Available via license: CC BY 4.0
Content may be subject to copyright.
Articles
240
www.thelancet.com Vol 394 July 20, 2019
New WHO prevalence estimates of mental disorders in
conflict settings: a systematic review and meta-analysis
Fiona Charlson, Mark van Ommeren, Abraham Flaxman, Joseph Cornett, Harvey Whiteford, Shekhar Saxena
Summary
Background Existing WHO estimates of the prevalence of mental disorders in emergency settings are more than a
decade old and do not reflect modern methods to gather existing data and derive estimates. We sought to update
WHO estimates for the prevalence of mental disorders in conflict-aected settings and calculate the burden per
1000 population.
Methods In this systematic review and meta-analysis, we updated a previous systematic review by searching MEDLINE
(PubMed), PsycINFO, and Embase for studies published between Jan 1, 2000, and Aug 9, 2017, on the prevalence of
depression, anxiety disorder, post-traumatic stress disorder, bipolar disorder, and schizophrenia. We also searched the
grey literature, such as government reports, conference proceedings, and dissertations, to source additional data, and
we searched datasets from existing literature reviews of the global prevalence of depression and anxiety and reference
lists from the studies that were identified. We applied the Guidelines for Accurate and Transparent Health Estimates
Reporting and used Bayesian meta-regression techniques that adjust for predictors of mental disorders to calculate
new point prevalence estimates with 95% uncertainty intervals (UIs) in settings that had experienced conflict less
than 10 years previously.
Findings We estimated that the prevalence of mental disorders (depression, anxiety, post-traumatic stress disorder,
bipolar disorder, and schizophrenia) was 22·1% (95% UI 18·8–25·7) at any point in time in the conflict-aected
populations assessed. The mean comorbidity-adjusted, age-standardised point prevalence was 13·0% (95% UI
10·3–16·2) for mild forms of depression, anxiety, and post-traumatic stress disorder and 4·0% (95% UI 2·9–5·5) for
moderate forms. The mean comorbidity-adjusted, age-standardised point prevalence for severe disorders
(schizophrenia, bipolar disorder, severe depression, severe anxiety, and severe post-traumatic stress disorder) was
5·1% (95% UI 4·0–6·5). As only two studies provided epidemiological data for psychosis in conflict-aected
populations, existing Global Burden of Disease Study estimates for schizophrenia and bipolar disorder were applied
in these estimates for conflict-aected populations.
Interpretation The burden of mental disorders is high in conflict-aected populations. Given the large numbers of
people in need and the humanitarian imperative to reduce suering, there is an urgent need to implement scalable
mental health interventions to address this burden.
Funding WHO; Queensland Department of Health, Australia; and Bill & Melinda Gates Foundation.
Copyright © 2019 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license.
Introduction
Currently, there are major conflict-induced humanitarian
crises in numerous countries, including Afghanistan,
Iraq, Nigeria, Somalia, South Sudan, Syria, and Yemen.
UN estimates suggest that more than 68·6 million people
worldwide have been forcibly dis placed by violence and
conflict, the highest number of people aected since
World War 2.1 This increase in people aected by conflict
coincides with a growing interest in mental health, as
exemplified by the recently approved 10-year extension of
the Mental Health Action Plan by 194 WHO member
states.2 Interest is especially high in the mental health of
people aected by humanitarian emergencies.3
In 2005, WHO estimated the prevalence of mental
disorders among people aected by humanitarian
emergencies.4 These estimates have been frequently
repeated in policy documents,3,5,6 news media,7 and
appeals and funding proposals for help for people living
through the world’s worst crises. WHO emphasised that
these estim ates represented averages across emergency
settings and that observed prevalence estimates would
vary by aected population and assessment method.4
However, WHO’s 2005 estimates were not based on
applicable systematic reviews of evidence.
Epidemiological studies in conflict settings typically
present varying results, making their interpretation
dicult,8 and their statistical heterogeneity is extremely
high.9,10 We sought to update WHO estimates of the
prevalence of mental disorders in conflict-aected
populations by updating systematic literature reviews for
post-traumatic stress disorder and depression, searching
for a wider range of disorders, and applying Bayesian
meta-regression techniques while adjusting for predictors
of mental disorders in conflict settings. Natural disasters
Lancet 2019; 394: 240–48
Published Online
June 11, 2019
http://dx.doi.org/10.1016/
S0140-6736(19)30934-1
See Comment page 192
Policy and Epidemiology
Group, Queensland Centre for
Mental Health Research
(F Charlson PhD, J Cornett BS,
Prof H Whiteford PhD), and
School of Public Health,
University of Queensland
(F Charlson, Prof H Whiteford),
QLD, Australia; Institute for
Health Metrics and Evaluation,
University of Washington,
Seattle, WA, USA (F Charlson,
A Flaxman PhD, H Whiteford);
Department of Mental Health
and Substance Abuse, WHO,
Geneva, Switzerland
(M van Ommeren PhD); and
T H Chan School of Public
Health, Harvard University,
Boston, MA, USA (S Saxena MD)
Correspondence to:
Dr Mark van Ommeren,
Department of Mental Health
and Substance Abuse, WHO,
Geneva 1211, Switzerland
vanommerenm@who.int
Articles
www.thelancet.com Vol 394 July 20, 2019
241
and public health emergencies, such as the Ebola
virus outbreak, were outside the scope of this study.
