Article

Discrepancies between survey and administrative data on the use of mental health services in the general population: Findings from a study conducted in Québec

Département de psychiatrie de l'université de Montréal, C,P, 6138 Succ, Centre-Ville, Montréal, H3C 3J7, Canada.
BMC Public Health (Impact Factor: 2.26). 10/2011; 11(1):837. DOI: 10.1186/1471-2458-11-837
Source: PubMed
ABSTRACT
Population surveys and health services registers are the main source of data for the management of public health. Yet, the validity of survey data on the use of mental health services has been questioned repeatedly due to the sensitive nature of mental illness and to the risk of recall bias. The main objectives of this study were to compare data on the use of mental health services from a large scale population survey and a national health services register and to identify the factors associated with the discrepancies observed between these two sources of data.
This study was based on the individual linkage of data from the cycle 1.2 of the Canadian Community Health Survey (CCHS-1.2) and from the health services register of the Régie de l'assurance maladie du Québec (RAMQ). The RAMQ is the governmental agency managing the Quebec national health insurance program. The analyses mostly focused on the 637 Quebecer respondents who were recorded as users of mental health services in the RAMQ and who were self-reported users or non users of these services in the CCHS-1.2.
Roughly 75%, of those recorded as users of mental health services users in the RAMQ's register did not report using mental health services in the CCHS-1.2. The odds of disagreement between survey and administrative data were higher in seniors, individuals with a lower level of education, legal or de facto spouses and mothers of young children. They were lower in individuals with a psychiatric disorder and in frequent and more recent users of mental health services according to the RAMQ's register.
These findings support the hypotheses that social desirability and recall bias are likely to affect the self-reported use of mental health services in a population survey. They stress the need to refine the investigation of mental health services in population surveys and to combine survey and administrative data, whenever possible, to obtain an optimal estimation of the population need for mental health care.

Full-text

Available from: Aline Drapeau, Jan 11, 2016
RESEARCH ARTIC LE Open Access
Discrepancies between survey and administrative
data on the use of mental health services in the
general population: findings from a study
conducted in Québec
Aline Drapeau
1,2,3*
, Richard Boyer
1,2
and Fatoumata Binta Diallo
2,3
Abstract
Background: Population surveys and health services registers are the main source of data for the management of
public health. Yet, the validity of survey data on the use of mental health services has been questioned repeatedly due
to the sensitive nature of mental illness and to the risk of recall bias. The main objectives of this study were to compare
data on the use of mental health services from a large scale population survey and a national health services register
and to identify the factors associated with the discrepancies observed between these two sources of data.
Methods: This study was based on the individual linkage of data from the cycle 1.2 of the Canadian Community
Health Survey (CCHS-1.2) and from the health services register of the Régie de lassurance maladie du Québec
(RAMQ). The RAMQ is the governmental agency managing the Quebec national health insurance program. The
analyses mostly focused on the 637 Quebecer respondents who were recorded as users of mental health services
in the RAMQ and wh o were self-reported users or non users of these services in the CCHS-1.2.
Results: Roughly 75%, of those recorded as users of mental health servi ces users in the RAMQs register did not
report using mental health services in the CCHS-1.2. The odds of disagreement between survey and administrative
data were higher in seniors, individuals with a lower level of education, legal or de facto spouses and mothers of
young children. They were lower in individuals with a psychiatric disorder and in frequent and more recent users
of mental health services according to the RAMQs register.
Conclusions: These findings support the hypotheses that social desirability and recall bias are likely to affect the
self-reported use of mental health services in a population survey. They stress the need to refine the investigation
of mental health services in population surveys and to combine survey and administrative data, whenever possible,
to obtain an optimal estimation of the population need for mental health care.
Background
The combined use of population surveys and health ser-
vices registers is a powerful tool for public health since
their respective limitations and assets can balance each
other. National health services registers are mostly imple-
mented to manage the payment of health services. There-
fore they can only supply data on to the population,
services and health professionals covered by the health
program. Population surveys pro vide data on topics that
are n ot usually documented in health services registers
and that can produce a more detailed description of ser-
vices users. A potential limitationofsurveydataistheir
questionable validity when they involve sensitive issues.
The validity of survey data on the use of mental health
services has been questioned repeatedly due to the sensitive
nature of mental illnes s [1-4]. Indeed, prejudices again st
mental illness are widespread [5-7] and, as a consequence,
people with a mental health problem are highly stigmatized
[5-9]. Indiv iduals with a negative attitude towards mental
illness and those who have been victimized because of
* Correspondence: aline.drapeau@umontreal.ca
Contributed equally
1
Département de psychiatrie de luniversité de Montréal, C.P. 6138 Succ.
Centre-Ville, Montréal, H3C 3J7, Canada
Full list of author information is available at the end of the article
Drapeau et al. BMC Public Health 2011, 11:837
http://www.biomedcentral.com/1471-2458/11/837
© 2011 Drapeau et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the C reative
Commons Attr ibution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cite d.
Page 1
their mental illness or the mental illness of a close relative
may be reluctant to report their symptoms and their use of
mental health services in a population s urvey. However,
although the prejudices against mental illness are wide-
spread, they are not universal. They tend to be more pro-
minent in some segments of the population and against
specific psychiatric disorders. On t he one hand, negative
attitudes towards mental illness and towards people with
mental health problems tend to increase with age [10-12]
and to be more apparent for schizophrenia than for depres-
sion [10,11,13], in me n than in women [10] and among
singles and individuals living with a partner than in
divorced, separated or widows [14]. On the other hand,
they appear to decrease with education [12,14,15] and
income [14,16] and to be uncommon in those exposed to
mental illness as carers or because of a personal or family
history of mental illness [10,12,17]. Findings regarding the
influence of gende r are conflicting; in some studies,
women express more negative attitudes than men [11,15]
whereas, in other studies, they appear more open-minded
than men towards mental illness and toward the mentally
ills [10,16,18]. Finally, no statistically significant association
has been found between employment status and negative
attitudes towards mental illness [12,16] despite the perva-
sive stigmatisation of people with mental health problems
in the work environment [19,20].
In addition to the effect of social desirability, survey
data on the use of health services may also be flawed by
recall bias. In population surveys, the l ength of the time
window used for the investigation of mental health ser-
vices is traditionally the 12 months preceding the survey
interview. Most individuals would find that recalling their
use of health services in the past year is quite a challenge.
