Mental disorders are the leading cause of disability worldwide,
accounting for 23% of all non-fatal burden.
1
Approximately 38%
of the EU population experience a mental disorder each year,
2
causing significant societal costs, estimated at e453 billion in Eur ope
in 2010; in the USA costs were $300 billion in 2002–2003.
3,4
Mental
disorders are treatable and potentially preventable.
5–7
However,
help-seeking is often delayed or completely absent.
8
The low
treatment rate further aggravates burden and costs,
9
as untreated
individuals are more likely to experience problematic interpersonal
and family functioning and have lower life expectancies.
10–13
Prevention of mental disorders through early intervention and
the encouragement of help-seeking are major challenges for public
health.
14,15
However, several factors influence help-seeking for
mental health problems. Desire to handle the problem on one’s
own, low perceived need, low mental health literacy and financial
factors are associated with a reduction in help-seeking.
16–20
Negative and stigmatising attitudes towards mental illness, and
towards help-seeking and people with mental illness, further referred
to as stigma, are other important barriers to help-seeking.
21–27
Commonly, four stigma types that influence help-seeking can be
distinguished: perceived public stigma (PublicS), personal stigma
(PersonS), self-stigma (SelfS) and attitudes towards help-seeking
(HelpA). PublicS and PersonS are two types of public stigma (also
referred to as social or enacted stigma), defined as the stigmatising
perception about a person who has a mental illness endorsed
collectively by members of the general population.
22,28–31
More
specifically, PublicS is understood to be the individual’s perception
of public stigma,
22
as measured by Link’s Perceived Devaluation
Discrimination Scale;
32
PersonS, on the other hand, describes
personal attitudes towards members of a stigmatised group ,
29,33–37
and can find a behavioural expression in the desire for social
distance.
38
When these two types of public stigma were compared,
endorsement of PublicS was substantially higher than PersonS.
33
SelfS (also called internalised or anticipated stigma) occurs when
an individual affected by a mental illness endorses stereotypes about
mental illness, anticipates social rejection, considers stereotypes to be
self-relevant and believes himself or herself to be a devalued member
of society.
28–30,39–42
HelpA includes the perception of a need for
help, stigma tolerance associated with seeking such services,
openness regarding one’s problems and confidence that the help
will be of assistance.
43
Overall, stigma is a multifaceted concept
andhas,therefore,beenmeasuredwithavarietyofinstruments.
28,44
Recent reviews of the influence of mental health-related stigma
on help-seeking have reported that stigma, in particular SelfS and
HelpA, had negative effects on help-seeking.
21–27
Many of these
studies did not distinguish between intended or recommended
and active help-seeking, thereby referring to the Theory of
Planned Behaviour,
45
which proposes that intentions correlate
strongly with behaviour.
46
In practice, however, although most
people would recommend seeking professional help for mental
problems,
47
or report an intention to seek help when affected by
mental problems themselves,
48
a considerably lower proportion
actually sought it.
8
Stigma might be one reason for not putting
help-seeking intentions into action. However, only active help-
seeking will reduce the burden of the disorder. We conducted, for
the first time, a systematic review and meta-analysis to estimate
the association of the four types of stigma with activ e help-seeking
in the general population. Additionally, we estimated the role of
potential moderating study characteristics such as sample source
or response rate.
Method
Our systematic review and meta-analysis was conducted in
accordance with the Preferred Reporting Items for Systematic
Reviews and Meta-Analyses (PRISMA) guidelines (see online
261
Association between mental health-related
stigma and active help-seeking: systematic
review and meta-analysis
Nina Schnyder, Radoslaw Panczak, Nicola Groth and Frauke Schultze-Lutter
Background
Mental disorders create high individual and societal costs
and burden, partly because help-seeking is often delayed or
completely avoided. Stigma related to mental disorders or
mental health services is regarded as a main reason for
insufficient help-seeking.
Aims
To estimate the impact of four stigma types (help-seeking
attitudes and personal, self and perceived public stigma) on
active help-seeking in the general population.
Method
A systematic review of three electronic databases was
followed by random effect meta-analyses according to the
stigma types.
Results
Twenty-seven studies fulfilled eligibility criteria. Participants’
own negative attitudes towards mental health help-seeking
(OR = 0.80, 95% CI 0.73–0.88) and their stigmatising attitudes
towards people with a mental illness (OR = 0.82, 95% CI 0.69–
0.98) were associated with less active help-seeking. Self-
stigma showed insignificant association (OR = 0.88, 95% CI
0.76–1.03), whereas perceived public stigma was not
associated.