Previous research has identified dierent mental health
consequences across these emergency settings11,12 (conflict
probably has more severe consequences), and our
selectivity was designed to limit heterogeneity within our
dataset. Our approach was in line with current WHO and
Inter-Agency Standing Committee humanitarian policies
and tools that include a broad multidisorder perspect-
ive.2,3,13,14 Furthermore, we aimed to estimate disease
burden in terms of years lived with disability (YLDs) per
1000 people aected by conflict.
Methods
Overview
For this systematic review and meta-analysis, we followed
the Guidelines for Accurate and Transparent Health
Estimates Reporting (GATHER) statement15 and used
methodologies developed for the Global Burden of
Diseases, Injuries, and Risk Factors Study (GBD) 2016.16
We refer to the generic term conflict as a substitute for
armed conflict and war. Current concepts and defi nitions
of conflict were extracted by searching peace and conflict
databases, organisation websites, and pub lished reports.
The relevance and usefulness of current concepts and
definitions of conflict were assessed to determine the
most appropriate database for our context. A critique of
the usefulness of each database identified five potentially
appropriate conflict databases. The Uppsala Conflict
Data Program,17 the Correlates of War project,18 the
Integrated Network for Societal Conflict Research Major
Episodes of Political Violence,19 and the Heidelberg
Institute for International Conflict Research Conflict
Barometer20 describe conflict as the existence of opposing
forces and stipulate a violence threshold described in
terms of number of deaths. The Political Terror Scale21
reports level of state terror according to state-perpetrated
human rights violations. We then did a quantitative
assessment of concordance between these five databases
using the kappa (κ) statistic.22 On the basis of the
assessments of usefulness and concordance, we elected
to use both the Uppsala Conflict Data Program and
Political Terror Scale databases. Further details of this
process can be found in the appendix and online.23
We based our dataset on a previous systematic review,10
which included studies published between 1980 and 2013
(search details in the appendix). We updated this review
by searching MEDLINE (PubMed), PsycINFO, and
Embase for studies published between Jan 1, 2000, and
Aug 9, 2017, to identify sources for the prevalence
Research in context
Evidence before this study
In 2005, in response to the Asian tsunami, WHO estimated
the prevalence of mental disorders among people affected by
humanitarian emergencies. These estimates were repeated in
policy documents, news media, and appeals and funding
proposals, but they did not have confidence intervals and
were not based on systematic reviews of evidence. We
searched MEDLINE (PubMed), PsycINFO, and Embase, to
identify studies published from Jan 1, 2000, to Aug 9, 2017,
to identify sources for the prevalence of post-traumatic stress
disorder, depression, and anxiety disorders using the
Diagnostic and Statistical Manual of Mental Disorders (DSM)
or the International Classification of Diseases (ICD) criteria
and variables known to be associated with prevalence (such
as exposure to trauma) to guide a predictor analysis. We used
the search string ((((((((“Warfare”[MESH]) OR “Warfare and
Armed Conflicts”[MESH]) OR “Torture”[MESH]) OR “Ethnic
Violence”[MESH]) OR “Exposure to Violence”[MESH]) OR
“Mass Casualty Incidents”[MESH]) OR “Civil
Disorders”[MESH])) AND (((((“Anxiety Disorders”[MESH]) OR
“Mood Disorders”[MESH]) OR “Trauma and Stressor Related
Disorders”[MESH]) OR “Stress, Psychological”[MESH]) OR
“Neurotic Disorders”[MESH]) AND
((((“Epidemiology”[MESH] OR “epidemiology” [Subheading])
OR “Prevalence”[MESH]) OR “Psychiatric Status Rating
Scales”[MESH])) for PubMed, and adapted it for the other
online databases. No language restriction was applied. We
also did a grey literature search using Google Scholar, datasets
from existing literature reviews, and reference lists from
studies identified.
Added value of this study
In this systematic review and meta-analysis, we updated WHO’s
2005 estimates for the prevalence of mental disorders in
conflict-affected low-income and middle-income settings,
focusing on depression, anxiety disorder, post-traumatic stress
disorder, bipolar disorder, and schizophrenia in settings that had
experienced conflict in the preceding 10 years. We estimated
that more than one in five people (22·1%) in post-conflict
settings has depression, anxiety disorder, post-traumatic stress
disorder, bipolar disorder, or schizophrenia and that almost one
in ten people (9·1%)in post-conflict settings has a moderate of
severe mental disorder at any point in time.
Implications of all the available evidence
Given that the prevalence of mental disorders was found to be
very high, there is a need to make available sustainable mental
health care in conflict-affected countries. This will require a
focus on investment in leadership and governance for mental
health; integrated and responsive mental health and social
care services in community-based settings; strategies for
promotion and prevention in mental health; and information
systems, evidence, and research for mental health in
conflict-affected countries. The well established links between
mental health, individual functioning, and country
development underscore the imperative to prioritise mental
health care in countries affected by conflict.
