook EH,
Leventhal EL, DiLavore PC, et al. The
autism diagnostic observation schedulegeneric: a standard measure of social and
communication deficits associated with the
spectrum of autism. J Autism Dev Disord.
2000;30:205–23.
subcommittee of the American academy of
neurology and the child neurology society.
Neurology. 2000;55:468–79.
20. Ash A, Shwartz M. R2: A useful measure
of model performance when predicting a
dichotomous outcome. Stat Med. 1999;
18:375–84.
21. Rice CE, Baio J, Van Naarden Braun K,
Doernberg N, Meaney FJ, Kirby RS. A public
health collaboration for the surveillance of
autism spectrum disorders. Paediatr Perinat
Epidemiol. 2007;21:179–90.
22. Autism and Developmental Disabilities
Monitoring Network Surveillance Year
2002 Principal Investigators; Centers for
Disease Control and Prevention. Prevalence
of autism spectrum disorders – autism
and developmental disabilities monitoring
network, 14 sites, United States, 2002.
MMWR Surveill Summ. 2007 Feb 9;
56(1):12–28.
23. Newschaffer CJ, Falb MD, Gurney JG.
National autism prevalence trends from
United States special education data.
Pediatrics. 2005;115:e277–82.
24. Croen LA, Grether JK, Selvin S. Descriptive
epidemiology of autism in a California
population: who is at risk? J Autism Dev
Disord. 2002;32:217–24.
17. American Psychiatric Association. Diagnostic
and statistical manual. 4th ed. Washington
(D.C.): American Psychiatric Association;
2000.
18. Filipek PA, Accardo PJ, Baranek GT,
Cook EH, Dawson G, Gordon B, et al.
The screening and diagnosis of autistic
spectrum disorders. J Autism Dev Disord.
1999;29:439–84.
19. Filipek PA, Accardo PJ, Ashwal S,
Baranek GT, Cook EH, Dawson G, et al.
Practice parameter: screening and diagnosis
of autism: report of the quality standards
107
Chronic Diseases in Canada
Associations between chronic disease, age and physical
and mental health status
W. M. Hopman, MA (1,2); M. B. Harrison, PhD (2,3,4); H. Coo, MSc (2); E. Friedberg, MHA (3,4); M. Buchanan, BScN (3);
E. G. VanDenKerkhof, DrPH (2,3,5)
Abstract
This paper examines the associations between chronic disease, age, and physical
and mental health-related quality of life (HRQOL), using data collected in 10 studies
representing five chronic conditions. HRQOL was measured using the SF-36 or the shorter
subset, SF-12. Physical Component Summary (PCS) and Mental Component Summary
(MCS) scores were graphed by condition in age increments of 10 years, and compared
to age- and sex-adjusted normative data. Linear regression models for the PCS and MCS
were controlled for available confounders. The sample size of 2418 participants included
129 with renal failure, 366 with osteoarthritis (OA), 487 with heart failure, 1160 with
chronic wound (leg ulcer) and 276 with multiple sclerosis (MS). For the PCS, there were
large differences between the normative data and the mean scores of those with chronic
diseases, but small differences for the MCS. Female gender and comorbid conditions
were associated with poorer HRQOL; increased age was associated with poorer PCS and
better MCS. This study provided additional evidence that, while physical function could
be severely and negatively affected by both chronic disease and advanced age, mental
health remained relatively high and stable.
Key words: age, chronic disease, mental health, physical health, HRQOL,
status, SF-36, SF-12
Background
Health-related quality of life (HRQOL) is a
primary concern with chronic conditions
and is often used as a research outcome
in both clinical trials and observational
studies. A useful characteristic of a gene­
ric measure of HRQOL is the ability to
compare across different diseases to assess
burden of illness.1 HRQOL is a particularly
relevant outcome in chronic disease, where
a cure is often unavailable and health goals
involve living with and managing one’s
condition.2,3
A growing body of research has examined
HRQOL in a variety of diseases, and
many studies have identified significant
impairments.4–10 HRQOL has also been
examined in the general population, pro­
viding normative data for comparative
purposes.4–6,11 However, while chronic
disease typically has a significant negative
impact on the physical aspects of health,
mental health status may remain relatively
unaffected. This has been demonstrated
in studies on individual conditions,12–17
as well as in two multiple-condition
studies: one focusing on allergies, arthri­
tis, congestive heart failure, chronic
lung disease, hypertension, diabetes and
ischemic heart disease,1 and another that
compared multiple sclerosis (MS), osteo­
arthritis (OA), renal disease and renal
transplants.2 Additional research into the
effect of multimorbidity, taking into account
the effect of both the number and severity
of comorbid conditions, also identified an
inverse association between the number
and severity of conditions and HRQOL,18,19
particularly in the physical domains.18,19
Cross-sectional evidence suggests that
while the physical aspects of HRQOL
decline as age increases,2,4 mental health
remains stable across age categories, or
may even improve.2,4,21 This observation is
further supported with longitudinal data.
HRQOL tends to be stable over three22 to
five23 years, but if there are changes, it is
the physical aspects of HRQOL which tend
to decline while mental aspects improve.23
The purpose of this study was to examine
the relationship between age and physical
and mental aspects of health for people
with different chronic conditions. Research
objectives included a comparison of the
physical and mental health status across
diseases, as well as an examination of
the association between age and HRQOL,
while controlling for key variables available
across all databases. We hypothesized that
the physical aspects of HRQOL would be
substantially lower in those with chronic
disease as compared to a normative popu­
lation, and that it would also be lower in
older versus younger age groups, while the
mental aspects of HRQOL would be similar
to the normative data and be relatively unaf­
fected by disease group or increased age.
Herein, we examined data collected in
ten Canadian studies representing five con­
ditions, including renal failure, hip and
knee OA, congestive heart failure (CHF,
Author References
1 Clinical Research Centre, Kingston General Hospital, Kingston, ON.
2 Department of Community Health and Epidemiology, Queen’s University, Kingston, ON.
3 School of Nursing, Faculty of Health Sciences, Queen’s University.
4 Ottawa Health Research Institute, Clinical Epidemiology Program, Ottawa, ON.
5 Department of Anesthesiology, Faculty of Health Sciences, Queen’s University.
Correspondence: Wilma M. Hopman, MA, Clinical Research Centre, Angada 4, Kingston General Hospital, Kingston, ON, Canada K7L 2V7,
Tel.: 613-549-6666, ext. 4941, Fax: 613-548-2428, Email: [email protected]
Chronic Diseases in Canada
108
Vol 29, No 2, 2009
Table 1
Characteristics of the ten studies
Study
Renal failure
n
129
Inclusion criteria
Age 18 years
> 6 months duration
Age Mean (SD)
PCS Mean (SD)
MCS Mean (SD)
Acute/reversible
cognitive impairment
Exclusion criteria
59.4 (14.7)
33.2 (11.8)
50.1 (11.2)
Osteoarthritis (hip)
177
Able to consent
Revisions, fractures
67.6 (11.2)
24.4 (6.6)
49.4 (12.5)
Osteoarthritis (knee)
189
Able to consent
Revisions, fractures
68.6 (8.8)
26.2 (7.9)
50.9 (12.4)
CHF (usual vs. transitional care)
191
Speak English or French
Unable to consent
75.7 (9.9)
29.9 (8.2)
51.0 (9.6)
CHF (partners in care)
296
Speak English or French
Unable to consent
72.7 (12.0)
31.5 (8.8)
46.7 (11.2)
Chr. wound (uptake of evidence)
117
Speak English or French
Unable to consent
74.1 (12.5)
32.3 (9.9)
48.8 (10.7)
Chr. wound (two models of care)
211
Speak English or French
Unable to consent
68.4 (13.9)
35.7 (9.8)
49.1 (11.2)
Chr. wound (bandaging RCT)
180
Speak English or French
Unable to consent
67.1 (16.1)
39.5 (10.9)
50.6 (10.2)
Chr. wound (new service delivery)
652
Speak English or French
Unable to consent
72.2 (13.7)
31.8 (9.6)
47.9 (12.0)
Multiple sclerosis*
276
Clinically definite MS
Communicate verbally
Cognitive impairment
duration > 12 months
46.5 (10.1)
33.5 (10.6)
46.0 (12.2)
CHF:
Chr.:
SD:
PCS:
MCS:
congestive heart failure;
chronic;
standard deviation;
physical component summary;
mental component summary
two studies), chronic wounds (leg ulcer,
four studies) and MS. While HRQOL data
do exist for these conditions, little of it is
Canadian, and the opportunity to compare
these five conditions is new, thus adding to
the body of knowledge about the impact of
chronic disease on HRQOL. These findings
will also be of interest to those who provide
care to patients with these conditions.
Methods
Details of the 10 studies are described
below and are presented in Table 1. Data
were collected at baseline through a
combination of patient interview (SF-36 or
SF-12, sociodemographic data) and chart
review (clinical data). Ethics approval for
each study was obtained from the Queen’s
University and Affiliated Teaching Hospitals
Research Ethics Board or the Ottawa Health
Research Institutes Ethics Board, as well as
site-specific institutional reviews, where
applicable. An application for the combined
analysis was approved by the Queen’s
Research Ethics Board (approval number
EPID-227-06).
Measures
The Medical Outcomes Trust 36-item
health survey (SF-36)4 and its 12-item
subset, the SF-12,5 are among the most
widely used instruments to measure
Vol 29, No 2, 2009
HRQOL.4,5 The SF-36 and SF-12 measure
eight self-reported aspects of HRQOL,
including physical function, physical role,
bodily pain, general health, vitality, social
function, emotional role and mental health.
The Physical Component Summary (PCS)
and Mental Component Summary (MCS)
are standardized to a mean of 50, with a
score above 50 representing better than
average function and below 50 poorer
than average function.5,6 Previous work
by Ware et al. has noted a high degree
of correspondence between the PCS and
MCS obtained from the SF-36 and SF-12.
Regression analyses to reproduce the PCS
and MCS scores for the SF-36 using the
SF-12 scores had R2 values in excess of 0.90
for both.5 In addition, an examination of
the actual scores across 17 population and
disease subgroup comparisons indicated
that the average SF-36 and SF-12 PCS and
MCS scores differed by less than one point,
suggesting that the interpretation is the
same and that comparisons are valid.5
Databases and participants
The renal failure database included all
consenting patients receiving hemodialysis
at Kingston General Hospital (KGH) and its
affiliated satellite units in Kingston, Ontario.
The SF-36 version 1.0 was administered at
a routine hemodialysis visit. The hip and
knee OA databases included all consenting
109
primary elective total hip and total knee
replacement patients on the waiting lists
of five orthopedic surgeons in Kingston.
The SF-36 version 1.0 was administered
at the time of the six-week pre-surgical
assessment.
The two CHF databases included all patients
who had a diagnosis or exacerbation of
CHF at hospital admission. Data for the
first study (i.e. Usual Care versus Transition
Care) were collected during hospitalization
at two medical units of the Civic Campus at
the Ottawa Hospital. Data for the second
study (i.e. Partners in Care: CHF Study)
were collected from patients recruited
from 10 sites, including inpatient units,
and community and specialty clinics in
Ontario, New Brunswick, Manitoba and
Illinois. HRQOL was assessed at the time
of study entry using the SF-12 version 1.0.
Although most were inpatients at the time
of enrolment, they were seen early in their
admission and the stays were typically
brief. As most of the items on the SF-12
reference the past four weeks, the data are
considered representative of the time when
they were not hospitalized.
The chronic wound database (i.e. leg ulcers)
was based on four studies, including the
Prospective Study of the Uptake of EvidenceBased Guidelines in the Community; the
Chronic Diseases in Canada
Effectiveness and Efficiency of Two Models
of Delivering Care to Chronic Wound
Population;16 the Chronic Leg Ulcers in the
Community Pre- and Post-Implementation
of a New Service Delivery Model;17 and
a Randomized Control Trial (RCT) of
the Effectiveness of Two Compression
Technologies. Patients were recruited from
sites in Ontario (i.e. Ottawa, Kingston,
Toronto, Hamilton, Niagara, KitchenerWaterloo and London), Manitoba (i.e.
Winnipeg) and Saskatchewan (i.e. Regina
and Saskatoon). SF-12 version 1.0 (version
2.0 for the RCT) data were collected as part
of the baseline assessment.
The MS database included all consenting
individuals with an appointment at the MS
Clinic in Kingston over a one-year period.
Two weeks before their appointment,
patients received a package containing
the SF-36 version 1.0 and a sociodemogra­
phic questionnaire. Those who consented
returned the completed package at their
appointment.
Data management and
statistical procedures
All project databases were entered into
SPSS (version 14.0 for Windows, Chicago,
Illinois, 2005) for scoring and analysis.
For the combined analysis, the variables
contained in each, as well as the associated
coding, were examined to find the common
variables across the 10 databases. Key
variables contained in each database were
age, gender, whether the patient lived
alone, cardiovascular disease, diabetes and
“additional” comorbidities.
The definition for cardiovascular disease
included hypertension, as this was impor­
tant for the renal failure population.
However, for the CHF group, it included
cardiovascular disease other than conges­
tive heart failure to avoid multicollinearity.
While this resulted in a somewhat different
adjustment for cardiovascular disease
across the chronic conditions, it was felt
that a crude adjustment was preferable to
no adjustment at all. The diversity of
the patient populations also resulted in the
collection of different comorbidities. For
example, very few comorbidities were col­
lected for the MS sample, while a lengthy list
was compiled for the heart failure studies.
Chronic Diseases in Canada
As a result, only two comorbidities plus a
category of “additional comorbidities” (i.e.
defined simply as yes/no) could be drawn
from each of the databases, and included
comorbidities ranging from depression
and sleep disorders to cancer, stroke and
myocardial infarction. Variables such as
education level, marital status, severity of
disease and socio-economic status were not
consistently collected across all databases.
Patients under 25 years were excluded, as
there were too few for comparison (i.e. two
OA, four renal failure, three MS).
To facilitate comparison, age was cate­
gorized in 10-year increments as in the
Canadian normative data for the SF-36.11
Once the 10 databases were collapsed
into 5 condition-specific databases, the
mean PCS and MCS scores were graphed
by age group and condition, and compared
to age- and sex-adjusted normative data.11
Linear regression models were developed
for the PCS and the MCS, controlling
for condition, age group, gender, living
circumstances, cardiovascular disease, dia­
betes and additional comorbidities. All twoway interactions were also assessed. The
condition with the highest mean age (i.e.
CHF) was used as the reference condition,
while the reference age group used was 25
to 34 years.
Results
Response rates and demographics
The 10 individual study sample sizes
ranged from 117 to 652 participants,
with a combined sample size of 2418.
The characteristics of patients in the five
chronic conditions are displayed in Table
2. In all studies, the participation rates
were high (i.e. > 77%). For the renal
failure database, 129 of 155 (83.2%)
provided consent, and age ranged from
25.5 to 89.8 years. For OA, 880 patients
were eligible and 673 agreed to participate,
for a response rate of 76.5%. However,
chart review was done after surgery and,
consequently, only 366 participants had
complete data, since 307 were still awaiting
surgery at the study’s end. Participant age
ranged from 30.0 to 89.0 years, with a
similar number of patients awaiting a total
hip or knee replacement (i.e. 177 and 189,
respectively).
110
The heart failure database contained 487
of 506 eligible patients (i.e. 96.2%) who
completed the HRQOL questionnaire.
Patients ranged in age from 31.0 to
102.0 years. Four databases were represented
in the chronic wound sample, and 1160 of
1470 (i.e. 78.9%) patients completed the
SF-12; ages for this sample ranged from
25.0 to 102.0. For MS, 300 of 363 patients
(i.e. 82.6%) agreed to participate, with
276 completing the SF-36. This sample
was the youngest, with ages ranging from
25.0 to 77.0 years, and only 10 individuals
over 65 years. It also had far more women
(i.e. 203) than men (i.e. 73), compared
to the other studies where gender was
more balanced.
Descriptive statistics, physical
component summary
Figure 1 contains the graph of the mean
values for the PCS for each age group by
condition; the means and 95% confidence
intervals are presented in the accomp­
anying table. The differences between the
normative data and the mean values of
those with each of the chronic diseases
were large, demonstrating a sig­nificant
burden of illness. The renal failure group
showed the greatest variation by age group,
while the OA sample consistently had the
lowest scores.
When examining all chronic diseases as
a group (n = 2418), an initial decline
levelled off as age increased. Starting with
the 25 to 34 year age group and ending with
the 75-plus age group, the mean values
for the PCS were 41.4 ± 10.1; 35.3 ±
11.7; 32.3 ± 10.3; 31.3 ± 9.3; 30.6 ±
9.6; 31.5 ± 10.0. The 10-point difference
between the youngest (i.e. 41.4) and oldest
(i.e. 31.5) groups was similar to the drop
in the normative sample. Examining each
disease group separately, this pattern was
less clear, with some conditions (e.g. OA 55
to 64 and chronic wound 65 to 74) showing
somewhat higher scores at older ages than
in the adjacent, younger group. However,
the confidence intervals were wide and
overlapped for some conditions, suggesting
that even though the differences were
sometimes large, they were not necessarily
statistically significant.
