Assignment 1

Anxiety and its disorders as risk factors for suicidal thoughts and behaviors: A meta-analytic review

Kate H. Bentleya,*, Joseph C. Franklinb, Jessica D. Ribeirob,c, Evan M. Kleimanb, Kathryn R. Foxb, and Matthew K. Nockb

aCenter for Anxiety and Related Disorders, Boston University, USA

bDepartment of Psychology, Harvard University, USA

cMilitary Suicide Research Consortium, USA

Abstract

Suicidal thoughts and behaviors are highly prevalent public health problems with devastating

consequences. There is an urgent need to improve our understanding of risk factors for suicide to

identify effective intervention targets. The aim of this meta-analysis was to examine the

magnitude and clinical utility of anxiety and its disorders as risk factors for suicide ideation,

attempts, and deaths. We conducted a literature search through December 2014; of the 65 articles

meeting our inclusion criteria, we extracted 180 cases in which an anxiety-specific variable was

used to longitudinally predict a suicide-related outcome. Results indicated that anxiety is a

statistically significant, yet weak, predictor of suicide ideation (OR=1.49, 95% CI: 1.18, 1.88) and

attempts (OR=1.64, 95% CI: 1.47, 1.83), but not deaths (OR=1.01, 95% CI: 0.87, 1.18). The

strongest associations were observed for PTSD. Estimates were reduced after accounting for

publication bias, and diagnostic accuracy analyses indicated acceptable specificity but poor

sensitivity. Overall, the extant literature suggests that anxiety and its disorders, at least when these

constructs are measured in isolation and as trait-like constructs, are relatively weak predictors of

suicidal thoughts and behaviors over long follow-up periods. Implications for future research

priorities are discussed.

Keywords

Suicide; Anxiety; Risk factor; Meta-analysis

1. Introduction

Suicidal behavior is a leading cause of injury and death across the globe (Centers for

Disease Control and Prevention, 2011), with upwards of one million individuals who take

their own lives annually (World Health Organization, 2012). In the United States alone, over

40,000 people die by suicide each year. In addition to completed suicides, approximately 3%

of individuals make a suicide attempt during their lifetime (Borges et al., 2010; Nock et al.,

*Corresponding author at: Center for Anxiety and Related Disorders, Boston University, 648 Beacon Street, 6th Floor, Boston, MA 02215, USA. [email protected] (K.H. Bentley).

HHS Public Access Author manuscript Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

Published in final edited form as: Clin Psychol Rev. 2016 February ; 43: 30–46. doi:10.1016/j.cpr.2015.11.008.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

2008). It is also estimated that approximately 9% of people report experiencing serious

thoughts of suicide during their lives. Unfortunately, the U.S. suicide rate has risen steadily

over the past decade (Centers for Disease Control and Prevention, 2011) and is projected to

continue increasing over upcoming years (Mathers & Loncar, 2006).

One promising avenue toward reducing the global impact of suicide is to identify risk

factors that predict suicidal thoughts and behaviors. Risk factors are not to be confused with

correlates; whereas correlates represent concomitants or consequences of a phenomenon of

interest and can be identified through cross-sectional methods, establishing risk factors

necessitates longitudinal designs (Kraemer et al., 1997). For example, from a study showing

that individuals who attempt suicide are more likely to be diagnosed with an anxiety

disorder than those who do not attempt suicide, one may conclude that anxiety is a correlate

of suicidal behavior. However, to determine that anxiety functions as a risk factor for

suicidal behavior, it would need to be established that anxiety disorders precede and

heighten future risk for suicide attempts. Establishing risk factors for suicidality is essential

for a number of reasons, including improved understanding of underlying mechanisms,

identification of at-risk individuals, and development of evidence-based prevention and

treatment programs. Most previous studies have focused exclusively on correlates, which are

unlikely to be as informative for prediction and intervention purposes. Accordingly, the

primary goal of this study was to conduct a meta-analytic review of prospective studies that

have evaluated anxiety and its disorders as risk factors for suicidal thoughts and behaviors.

In a narrative review, it can be difficult to provide a comprehensive account of all relevant

studies, reconcile contradictory findings, account for methodological variations across the

literature, and ascertain overall magnitudes of risk factors under investigation. A meta-

analysis can overcome these limitations, and thus would be very helpful in summarizing

current knowledge about anxiety as a risk factor for suicidality.

There are several reasons why we chose to focus on anxiety in this meta-analysis. First,

anxiety and its disorders are listed as important risk factors for suicide by a number of

national organizations (e.g., American Association of Suicidology, 2015; American

Foundation for Suicide Prevention, 2015; National Suicide Prevention Lifeline, 2015).

Determining the strength of empirical evidence to support this information, which is widely

disseminated to clinicians, researchers, and the public, is necessary.

Second, anxiety is implicated in many prominent theories of suicide. For example, according

to Beck's cognitive model of suicide, once a suicide schema is activated, anxiety (and

agitation) can serve as an expression of attentional fixation on suicide, which interacts with

hopelessness to increase suicide risk (e.g., Wenzel & Beck, 2008; Wenzel, Brown, & Beck,

in press). Although anxiety is not explicitly addressed in Joiner's interpersonal theory of

suicide (Joiner, 2005; Van Orden et al., 2010), this model's emphasis on the fearsome nature

of suicidal behavior is consistent with evidence showing that acute anxious states (e.g.,

heightened arousal/agitation, severe panic attacks) are often present immediately prior to

lethal or near-lethal suicidal acts (e.g., Britton, Ilgen, Rudd, & Conner, 2012; Busch,

Fawcett, & Jacobs, 2003; Conrad et al., 2009; Fawcett et al., 1990; Hall, Platt, & Hall, 1999;

Ribeiro et al., 2015; Ribeiro, Silva, & Joiner, 2014). These findings align with Fawcett's

influential conceptualizations of anxiety/agitation as a determinant for acute suicide risk

Bentley et al. Page 2

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

(e.g., Fawcett, 2001; Fawcett, Busch, Jacobs, Kravitz, & Fogg, 1997), and expert clinical

consensus identifying agitation as a “warning sign” for suicide (e.g., Rudd et al., 2006).

Furthermore, Baumeister (1990) conceptualizes suicide as the ultimate escape from aversive

self-awareness and associated negative affect, which often includes anxiety. Along similar

lines, Riskind and colleagues have theorized that a specific cognitive risk factor for anxiety

(looming vulnerability), when coupled with hopelessness, enhances urges to escape

psychological pain and elevates risk for suicide (e.g., Rector, Kamkar, & Riskind, 2008;

Riskind, 1997; Riskind, Long, Williams, & White, 2000).

Third, anxiety and related disorders (broadly defined here as anxiety, obsessive–compulsive,

trauma and stressor-related, and somatic symptom disorders) have received theoretical and

empirical attention as potential risk factors for suicide. First and foremost, these disorders

are characterized by aversive, avoidant reactions to emotional experiences (Barlow, Sauer-

Zavala, Carl, Bullis, & Ellard, 2014). Suicidal thoughts and behaviors have similarly been

conceptualized as avoidant or escape-based responses to the experience of strong emotions

(Baumeister, 1990; Boergers, Spirito, & Donaldson, 1998; Briere, Hodges, & Godbout,

2010; Bryan, Rudd, & Wertenberger, 2013; Shneidman, 1993), highlighting the potential

functional similarities of these phenomena. Further, behavioral avoidance, a hallmark

feature of anxiety disorders, often results in significant isolation, reduced quality of life, and

impaired functioning (e.g., Massion, Warshaw, & Keller, 1993; Olatunji, Cisler, & Tolin,

2007), which confer additional risk to suicidality (Kanwar et al., 2013; Kaplan, McFarland,

Huguet, & Newsom, 2007). Other avoidance strategies (e.g., suppression) often used by

individuals with an anxiety disorder have also been shown to exacerbate vulnerability to

suicidal thoughts and behaviors (Amir, Kaplan, Efroni, & Kotler, 1999).

In addition to promising theoretical relevance of anxiety disorders to suicidality, empirical

investigations of this relationship generally show that anxiety disorders are independently

associated with suicidal thoughts and behaviors. Although several earlier studies indicated

that anxiety disorders are not related to suicidality (e.g., Hornig & McNally, 1995;

Warshaw, Massion, Peterson, Pratt, & Keller, 1995), converging empirical evidence

suggests that a broad range of anxiety and related disorders (including panic disorder,

posttraumatic stress disorder [PTSD], social anxiety disorder, and generalized anxiety

disorder [GAD]) function as statistically significant risk factors for suicidal thoughts and

behaviors (Boden, Fergusson, & Horwood, 2007; Bolton et al., 2008; Borges, Angst, Nock,

Ruscio, & Kessler, 2008; Gradus et al., 2010; Nepon, Belik, Bolton, & Sareen, 2010; Nock

et al., in press, 2009, 2010a; Noyes, 1991; Sareen, 2011; Sareen et al., 2005; Thibodeau,

Welch, Sareen, & Asmundson, 2013; Weissman, Klerman, Markowitz, & Ouellette, 1989;

Wilcox, Storr, & Breslau, 2009).

For example, Nock and colleagues (2010a) observed that anxiety disorders significantly

predicted future suicide attempts in a large, nationally representative sample (N = 9282),

when controlling for psychiatric comorbidity. Indeed, anxiety disorders produced larger

population attributable risk proportions (PARPs), which take into account the prevalence of

predictors and distribution of comorbidity, for suicide ideation and attempts than impulse-

control and substance use disorders. Among suicide ideators, anxiety disorders played a

larger role than mood disorders, impulse-control disorders, and substance use disorders in

Bentley et al. Page 3

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

accounting for subsequent suicide attempts. Findings from a cross-national study by Nock

and colleagues (2009) are similar, with anxiety serving as a more powerful predictor of

suicide attempts among ideators than mood, impulse-control, and substance use disorders. A

recent meta-analysis of 42 studies further supports the notion that anxiety disorders and

suicidal behavior are related (Kanwar et al., 2013); however, by including a combination of

cross-sectional and longitudinal studies, the conclusions that can be drawn from this

previous meta-analysis regarding anxiety (and specific disorders) as risk factors are

tempered.