Conclusions
Personal attitudes towards mental illness or help-seeking are
associated with active help-seeking for mental problems.
Campaigns promoting help-seeking and fighting mental
illness-related stigma should target these personal attitudes
rather than broad public opinion.
Declaration of interest
None.
Copyright and usage
B The Royal College of Psychiatrists 2017.
The British Journal of Psychiatry (2017)
210, 261–268. doi: 10.1192/bjp.bp.116.189464
Review article
https://doi.org/10.1192/bjp.bp.116.189464 Published online by Cambridge University Press
supplement DS1).
49
Inclusion and exclusion criteria were
specified and documented in advance by F.S.L. and N.S. (see
online supplement DS2). We included only studies with general
population rather than clinical samples to reduce potential selection
bias towards active help-seeking. Quantitative, cross-sectional or
longitudinal surveys examining the impact of at least one of the
four stigma types on actual help-seeking were eligible. We
searched three electronic databases (PubMed, PsycINFO and
EMBASE) with no language restriction. The last search was carried
out on 10 July 2015. Potentially relevant studies published in peer-
reviewed journals since 1990 were identified using keywords
(adapted to the respective database) related to mental disorder
AND stigma AND help-seeking (see online supplement DS3 for
full search strategies and details of keywords). We also scrutinised
the reference lists of relevant papers,
21–27
and contacted expert
researchers for potential additional studies.
Study selection and data extraction
We screened the titles and abstracts of all studies that met the
search criteria and then consulted the full text to determine
eligibility. We revised the data extraction sheet during the
extraction process until it was applicable to all studies. Authors
N.S. and N.G. extracted data independently, with potential
disagreements resolved by discussion with F.S.L.. Authors of
eligible studies were contacted for additional information or
missing data, if necessary. We extracted the following information:
(a) publication details: author, year of publication, location and
time of survey, setting and design;
(b) source of study population: general population sample (GPS)
or subgroups of GPS such as students or military personnel
(further referred to as selective GPS samples), total number
of survey participants, number of participants used in
analyses, random selection and representativeness;
(c) stigma measure: scale/items, reliability of scale and classi-
fication into one of the five stigma types – four specific
stigmas, and ‘general stigma’ (GenS) for studies that did not
survey a distinct stigma but combined more than one type
into a single variable;
(d) help-seeking time-frame: help-seeking w ithin the past 12
months v. lifetime help-seeking;
(e) statistical method;
(f) results: effect size of association with corresponding
confidence interval or coefficient of association with
corresponding standard error and covariates.
If a study reported more than one stigma type, we extracted all
of them. We used estimates from the fully adjusted models. We
recorded the direction of the stigma measure (e.g. hig her scores
indicate more stigma) and its range, as well as the direction of
the association. Finally, we rated the quality of reporting according
to the Strengthening the Reporting of Observational Studies in
Epidemiology (STROBE) statement,
50
with a higher number of
reported items representing a higher quality score.
Statistical analysis
The odds ratio (OR) for stigma effect on help-seeking was the
main outcome. We calculated ORs and 95% confidence intervals
if only regression coefficients and standard errors (s.e.) were
provided. We combined ORs in random effect meta-analyses.
We conducted separate meta-analyses for each stigma measure
to detect their independent effect on help-seeking. ORs of studies
reporting lower levels of stigma increasing (rather than hig her
levels of stigma decreasing) the odds of help-seeking were
inversed.
51
Heterogeneity was assessed using the I
2
statistic; this
provides information about the percentage of total variation
across individual studies that cannot be explained by chance.
52
Values range from 0% to 100%, with hig her values showing an
increase in heterogeneity: 25%, 50% and 75% have been
commonly used to represent low, moderate and high heterogeneity,
respectively.
52
We additionally examined the heterogeneity using
t
2
statistics. Unlike I
2
, t
2
is not affected by the number of
participants included in the meta-analysis.
53
Its values range from
0 to infinity, with hig her values indicating higher heterogeneity.
Values of 0.04, 0.16 and 0.36 have been commonly used to
represent low, moderate and high heterogeneity, respectively.
54
We assessed bias of small study effects with funnel plots and
Egger’s test.