Vol 29, No 2, 2009
Figure 1
Physical Component Summary scores by disease and age group
60
Mean PCS Score
50
Normative
40
Renal failure (129)
Osteoarthritis (366)
30
Heart failure (487)
Chronic wound (1160)
Multiple sclerosis (276)
20
10
0
25 to 34
35 to 44
45 to 54
55 to 64
65 to 74
75+
Age Group in Years
Footnote: Means and 95% CIs by age group, physical component summary
65 to 74 years
75+ years
Normative
Sample (n)
25 to 34 years
53.0
52.2, 53.7
35 to 44 years
52.0
51.3, 52.7
45 to 54 years
51.3
50.9, 51.7
55 to 64 years
49.0
48.6, 49.3
47.2
46.8, 47.6
42.0
41.5, 42.5
Renal failure (129)
45.3
36.8, 53.7
31.9
25.0, 38.7
33.2
27.0, 39.3
35.0
31.4, 38.6
29.7
26.1, 33.2
30.4
23.8, 37.1
Osteoarthritis (366)
n/a
27.3
21.0, 33.6
25.2
23.1, 27.3
27.9
25.6, 30.2
25.1
24.1, 26.2
24.3
22.8, 25.7
Heart failure (487)
n/a
34.0
21.6, 46.3
32.5
28.8, 36.2
31.1
28.8, 33.4
30.9
29.4, 32.3
30.6
29.6, 31.6
Chronic wound (1160)
42.2
37.2, 47.3
37.3
33.2, 41.4
35.0
33.1, 36.9
32.0
30.6, 33.5
33.9
32.7, 35.1
33.4
32.6, 34.2
Multiple sclerosis (276)
41.1
37.6, 44.6
35.6
33.4, 37.9
31.6
29.6, 33.7
30.2
27.7, 32.8
28.7
22.2, 35.3
n/a
Data were not graphed and 95% CIs were not calculated when the sample size was < 5.
Vol 29, No 2, 2009
111
Chronic Diseases in Canada
Figure 2
Mental component summary scores by disease and age group
60
Mean MCS Score
50
Normative
40
Renal failure (129)
Osteoarthritis (366)
30
Heart failure (487)
Chronic wound (1160)
Multiple sclerosis (276)
20
10
0
25 to 34
35 to 44
45 to 54
55 to 64
65 to 74
75+
Age Group in Years
Footnote: Means and 95% CIs by age group, physical component summary
65 to 74 years
75+ years
Normative
Sample (n)
25 to 34 years
50.1
49.2, 51.1
35 to 44 years
50.9
50.1, 51.7
45 to 54 years
51.4
51.0, 51.8
55 to 64 years
53.7
53.4, 54.0
53.7
53.4, 54.0
54.5
54.1, 54.9
Renal failure (129)
52.3
45.5, 59.0
46.0
38.6, 53.4
51.4
47.1, 55.6
48.6
44.5, 52.6
51.0
46.8, 55.2
52.8
47.6, 58.1
Osteoarthritis (366)
n/a
52.3
37.6, 67.0
46.1
42.4, 49.8
49.9
46.2, 53.5
50.9
49.0, 52.7
50.7
48.2, 53.1
Heart failure (487)
n/a
45.8
34.7, 56.8
44.7
39.2, 50.1
47.1
44.1, 50.1
48.1
46.4, 49.9
49.3
47.9, 50.6
Chronic wound (1160)
48.8
42.8, 54.9
46.3
42.5, 50.2
46.4
44.4, 48.4
45.4
43.3, 47.5
48.3
47.0, 49.7
50.3
49.4, 51.2
Multiple sclerosis (276)
43.1
38.6, 47.7
46.1
43.7, 48.5
43.7
41.1, 46.4
50.2
46.8, 53.5
52.1
41.7, 62.4
n/a
Data were not graphed and 95% CIs were not calculated when the sample size was < 5.
Chronic Diseases in Canada
112
Vol 29, No 2, 2009
Table 2
Sample characteristics for the five conditions
Characteristic
Renal Failure n = 129
Osteoarthritis n = 366
Heart Failure n = 487
Chronic Wound
n = 1160
Multiple Sclerosis
n = 276
n (%)
n (%)
n (%)
n (%)
n (%)
Age group
25 to 34 years
9 (7.0)
1 (0.3)
2 (0.4)
17 (1.5)
29 (10.5)
35 to 44 years
14 (10.9)
7 (1.9)
7 (1.4)
42 (3.6)
94 (34.1)
45 to 54 years
20 (15.5)
40 (10.9)
22 (4.5)
120 (10.3)
90 (32.6)
55 to 64 years
38 (29.5)
56 (15.3)
58 (11.9)
150 (12.9)
53 (19.2)
65 to 74 years
31 (24.0)
157 (42.9)
136 (27.9)
267 (23.0)
8 (2.9)
75+ years
17 (13.2)
105 (28.7)
262 (53.8)
564 (48.6)
2 (0.7)
Female
54 (41.9)
202 (55.2)
236 (48.5)
649 (55.9)
203 (73.6)
Live alone
25 (19.4)
79 (21.6)
195 (51.5)
420 (36.2)
35 (12.7)
Comorbidities
110 (85.3)
152 (41.5)
373 (76.6)‡
701 (60.4)
47 (17.0)
Diabetes
Cardiovascular*
45 (34.9)
13 (3.6)†
158 (32.4)
368 (31.7)
12 (4.3)
Additional
91 (70.5)
79 (21.6)
435 (89.3)
805 (69.4)
122 (44.2)
Mean ± SD
Mean ± SD
Mean ± SD
Mean ± SD
Mean ± SD
Age
59.4 ± 14.7
68.1 ± 10.0
73.9 ± 11.3
70.9 ± 14.2
46.5 ± 10.1
PCS
33.2 ± 11.8
25.3 ± 7.3
30.9 ± 8.6
33.8 ± 10.2
33.5 ± 10.6
MCS
50.1 ± 11.2
50.2 ± 12.5
48.4 ± 10.9
48.7 ± 11.5
46.0 ± 12.2
SD: standard deviation;
PCS: physical component summary;
MCS: mental component summary
* Cardiovascular includes hypertension
† Insulin-dependent diabetes only for the OA group
‡ Other than heart failure for the heart failure group
Descriptive statistics, mental
component summary
Figure 2 contains the graph of the mean
values for the MCS; the means and 95%
confidence intervals are presented in the
accompanying table. For those under
45 years, the sample sizes and resulting
wide confidence intervals preclude valid
comparisons for all but the MS sample
aged 25 to 34 years, and both the MS and
the chronic wound sample aged 35 to
44 years. In all three comparisons,
those with the chronic disease scored
substantially lower than the normative
sample. The highest mean value for age
groups over 55 years was in the normative
sample, but the differences between the
normative and condition-specific samples
were small and, therefore, of questionable
clinical significance. The MS sample had
the lowest mean values for two age groups
(i.e. 25 to 34 and 45 to 54 years), while
Vol 29, No 2, 2009
the chronic wound sample had the lowest
values for those in the 55 to 64 year age
group. Several disease groups overlap­
ped at the lowest value for the remaining
three age groups.
increased after 55 years of age. However, as
for the PCS results above, the confidence
intervals were often wide and overlapping,
and thus the data must be interpreted
with caution.
When all samples were combined (i.e. n =
2418), mean MCS values were fairly stable
for participants aged 25 to 54 years, after
which there was a steady improvement
by age category. Starting with the 25 to 34 year
age group and ending with the 75-plus age
group, the mean values for the MCS were
46.6 ± 11.7; 46.4 ± 11.9; 45.8 ± 11.7; 47.4 ±
12.7; 49.1 ± 11.1; 50.1 ± 11.2. This
increase in the older groups was similar to
the normative sample. Within a disease,
however, this pattern was less clear, with
most diseases showing declines until the
age group of 45 to 54 years. At this point,
the mean MCS appeared to increase for
all but the chronic wound sample, which
Data were also grouped by gender within
each disease to see if the pattern held
true for both men and women. This was
the case (data not shown); therefore, the
results were not reported separately for
men and women.
113
Regression analyses
Tables 3 and 4 contain the linear regres­
sion model for the PCS and MCS, respec­
tively. Although all two-way interactions
were tested, only two attained statistical
significance in each model, all with
negative coefficients (i.e. renal* additional
comorbidities and chronic wound*
additional comorbidities for the PCS; and
Chronic Diseases in Canada
Table 3
Linear regression model for the physical component summary
95% CI
p-value
44.1
41.3, 46.9
< 0.001
Renal Failure
0.9
-1.0, 2.8
0.367
Osteoarthritis
-8.6
-10.1, -7.1
< 0.001
Chronic Wound
1.9
0.9, 3.0
< 0.001
Multiple Sclerosis
-1.5
-3.2, 0.3
0.099
Physical Component Summary (r2 = 0.14)
Constant
Coefficient
Condition (reference = heart failure) (0 = no, 1 = yes)
Age group (reference = 25 to 34 years)
35 to 44 years
-5.4
-8.2, -2.5
< 0.001
45 to 54 years
-7.7
-10.4, -5.0
< 0.001
55 to 64 years
-8.1
-10.8, -5.4
< 0.001
65 to 74 years
-7.9
-10.6, -5.2
< 0.001
75+ years
-8.2
-10.9, -5.5
< 0.001
Gender (0 = male)
-1.3
-2.1, -0.5
0.002
Cardiovascular Disease (0 = no)
-1.1
-1.8, -0.4
0.001
Diabetes (0 = no)
-2.1
-3.1, -1.2
< 0.001
Additional Comorbidities (0 = no)
-3.5
-4.3, -2.6
< 0.001
Living circumstance was not significant (p = 0.52)
OA* additional comorbidities and MS*
additional comorbidities for MCS). Given
the limitations of the variable for additional
comorbidities (described earlier), only the
main effects were presented in the models.
The PCS model accounted for 14.4% of the
variation in outcome. The OA sample scored
significantly lower than the heart failure
sample, while the chronic wound sample
scored significantly higher than the heart
failure sample. The difference between the
MS and the heart failure sample approached
significance. The renal failure sample did
not differ significantly from the heart failure
sample. All age groups scored significantly
lower than the reference age group of 25 to
34 years. Men tended to have higher scores
than women. Cardiovascular disease, dia­
betes and additional comorbidities were all
associated with significantly lower mean
PCS scores. Living circumstance was not
a significant predictor of PCS.
CI: confidence interval
Table 4
Linear Regression Model for the Mental Component Summary
95% CI
p-value
48.4
45.0, 51.7
<0.001
Renal Failure
3.2
0.9, 5.5
0.006
Osteoarthritis
0.5
-1.3, 2.3
0.572
Chronic Wound
0.1
-1.1, 1.4
0.843
Multiple Sclerosis
-0.3
-2.4, 1.8
0.755
35 to 44 years
0.3
-3.1, 3.7
0.861
45 to 54 years
-0.1
-3.4, 3.2
0.963
55 to 64 years
2.1
-1.2, 5.3
0.218
65 to 74 years
3.9
0.7, 7.1
0.019
75+ years
5.3
2.1, 8.5
0.001
Gender (0 = male)
-1.3
-2.3, -0.4
0.007
Cardiovascular Disease (0 = no)
-0.8
-1.7, -0.1
0.042
Diabetes (0 = no)
-1.6
-2.7, -0.5
0.006
Additional Comorbidities (0 = no)
-2.6
-3.7, -1.5
<0.001
Mental Component Summary (r2 = 0.05)
Constant
Coefficient
Condition (reference = heart failure) (0 = no, 1 = yes)
Age group (reference = 25 to 34 years)
Living circumstance was not significant (p = 0.85)
CI: confidence interval
The MCS model accounted for only 4.6%
of the variation in outcome. The similarity
between the scores seen in Figure 2
was evident here as well, with only the
renal failure sample scoring statistically
significantly better than the heart failure
sample. For the most part, the effect
of increased age was positive and was
statistically significant for age groups over
65 years. Women tended to score more
poorly than men, as they did on the PCS.
The effect of comorbidities was negative,
although of borderline significance for car­
diovascular disease. Living circumstances
were not significantly associated with MCS.
Discussion
These data suggest a strong negative
association between physical health status
and both chronic disease and advanced
age. However, mental health status remains
relatively stable across disease groups and
age groups. This phenomenon has been
identified in other health conditions1,2,4,13–21
and the Canadian normative data,11 and is
confirmed by the results of our analysis
of five chronic conditions studied here.
The effect of advanced age on the PCS is
strikingly negative, with effect sizes ranging
from a five-point to an eight-point drop
Chronic Diseases in Canada
114
Vol 29, No 2, 2009
even after controlling for condition, gender
and comorbidities. This is supported by
the literature2,4 and is not only statistically
significant, but highly clinically relevant,
given that a two- to three-point difference
is likely to be clinically important.6 Only
two other variables had a large effect,
with the OA sample scoring an average of
8.6 points lower than the reference sample
of heart failure; those with additional
comorbidities also scored 3.5 points lower
than those who did not have other comor­
bidities. However, it is likely that these
estimates are conservative, given that the
differences are relative to heart failure
(i.e. used as reference group). Comparisons
with a healthy population as reference
category would likely show greater diffe­
rences, but these data are only available in
aggregate form.
Few variables had an effect size that
exceeded two to three points on the MCS.
Renal failure patients scored an average of
3.2 points higher than the reference category.
Those with advanced age (i.e. 65 to 74 years
and 75-plus years) scored higher (i.e. four
points and five points, respectively) than
the reference age group, which supports the
literature on the effect of age on mental
health.2,4,21 Additional comorbidities had a
large negative association with the MCS, as
those with additional comorbidities scored
an average of 2.6 points lower than those
without additional comorbidities. These fin­
dings are consistent with results from other
studies that assessed the impact of the
number of comorbidities on HRQOL.18–20, 24,25
These results provide useful insights
into the burden of illness experienced by
persons with these chronic conditions. The
finding that physical health status declines
with increased age and disease burden is
not new.1–3 However, these data provide
useful estimates of the relative effect of age
across five different diseases and confirm
previous findings that identify declining
physical function, but stable mental
function, in those with chronic disease
and/or increased age. These findings can
also have important implications for the
care and treatment of persons with these
conditions. While it is not possible to
predict when physical function is likely
to decline for a specific case, the results can
Vol 29, No 2, 2009
identify those at greater risk. In addition,
evidence of better mental health in older
age groups and in those whose illness has
been diagnosed for some time may allow
health care providers to focus in particular
on the mental health of those recently
diagnosed with a chronic disease.21 The
finding that, on average, women had lower
scores than men on both the PCS and the
MCS suggests that they may be particularly
vulnerable. Finally, the strong negative
association between the comorbidities (i.e.
cardiovascular, diabetes and “other comor­
bidities”) and both mental and physical
health status has been noted in other
studies,18–20,24 suggesting that those with
multiple comorbidities may be at greater
risk for poor HRQOL outcomes.
adjustment for illness severity seemed
preferable to no adjustment.
The results should be interpreted within the
limitations of the study. These data were
obtained from 10 databases, with under­
lying study designs that varied in both
purpose and methodology. As a result, only
six variables were consistently collected
across all databases; no consistent infor­
mation was collected regarding illness
severity, socio-economic status, education
and social support, which are commonly
associated with HRQOL. As a result, varia­
bles that are important determinants of
physical (e.g. severity of illness) and/or
mental health status (e.g. education and
social support) could not be tested, thereby
limiting our ability to develop the predictive
models. Moreover, almost half (i.e. 48%) of
the subjects had chronic leg ulcers as the
chronic disease, which limits the ability
to generalize the findings.
Nevertheless, these data provide compelling
evidence that, while physical function can
be severely and negatively affected by
both chronic disease and advanced age,
mental health remains relatively high and
stable, adding to the growing body of know­
ledge regarding the impact of increased age
and chronic disease on HRQOL. Additional
research with other disease groups, and
longitudinal research in particular, will
provide further insight into the complex
relationship between chronic disease,
physical health status, mental health status
and advancing age.
In addition, one of the six variables (i.e.
additional comorbidities) was based on the
comorbidities collected within each study.
Since some studies collected more than
others, participants in those studies would
be more likely to have a positive value for
this variable. Moreover, there is increa­
sing evidence that the severity as well
as the number of comorbidities was an
important consideration,19,20 and these
data were not consistently collected within
our studies. Future research would benefit
from considering both factors, preferably
with the use of a validated comorbidity
index. Despite these limitations, this crude
115
Furthermore, sample sizes within the age
groups for certain diseases were quite low.
There were too few young patients with OA
and heart failure, and too few older patients
with MS to graph these age/disease groups.
Even in cells with greater than five patients,
some of the numbers were quite low.
Consequently, large confidence intervals
often overlapped, indicating that the
results did not necessarily attain statistical
significance even when the difference
appeared large. Finally, our data are crosssectional and the age stratification is not
the equivalent of a cohort that is followed
over time.
References
1.
Alonso J, Ferrer M, Gandek B, et al.
Health-related quality of life associated
with chronic conditions in eight countries:
results from the international quality of life
assessment (IQOLA) project. Qual Life Res.
2004;13:283–98.
2.
Singer MA, Hopman WM, MacKenzie TA.
Psychological adjustment in four chronic
medical conditions. Qual Life Res. 1999;
8:687–91.
3.
Brunet DG, Hopman WM, Singer MA,
Edgar CM, MacKenzie TA. Measurement
of health-related quality of life in multiple
sclerosis patients. Can J Neurol Sci. 1996;
23:99–103.
Chronic Diseases in Canada
4.
Ware JE, Snow KK, Kosinski M. SF-36
health survey: manual and interpretation
guide. Boston (MA): The Health Institute,
New England Medical Center; 1993.
5.
Ware JE, Kosinski M, Keller SD. SF-12: how
to score the SF-12 physical and mental
health summary scales. Second ed. Boston
(MA): The Health Institute, New England
Medical Centre; 1995.
6.
7.
8.
9.