In summary, researchers, theorists, and clinicians have evidenced a longstanding interest in

the role of anxiety and its disorders in suicide. A meta-analysis of longitudinal studies is

necessary to synthesize the available literature on anxiety as a risk factor for suicidal

thoughts and behaviors. Specifically, we aimed to: (1) estimate the predictive power (i.e.,

risk factor magnitude and accuracy) of anxiety as a risk factor for suicide-related outcomes,

(2) compare the strength of effect sizes across types of anxiety (e.g., disorders, symptoms),

and (3) examine potential moderators of the association between anxiety and suicide-related

outcomes, including sample type, sample age, and length of follow-up period. To our

knowledge, this represents the first effort to provide a comprehensive, quantitative review of

the predictive ability of anxiety and its disorders for suicidal thoughts and behaviors.

2. Method

This meta-analysis complies with the Preferred Reporting Items for Systematic Reviews and

Meta-Analyses statement (PRISMA; Moher, Liberati, Tetzlaff, & Altman, 2009).

2.1. Search strategy and selection

We conducted a comprehensive literature search of PsycINFO, PubMed, and Google

Scholar from database inception through December 2014 to identify studies eligible for the

current review. The prediction-related terms longitudinal, longitudinally, predicts,

prediction, prospective, prospectively, future, later, and follow-up were entered

simultaneously with the self-injury-related terms self-injury, suicidality, self-harm, suicide,

suicidal behavior, suicide attempt, suicide death, suicide plan, suicide thoughts, suicide

ideation, suicide gesture, suicide threat, self-mutilation, self-cutting, cutting, self-burning,

self-poisoning, deliberate self-harm, DSH, nonsuicidal self-injury, and NSSI. Given the

inconsistent and ambiguous terminology used to describe self-injury (with or without

suicidal intent) in the literature (Nock, 2010), we employed a wide range of self-injury-

related terms to reduce the likelihood of missing relevant studies. The reference sections of

identified articles were also searched for other relevant publications.

Fig. 1 presents a PRISMA diagram of the study selection process. A total of 2385 unique

publications were identified by initial literature searches; after reviewing each abstract, 642

studies were deemed eligible for further review. The authors then reviewed each full-text

study and identified 65 publications that met criteria for inclusion in the meta-analysis (see

Appendix A for references for articles included in meta-analyses).

Bentley et al. Page 4

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

2.2. Inclusion and exclusion criteria

To be included in this meta-analysis, studies had to first include a longitudinal analysis

predicting one or more of the following suicide-related outcomes: suicide ideation, suicide

plan, suicide gesture, suicide attempt, or suicide death. Correspondingly, we excluded

studies with outcomes that included nonsuicidal self-injury (NSSI) only. We also excluded

studies that lumped nonsuicidal and suicidal behavior together into a single outcome (e.g.,

“deliberate self-harm”) because including these combined variables would render it difficult

to draw strong conclusions about the association between anxiety and suicidal outcomes

specifically. Second, included studies had to use one or more of the following anxiety-

specific predictors: any individual anxiety diagnosis (i.e., agoraphobia, anxiety disorder not

otherwise specified, GAD, panic disorder, social anxiety disorder, specific phobia) or

anxiety diagnosis grouping (e.g., “any anxiety disorder” as defined by study authors), any

obsessive–compulsive or related disorder (e.g., OCD), any trauma or stressor-related

disorder (e.g., adjustment disorder, PTSD), any somatic symptom disorder, symptoms of

anxiety (diagnosis-specific and general; e.g., scores on self-report and clinician-rated

measures such as the Beck Anxiety Inventory [BAI; Beck, Epstein, Brown, & Steer, 1988],

Hamilton Anxiety Rating Scale [HARS; Hamilton, 1959], or Yale-Brown Obsessive

Compulsive Scale [Y-BOCS; Goodman et al., 1989], panic attacks). Correspondingly, we

excluded studies only examining constructs or symptoms broadly related but not specific to

anxiety (e.g., neuroticism, negative affect, emotion dysregulation). We also excluded studies

combining anxiety and mood disorders into a single variable (e.g., internalizing problems,

“mood or anxiety diagnosis”) because including these overlapping predictors would

undermine our conclusions regarding the specific effect of anxiety on suicidal thoughts and

behaviors. Finally, studies had to provide sufficient statistical information to conduct the

present analyses, appear in a publication, and present original data not already reported in

another study.

2.3. Data abstraction

Authors systematically extracted relevant information from each article included in the

present meta-analysis using a predefined coding strategy. Recorded information included:

(a) year of study publication, (b) sample age (i.e., adult, adolescent, mixed), (c) sample size,

(d) sample type (i.e., general, clinical, or history of prior SITBs), (e) length of study follow-

up (in months), (f) statistical information for each unique prediction case (i.e., any instance

using an anxiety-specific variable to longitudinally predict a suicide-related outcome), and

(g) whether the prediction case was hypothesized to be a risk or protective factor (e.g., no

psychopathology, reduction in anxiety). Whenever possible, we used the most unadjusted

estimate(s) or extracted raw diagnostic data for our meta-analytic calculations; of note, only

10% of cases included our analyses were coded as adjusted (e.g., controlling for related

variables such as age, gender, etc.).

2.4. Data analysis

Our outcomes of interest were suicidal thoughts and behaviors (i.e., suicide ideation, suicide

attempt, suicide death); due to the fact that only one study reported on suicide gestures and

plans (Borges et al., 2008), we did not conduct any analyses specific to these outcomes. Our

Bentley et al. Page 5

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

predictors of interest were anxiety-specific constructs. From each study, we extracted

relevant statistics (e.g., zero-order correlation, 2 × 2 contingency table, odds ratio, hazard

ratio, independent group means) for each unique predictor case. Protective factor cases (i.e.,

when a variable could be reasonably assumed to correspond to a reduction in risk for future

suicidal behavior, such as absence of PTSD) and redundant cases (e.g., a non-unique case

reported in two studies using the same sample) were removed. In addition, when one

publication reported cases using the same predictor and outcome at multiple time points, we

selected only the longest time point to reduce data dependency. Accordingly, we removed 6

cases, which resulted in a total of 180 unique prediction cases used in this review.

Prior to conducting our meta-analytic analyses, all relevant study statistics were converted to

ORs and their 95% confidence interval (CI) in the program Comprehensive Meta-Analysis

(CMA, Version 2; Borenstein, Hedges, Higgins, & Rothstein, 2005). Cases that reported

hazard ratio (HRs) and lacked the raw data necessary to be converted to ORs were analyzed

separately. We used random-effects models (Field, 2001; Hedges & Olkin, 1985; Hedges &

Vevea, 1998) for all calculations. These models assume that reported effect sizes across

studies represent a random sampling of studies distributed around a true effect size for the

population whereas the assumption of fixed effects models is that included studies reflect the

true population effect size. Thus, we chose random-effects models to take into account both

between- and within-study heterogeneity across the publications included in our meta-

analysis.

First, we pooled all ORs from included studies to estimate overall effect sizes for any

anxiety construct (i.e., any type of anxiety diagnosis and symptoms) predicting suicide

ideation, attempt, and death. Then, we estimated effect sizes for anxiety symptoms (e.g., total

scores on diagnosis-specific or general anxiety measures, panic attacks), excluding anxiety

diagnoses, on the same suicide-related outcomes. Next, we estimated effect sizes for any

type of anxiety diagnosis (i.e., adjustment disorder, agoraphobia, anxiety disorder not

otherwise specified, GAD, OCD, panic disorder, PTSD, social anxiety disorder, somatic

symptom disorder, specific phobia, “any anxiety diagnosis” as defined by the study authors)

predicting suicide-related outcomes. Finally, we conducted diagnosis-specific analyses for

the following individual diagnoses on suicide ideation, attempt, and death: adjustment

disorder, agoraphobia, GAD, OCD, panic disorder, PTSD, social anxiety disorder, somatic

symptom disorder, and specific phobia. Diagnosis-specific analyses were not conducted for

anxiety disorder not otherwise specified because there was only one case using this

particular diagnosis. A total of 20 cases used HRs only, which all predicted either attempt or

death. We pooled these cases separately to estimate overall effect sizes for any anxiety

construct predicting suicide attempt and then death.

Several tests were used to assess for potential publication bias in our meta-analysis. First,

Orwin's fail-safe N was used to estimate the number of missing studies with ORs of 1.00

necessary to lower the magnitude of observed overall effect sizes to a negligible value; for

these analyses, we set the alternative OR to 1.01. We also employed Egger's test of the

intercept, which uses precision (i.e., inverse of the standard error) to predict the standardized

effect (i.e., effect size divided by the standard error). In Egger's equation, the slope of the

regression line indicates the size of the effect and the intercept indicates bias. Last, we used

Bentley et al. Page 6

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

a funnel plot, which depicts the standard error of each included study on the vertical axis

relative to the effect size magnitude on the horizontal axis. When studies are symmetrically

dispersed around the combined mean effect size, it is likely the meta-analysis captured all

relevant studies, whereas an asymmetrical distribution of studies indicates publication bias.

Duval and Tweedie's Trim and Fill test was used to quantify the nature of publication bias.

Specifically, this analysis identifies the number of studies missing due to publication bias,

estimates the effects of these missing studies, adds the unpublished studies to the analysis,

and re-estimates the overall effect size estimate. Should the initial and re-estimated ORs

significantly differ, one can conclude that the effect observed in the meta-analysis is biased.