55
Subgroup analyses were pre-specified to investigate
whether effects of stigma on help-seeking depended on specific
study characteristics. We stratified analyses according to study
population (GPS v. selective GPS sample), time of help-seeking
(within the past 12 months v. lifetime), survey period (before
2006 v. 2006 and after), response rate (570% v. 570%), quality
of reporting (higher v. lower quality based on the median STROBE
checklist score, with studies scoring 25 or above deemed higher
quality) and setting (questionnaire v. interview). We defined
two stratifications post hoc according to healthcare systems –
private (USA) v. state-involved (other countries) – and study
design (cross-sectional v. prospective). Stratification was only
conducted if more than one study per group was found. All
statistical analyses were done in Stata version 14 (Stata Corporation,
College Station, Texas, USA).
Results
We identified 7968 papers in the initial search of databases and the
reference lists of previous reviews (Fig. 1).
21–27
After removing
1163 duplicates we screened the titles and abstracts of 6805
potentially eligible studies. We assessed the full text of 201 articles.
We contacted authors of nine studies for additional data, five of
whom responded and their findings were thus included. One of
the studies with missing data provided data for only one stigma
type,
56
and was therefore only partially included. Two were
excluded owing to missing data. One study used robust standard
errors (RSE), did not report CIs, and the authors were not able to
provide parametric standard errors or confidence intervals. This
study was excluded because the calculation of confidence intervals
from RSEs leads to different results from those when standard
errors are used. One study reported a lower CI limit equal to
the estimate;
64
we assumed it to be a rounding problem and with
lack of an author response used data ‘as is’. A final total of 27
studies were included in the meta-analyses.
16,33,56–80
General study characteristics
Altogether, the 27 studies included 31 677 participants aged 15
years or older. They included GPS (13 studies) or subsamples of
non-clinical GPS (14 studies). Four studies used a prospective
design. All studies but one, from Singapore,
75
were conducted
in Western societies (Europe, Australia or USA). Included studies
investigated at least one of the four types of stigma, but varied
greatly in their assessment (online supplement DS4). Thirteen
assessed PublicS, with six of them using the Perceived Devaluation
Discrimination (D-D) Scale,
32
or its adaptation.
81
Of the six
studies that investigated PersonS, two used a social distance scale
and two employed an adaptation of the D-D scale (‘most people’
replaced with ‘I’). Three of the five studies investigating SelfS used
a single-item assessment. Four of the 13 studies investigating
262
Schnyder et al
https://doi.org/10.1192/bjp.bp.116.189464 Published online by Cambridge University Press
Mental health stigma and help-seeking
HelpA used Fischer’s Attitudes Towards Seeking Professional
Psychological Help scale,
43
and two used a single item scale. Seven
studies used a non-specific general stigma measure (see online
Table DS1). All studies reported help-seeking from a formal,
professional source such as a psychiatrist, psychotherapist or
general practitioner. Only one study also investigated informal,
lay sources of help, such as family or a priest. To improve the
homogeneity of our outcome measure we only extracted data
for formal, professional sources. Twenty studies reported recent
help-seeking (within the past 12 months), seven reported lifetime
help-seeking and one study reported both.
64
From the latter we
extracted only data for lifetime help-seeking.
Influence of stigma type on help-seeking
Figure 2 shows the results of the five random effect meta-analyses
for each of the stigma types, as well as general stigma. Negative
HelpA (OR = 0.80, 95% CI 0.73–0.88) and higher PersonS
(OR = 0.82, 95% CI 0.69–0.98) were associated with less active
help-seeking for mental health problems. Higher SelfS
(OR = 0.88, 95% CI 0.76–1.03) showed an indication of less active
help-seeking, but the results were not statistically significant.
PublicS (OR = 0.97, 95% CI 0.93–1.02) and the unspecific GenS
(OR = 0.98, 95% CI 0.84–1.15) were not associated with active
help-seeking. There was substantial between-study heterogeneity
in each of the meta-analyses, with I
2
ranging from 58% for
PublicS to 91% for PersonS. Between-study variance t
2
,by
contrast, was low to moderate,
81,82
ranging from 0.003 for PublicS
to 0.044 for PersonS. Only HelpA showed evidence of small-study
bias (Egger’s test, P50.01; all other stigma measures P40.294; see
online figure DS1 for funnel plots and P values).