Ware JE, Kosinski M, Keller SD. SF-36
Physical and mental health summary
scales: a user manual and interpretation
guide. Boston (MA): The Health Institute,
New England Medical Center; 1994.
Hobbs FD, Kenkre JE, Roalfe AK, Davis RC,
Hare R, Davies MK. Impact of heart failure
and left ventricular systolic dysfunction
on quality of life: a cross-sectional study
comparing common cardiac and medical
disorders and a representative adult pop­
ulation. Eur Heart J. 2002;23:1867–76.
Jolly M. How does quality of life of patients
with systemic lupus erythematosis compare
with that of other common illnesses?
J Rheumatol. 2005;32:706–8.
van der Wall JM, Terwee CB, van der Windt DA,
Bouter LM, Dekker J. Health-related and
overall quality of life of patients with
chronic hip and knee complaints in general
practice. Qual Life Res. 2005;14:95–803.
10. Salaffi F, Carotti M, Stancati A, Grassi
W. Health-related quality of life in older
adults with symptomatic hip and knee
osteoarthritis: a comparison with matched
healthy controls. Aging Clin Exp Res.
2005;17:255–63.
11. Hopman WM, Towheed T, Anastassiades T,
et al. Canadian normative data for the SF-36
health survey. CMAJ. 2000;63:265–71.
12. Yost KJ, Haan MN, Levine RA, Gold EB.
Comparing SF-36 scores across three groups
of women with different health profiles.
Qual Life Res. 2005;14:1251–61.
13. Kusek JW, Greene P, Wang SR, et al. Crosssectional study of health-related quality of
life in African Americans with chronic renal
insufficiency: the African American study
of kidney disease and hypertension trial.
Am J Kidney Dis. 2002;39:513–24
14. Groothoff JW, Grootenhuis MA, Offringa M,
Gruppen MP, Korevaar JC, Heymans HSA.
Quality of life in adults with end-stage
renal disease since childhood is only
partially impaired. Nephrol Dial Transplant.
2003;18:310–7.
15. Kaplan RM, Criqui MH, Denenberg JO,
Bergan J, Fronek A. Quality of life in
patients with chronic venous disease:
San Diego population study. J Vasc Surg.
2003;37:1047–53.
16. Harrison MB, Browne GB, Roberts J,
Tugwell P, Gafni A, Graham, ID. Quality
of life of individuals with heart failure:
a randomized trial of the effectiveness of
two models of hospital-to-home transition.
Med Care. 2002;40:271–82.
22. Hopman WM, Berger C, Joseph L, et al.
Stability of normative data for the SF-36:
results of a three-year prospective study
in middle-aged Canadians. Can J Public
Health. 2004;95:387–91.
23. Hopman WM, Berger C, Joseph L, et al.
The natural progression of health-related
quality of life: results of a five-year pro­
spective study of SF-36 scores in a normative
population from the Canadian multicentre
osteoporosis study (CaMos). Qual Life Res.
2006;15:527–36.
24. Bayliss EA, Bayliss MS, Ware JE Jr,
Steiner JF. Predicting declines in physical
function in persons with multiple chronic
medical conditions: what we can learn from
the medical problems list. Health Qual Life
Outcomes. 2004;7(2):47.
25. Gadalla T. Association of comorbid mood
disorders and chronic illness with disability
and quality of life in Ontario, Canada.
Chronic Dis Can. 2008;28(4):148-54.
17. Harrison MB, Graham ID, Lorimer K,
Friedberg E, Pierscianowski T, Brandys T.
Leg-ulcer care in the community, before
and after implementation of an evidencebased service. CMAJ. 2005;172:1147–52.
18. Fortin M, Lapointe L, Hudon C, Vanasse A,
Ntetu AL, Maltais D. Multimorbidity and
quality of life in primary care : a systematic
review. Health Qual Life Outcomes.
2004;2:51.
19. Fortin M, Bravo G, Hudon C, Lapointe L,
Almirall J, Dubois MF, Vanasse A.
Relationship
between
multimorbidity
and health-related quality of life of
patients in primary care. Qual Life Res.
2006;15(1):83–91.
20. Fortin M, Dubois MF, Hudon C, Soubhi H,
Almirall J. Multimorbidity and quality
of life: a closer look. Health Qual Life
Outcomes. 2007;5:52.
21. Cassileth BR, Lusk EJ, Strouse TB, et al.
Psychosocial status in chronic illness:
a comparative analysis of six diagnostic
groups. N Engl J Med. 1984;311: 506–11
Chronic Diseases in Canada
116
Vol 29, No 2, 2009
Statistical modelling of mental distress among
rural and urban seniors
C. P. Karunanayake, PhD (1); P. Pahwa, PhD (1,2)
Abstract
The senior population is growing rapidly in Canada. Consequently, there will be an
increased demand for health care services for seniors who have mental illness. Seniors are
more likely to live in rural areas than younger people; therefore, it is important to identify
the differences between rural and urban seniors in order to design and deliver mental
health services. The main objective of this paper was to use the National Population Health
Survey (NPHS) to examine the differences with regard to mental distress between rural
and urban seniors (i.e. 55 years and older). The other objectives were to investigate the
long-term association between smoking and mental health and the long-term association
between unmet health care needs and the mental health of seniors in rural and urban
areas. The mental distress measure was examined as a binary outcome. The analysis
was conducted using a generalized estimating equation approach that accounted for the
complexity of a multi-stage survey design. Rural seniors reported a higher proportion of
mental distress [OR=1.16; 95% CI: 0.98, 1.37] with a borderline statistical significance
than urban seniors. This finding was based on a final multivariate model to study the
relationship between mental distress and location of residence (i.e. rural or urban) as
well as between smoking and self-perceived unmet health care needs, adjusting for other
important covariates and missing outcome values. A significant correlation was noted
between smoking and mental health problems among seniors after adjusting for other
covariates [OR = 1.26; 95% CI: 1.00, 1.60]. Participants who reported self-perceived
unmet health care needs reported a higher proportion of mental distress [OR = 1.72;
95% CI: 1.38, 2.13] compared to those who were satisfied with their health care.
Key words: mental health, rural seniors, longitudinal data, National Population Health
Survey, generalized estimating equations, bootstrap weights, missing data
Introduction
Seniors are one of the fastest growing
population groups in Canada, as reported
by Statistics Canada. By 2021, there will be
almost seven million seniors (i.e. 65 years
of age or older), representing 19% of the
population.1 Seniors are more likely than
younger people to live in rural areas (24%
versus 21%) and are also more likely to
reside in smaller urban areas.1 Rural seniors
often live in isolation, and due to a lack of
social interaction, they may be at a higher
risk of developing mental health problems
compared to their urban counterparts.
Furthermore, this isolation may increase
the likelihood of rural seniors reporting
lifestyle habits such as smoking and alcohol
consumption, which helps aggravate
mental health problems.2–4 Mental illness
accounts for 30% of disability claims, i.e.
$15 to $33 billion annually in Canada.5 A
recent Canadian study estimates that the
annual cost of treated and non-treated
mental health problems in Canada is
$14.4 billion.6 By 2020, depression will
be the second leading cause of the overall
world illness burden, after ischemic dis­
ease.7-8 Advances in neuroscience and
behavioural medicine have shown that
mental disorders are the result of complex
interactions among biological, psycho­
logical and social factors.8 There has been
adequate research on the mental health
of rural seniors, but very few rural-urban
comparison studies have been conducted
on seniors. It is important to identify
differences between rural and urban seniors
in order to design and deliver appropriate
mental health services. Proper statistical
analysis of available national longitudinal
datasets allows for the investigation of
important risk factors for mental distress
These factors can lead us to identify highrisk groups at early stages and will help us
target our preventive measures to lessen
the future economic burden on the health
care system.9
The authors of this paper will use the
National Population Health Survey (NPHS)10
from Statistics Canada to examine (1)
the rural and urban differences in mental
distress; (2) the long-term association
between smoking and mental distress;
and (3) the long-term association between
unmet health care needs and mental
health among seniors who live in rural
and urban areas. Urban areas are defined
as continuously built-up areas with a
population concentration of 1000 or more,
and a population density based on the
previous census11 of 400 or more people
per square kilometre; other areas are
defined as rural.
Author References
1 Canadian Centre for Health and Safety in Agriculture, University of Saskatchewan, Saskatoon, SK, Canada.
2 Department of Community Health & Epidemiology, University of Saskatchewan, Saskatoon, SK, Canada.
Correspondence: Chandima Karunanayake, PhD, Canadian Centre for Health and Safety in Agriculture, University of Saskatchewan, Royal University Hospital,
103 Hospital Drive, Saskatoon, SK, Canada S7N 0W8, Telephone: 306-966-1647, Fax: 306-966-8799, Email: [email protected]
Chronic Diseases in Canada
118
Vol 29, No 3, 2009
Methods
Longitudinal NPHS dataset
Data from the National Population Health
Survey (NPHS) were used in this analysis.
The NPHS is a longitudinal study10 of a
Canadian national sample. The original
survey included 17 626 subjects sampled
in 1994/95 (i.e. cycle 1), with the aim of
following or recontacting them every 2 years
for up to 20 years. To be included in the
survey, respondents must have completed
at least the general component of the ques­
tionnaire in 1994/95.12 Except in the
province of Quebec, the NPHS employed
a stratified two-stage design (i.e. clusters
and dwellings) based on the Labour Force
Survey (LFS) from Statistics Canada. In
Quebec, the NPHS sample was selected
using a two-stage design similar to that of
the LFS from dwellings participating in a
health survey organized by Santé Québec
(the 1992/1993 Enquête sociale et de santé
[ESS]).13 Base sample sizes for each province
were determined using the Kish allocation,
which balanced the reliability require­
ments at national and provincial levels.
A minimum of 1200 households in each
province was needed to ensure a specified
reliability by sex and broad age groups.
Populations on First Nations reserves,
on Canadian Forces bases and in some
remote areas of Quebec and Ontario were
excluded from the household components
of the survey. Data were weighted to reflect
the sample design, non-response adjust­
ments and post-stratification.
The same individuals were surveyed
repeatedly, which allowed for the inves­
tigation of the effects of baseline- and timevarying risk factors on longitudinal changes
in mental health. Conclusions drawn from
such longitudinal studies are stronger
compared to cross-sectional studies,
because some information on the sequence
of events is available.14 In most situations,
longitudinal data are incomplete. There
are several approaches for the analysis of
incomplete longitudinal data. A binary
logistic regression model can be fitted
using a method based on improbability,
i.e. the Generalized Estimating Equation
(GEE) approach,15 assuming that the
dropouts are not missing completely at
random (MCAR). Another approach for
Vol 29, No 3, 2009
analysis of incomplete longitudinal data
are pattern mixture models formulated
by Little,16–18 assuming that dropouts are
not MCAR. The pattern mixture model is
a solution to the non-response problem
in survey data. The first step in applying
the pattern mixture model approach is to
divide the subjects into groups depending
on their missing data pattern. If subjects
are measured at six time points, then there
are 64 (26) possible missing data patterns.
A between-subject variable is created by
grouping the missing data patterns. This
between-subject variable can be used in the
longitudinal data analysis as another covari­
ate. In this paper, pattern mixture models
were examined with GEE-based models for
National Population Health Survey data.
mental health studies20–23 that are included
in these analyses are sex, age, marital status,
education level, total household income,
self-reported general health index, geo­
graphical area, any chronic condition, physi­
cal activity within the last three months,
and self-perceived unmet health care needs.
Modelling distress as a binary response
From the relatively wide range of mental
health indicators available in the NPHS,
we chose the distress measure based on
a subset of items from the Composite
International Diagnostic Interview (CIDI).
The outcome of interest consisted of
six questions developed by Kessler and
Mroczek of the University of Michigan.19
The distress scale is comprised of various
CIDI items that inquired about feelings of
sadness, nervousness, restlessness, hope­
lessness, worthlessness and the feeling that
everything was an effort.11 Additional items
clarified whether these symptoms occurred
“a lot,” “somewhat,” “a little,” “more than
usual,” “the same,” or “less than usual”
compared to the previous month. Based
on the above questions, a distress scale
was derived for each of the six cycles. This
derived variable determines the respon­
dent’s distress scale. Scores on the distress
scale range from 0 (i.e. no distress) to 24
(i.e. highly distressed). Details can be found
in the NPHS derived variable directory.11
It was interesting to investigate how
the response vector evolved over time
and to observe how it related to a set of
explanatory variables. The distress scale
was highly skewed, and we decided to
recode it according to the literature23–24 and
the suggestions of a geriatric psychiatrist.
As distress was recoded to a binary scale
(i.e. categories: no/low [0-5 scale] and
moderate/high [6-24 scale]), it seemed
natural to consider a binary model. We
fitted the GEE-based binary regression
model25–26 using the GENMOD procedure
in SAS.25–29 This procedure allowed us to
select different specifications of working
correlation matrices (i.e. independent, firstorder autoregressive [AR(1)], exchangeable
and unstructured). We selected the model
with an unstructured covariance structure,
which gave us the smallest standard
errors.30 The GENMOD is based on Liang
and Zeger’s method,28–29 which accounts
for the within-subject dependencies only,
due to the repeated measurements over
time. To account for the complexities of
the multi-stage stratified clustered design, the
bootstrap resampling method was used to
calculate the correct variance around a given
estimate. This was achieved using both the
“Bootvar” SAS macro30–32 and the bootstrap
weights provided by Statistics Canada.10,12
The “Bootvar” macro was modified to apply
to the generalized estimating equations
method.31,34 The approach used to study the
effect of missing data was the GEE-based
pattern mixture model.
Subpopulation
Statistical analysis
This study was limited to the population group
aged 55 years and older as of the initial
survey in 1994/95. There were 4444 par­
ticipants in the subpopulation and
16 052 observations in the longitudinal
analysis. The main factors of interest
are location of residence, smoking status
and drinking status. Other demographic and
socio-economic variables in previous
Univariate analyses were conducted to
examine the relationship between the dis­
tress scale and the main factors of interest,
as well as demographic and socio-economic
variables at α = 0.20 significance level. The
next step was to conduct the multivariable
analysis to determine the effects of all
potential covariates and/or interactions
Distress scale: National Population
Health Surveys (NPHS)
119
Chronic Diseases in Canada
on the distress scale. All potential covari­
ates and interaction terms were included
concurrently in the model. Variables that
were significant at α = 0.05 level or of
scientific interest, as well as missing data
patterns, were retained in the final model.
Results
In 1994, 20% of the general population
of Canada were seniors. Among seniors,
there were more female seniors (i.e.
56%) compared to male seniors. In rural
populations, 22% of residents were seniors
and in urban populations, 20% of residents
were seniors. There was a higher percentage
of male seniors (i.e. 53%) in rural areas,
but a higher percentage of female seniors
(i.e. 58%) in urban areas. Our main interest
in conducting this analysis was to compare
mental distress between rural and urban
seniors. We started the statistical analysis
by exploring these differences in baseline
characteristics, presented in Table 1 and
summarized below.
Comparison of mental distress between
rural and urban seniors
The percentage of the moderate- or highdistress category among seniors was
17% in rural areas and 16% in urban
areas, respectively. The proportion of the
moderate- or high-distress category for all
age categories varies from 12% to 25% for
rural seniors and from 14% to 19%
for urban seniors. In the moderate- or
high-distress category, there was a slightly
higher percentage (i.e. 22%) of rural female
seniors compared to urban female seniors
(i.e. 20%). In addition, female seniors had
a higher distress level compared to male
seniors in both rural and urban areas. The
proportion of moderate or high distress
levels was higher for rural seniors who
were single, married, common-law spouses
or in a partnership compared to their
urban counterparts. Both rural and urban
Quebec residents had a higher proportion
of moderate or high distress than rural and
urban seniors in other regions. In both rural
and urban areas, respondents with low
education levels had a higher proportion
of moderate or high distress compared to
post secondary graduates, and a higher
proportion of respondents with low
income were in this distress category than
Chronic Diseases in Canada
high income seniors. Moreover, seniors
who were less involved in social activities
reported a higher proportion of moderate or
high distress in both rural and urban areas.
The seniors who were current smokers had
a higher prevalence of moderate or higher
distress in both rural and urban areas
compared to non-smoking seniors. Nondrinkers had a higher proportion (i.e. 21%)
of moderate or high distress compared to
current drinkers in rural and urban areas.
There was a higher prevalence of moderate
or high distress in respondents, (1) with
any chronic condition compared to persons
without a chronic condition; and (2),
without any physical activity within the
last three months compared to respondents
with any physical activity. In addition,
seniors with self-perceived unmet health
care needs had a higher prevalence of
moderate or high distress than those whose
health care needs were perceived to have
been met. This proportion was higher in
urban areas (i.e. 49%) compared to rural
areas (i.e. 43%).
Figure 1 illustrates the rural and urban
comparison of self-perceived unmet health
care needs. This suggests that seniors in
rural areas were more likely to have unmet
health care needs than their counterparts
in urban areas from 1996 to 2002. This
apparent increase over time could be due
to aging. Figure 2 illustrates the reasons
for not getting self-perceived needed health
care for rural and urban seniors from
1994 to 2004. The most common reasons
for not meeting the health care needs
of rural seniors were difficult access to
health professionals (i.e. 40%) and seniors
choosing not to see health professionals
(i.e. 25%). For their urban counterparts,
the most common reasons for a lack of
seniors’ health care were difficult access to
health professionals (i.e. 20%) and other
reasons (i.e. 45%), which included too
busy, didn’t get around to it, didn’t know
where to go, transportation problems,
language problems, and personal or family
responsibilities. It was also interesting to
note that both rural and urban seniors (i.e.