Those with a declin ing memory, such as s eniors [21] and
individuals who have symptoms or who take medication
that affect their memory [9,22] may be especially at risk
of ina ccurate recall. Bhanda ri and Wagner [21] carried
out a meta-analysis of 42 studies based on the linkage of
individual data from population surveys or patients-based
studies and administrative registers. Only one of these
focused on mental health services in the general popula-
tion. Bhandari and Wagner [21] found that the probabil-
ity of under-reporting health services was sizeabl e, that it
increased with the length of the time window used to
document these services (1 to 4 months: 26.0%; 6
months: 39.6%; 12 months: 50.3%) and that it was higher
in seniors. They also noted that the reporting of health
services was more a ccurate for salient services (e.g., hos-
pitalisation vs. outpatient services) [21].
Although the comparison of survey and administrative
data is the best available strategy to validate survey data
on the use of mental health services, it is not perfect
since it fails to recognise that consumers, services
providers and decision-makers may not always agree on
the definition of mental health services [1-3]. Tradition-
ally, respondents are asked whether or not they have con-
sulted a health professional in the past year for a mental
health problem and, if they did, what category of health
professional they have consulted. Thus, thei r answer
depends, not only on th eir w illingness or capacity to
report their use of mental health services but also on
their perception of their motive for their past consulta-
tion and on their definition of a mental health problem
and of a mental health service. Services providers inter-
pret the symptoms disclosed by their patients and classify
them into one or more standardized diagnostic s that may
or may not fit with the patients motives for the consulta-
tion. For example, a diabetic patient might consult a gen-
eral practitioner (GP) on a regular basis. According to
the patient, the motive for these consultations is diabetes.
During a c onsultation, the patient mentions that he or
she feels tired and nervous and does not sleep well. Upon
further i nquiry, the GP detects the first signs of depres-
sion but decide to not prescribe a n anti-depressa nt for
the moment. On his claim for payment, the GP records
two diagnostics: diabetes and depression. The recording
of a depression diagnostic is in line with the rules defined
by decision-makers for the payment of health services
since the patients psychological symptoms w ere exam-
ined. Yet, the patient did not consult this health profes-
sional for a mental health problem and he or sh e was not
treated by medication or therapy for a mental health pro-
blem. In all likelihood, this patient would not r eport this
consultation as a mental health service in a popula tion
survey. This example is typical of occasional differences
in the definition of mental health services by respondents
and services providers that would generate a random var-
iation in the discrepanc y between survey and administra-
tive data. In a publi c health persp ective, the main issue is
not so much random variation but rather systematic var-
iations acro ss specific segments of t he population (e.g .,
those who are highly prejudiced against mental illness or
who have a failing memory) since systematic variations
raise some doubt regarding the validity of survey data on
the use of mental health service.
The main objective of this study was to investigate the
validity of survey data on the use of mental health ser-
vices. More precisely, this study aimed to compare the
use of mental health s ervices recorded in a large scale
population survey (cycle 1.2 of the Cana dian Community
Health Services Survey (CCHS-1.2)) and in a national
health service s re gister (register of the Régie de l assur-
ance mala die du Québec (RAMQ)). Gi ven t he hypothe-
sised effect of social desirability and recall bias on the
self-reported use of mental health services in surveys, sys-
tematic variations in the discrepancies between the
Drapeau et al. BMC Public Health 2011, 11:837
http://www.biomedcentral.com/1471-2458/11/837
Page 2 of 10
Page 2
CCHS-1.2 and the RAMQ were expected. Thus, a second
objective was to identify the factors associated with the
discrepancies vis-à-vis the use of mental health services
between survey and administrative data. The select ion of
the factors under study was based, fir st, on the empirical
evidence concerning factors associated with the preju-
dices against mental illness, the stigmatization of t he
mentally ills, and the under-reporting of mental health
services in surveys compared to administrative registers
and, second, on the assumption that a recent, salient or
frequent use of mental health services was more likely to
be recalled.
Methods
Sample
This study is based on the individual linkage of data
from the CCHS-1.2 and from the RAMQs health ser-
vices register. The CCHS is a repeated cross-sectional
population survey carried out by Statistics Canada.
Cycle 1.2 of the CCHS was conducted in 2001-2002 and
was designed to document the mental health of the
population, the use of formal and informal mental
health services and a number of factors traditionally
associated with mental illness and with the use of men-
tal health services. The target population was made up
of individuals aged 15 ye ars and more and living in pri-
vate households. Respondents were randomly selected
using a multistage stratified cluste r sampling design.
Additional information on the CCHS-1.2 can be found
in Gravel et al. [23]. The RAMQ is the governmental
agency managing the Quebec national health in surance
program, which covers medical services provided by
physicians (i.e., general pract itione rs an d spe cialis ts) to
resident Quebecers. The RAMQsregisterisfedbythe
claims for payment of these physicians and the accuracy
of RAMQ data is routinely ascertained to ensure t hat
these claims are just ified. The RAMQ register contains
data on the patients (e.g., date of birth; sex; residence),
the servic es (e.g. , type of servic es; loca tion; diagn osis)
and the physicians (e.g., specialty) covered by the
program.
This study is restricted to Quebecer respondents aged 18
and more who agreed that the data that they had provided
to the CCHS-1.2 be shared with governmental agencies for
research purposes. The rate of agreement was high (n =
4796/5047: 95%) (Figure 1). Refusals were slightly higher in
non employed (6.5%) than in employed (4.3%), in those
without(5.6%)thanwith(4.1%)aspouse(legalordefacto)
and in those with children in the household (6.0%) th an
without (4.4%) [24]. They did not differ by age, gender or
the presence of a psychiatric disorder. Responde nts aged
15 to 17 were excluded from this study because some of
the questions related to the use of mental health services
were not addressed to this segment of the population.
Linkage procedure
LinkageofthedatafromtheCCHS-1.2andtheRAMQ
was carried out by the Institut de la statistique du Qué-
bec. It wa s based on the name, health insurance numb er,
date of birth, sex of res ponde nts and po stal code of their
residence. Most respondents (i.e., 4234/4796; 88 %) w ere
successfully linked with their RAMQsrecord(Figure1).
Thosewhowerenotlinkedweremorelikelytobeolder
than 45 years of age and to have a negative perception of
their health. [24].
Use of mental health services in the CCHS-1.2 and the
RAMQs register
In the CCHS-1.2, questions pertaining to the use of mental
health services were addressed to all adult r espondents
whether or not they had experienced mental health pro-
blems in their lifetime or in the year preceding the survey
interview. Respondents who reported talking to a general
practitioner, a psychiatrist or another medical specialist for
problems related to their emotions, their mental health or
their drugs or alcohol intake in the twelve months preced-
ing their interview in the CCHS-1.2 or following suicidal
ideation or suicide attempt over the same period were
coded as self-reported users of mental health s ervices.