Subgroup analyses
The stratified meta-analyses for the most part did not demonstrate
any major influence of study characteristics (Fig. 3; online
supplement DS5). Associations between HelpA and help-seeking
were weakly influenced by type of study population, time of
help-seeking, setting, response rate, design and quality of
reporting. We found stronger negative associations in surveys with
random v. selective general population samples, recent v. lifetime
help-seeking, personal assessments v. questionnaires, higher v.
lower response rates, prospective v. cross-sectional design and
lower v. higher reporting quality. Associations between SelfS and
help-seeking were weakly influenced by study setting, survey
period and response rate. We found stronger negative associations
in surveys with personal assessments v. questionnaires, those
conducted before v. after 2006, and with higher v. lower response
rates. There was a small effect of year of study publication on the
association between GenS and help-seeking , with older studies
reporting slightly stronger effects. Associations between PersonS
and help-seeking were weakly influenced by study design, with
cross-sectional studies reporting negative associations whereas
prospective studies did not report significant associations.
Stratification by the country’s type of healthcare insurance did
not show any effect. Results of stratified analyses of PublicS and
PersonS were robust across all the investigated strata. A decline
in between-study heterogeneity was observed in some
stratification analyses. None of the stratification analyses could
fully explain the observed heterogeneity in all of the associations
between stigma types and help-seeking.
Discussion
Our results confirm the notion that stigma related to mental
illness or mental health services is directly associated with less
active help-seeking for mental problems in the general population.
The strength of association depends on the type of stigma, rather
than being the case for stigma in general. We found associations
between less active help-seeking and participants’ levels of HelpA
and PersonS. SelfS showed insignificant associations. PublicS and
unspecific GenS showed no association. These findings are in line
with social psychological studies demonstrating that attitudes
towards a behaviour are associated with engaging in the behaviour
itself in other situations.
84
Persons with pronounced PersonS
might try to avoid contact with the stigmatised group,
85–87
and
therefore refrain from help-seeking. PublicS and SelfS failed to
show significant associations, but both pointed to the expected
direction of more stigma predicting less active help-seeking. The
majority of studies surveying SelfS used a single item asking about
a person’s embarrassment when thinking about help-seeking for
his or her mental health problems. Even though embarrassment/
shame seems to be a barrier to help-seeking intentions,
88
it is unclear
whether this facet of SelfS can fully capture this stigma type.
89
Although a recent systematic review found a small association
between SelfS and help-seeking (intentions/recommendations
and active),
21
the influence of SelfS on active help-seeking in the
general population needs further exploration. To assess stigma
related to mental illness and its impact on help-seeking, future
studies using GenS might also consider assessing one of the more
specific stigma types.
Although the four stigma types revealed independent effects
on help-seeking, they are interrelated.
66,90–94
Self-stigma seems
263
Records identified through
database search
7347
Additional records identified
through other sources
621
7968 records
Records screened
6805
Full-text articles assessed
for eligibility
201
Studies included in quantitative
synthesis (meta-analysis)
27
Duplicates removed
1163
Records excluded
6604
174 full-text articles excluded:
90 no active help-seeking
as outcome
51 no regression analysis
16 stigma only mentioned
in discussion
11 qualitative study
2 no response from authors
for missing data
2 stigma measure does not fit
the definition
1 used robust standard error
1 intervention study
Fig. 1 PRISMA flowchart of selection of eligible studies, with
reasons for full-text exclusion.
https://doi.org/10.1192/bjp.bp.116.189464 Published online by Cambridge University Press
Schnyder et al
to arise from an individual’s own attitudes towards people with
mental illness, as well as from (perceived) public stigma.
66,89–91
Perceived public stigma, personal stigma and self-stigma seem to
predict attitudes towards help-seeking.
66,92–95
Furthermore, studies
have suggested that stigma is associated with a low perceived need
for help,
66,96,97
and a strong desire to handle the problem on one’s
own.
98
These two factors were proposed as important barriers in
considering delayed or no help-seeking.
97,98
Future studies might
consider them as additional moderators of active help-seeking and
in interaction with stigmatising attitudes.
66
It is crucial to
understand the complexity of various ty pes of stigma, their role
in help-seeking for mental health problems, and their direct
impact on mental problems such as suicidality,
99
in order to
develop efficient public campaigns promoting help-seeking.