33% and 26%, respectively) who reported
moderate or high distress were more
likely (i.e. more than 6 times within past
12 months) to visit their family doctor.
120
Participants in rural areas were less likely
to see their family doctor (i.e. 18%)
com­pared to their urban counterparts
(i.e. 14%).
Univariate analysis results
Analyses were conducted to examine the
relationship between the distress scale,
the main interest factors and the demo­
graphic and socio-economic variables men­
tioned above. The preliminary analysis
showed that the variables of sex, education
level, age group, marital status, income level,
general health, geographic area, smoking
status, any chronic condition, physical
activity, self-perceived unmet healthcare
needs and location of residence were related
to the mental distress scale at significance
level of α = 0.20. Alcohol consumption
was not shown as a risk for mental distress
in the preliminary analysis and it was not
used in the model.
Multivariable analysis results
Table 2 explains the four missing patterns.
We can contrast completers (i.e. those who
completed all six cycles) versus those
who missed one cycle, completers versus
those who missed two or more cycles, and
completers versus people who died within
six cycles.
We included the missing patterns and
covariates to the multivariate model. This
multivariate analysis was based on a gener­
alized estimating equations approach, with
the results given in Table 3. This model is
called GEE-based pattern mixture model.
In this analysis, several important interac­
tion terms were tested (i.e. sex and smo­
king, sex and physical activity, etc.) and
none were significant.
Included in the model for the purpose
of the generalized estimating equations
procedure were age, sex, marital status,
location of residence, geographic area,
income level, education level, smoking
status, general health status, any chronic
condition, physical activity within the last
three months and self-perceived unmet
health care needs.
All of these variables were retained for the
final model of the relationship between
covariates and mental distress among
Vol 29, No 3, 2009
Figure 1
Self-perceived unmet health care needs of rural and urban seniors over time
seniors (i.e. 55 years and older). The odds
ratios reported for all covariates predicting
mental distress took into account the
relationship of each of these variables with
each outcome at each cycle. All reported
odds ratios were adjusted for all other
variables in the model. The odds ratios
demonstrate the likelihood that those with
poor self-rated general health status would
have greater mental distress compared
to those with excellent self-rated general
health status. This takes into account the
changes in the status of self-rated general
health over a two-year follow-up period
to produce an overall estimate of the
association for each relationship. A similar
interpretation can be applied to each
of the other variables in the model for
mental distress.
The following results were obtained based
on the final multivariate model used to
study the relationship between location of
residence, smoking and mental distress,
adjusting for other important covariates and
the pattern of missing data. Rural seniors
(i.e. 55 years and older) reported a higher
proportion of mental distress [OR = 1.16;
95% CI (0.98 to 1.37)] than urban seniors.
A significant association was evident among
seniors with mental health problems and
smoking after adjusting for other covariates
Vol 29, No 3, 2009
[OR = 1.26; 95% CI (1.00 to 1.60)]. Senior
non-completers had significantly higher
proportions of mental distress compared
to completers [OR = 1.44 (1.13 to 1.82),
OR = 1.39 (1.10 to 1.76) and OR = 1.68
(1.36 to 2.06) respectively]. A significantly
higher proportion of female participants
reported mental distress compared to male
participants [OR = 1.79 (1.47 to 2.17)].
Separated, widowed or divorced parti­
cipants reported a significantly higher
proportion of mental distress compared
to single participants [OR = 1.48 (1.08 to
2.03)]. Participants from Quebec reported
a significantly higher prevalence [OR =
1.57 (1.20 to 2.06)] of distress compared
to Ontario participants, while Atlantic
residents reported a lower prevalence [OR =
0.68 (0.53 to 0.87)] of distress compared
to Ontario participants. Senior participants
who did not complete their secondary
school education had a higher prevalence
[OR = 1.36 (1.07 to 1.72)] of distress
compared to post-secondary graduates.
Participants who reported their general
health index as “poor,” “fair” or “good”
had a higher prevalence of mental distress
compared to those who reported “excellent”
general health. Senior participants who had
any kind of chronic condition had a higher
prevalence of mental distress [OR = 1.60
(1.29 to 1.99)] compared to those without
121
chronic conditions. Participants who had
engaged in physical activity within the
last three months had a lower prevalence
[OR = 0.82 (0.70 to 0.95)] of mental
distress compared to those who had not.
For variety of reasons, participants who
reported self-perceived unmet health care
needs had a higher proportion of reporting
mental distress compared to those who
reported satisfied health care needs [OR =
1.72 (1.38 to 2.13)]. The highest proportion
of respondents listed the main reason for
unsatisfied health care needs as difficulty
getting access to health care professionals.
The model, adjusted for the pattern of
missing values, is presented here. With
and without adjusting for missing values,
there were slight differences in para­
meter estimates, standard errors and ORs.
Therefore, a GEE-based pattern mixture
model helps us to remove the bias of
estimates due to missing outcome values.
Discussion
Until now, there has been no proper
statistical analysis of the mental health of
Canadian seniors that accounts for both
the complexities of longitudinal NPHS data
(i.e. six cycles) and the hierarchical nature
formed by using a multi-stage complex
Chronic Diseases in Canada
Figure 2
Reasons for not getting needed health care in rural and urban areas
Rural
Urban
survey design. Moreover, this analysis
was adjusted for missing outcome values,
which will remove the bias of estimates.
There were significant differences among
baseline characteristics (i.e. sex, marital
status, geographic areas, education level,
income level, social support, smoking
status, drinking status, general health index,
physical activity and self-perceived unmet
health care needs) and among location
of residence (i.e. urban and rural).
We observed that rural seniors reported a
higher proportion of high mental distress
than urban seniors. In addition, there was
Chronic Diseases in Canada
a significant long-term association between
smoking and mental distress. Our results
revealed that there was a significant longterm association between self-perceived
unmet health care needs and mental health
among seniors (i.e. 55 years and older)
who live in both rural and urban areas.
Most mental health studies using longi­
tudinal data sets as afforded by the NPHS
have focussed on depression. In our study,
the outcome is mental distress. Other
findings of our study were consistent with
those studies. Stephens et al.22 reported that
there is no relationship between mental
122
health and adequate incomes; our results
also revealed this finding. According to
Stephens et al.,23 physical and mental health
problems were related. We also observed
that respondents who reported lower general
health status were associated with mental
distress. Our observation of better men­
tal health in males than in females was
consistent with the findings of Stephens et
al.23 and Østbye et al.35 Similar to the fin­
dings of Stephens et al.,23 chronic physical
health problems were closely associated
with mental health.
Vol 29, No 3, 2009
Table 1
Baseline demography and other information by location of residence for 1994
Mental health
(moderate/high) –
rural counterparts
Mental health
(moderate/high) –
urban counterparts
p-value
< 0.0001
Demographic information
Age group
55 to 59
19.1
15.7
60 to 64
13.5
17.4
< 0.0001
65 to 69
16.6
14.8
< 0.0001
70 to 74
12.5
16.3
< 0.0001
75 to 79
24.6
13.8
< 0.0001
80 and older
18.2
19.4
0.1579
Male
11.8
11.0
< 0.0001
Female
22.2
19.8
< 0.0001
Married/common-law/partnership
17.1
13.7
< 0.0001
Separated/widowed/divorced
14.5
21.6
< 0.0001
Single
22.6
13.9
< 0.0001
Sex
Marital Status
Geographical area
Atlantic
19.8
12.0
< 0.0001
Quebec
24.7
23.0
< 0.0001
Ontario
13.1
14.4
< 0.0001
Prairies
14.6
14.9
0.0039
British Columbia
10.2
11.3
< 0.0001
Less than secondary school graduation
19.1
22.5
< 0.0001
Socio-economic status
Education level
Secondary school graduation
13.9
13.1
< 0.0001
Some post-secondary
13.8
10.5
< 0.0001
Post-secondary graduation
13.5
11.1
< 0.0001
Low
22.7
22.7
1.0000
Middle
16.6
14.8
< 0.0001
4.4
10.7
< 0.0001
Low
19.6
19.6
1.0000
Moderate
16.4
16.8
< 0.0001
High
15.1
11.4
< 0.0001
Income level
High
Social support
Social involvement score
Lifestyle
Smoking status
Current smoker
19.2
20.5
< 0.0001
Ex-smoker
15.5
15.5
1.0000
Non-smoker
17.4
14.7
< 0.0001
Vol 29, No 3, 2009
123
Chronic Diseases in Canada
Table 1 (continued)
Baseline demography and other information by location of residence for 1994
Mental health
(moderate/high) –
rural counterparts
Mental health
(moderate/high) –
urban counterparts
p-value
Drinking status
Current drinker
15.5
13.7
< 0.0001
Ex-drinker
19.1
21.3
< 0.0001
Non-drinker
21.3
21.2
0.4252
Poor
65.1
58.9
< 0.0001
Fair
32.0
33.5
< 0.0001
Good
13.6
14.5
< 0.0001
Very good
6.4
6.2
< 0.0001
Excellent
2.2
7.3
< 0.0001
Yes
18.8
18.8
0.2812
No
10.9
7.6
< 0.0001
Yes
14.3
14.7
< 0.0001
No
31.5
24.1
< 0.0001
Health-related:
General health status
Any chronic condition?
Physical activity within the last three months?
Self-perceived unmet health care needs
Yes
42.5
48.9
< 0.0001
No
16.2
15.0
< 0.0001
Reasons for not getting health care†
Difficulty getting access to health professionals
Financial constraints
47.2
61.5
F*
30.8
Felt inadequate health care provided
F*
31.7
Chose not to see health professionals
29.2
30.2
Other
66.1
58.2
Number of consultations – family doctor within last 12 months
None
11.9
8.2
< 0.0001
1 to 6 times
13.4
15.1
< 0.0001
More than 6 times
32.8
25.8
< 0.0001
F*- Due to confidentiality small percentages are not reported.
†p-values are not reported.
Table 2
Missing distress data patterns over six cycles
Description
Percentage
Completed all six cycles
38.43
One cycle missing
11.87
Two or more cycles missing
20.80
People who died within six cycles
28.90
Chronic Diseases in Canada
124
Vol 29, No 3, 2009
Table 3
Odds ratio (OR) and their 95% confidence interval (95% CI) based on multivariate binary
logistics regression (GEE-based pattern mixture model) of the prevalence of mental distress
OR (95% C.I.)
Drop:
One missing
1.44 (1.13, 1.82)
Two or more missing
1.39 (1.10, 1.76)
Died within six cycles
1.68 (1.36, 2.06)
Completers
Reference
Age Group: 55 to 59
1.76 (1.31, 2.35)
60 to 64
1.49 (1.13, 1.97)
65 to 69
1.21 (0.93, 1.59)
70 to 74
1.07 (0.84, 1.38)
75 to 79
1.01 (0.80, 1.27)
80 and older
Reference
Sex:
Female
1.79 (1.47, 2.17)
Male
Reference
Marital Status
Married/common-law/partnership
1.19 (0.87, 1.64)
Separated/widowed/divorced
1.48 (1.08, 2.03)
Single
Reference
Location of residence
Rural
1.16 (0.98, 1.37)
Urban
Reference
Geographical area
Atlantic
0.68 (0.53, 0.87)
Quebec
1.57 (1.20, 2.06)
Ontario
Reference
Prairies
1.00 (0.80, 1.25)
British Columbia
0.87 (0.67, 1.13)
Socio-economic status
Education level
Less than secondary school graduation
1.36 (1.07, 1.72)
Secondary school graduation
1.28 (0.94, 1.73)
Some post-secondary
1.11 (0.84, 1.47)
Post-secondary graduation
Reference
Income level
Low
1.30 (0.89, 1.89)
Middle
1.20 (0.88, 1.64)
High
Reference
Life-style
Smoking Status
Current smoker
1.26 (1.00, 1.60)
Ex-smoker
1.10 (0.92, 1.31)
Non-smoker
Reference
Vol 29, No 3, 2009
125
In contrast to the findings of Stephens et
al.23 about the province of residence, we
found that there is a significant diffe­
rence in the mental distress of seniors in
some geographic areas. Participants from
Quebec reported a high proportion of high
mental distress compared to their Ontario
participants; Atlantic residents reported a
lower proportion of high mental distress
compared to their Ontario participants.
There is a possibility of reverse causation,
which is shown by other researchers.36–40
Murphy et al.36 reported that smoking at
baseline was not related to a subsequent
incidence of depression. In addition,
they found that participants who become
depressed are more likely to begin or
continue smoking compared to participants
who never become depressed. Lasser et al.39
reported that persons with mental health
problems are about twice as likely to smoke.
Saffer et al.40 found that persons with a
history of mental health problems are 94%
more likely to smoke compared to persons
with no history of mental health problems.
This paper focused on investigating the
long-term association between mental
health and smoking. To determine the
direction of any causation, a special
analysis is required. The NPHS measures
self-reported, unmet health care needs by
asking, “During the past 12 months, was
there ever a time when you felt that you
needed health care, but you didn’t receive
it?” A “yes” response was tabulated as an
unmet need. Because of the wording of the
question addressing unmet needs, it is not
possible to distinguish situations in which
people did not receive services at all from
situations in which they were not received
in a timely manner. Chen et al.41 (2002)
and Sanmartin et al.42 (2002) reported
that individuals with chronic conditions,
including pain or distress were more
likely to report problems with the health
care delivery system. Several studies43 of
seniors’ health reported that the health
care system only marginally improved the
overall health of the senior population. Our
results, which correspond with the findings
of the above studies, suggested a possible
reverse causation. In this sense, the unmet
needs are the effect of the distress, not the
cause. To determine the direction of any
causation, further analysis is required.
Chronic Diseases in Canada
Table 3 (continued)
Odds ratio (OR) and their 95% confidence interval (95% CI) based on multivariate binary
logistics regression (GEE-based pattern mixture model) of the prevalence of mental distress
6.
Stephens T, Joubert N. The economic bur­
den of mental health problems in Canada.
Chronic Dis Can. 2001;22(1):18–23.
7.
Murray CJL, Lopez AD, eds. The global
burden of disease: A comprehensive
assessment of mortality and disability from
diseases, injuries, and risk factors in 1990
and projected to 2020. Vol. 1. Cambridge
(MA): Harvard University Press; 1996. 990 p.
8.
World Health Organization. The world
health report 2001 – mental health: new
understanding, new hope. Geneva: World
Health Organization; 2001. 178 p.
9.
Stephens T, Joubert N. The economic bur­
den of mental health problems in Canada.
Chronic Dis Can. 2001; 22(1):18-23.
OR (95% C.I.)
Health-related:
General health status
Poor
12.14 (7.69, 19.18)
Fair
5.26 (3.51, 7.88)
Good
2.74 (1.85, 4.06)
Very good
1.31 (0.89, 1.95)
Excellent
Reference
Any chronic condition*
Yes
1.60 (1.29, 1.99)
No
Reference
Physical activity within the last three months
Yes
0.82 (0.70, 0.95)
No
Reference
Self-perceived unmet health care needs
Yes
1.72 (1.38, 2.13)
No
Reference
* denotes one or more chronic conditions
These results can be used to improve
the design and delivery of mental health
services to rural and urban seniors. The
results can also be used to target methods
to reduce smoking among seniors who
live in rural and urban areas, and address
the causes of unmet health care needs.
Better design and delivery of services may
result in cost savings in terms of seniors’
psychotherapy appointments, emergency
room visits, medication use and consequent
productivity loss.
Acknowledgements
We would like to thank the Public Health
and the Agricultural Rural Ecosystem
(PHARE) program of the CIHR for funding
for this research. We would also like to
thank the Remote Data Access services of
Statistics Canada. Our special thanks to
Dr. Lilian Thorpe, Geriatric Psychiatrist, for
her valuable comments to improve this paper.
www.phac-aspc.gc.ca/seniors-aines/pubs/
factoids/2001/toc_e.htm
2.
BC Partners for Mental Health and
Addictions Information. The primer: facts
sheets on mental health and addictions
issues [Internet]. British Columbia:
Canadian Mental Health Association –
British Columbia Division; 2003. 127 p.
Available from: www.bcss.org/documents/
primer.pdf
3.
Stotts RC, Smith CK. Smoking patterns
among rural elderly [Internet]. South J
Nurs Res. 2002; 3(4):1-14. Available from:
http://www.snrs.org/publications/SOJNR_
articles/iss04vol03.htm#inter
4.
References
1.
Public Health Agency of Canada. Statistical
snapshots of Canada’s seniors – No. 1 – No. 37
[Internet]. Ottawa: Public Health Agency
of Canada; 2005. Available from: http://
Chronic Diseases in Canada
5.
Spencer C. Older adults, alcohol and
depression [Internet]. National project
report: seeking solutions: Canadian
community action on seniors and alcohol
issues. Vancouver: Gerontology Research
Centre, Simon Fraser University; 2003 May.
Available from: http://www.agingincanada.
ca/Alcohol and Depression_7.pdf
Dewa CS, Lesage A, Goering P, Caveen M.
Nature and prevalence of mental illness
in the work place. Healthc Pap. 2004;
5(2):12–25.
126
10. Statistics Canada. National population health
survey – household component, cycle 6
(2004/2005): longitudinal documentation
[Internet]. Ottawa: Statistics Canada; 2006.