Recorded use of mental health services in the RAMQsreg-
ister was based on three variables: medical specialty of the
physician; type of medical act; and diagnosis. Individuals
satisfying at least one of the following criteria in the twelve
months preceding their inte rview in t he CCHS-1.2 were
coded as recorded users of mental health services: (1)
health service provided by a psychiatrist; (2) psychiatric act
(codes 360 to 394 of the RAMQ; these codes refers to psy-
chiatric consultations, complete psychiatric examinations
and psychiatric treatm ents in outpatient clinics or hospi-
tals); (3) psychiatric diagnosis (codes 290 to 319 and 2900
to 3199 of the International Classification of Disease - 9th
version (ICD-9)). The interval between each mental health
service recorded in the RAMQ sregisterandthetwelve
month preceding the interview in the CCHS-1.2 was calcu-
lated individually for each respondent to ensure that the
period of reference for the self-reported and recorded use
was similar.
Self-reported use in the CCHS-1.2 and recorded use in
the RAMQ s register were combined to produce four
categories of respondents. Most respondents (82.8%)
were non-users according to the CCHS-1.2 and the
RAMQs register; 2.1% were self-reported users but non
users according to the RAMQs register; 3.7% were self-
reported and recorded users; and 11.4% were self-
reported non users but recorded users (Figure 1). The
analyses mos tly focus on the la st two categories (i.e. 637
respondents who were recorded as users of mental
health services in th e RAMQ sregisterandwhowere
self-reported users or non users).
Drapeau et al. BMC Public Health 2011, 11:837
http://www.biomedcentral.com/1471-2458/11/837
Page 3 of 10
Page 3
Description of independent variables
Data on age, education (less than high school diploma;
high school diploma; post-high school diploma; univer-
sity diploma), employment status (with or without part-
time of full-time employment), gender, marital status
(with or without a legal or de facto spouse), parent al sta-
tus (with or without child ren aged 1 2 years or less in the
household) were extracted from the CCHS-1.2 file.
Data on the psychiatric profile of respondents were
provided by the CCHS-1.2 and the RAMQsregister.In
the CCHS-1.2, psychiatric symptoms were elicited with
the Composite International Diagnosis Interview (CIDI)
for mo od disorders (i.e., major depression and mania),
and anxiety disorders (i.e., agoraphobia, panic attack,
social phobia). The CIDI is based on the DSM-IV criteria.
Three dichotomous variables were constructed from
these data: presence/absence of self-reported symptoms
compatible with a diagnosis of mood disorder; presence/
absence of self-re ported symptoms compatible with a
diagnosis of anxiety disorder; neither mood nor anxiety
disorder. In the RAMQ, diagnoses are specified by physi-
cians on their claim for payment and are classified
according to the ICD-9. Three dichotomous variables
were created from these data: presence/absence of neuro-
tic disorder; presence/absence of psychotic disorder; pre-
sence/absence of any other psychiatric diagnosis.
Recorded use of mental health services in the RAMQs
register was described by four variables: the number of
mental health services provided in the year preceding the
CCHS-1.2; the maximum number of criteria (i.e., service
provided by a psychiatrist; psychiatric act; psychiatric
diagnosis) satisfied in a single medical visit; the medical
specialty of the physician (i.e., general practitioner; psy-
chiatrist; other specialist); and the type of psychiatric act
(i.e., consultation; complete exam; psychiatric treatment).
In addition, the interval between the CCHS-1.2 interview
CCHS-1.2
Quebecer
respondents
Aged 18 and over
5047 (100%)
Refused linkage
251 (5%)
Accepted linkage
4796 (95%)
Linked
4234 (88
%)
Not linked
562 (12%)
Over-reporters
CCHS Users
RAMQ Non users
90 (2.1%)
Under-reporters
CCHS Non users
RAMQ Users
481 (11.4%)
Users
CCHS and RAMQ
156 (3.7%)
Non users
CCHS and RAMQ
3507 (82.8%)
Figure 1 Flow-chart of study participants.
Drapeau et al. BMC Public Health 2011, 11:837
http://www.biomedcentral.com/1471-2458/11/837
Page 4 of 10
Page 4
and the most recent mental health recorded in the
RAMQs register was computed and split into four cate-
gories (1-3 months; 3-6 months; 6-9 months; 9-12
months).
Statistical analysis
Percentages of agr eemen t and disagreement between the
CCHS-1.2 a nd the RAMQ data and Kappa inter-raters
coefficients between these two sources of data were com-
puted for the whole sample and f or each variable under
study. Logistic regressions were performed to estimate
the odds ratio for disagreeme nt vs. agreement between
the CCHS-1.2 and the RAMQs register for each variable.
Confidence intervals were estimated at the 95% confi-
dence level. All analyses were based o n weighted d ata to
control for non response and for the complex survey
design of the CCHS-1.2. The weights w ere calculated by
Statistics Canada.
Ethical considerations
This study was restricted to respondents who agreed that
the data that they had provided in the CCHS-1.2 be
shared with governmental agencies for research purposes.
Access to the CCHS-1.2 data file was granted by the
Social Sciences and Humanities Research Council of
Canada and Statistics Canada. Linkage of the CCHS-1.2
and the RAMQ files was authorized by the Commission
daccès à linfo rmation du Québec. Data were analysed by
the authors at the secured environment of the Centre
daccès des données de recherche of the Institut de la sta-
tistique du Québec (CADRISQ). This study was approved
by the ethical comm ittee of the Centre de recherche Fer-
nand-Segui n of the Hôpital Louis-H. Lafo ntaine, which is
the main research affiliation of the first author.
Results
The sample was made up of 240 men and 398 women
and the mean age was 49.3 (sd 15.4) years old. The mean
number of recorded mental health services in the year
preceding the survey was 4.1 7.2) and 49.6% of respon-
dents satisfied 2 o r 3 criteria for a mental health service
in a single medical visit. General practitioners were the
most f requent mental health se rvices providers (94.7%)
and psychiatric treatment was the most frequent psychia-
tric act (71.0%). The interval b etween the CCHS-1.2
interview and the most recent mental health service
recorded in the RAMQ s file was relatively short for
66.8% of respondents (1-3 months: 37.1%; 3-6 months:
29.7%). Addit ional descriptive dat a are presented i n
Tables 1 and 2.