The program MetaDiSc Version 1.4 (Zamora, Muriel, & Abraira, 2014) and random-effects

models were used to conduct diagnostic accuracy analyses in order to provide detail in

addition to overall ORs regarding patterns of included data. Our diagnostic accuracy

statistics involved generating sensitivity (true positive rate; e.g., of all individuals who

attempt suicide, the percentage that we would predict will make an attempt based on the

presence of an anxiety disorder) and specificity (true negative rate; e.g., of all individuals

who do not attempt suicide, the percentage that we would predict will not make an attempt

based on the absence of an anxiety disorder) values. We also generated receiver operating

characteristic (ROC) curves, which plots the sensitivity against the false positive rate (1 —

specificity) for each prediction case; areas under the curve (AUCs) and corresponding

standard errors were then calculated and used as our primary indicator of diagnostic

accuracy. We adhered to the following guidelines in interpreting AUCs: >.90 suggesting

excellent prediction, .80–.89 good prediction, .70–.79 fair prediction, .61–.69 poor

prediction, and .50–.59 extremely poor prediction (Šimundić, 2008).

To be included in diagnostic accuracy analyses, prediction cases had to provide raw data that

could be converted into a 2 × 2 table with cells that indicate true positive, false positive, true

negative, and false negative events. Whenever possible, raw data was obtained to maximize

the number of prediction cases included in these analyses. Given the limited number of

prediction cases for suicide ideation that provided sufficient information to run these

analyses, diagnostic accuracy estimates were computed for outcomes of suicide attempt and

death only. For suicide attempt, 39 prediction cases (spanning all anxiety constructs) were

included, and for death, 26 cases (also spanning all anxiety constructs) were included in

diagnostic accuracy analyses.

We used the I2 statistic to examine heterogeneity within the meta-analysis, thereby

justifying our use of random-effects models, which, as previously noted, account for

important methodological variations across studies. An I2 statistic >50 indicates high

variance in observed effect sizes (Higgins, Thompson, Deeks, & Altman, 2003) and

suggests that characteristics of individual studies should be examined as potential

moderators of the observed effects of anxiety on suicide-related outcomes. The following

variables were tested as potential moderators: sample type (i.e., general, clinical, history of

self-injurious thoughts or behaviors [SITBs]), sample age (i.e., adult, adolescent, mixed),

and length of follow-up period. For categorical variables (sample type and sample age), we

examined the magnitude of effect size estimates for any anxiety construct predicting suicide

ideation, attempt, and death at each level of the potential moderator. We treated length of

Bentley et al. Page 7

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

follow-up period as a continuous variable, and employed unrestricted maximum likelihood

meta-regression using a random effects model to estimate the effect of length of follow-up

period on the predictive ability of anxiety. For these analyses, we used R2 as an indicator of

moderator strength.

3. Results

3.1. Description of included studies

Characteristics of studies are presented in Table 1. Years of publication spanned 1982 to

2015. The 65 included articles yielded a total of 69 samples, as four publications reported

data separately for males and females (i.e., Berglund & Nilsson, 1987; Lewinsohn et al.,

2001; Reinherz et al.,1995; Tidemalm et al., 2014). Across all samples, there were a total of

852,159 participants. The median number of participants per study was 249 (M = 12,350, SD

= 44,712, range = 26–309,861). Just over half of samples (52.2%) were classified as clinical

(e.g., psychiatric patients), whereas 26.1% were drawn from the general community and

21.7% included only individuals with a history of prior SITBs. One study (Borges et al.,

2008) used a combination of community-based participants and those with a SITB history.

The majority of studies (71.0%) utilized adult samples; 21.7% of studies were conducted in

adolescent-only samples and only 5 studies (7.2%) used mixed samples. Over half of

included studies (55.1%) utilized samples from countries outside North America, most

commonly in Europe.

Across all included studies, the median follow-up interval was 60 months (M = 113.1, SD =

126.2, range = 1–708 months). The most common follow-up period length was also 60

months (9 studies). Only 4 included studies had a follow-up interval shorter than 12 months,

whereas just under one-third (30.4%) of studies had a follow-up period of 12 years or

longer. With regard to predictors, the vast majority of studies (85.5%) reported at least one

case in which an anxiety diagnosis (either a specific diagnosis [e.g., GAD, PTSD] or

diagnosis grouping [e.g., any anxiety disorder]) was used to predict a suicide-related

outcome. Of these, 11 studies also reported one or more additional case(s) using anxiety

symptoms to predict a suicide-related outcome. Only 10 studies only included symptom-

level anxiety predictors. For suicide-related outcomes, most studies (60.9%) reported at least

one usable suicide attempt prediction case. Death was the second most common outcome

examined, with 37.7% of studies presenting one or more cases in which an anxiety predictor

was used to predict suicide death. Only 13.0% of studies included a prediction case for

ideation. There was an average of 2.61 usable prediction cases per study.

3.2. Overall anxiety effect sizes and publication bias

3.2.1. Suicide ideation—Overall effect sizes for any anxiety construct predicting outcomes of suicide ideation, suicide attempt, and suicide death are presented in Table 2;

these (and other overall OR) findings are also presented in Fig. 2. In total, 26 prediction

cases included suicide ideation as an outcome. The overall OR for any anxiety construct

predicting suicide ideation was statistically significant, but small. Heterogeneity among

these cases was high, suggesting significant between-study variance. We employed multiple

analyses to test for publication bias among suicide ideation prediction cases. Orwin's fail-

Bentley et al. Page 8

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

safe N value suggested that 884 cases with a mean OR of 1.00 would be necessary to lower

the overall mean OR for any anxiety construct below our predefined threshold of 1.01. In

addition, Egger's test demonstrated an intercept (B0) of 0.91 (95% CI: −2.02, 3.83), t(24) =

0.64, p = 0.27, which provides no clear indication of publication bias. However, the funnel

plot appeared asymmetrical, with a disproportionate number of studies falling on the right

side of the mean effect (see Fig. 3). Duval and Tweedie's trim and fill analysis also detected

9 missing cases, and indicated that if these were included, the overall weighted OR would be

slightly smaller (OR =1.24, 95% CI: 1.01, 1.53).

As previously noted, there were no HR cases for suicide ideation. We also did not examine

diagnostic accuracy for suicide ideation due to an insufficient number of cases with the

information required for these analyses.

3.2.2. Suicide attempt—A total of 80 prediction cases that used (or could be converted into) an OR included suicide attempt as an outcome. The overall effect size estimate for any

anxiety construct also reached the level of statistical significance, but was small (Table 2,

Fig. 2). Similar to suicide ideation, heterogeneity among suicide attempt prediction cases

was high. With regard to publication bias, Orwin's fail-safe N indicated that 838 cases with

an OR of 1.00 would be needed to bring the combined weighted OR for suicide ideation

below our threshold of 1.01. Egger's regression test produced an intercept (B0) of 1.13 (95%

CI: 0.66, 1.59), t(78) = 4.83, p < .0001, indicating that the less precise studies reported larger

effect sizes, thereby suggesting significant publication bias. The funnel plot also appeared

asymmetrical, with more studies falling to the right of the mean (see Fig. 3). Duval and

Tweedie's trim and fill test indicated that 21 prediction cases were missing from our meta-

analysis, and should these studies have been included, the combined mean OR for any

anxiety construct predicting suicide attempt would be reduced to 1.20 (95% CI: 1.07, 1.34).

There also were 10 HR cases included in the meta-analysis that used suicide attempt as an

outcome. For any anxiety construct predicting suicide attempt, the overall weighted mean

HR did not reach statistical significance; these cases also evidenced high heterogeneity

(Table 2).

As previously noted, there were 39 suicide attempt prediction cases with sufficient

information to conduct diagnostic accuracy analyses. Overall, the accuracy of anxiety in

prediction suicide attempt was poor, indicating that although anxiety may confer risk for

attempt, it is only slightly better than chance. Specifically, the AUC was .56 (SE = .02),

indicating extremely poor prediction (see Fig. 4). Further, sensitivity (i.e., true positive rate)

was very low at .12 (95% CI: .11, .13, p < .001). Although specificity (i.e., true negative

rate) was acceptable at .88 (95% CI: .87, .88, p < .001), this is likely due to a rare outcome

(suicide attempt) crossed with a relatively rare predictor (e.g., anxiety disorder).

3.2.3. Suicide death—There were 39 prediction cases with suicide death as an outcome included in OR analyses. The overall weighted mean OR for any anxiety construct did not

reach statistical significance (Table 2, Fig. 2). Heterogeneity among these cases was lower

than suicide ideation and attempt cases; indeed, the heterogeneity statistic suggests that less

than half of variance was accounted for by between-study variance. Regarding tests for

Bentley et al. Page 9

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

publication bias, Orwin's fail-safe N suggested that 14 cases with mean ORs of 1.00 would

be needed to bring the overall estimated OR for suicide death under our predefined value; of

note, the overall effect size was already at our predefined value for a trivial magnitude.

Egger's test resulted in an intercept (B0) of —0.21 (95% CI: -0.67, 0.24), t(37) = 0.95, p =

0.18, which is not indicative of substantial publication bias. The funnel plot for suicide death

cases appeared asymmetrical (see Fig. 3), with more studies on the left side of the mean, but

a Duval and Tweedie's trim and fill test indicated that no prediction cases were missing.

A total of 10 HR cases included suicide death as an outcome. The overall weighted mean

HR for any anxiety construct predicting suicide death was statistically significant, yet small

(Table 2). There was extremely high heterogeneity among these cases, suggesting

substantial between-study variance.

Diagnostic accuracy analyses were conducted using the 26 suicide death cases that provided

the necessary raw data. Similar to suicide attempt, specificity (.75; 95% CI: .75, .75, p < .

001) was much higher than sensitivity, which was very poor (.05; 95% CI: .04, .05, p < .