Available from: www.statcan.gc.ca/imdbbmdi/document/3225_D5_T1_V3-eng.pdf
11. Statistics Canada. National population
health survey – household component,
cycle 6 (2004/2005): documentation for
the derived variables and the constant
longitudinal variables [Internet]. Ottawa:
Statistics Canada; 2006. Available from:
www.statcan.gc.ca/imdb-bmdi/document/
3225_D10_T9_V2-eng.pdf
12. Statistics Canada. Public use microdata file
(PUMF): national population health survey –
1994–1995. Ottawa: Statistics Canada;
1995. 64 p.
13. Bellerose C, Lavallée C, Tremblay D. Cahier
technique et méthodologique. Enquête
sociale et de santé 1992–1993.Vol. 1,
Montréal: Gouvernement du Québec,
Ministère de la Santé et des Services
sociaux; 1995. 134 p.
14. Buckley NJ, Denton FT, Robb AL, Spencer
BG. Socio-economic influence on the
health of older people: estimates based
on two longitudinal surveys. Hamilton:
Research Institute for Quantitative Studies
in Economics and Population (QSEP); 2003.
Report No.: 387.
Vol 29, No 3, 2009
15. Michiels B, Molenberghs GM, Bijnens L,
Vangenengden T, Thijs H. Selection models
and pattern mixture models to analyze
longitudinal quality of life data subject to
dropout. Stat Med. 2002;21:1023–41.
26. SAS Institute Inc. SAS/STAT 9.1 user’s
guide. Cary (NC): SAS Institute Inc.; 2005.
5136 p. Available from: http://support.
sas.com/documentation/onlinedoc/91pdf/
index_913.html
36. Murphy JM, Horton NJ, Monson RR,
Laird NM, Sobol AM, Leighton AH.
Cigarette smoking in relation to depression:
historical trends from the Sterling Country
Study. Am J Psychiatry. 2003;160:1663–9.
16. Little RJ, Rubin DB. Statistical analysis with
missing data. Chapters 14 and 15. New York:
John Wiley and Sons; 2002. p. 292-348.
27. Diggle PJ, Liang K-Y, Zeger SL. Analysis
of longitudinal data. New York: Oxford
University Press; 1994. 253 p.
17. Little RJ. Pattern – mixture models for
multivariate incomplete data. J Am Stat
Assoc. 1993;88:125–34.
28. Liang K-Y, Zeger SL. Longitudinal data
analysis using generalized estimating
equations. Biometrika. 1986;73:13–22.
37. Anda RF, Williamson DF, Escobedo LG,
Mast EE, Giovino GA, Remington PL.
Depression and the dynamics of smoking: a
national perspective. JAMA. 1990;264(12):
1541–5.
18. Little RJ. A class of pattern–mixture models
for normal missing data. Biometrika.
1994;81:471–83.
29. Zeger SL, Liang K-Y. Longitudinal data
analysis for discrete and continuous
outcomes. Biometrics. 1986;42:121–30.
19. Kessler R, Mroczek D. Final versions of our
non-specific psychological distress scale.
Ann Arbor (MI): Survey Research Centre of
the Institute for Social Research, University
of Michigan. Memo dated March 10, 1994.
30. SAS Institute Inc. Longitudinal data analysis
with discrete and continuous responses:
instructor based training. Cary (NC): SAS
Institute Inc.; 2002. p. 3–32.
39. Lasser K, Boyd JW, Woolhandler S,
Himmelstein DU, McCormick D, Bor DH.
Smoking and mental illness: a populationbased prevalence study. JAMA. 2000;
284(20):2606–10.
31. Statistics Canada. Estimation of the variance
using the bootstrap weights. User’s guide for
the BOOTVARE_V21.SPS program. Version
2.1. Ottawa: Statistics Canada; 2005.
40. Saffer H, Dave D. Mental illness and the
demand for alcohol, cocaine and cigarettes.
Economic Inquiry, Oxford University Press,
2005 April; Vol. 43(2):229-246.
32. Rao JN. Interplay between sample survey
theory and practice: an appraisal. Surv
Methodol. 2005;31(2):117–38.
41. Chen J, Hou F. Unmet needs for health
care. Health Rep. Ottawa: Statistics Canada;
2002. 2002;13(2)23–34. Catalogue. No.:
82-003-XIE.
20. Patten SC, Beck CA. Major depression and
mental health care utilization in Canada:
1994–2000. Can J Psychiatry. 2004;49(5):
303-9.
21. Wang J, El-Guebaly N. Sociodemographic
factors associated with comorbid major
depression episodes and alcohol depend­
ence in the general population. Can
J Psychiatry. 2004 Jan;49(1):37–44.
22. Wilkins K, Beaudet MP. Work stress and
health. Health Rep. 1998;10(3):47–62.
23. Stephens T, Dulberg C, Joubert N. Mental
health of the Canadian population:
a comprehensive analysis. Chronic Dis
Can. 2000;20(3):118–26.
24. Baggaley RF, Ganaba R, Fillippi V, Kere M,
Marshall T, Sombie I, Storeng KT, Patel V.
Detecting depression after pregnancy: the
validity of the K10 and K6 in Burkina Faso.
Trop Med Int Health. 2007;12(10):1225–9.
25. Allison, PD. Logistic regression using SAS:
theory and application. Cary (NC): SAS
Institute; 1999. p. 5–78, 179–213.
Vol 29, No 3, 2009
33. Binder DA, Roberts GR. Statistical inference
in survey data analysis: where does the
sample design fit in? Paper presented at:
Statistics Canada Research Data Centre
Conference Program, University of
McMaster; 2003 Sep 24-25; Hamilton, ON.
[cited 2008 Mar 05]. Available from: http://
socserv.socsci.mcmaster.ca/rdc2003/
binderoberts.pdf
38. Hughes JR, Hatsukami DK, Mitchell JE,
Dahlgren LA. Prevalence of smoking among
psychiatric outpatients. Am J Psychiatry.
1986;143:993–7.
42. Sanmartin C, Houle C, Tremblay S,
Berthelot JM. Changes in unmet health
care needs. Health Rep. 2002;13(3):15–21.
Catalogue. No.: 82-003-XIE.
43. Martin-Matthews A. Sharing the learning:
Health transition fund: Synthesis series.
Ottawa: Health Canada; 2002. Catalogue
No: H13-6/2002-7.
34. Fleming SA, Bains N, Hunter DJ, Lam M.
Social support and health care use
among a sample of healthy Canadians:
a longitudinal analysis of the national
population health survey. Kingston (ON):
Health information partnership, Eastern
Ontario Region; 2004. 58 p.
35. Østbye T, Kristjansson B, Hill G, Newman SC,
Brouwer RN, NcDowell I. Prevalence
and predictors of depression in elderly
Canadians: the Canadian study of health and
aging. Chronic Dis Can. 2005;26(4):93–9.
127
Chronic Diseases in Canada
Factors associated with the adoption of a smoking ban
in Quebec households
É. Ouedraogo, MD (1); F. Turcotte, MD (2); M. J. Ashley, MD, DPH (3); J. M. Brewster, PhD (4); R. Ferrence, PhD (5)
Abstract
The home represents an important source of exposure to environmental tobacco smoke
for non-smokers, including children, who live with smokers. Our goal is to identify the
sociodemographic factors associated with the adoption of smoking bans in “smoker
households” in Quebec. Selected associations are compared with three other Canadian
provinces (Ontario, British Columbia and Nova Scotia). This is a cross-sectional study
involving 2648 respondents. Logistic regression analysis is employed. Few smoker
households in Quebec (21%) have a ban on smoking; the presence of a non-smoker
is strongly linked to the existence of such a ban; the presence of a child under the age
of 6 is less strongly associated with the adoption of a ban in Quebec than in the other
provinces, and the presence of an adolescent shows no association whatsoever. In
addition to the child health benefits of household smoking bans, greater emphasis should
be placed on the impact that such bans can have on children’s future smoking behaviour.
One option from a health promotion standpoint might be to organize a campaign aimed
at non-smokers who live with smokers, in order to urge them to be less tolerant of
environmental tobacco smoke.
Key words: environmental tobacco smoke, smoke-free home, sociodemographic factors,
Quebec households
Introduction
The harmful health effects of environmental
tobacco smoke (ETS) on non-smokers
are now well-established.1-3 In Canada,
measures to limit exposure to ETS have
been introduced in workplaces and in a
variety of public places as well.i Numerous
studies have shown that such measures not
only improve the health and comfort level
of non-smokers, but also reduce smokers’
consumption of tobacco products.4
However, the home remains an important
source of exposure to ETS, particularly
for pre-school children. The favourable
response to smoking restrictions in
public places and places of work and
the diminishing social acceptability of
smoking suggest that such measures
could be extended to the private sphere.
In Ontario, the proportion of non-smokers
who favour a complete ban on smoking in
the presence of small children increased
by 15.4% between 1992 and 1996, while
opposition to such measures on the part
of smokers decreased by 8.6% during the
same period.5
There is only limited data on smoking
bans in Canadian households. A 1995
study carried out by Health Canada
with households that included children
aged 12 or younger found that 19% of
such households had a complete ban on
smoking, 44% had a partial ban, and
37% had no restrictions.6-7 However, the
response rate for this study was only 50%.
Despite a remarkable reduction in the
prevalence of smoking, Quebec remains
above the Canadian average in terms
of ETS exposure. According to the most
recent Canadian Tobacco Use Monitoring
Survey (CTUMS) conducted in 2006, the
prevalence of Quebecois children aged 0
to 17 who are exposed to ETS in the home
is 21.6%, compared to 11.2% in Canada
as a whole.8
As far as we know, there are no pop­
ulation data describing the effect of
sociodemographic factors on the adoption
of household smoking bans in Quebec.
The objective of this study is to identify
the sociodemographic characteristics of
households and respondents most strongly
associated with the existence of a smoking
ban in Quebec households that include
at least one smoker, and to establish
comparisons with three other Canadian
provinces (Ontario, British Columbia and
Nova Scotia).
i Measures to limit exposure to ETS are in place even in the “private” environment of the car: under a new Ontario law that came into effect on January 21, 2009,
charges can be laid against any person found smoking in a vehicle that is also carrying a child under the age of 16.
Author References
1 Département Médecine Sociale et Préventive, Faculté de médecine, Université Laval, QC
2 Département Médecine Sociale et Préventive, Faculté de médecine, Université Laval, QC
3 Dalla Lana School of Public Health, University of Toronto, ON
4 Ontario Tobacco Research Unit, Dalla Lana School of Public Health, University of Toronto, ON
5 Ontario Tobacco Research Unit, Centre for Addiction and Mental Health University of Toronto, ON
Correspondence: Éva Ouedraogo, MD, Département Médecine Sociale et Préventive, Faculté de médecine, Université Laval, Tel.: 418-666-7000 ext. 236, Fax: 418-666-2776 ;
Email: [email protected]
Chronic Diseases in Canada
128
Vol 29, No 3, 2009
Methodology
This is a cross-sectional population study.
The data are derived from the National
Survey on Environmental Tobacco Smoke
in the Home and were collected between
June 2001 and January 2002 by the Institute
for Social Research at York University,
under the direction of the Ontario Tobacco
Research Unit (OTRU).
The sampling and questionnaires have
already been described.9 In brief, the initial
sample comprised 14 600 households;
respondents were asked to complete an
initial questionnaire to determine the
smoking status of their households, as well
as the presence of children under the age
of 18. In order to gather information on
smoking bans, a second sample of “smoker
households” was selected. In this sample,
households that included at least one adult
smoker and a child were over-represented
since the goal was to obtain a sample size
that was sufficiently large for statistical
analysis; on the basis of these criteria,
5000 households from every province in
Canada were selected. From these house­
holds, 2648 smoker households with
and without children were selected in
four provinces (Quebec, Ontario, British
Columbia and Nova Scotia). While the
original survey extended to all parts of
Canada, Ontario and British Columbia
were selected in order to compare Quebec
with provinces that show better smoking
ban prevalence rates in the literature, and
Nova Scotia was selected because it is
comparable to Quebec, with the exception
of its cultural context.
A smoker household is defined as a house­
hold in which at least one person aged 18 or
over smokes cigarettes, cigars, cigarillos or a
pipe on a daily or occasional basis. Although
information was collected on every smoker
in any given household, few people under
the age of 18 live away from their parents
and are in a position to establish their own
household smoking rules.9
Participants were selected on the basis of
computer-generated telephone numbers.
Information was collected using a ques­
tionnaire that was administered with
the aid of computer-assisted telephone
Vol 29, No 3, 2009
interviewing (CATI) technology. To ensure
an optimal response rate, a maximum of
14 calls were made to any number for
which contact was not established on the
first call; 10 of these calls were made in
the evening or on weekends.
The dependent variable is the household
smoking ban. A smoke-free household is a
home in which all occupants refrain from
smoking inside at all times. The beha­
viour of visitors and guests with respect
to household smoking bans was not
taken into account. Smoking bans were
measured on the basis of people’s response
to the following question: “Do you smoke
cigarettes, cigars, cigarillos or a pipe at
home every day, from time to time, or not
at all?” In cases where the household had
more than one smoker, the question was
put to the respondent for every other
smoker. This information provided a means
of defining a variable with two response
levels: (1) a complete household smoking
ban; or (2) a partial ban or no ban.
The results are expressed as weighted
prevalence ratios. A weighting system was
introduced in order to take into account the
unequal probability of households being
selected on the basis of their composition,
since smoker households with children
were more likely to be selected than those
with no smokers or children. We also
needed to take into account the unequal
distribution of households by province of
origin. A weighting coefficient was assigned
to each type of household.
Initially, univariate unconditional logistic
regression was used to identify the vari­
ables most strongly associated with the
adoption of a household smoking ban.
Then, a multivariate analysis was per­
formed. Provincial variations in the effect
of a sociodemographic characteristic on
the adoption of a smoking ban were evalu­
ated using a logistic regression model
that was stratified for the province and
adjusted for the other characteristics. To
avoid the problem of colinearity, variables
contributing the same information were
not included in the model at the same
time. The weighted prevalence ratios thus
obtained were used to estimate the asso­
ciation between each variable and the
129
presence of a household smoking ban with
a confidence interval of 95%.
These statistical analyses were performed
using SAS software.
Results
Sociodemographic characteristics and
household smoking ban prevalence
A total of 2648 respondents representing
households with at least one adult smoker
agreed to complete the questionnaire in
the four provinces. The survey response
rate was 62%. Tables 1 and 2 present
the characteristics of respondents and
households respectively. The proportion
of female respondents was slightly greater
than that of male respondents; more than
half of all respondents were in a couple
relationship. Close to 50% had a highschool education or less. Sixty-three per­
cent of respondents were cigarette and/
or pipe, cigar or cigarillo smokers; 80% of
these smokers smoked every day. Slightly
over one third of the households had
at least one child; in 37% of cases that
child was under the age of 6. In Quebec,
949 smoker households agreed to complete
the questionnaire. The characteristics of the
Quebec sample are comparable to those of
the total sample. Given the large number of
missing values for household income, the
effect of this variable was not included in
the analysis of associations.
The weighted prevalence of household
smoking bans in Quebec is 21%; this is sig­
nificantly different from the rates observed
in Ontario and British Columbia which
were, respectively, 43.7% (p < 0.0001) and
52.1% (p < 0.0001), but not statistically
different from the rate observed in Nova
Scotia (32%) (p = 0.64).
Factors associated with the adoption
of a household smoking ban in Quebec
Table 3 presents the results of the univariate
and multivariate analyses. Several
respondent characteristic variables are
significantly associated with the adoption
of a household smoking ban in a univariate
analysis, but this is no longer the case once
other variables are taken into account.
Thus, the age, gender and smoking status
of respondents have no impact on the
Chronic Diseases in Canada
Table 1
Characteristics of respondents in the provinces of Quebec, Ontario, British Columbia and Nova Scotia
Variables
Quebec
Ontario
British Columbia
Nova Scotia
Totala
nb
%c
%Wd
n
%
%W
n
%
%W
n
%
%W
n
%
%W
18-24
156
16.5
22.0
165
14.0
19.2
52
14.7
19.2
18
13.8
19.1
391
15.0
20.1
25-64
718
76.1
70.5
949
80.8
75.1
267
75.6
71.1
103
79.2
73.5
2037
78.3
72.9
≥ 65
69
7.3
7.5
61
5.2
5.7
34
9.6
9.6
9
6.9
7.4
173
6.6
7.0
Total
943
Age group (in years)
1175
353
130
2601
Gender
Women
549
57.8
56.0
669
55.5
52.8
194
53.7
49.2
66
50.0
42.9
1478
55.8
52.9
Men
400
42.1
44.0
537
44.5
47.2
167
46.3
50.8
66
50.0
57.1
1170
44.2
47.1
Total
949
1206
361
132
2648
Smoking status
Current smoker
646
68.4
63.5
814
68.0
62.9
222
61.7
57.0
97
74.0
71.8
1779
67.5
62.6
Former smoker
137
14.5
15.8
157
13.1
15.1
59
16.4
47.0
16
12.2
12.2
369
14.0
15.5
Non-smoker
162
17.1
20.7
227
18.9
22.0
79
21.9
26.0
18
13.8
16.0
486
18.5
21.9
Total
945
1198
360
131
2634
Marital status
Living with
a spouse
560
59.4
58.9
695
58.1
57.0
205
57.1
56.5
84
63.6
63.6
1544
58.7
57.8
Single/separated/
divorced
383
40.7
41.1
502
41.9
43.0
154
42.9
43.5
48
36.4
36.4
1087
41.3
42.2
Total
943
1197
359
132
2631
Education level
Bachelor’s degree
or higher
132
14.0
13.3
218
18.4
18.6
70
19.5
20.5
15
11.6
14.5
435
16.6
17.0
Post secondary
318
33.8
35.6
564
47.6
47.1
186
51.8
51.1
25
19.4
18.9
1093
32.4
33.1
Secondary
or lower
491
52.2
51.1
402
34.0
34.3
103
28.7
28.4
89
69.0
66.6
1085
51.0
49.9
Total
941
1184
359
129
2613
Perception of ETS as
An important
health problem
566
59.9
59.8
844
70.4
70.0
257
71.4
71.4
96
73.3
74.2
1763
67.9
67.1
A health problem
of little or
no importance
380
40.1
40.2
354
29.6
30.0
103
28.6
28.6
35
27.7
25.8
872
33.1
32.9
Total
946
1198
360
131
2635
ETS in the workplace
a
b
c
d
Complete ban
370
70.9
70.3
472
78.3
78.9
143
90.5
89.6
31
68.9
66.0
1016
76.8
77.3
Partial ban
131
29.1
29.7
131
21.7
21.1
15
9.5
10.4
14
31.1
34.0
291
23.2
22.7
Total
501
603
158
45
1307
The total may differ from the sample size (2648) due to missing values.
n represents the size of the sample.