See Online for appendix
Articles
242
www.thelancet.com Vol 394 July 20, 2019
of post-traumatic stress disorder, depression, anxiety
disorders, schizophrenia, and bipolar disorder diagnosed
using the Diagnostic and Statistical Manual of Mental
Disorders (DSM) or the International Classification of
Diseases (ICD) criteria, and for variables known to be
associated with prevalence (such as exposure to trauma)
to guide a predictor analysis. No language restriction
was applied. We also searched the grey literature, such
as government reports, conference proceedings, and
dissertations, to source additional data. Sources included
Google search engines (eg, Google Scholar) and ProQuest
digital dissertations. Additionally, we searched datasets
from existing literature reviews of the global prevalence
of depression and anxiety24,25 and reference lists from the
studies that were identified. All grey-literature sources
identified were in English. We sought data on the
prevalence of schizophrenia and bipolar disorder in
conflict-aected populations from existing systematic
reviews.26,27
We included study samples that were representative of
the general conflict-aected population, defined as being
within a described geographical location and having
been in a state of conflict within 10 years preceding data
collection, as documented by the Uppsala Conflict Data
Program database.17 We only included studies of partici-
pants residing in their country of origin, or displaced
or resettled in a neighbouring low-income or middle-
income country; populations resettled in a high-income
country were excluded because there is evidence that the
hetero geneity might be considerable because of exposure
to external and environmental factors during the
resettling process.9 We included studies that reported
point or past-year prevalence estimates from either
cross-sectional or longitudinal population-based surveys.
Figure 1: Map of number of depression studies, 1980–2017
1
2
4
6
8
Depression datapoints
Figure 2: Map of number of any anxiety studies, 1980–2017
1
2
3
4
Any anxiety datapoints
Articles
www.thelancet.com Vol 394 July 20, 2019
243
Survey instru ments had to map to DSM or ICD
diagnostic criteria. A complete list of inclusion and
exclusion criteria can be found in the appendix.
Two of the authors (FC and JC) were responsible for the
searches and data extraction. Identified data sources were
reviewed by both authors and, if a consensus was not
reached, a third author (HW).
Data analysis
Data were extracted into a standardised excel template.
Duplicate data from the same study samples and reported
in multiple studies were identified and removed. We used
a Bayesian meta-regression model and the adaptive
Metropolis Markov-chain Monte Carlo method to draw
samples from the posterior distribution of all model
parameters simultaneously, with the modelling software
package DisMod-MR 1.0.16,28 To explain between-study
variability in prevalence, we included a range of study-level
and war-related covariates that had previously been
shown to have significant associations with mental
disorder prevalence.10 We reported age-standardised point
estimates based on the means of functions of these
parameter draws, and uncertainty intervals (UIs)
corresponding to the 2·5–97·5 percentile values. The UI
provides an upper and lower bound that the model predicts
to contain the true value with 95% certainty. Details on
covariate selection can be found in the appendix.
We conducted quality assessment at the time of data
extraction through our inclusion and exclusion criteria.
To adjust for comorbidities and severity splits in
depression, anxiety, and post-traumatic stress disorder,
we applied the prevalence of 41·6% (95% UI 39·8–43·4)
of indi viduals with depressive disorder who also had
comorbid anxiety, as previously identified from the
literature.29 Distributions of depression and anxiety
severity were taken from GBD 2016,16 which considers
several health states within a particular disease that are
reflective of dierent levels of functional impairment
(ie, none, mild, moderate, or severe) once disability
attributable to comorbid disorders is portioned out.30
In the absence of severity splits for post-traumatic
stress disorder, we relied on severity distributions for
anxiety disorders. More detail on GBD severity splits and
disability weights can be found in the appendix.
Figure 3: Map of number of post-traumatic stress disorder studies, 1980–2017
1
5
10
15
18
Post-traumatic stress
disorder datapoints
Depression Any anxiety disorder
(including post-traumatic
stress disorder)
Post-traumatic stress
disorder
Severe disorder 1·1% (0·3–2·2) 2·8% (1·8–4·0) 2·0% (1·1–3·2)
Moderate disorder 1·8% (1·2–2·6) 4·1% (2·9–5·6) 2·9% (1·7–4·4)
Mild disorder 6·4% (4·4–8·6) 8·5% (6·2–11·1) 6·1% (3·5–9·1)
Disorder without
functional impairment
1·4% (0·9–2·0) 6·2% (4·6–7·9) 4·4% (2·7–6·5)
Total 10·8% (8·1–14·2) 21·7% (16·7–28·3) 15·3% (9·9–23·5)
All severity splits taken from Global Burden of Diseases, Injuries, and Risk Factors Study 2016. Disorder without
functional impairment indicates cases with disability weight equal to zero once disability attributable to comorbid
disorders is portioned out.