001). Again, this specificity value was likely an artifact of a rare outcome (suicide death)

crossed with a relatively rare predictor (e.g., anxiety disorder). Further, the AUC did not

reach statistical significance (.42, SE = .02; see Fig. 4), suggesting extremely poor overall

prediction.

3.3. Anxiety symptoms effect sizes

3.3.1. Suicide ideation—Results from analyses corresponding to any anxiety symptoms (i.e., excluding anxiety and related diagnoses) predicting suicide ideation, attempt, and death

are presented in Table 3 and Fig. 2. Analyses showed a small, but statistically significant

overall OR. Heterogeneity among these cases was relatively low.

3.3.2. Suicide attempt—The overall OR for anxiety symptoms predicting suicide attempt was small, but reached statistical significance, with high between-cases variance (Table 3,

Fig. 2).

3.3.3. Suicide death—The estimated overall OR for anxiety symptom cases predicting suicide death did not reach the level of statistical significance. There was low heterogeneity

among these cases (Table 3, Fig. 2).

3.4. Anxiety diagnosis effect sizes

3.4.1. Suicide ideation—Findings from analyses of anxiety and related diagnoses predicting suicide ideation, suicide attempt, and suicide death are presented in Table 4 and

Fig. 2. First, the combined effect of any type of anxiety diagnosis significantly predicted

suicide ideation; there was also high heterogeneity among cases included in this analysis.

Diagnosis-specific analyses indicated that the strongest predictor of suicide ideation was

PTSD; however, one PTSD case with a notably large magnitude (OR = 4.21) may have been

partially driving these findings, as no other suicide ideation prediction cases exceeded an

OR of 2.33. Diagnoses of GAD, specific phobia, and social anxiety disorder were also

significant predictors of suicide ideation. Corresponding I2 statistics also suggest substantial

Bentley et al. Page 10

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

heterogeneity among PTSD and GAD cases, but not specific phobia or social anxiety

disorder cases (Table 4).Of note, only 4 suicide ideation prediction cases were used to

generate overall ORs for PTSD, specific phobia, and social anxiety disorder, and only 2

cases were used for GAD, thereby limiting the conclusions we can draw from this set of

analyses. Overall effect size estimates were not computed for adjustment disorder,

agoraphobia, OCD, panic disorder, or somatic symptom disorder because there was no more

than one relevant case for each of these diagnoses.

3.4.2. Suicide attempt—The combined effect of any type of anxiety diagnosis on suicide attempt was also statistically significant, yet relatively small in magnitude (Table 4, Fig. 2).

High heterogeneity among these cases was also observed. Diagnosis-specific analyses

indicated that PTSD, panic disorder, and social anxiety disorder were significant predictors

of subsequent suicide attempt; the strongest predictor of suicide attempt was PTSD followed

by panic disorder. Of note, two prediction cases with ORs above 4.0 may have been driving

the relatively high estimate for panic disorder, as no other cases included in any diagnosis-

specific analysis for suicide attempt exceeded 3.3. Agoraphobia, GAD, OCD, somatic

symptom disorder, and specific phobia did not reach the level of statistical significance.

Adjustment disorder was a significant predictor of fewer suicide attempts. It is again

important to consider the relatively small number of cases included in diagnosis-specific

analyses for adjustment disorder, agoraphobia, OCD, somatic symptom disorder, and

specific phobia (ranging from 2 to 5). Among the diagnosis-specific cases that produced

statistically significant effect sizes, substantial heterogeneity was observed for PTSD,

whereas relevant cases for panic disorder, social anxiety disorder, and adjustment disorder

evidenced relatively low heterogeneity.

3.4.3. Suicide death—For any type of anxiety diagnosis predicting suicide death, the overall OR did not reach the level of statistical significance, and heterogeneity among these

cases was present (Table 4, Fig. 2). Diagnosis-specific overall ORs for panic disorder and

OCD were weak and nonsignificant, whereas adjustment disorder was shown to be a

statistically significant predictor of fewer suicide deaths. I2 statistics also suggested

substantial heterogeneity was present among cases included in the overall “any anxiety

diagnosis” analysis, but relatively low heterogeneity across cases included in diagnosis-

specific analyses. We did not generate weighted suicide death ORs for agoraphobia, GAD,

PTSD, social anxiety disorder, somatic symptom disorder, or specific phobia due to the fact

that there was no more than one relevant case for each diagnosis.

3.5. Moderation analyses

We examined three potential moderators of the relationship between anxiety and suicide

ideation, attempt, and death: sample type (i.e., general, clinical, history of prior SITBs),

sample age (i.e., adult, adolescent, mixed), and length of follow-up period.

3.5.1. Sample type—Sample type did not moderate the relationship between any anxiety construct and suicide ideation or suicide death. However, suicide attempt prediction cases

drawn from studies using general samples generated a significantly stronger overall effect

size (OR = 2.98, 95% CI: 2.37, 3.75, z = 9.35, p < .001) than clinical populations (OR =

Bentley et al. Page 11

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

1.55, 95% CI: 1.38, 1.75, z = 7.37, p < .001) and individuals with a history of SITBs (OR =

1.16, 95% CI: 0.85, 1.59, z = 0.94, p = .35).

3.5.2. Sample age—The age of the study sample also did not affect the predictive ability of anxiety for suicide ideation or death. Similar to sample type, however, there was a

moderation effect of sample age on the relationship between any anxiety construct and

suicide attempt. Specifically, weighted mean effect sizes from studies using samples

comprised of all adults (OR = 1.79, 95% CI: 1.56, 2.06, z = 8.24, p < .001) were

significantly stronger than those using mixed samples (OR = 0.83, 95% CI: 0.62, 1.11, z =

−1.23, p = .22). Of note, there were no significant differences between adult-only or mixed

samples in comparison to adolescent-only samples (OR = .59, 95% CI: 0.99, 2.56, z = 1.92,

p = .06).

3.5.3. Length of follow-up period—Results from maximum likelihood meta-regression analyses indicated that length of follow-up period significantly moderated the relationship

between anxiety and both suicide ideation and attempt (see Fig. 5), but not suicide death.

For suicide ideation, there was a small (but steady) decline in magnitude of combined ORs

steadily as the length of follow-up period increased (b = −0.005, SE = 0.00, 95% CI: −0.01,

−0.00, z = −2.91, p < .01), with follow-up length explaining 32% of the between-study

variance. The same trend was observed for suicide attempt (b = −0.004, SE = 0.00, 95% CI:

−0.01, −0.00, z = −2.81, p < .01), with follow-up length explaining 5% of between-study

variance. Together, these findings indicate that with longer follow-up periods, the predictive

ability of anxiety disorders and symptoms for suicide ideation and attempt weakens slightly.

4. Discussion

Suicidal thoughts and behaviors are public health problems that affect millions across the

globe. Despite receiving increased empirical attention over the past 50 years, the rates of

suicidal thoughts and behaviors have not abated. Identifying powerful and accurate risk

factors is one promising means toward effective prevention and treatment, and thus eventual

reduction, of these phenomena. Anxiety is one potential risk factor for suicidality that has

been of particular interest to date. For one, anxiety is listed as a top risk factor for suicide by

leading national organizations (e.g., AAS, AFSP, National Suicide Prevention Lifeline).

Prominent theories of suicide (e.g., Baumeister, 1990; Joiner, 2005; Wenzel & Beck, 2008)

also either directly address anxiety and/or are consistent with findings implicating the role of

acute anxiety/agitation in suicidal behavior. In addition, converging evidence from large-

scale longitudinal studies suggests that anxiety disorders significantly predict suicidal

ideation and behaviors, even when controlling for comorbidity (e.g., Boden et al., 2007;

Nock et al., 2009, 2010; Sareen et al., 2005). As such, the aim of this meta-analysis was to

synthesize the existing knowledge of anxiety as a risk factor for suicidal thoughts and

behaviors. Analyses indicated that anxiety and its disorders are statistically significant

predictors of suicide ideation and attempts, but not deaths.1 The strength of observed

relationships was weak across all categories examined.

1This refers to findings from weighted mean OR analyses, which were based on 80 and 39 cases (respectively), given that HR analyses for attempt and death were based on only 10 prediction cases each, and thus less reliable.

Bentley et al. Page 12

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Although overall, anxiety and its disorders showed statistically significant prospective

associations with suicide ideation and attempts, these findings do not translate into high

clinical utility. Across all anxiety-related predictors and suicide-related outcomes, combined

effect sizes were small, and no one predictor stood out as being significantly stronger than

the rest. Specifically, the largest combined OR just exceeded 2.0, and most effect sizes that

reached statistical significance clustered around 1.5. Given that the absolute risk of

experiencing suicidal thoughts and behaviors (and especially attempts and deaths) is low,

these estimates do not translate into clinically practical information. Consider an example

using PTSD, the diagnosis that produced the largest overall effect on suicide attempts in our

meta-analysis. Recent data suggest that in the United States, 0.4% of adults make a suicide

attempt in a given year (Borges et al., 2006). Even when these odds are multiplied by 2.25

for individuals with PTSD, as our meta-analysis would suggest, the resultant odds are still

close to zero. Determining that an individual has a 0.9% probability of attempting suicide in

the next year based on a PTSD diagnosis (as opposed to a 0.4% probability for individuals

without PTSD) does not provide meaningful information for clinicians facing critical

decisions about suicide risk. This pattern of low clinical significance also holds for

adjustment disorder, the only diagnosis evidencing a statistically significant protective effect

against suicidal behavior (0.1% versus 0.4% probability of attempting suicide for individuals

with and without this diagnosis, respectively). The limited clinical utility of our findings is

even more striking when one considers the fact that clinicians are usually tasked with

determining who is at risk for suicidal behavior over the next few hours or days, rather than

months or years. In short, anxiety and its disorders, at least as these constructs have been

studied to date, are unlikely to serve as powerful “real-world” indicators of risk for suicidal

thoughts and behaviors.