Several anti-stigma and information campaigns aiming to
improve people’s knowledge about mental illness (mental health
264
Study
Help-seeking attitudes
Mojtabai et al (2002)
73
Thoits (2005)
70
Judd et al (2006)
68
Komiti et al (2006)
70
Rusch et al (2008)
76
Interian et al (2010)
65
Aromaa et al (2011)
58
Kim et al (2011)
69
Mojtabai & Crum (2013)
79
Vogt et al (2014)
80
Adler et al (2015)
16
Smith et al (2004)
77
Elhai et al (2008)
60
Total (I
2
= 88%, 95% CI 82–93, P = 0.001)
Personal stigma
Eisenberg et al (2009)
33
Interian et al (2010)
65
Aromaa et al (2011)
58
Downs & Eisenberg (2012)
59
Vogt et al (2014)
88
Jorm et al (2000)
67
Total (I
2
= 91%, 95% CI 84–95, P = 0.001)
Self-stigma
Thoits (2005)
79
Bambauer & Prigerson (2006)
37
Rusch et al (2008)
76
Nyunt et al (2009)
75
ten Have et al (2010)
78
Vogt et al (2014)
80
Jennings et al (2015)
66
Total (I
2
= 81%, 95% CI 82–91, P = 0.001)
Perceived public stigma
Bambauer & Prigerson (2006)
57
Judd et al (2006)
68
Komiti et al (2006)
70
Golberstein et al (2008)
62
Golberstein et al (2009)
63
Rusch et al (2008)
76
Eisenberg et al (2009)
33
Interian et al (2010)
65
Downs & Eisenberg (2012)
Green et al (2012)
64
Vogt et al (2014)
80
Adler et al (2015)
16
Jennings et al (2015)
66
Total (I
2
= 58%, 95% CI 23–77, P = 0.004)
General stigma
Bambauer & Prigerson (2006)
57
Nadeem et al (2007)
74
Menke & Flynn (2009)
71
Kim et al (2011)
69
Elnitsky et al (2013)
61
Mojtabai & Crum (2013)
72
Blais et al (2015)
56
Total (I
2
= 69%, 95% CI 30–88, P = 0.004)
n (analysed)
1792
1712
350
267
92
200
507
3380
195
601
160
393
279
5555
200
507
519
601
422
1712
135
92
1092
8796
601
95
135
350
267
302
726
92
5555
200
519
124
601
160
95
135
129
1013
3380
799
195
2025
OR (95% CI)
0.45 (0.31, 0.62)
0.68 (0.61, 0.77)
0.97 (0.93, 1.01)
0.94 (0.89, 1.00)
1.07 (0.94, 1.23)
0.60 (0.39, 0.92)
0.62 (0.54, 0.72)
0.63 (0.45, 0.88)
0.08 (0.01, 0.72)
0.90 (0.83, 0.98)
0.61 (0.39, 0.91)
0.82 (0.75, 0.90)
0.92 (0.87, 0.97)
0.80 (0.73, 0.88)
0.57 (0.51, 0.64)
1.15 (0.95, 1.41)
0.81 (0.73, 0.90)
0.73 (0.62, 0.85)
0.88 (0.82, 0.96)
0.98 (0.78, 1.22)
0.82 (0.69, 0.98)
0.87 (0.75, 1.02)
0.34 (0.15, 0.75)
1.07 (1.01, 1.14)
0.94 (0.24, 3.66)
0.76 (0.61, 0.96)
0.88 (0.82, 0.94)
1.15 (0.41, 3.25)
0.88 (0.76, 1.03)
0.81 (0.45, 1.45)
0.98 (0.92, 1.04)
0.96 (0.91, 1.02)
0.99 (0.91, 1.09)
1.01 (0.93, 1.09)
0.77 (0.66, 0.90)
1.02 (0.90, 1.14)
1.08 (0.93, 1.25)
1.19 (1.03, 1.36)
0.90 (0.90, 1.00)
0.96 (0.91, 1.01)
0.96 (0.65, 1.43)
1.16 (0.54, 2.50)
0.97 (0.93, 1.02)
0.72 (0.50, 1.04)
0.95 (0.61, 1.46)
1.03 (1.01, 1.05)
1.09 (0.85, 1.41)
1.58 (1.09, 2.30)
1.06 (0.28, 4.05)
0.80 (0.68, 0.93)
0.98 (0.84, 1.15)
% Weight
4.30
9.94
11.71
11.40
9.36
3.20
9.16
4.52
0.17
10.87
3.26
10.61
11.50
100.00
17.54
15.38
17.72
16.47
18.21
14.69
100.00
21.98
3.13
27.57
1.15
17.06
27.17
1.93
100.00
0.52
12.91
12.91
9.73
10.77
5.22
7.45
5.65
6.23
13.56
13.62
1.11
0.31
100.00
11.27
8.99
28.57
16.42
11.05
1.33
22.38
100.00
0.05 0.25 0.5 0.75 1 1.5 2
8
7
7
7
7
7
Fig. 2 Forest plot of the results of meta-analyses of five stigma types on active help-seeking.
