http://journals.cambridge.org/PSM
AdditionalservicesforPsychologicalMedicine:
Emailalerts:Clickhere
Subscriptions:Clickhere
Commercialreprints:Clickhere
Termsofuse:Clickhere
Spiritualandreligiousbeliefsasriskfactorsfortheonsetofmajor
depression:aninternationalcohortstudy
B.Leurent,I.Nazareth,J.BellnSaameo,M.I.Geerlings,H.Maaroos,S.Saldivia,I.vab,F.TorresGonzlez,M.Xavier
andM.King
PsychologicalMedicine/FirstViewArticle/January2013,pp112
DOI:10.1017/S0033291712003066,Publishedonline:
Linktothisarticle:http://journals.cambridge.org/abstract_S0033291712003066
Howtocitethisarticle:
B.Leurent,I.Nazareth,J.BellnSaameo,M.I.Geerlings,H.Maaroos,S.Saldivia,I.vab,F.TorresGonzlez,M.Xavier
andM.KingSpiritualandreligiousbeliefsasriskfactorsfortheonsetofmajordepression:aninternationalcohortstudy.
PsychologicalMedicine,AvailableonCJOdoi:10.1017/S0033291712003066
RequestPermissions:Clickhere
Downloadedfromhttp://journals.cambridge.org/PSM,IPaddress:128.40.240.4on30Jan2013
O R I G I N A L AR T I C LE
Mental Health Sciences Unit, Faculty of Brain Sciences, University College London Medical School, UK
Research Department of Primary Care and Population Health, University College London Medical School, UK
3
Department of Preventive Medicine, El Palo Health Centre, Malaga, Spain
4
University Medical Centre, Utrecht, The Netherlands
5
Faculty of Medicine, University of Tartu, Estonia
6
Departamento de Psiquiatra y Salud Mental, Universidad de Concepcion, Chile
7
Department of Family Medicine, University of Ljubljana, Slovenia
8
CIBERSAM-Granada University, Granada, Spain
9
Department of Mental Health, Faculdade Ciencias Medicas, CEDOC, Lisboa, Portugal
2
Background. Several studies have reported weak associations between religious or spiritual belief and psychological
health. However, most have been cross-sectional surveys in the USA, limiting inference about generalizability. An
international longitudinal study of incidence of major depression gave us the opportunity to investigate this
relationship further.
Method. Data were collected in a prospective cohort study of adult general practice attendees across seven countries.
Participants were followed at 6 and 12 months. Spiritual and religious beliefs were assessed using a standardized
questionnaire, and DSM-IV diagnosis of major depression was made using the Composite International Diagnostic
Interview (CIDI). Logistic regression was used to estimate incidence rates and odds ratios (ORs), after multiple
imputation of missing data.
Results. The analyses included 8318 attendees. Of participants reporting a spiritual understanding of life at
baseline, 10.5 % had an episode of depression in the following year compared to 10.3 % of religious participants
and 7.0 % of the secular group (p<0.001). However, the ndings varied signicantly across countries, with
the dierence being signicant only in the UK, where spiritual participants were nearly three times more likely
to experience an episode of depression than the secular group [OR 2.73, 95 % condence interval (CI) 1.594.68].
The strength of belief also had an eect, with participants with strong belief having twice the risk of
participants with weak belief. There was no evidence of religion acting as a buer to prevent depression after
a serious life event.
Conclusions. These results do not support the notion that religious and spiritual life views enhance psychological
well-being.
Received 13 July 2012 ; Revised 19 November 2012 ; Accepted 6 December 2012
Key words : General practice, longitudinal, major depression, religion, spirituality.
Background
Research ndings, most originating from the USA,
have generally reported a positive, albeit weak, relationship between higher levels of religious involvement and better health once other inuences, such as
age, sex and social support, have been taken into account (Koenig et al. 1998, 2001 ; McCullough & Larson,
1999 ; Johnstone et al. 2008). Nevertheless, many other
studies do not nd such an association (Payne et al.
1991 ; Schaefer, 1997 ; Lewis et al. 2000 ; Gartner et al.
2012). There have been at least two meta-analyses of
relevant studies. The rst examined the relationship
between religiosity and psychological adjustment
(Hackney & Sanders, 2003), while excluding studies
measuring a wider concept of spirituality or those
examining the relationship with mental disorders.