There are several potential explanations for our findings that anxiety (diagnoses or

symptoms) does not significantly predict suicide deaths. The interpersonal theory of suicide

(Joiner, 2005; Van Orden et al., 2010) posits that for lethal suicidal behavior to occur,

individuals must develop high levels of acquired capability through repeated experiences

with painful and provocative events (e.g., Ribeiro et al., 2014). Indeed, some anxiety

disorders examined here (e.g., agoraphobia, GAD) may be associated with significant fear

and avoidance of the experiences required to develop this suicidal capability and more

frightening, lethal suicidal behavior (e.g., Silva, Ribeiro, & Joiner, 2015). Thus, consistent

with Joiner's model, hallmark features of anxiety disorders may protect against the transition

from ideation (and even non-lethal suicide attempts) to death by suicide. Another related

explanation is that the constructs of anxiety captured in this review are not as relevant to

suicide deaths as specific, more acute anxiety-related symptoms. Given findings indicating

that severe, heightened arousal is often present immediately prior to suicide (e.g., Busch et

al., 2003; Robins, 1981; Way, Miraglia, Sawyer, Beer, & Eddy, 2005), and facilitates lethal

suicidal behavior among at-risk individuals (e.g., Ribeiro et al., 2015), there is likely an

important distinction between the role of anxiety disorders and general affective states of

anxiety (e.g., trait-like anxiety, future-oriented worry) versus acute symptoms of anxiety and

agitation (e.g., excessive motor and mental arousal) in predicting suicide deaths specifically.

Bentley et al. Page 13

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

We also observed several significant moderation effects in this meta-analysis. First, general

samples generated larger effect sizes for prediction of suicide attempts than those drawn

from clinical settings or with a history of SITBs. This is likely due to the fact that in samples

of psychiatric patients and/or with a SITB history, confounding factors (e.g.,

psychopathology severity, prior suicidal behavior) were more likely to be controlled for (as

a function of the sample selection) in comparisons between individuals with and without

anxiety. In contrast, studies of community samples did not account for these related

variables, and thus were likely to produce stronger effects. We also found that adult samples

produced larger effects than samples combining adolescents and adults; however, the

conclusions that can be drawn here are limited given the very small number of cases

included in the mixed and adolescent-only analyses, and overlapping CIs for effect sizes

produced in adolescent- and adult-only samples. Finally, studies with shorter follow-up

periods demonstrated larger effects than longer follow-up intervals. This suggests that

anxiety and its disorders may be stronger predictors of suicidal thoughts and behaviors over

shorter amounts of time.

4.1. Limitations and future directions

Overall, the current evidence suggests that anxiety and its disorders do not substantially

increase risk for future suicidal thoughts and behaviors; however, this conclusion only

applies to the relationship between anxiety and suicide as it has traditionally been examined.

The present meta-analysis identified several methodological limitations of existing research

on anxiety as a risk factor for suicidality. First, the vast majority of studies have measured

anxiety in terms of specific diagnoses or broad, trait-like symptoms, rather than dynamic,

acute constructs. Second, the existing literature has generally analyzed anxiety in isolation,

rather than in conjunction with other known risk factors. Finally, studies have tended to

employ long follow-up periods, rather than the short follow-up periods more relevant to

“real-world” assessment of suicide risk. Despite the fact that almost all previous studies

have used these methods, few would expect any variable (anxiety-related or otherwise),

when measured as a trait-like, static construct and studied in isolation, to accurately predict

suicidal thoughts and behaviors over many years.

Thus, the possibility remains that anxiety is a very important risk factor for suicide, but it

has not yet been studied under the conditions necessary to detect this relationship. Future

studies must focus on directly testing prominent theories and models of suicide that

conceptualize suicidal thoughts and behaviors as resulting from complex combinations of

risk factors (distal and proximal, trait-like and unstable) interacting to heighten risk over

short periods of time. For instance, rapid elevations in specific, acute anxiety-related

constructs (e.g., heightened arousal/agitation, sudden onset of severe panic-like symptoms)

may substantially increase risk for suicidal behavior when such variables are combined with

other potentially important factors (e.g., acquired capability, lack of social support, prior

engagement in SITBs, triggering events, psychiatric comorbidities). Even in these specific

contexts, these constructs may only substantially enhance risk for short periods of time.

In order to better approximate these hypotheses, a number of methodological changes are

needed in future research. For one, studies must employ more dynamic (including

Bentley et al. Page 14

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

physiological, implicit, and/or behavioral) measures of specific anxiety-related constructs

hypothesized to be most pertinent to prediction of suicidal behavior as well as examine

potential interactive effects with other known and more novel risk factors (e.g., implicit

associations with suicide; Nock et al., 2010). It is also recommended that in future studies

aiming to isolate the unique effects of a single risk factor (or specific combination of risk

factors), researchers consistently adjust for factors that may influence the magnitude of their

findings (e.g., gender, depression severity). Future investigations must also assess both

putative risk factors and suicide-related outcomes much more frequently (rather than letting

many months or years elapse between time points), and over much shorter periods of time

(e.g., minutes, hours, days). It is also encouraged that researchers examine more fine-grained

features of suicidal behavior (e.g., attempt frequency, lethality, intent) in order to refine our

understanding of the longitudinal relationships between hypothesized risk factors and

specific outcomes, including characteristics of those outcomes. Finally, very large numbers

of participants must be included in these studies in order to detect a sufficient number of

suicidal acts over follow-up intervals.

Previously, this type of research has not been feasible for the majority of researchers. Recent

advancements in technology, however, have rendered it possible for many to conduct these

studies much more easily and for relatively low cost. For example, ecological momentary

assessment and other mobile technologies delivered through smart-phone or web-based

methods serve as easily accessible, economically viable tools for researchers and

participants alike. Future investigations that capitalize on current technological innovations

have great potential for improving our understanding of effective prediction, prevention, and

treatment of suicidal thoughts and behaviors, and thus reduce the impact of suicide

worldwide.

5. Conclusions

This meta-analysis indicates that anxiety disorders and symptoms are statistically significant

yet weak risk factors for future suicide ideation and attempts but not deaths. Given the

methodological limitations of existing research in this area, however, results do not

necessarily imply that anxiety plays a trivial or nonexistent role in suicide prediction.

Rather, anxiety does not appear to confer meaningful risk for suicidal thoughts and

behaviors under the specific conditions of how it has been examined to date. Future research

must take advantage of recent technological advancements to generate potentially

informative and practical knowledge about the role of anxiety-related constructs in

conferring risk for suicidal thoughts and behaviors.

Appendix A. References for articles included in meta-analyses

References

Allgulander C, Allebeck P, Przybeck T, Rice JP. Risk of suicide by psychiatric diagnosis in Stockholm County. Euorpean Archives of Psychiatry and Clinical Neuroscience. 1992; 241:323–326.

Alonso P, Segalàs C, Real E, Pertusa A, Labad J, Jiménez-Murcia S, et al. Menchón JM. Suicide in patients treated for obsessive–compulsive disorder: A prospective follow-up study. Journal of Affective Disorders. 2010; 124:300–308. http://dx.doi.org/10.1016/j.jad.2009.12.001. [PubMed: 20060171]

Bentley et al. Page 15

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Angst J, Clayton PJ. Personality, smoking, and suicide: A prospective study. Journal of Affective Disorders. 1998; 51:55–62. [PubMed: 9879803]

Bakken K, Vaglum P. Predictors of suicide attempters in substance-dependent patients: A six-year prospective follow-up. Clinical Practice and Epidemiology in Mental Health. 2007; 3(20) http:// dx.doi.org/10.1186/1745-0179-3-20http://dx.doi.org/10.1186/1745-0179-3-20.

Beautrais AL. Further suicidal behavior among medically serious suicide attempters. Suicide and Life- Threatening Behavior. 2004; 34(1):1–11. [PubMed: 15106883]

Berglund M. Suicide in alcoholism: A prospective study of 88 suicides: The multidimensional diagnosis at first admission. Archives of General Psychiatry. 1984; 41:888–891. [PubMed: 6466048]

Berglund M, Nilsson K. Mortality in severe depression: A prospective study including 103 suicides. Acta Psychiatrica Scandinavica. 1987; 76:372–380. [PubMed: 3425363]

Blumenthal S. Mortality and rate of suicide of first admission psychiatric patients. Psychopathology. 1989; 22(1):50–56. http://dx.doi.org/10.1111/add.12381. [PubMed: 2727228]

Bolton JM, Pagura J, Enns MW, Grant B, Sareen J. A population-based longitudinal study of risk factors for suicide attempts in major depressive disorder. Journal of Psychiatric Research. 2010; 44(13):817–826. http://dx.doi.org/10.1016/j.jpsychires.2010.01.003. [PubMed: 20122697]

Borg SE, Ståhl M. Prediction of suicide: A prospective study of suicides and controls among psychiatric patients. Acta Psychiatrica Scandinavica. 1982; 65:221–232. [PubMed: 7072514]

Borges G, Angst J, Nock MK, Ruscio AM, Kessler RC. Risk factors for the incidence and persistence of suicide-related outcomes: A 10-year follow-up study using the National Comorbidity Surveys. Journal of Affective Disorders. 2008; 105:25–33. http://dx.doi.org/10.1016/j.jad.2007.01.036. [PubMed: 17507099]

Brådvik L, Mattisson C, Bogren M, Nettelbladt P. Mental disorders in suicide and undetermined death in the Lundby Study. The contribution of severe depression and alcohol dependence. Archives of Suicide Research. 2010; 14(3):266–275. http://dx.doi.org/10.1080/13811118.2010.494146. [PubMed: 20658380]

Breslau N, Schultz L, Lipton R, Peterson E, Welch KMA. Migraine headaches and suicide attempt. Headache: The Journal of Head and Face Pain. 2012; 52(5):723–731. http://10.10.1111/j. 1526-4610.2012.02117.x.