Table 2: Age-standardised point prevalence with 95% uncertainty intervals, unadjusted for comorbidity
Point prevalence
(95% uncertainty
interval)
Severe disorder (severe anxiety, severe
post-traumatic stress disorder, severe
depression, schizophrenia, and bipolar disorder)
5·1% (4·0–6·5)
Moderate disorder (moderate anxiety,
moderate post-traumatic stress disorder, and
moderate depression)
4·0% (2·9–5·5)
Mild disorder (mild anxiety, mild post-traumatic
stress disorder, and mild depression)
13·0% (10·3–16·2)
Total 22·1% (18·8–25·7)
Table 1: Point prevalence estimates for mental disorders in
conflict-affected populations, adjusted for comorbidity
Articles
244
www.thelancet.com Vol 394 July 20, 2019
YLDs were derived by multiplying the number of
prevalent cases associated with each disorder by their
associated GBD disability weight. In place of GBD
prevalence estimates, we used prevalence estimates of
conflict-aected population mental disorder (derived as
described previously) as a primary input for YLD
estimation. Post-traumatic stress disorder was concep-
tualised as an anxiety disorder until the most recent
version of DSM (fifth edition) and, accordingly, was not
assessed as a separate disorder in GBD 2016, so we did not
calculate burden of disease estimates for post-traumatic
stress disorder. In our analyses, we considered all prevalent
cases of schizophrenia and bipolar disorder as severe. We
used Monte Carlo simulation–modelling techniques to
present 95% UIs around estimates reflecting the main
sources of sampling uncertainty in the calculations using
Ersatz software, version 1.2.31 More detailed information
on GBD burden of disease estimation can be found
elsewhere.16,32
Role of the funding source
WHO provided funding for this study and had a role in
study design, data inter pretation, and writing of the
report. The other funders had no role in study design,
data collection, data analysis, data interpretation, or
writing of the report. The corresponding author had
full access to all the data in the study and had final
responsibility for the decision to submit for publication.
Results
We identified 129 studies published between Jan 1, 1980,
and Aug 9, 2017, providing 96 studies with prevalence
estimates for post-traumatic stress disorder, 70 studies
with prevalence estimates for depression, and 38 studies
with prevalence estimates for any anxiety disorder
(appendix); 51 of these were studies published between
Jan 1, 2000, and Aug 9, 2017. 39 countries were represented
in the dataset; 34 had data for depression, 34 had data for
post-traumatic stress disorder, and 25 had data for anxiety
(figures 1–3; appendix).
We estimated that the prevalence of mental disorders
(depression, anxiety, post-traumatic stress disorder,
bipolar disorder, and schizophrenia) was 22·1%
(95% UI 18·8–25·7) at any point in time in the conflict-
aected populations assessed (table 1). Age-standardised
prevalence for depression, post-traumatic stress dis-
order, and anxiety disorders was elevated in conflict-
aected populations compared with global mean
prevalence (10·8% [95% UI 8·1–14·2] for depression,
15·3% [9·9–23·5] for post-traumatic stress disorder, and
21·7% [16·7–28·3] for any anxiety disorders; table 2).
The mild forms of all three disorders were the most
prevalent. Adjusting for comorbidity between depression
and anxiety, the mean, combined age-standardised
preva lence of mild, moderate, and severe depression,
post-traumatic stress disorder, and any anxiety disorders
was 21·2% [95% UI 17·7–24·7] in conflict-aected
popula tions (table 3). The mean comorbidity-adjusted,
age-stand ardised point prevalence was 13·0% (95% UI
10·3–16·2) for mild forms of depression, anxiety, and
post-traumatic stress disorder and 4·0% (95% UI
2·9–5·5) for moderate forms. The mean comorbidity-
adjusted, age-standardised point prevalence for severe
disorders (schizophrenia, bipolar disorder, severe
depression, severe anxiety, and severe post-traumatic
stress disorder) was 5·1% (95% UI 4·0–6·5). By
aggregating the prevalence of mental dis orders in
conflict-aected populations by severity, we estimated
Depression
(without comorbid
anxiety disorder)
Any anxiety disorder
(without comorbid
depression)
Any anxiety
disorder with
comorbid
depression*
Total
Severe disorder 0·6% (0·2–1·3) 3·3% (2·1–4·7) 0·4% (0·1–1·0) 4·3% (3·1–5·6)
Moderate disorder 1·1% (0·7–1·5) 2·2% (1·3–3·3) 0·8% (0·5–1·1) 4·0% (2·9–5·5)
Mild disorder 3·7% (2·6–5·1) 6·8% (4·4–9·6) 2·6% (1·9–3·6) 13·0% (10·3–16·2)
Total 5·3% (4·0–6·9) 12·1% (9·4–15·4) 3·8% (2·8–4·9) 21·2% (17·7–24·7)
Estimates of any anxiety disorder include post-traumatic stress disorder. Totals might not equal sum of parts due to
rounding. *Applying a rate of 41·6% (95% uncertainty interval 39·8–43·4) of depression cases with comorbid anxiety.
Global Burden of Diseases, Injuries, and Risk Factors Study 2016 severity splits applied.
Table 3: Age-standardised point prevalence with 95% uncertainty intervals, adjusted for comorbidity
Figure 4: Age-specific prevalence (mean) of depression and any anxiety and
post-traumatic stress disorder in conflict-affected populations, 2016
GBD 2016=Global Burden of Diseases, Injuries, and Risk Factors Study 2016.