B. Leurent et al.
Method
Study setting and design
The study, described in detail elsewhere (King et al.
2006b, 2008), was approved by research ethics committees in each country. It was a prospective cohort
study conducted in (1) 25 general practices in the
Medical Research Council (MRC) General Practice
Research Framework (GPRF) in the UK ; (2) nine large
primary care centres in Andaluca, Spain ; (3) 74 general practices nationwide in Slovenia ; (4) 23 general
practices nationwide in Estonia ; (5) seven large general practice centres near Utrecht, The Netherlands ;
(6) two large primary care centres in the Lisbon area of
Portugal ; and (7) 78 general practices in Concepcion
and Talcahuano in the Eighth Region of Chile. General
practices covered urban and rural populations with
considerable socio-economic variation.
B. Leurent et al.
obtained on imputed data by univariable logistic regression and presented graphically. Odds ratios (ORs)
unadjusted and adjusted for age, sex, education, employment, social support, past history of depression
and country were computed using logistic regression.
Interactions between the predictor and each covariate
were tested and the model was stratied if an interaction was found. In view of multiple testing, we
considered an interaction signicant at the level of
p<0.01. All other p values reported are two-sided, and
considered signicant at the level of p<0.05. Incidence
rates and ORs are reported along with their 95 % condence intervals (CIs). Standard errors are based on
robust sandwich estimates to account for the clustering eect of each general practice (Huber, 1967).
For the analyses on imputed data, the exact number
of participants in each category of the exposure
variables cannot be reported as it varied slightly
with each imputation. In the presence of missing
data, CIs are a more reliable guide to the precision
around estimates than estimated frequencies. Statistical analyses were performed using Stata Release 11
(Stata Corp, 2009).
Results
Response rates and prevalence of DSM-IV major
depression at recruitment
Response to recruitment was high in Portugal (76 %),
Estonia (80 %), Slovenia (80 %) and Chile (97 %) but
lower in the UK (44 %) and The Netherlands (45 %).
Ethical constraints did not allow collection of data on
non-responders at baseline. Across all countries the
response to follow-up at 6 months was at 89.5 %. A
total of 10045 people took part : 219 were excluded on
grounds of age, 143 had a missing CIDI diagnosis
and ve a missing practice identication at baseline.
Of the remaining 9678, 8318 without DSM-IV major
depression (86 %) at baseline were analysed. The median age was 49 years, two-thirds were women,
and 75 % of participants held a religious or spiritual
understanding of life. Women were more likely
than men to have a religious or spiritual understanding of life, and a past history of depression was least
common in people with a spiritual understanding
(Table 1). The characteristics of participants by country are reported in Supplementary Table S1 (available
online). Eleven per cent of participants did not complete the 6-month follow-up and 16 % did not complete the 12-month follow-up. Non-completers tended