Britton PC, Ilgen MA, Valenstein M, Knox K, Claassen CA, Conner KR. Differences between veteran suicides with and without psychiatric symptoms. American Journal of Public Health. 2012; 102(S1):S125–S130. http://dx.doi.org/10.2105/AJPH.2011.300415. [PubMed: 22390586]

Brown GK, Beck AT, Steer RA, Grisham JR. Risk factors for suicide in psychiatric outpatients: A 20- year prospective study. Journal of Consulting and Clinical Psychology. 2000; 68(3):371–377. http://dx.doi.org/10.1037/0022-006X.68.3.371http://dx.doi.org/10.1037/0022-006X.68.3.371. [PubMed: 10883553]

Bryan CJ, Rudd MD, Wertenberger E, Young-McCaughon, Peterson A. Nonsuicidal self-injury as a prospective predictor of suicide attempts in a clinical sample of military personnel. Comprehensive Psychiatry. 2015; 59:1–7. http://dx.doi.org/10.1016/j.comppsych.2014.07.009. [PubMed: 25749478]

Cedereke M, Öjehagen A. Prediction of repeated parasuicide after 1–12 months. European Psychiatry. 2005; 20(2):101–109. http://dx.doi.org/10.1016/j.eurpsy.2004.09.008. [PubMed: 15797693]

Chan LF, Shamsul AS, Maniam T. Are predictors of future suicide attempts and the transition from suicidal ideation to suicide attempts shared or distinct: A 12-month prospective study among patients with depressive disorders. Psychiatry research. 2014; 220(3):867–873. [PubMed: 25240940]

Clarke DE, Eaton WW, Petronis KR, Ko JY, Chatterjee A, Anthony JC. Increased risk of suicide ideation in smokers and former smokers compared to never smokers: Evidence from the Baltimore ECA follow-up study. Suicide and Life-Threatening Behavior. 2010; 40(4):307–318. http:// dx.doi.org/10.1521/suli.2010.40.4.307. [PubMed: 20822357]

Conner KR, McCarthy MD, Bajorska A, Caine ED, Tu XM, Knox KL. Mood, anxiety, and substance- use disorders and suicide risk in a military population cohort. Suicide and Life-Threatening

Bentley et al. Page 16

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Behavior. 2012; 42(6):699–708. http://dx.doi.org/10.1111/j.1943-278X.2012.00125.x. [PubMed: 23094649]

Cougle JR, Resnick H, Kilpatrick DG. PTSD, depression, and their comorbidity in relation to suicidality: Cross-sectional and prospective analyses of a national probability sample of women. Depression and Anxiety. 2009; 26(12):1151–1157. http://dx.doi.org/10.1002/da.20621. [PubMed: 19842171]

Courtet P, Picot MC, Bellivier F, Torres S, Jollant F, Michelon C, et al. Malafosse A. Serotonin transporter gene may be involved in short-term risk of subsequent suicide attempts. Biological Psychiatry. 2004; 55:46–51. http://dx.doi.org/10.1016/j.biopsych.2003.07.004. [PubMed: 14706424]

Cullberg J, Wasserman D, Stefansson CG. Who commits suicide after a suicide attempt?: An 8 to 10 year follow up in suburban catchment area. Acta Psychiatrica Scandinavica. 1988; 77(5):598–603. [PubMed: 3407429]

Darke S, Williamson A, Ross J, Teesson M. Attempted suicide among heroin users: 12-month outcomes from the Australian Treatment Outcome Study (ATOS). Drug and Alcohol Cependence. 2005; 78(2):177–186. http://dx.doi.org/10.1016/j.drugalcdep.2004.10.009.

Dennehy EB, Marangell LB, Allen MH, Chessick C, Wisniewski SR, Thase ME. Suicide and suicide attempts in the systematic treatment enhancement program for bipolar disorder (STEP-BD). Journal of Affective Disorders. 2011; 133(3):423–427. http://dx.doi.org/10.1016/j.jad. 2011.04.036. [PubMed: 21601286]

Desai RA, Dausey DJ, Rosenheck RA. Mental health service delivery and suicide risk: The role of individual patient and facility factors. The American Journal of Psychiatry. 2005; 162(2):311–318. [PubMed: 15677596]

Fawcett J, Scheftner WA, Fogg L, Clark DC, Young MA, Hedeker D, Gibbons R. Time-related predictors of suicide in major affective disorder. The American Journal of Psychiatry. 1990; 147(9):1189–1194. http://dx.doi.org/10.1159/000327754. [PubMed: 2104515]

Flensborg-Madsen T, Knop J, Mortensen EL, Becker U, Sher L, Grønbæk M. Alcohol use disorders increase the risk of completed suicide—Irrespective of other psychiatric disorders. A longitudinal cohort study. Psychiatry Research. 2009; 167:123–130. http://dx.doi.org/10.1016/j.j.psychres. 2008.01.008. [PubMed: 19359047]

Fridell EJ, Öjehagen A, Träskman-Bendz I. A 5-year follow-up study of suicide attempts. Acta Psychiatrica Scandinavica. 1996; 93(3):151–157. [PubMed: 8739657]

Gallagher M, Prinstein MJ, Simon V, Spirito A. Social anxiety symptoms and suicidal ideation in a clinical sample of early adolescents: Examining loneliness and social support as longitudinal mediators. Journal of Abnormal Child Psychology. 2014; 42:871–883. http://dx.doi.org/10.1007/ s10802-013-9844-7. [PubMed: 24390470]

Goldstein TR, Ha W, Axelson DA, Goldstein BI, Liao F, Gill MK, Birmaher B. Predictors of prospectively examined suicide attempts among youth with bipolar disorder. Archives of General Psychiatry. 2012; 69(11):1113–1122. http://dx.doi.org/10.1001/archgenpsychiatry.2012.650. [PubMed: 22752079]

Goldston DB, Daniel SS, Erkanli A, Reboussin BA, Mayfield A, Frazier PH, Treadway SL. Psychiatric diagnoses as contemporaneous risk factors for suicide attempts among adolescents and young adults: Developmental changes. Journal of Consulting and Clinical Psychology. 2009; 77(2):281–290. http://dx.doi.org/10.1037/a0014732. [PubMed: 19309187]

Hayashi N, Igarashi M, Imai A, Yoshizawa Y, Utsumi K, Ishikawa Y, et al. Okazaki Y. Post- hospitalization course and predictive signs of suicidal behavior of suicidal patients admitted to a psychiatric hospital: A 2-year prospective follow-up study. BMC Psychiatry. 2012; 12 http:// dx.doi.org/10.1186/1471-244X-12-186.

Holma KM, Melartin TK, Haukka J, Holma IAK, Sokero TP, Isometsä ET. Incidence and predictors of suicide attempts in DSM-IV major depressive disorder: A five-year prospective study. The American Journal of Psychiatry. 2010; 167(7):801–808. [PubMed: 20478879]

Hung GC, Kwok C, Yip PS, Gunnell D, Chen Y. Predicting suicide in older adults – a community- based cohort study in Tapei City, Taiwan. Journal of Affective Disorders. 2014; 172:165–170. http://dx.doi.org/10.1016/j.jad.2014.09.037. [PubMed: 25451412]

Bentley et al. Page 17

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Keane EM, Dick RW, Bechtold DW, Manson SM. Predictive and concurrent validity of the Suicide ideation Questionnaire among American Indian adolescents. Journal of Abnormal Child Psychology. 1996; 24(6):735–747. http://dx.doi.org/10.1007/BF01664737. [PubMed: 8970907]

Keilp JG, Oquendo MA, Stanley BH, Burke AK, Cooper TB, Malone KM, Mann JJ. Future suicide attempt and responses to serotonergic challenge. Neuropsychopharmacology. 2010; 35(5):1063– 1072. http://dx.doi.org/10.1038/npp.2008.28. [PubMed: 18354392]

Kovacs M, Goldston D, Gatsonis C. Suicidal behaviors and childhood-onset depressive disorders: A longitudinal investigation. Journal of the American Academy of Child & Adolescent Psychiatry. 1993; 32(1):8–20. http://dx.doi.org/10.1097/00004583-199301000-00003. [PubMed: 8428888]

Laan W, Termorshuizen F, Smeets HM, Boks MPM, de Wit NJ, Geerlings MI. Acomorbid anxiety disorder does not result in an excess risk of death among patients with a depressive disorder. Journal of Affective Disorders. 2011; 135(1-3):284–291. http://dx.doi.org/10.1016/j.jad. 2011.08.026. [PubMed: 21937123]

Leadholm AKK, Rothschild AJ, Nielsen J, ech P, Østergaard SD. Risk factors for suicide among 34,671 patients with psychotic and non-psychotic severe depression. Journal of Affective Disorders. 2014; 156:119–125. http://dx.doi.org/10.1016/j.jad.2013.12.003. [PubMed: 24388683]

Lewinsohn PM, Rohde P, Seeley JR, Baldwin CL. Gender differences in suicide attempts from adolescence to young adulthood. Journal of the American Academy of Child & Adolescent Psychiatry. 2001; 40(4):427–434. [PubMed: 11314568]

Links PS, Kolla NJ, Guimond T, McMain S. Prospective risk factors for suicide attempts in a treated sample of patients with borderline personality disorder. Canadian Journal of Psychiatry. 2013; 58(2):99–106. [PubMed: 23442897]

May AM, Klonsky ED, Klein DN. Predicting future suicide attempts among depressed suicide ideators: a 10-year longitudinal study. Journal of Psychiatric Research. 2012; 46(7):946–952. http://dx.doi.org/10.1016/j.jpsychires.2012.04.009. [PubMed: 22575331]