% represents unweighted proportions.
%W represents weighted proportions. A weighting coefficient was assigned to each type of household to take into account the unequal probability of a household being selected on
the basis of its composition.
Chronic Diseases in Canada
130
Vol 29, No 3, 2009
Table 2
Characteristics of households in the provinces of Quebec, Ontario, British Columbia and Nova Scotia
Variables
Quebec
nb
Ontario
British Columbia
Nova Scotia
Totala
%c
%Wd
n
%
%W
n
%
%W
n
%
%W
n
%
%W
719
59.6
68.8
225
62.3
70.0
72
54.5
61.7
1592
60.1
68.7
487
40.4
31.2
136
37.7
30.0
60
45.5
38.3
1056
39.9
31.3
Presence of ≥ 1 adult non-smoker
Yes
576
60.7
68.9
No
373
39.3
31.1
Total
949
1206
361
132
2648
Presence of a child
Yes
449
47.3
35.5
No
500
52.7
64.5
Total
949
621
51.5
39.8
171
47.4
35.2
585
48.5
60.2
190
52.6
64.8
1206
361
64
48.5
37.7
1305
49.3
37.7
68
51.5
62.3
1343
50.7
62.3
132
2648
Age of the child
5 or under
155
34.6
33.3
236
38.5
38.1
69
41.1
41.0
20
31.7
32.3
480
37.2
36.8
6-17
293
65.4
66.7
377
61.5
61.9
99
58.9
59.0
43
68.3
67.7
812
62.8
63.2
Total
448
613
168
63
1292
Family context
Adult non-smoker
and child
282
29.7
24.8
370
30.9
26.6
108
30.2
26.6
29
22.1
17.4
789
29.9
25.6
Adult non-smoker,
no child
293
30.9
44.1
342
28.5
42.0
114
31.8
43.0
42
32.1
44.3
791
30.0
42.9
Adult smoker
and child
166
17.5
10.7
243
20.3
12.8
60
16.8
9.7
34
25.9
17.0
503
19.1
11.8
Adult smoker,
no child
207
21.8
20.4
243
20.3
18.6
76
21.2
20.6
26
26.9
21.3
552
21.0
19.6
Total
948
1198
358
131
2635
Family income
63
8.6
10.5
144
15.7
19.8
30
11.1
12.3
10
9.7
11.8
247
12.2
15.2
55,000-95,000
158
21.3
22.8
256
28.0
27.7
79
29.4
31.4
16
15.5
16.8
509
25.1
26.2
≤ 55,000
519
70.1
66.7
516
56.3
52.5
160
59.5
56.3
77
74.8
71.4
1272
62.7
58.6
Total
740
> 95,000
a
b
c
d
916
269
103
2028
The total may differ from the sample size (2648) due to missing values.
n represents the size of the sample.
% represents unweighted proportions.
%W represents weighted proportions. A weighting coefficient was assigned to each type of household to take into account the unequal probability of a household being selected on
the basis of its composition.
adoption of a smoking ban, nor does the
respondent’s status as a single or attached
person. The perception of ETS as a major
health problem has an effect that falls
within the range of statistical significance.
The education level of respondents has an
influence on the adoption of a smoking
ban, with an adjusted prevalence ratio (PR)
of 3.26 (95% confidence interval [CI] 2.0
to 5.3).
The impact of each household characteris­
tic on the adoption of a household smoking
ban remains constant regardless of whether
the other variables are taken into account.
The most influential characteristic of all
Vol 29, No 3, 2009
is the presence of an adult non-smoker,
which is associated with a fourfold increase
in the probability that a household smoking
ban will be in place when compared to
households in which all the adults are
smokers. The presence of a child appears to
have little influence on the decision to adopt
a household smoking ban. In fact, the impact
of this characteristic varies depending on
the age of the child: the presence of a child
under the age of 6 significantly doubles
the probability that smoking restrictions
will be in place, while the presence of a
pre-adolescent or adolescent child does not
appear to have a significant impact. When
both a child and an adult non-smoker live
131
in a given household, the chances that a
household smoking ban will be in place are
seven times greater than in households that
do not include one or the other.
Comparisons between Quebec and the
other provinces
Table 4 presents provincial variations in the
adjusted effect of each sociodemographic
characteristic on the adoption of a house­
hold smoking ban. Certain variables were
found to have an effect in all four pro­vinces,
although the degree of influence varied.
Thus, the presence of an adult non-smoker
in the household is the characteristic
most strongly associated with household
Chronic Diseases in Canada
Table 3
Crude and adjusted prevalence ratios (PRs) related to the adoption of a smoking ban (SB)
in Quebec households, by sociodemographic variable Variables
na
SB (%W)b
Crude PRs
Adjusted PRsc
18-24
156
29.5
2.4 (1.2-5.0)
0.71 (0.3-1.8)
25-64
718
18.1
1.3 (0.7-2.5)
0.65 (0.3-1.4)
≥ 65
69
13.0
1.0
1.0
Women
549
21.1
1.24 (0.9-1.7)
1.38 (0.9-1.9)
Men
400
17.7
1.0
1.0
162
29.6
2.45 (1.6-3.7)
1.09 (0.7-1.7)
Respondent variables
Age group (in years)
Gender
Smoking status
Non-smoker
Former smoker
138
31.9
2.73 (1.8-4.1)
1.46 (0.9-2.4)
Current smoker
649
14.6
1.0
1.0
560
19.8
1.76 (1.0-3.0)
0.82 (0.4-1.5)
Marital status
Living with a spouse
Single
241
24.1
2.24 (1.3-4.0)
1.21 (0.6-2.5)
Separated/divorced
142
11.3
1.0
1.0
134
31.0
3.5 (2.2-5.5)
3.26 (2.0-5.3)
Education level
Bachelor’s degree or higher
Postsecondary
319
28.2
3.1 (2.1-4.5)
2.73 (1.8-4.1)
Secondary or lower
438
11.0
1.0
1.0
Perception of ETS as
An important health problem
566
23.3
1.8 (1.3-2.6)
1.46 (1.0-2.1)
A health problem of little
or no importance
380
14.2
1.0
1.0
Yes
576
27.6
4.70 (3.1-7.2)
3.81 (2.3-6.4)
No
373
7.5
1.0
1.0
Household variables
Presence of ≥ 1 adult non-smoker
Presence of children
Yes
449
21.8
1.29 (0.9-1.8)
1.32 (0.9-1.9)
No
500
17.8
1.0
1.0
155
27.7
1.79 (1.2-2.7)
1.9 (1.1-3.1)
Age of the child (in years)
≤5
6-17
293
18.8
1.08 (0.7-1.6)
1.1 (0.7-1.6)
No children
500
20.7
1.0
1.0
283
28.6
7.1 (3.7-13.8)
6.5 (3.0-13.9)
Effect of family context
Adult non-smoker and child
Adult non-smoker, no child
293
26.6
6.5 (3.3-12.5)
5.6 (2.7-11.6)
Adult smoker and child
166
10.2
2.0 (0.9-4.5)
2.4 (1.0-5.5)
207
5.3
1.0
Adult smoker, no child
a The total may differ from 949, due to missing values.
b SB proportions are weighted.
c Prevalence ratios have been adjusted for the other variables.
Chronic Diseases in Canada
132
smoking bans; this charac­teristic is
associated with an almost fourfold
increase in the probability that smoking
will be banned in the home in Quebec
and Nova Scotia, but only a twofold
increase in British Columbia. The second
variable associated with the adoption of a
household smoking ban is the presence of
a child under the age of 6; however, the
association is weaker and non-significant
in Quebec, whereas in Nova Scotia and
Ontario it significantly triples the likelihood
that a household smoking ban will be in
place. The presence of a child over the age
of 6 was found to have an effect only in
British Columbia, where the presence of
children was observed to have a remar­
kably uniform impact, regardless of age.
Other characteristics were found to have
an influence that varies by province. The
presence of a child of any age is associated
with the adoption of a household smoking
ban in Ontario and British Columbia;
although non-significant, the strength of this
association in Nova Scotia approaches 2.
In British Columbia, the probability that a
household smoking ban will be in place is
4 times greater among adults aged 18 to 24
and 25 to 64 than among those aged 65 or
over. The influence of a person’s smoking
status can be observed in Ontario, where
former smokers are significantly more likely
to adopt a smoking ban than non-smokers
and current smokers. The education level
of respondents was significantly associated
with the adoption of household smoking
restrictions in Quebec and Ontario. Neither
gender, nor the presence or absence of a
spouse appeared to influence the adoption
of a smoking ban in any of the provinces
included in the study.
Discussion
The presence of an adult non-smoker in
households that include at least one smoker
is the factor most often associated with the
adoption of a household smoking ban in
Quebec and the three other provinces
in this study. This confirms the findings
of Borland in Victoria, Australia,10 who
found that smokers in mixed households
were 4.7 times more likely to always smoke
outside the home than smokers in an allsmoker household (95% CI 3.7 to 6.0).
However, the impact of this variable is less
Vol 29, No 3, 2009
Table 4
Variations in the effects of sociodemographic characteristics on the adoption
of a smoking ban, by province
Variables
Quebec
PRa (95% CI)
Ontario
PR (95% CI)
BC
PR (95% CI)
Nova-Scotia
PR (95% CI)
Respondent variables
Age group (in years)
18-24
0.7 (0.4-1.2)
1.86 (1.09-3.2)
3.6 (1.5-8.6)
1.0 (0.4-2.4)
25-64
0.5 (0.3-0.8)
1.42 (0.91-2.2)
3.5 (1.7-7.2)
1.3 (0.3-5.1)
1.0
1.0
1.0
1.0
1.3 (0.9-1.8)
1.3 (1.0-1.6)
1.1 (0.7-1.7)
0.8 (0.4-1.7)
1.0
1.0
1.0
1.0
Non-smoker
1.2 (0.8-1.8)
1.2 (0.8-1.6)
0.5 (0.3-0.9)
0.5 (0.1-1.5)
Former smoker
1.5 (0.9-2.3)
1.6 (1.1-2.3)
0.5 (0.2-0.9)
1.5 (0.5-4.8)
Current smoker
1.0
1.0
1.0
1.0
Living with a spouse
0.6 (0.4-1.1)
1.0 (0.7-1.4)
0.8 (0.4-1.4)
0.9 (0.3-2.3)
Single
1.0 (0.5-1.7)
1.0 (0.6-1.5)
0.9 (0.5-1.8)
0.9 (0.3-2.8)
1.0
1.0
1.0
1.0
≥ 65
Gender
Women
Men
Smoking status
Marital status
Separated/divorced
Education level
Bachelor’s degree or higher
2.8 (1.7-4.5)
2.4 (1.7-3.4)
1.6 (0.9-2.9)
1.5 (0.5-4.6)
Postsecondary
2.4 (1.6-3.4)
1.5 (1.1-2.0)
2.1 (1.2-3.5)
2.2 (0.5-5.7)
1.0
1.0
1.0
1.0
1.5 (1.0-2.1)
1.3 (0.9-1.7)
1.5 (0.9-2.5)
1.6 (0.6-4.1)
1. 0
1.0
1.0
1.0
Yes
3.8 (2.4-5.9)
3.5 (2.6-4.7)
2.4 (1.5-3.9)
3.7 (1.6-8.8)
No
1.0
1.0
1.0
1.0
Yes
1.2 (0.9-1.8)
1.8 (1.4-2.3)
2.6 (1.6-4.1)
1.9 (0.9-4.3)
No
1.0
1.0
1.0
1.0
≤5
1.6 (1.0-2.6)
3.0 (2.1-4.2)
2.5 (1.4-4.7)
3.5 (1.2-10.5)
6-17
1.01 (0.7-1.5)
1.3 (0.9-1.8)
2.7 (1.6-4.6)
1.4 (0.6-3.5)
1.0
1.0
1.0
1.0
Adult non-smoker and child
5.7 (2.8-11.3)
6.6 (4.3-10.2)
6.3 (3.1-12.5)
7.9 (1.9-32.8)
Adult non-smoker, no child
5.3 (2.7-10.5)
4.1 (2.6-6.3)
2.4 (1.2-4.7)
4.1 (1.0-16.3)
Adult smoker and child
2.0 (0.9-4.5)
2.2 (1.4-3.5)
2.5 (1.2-5.3)
2.2 (0.5-9.6)
Adult smoker, no child
1.0
1.0
1.0
1.0
Secondary or lower
Perception of ETS as
An important health problem
A health problem of little
or no importance
Household variables
Presence of ≥ 1 adult non-smoker
Presence of children
Age of the child (in years)
No children
Effect of family pressure
a Prevalence ratios have been adjusted for the other variables.
Vol 29, No 3, 2009
133
pronounced in British Columbia, which
suggests that smokers are themselves
sufficiently aware of the harmful effects
of ETS to adopt a household smoking ban
of their own accord, or have a greater
willingness to smoke outdoors; British
Columbia has a more temperate climate
than the other provinces included in this
study. From a health promotion standpoint,
this suggests a need to consider campaigns
that target non-smokers who live with
smokers. Unlike classic campaigns that
exhort smokers to go outside to smoke, a
campaign urging non-smokers to show less
tolerance with respect to environmental
tobacco smoke might well contribute to a
reduction in ETS exposures.
Surprisingly, the presence of children (all
ages combined) in smoker households in
Quebec is not linked to the adoption of
smoking bans, as was found to be the case
in British Columbia and Ontario. However,
the situation is somewhat improved
if the child is under the age of 6. While
the presence of a child appears to have
only a weak influence on the adoption
of household smoking bans in Quebec,
progress has nonetheless been made since
1996, when a Statistics Canada study found
that neither the presence of a child, nor
the age of the child, had any influence on
smoking in the home.7 Still, these data are
not particularly encouraging in light of the
positive impact that household smoking
bans have been shown to have on the
prevalence of smoking in adolescence.
Indeed, a number of authors, including
Farkas,11 have found that adolescents who
live in smoke-free households were only
74% as likely to be smokers as adolescents
who live in households where there are no
smoking restrictions. Farkas also observed
that adolescents who already smoke are
twice as likely to stop smoking if they live
in a smoke-free household. Wakefield12
has observed that transition through the
different stages of tobacco addiction was
significantly slower for adolescents who
reported that their household had a
smoking ban. The results of our study
have important operational implications
for public health programs, since they
suggest parents are more inclined to adopt
a smoking ban when their children are
young but become less inclined to do so
Chronic Diseases in Canada
as their children age. In addition to the
link between child health and household
smoking bans, greater emphasis should be
placed on the impact of smoking bans on
children’s future smoking behaviour.
The perception of ETS as an important
health issue also has an impact that falls
within the range of statistical significance.
The proportion of respondents who viewed
ETS as a health problem of little or no
importance was substantial in 2001 (40%).
Although that proportion has probably
declined since the data were collected, it
remains a variable of considerable interest
from a public health perspective by virtue
of the fact that it is modifiable.
Contrary to the findings of some authors,
the adoption of household smoking bans
was not linked to the gender of respondents
in Quebec or in the three other provinces in
this study. These results differ from those
of Gilpin13 and Merom14 who found that
women adopt such household policies
significantly less often than men, with
odds ratios of 0.72 CI (0.61 to 0.84) and
0.84 CI (0.72 to 0.96) respectively. As far
as the age of respondents is concerned, we
observed that this characteristic had no
influence on the adoption of smoking bans
in Quebec households. This is surprising,
given that participants aged 18 to 24 belong
to a generation whose members are better
informed about the harmful effects of ETS
and are more likely to have small children
at home.
The strength of this study resides in the
potential to identify the factors which are
associated with the presence of a smoking
ban in “smoker households” in Quebec.
In public health, it is crucial to be able to
identify factors that can explain at least part
of a problem that affects the population and
are viewed as modifiable. The results of
this study suggest that ETS in the home is
at least partly modifiable if a less tolerant,
non-smoking adult lives with a smoker and
if the perception of ETS as an important
health problem increases. The results also
suggest that parents are less likely to
perceive the benefits of a household smoking
ban when their children are 6 or older. These
hypotheses, once confirmed, will serve as
Chronic Diseases in Canada
a basis for modifying health promotion
interventions. In fact, a Chinese study15
on the predictive factors of lax household
smoking bans points to the presence of
a smoking partner (odds ratio = 2.78,
p < 0.05).
One might argue that these data are
too old, since they date back to 2001.