The one-year prevalence of the use of mental health
services in adults was higher when estimated from the
RAMQs register (15%) than from the CCHS-1.2 (6%).
The overall agreement between data from the CCHS-1.2
and the RAMQ was low (kappa = 0.29; CI
95%
0.25 to
0.34) for the use of mental health services and it was
lowest in seniors (kappa = 0.05; CI
95%
0.00 to 0.12). In
effect, 75.5% (481/637) of the respondents who, accord-
ing to the RAMQ s register, had received a mental
health service in the 12 months pr eceding their inter-
view in the CCHS-1.2 report that they had not talked to
a physician about their emotions, their mental health or
their alcohol or drug problems during that period.
Several variables were a ssociated with the discre pancy
between the CCHS-1.2 and the RAMQ sregisterdata
(Table 3). For instance, the odds of disagreement
decreased with education level and were higher in seniors
and in respondents living with a partner. The lack of
association with gender was unexpected so interactions
between gender and the other socio-demographic vari-
ables were investigated. A significant i nteraction was
found between gender and p arental s tatus: women and
men who had no children at home did not differ i n their
odds of disagreement whereas a higher percentage of
women (87%) than of men (63%) with young children did
not r eport using mental health servic es although some
mental health services were recorded in the RAMQs reg-
ister. The odds of disagreement were lower in respon-
dents who self-reported symptoms compatible with a
diagnosis of mood or anxiety disorder and in those who
had a neurotic, psychotic or other psychiatric disorder
according to the RAMQs register.
As expected, discrepancies between s elf-reported and
recorded use of me ntal health services were lower in
individuals with a more frequent, more salient or more
recent exposure to mental health services according to
the RAMQs register (Table 3). The odds of disagreement
decreased with the number of recorded mental health
services and with the number of criteria satisfied in a sin-
gle medical visit and it increased with the length of the
interval between the CCHS-1.2 interview and the most
recent recorded mental health service. These odds were
lower for services provided by a psychiatrist and higher
for mental health services provided by general practi-
tioners than for those provided by other physicians. They
were also lower for psychiatric consultation and complete
examination but were not sta tistically significant for psy-
chiatric treatment.
Lastly, the over-reporting of mental health services was
mod erately high wit h 37% of self-reported users of men-
tal health services having no recorded use in the RAMQ
for the year preceding their interview in the CCHS-1.2.
This over-reporting was high in seniors (OR = 4.5 CI
95%
1.7 to 12.1) bu t low in respond ents with self- reported
symptoms compatible with a mood disorder (OR = .32
CI
95%
0.15 to 0.70). Most cases of over-reporting were
attributable to telescoping. Telescoping consists in
reporting the use of m ental health services that took
Drapeau et al. BMC Public Health 2011, 11:837
http://www.biomedcentral.com/1471-2458/11/837
Page 5 of 10
Page 5
place beyond the time window selected for the documen-
tation of these services in the su rvey. Over-reporters
were recorded as users of mental h ealth services in the
RAMQ in the 6 months (83.0% CI
95%
77.8 to 88.2) or 12
months (91.1% CI
95%
87.4 to 93.8) beyond the reference
period of one year used in the CCHS-1.2.
Discussion
Findings from this study show that the level of discre-
pancy bet ween self-reported vs. recorded use of mental
health service s is high and that the odds of disagreement
between the survey an d a dministrat ive d ata under study
are not random. The systematic va riation of the discre-
pancies between survey and administrative data across
specific segments of the population suggests that some
biases, such as social desirability and recall bias, affect the
validity of the self-repo rted use of mental health services
in surveys. This study also shows that this disagreement
is lower in individuals wit h a psychiatric disorder (either
self-reported or according to the RAMQ) and in those
recorded as frequent and more recent use rs of me ntal
health services in the RAMQ register.
The limitations of this study must be kept in mind to
better qualify these findings. First, as stressed previously,
administrative data are not a perfect Gold Standard for the
use of mental health s ervices since patients may not be
aware that a mental health service (as defined by services
providers a nd decision-makers) has been provided. Sec-
ond, this study is restricted to the mental he alth services
and physicians covered by Quebecs national health insur-
ance program. Thus findings from this study may not be
generalized to services provided by other health
Table 1 Socio-demographic and clinical description of the sample
Agreement
CCHS-RAMQ
(n = 157)
a
Disagreement
CCHS-RAMQ
(n = 481 )
a
Total sample
(n = 637)
a
Socio-demographic profile
Age
Mean 42.85 ± 13.24 51.35 ± 15.53 49.26 ± 15.43
18-64 years 150 (95.5%) 370 (76.9%) 520 (81.5%)
65 and over 7 (4.5%) 111 (23.1%) 118 (18.5%)
Gender
Men 50 (31.8%) 190 (39.5%) 240 (37.6%)
Women 107 (68.2%) 291 (60.5%) 398 (62.4%)
Education
No high school diploma 30 (19.4%) 144 (31.0%) 174 (28.1%)
High school diploma 32 (20.6%) 93 (20.0%) 125 (20.2%)
Post-high school diploma 54 (34.8%) 142 (30.5%) 196 (31.6%)
University diploma 39 (25.2%) 86 (18.5%) 125 (20.2%)
Employment status
Workers 103 (66.0%) 299 (62.3%) 402 (63.2%)
Non workers 53 (34.0) 181 (37.7%) 234 (36.8%)
Marital status
With spouse 61 (39.1%) 295 (61.3%) 356 (55.9%)
Without spouse 95 (60.9%) 186 (38.7%) 281 (44.1%)
Parental status
With young children 51 (32.7%) 132 (27.4%) 183 (28.7%)
Without young children 105 (67.3%) 349 (72.6%) 454 (71.3%)
Psychiatric disorder
Based on CCHS-1.2 (DSM-IV)
b
Mood disorder 73 (46.5%) 27 (5.6%) 100 (15.7%)
Anxiety disorder 35 (22.2%) 19 (4.0%) 54 (8.5%)
Neither mood nor anxiety disorder 70 (44.6%) 439 (91.3%) 509 (79.9%)
Based on RAMQ (ICD-9)
a
Neurotic disorder 83 (52.9%) 190 (39.5%) 236 (37.0%)
Psychotic disorder 31 (19.7%) 16 (3.3%) 37 (5.8%)
Other psychiatric disorder 26 (16.6%) 55 (11.4%) 66 (10.4%)
a Un-weighted number of respondents. The other numbers in the Table are weighted.
b The percentages for psychiatric disorders are not additive since an individual can have more than one disorder
Drapeau et al. BMC Public Health 2011, 11:837
http://www.biomedcentral.com/1471-2458/11/837
Page 6 of 10
Page 6
professionals (e.g., psychologists and therapists) and
to informal services (e.g., support groups). Among the 481
respondents who declared that they did not consult
a physician for a mental health problem in the year pre-
ceding their interview in the CCHS-1.2, a small percent
reported using other types of mental health service during
that period (Psychologist: 5.0%; Informal service - e .g.,
Internet;helpline-:4.4%;Other health professional -
Nurse or social worker -: 4.0%; Alternative medicine - e.g.,
Acupuncturist; massotherapist -: 0.4%). Third, in principle,
the apparent under-reporting of mental health services in
the CCHS-1.2 compared to the RAMQs register could be
partly attributable to the respondents burden bias. Indeed
to decrease the bu rden of respondents, it is customary to
administer a series of questions on a specific topic only to
respondents who answered positively to a filter question.