to be younger and less educated and had experienced
more serious life events in the 6 months before baseline.
Spiritual
(n=1746, 21 %)
Neither
(n=2087, 25 %)
Totala
(n=8318)
p valueb
462 (10.6)
714 (16.4)
322 (7.4)
189 (4.3)
358 (8.2)
833 (19.2)
1470 (33.8)
300 (17.2)
136 (7.8)
328 (18.8)
227 (13.0)
143 (8.2)
75 (4.3)
537 (30.8)
354 (17.0)
153 (7.3)
387 (18.5)
506 (24.2)
469 (22.5)
97 (4.6)
121 (5.8)
1131 (13.6)
1006 (12.1)
1048 (12.6)
923 (11.1)
1077 (12.9)
1005 (12.1)
2128 (25.6)
<0.001
3045 (70.0)
1303 (30.0)
1235 (70.7)
511 (29.3)
1235 (59.2)
852 (40.8)
5599 (67.3)
2719 (32.7)
<0.001
498 (11.5)
585 (13.5)
764 (17.6)
946 (21.8)
1022 (23.5)
533 (12.3)
332 (19.0)
356 (20.4)
356 (20.4)
333 (19.1)
266 (15.2)
103 (5.9)
406 (19.5)
429 (20.6)
398 (19.1)
435 (20.8)
293 (14.0)
126 (6.0)
1244 (15.0)
1380 (16.6)
1540 (18.5)
1745 (21.0)
1619 (19.5)
790 (9.5)
<0.001
1333 (30.7)
3005 (69.3)
10
615 (35.3)
1129 (64.7)
2
634 (30.5)
1445 (69.5)
8
2614 (31.5)
5681 (68.5)
23
0.001
656 (15.1)
1324 (30.5)
2011 (46.3)
349 (8.0)
8
533 (30.5)
627 (35.9)
388 (22.2)
197 (11.3)
1
749 (36.0)
820 (39.4)
343 (16.5)
167 (8.0)
8
1965 (23.7)
2844 (34.3)
2769 (33.4)
718 (8.7)
22
<0.001
1663 (38.4)
1017 (23.5)
1654 (38.2)
14
975 (56.1)
276 (15.9)
487 (28.0)
8
1313 (63.5)
336 (16.2)
420 (20.3)
18
4003 (48.4)
1662 (20.1)
2610 (31.5)
43
<0.001
1535 (35.4)
2806 (64.6)
7
563 (32.3)
1180 (67.7)
3
162 (7.8)
1921 (92.2)
4
2265 (27.3)
6039 (72.7)
14
<0.001
2021 (46.5)
2322 (53.5)
5
728 (41.8)
1015 (58.2)
3
1039 (49.9)
1046 (50.2)
2
3863 (46.5)
4445 (53.5)
10
<0.001
12.7 (2.4)
34
12.3 (2.7)
10
12 (2.7)
19
12.4 (2.5)
77
<0.001
2680 (62.2)
1248 (29.0)
243 (5.6)
137 (3.2)
40
563 (33.0)
317 (18.6)
118 (6.9)
708 (41.5)
40
3243 (53.9)
1565 (26.0)
361 (6.0)
845 (14.1)
80
<0.001
4.3 (1.5)
90
3.7 (1.6)
30
4.1(1.6)
120
<0.001
N.A.
N.A.
B. Leurent et al.
Table 2. Odds ratios (ORs) for onset of major depression over 12 months by understanding
of life, compared to neither religious nor spiritual
Unadjusted OR (95 % CI)
Overall
Religious
Spiritual
1.52 (1.191.93)**
1.56 (1.212.02)**
1.14 (0.871.50)
1.32 (1.021.70)*
UK
Religious
Spiritual
1.35 (0.682.68)
2.73 (1.594.68)***
1.86 (0.883.92)
2.68 (1.524.71)**
Spain
Religious
Spiritual
1.41 (0.782.54)
1.53 (0.733.20)
1.34 (0.732.46)
1.50 (0.733.07)
Slovenia
Religious
Spiritual
0.62 (0.261.46)
1.06 (0.522.16)
0.69 (0.271.73)
1.10 (0.522.34)
Estonia
Religious
Spiritual
1.16 (0.632.13)
1.08 (0.591.98)
1.06 (0.512.22)
1.14 (0.622.12)
The Netherlands
Religious
Spiritual
0.63 (0.311.28)
1.15 (0.622.13)
0.69 (0.351.37)
1.07 (0.572.02)
Portugal
Religious
Spiritual
2.40 (0.5510.43)
1.96 (0.3810.05)
1.78 (0.398.08)
1.52 (0.278.48)
Chile
Religious
Spiritual
1.12 (0.582.17)
1.03 (0.532.00)
1.08 (0.542.14)
1.04 (0.561.96)
16
Incidence
95% CI
14
12.5
12
11.7
10.1
10
9.2
8
7.4
6.2
6
4
2
n*= 520
443
1075
1255
947
1734
Weakly
Strongly
Fig. 1. Incidence of major depression versus strength of spiritual or religious belief. Based on imputed data. * Frequencies based
on observed data, not numbers included in analysis. CI, Condence interval.
2.21
2
1.47
1.25
1.00
0.94
0.93
0.92
0.77
0.79
0.5
0.25
n*= 51
5/4
140
328
844
2193
806
405
171
113
0 (Ref.)