0
10
20
30
40
50
60
70
80
90
100
Prevalence (%)
0 10 20 30 40 50 60 70
0
10
20
30
40
50
60
70
80
90
100
Prevalence (%)
Age (years)
Depression in conflict
Depression GBD 2016
Anxiety in conflict
Anxiety GBD 2016
Post-traumatic stress disorder in conflict
Articles
www.thelancet.com Vol 394 July 20, 2019
245
that at any point in time about 9% of the conflict-aected
population has moderate to severe mental disorders
(schizophrenia, bipolar dis order, moderate to severe
anxiety, moderate to severe post-traumatic stress
disorder, and moderate to severe depression; table 1).
We only identified two studies that provided epi-
demiological estimates for psychosis in conflict-aected
populations. A cross-sectional study of an internally
displaced population in South Darfur reported a pre-
valence of schizophrenia of 4·1%,33 and a general
population survey in Timor-Leste reported a schizophrenia
point prevalence of 0·34%.34 We did not identify any
studies that reported epidemiological estimates for bipolar
disorder in conflict-aected populations. This small
number of studies precluded pooling of estimates, and
we conservatively defaulted to global mean prevalence
estimates as derived by GDB 2016 for schizophrenia
(0·3% [95% UI 0·2–0·3]) and bipolar disorder
(0·6% [0·5–0·7]).35 Therefore, we were unable to take into
account any increase in psychosis or bipolar disorder
prevalence in populations aected by conflict.
In conflict settings, trends of depression and anxiety
prevalence increased with age. Mean prevalence of post-
traumatic stress disorder declined in the older age groups,
although there are large ranges of uncertainty surrounding
these estimates (figure 4). Our data suggest prevalence of
depression, post-traumatic stress disorder, or any anxiety
disorder is higher in women than in men, although this
finding was only significant for depression (appendix).
Examination of covariate coecients in our modelling
showed that symptom scales significantly overestimate
prevalence by about 1·5 to 2 times in conflict-aected
populations as compared with diagnostic tools in all
three disorder models (appendix).
Heterogeneity in our datasets was large. The median
value of the negative binomial model overdispersion
parameter calculated by DisMod-MR was 1·2 for anxiety,
0·95 for post-traumatic stress disorder, and 0·96 for
depression (where zero is completely uninformative, and
infinity is a Poisson distribution).
Age-specific YLDs in conflict-aected populations
showed elevated and significant dierences across most
age groups compared with estimated global YLDs in
GBD 2016 (figure 5). We estimated age-standardised
YLDs for depression in conflict-aected populations at a
rate of 24·8 YLDs per 1000 population (95% UI
16·4–36·0), in contrast to the GBD 2016 global age-
standardised estimate of 4·6 YLDs per 1000 population
(3·2–6·2). Age-standardised estimates of YLDs for any
anxiety disorder in conflict-aected populations were
23·2 YLDs per 1000 population (95% UI 17·0–29·9), as
compared with the GBD 2016 estimates of 3·5 YLDs per
1000 population (2·5–4·8).
Discussion
By updating our previous systematic review on
depression and post-traumatic stress disorder10 to include
more recent data and data on schizophrenia, bipolar, and
anxiety disorders, we identified an additional 45 studies
published over a 4-year period; this reflects a substantial
increase in psychiatric epidemiological research taking
place in conflict-aected contexts.
We estimated that approximately one in five people in
post-conflict settings has depression, anxiety disorder,
post-traumatic stress disorder, bipolar dis order, or
schizophrenia. This finding is in contrast to data from
GBD 2016,16 which suggest a mean global prevalence of
one in 14. Our empirically derived estimates show higher
prevalence of severe mental disorders than the previous
WHO estimates (about 5·1% point prevalence in current
estimate compared with 3–4% 12-month prevalence in
previous estimates) and higher prevalence of mild to
moderate mental disorders (approximately 17% point
prevalence in the revised estimates, compared with
15–20% 12-month prevalence in previous estimates). Our
estimates of YLDs per 1000 people for depression and
post-traumatic stress disorder were more than five times
higher than the existing global mean burden of disease
Figure 5: Age-specific years lived with disability (YLDs) per 1000 population (95% uncertainty interval) of
depression and any anxiety in conflict-affected populations, 2016
GBD 2016=Global Burden of Diseases, Injuries, and Risk Factors Study 2016.
0
10
20
30
40
50
60
70
80
90
100
YLDs per 1000 population
0–9 10–19 20–29 30–39 40–49 50–59 60–69 ≥70
0
10
20
30
40
50
60
70
80
90
100
YLDs per 1000 population
Age group (years)
Depression in conflict
Major depression GBD 2016
Anxiety in conflict
Anxiety GBD 2016
Articles
246
www.thelancet.com Vol 394 July 20, 2019
estimates. One previous study10 reported an age-
standardised pooled prevalence of 7·6% for depression
and 12·9% for post-traumatic stress disorder.