When this strategy is used repeatedly, respondents learn
to recognize the procedure and tend to answer negatively
to a s uspected filter question to avoid the burden of
answering the ensuing series of question. In the CCHS-
1.2, questions on the use of mental health services
followed the quest ions on a number of topics (e.g.,
psychiatric symptoms for mood and anxiety disorders)
that were introduced by a filter question using the same
format (i.e., in your lifetime, have you ever...) as that used
for the investigation of mental health services. Some
respondents are likely to have recogni sed the filter q ues-
tion and to have answered neg atively to avoid the load of
questions on the use of mental health services. The influ-
ence of respondents burden bias could not be investigated
with the data at hand since the same questionnaire format
was used for all respondents. However, the higher odds of
disagreement between the CCHS-1.2 and the RAMQ in
seniors, in less educated people, in individuals with a
spouse and in women with young children go against the
hypothesis of respondents burden bias unless one assumes
that these segments of the population are more likely to
have recognized the filter question for the use of mental
health services. Despite its limitations, this study has
strengths that must also be taken into account in the
appreciation of findi ngs. In parti cular, the CCHS-1.2 i s a
population survey conducted by Statistics Canada instead
Table 2 Characteristics of the use of mental health services based on the RAMQ
Agreement
CCHS-RAMQ
(n = 157 )
a
Disagreement
CCHS-RAMQ
(n = 481 )
a
Total sample
(n = 637 )
a
Number of mental health services
Mean 8.61 ± 11.73 2.67 ± 3.92 4.12 ± 7.19
1 32 (20.4%) 215 (44.7%) 247 (38.7%)
2 21 (13.4%) 117 (24.3%) 138 (21.6%)
3 and more 104 (66.2%) 149 (31.0%) 253 (39.7%)
Number of criteria satisfied
1 44 (28.2%) 277 (57.6%) 321 (50.4%)
2 75 (48.1%) 184 (38.3%) 259 (40.7%)
3 37 (23.7%) 20 (4.2%) 57 (8.9%)
Type of criterion satisfied
b
Specialty: psychiatry 40 (25.5%) 23 (4.8%) 63 (9.9%)
Psychiatric act 129 (82.2%) 352 (73.2%) 481 (75.5%)
Psychiatric diagnostic 139 (85.51%) 351 (73.0%) 490 (76.9%)
Medical specialty of physician
b
General practitioner 137 (87.3%) 466 (96.9%) 603 (94.7%)
Psychiatrist 40 (25.5%) 23 (4.8%) 63 (9.9%)
Type of psychiatric act
b
Consultation 23 (14.6%) 17 (3.5%) 40 (6.3%)
Complete exam 48 (30.8%) 36 (7.5%) 84 (13.2%)
Psychiatric treatment 119 (75.8%) 334 (69.4%) 453 (71.0%)
Interval between most recent mental health service and survey interview
1-3 months 92 (59.4%) 144 (29.9%) 236 (37.1%)
3-6 months 39 (25.2%) 150 (31.2%) 189 (29.7%)
6-9 months 13 (8.4%) 100 (20.8%) 113 (17.8%)
9-12 months 11 (7.1%) 87 (18.1%) 98 (15.4%)
a Un-weighted number of respondents. The other numbers in the Table are weighted.
b The percentages for type of criteria, medical specialty of physician and type of psychiatric act are not additive since an individua l can h ave satisfied different
criteria, meet different physicians and had different act in different medical visits.
Drapeau et al. BMC Public Health 2011, 11:837
http://www.biomedcentral.com/1471-2458/11/837
Page 7 of 10
Page 7
of a study based on patients and the sample under study is
sizeable. In addition, selection bias is unlikely to have
affected the results since the refusal rate for the individual
linkage of survey and administrative data for research pur-
poses was low and the rate of successful linkage was high.
Some authors have hypothesised that individu als with a
psychiatric disorder are more likely to conceal their use
of mental health services in a survey interview for fear of
the prejudices against mental illness or to inaccurately
report their use of these services because they may take
medication or have symptoms that affect their memory
[3,9]. In contradiction with this hypothesis, a number of
studies have shown that people with a personal experi-
ence of psychopathology are more open-minded about
mental illness [10,12,14,17]. In this study, the odds of dis-
agreement were indeed lowe r in respondents with a self-
reported or recorded psychiatric disorder than in those
without. In addition, the odds of disagreement between
the CCHS-1.2 and the RAMQsregisterwaslowerfor
psychotic disorder than for other psychiatric disorders
although the empirical evidence show that t he prejudices
and stigmatisation tend to be higher for schizophrenia
[10,11,13]. In agree ment w ith the studies carrie d out
among psychiatric patients by Heinrich et al, [25], Kash-
ner et al. [26] and Killeen et al. [1], findings from this
study suggest that the self-reported use of mental health
services among individuals with a psychiatric disorder is
largely consistent with the use recorded in administrative
registers. The agreement between the CCHS-1.2 and the
RAMQs register w as especially low among the respon-
dents who had no mood or anxiety disorder according to
their self-reported symptoms. In consequence, the accu-
racy of self-reported use of mental health services may be
a pu blic health concern not so much for individual s with
a psychiatric disorder but rather for those who do not
meetthediagnosticcriteria(i.e., sub-clinical cases) or
who do not report their symptoms for fear of the preju-
dice against mental illness.