However, the results of this study remain
fully relevant for a number of reasons. To
begin with, the study constitutes the only
epidemiological index currently at our
disposal, as it is the first study of its kind
to include a representative sample of the
Canadian population; also, as far we know,
no other study has ever been published on
the factors associated with the adoption
of household smoking bans in Quebec.
As such, the results of this study will
ultimately lend themselves to comparisons
with future studies. Finally, the results of
the 2006 Canadian Tobacco Use Monitoring
Survey (CTUMS) show that these data are
still relevant, since Quebec remains above
the Canadian average in terms of household
second-hand smoke exposure for children
aged 0 to 17; in Quebec, 18.4% of chil­
dren aged 0 to 11 and 25.8% of children
aged 11 to 17 are regularly exposed to ETS,
compared to 3% and 7% respectively in
British Columbia, 5% and 8.9% in Ontario,
and 10.5% and 18% in Nova Scotia.8
Quebec’s Tobacco Act, which prohibits
smoking in many places (health care
facilities, daycare centres, social services,
schools and institutions of higher learning,
sports and recreational facilities, arts and
cultural facilities, public transit) has been
in place since 2001. On May 31, 2006, smo­
king was prohibited in bars and restaurants
as well. This new measure, combined with
the Tobacco Act, will no doubt have an
impact on the social acceptability of smo­
king, but its effect on household smoking
bans is unlikely to be so great as to invali­
date the results of the present study.
This study comprises a few limitations that
should be noted. The response rate of 62%
is modest and, as such, may cast doubt on
the reliability of household smoking ban
data. However, it is well within the range
of response rates routinely obtained with
134
the method used. Indeed, survey response
rates have been trending downward in
recent years, compared to the 1970s, 1980s
and 1990s. 15,16
Moreover, dividing the household smoking
ban variable into two categories (complete
household smoking ban and occasional/
no ban) tends to group into the latter
category both households that have never
had a ban and households in which a ban
exists but is occasionally disregarded.
Although having three categories might
have provided a more accurate reflection of
reality, our interest lies in households that
have a complete smoking ban, since our
goal is to identify the variables associated
with such bans; furthermore, the small size
of certain sub-groups limited our options in
this regard.
As for the weighting of prevalence esti­
mates in this study, the coefficients should
have been calculated on the basis of the
distribution of households as reflected
in national statistics. In the absence of
such statistics, these coefficients were
obtained on the basis of the distribution
of households participating in the survey,
which constitutes a reliable estimate of
national data. Consequently, the estimated
prevalences in our study are probably close
to the real numbers.
Finally, the results for Nova Scotia should
be interpreted with prudence, given the
small sample size (n = 132). This factor
makes it difficult to distinguish between
an absence of association and the study’s
inability to detect an association where
one exists.
Conclusion
The performance of Quebec’s tobacco
control strategy, the diminishing social
acceptability of smoking, and legislation
that prohibits smoking in public places and
workplaces have brought about a reduction
in smoking prevalence. However, Quebec’s
children do not all enjoy the benefits of
these changes and remain more exposed to
ETS than children in other parts of Canada.
The results of this study provide a more
accurate profile of children’s exposure
Vol 29, No 3, 2009
to ETS in the home. They suggest that
children who live in smoker households
that do not include any adult non-smokers
are at greater risk of exposure to ETS than
those who live with at least one adult nonsmoker. These findings also suggest that
the adoption of a smoking ban is viewed
as acceptable for younger children but
less so for older children, which could
have important operational implications
for public health programs. The strong
association between the presence of adult
non-smokers and the adoption of household
smoking bans also raises health promotion
implications and suggests that intervention
campaigns aimed at non-smokers who live
with smokers should incite the former to
be less tolerant of environmental tobacco
smoke in the home.
References
1.
Centre National de Documentation sur le
Tabac et la Santé. Faits saillants – la fumée
de tabac dans l’environnement : ses consé­
quences générales pour la santé. Ottawa
ON: Canadian Council for Tobacco Control;
1996 Mar.
2.
Glantz SA, Parmley WW. Passive smoking
and heart disease. Mechanisms and risk.
JAMA. 1995 Apr 5;273(13):1047–53.
3.
Sandler DP, Wilcox AJ, Everson RB.
Cumulative effects of lifetime passive
smoking on cancer risk. Lancet. 1985 Feb 9;
1(8424):312–5.
4.
5.
6.
Pizacani BA, Martin DP, Stark MJ, et
al. Household smoking bans: which
households have them and do they work?
Prev Med. 2003;Jan;36(1):99–107.
Ashley MJ, Cohen J, Ferrence R, et al.
Smoking in the home: changing attitudes
and current practices. Am J Public Health.
1998 May;88(5):797–800.
EKOS Research Associates Inc. An
assessment of knowledge, attitudes and
practices concerning environmental tobacco
smoke. Final report. Ottawa, Ontario.
1995 Mar 31. Submitted to Health Canada.
Vol 29, No 3, 2009
7.
Ashley MJ, Ferrence R. Environmental
tobacco smoke (ETS) in home environments:
a discussion paper prepared for Health
Canada’s Strategic Planning Workshop to
Reduce ETS. Ottawa, Ontario: The Ontario
Tobacco Research Unit; 1995 October 19-20.
8.
Health Canada. Canadian tobacco use mon­
itoring survey (CTUMS): section on house­
holds. Ottawa (ON): Health Canada; 2006.
Available from: http://www.hc-sc.gc.ca/
hl-vs/tobac-tabac/research-recherche/stat/
_ctums-esutc_2006/ann-table9-eng.php
9.
Northrup DA. Environmental tobacco
smoke in the home: a national survey
technical documentation. Toronto: Institute
for Social Research, York University; 2002.
16. Dunkelberg WC, Day GS. Non response
bias and call backs in sample surveys.
J Mark Res. 1973;10:160–8.
17. Dillman DA. Mail and internet surveys: the
tailored design method. New York: John
Wiley & Sons; 2000. 464 p.
10. Borland R, Mullins R, Trotter L, et al.
Trends in environmental tobacco smoke
restrictions in the home in Victoria,
Australia. Tob Control. 1999;8(3):266–71.
11. Farkas AJ, Gilpin EA, White MM, et al.
Association between household and work­
place smoking restrictions and adolescent
smoking. JAMA. 2000;284(6):717–22.
12. Wakefield MA, Chaloupka FJ, Kaufman NJ,
et al. Effect of restrictions on smoking at
home, at school, and in public places on
teenage smoking: cross sectional study.
BMJ. 2000;321(7257):333–7.
13. Gilpin EA, White MM, Farkas AJ, et al.
Home smoking restrictions: which smokers
have them and how they are associated
with smoking behavior. Nicotine Tob Res.
1999;1(2):153–62.
14. Merom D, Rissel C. Factors associated with
smoke-free homes in NSW: results from the
1998 NSW Health Survey. Aust N Z J Public
Health. 2001;25(4):339–45.
15. Mak YW, Loke AY, Abdullah AS, Lam TH.
Household smoking practices of parents
with young children, and predictors of
poor household smoking practices. Public
Health. 2008 Nov;122(11):1199-209. Epub
2008 Jul 10.
135
Chronic Diseases in Canada
Workshop/conference report
Myalgic Encephalomyelitis/Chronic Fatigue Syndrome
(ME/CFS) program and Interdisciplinary Research Symposium
on Disabling Fatigue in Chronic Illness
E. Stein, MD (1); M. MacQuarrie, BSc MRP LLB (2)
Background
This program at the University of Calgary
was the first comprehensive program on
myalgic encephalomyelitis/chronic fatigue
syndrome (ME/CFS) held in Canada. There
were three parts: a continuing medical
education program for physicians, a
research symposium on fatiguing illnesses
and a public lecture. ME/CFS is one cause
of disabling fatigue. ME/CFS alone is
a serious medical condition that affects
between 150 0001 and 340 0002 Canadians.
The core symptoms have been described
by Fukuda et al.3 and were further refined
in the Canadian Consensus Guidelines for
the Diagnosis and Treatment of ME/CFS.4
Program
The physician program was divided into two
segments: Part I (October 24, 2008) covered
the diagnosis and assessment of patients
with ME/CFS and Part II (November 7,
2008) covered clinical management.
A full-day interdisciplinary research sym­
posium on disabling fatigue in chronic
illness took place on November 8, 2008, in
a mix of plenary and concurrent sessions.
This symposium was organized to assist
in both the development of a collaborative
research agenda to understand disabling
fatigue in chronic disease as well as in the
knowledge transfer process among health
care professionals, researchers and the
Canadian public. Disabling fatigue exists
in numerous chronic conditions, including
autoimmune disorders, chronic infection,
chronic liver disease, pulmonary and heart
disease, ME/CFS, overtraining and stress
syndromes, some psychiatric conditions
and fatigue due to unknown causes.
The plenary sessions included Karin Olson,
RN, PhD, University of Alberta (Conceptual
Model of Fatigue: The Edmonton Fatigue
Framework); Leonard Jason, PhD, DePaul
University (Epidemiology and Case
Definition of ME/CFS); Nancy Klimas, MD,
University of Miami (Biological Markers
in Disabling Fatigue); Bryan Kolb, PhD,
University of Lethbridge (Neuroplasticity
and Implications for Disabling Fatigue);
and a video of Gerry Thomas, with a
patient perspective.
Dr. Olson presented the Edmonton Fatigue
Framework (EFF), a proposed etiological
model of fatigue based on 15 years of
research with five populations experien­
cing fatigue: those with cancer, depression
and ME/CFS, as well as shift-workers and
recreational runners. In this model,
fatigue (subtyped as tiredness, fatigue and
exhaustion) is considered a behavioural
marker of the inability to adapt to stress
and is secondary to changes in muscle
endurance, sleep quality, cognitive func­
tion, dietary intake and other factors.5
Dr. Klimas, an internationally renowned
research immunologist and clinician, pro­
filed current work, including dynamic
modelling using an exercise stressor model,
showing that one must stress a subject
with ME/CFS to get reliable differences
versus controls. Mathematical modelling
by Gordon Broderick, PhD (Computational
Biology, University of Alberta and one of
Dr. Klimas’ team) is identifying which
biomarkers could serve as a diagnostic test
for ME/CFS.6
Dr. Jason spoke about the definition,
prevalence and social impact of ME/
CFS and, in the concurrent session,
about a four-arm, non-pharmacological
intervention study in ME/CFS. Both of his
presentations underscored the importance
of properly identifying and subtyping ME/
CFS patients. Not every ME/CFS patient
reacts to treatment in the same way.
Dr. Bryan Kolb (Director of the Canadian
Centre for Behavioural Neuroscience at
the University of Lethbridge and author
of the classical textbook, Fundamentals of
Human Neuropsychology7) reviewed
the literature on brain plasticity and
implications for ME/CFS. He made the
Author References
1 Private practice, Calgary, AB, Canada
2 Myalgic Encephalomyelitis Association of Ontario, Toronto, ON, Canada and member of the National ME/FM Action Network
Submitted by Ellie Stein, MD, (Chair, Program Committee) on behalf of the Planning Committee: Drs. Terrie Brandon (Family Medicine, Calgary); Brian MacIntosh
(Kinesiology, University of Calgary); Karin Olson (Nursing, University of Alberta); Steve Simpson (Medicine, University of Calgary); Elaine Stapon (Family Medicine, Calgary);
and Ms. Glenda Wong (Department of CME, University of Calgary).
Correspondence: Ellie Stein, MD, 4523-16A St. SW Calgary, AB, T2T 4L8, Tel.: 403-287-9941, Fax: 403-287-9958, Email: [email protected]
Chronic Diseases in Canada
136
Vol 29, No 3, 2009
provocative hypothesis that the increased
prefrontal volume post “effective cognitive
behavioural therapy” reported by de Lange
et al.8 may have been due to the impact of
the therapy on depression rather than ME/
CFS. Many symptoms, including depression
and stress, correlate with structural changes
in the prefrontal cortex.
• Neil Skjodt, MD (Medical Director,
Edmonton Sleep Institute; Director of
Research, Canadian Sleep Institute),
noted that the sleep irregularities in
ME/CFS were, for various reasons,
not getting specialized attention. He
suggested a more appropriate sleep
assessment protocol for these patients.
The concurrent sessions were also filled with
some very stimulating findings, including:
Other thought-provoking and informative
presentations were given by:
• Bruce Dick, PhD (Departments of
Anesthesiology and Pain Medicine
and Psychiatry, University of Alberta),
presented work on cognitive function
in fibromyalgia.9 He reported that the
spatial span test was the most difficult
task for people with fibromyalgia and
that cognitive function did not improve
in the short term with pain interven­
tion. However, patients on long-term
opiate treatment did better on cognitive
tests overall than those who were not
on opiates.
• Denise Adams, BSc (PhD candidate,
University of Alberta), on Traditional
Chinese Medicine for the Treatment of
Chronic Fatigue: A Systematic Review;
• Patrick Neary, PhD (Kinesology and
Health Studies, University of Regina),
presented his data on prefrontal cortical
oxygenation, as measured by nearinfrared spectrophotometry during
exercise to exhaustion.10 There was no
difference at rest between the ME/CFS
and control groups, but with exercise,
both the total haemoglobin and oxy­
genated haemoglobin were reduced in
the ME/CFS group. He has shown
in unpublished work that oxygen flow
to the brain is slow to recover when
patients with ME/CFS stand up. This
reinforces the need for provocation
testing in ME/CFS.
• Kathleen Pierson, MD, PhD (Department
of Psychiatry, University of Calgary), on
Measuring Fatigue in Early Psychosis;
• Carey Johnson, MD (private practice,
Calgary), presented his observations
that approximately 50% of patients
with Erlers Danlos (ED) syndrome
have clinical features meeting the
Canadian Consensus Guidelines for the
Diagnosis and Treatment of ME/CFS;
the other 50% have the typical features
of connective tissue disorder, but not
chronic fatigue, sleep disorder, pain
or sensory sensitivity. He has a study
in progress to identify genetic markers
for this subgroup of ED patients.
Vol 29, No 3, 2009
• Brian MacIntosh, PhD (Kinesology,
University of Calgary), on Measuring
Peripheral vs. Central Fatigue;
• Lynn Marshall, MD (Environmental
Health Clinic, Women’s College Hospital,
Toronto), on Functional Impairment
in an Environmental Clinic Sample;
• Steve
Simpson,
MD,
FRCP(C)
(Psychiatry, University of Calgary;
Consulting Psychiatrist, Tom Baker
Cancer Centre), on the Management
of Cancer Fatigue;
• Mark Swain, MD, FRCP(C) (Professor
of Medicine, University of Calgary),
on Disabling Fatigue in Inflammatory
Disorders; and
• Mark van Ness, PhD, Staci Stevens, MSc,
and Kylie Kumasaka, PhD candidate
(Pacific Fatigue Laboratory, University
of the Pacific, California), on Metabolic
Dysfunction in ME/CFS.
Podcasts (i.e. audio and slides only) of
the ME/CFS continuing medical education
program and the public lecture are cur­
rently available free of charge. To view
the podcasts go to: http://podcast.
med.ucalgary.ca/groups/cfs/weblog/. The
research day is not a podcast, although
handouts from many of the presentations
are available on the University of Calgary
Continuing Medical Education site, at
www.cme.calgary.ca.
Conclusion
There were many coinciding concepts
from various presenters in different fields,
specifically, the interrelation of all body
systems and the need to take an inter­
disciplinary approach to understanding
ME/CFS and disabling fatigue in chronic
illness. The opportunity for so many dis­
ciplines to gather and benefit from “crosstalking” and the fertilization of ideas was
immense, and new ideas and collaborations
are ongoing. The objective of the symposium
to develop a collaborative research agenda
was ambitious. Meeting this objective will
take time, effort and funding.
Acknowledgements
This program was co-endorsed by the
University of Calgary, Faculty of Medicine,
and the University of Alberta, Faculty of
Medicine and Dentistry. It was supported
by Alberta Health and Wellness, the Public
Health Agency of Canada and the Canadian
Institutes of Health Research. The ME-FM
Society of Alberta was a major sponsor.
Other sponsors (in alphabetical order) were
Bayda Ludwar Law Firm, Cambrian Drug
Mart, Ferring Pharmaceuticals, Genuine
Health, Metagenics, Pfizer Canada, Script
Pharmacy and Varsity Natural Health
Center. The podcasts are available thanks
to sponsorship from the Alison Hunter
Memorial Foundation (www.ahmf.org).
The public lecture on November 9, 2008
featured Nancy Klimas, MD, giving an
update on ME/CFS research, Alison Bested,
MD, FRCP(C) (private practice, Toronto),
providing clinical pearls11 and a tag team of
Stevens, van Ness and Kumasaka presen­
ting on Exercise Tolerance in ME/CFS.
137
Chronic Diseases in Canada
References
1.
Jason LA, Richman JA, Rademaker F, et
al. A community-based study of chronic
fatigue syndrome. Arch Intern Med. 1999;
159(18):2129–37.
2.
Park J, Knudson S. Medically unexplained
physical symptoms. Health Rep. 2007;
18:45–9.
3.
Fukuda K, Straus SE, Hickie I, Sharpe MC,
Dobbins JG, Komaroff A. The chronic fatigue
syndrome: a comprehensive approach to
its definition and study. Ann Intern Med.
1994;121:953–9.
4.
Carruthers BM, Jain AK, DeMeirleir K et al.
Myalgic Encephalomyelitis/Chronic Fatigue
Syndrome: clinical working case definition,
diagnostic and treatment protocols – a
consensus document. J Chronic Fatigue
Syndr. 2003;11(1):7–115.
5.