A useful finding from our study for field researchers
who use self-report or symptom-based measures to
ascertain mental disorder prevalence estimates is that
these instruments were shown to significantly over-
estimate the prevalence of depression, post-traumatic
stress disorder, and anxiety by 1·5 to 2 times. Most of
these instruments do not assess clinical significance
or function, and hence can overestimate prevalence of
disorders compared with diagnostic instruments.
Our study methodology has several strengths. By
contrast with previously published reviews, we applied
more stringent inclusion and exclusion criteria to our
literature search, optimised search strategies, and used
updated statistical methods.9,36 We sought to address
heterogeneity in epidemiological studies by use of
Bayesians approaches to allow for a more consistent set
of estimates. We made separate estimates for mild,
moderate, and severe mental disorders. Although the
clinical significance of mild mental disorders in
emergencies can be contested,8 the clinical needs of
people with severe mental disorders are too often
neglected.37 An important limitation in this study was the
raw data. Even with relatively strict inclusion criteria,
there was considerable heterogeneity in the mental
disorder datasets and their reported estimates, which
created large uncertainty around the predicted estimates.
This heterogeneity stemmed partly from dierences
across study designs—an issue inherent to psychiatric
epidemiology, particularly research following major
emergencies8—and partly from the myriad of factors that
aect the experience and expression of mental distress in
these settings. Many studies failed to report a robust
process of translation, cultural adaptation, or validity
testing of their instruments. However, a key strength
of the DisMod-MR approach is how it addresses
heterogeneity through adjustments to the data, which
allowed us to create a robust epidemiological profile of
mental disorders in conflict-aected populations.
Although not unique to the field of psychiatric epi-
demiology, issues related to the case definitions of mental
disorders warrant consideration in the context of the
settings represented in our study. Although reliable
systems of classification (DSM and ICD) make it possible
to determine prevalence estimates and, therefore, to guide
decisions about the development of services, these models
of mental disorders assume universality and might not
be the most useful way to describe the experience
and expression of psychological distress in many of
the contexts captured in our study.38 Further to the concept
of cultural variation are issues presented by shifts in
diagnostic criteria. Data included our study are pre-
dominantly based on DSM-IV; no studies using DSM-5
were identified. It is apparent that epidemiological
research, at least in this context, is yet to move on to most
recent version of DSM. In the event of such a transition, it
might be prudent to revise anxiety and post-traumatic
stress disorder prevalence and YLD estimates.
We only identified two studies on schizophrenia and
found no studies on bipolar disorder in conflict-aected
populations—too few to pool estimates using meta-
regression methods, especially given that one of the
studies estimated a ten times higher prevalence estimate
than the GBD 2016 prevalence estimate of schizophrenia.33
Therefore, we conservatively defaulted to global mean
prevalence estimates as derived by GDB 2016.16 The
estimates for psychosis we report here might thus be
underestimates and do not take into account the studies
we had to exclude from our systematic search that suggest
an increase in psychoses in populations aected by
conflict.39 Because of the paucity of data, we had to use
several assumptions and proxy inputs—such as a
comorbidity adjustment informed by a single study from a
conflict-aected population and the proxy use of GBD 2016
disability weights—which should be considered when
interpreting our findings, until more and better-quality
epidemiological data become available. Furthermore, the
study did not include comorbid disorders, such as alcohol
use disorders and epilepsy, which are frequently addressed
within mental health programmes.14
Nonetheless, our study identified the sustained
presence of high prevalence of mental disorders in
conflict-aected countries, making a compelling case for
global humanitarian, development, health, and mental
health communities to prioritise development of mental
health services in conflict and post-conflict settings.
Evidence for building systems for mental health care
after conflict shows that emergencies, which can
generate political interest and funding for mental health,
can be a catalyst for the meaningful development of
mental health care.3 A review of lessons learned from
such work in ten countries showed that focusing on
system-wide reform to address both new-onset and pre-
existing mental disorders is crucial.3 Practical guidance
for management of disorders that should be scaled up in
conflict-aected coun tries already exists. WHO and UN
High Commissioner for Refugees have designed the
mhGAP Humanitarian Intervention Guide,14 which
addresses the assessment and management of moderate
and severe mental disorders in non-specialised health-
care settings, such as general hospitals and primary
health care. Moreover, a variety of packages designed to
address multiple mental disorders, such as Problem
Management Plus, Common Elements Treatment
Approach and Self-Help Plus, have been used with
promising results among conflict-aected Pakistanis,
Burmese refugees, and South Sudanese refugees.40–42 It
should be noted that there is wide consensus that mental
health and psychosocial support for aected populations
should go beyond psychological and medical treatments
for mental disorders, and that such support should
include psychosocial intervention that strengthens
Articles
www.thelancet.com Vol 394 July 20, 2019
247
community self-help and support13 and advocacy for
security and protection and for adequate humanitarian
aid, including basic health services and livelihood
support.
Our findings highlight the need to prioritise conflict-
aected countries for implementation of the WHO
Mental Health Action Plan.2 This will require a focus on
investment in leadership and governance for mental
health, and the development of integrated, responsive
mental health and social care services in community-
based settings. Strategies for promotion and prevention
in mental health, and building and strengthening of
information systems, evidence, and research for mental
health in conflict-aected countries, are also needed.