In this study, the disagreement between survey and
administrative dat a on the use of mental health services
was higher for less educated individuals and people living
with a partner. These findings tend to support the
hypothesis that social desirability may affect the accuracy
of self-reported use of mental health services since lower
education [12,14,15] and living with a partner [14] have
been associated with negative attitudes toward mental ill-
ness and the mentally ills. The higher odds of disagree-
ment between the CCHS-1.2 and the RAMQsregister
observed in women with young children than in men
sharing the same fami ly responsibility is somewhat puz-
zling. A tentative explanation may be that parenthood is
moresalientforwomenthanformensothattheneed
Table 3 Odds ratio for disagreement vs. agreement between data from the CCHS-1.2 and the RAMQ s register
Odds ratio 95% confidence interval
Age (ref., 18-64 years) 6.71 3.01 - 15.00
Gender (ref., Women) 1.40 0.956- 2.06
Education 0.78 0.66 - 0.93
Employment status (ref., Non employed) 0.85 0.58 - 1.24
Marital status (ref., Without a spouse) 2.48 1.72 - 3.60
Parental status (ref., (Without children in household) 0.77 0.52 - 1.14
Clinical profile (ref.: No psychiatric diagnosis)
Based on CCHS-1.2
Mood disorder 0.06 0.04 - 0.10
Anxiety disorder 0.09 0.05 - 0.16
Based on RAMQ
Neurotic disorder 0.36 0.20 - 0.63
Psychotic disorder 0.11 0.05 - 0.22
Other psychiatric disorder 0.40 0.21 - 0.79
Number of recorded mental health services 0.44 0.35 - 0.55
Number of criteria satisfied in a single medical visit 0.32 0.24 - 0.42
Medical specialty of physician
General practitioner (vs. other) 4.31 2.14 - 8.68
Psychiatrist (vs. other) 0.15 0.09 - 0.26
Type of psychiatric act
Consultation (vs. other) 0.22 0.11 - 0.42
Complete exam (vs. other) 0.18 0.11 - 0.29
Psychiatric treatment (vs. other) 0.73 0.48 - 1.10
Interval between most recent mental health service and survey interview 1.90 1.55 - 2.33
Drapeau et al. BMC Public Health 2011, 11:837
http://www.biomedcentral.com/1471-2458/11/837
Page 8 of 10
Page 8
forsocialapprovalregardingone scompetencyinthe
role of parent would be stronger for women than for
men.
In agreement with the meta-analysis carried out by
Bhandari and Wagner [21] of 42 studies based on the link-
age of surveys data or patients reports and administrative
data, the odds of disagreement between the CCHS-1.2 and
the RAMQ were lower for more salient (i.e., services pro-
vided by psychiatrists), more frequent and more recent
mental health services. Thus occasional and less recent
services were less l ikely to be recalled. All but one of the
studies reviewed by Bhandari and Wagner [21] focus on
somatic illness thus suggesting that recall bias is a threat
to the validity of survey data not only for the use of mental
health services but also for the use of general health ser-
vices. In this study, the hypothesis of recall bias is further
supported by the higher odds of both under- and over-
reporting in seniors, who are at higher risk of memory def-
icit, than in younger respondents. The higher odds of
under-reporting in seniors is consistent with the findings
of a study conducted by Rhodes and her colleagues [3,22],
which was based on the linkage of data from a large scale
population survey and data from the register maintained
by the Ontario Health Insurance Program (OHIP). A
number of strategies have been developed to foster the
recall of past events in a rese arch interview, including
decreasing th e length of the reference period and using a
calendar showing major events that have occurred during
that period. These strategies should be routinely applied in
surveys dealing with the use of health services.
Thisstudyprovidessomesupportforthehypothesis
that consumers and services providers or decision-makers
may have a different perspective on the definition of men-
tal health services. The point-of-views of patients, services
prov iders and decision-makers rega rding the na ture (i.e.,
mental health vs. other health problem) of the health ser-
vices received by patients are more likely to converge if
these services are salien t or fr equent than i f they are less
significant or o ccasi onal. As expected, the disagreement
between the CCHS-1.2 and the RAMQ was low for ser-
vices provided by psychiatrists and it decreased with the
number of s ervices recorded as mental health services in
the RAMQs register. Patients who were not examined by
a psychiatrist or who wer e recorded only one or twice as
users of mental health services in a 1-year period may not
have viewed these services as mental health services, espe-
cially if their motive for the consultation was no t r elated
to their mental health and if they were not treated (e.g.,
medication or therapy) for a mental health problem. Qua-
litat ive studies could contribute to clarify the consumers,
providers and decision-makers point-of-views regarding
the definition of a mental health service.
To our knowledge, t his study and the study conducted
by Rhodes and her colleagues [3,22] are the only published
studies that have ascertained the accuracy of mental health
services through the linkage of data from a population sur-
vey and a health services register. Both studies were con-
ducted in Canada. Other studies comparing the self-
reported use of mental health services and the use
recorded in administrative registers were based on psy-
chiatric patients [1,9,25,26] or Medicaid beneficiaries [27].
Our study replicates findings from these studies regarding
the effect of age and psychiatric disorder on the accuracy
of the self-reported use of mental health services. In addi-
tion, the profile of under-reporters observed in this study
coincides with the profile of people with negative attitude
towards mental illness in Canada and in other countries.
Still, additional studies linking data from population sur-
veys and administrative registers are needed to verify to
what extent the under-reporting of the use of mental
health services in population surveys is a widespread
phenomenon.
Conclusions
In conclusion, fin dings from this stu dy stress the need to
refine the investigation of mental health services in popu-
lation surveys and to combine survey and administrative
data, whenever possible, to obtain an optim al estimation
of the population need for mental health care and a more
detailed profile of the users of mental health services.
Indeed, in all likelihood, the systematic variatio n
observed in the discrepancies between the CCH S-1.2 and
the RAMQs register can, at least partly, be attributed to
the pooled effect of social desirability a nd r ecall bias
since the profile of under-reporters coincide with the
general profile of people expressing negative attitudes
towards mental illness and towards people with mental
health problems (e.g., seniors; less educated individuals;
people living with a partner) and of people with a mem-
ory deficit (e.g ., seniors). Thus survey data would tend to
under-estimate the use of mental health services in the
population and to bias the profile of users. The next step
would be to assess to what extent self-reported and
recorded data differ in their estimation of the use of men-
tal health services and in the description of the users.
Acknowledgements
This study was funded by the Canadian Institutes for Health Research
(#167330) and a researcher award to the first author (A Drapeau) from the
Fonds de recherche en santé du Québec (#12230). We are thankful to the
Institut de la statistique du Québec and the Régie de lassurance maladie du
Québec for their collaboration and to Anne-Marie Parent who carried out
the preliminary analyses under the supervision of the first author.