Monnin J, Thiemard E, Vandel P, Nicolier M, Tio G, Courtet P, et al. Haffen E. Sociodemographic and psychopathological risk factors in repeated suicide attempts: gender differences in a prospective study. Journal of Affective Disorders. 2012; 136(1):35–43. http://dx.doi.org/10.1016/ j.jad.2011.09.001. [PubMed: 21975134]

Nimeus A, Alsen M, Träskman-Bendz L. The suicide assessment scale: an instrument assessing suicide risk of suicide attempters. European Psychiatry. 2000; 15(7):416–423. [PubMed: 11112934]

Nordström P, Gustavsson P, Edman G, Åsberg M. Temperamental vulnerability and suicide risk after attempted suicide. Suicide and Life-Threatening Behavior. 1996; 26(4):380–394. [PubMed: 9014268]

Preuss UW, Schuckit MA, Smith TL, Danko GP, Bucholz KK, Hesselbrock MN, et al. Kramer JR. Predictors and correlates of suicide attempts over 5 years in 1,237 alcohol-dependent men and women. American Journal of Psychiatry. 2003; 160(1):56–63. [PubMed: 12505802]

Rao U, Weissman MM, Martin JA, Hammond RW. Childhood depression and risk of suicide: a preliminary report of a longitudinal study. Journal of the American Academy of Child & Adolescent Psychiatry. 1993; 32(1):21–27. [PubMed: 8428874]

Reinherz HZ, Giaconia RM, Silverman AB, Friedman A, Pakiz B, Frost AK, Cohen E. Early psychosocial risks for adolescent suicide ideation and attempts. Journal of the American Academy of Child & Adolescent Psychiatry. 1995; 34(5):599–611. [PubMed: 7775355]

Ribeiro JD, Pease JL, Gutierrez PM, Silva C, Bernert RA, Rudd MD, Joiner TE. Sleep problems outperform depression and hopelessness as cross-sectional and longitudinal predictors of suicide ideation and behavior in young adults in the military. Journal of Affective Disorders. 2012; 136(3): 743–750. http://dx.doi.org/10.1016/j.jad.2011.09.049. [PubMed: 22032872]

Riihimäki K, Vuorilehto M, Melartin T, Haukka J, Isometsä E. Incidence and predictors of suicide attempts among primary-care patients with depressive disorders: a 5-year prospective study. Psychological Medicine. 2014; 44(02):291–302. http://dx.doi.org/10.1017/S0033291713000706. [PubMed: 23570583]

Sanchez-Gistau V, Baeza I, Arango C, Gonzalez-Pinto A, de la Serna E, Parellada M, et al. Castro- Fornieles J. Predictors of suicide attempt in early-onset, first-episode psychoses: a longitudinal 24-

Bentley et al. Page 18

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

month follow-up study. The Journal of Clinical Psychiatry. 2013; 74(1):59–66. http://dx.doi.org/ 10.4088/JCP.12m07632. [PubMed: 23217983]

Sani G, Tondo L, Koukopoulos A, Reginaldi D, Kotzalidis GD, Koukopoulos AE, et al. Tatarelli R. Suicide in a large population of former psychiatric inpatients. Psychiatry and Clinical Neurosciences. 2011; 65(3):286–295. http://dx.doi.org/10.1111/j.1440-1819.2011.02205.x. [PubMed: 21507136]

Schneider B, Philipp M, Müller MJ. Psychopathological predictors of suicide in patients with major depression during a 5-year follow-up. European Psychiatry. 2001; 16(5):283–288. [PubMed: 11514130]

Sokero TP, Melartin TK, Rytsälä HJ, Leskelä US, Lestelä-Mielonen PS, Isometsä ET. Prospective study of risk factors for attempted suicide among patients with DSM–IV major depressive disorder. The British Journal of Psychiatry. 2005; 186(4):314–318. [PubMed: 15802688]

Tejedor MC, Diaz A, Castillon JJ, Pericay JM. Attempted suicide: repetition and survival findings of a follow-up study. Acta Psychiatrica Scandinavica. 1999; 100(3):205–211. http://dx.doi.org/ 10.1097/chi.0b013e318149e66f. [PubMed: 10493087]

Tidemalm D, Haglund A, Karanti A, Landen M, Runeson B. Attempted suicide in bipolar disorder: Risk factors in a cohort of 6086 patients. PLoS ONE. 2014; 9(4):e94097. http://dx.doi.org/ 10.1371/journal.pone.0094097. [PubMed: 24705630]

Tuisku, V.; Kiviruusu, O.; Pelkonen, M.; Karlsson, L.; Strandholm, T.; Marttunen, M. Depressed adolescents as young adults–Predictors of suicide attempt and non-suicidal self-injury during an 8- year follow-up. Journal of Affective Disorders. 2013. http://dx.doi.org/10.1016/j.jad.2013.09.031

Valtonen HM, Suominen K, Mantere O, Leppämäki S, Arvilommi P, Isometsä ET. Prospective study of risk factors for attempted suicide among patients with bipolar disorder. Bipolar Disorders. 2006; 8:576–585. [PubMed: 17042831]

Valtonen HM, Suominen K, Haukka J, Mantere O, Leppämäki S, Arvilommi P, Isometsä ET. Differences in incidence of suicide attempts during phases of bipolar I and II disorders. Bipolar Disorders. 2008; 10(5):588–596. [PubMed: 18657243]

Wedig MM, Silverman MH, Frankenburg FR, Reich DB, Fitzmaurice G, Zanarini MC. Predictors of suicide attempts in patients with borderline personality disorder over 16 years of prospective follow-up. Psychological Medicine. 2012; 42(11):2395–2404. http://dx.doi.org/10.1017/ S0033291712000517. [PubMed: 22436619]

Wenzel A, Berchick ER, Tenhave T, Halberstadt S, Brown GK, Beck AT. Predictors of suicide relative to other deaths in patients with suicide attempts and suicide ideation: a 30-year prospective study. Journal of Affective Disorders. 2011; 132(3):375–382. http://dx.doi.org/10.1016/j.jad. 2011.03.006. [PubMed: 21481944]

Wolk SI, Weissman MM, Puig J. Suicidal Behavior in Depressed Children Grown. Psychiatric Annals. 1996; 26(6) http://dx.doi.org/10.1097/chi.0b013e318149e66f.

Yaseen ZS, Chartrand H, Mojtabai R, Bolton J, Galynker II. Fear of yding in panic attacks predicts suicide attempt in comorbid depressive illness: Prospective evidence from the national epidemiological survey on alcohol and related conditions. Depression and Anxiety. 2013; 30(10): 930–939. http://dx.doi.org/10.1002/da.22039. [PubMed: 23281011]

Yen S, Shea T, Pagano M, Sanislow CA, Grilo CM, McGlashan TH, et al. Morey LC. Axis I and axis II disorders as predictors of prospective suicide attempts: findings from the collaborative longitudinal personality disorders study. Journal of Abnormal Psychology. 2003; 112(3):375. http://dx.doi.org/10.1037/0021-843X.112.3375. [PubMed: 12943016]

References

American Association of Suicidology. [Accessed October 16, 2015] Warning Signs & Risk Factors [online]. 2015. http://www.suicidology.org/ncpys/warning-signs-risk-factors

American Foundation for Suicide Prevention. [Accessed October 16, 2015] Suicide Risk factors [online]. 2015. https://www.afsp.org/understanding-suicide/suicide-risk-factors

Amir M, Kaplan Z, Efroni R, Kotler M. Suicide risk and coping styles in posttraumatic stress disorder patients. Psychotherapy and Psychosomatics. 1999; 68(2):76–81. [PubMed: 10026458]

Bentley et al. Page 19

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Barlow DH, Sauer-Zavala S, Carl JR, Bullis JR, Ellard KK. The nature, diagnosis, and treatment of neuroticism: Back to the future. Clinical Psychological Science. 2014; 2(3):344–365.

Baumeister R. Suicide as escape from self. Psychological Review. 1990; 97:90–113. http://dx.doi.org/ 10.1037/0033-295X.97.1.90. [PubMed: 2408091]

Beck AT, Epstein N, Brown G, Steer RA. An inventory for measuring clinical anxiety: Psychometric properties. Journal of Consulting and Clinical Psychology. 1988; 56:893–897. [PubMed: 3204199]

Boden JM, Fergusson DM, Horwood LJ. Anxiety disorders and suicidal behaviours in adolescence and young adulthood: Findings from a longitudinal study. Psychological Medicine. 2007; 37:431–440. [PubMed: 17109776]

Boergers J, Spirito A, Donaldson D. Reasons for adolescent suicide attempts: Associations with psychological functioning. Journal of the American Academy of Child and Adolescent Psychiatry. 1998; 37(12):1287–1293. [PubMed: 9847501]

Bolton JM, Cox BJ, Afifi TO, Enns MW, Bienvenu OJ, Sareen J. Anxiety disorders and risk for suicide attempts: Findings from the Baltimore epidemiologic catchment area follow-up study. Depression and Anxiety. 2008; 25:477–481. [PubMed: 17541978]

Borenstein, M.; Hedges, L.; Higgins, J.; Rothstein, H. Comprehensive meta-analysis version 2. Englewood, New Jersey: Biostat; 2005.