These services could be initiated with short-term
emergency funds that are often available during crises.
Demonstration projects can provide proof of concept and
attract the further support and funds necessary for system
development to reduce the burden of mental disorders
among people aected by war and other conflict.3
Our study shows that the impact of conflict on people’s
mental health is higher than previous estimates suggest.
Mental health care must be prioritised in countries aected
by conflict, not least for the well established links between
mental health, functioning, and country development.
Contributors
FC and JC undertook the systematic review and data collection. FC and
AF undertook statistical analyses and modelling. FC, HW, and AF were
responsible for study design. FC, AF, MvO, and SS were responsible for
interpretation of results. All authors were responsible for writing and
editing of the manuscript.
Declaration of interests
We declare no competing interests.
Acknowledgments
We thank Dan Chisholm (WHO, Geneva, Switzerland), Domenico Giacco
(Queen Mary University of London, UK), Derrick Silove (University of
New South Wales, Sydney, Australia), and Peter Ventevogel (United
Nations High Commissioner for Refugees, Geneva, Switzerland) for
comments on an earlier version of this manuscript. The authors alone are
responsible for the views expressed in this Article and they do not
necessarily represent the views, decisions, or policies of the institutions
with which they are aliated.
References
1 United Nations Oce for the Coordination of Humanitarian Aairs
(OCHA). 2019 Global Humanitarian Overview. Geneva: OCHA, 2018.
2 WHO. Mental Health Action Plan 2013–2020. Geneva: World
Health Organization, 2013.
3 WHO. Building back better: sustainable mental health care after
emergencies. Geneva: World Health Organization, 2013.
4 van Ommeren M, Saxena S, Saraceno B. Aid after disasters.
BMJ 2005; 330: 1160–61.
5 Oce of the United Nations High Commissioner for Refugees.
Operational Guidance Mental Health and Psychosocial Support
Programming for Refugee Operations. Geneva: United Nations
High Commissioner for Refugees, 2013.
6 World Bank. Mental health among displaced people and refugees:
making the case for action at the World Bank Group. Washington:
World Bank, 2017.
7 Miller G. The tsunami’s psychological aftermath. Science 2005;
309: 1030.
8 Rodin D, van Ommeren M. Commentary: explaining enormous
variations in rates of disorder in trauma-focused psychiatric
epidemiology after major emergencies. Int J Epidemiol 2009;
38: 1045–48.
9 Steel Z, Chey T, Silove D, Marnane C, Bryant RA, van Ommeren M.
Association of torture and other potentially traumatic events with
mental health outcomes among populations exposed to mass
conflict and displacement: a systematic review and meta-analysis.
JAMA 2009; 302: 537–49.
10 Charlson FJ, Flaxman A, Ferrari AJ, Vos T, Steel Z, Whiteford HA.
Post-traumatic stress disorder and major depression in
conflict-aected populations: an epidemiological model and
predictor analysis. Glob Mental Health 2016; 3: e4.
11 Myles P, Swenshon S, Haase K, et al. A comparative analysis of
psychological trauma experienced by children and young adults in
two scenarios: evacuation after a natural disaster vs forced
migration to escape armed conflict. Public Health 2018;
158: 163–75.
12 Norris F, Friedman M, Watson P, Byrne C, Diaz E, Kaniasty K.
60 000 disaster victims speak: Part I. An empirical review of the
empirical literature, 1981–2001. Psychiatry 2002; 65: 207–39.
13 Inter-Agency Standing Committee. IASC guidelines on mental
health and psychosocial support in emergency settings. Geneva:
Inter-Agency Standing Committee, 2007.
14 WHO and United Nations High Commissioner for Refugees.
mhGAP Humanitarian Intervention Guide (mhGAP-HIG):
clinical management of mental, neurological and substance use
conditions in humanitarian emergencies. Geneva: World Health
Organization, 2015.
15 Stevens GA, Alkema L, Black RE, et al. Guidelines for accurate and
transparent health estimates reporting: the GATHER statement.
PLoS Med 2016; 13: e1002056.
16 Vos T, Abajobir AA, Abate KH, et al. Global, regional, and national
incidence, prevalence, and years lived with disability for
328 diseases and injuries for 195 countries, 1990–2016: a systematic
analysis for the Global Burden of Disease Study 2016. Lancet 2017;
390: 1211–59.
17 Uppsala Conflict Data Program. UCDP/PRIO armed conflict dataset.
2017. http://www.prio.no/Data/Armed-Conflict/UCDP-PRIO/
(accessed July 6, 2017).
18 Sarkees MR. The COW typology of war: defining and categorizing
wars (version 4 of the data). 2010. http://www.correlatesofwar.org/
data-sets/COW-war/the-cow-typology-of-war-defining-and-
categorizing-wars/view (accessed Feb 13, 2013).
19 Integrated Network for Societal Conflict Research. Major Episodes
of Political Violence (MEPV) and conflict regions, 1946–2008. 2012.
http://www.systemicpeace.org/inscr/inscr.htm (accessed
Feb 13, 2013).