Odds ratios (OR) and 95% confidence intervals (CI) of individuals studies and pooled estimates of separate random effects meta-analyses. OR51 indicates negative associations
between stigma or attitudes and help-seeking, i.e. higher levels of stigma are associated iwth less help-seeking. Estimates of between-study variance: t
2
= 0.018 for HelpA, t
2
= 0.044
for PersonS and t
2
= 0.023 for GenS. The study by ten Have et al (2010) estimated relative risk ratio; OR estimate was not reported and not available from study authors.
https://doi.org/10.1192/bjp.bp.116.189464 Published online by Cambridge University Press
Mental health stigma and help-seeking
literacy) and to reduce stigma associated with mental illness have
been conducted in recent years.
100
Whereas knowledge about the
causes and treatment of mental illness seemed to improve over
time and after campaigns,
101–104
reducing negative attitudes has
proved to be more difficult.
102
Only 7% of the world population
reported a belief that mental illness can be overcome,
105
and those
most reluctant to seek help perceived the lowest benefits in
engaging in this behaviour.
106
To promote help-seeking, findings
from these meta-analyses suggest that campaigns should address
negative personal attitudes by strengthening beliefs in the
treatability of mental illness. Advanced, biologically oriented
mental health literacy,
107
and activation of fear due to media
reports,
108
can increase the desire for social distance towards
people with mental illness. Therefore, the content of campaigns
should be chosen thoughtfully to avoid unintended effects.
109
Future studies
Subgroup analyses suggested that associations between stigma and
help-seeking can depend on certain study characteristics, in
particular response rate and assessment setting. Higher response
rates were generally associated with stronger negative effects of
stigma. As higher response rates can reduce a potential non-
responder bias,
110
they lead to more reliable results.
111
Consequently, reporting of response rates is crucial for assessing
the validity and reliability of research findings,
111
which should
be considered in future surveys. With regard to differences in
setting, face-to-face assessments were associated with stronger
negative effects than were self-reports by questionnaire. Since
the expression of stigmatising attitudes towards people with
mental illness or towards mental health services might be affected
by social desirability bias,
112
this is a surprising finding. Social
desirability should have a greater role in personal contact. Surveys
investigating social taboos (such as stigmatising attitudes) showed
increased levels of response accuracy when data were assessed
using self-administration (such as questionnaires), compared with
interviewer administration.
112
Questionnaires might therefore be
more suitable when researching stigma. In our analyses three
out of four studies with personal assessment reported high
response rates;
72,73,79
the fourth did not report a response rate.
65
Four studies using questionnaires reported low rates,
58,67,77,80
two reported none,
60,76
and only three reported high response
rates.
16,69,70
Inspection of single study effects indicates that across
these questionnaire studies, those with high response rates
reported a stronger negative association. Sampling bias associated
with lower response rates might therefore have a more crucial role
in detecting associations between stigma and active help-seeking
than the mode of assessment.
The association between HelpA and help-seeking was stronger
when recent rather than lifetime help-seeking was considered.
Furthermore, the association between HelpA and help-seeking
was stronger in prospective studies, whereas the association
between PersonS and help-seeking disappeared in prospective
studies. These results indicate the importance of a timely
265
HelpA PersonS SelfS PublicS GenS
Participants
Time of
help-seeking
Response
rate
Reporting
quality
Setting
Survey
time
Insurance
Design
Unstratified
GPS
Selective GPS
sample
Present
Lifetime
570%
570%
Missing
Lower
Higher
Personal
contact
Questionnaire
Before 2006
After 2006
Private
State involved
Cross-sectional
Prospective
0.7 1.0 1.5 0.7 1.0 1.5 0.7 1.0 1.5
OR (95% CI)
0.7 1.0 1.5 0.7 1.0 1.5
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
–
Fig. 3 Forest plot of the results of stratified meta-analyses of five stigma types on active help-seeking.
Pooled estimates (odds ratio; OR) and 95% confidence interval (CI) of each strata are reported. Dashed lines represent 95% CI of non-stratified analyses (top row). GenS, general
stigma; HelpA, help-seeking attitude; PersonS, personal stigma; PublicS, perceived public stigma; SelfS, self-stigma. OR51 indicates negative associations between stigma or
attitudes and help-seeking, i.e. higher levels of stigma are associated with less help-seeking.
https://doi.org/10.1192/bjp.bp.116.189464 Published online by Cambridge University Press