+1
+2
+3
+4/+5
B. Leurent et al.
Table 3. Odds ratios (ORs) for onset of major depression between 6 and 12 months, after a
change in understanding of life between baseline and 6 months
Understanding of life
At baseline
At 6 months
Frequencya
Unadjusted OR
(95 % CI)
Adjusted ORb
(95 % CI)
Religious
Religious
Spiritual
Neither
3004
658
175
1.00 (Reference)
0.98 (0.621.53)
0.64 (0.261.56)
1.00 (Reference)
0.97 (0.611.55)
0.87 (0.352.16)
Spiritual
Religious
Spiritual
Neither
436
887
253
1.40 (0.842.32)
1.00 (Reference)
1.05 (0.542.03)
1.51 (0.872.61)
1.00 (Reference)
1.09 (0.562.14)
Neither
Religious
Spiritual
Neither
196
265
1413
1.53 (0.703.36)
1.46 (0.722.96)
1.00 (Ref)
1.31 (0.533.19)
1.35 (0.662.78)
1.00 (Ref)
20
Religious
Spiritual
Neither
15
10
n* = 3474
0
1749
2235
1
>1
Discussion
Main ndings
We found that people who held a religious or spiritual
understanding of life had a higher incidence of depression than those with a secular life view. However,
this nding varied by country ; in particular, people in
the UK who had a spiritual understanding of life were
the most vulnerable to the onset of major depression.
Regardless of country, the stronger the spiritual or religious belief at baseline, the higher the risk of onset of
depression. Although our main nding of an association between religious life understanding and onset
of depression varied by country, we found no
evidence that spirituality may protect people, and
only weak evidence that a religious life view was
possibly protective in two countries (Slovenia and The
Netherlands). Finally, there was no moderating eect
of religious and spiritual understanding of life on the
impact of life events on onset of major depression.
Strengths and limitations
The main strengths of our study are its prospective
cohort design, the involvement of several countries,
the large sample size and the use of standardized assessments of religious/spiritual belief, other risk variables and major depression. However, combining
people from dierent cultures creates heterogeneity
and runs the risk of missing real dierences within
countries. Despite a large sample size, we lacked
power for some of the analyses, particularly the assessment of statistical interactions. However, to our
knowledge there are very few data sets available containing variables on spiritual and religious beliefs that
are large enough to undertake interaction analyses.
Thus we consider it is important to show the results to
avoid a reporting bias that favours statistically signicant results (Dwan et al. 2008). Another strength
of this study is the use of multiple imputation to take
account of participants who dropped out. Assuming
predictors of missingness have been examined at
baseline, this should give unbiased ndings, as opposed to restricting the analysis to participants who
completed both follow-ups only, which may represent
a sample at lesser risk of depression than the original
one. However, participations rates were low in some
countries, and we cannot assume that ndings in
10
B. Leurent et al.
Implications
Why a religious or spiritual life view might place
people at risk of depression remains unclear. One explanation is that people predisposed to depression at
baseline may seek meaning in spiritual or religious
sources. The possibility that people predisposed to
depression increase their search for existential meaning in religion and spirituality is supported by our
nding that change in belief over time towards greater
religiosity did seem to be related to greater risk of depression.
Supplementary material
For supplementary material accompanying this paper
visit http://dx.doi.org/10.1017/S0033291713003066.
Acknowledgements
The study in Europe was funded by a European
Commission Vth Framework grant (PREDICT-QL4CT2002-00683). Funding in Chile was provided by
project FONDEF DO2I-1140. We are grateful for part
support in Europe from : the Estonian Scientic
Foundation (grant 5696) ; the Slovenian Ministry for
Research (grant 4369-1027) ; the Spanish Ministry of
Health (FIS references : PI041980, PI041771, PI042450)
and the Spanish Network of Primary Care Research,
redIAPP (ISCIII-RETIC RD06/0018) and SAMSERAP
group ; and the UK NHS Research and Development
oce for service support costs in the UK. The funders
had no direct role in the design or conduct of the
study, interpretation of the data or review of the
manuscript.
M. King had full access to the data and takes responsibility for their integrity and the accuracy of
the data analysis. We thank all patients and general
practice sta who took part ; the European Oce at
University College London for their administrative
assistance at the coordinating centre ; K. McCarthy,
the projects scientic ocer in the European
Commission, Brussels, for his helpful support and
guidance ; the UK MRC GPRF ; L. Letley from the MRC
GPRF ; the GPs of the Utrecht General Practitioners
Network ; and the Camden and Islington Mental
Health and Social Care Trust. We also acknowledge
the Maristan network, through which the collaboration in Spain, Portugal, the UK and Chile rst developed.
Declaration of Interest
None.
11
12
B. Leurent et al.