20 Heidelberg Institute for International Conflict Research. Conflict
barometer 2012: disputes - non-violent crises - violent crises - limited
wars - wars. 2013. https://friedensbildung-schule.de/sites/
friedensbildung-schule.de/files/anhang/medien/fbs-conflict-
barometer-2012-204.pdf (accessed July 6, 2017).
21 Political Terror Scale. The Political Terror Scale. 2017. http://www.
politicalterrorscale.org/Data/Data-Archive.html (accessed July 6, 2017).
22 Sim J, Wright CC. The kappa statistic in reliability studies:
use, interpretation, and sample size requirements. Phys Ther 2005;
85: 257–68.
23 Queensland Centre for Mental Health Research. A review of the
operational definitions of conflict: implications for epidemiological
research. 2017. https://qcmhr.uq.edu.au/wp-content/uploads/2019/
04/Operationalisation-of-a-definition-of-conflict.pdf (accessed
April 1, 2019).
24 Ferrari A, Somerville A, Baxter A, et al. Global variation in the
prevalence and incidence of major depressive disorder: a systematic
review of the epidemiological literature. Psychol Med 2013;
43: 471–81.
25 Baxter A, Scott K, Vos T, Whiteford H. Global prevalence of anxiety
disorders: a systematic review and meta-regression. Psychol Med
2013; 43: 897–910.
26 Saha S, Chant D, Welham J, McGrath J. A systematic review of the
prevalence of schizophrenia. PLoS Med 2005; 2: e141.
27 Ferrari AJ, Saha S, McGrath JJ, et al. Health states for schizophrenia
and bipolar disorder within the Global Burden of Disease 2010
Study. Popul Health Metr 2012; 10: 16.
28 Flaxman AD, Vos T, Murray CJL, eds. An integrative metaregression
framework for descriptive epidemiology. Seattle, WA: University of
Washington Press; 2013.
Articles
248
www.thelancet.com Vol 394 July 20, 2019
29 Kessler RC, Sampson NA, Berglund P, et al. Anxious and
non-anxious major depressive disorder in the World Health
Organization World Mental Health Surveys. Epidemiol Psychiatr Sci
2015; 24: 210–26.
30 Ferrari AJ, Charlson FJ, Norman RE, et al. Burden of depressive
disorders by country, sex, age, and year: findings from the global
burden of disease study 2010. PLoS Med 2013; 10: e1001547.
31 Barendregt JJ. Ersatz Version 1.2. 2012. http://www.epigear.com/
index_files/ersatz.html (accessed Nov 14, 2017).
32 Whiteford HA, Degenhardt L, Rehm J, et al. Global burden of disease
attributable to mental and substance use disorders: findings from the
Global Burden of Disease Study 2010. Lancet 2013; 382: 1575–86.
33 Elhabiby MM, Radwan DN, Okasha TA, El-Desouky ED.
Psychiatric disorders among a sample of internally displaced
persons in South Darfur. Int J Soc Psychiatry 2014; 61: 358–62.
34 Silove D, Bateman CR, Brooks RT, et al. Estimating clinically
relevant mental disorders in a rural and an urban setting in
postconflict Timor Leste. Arch Gen Psychiatry 2008; 65: 1205–12.
35 Charlson F, Ferrari A, Santomauro D, et al. Global epidemiology
and burden of schizophrenia: findings from the Global Burden of
Disease Study 2016. Schizophr Bull 2018; 44: 1195–203.
36 Morina N, Akhtar A, Barth J, Schnyder U. Psychiatric disorders in
refugees and internally displaced persons after forced displacement:
a systematic review. Front Psychiatry 2018; 9: 433.
37 Jones L, Asare JB, El Masri M, Mohanraj A, Sherief H,
Van Ommeren M. Severe mental disorders in complex
emergencies. Lancet 2009; 374: 654–61.
38 Patel V, Cohen A, Kohrt B, Whiteford H, Lund C. Global mental
health. In: Merson MH, Black R, Mills AJ, eds. Global health:
diseases, programs, systems and policies. Burlington, MA:
Jones & Bartlett Learning, 2018: 423–76.
39 Llosa AE, Ghantous Z, Souza R, et al. Mental disorders, disability
and treatment gap in a protracted refugee setting. Br J Psychiatry
2014; 204: 208–13.
40 Rahman A, Hamdani SU, Awan NR, et al. Eect of a multicomponent
behavioral intervention in adults impaired by psychological distress in
a conflict-aected area of Pakistan: a randomized clinical trial. JAMA
2016; 316: 2609–17.
41 Brown FL, Carswell K, Augustinavicius J, et al. Self Help Plus: study
protocol for a cluster randomised controlled trial of guided self-help
with South Sudanese refugee women in Uganda. Glob Ment Health
2018; 5: e27.
42 Bolton P, Lee C, Haroz EE, et al. A transdiagnostic
community-based mental health treatment for comorbid disorders:
development and outcomes of a randomized controlled trial among
Burmese refugees in Thailand. PLoS Med 2014; 11: e1001757.