Olson K, Turner AR, Courneya KS, et al.
Possible links between behavioural and
psychological indices of tiredness, fatigue,
and exhaustion in advanced cancer. Support
Care Cancer. 2008;16(3):241–9.
6.
Broderick G, Craddock RC, Whistler T,
Taylor R, Klimas N, Unger ER. Identifying
illness parameters in fatiguing syndromes
using classical projection methods.
Pharmacogenomics. 2006;7(3):407–19.
7.
Kolb B, Whishaw IQ. Fundamentals of
human neuropsychology. 6th ed. New York:
Freeman-Worth; 2008. 763 p.
8.
de Lange FP, Koers A, Kalkman JS, et al.
Increase in prefrontal cortical volume
following cognitive behavioural therapy
in patients with chronic fatigue syndrome.
Brain. 2008;131:2172–80.
9.
Dick BD, Verrier MJ, Harker KT, Rashiq S.
Disruption of cognitive function in
Fibromyalgia Syndrome. Pain. 2008;
139(3):610–6.
Chronic Diseases in Canada
10. Neary JP, Roberts AD, Leavins N,
Harrison MF, Croll JC, Sexsmith JR.
Prefrontal cortex oxygenation during
incremental exercise in chronic fatigue
syndrome. Clin Physiol Funct Imaging.
2008;28(6):364–72.
11. Bested, AC, Logan AC. Hope and help for
chronic fatigue syndrome and fibromyalgia.
2nd ed. Nashville (TN): Cumberland
House; 2008. 267 p.
138
Vol 29, No 3, 2009
Book review
Dissonant disabilities: women with chronic illnesses
explore their lives
M. Rezai, DC, PhD (Student), University of Toronto
Editors: Diane Driedger and Michelle Owen
Published: Canadian Scholars’ Press Inc./Women’s Press: April, 2008: Toronto
Format: Paperback; 258 pages
ISBN: 978-0-88961-464-2
As the title implies, this compelling col­
lection of essays examines the discordant
lives of women living with chronic illnesses.
The content is meaningful, not only to
those afflicted with similar conditions,
but also to a wide audience—including
physicians, researchers, policy-makers and
the general public—who will benefit from
both the scholarship of this presentation
and the unique perspectives that detail
each enlightening story.
This anthology by women with chronic
illnesses provides a forum for the discussion of
shared barriers and is the first anthology
of its kind in Canada. The distinguishing
feature of this book is the first-hand
accounts of illnesses shared by those directly
experiencing the disease versus accounts of
the disease process or self-help treatments
based on a medical model. While the
authors had every intention to represent a
variety of women, they acknowledge the
level of privilege with respect to time and
ability required to submit an essay and the
resulting disproportionate representation
of women from academia and women with
post-secondary educations.
The authors’ selection of articles poignantly
identifies issues significant to the lives
of women living with a range of chronic
illnesses, including physical, cognitive,
visible, invisible and contested illnesses.
The authors chose to restrict this anthology
to women to highlight the increased
prevalence of chronic illnesses among
women and to raise awareness of prevailing
risk factors, including psychosocial and
socioeconomic determinants. Each essay
Chronic Diseases in Canada
shares the personal story of a woman
with a particular illness, her challenges
and accomplishments with the illness
itself and her environment, including the
institutional policies that affect her home
and working life.
The definition of disability used through­
out the book is based on the social model
where disability is viewed as the inability
of society to account for those with impair­
ments, thereby excluding them from main­
stream society. The social model sees
the lack of a ramp as the problem and
not the use of a wheelchair. This form of
discrimination and social oppression is
paralleled to racism or sexism. The authors
further discuss historical perspectives of
disability that saw a person with a disability
as being in a constant state of sickness,
lacking independence and wanting to get
well. These attitudes linger today and come
to surface as a common theme uniting the
stories in the book shared by the women
from various cultural and socioeconomic
backgrounds. For the reader with a chronic
illness, these common struggles are easily
identified with and when their outcomes
are positive, they serve as a source of
inspiration; when negative, they arouse
empathy. For the health policy-maker,
these stories should inspire change.
The authors’ portrayal of the lives of
women with chronic illnesses was not
meant as an in-depth description of the
epidemiology of each condition, but
a presentation of the facts of daily life
for each woman, supported by current
research and legislative evidence. To
140
balance the discussion of barriers faced by
women with chronic illnesses, the authors
saw fit to include essays demonstrating the
strong will and resistance these women
have to existing social ideas.
The book is divided into five parts. In each
section, different women with chronic
illnesses share key concepts that form the
barriers they face and show their resistance
to prevailing social norms. In Part One:
“Clashing Expectations,” the focus is on
societal attitudes towards women with
chronic illnesses and the isolating feelings
of shame, doubt and powerlessness evoked
when expectations of continual production
are unmet. In this section, the authors raise
awareness of the “changing landscape” of
health experienced by women with chronic
illnesses and society’s lack of acceptance
of this fluctuation in functioning and, to
a greater extent, society’s expectation of
“soldiering on” despite illness.
In Part Two: “Unpredictable Bodies,” the
focus is on idealizations of the female
body and the impact of chronic illness.
To emphasize the far-reaching and global
nature of women’s struggles with chronic
illnesses, including body dissatisfaction,
the authors wisely included essays dis­
cussing the effects of Western society’s
preoccupation with weight on the cultural
expectations of Asian nations. An essay
focusing on women with chronic fatigue
syndrome, fibromyalgia and multiple
chemical sensitivities recognizes the
duality of experiencing a chronic illness:
the knowledge that you are the same
person while you become a different
Vol 29, No 2, 2009
person. When dealing with a contested
illness the situation only intensifies as the
medical community or employers fail to
validate the limitations of these women.
The authors recognize the importance
of association with others who have the
same illness in order to overcome such
ambiguities and gain comfort in the shared
experiences of others.
In Part Three: “Disturbing Work,” the
authors explore how women with chronic
illnesses both disturb work and find work
disturbing. The authors chose an essay
describing the life of a driven researcher
whose ambitions eventually led to a chronic
state of anxiety. For anyone associated with
academia, this story draws many parallels
and helps to identify the early warning signs
of mental and physical exhaustion and the
steps to take to prioritize one’s health and
wellness. In contrast, the next essay in this
section describes the challenges faced by a
young woman with a chronic illness seeking
a graduate degree. This story highlights
the inflexibility of institutional policies
to recognize the uncertainty germane to
chronic illness.
In Part Four: “Shifting Relationships,” the
many relationships women with chronic
illnesses must develop and negotiate are
examined. The authors highlight the impact
of chronic illness not only on the person
directly affected, but also on how those
who provide financial, emotional or physical
support are challenged and adapt. Finally,
in Part Five: “Traversing Dissonance,”
the authors inspire readers with uplifting
stories of how women with chronic
illnesses deal with the often contradictory
barriers to societal participation and how
some barriers are transformed into new
opportunities for growth, such as gaining
a sense of control, experiencing new
challenges and setting new goals. The
book concludes by leaving the reader with
a philosophical dilemma: “Can a society
that is ideologically (if not economically)
committed to preventing, avoiding or
ending most forms of involuntary suffering
appreciate people who are suffering?”
This highlights the common theme of the
book, namely, the need to reform society’s
Vol 29, No 2, 2009
structural (environmental) and concep­
tual acceptance of women with chronic
illnesses.
The authors’ purpose for compiling this
series of stories told by women with chronic
illnesses was to portray the many different
ways a disability may infiltrate the lives
of those affected. Each personal narrative
gives a glimpse into the life of a different
woman—young or old, early or late into her
career. The reader grows to appreciate how
chronic illness affects her daily routine,
her life ambitions and all those around her.
What sets this book apart is its ability to
relay the limitations faced by women with
chronic illnesses while simultaneously
demonstrating their strength and resilience,
combining strong feminist ideals with
critical disability theories.
This unique perspective serves many who
interact with women with chronic ill­
nesses. For the physician, it emphasizes
the needs of a potential patient – needs
that may reach beyond medication, such
as life- or stress-management, or exercise
and nutritional counselling. For the
policy-maker, it introduces the concept
of uncertainty in one’s daily physical and
mental functioning, for which there should
be some flexibility in place. Researchers
and epidemiologists will find the contents
of this book useful when developing con­
ceptual models, for example, examining
the role of psychosocial barriers to recovery
for those with chronic illnesses. Finally, for
the general public, the stories shared in this
book foster an understanding and respect
for the challenges faced by women with
chronic illnesses.
The authors were successful in their
attempt to provoke thought among
their readers. They acknowledge the gaps
in the literature and share their hope that
more work in the area of critical disability
studies will follow.
141
Chronic Diseases in Canada
Announcements
Chronic Disease Update listserv
Conferences
The Public Health Agency of Canada
encourages you to subscribe to the Chronic
Disease Update listserv. We are pleased
to offer you a way to receive information
on the work of the chronic disease team.
Keeping our colleagues and clients informed
about work in progress, new projects and
programs, and opportunities for collabo­
ration is a priority for us. Thank you for
your interest in our work.
7th International Conference on Diet
and Activity Methods
June 5-7, 2009
Washington, D.C.
http://icdam.org/
http://www.phac-aspc.gc.ca/cd-mc/
maillist-eng.php
Canadian Public Health
Association Conference
June 7-10, 2009
Winnipeg, Manitoba
http://www.cpha.ca/en/conferences/
conf2009.aspx
International Scientific Conference on
Nutraceuticals and Functional Foods
June 9-11, 2009
Zilina, Slovakia
http://www.foodandfunction.com/
Chronic Diseases in Canada
142
20th World Diabetes Congress
October 18-22, 2009
Montreal, Quebec
http://www.worlddiabetescongress.org/
Canadian Cardiovascular Congress
October 24-28, 2009
Edmonton, Alberta
http://www.cardiocongress.org/English/
Home_EN.html
Third International Chronic
Disease Conference
November 23-26, 2009
Calgary, Alberta
http://www.cdmcalgary.ca/index.
php?lang=english
Vol 29, No 2, 2009
CDIC: Information for Authors
Chronic Diseases in Canada (CDIC) is a quarterly
Book/Software Review: Usually solicited by the
References: In Vancouver style (consult a recent
scientific journal focussing on the prevention and
editors (500B1,300 words), but requests to review
CDIC issue for examples); num­bered in superscript
control of non‑ communicable diseases and injuries
are welcomed. Abstract not required.
in the order cited in text, tables and figures; listing
in Canada. Its feature articles are peer reviewed. The
up to six authors (first three and et al. if more);
content of articles may include research from such
Submitting Manuscripts
fields as epidemiology, public/community health,
Submit manuscripts to the Managing Editor, Chronic
used
biostatistics, the behavioural sciences, and health
Diseases in Canada, Public Health Agency of Canada,
observations/data or personal communications
services or economics. CDIC endeav­ours to foster
785 Carling Avenue, Address Locator 6805B, Ottawa,
used (discouraged) to be cited in the text in
communication on chronic diseases and injuries
Ontario K1A 0K9, e-mail: [email protected]
parentheses (authors responsible for obtaining
without any automatic reference numbering feature
among public health practitioners, epidemiolo­gists
in
word
processing;
any
unpublished
written per­mis­sion); authors are responsible for
and researchers, health policy plan­ners and health
Since
educators. Submissions are selected based on
on illustrations not applicable) to the “Uniform
scientific quality, public health relevance, clarity,
Requirements for Manuscripts Submitted to
Tables and Figures: Send vector graphics only.
concise­ness and technical accuracy. Although CDIC
Biomedical
the
Each on a separate page and in electronic file(s)
is a publication of the Public Health Agency of
International Committee of Medical Journal Editors,
separate from the text (not imported into the text
Canada, contributions are welcomed from both
authors should refer to this document for complete
body); as self‑ explanatory and succinct as possible;
the public and pri­vate sectors. Authors retain
details before submitting a manuscript to CDIC (see
not too numerous; numbered in the order that they
responsibility for the contents of their papers, and
<www.icmje.org>.
are mentioned in the text; explanatory material for
opinions expressed are not necessarily those of the
CDIC editorial committee nor of the Public Health
Agency of Canada.
Article Types
CDIC
adheres
Journals”
in
as
general
approved
(section
by
verifying accuracy of references.
tables
Checklist for Submitting
Manuscripts
in
footnotes,
identified
by
lower‑case
superscript letters in alpha­betical order; figures
limited to graphs or flow charts/templates (no
Cover letter: Signed by all authors, stating that all
photographs), with software used specified and
have seen and approved the final manuscript and
titles/footnotes on a separate page.
have met the authorship including a full statement
regarding any prior or duplicate publication or
Number of copies: If submitting by mail, one
submission for publication.
complete copy, including tables and figures; one
figures, references) in the form of original research,
First title page: Concise title; full names of all
copy of the manuscript on diskette. If submitting by
surveillance reports, meta‑analyses or methodological
authors and institutional affiliations; name, postal
e‑mail to cdic‑[email protected]‑aspc.gc.ca, please fax or
papers.
and e‑mail addresses, tele­phone and fax numbers
mail the covering letter to the address on the inside
for corresponding author; separate word counts for
front cover.
Peer‑reviewed Feature Article: Maximum 4,000
words for main text body (excluding abstract, tables,
Status Report: Describe ongoing national programs,
copy of any related supple­men­tary material, and a
abstract and text.
studies or information systems bearing on Canadian
public health (maximum 3,000 words). Abstract not
Second title page: Title only; start page numbering
required.
here as page 1.
Workshop/Conference
Report:
Summa-rize
Abstract:
Unstructured
(one
paragraph,
no
significant, recently held events relat­ing to national
headings), maximum 175 words (100 for short
public health (maximum 1,200 words). Abstract not
reports); include 3B8 key words (preferably from
required.
the Medical Subject Headings (MeSH) of Index
Medicus).
Cross‑Canada Forum: For authors to present or
exchange information and opin­ions on regional or
Text: Double‑spaced, 1 inch (25 mm) margins, 12
national surveillance findings, programs under
point font size.
development or public health policy initiatives
(maximum 3,000 words). Abstract not required.
Acknowledgements: Include disclosure of financial
and material support in acknowledgements; if
Letter to the Editor: Comments on articles recently
anyone is credited in acknowledgements with
published in CDIC will be consid­ered for publication
substantive scientific contributions, authors should
(maximum 500 words). Abstract not required.
state in cover letter that they have obtained written
permission.
Vol 29, No 2, 2009
143
Chronic Diseases in Canada
Chronic Diseases in Canada
a publication of the Public Health Agency
of Canada
Howard Morrison
Principal Scientific Editor
(613) 941-1286
Robert A Spasoff
Associate Scientific Editor
Claire Infante-Rivard
Associate Scientific Editor
Elizabeth Kristjansson
Associate Scientific Editor
Michelle Tracy
Managing Editor
CDIC Editorial Board
Jacques Brisson
Laval University
Neil E Collishaw
Physicians for a Smoke-Free Canada
James A Hanley
McGill University
Clyde Hertzman
University of British Columbia
C Ineke Neutel
University of Ottawa Institute on
Care of the Elderly
Kathryn Wilkins
Health Statistics Division
Statistics Canada
Chronic Diseases in Canada (CDIC) is a quar­
terly scientific journal focussing on cur­rent
evidence relevant to the control and pre­
vention of chronic (i.e. non-communicable)
diseases and injuries in Canada. Since 1980
the journal has published a unique blend of
peer-reviewed feature articles by authors
from the public and private sectors and
which may include research from such fields
as epidemiology, public/community health,
bio­statistics, the behavioural sciences, and
health services or economics. Only feature
articles are peer reviewed. Authors retain
responsibility for the content of their arti­
cles; the opinions expressed are not neces­
sarily those of the CDIC editorial committee
nor of the Public Health Agency of Canada.
Chronic Diseases in Canada
Public Health Agency of Canada
130 Colonnade Road
Address Locator 6501A
Ottawa, Ontario K1A 0K9
Fax: (613) 941-9502
E-mail: [email protected]
Indexed in Index Medicus/MEDLINE
Vérifier les informations
si elles doivent
être modifiées
To promote and protect the health of Canadians through leadership, partnership, innovation and action in public health.
— Public Health Agency of Canada
Published by authority of the Minister of Health.
© Her Majesty the Queen in Right of Canada, represented by the Minister of Health, 2009
ISSN 0228-8699
This publication is also available online at www.publichealth.gc.ca/cdic
Également disponible en français sous le titre : Maladies chroniques au Canada
Chronic Diseases in Canada
Volume 29 · Number 3 · 2009
Inside this issue
96
Validation of perinatal data in the discharge
Abstract Database of the Canadian Institute for
Health Information
K. S. Joseph, J. Fahey, MMath for the Canadian Perinatal
Surveillance System
101 Validity of autism diagnoses using administrative
health data
L. Dodds; A. Spencer; S. Shea; D. Fell; B. A. Armson, MD A. C. Allen;
S. Bryson
107 Associations between chronic disease, age and
physical and mental health status
W.M. Hopman; M.B. Harrison; H. Coo; E. Friedberg; M. Buchanan;
E.G. VanDenKerkhof
115 Statistical modelling of mental distress among
rural and urban seniors
C.P. Karunanayake; P. Pahwa
125 Factors associated with the adoption of a smoking
ban in Quebec households
É. Ouedraogo; F. Turcotte; M. J. Ashley; J. M. Brewster; R. Ferrence
133 Myalgic Encephalomyelitis/Chronic Fatigue
Syndrome program
E. Stein; M. MacQuarrie
136 Book review – Dissonant, disabilities: Women with
chronic illnesses explore their lives
M. Rezai
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