Author details
1
Département de psychiatrie de luniversité de Montréal, C.P. 6138 Succ.
Centre-Ville, Montréal, H3C 3J7, Canada.
2
Centre de recherche Fernand-
Seguin, 7331 rue Hochelaga, Montréal, H1N 3V2, Canada.
3
Département de
médecine sociale et préventive de luniversité de Montréal, C.P. 6138 Succ.
Centre-Ville, Montréal, H3C 3J7, Canada.
Drapeau et al. BMC Public Health 2011, 11:837
http://www.biomedcentral.com/1471-2458/11/837
Page 9 of 10
Page 9
Authors contributions
AD planned the study, reviewed the literature, supervised data analyses and
the interpretation of findings, and assumed leadership for the writing of the
manuscript. RB contributed to the planning of the study and of the analyses.
FBD took part in the review of the literature and carried out the analyses.
AD wrote the manuscript. RB and FBD participated in the interpretation of
findings and commented the first version of the manuscript. All authors read
and approved the final manuscript.
Competing interests
The authors declare that they have no competing interests.
Received: 10 June 2011 Accepted: 31 October 2011
Published: 31 October 2011
References
1. Killeen TK, Brady KT, Gold PB, Tyson C, Simpson KN: Comparison of self-
report versus agency records of service utilization in a community
sample of individuals with alcohol use disorders. Drug Alcohol
Dependence 2004, 73:141-147.
2. Marshall SF, Deapen D, Allen M, Anton-Culver H, Bernstein L, Horn-Ross PL,
Peel D, Pinder R, Reinolds P, Ross RK, et al: Validating California Teachers
Study self-reports of recent hospitalization: Comparison with California
hospital discharge data. American Journal of Epidemiology 2003,
158:1012-1020.
3. Rhodes AE, Lin E, Mustard CA: Self-reported use of mental health services
versus administrative records: should we care? International Journal of
Methods in Psychiatric Research 2002, 11:125-133.
4. Taube CA, Schlenger WE, Rupp A, Whitmore RW: Validity of Medicaid
household respondent reporting of ambulatory visits for mental
disorders. Journal of Economic and Social Measurement 1986, 14:243-256.
5. Kirby MJL, Keon WJ: De l ombre à la lumière. La transformation des
services concernant la santé mentale, la maladie mentale et la
toxicomanie au Canada. Canada: Comité sénatorial permanent des affaires
sociales, des sciences et de la technologie; 2006.
6. Hogan MF: Achieving the Promise: Transforming Mental Health Care in
America. Maryland (USA): New Freedom Commission on Mental Health;
2003.
7. Mental Health Commission of Canada: Toward Recovery & Well-Being. A
Framework for a Mental Health Strategy for Canada. Ottawa (Ontario):
Mental Health Commission of Canada; 2009.
8. Corrigan PW, Kleinlein P: The impact of mental illness stigma. In On the
Stigma of Mental Illness. Edited by: Corrigan PW. Washington (DC): American
Psychological Association; 2005:11-44.
9. Hennessy KD, Reed K: Validating self-reports of mental health service use
in a chronic population. Journal of Nervous and Mental Disease 1992,
180:399-400.
10. Jorm A, Griffiths K: The public s stigmatizing attitudes towards people
with mental disorders: how important are biomedical
conceptualizations? Acta Psychiatr Scand 2008, 118:315-321.
11. Lauber C, Nordt C, Falcato L, Rossler W: Factors influencing social distance
toward people with mental illness. Community Mental Health Journal 2004,
40(3):265-274.
12. Van t Veer J, Kraan H, Drosseart S, Modde J: Determinants that shape
public attitudes towards the mentally ill: A Dutch public study. Social
Psychiatry and Psychiatric Epidemiology 2006, 41:310-317.
13. Link B, Phelan J, Bresnahan M, Stueve A, Pescosolido B: Public conceptions
of mental illness: Labels, causes, dangerousness, and social distance.
American Journal of Public Health 1999, 89(9):1328-1333.
14. Jagdeo A, Cox BJ, Stein MB, Sareen J: Negative attitudes toward help
seeking for mental illness in 2 population-based surveys from the
United States and Canada. Can J Psychiatry 2009, 54(11):757-766.
15. Wang JL, Fick G, Adair C, Lai D: Gender specific correlates of stigma
toward depression in a Canadien general population sample. Journal of
affective disorders 2007, 103:91-97.
16. ten Have M, de Graaf R, Ormel J, Vilagut G, Kovess V, Alonso J, the
ESEMeD/MHEDEA 2000 Investigators: Are attitudes towards mental health
help-seeking associated with service use? Results from the European
Study of Epidemiology of mental Disorders. Soc Psychiatry Psychiatr
Epidemiol 2010, 45:153-163.
17. Aromaa E, Tolvanen A, Tuulari J, Wahlbeck K: Attitudes towards people
with mental disorders: the psychometric characteristics of a Finnish
questionnaire. Social Psychiatry and Psychiatric Epidemiology 2010,
45:265-273.
18. Gonzalez J, Alegria M, Prihoda T: How do attitudes toward mental health
treatment vary by age, gender, and ethnicity/race in young adults?
Journal of Community Psychology 2005, 33:611-629.
19. Baldwin ML, Marcus SC: Perceived and Measured Stigma Among Workers
With Serious Mental Illness. Psychiatric Services 2006, 57(3):388-392.
20. Baldwin ML, Marcus SC: Labor market outcomes of persons with mental
disorders. Industrial Relations 2007, 46:481-510.
21. Bhandari A, Wagner T: Self-reported utilization of health care services:
Improving measurement and accuracy. Medical Care Research and Review
2006, 63:217-235.
22. Rhodes AE, Fung K: Self-reported use of mental health services versus
administrative records: care to recall? International Journal of Methods in
Psychiatric Research 2004, 13:165-175.
23. Gravel R, Béland Y: The Canadian Community Health Survey: Mental
Health and Well-Being. Canadian Journal of Psychiatry 2005, 50:573-579.
24. Baulne J: Influence de lancrage social sur le comportement et lattitude
des femmes et des hommes par rapport à la maladie mentale:
Lexemple de la sous-déclaration du recours aux services psychiatriques
dans une enquête populationnelle. Projet dappariement EPSEBE -
Méthodologie et résultats. Direction de la méthodologie et de la qualité
Institut de la statistique du Québec; 2009.
25. Heinrich S, Deiser A, Birker T, Hierholzer C, Weigelt I, Zeichner D,
Angermeyer MC, Roick C, König H-H: Accuracy of self-reports of mental
health care utilization and calculated costs compared to hospital
records. Psychiatry Research 2011, 185:261-8.
26. Kashner TM, Suppes T, Rush AJ, Altshuler KZ: Measuring use of outpatient
care among mentally ill individuals: a comparison of self reports and
provider records. Evaluation and Program Planning
1999, 22:31-39.
27. Beebe TJ, McRae JA, Barnes SA: A comparison of self-reported use of
behavioral Health services with Medicaid agency records in Minnesota.
Psychiatric Services 2006, 57:1652-1654.
Pre-publication history
The pre-publication history for this paper can be accessed here:
http://www.biomedcentral.com/1471-2458/11/837/prepub
doi:10.1186/1471-2458-11-837
Cite this article as: Drapeau et al.: Discrepancies between survey and
administrative data on the use of mental health services in the general
population: findings from a study conducted in Québec. BMC Public
Health 2011 11:837.
Submit your next manuscript to BioMed Central
and take full advantage of:
Convenient online submission
Thorough peer review
No space constraints or color figure charges
Immediate publication on acceptance
Inclusion in PubMed, CAS, Scopus and Google Scholar
Research which is freely available for redistribution
Submit your manuscript at
www.biomedcentral.com/submit
Drapeau et al. BMC Public Health 2011, 11:837
http://www.biomedcentral.com/1471-2458/11/837
Page 10 of 10
Page 10
  • Source
    • "A number of limitations should be considered in the interpretation of the findings. First, the current dataset is based on self-reported data from a cross-sectional primary care mental health survey and differences in the reporting of mental health service use in surveys compared to administrative data has been highlighted in previous studies [58, 59] . Second, our results offer a partial view of the correlates of treatment adequacy from service utilization data that could be complemented by research on perceived needs for care and provider and clinic characteristics [60]. "
    [Show abstract] [Hide abstract] ABSTRACT: Purpose: Generalized Anxiety Disorder (GAD) is a common mental disorder in the primary care setting, marked by persistent anxiety and worries. The aims of this study were to: 1) examine mental health services utilisation in a large sample of primary care patients; 2) explore detection of GAD and minimal standards for pharmacological and psychological treatment adequacy based on recommendation from clinical practice guidelines; 3) examine correlates of treatment adequacy, i.e. predisposing, enabling and needs factors according to the Behavioural Model of Health Care Use. Methods: A sample of 373 adults meeting DSM-IV criteria for Generalized Anxiety Disorder in the past 12 months took part in this study. Data were drawn from the "Dialogue" project, a large primary care study conducted in 67 primary care clinics in Quebec, Canada. Following a mental health screening in medical clinics (n = 14833), patients at risk of anxiety or depression completed the Composite International Diagnostic Interview-Simplified (CIDIS). Multilevel logistic regression models were developed to examine correlates of treatment adequacy for pharmacological and psychological treatments. Results: Results indicate that 52.5 % of participants were recognized as having GAD by a healthcare professional in the past 12 months, and 36.2 % of the sample received a pharmacological (24.4 %) and/or psychological treatment (19.2 %) meeting indicators based on clinical practice guidelines recommendations. The detection of GAD by a health professional and the presence of comorbid depression were associated with overall treatment adequacy. Conclusions: This study suggests that further efforts towards GAD detection could lead to an increase in the delivery of evidence-based treatments. Key targets for improvement in treatment adequacy include regular follow up of patients with a GAD medication and access to psychotherapy from the primary care setting.
    Full-text · Article · Oct 2015 · BMC Family Practice
  • Source
    • "Also, nearly all the data collected for the NSMHWB relied on the participants' recall—asking participants to recall events from the previous 12 months. It is cognitively taxing to recall experiences over a 12 month period; this is an even more salient issue for people with depression and may introduce memory bias (Drapeau et al. 2011). Consideration of more proximal time frames is suggested for future work. "
    [Show abstract] [Hide abstract] ABSTRACT: Despite the recognised importance of accurate mental illness information in help-seeking and improving recovery, little is known about the dissemination of such information to people with depression. With a view to informing effective communication to those most in need, we explored the extent to which mental illness information is received by people with depression, its perceived helpfulness and we characterise those who do not receive such information. Using data from the Australian National Survey of Mental Health and Wellbeing we observed that mental illness information was received by 54.7 % of those with depression. Most (76.7 %) found it helpful. Pamphlets were the most frequently cited source of information. People who did not receive information were less educated, unlikely to have accessed mental health services and unlikely to believe they had mental health needs. Targeted information campaigns which shape perceptions of need in relation to depression have the potential to reduce the resultant disease burden.
    Full-text · Article · Jul 2015 · Community Mental Health Journal
  • Source
    • "Consequently, we were also unable to make statements about subsequent health care expenditures. Fourthly, health care use, work impairments, and work ability were all selfreported and thus vulnerable to recall bias [44, 45] . However , self-reported work ability is a widely used measure in the field of occupational health and it has shown to be a predictor for long-term sick leave, productivity loss, and disability benefit212223. "
    [Show abstract] [Hide abstract] ABSTRACT: Purpose: The aim of this study was to explore how work impairments and work ability are associated with health care use by workers with musculoskeletal disorders (MSD), cardiovascular disorders (CVD), or mental disorders (MD). Methods: In this cross-sectional study, subjects with MSD (n = 2,074), CVD (n = 714), and MD (n = 443) were selected among health care workers in 12 Dutch organizations. Using an online questionnaire, data were collected on individual characteristics, health behaviors, work impairments, work ability, and consultation of a general practitioner (GP), physiotherapist, specialist, or psychologist in the past year. Univariate and multivariate logistic regression analyses were performed to explore the associations of work impairments and work ability with health care use. Results: Lower work ability was associated with a higher likelihood of consulting any health care provider among workers with common disorders (OR 1.05-1.45). Among workers with MSD work impairments increased the likelihood of consulting a GP (OR 1.55), specialist (OR 2.05), and physical therapist (OR 1.98). Among workers with CVD work impairments increased the likelihood of consulting a specialist (OR 1.94) and physical therapist (OR 2.73). Among workers with MD work impairments increased the likelihood of consulting a specialist (OR 1.79) and psychologist (OR 1.82). Conclusion: Work impairments and reduced work ability were associated with health care use among workers with MSD, CVD, or MD. These findings suggest that addressing work-related problems in workers with common disorders may contribute in reducing health care needs.
    Full-text · Article · Jan 2014 · Journal of Occupational Rehabilitation
Show more