Borges G, Angst J, Nock MK, Ruscio AM, Kessler RC. Risk factors for the incidence and persistence of suicide-related outcomes: A 10-year follow-up study using the National Comorbidity Surveys. Journal of Affective Disorders. 2008a; 105:25–33. http://dx.doi.org/10.1016/j.jad.2007.01.036. [PubMed: 17507099]

Borges G, Angst J, Nock MK, Ruscio AM, Walters EE, Kessler RC. A risk index for 12-month suicide attempts in the National Comorbidity Survey Replication (NCS-R). Psychological Medicine. 2006; 36(12):1747–1757. [PubMed: 16938149]

Borges G, Nock MK, Haro Abad JM, Hwang I, Sampson NA, Alonso J, et al. Kessler RC. Twelve month prevalence of and risk factors for suicide attempts in the WHO Mental Health Surveys. Journal of Clinical Psychiatry. 2010; 71(12):1617–1628. http://dx.doi.org/10.4088/JCP. 08m04967blu. [PubMed: 20816034]

Briere J, Hodges M, Godbout N. Traumatic stress, affect dysregulation, and dysfunctional avoidance: A structural equation model. Journal of Traumatic Stress. 2010; 23:767–774. [PubMed: 21171138]

Britton PC, Ilgen MA, Rudd MD, Conner KR. Warning signs for suicide within a week of healthcare contact in Veteran decedents. Psychiatry Research. 2012; 200:395–399. [PubMed: 22796102]

Bryan CJ, Rudd MD, Wertenberger E. Reasons for suicide attempts in a clinical sample of active duty soldiers. Journal of Affective Disorders. 2013; 144:148–152. [PubMed: 22858207]

Busch KA, Fawcett J, Jacobs DG. Clinical correlates of inpatient suicide. Journal of Clinical Psychiatry. 2003

Centers for Disease Control and Prevention. Web-based Injury Statistics Query and Reporting System (WISQARS) [online]. 2011. Retrieved from www.cdc.gov/injury/wisqars/index.html

Conrad AK, Jacoby AM, Jobes DA, Lineberry TW, Shea CE, Arnold Ewing TD, et al. Kung S. A psychometric investigation of the Suicide Status Form II with a psychiatric inpatient sample. Suicide and Life-threatening Behavior. 2009; 39(3):307–320. [PubMed: 19606922]

Fawcett J. Treating impulsivity and anxiety in the suicidal patient. Annals of the New York Academy of Sciences. 2001; 932:94–102. [PubMed: 11411193]

Fawcett J, Busch KA, Jacobs D, Kravitz HM, Fogg L. Suicide: A four-pathway clinical-biochemical model. Annals of the New York Academy of Sciences. 1997; 836:288–301. [PubMed: 9616805]

Fawcett J, Scheftner WA, Fogg L, Clark DV, Young MA, Hedeker D, Gibbons R. Time-related predictors of suicide in major affective disorder. American Journal of Psychiatry. 1990b; 147:1189–1194. [PubMed: 2104515]

Field AP. Meta-analysis of correlation coefficients: A Monte Carlo comparison of fixed- and random- effects methods. Psychological Methods. 2001; 6:161–180. http://dx.doi.org/10.1037/1082-989X. 6.2.161. [PubMed: 11411440]

Goodman WK, Price LH, Rasmussen SA, Mazure C, Delgado P, Heninger GR, Charney DS. The Yale-Brown obsessive compulsive scale. II. Validity. Archives of General Psychiatry. 1989; 46(11):1012–1016. [PubMed: 2510699]

Bentley et al. Page 20

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Gradus JL, Qin P, Lincoln AK, Miller M, Lawler E, Sorensen, Lash TL. Post-traumatic stress disorder and completed suicide. American Journal of Epidemiology. 2010; 171(6):721–727. [PubMed: 20160171]

Hall R, Platt D, Hall R. Suicide risk assessment: A review of risk factors for suicide in 100 patients who made severe suicide attempts. Psychosomatics. 1999; 40:18–27. [PubMed: 9989117]

Hamilton M. The assessment of anxiety states by rating. British Journal of Medical Psychology. 1959; 32:50–55. [PubMed: 13638508]

Hedges, LV.; Olkin, I. Statistical methods for meta-analysis. Orlando, FL: Academic Press; 1985.

Hedges LV, Vevea JL. Fixed- and random effects models in meta-analysis. Psychological Methods. 1998; 3:486–504. http://dx.doi.org/10.1037/1082-989X.3.4.486.

Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. British Medical Journal. 2003; 327(7414):557–560. http://dx.doi.org/10.1136/bmj.327.7414.557. [PubMed: 12958120]

Hornig CD, McNally RJ. Panic disorder and suicide attempts: A reanalysis of data from the epidemiologic catchment area study. British Journal of Psychiatry. 1995; 167:76–79. [PubMed: 7551614]

Joiner, TE. Why people die by suicide. Cambridge, MA, US: Harvard University Press; 2005.

Kanwar A, Malik S, Prokop LJ, Sim LA, Feldstein D, Wang Z, Murad MH. The association between anxiety disorders and suicidal behaviors: A systematic review and meta-analysis. Depression and Anxiety. 2013; 30(10):917–929. http://dx.doi.org/10.1002/da.22074. [PubMed: 23408488]

Kaplan MS, McFarland BH, Huguet N, Newsom JT. Physical illness, functional limitations, and suicide risk: A population-based study. American Journal of Orthopsychiatry. 2007; 77(1):56–60. [PubMed: 17352585]

Kraemer HC, Kazdin AE, Offord DR, Kessler RC, Jensen PS, Kupfer DJ. Coming to terms with the terms of risk. Archives of General Psychiatry. 1997; 54:337–344. [PubMed: 9107150]

National Suicide Prevention Lifeline. [Accessed October 16, 2015] Suicide Risk Factors [online]. 2015. http://www.suicidepreventionlifeline.org/learn/riskfactors.aspx

Massion AO, Warshaw MG, Keller MB. Quality of life and psychiatric morbidity in panic disorder and generalized anxiety disorder. American Journal of Psychiatry. 1993; 150(4):600–607. [PubMed: 8465877]

Mathers CD, Loncar D. Projections of global mortality and burden of disease from 2002 to 2030. PLoS Medicine. 2006; 3(11):e442. http://dx.doi.org/10.1371/journal.pmed.0030442. [PubMed: 17132052]

Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement. Annals of Internal Medicine. 2009; 151(4):264–269. http://dx.doi.org/10.7326/0003-4819-151-4-200908180-00135. [PubMed: 19622511]

Nepon J, Belik SL, Bolton J, Sareen J. The relationship between anxiety disorders and suicide attempts: findings from the National Epidemiologic Survey on Alcohol and Related Conditions. Depression and Anxiety. 2010; 27(9):791–798. [PubMed: 20217852]

Nock MK. Self-injury. Annual Review of Clinical Psychology. 2010; 6:339–363.

Nock MK, Hwang I, Sampson NA, Kessler RC. Mental disorders, comorbidity and suicidal behavior: Results from the national comorbidity survey replication. Molecular Psychiatry. 2010a; 15(8): 868–876. [PubMed: 19337207]

Nock MK, Borges G, Bromet EJ, Cha CB, Kessler RC, Lee S. Suicide and suicidal behavior. Epidemiologic Reviews. 2008; 30(1):133–154. [PubMed: 18653727]

Nock MK, Hwang I, Sampson N, Kessler RC, Angermeyer M, Beautrais A, et al. Williams DR. Cross- national analysis of the associations among mental disorders and suicidal behavior: Findings from the WHO world mental health surveys. PLoS Medicine. 2009; 6(8):e1000123. [PubMed: 19668361]

Nock MK, Ursano RJ, Heeringa SG, Stein MB, Jain S, Raman R, et al. Kessler RC. Mental disorders, comorbidity, and pre-enlistment suicidal behavior among new soldiers in the U.S. Army: Results from the Army Study to Assess Risk and Resilience in Servicemembers (ARMY STARRS). Suicide and Life-threatening Behavior. 2015 in press.

Bentley et al. Page 21

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Noyes R. Suicide and panic disorder: A review. Journal of Affective Disorders. 1991; 22:1–11. [PubMed: 1880304]

Olatunji BO, Cisler JM, Tolin DF. Quality of life in the anxiety disorders: A meta-analytic review. Clinical Psychology Review. 2007; 27(5):572–581. http://dx.doi.org/10.1016/j.cpr.2007.01.015. [PubMed: 17343963]

Rector NA, Kamkar K, Riskind JH. Misappraisal of time perspective and suicide in the anxiety disordres: The multiplier effect of looming illusions. International Journal of Cognitive Therapy. 2008; 1(1):69–79.

Ribeiro JD, Silva C, Joiner TE. Overarousal interacts with a sense of fearlessness about death to predict suicide risk in a sample of clinical outpatients. Psychiatry Research. 2014a; 218(1–2): 106–112. [PubMed: 24780448]

Ribeiro JD, Bender TW, Buchman JM, Nock MK, Rudd MD, Bryan CJ, et al. Joiner TE. An investigation of the interactive effects of the capability for suicide and acute agitation on suicidality in a military sample. Depression and Anxiety. 2015

Riskind JH. Looming vulnerability to threat: A cognitive paradigm for anxiety. Behaviour Research and Therapy. 1997; 35:685–702. [PubMed: 9256513]

Riskind, JH.; Long, DG.; Williams, NL.; White, J. Desperate acts for desperate times: Looming vulnerability and suicide. In: Joiner, T., editor. Suicide science. New York: Plenum Press; 2000. p. 105-115.

Robins, E. The final months: A study of the lives of 134 persons who committed suicide. New York, NY: Oxford University Press; 1981.

Rudd MD, Berman AL, Joiner TE, Nock MK, Silverman MM, Mandrusiak M, et al. Witte T. Warning signs for suicide: Theory, research, and clinical applications. Suicide and Life-threatening Behavior. 2006; 36:255–265. [PubMed: 16805653]

Sareen J. Anxiety disorders and risk for suicide: Why such controversy? Depression and Anxiety. 2011; 28(11):941–945. [PubMed: 22076969]

Sareen J, Cox BJ, Afifi TO, de Graaf R, Asmundson GJ, ten Have M, Stein MB. Anxiety disorders and risk for suicide ideation and suicide attempts: A population-based longitudinal study of adults. Archives of General Psychiatry. 2005; 62(11):1249–1257. [PubMed: 16275812]

Shneidman, ES. Suicide as psychache: A clinical approach to self-destructive behavior. Northvale, NJ: Jason Aronson; 1993.

Silva C, Ribeiro JD, Joiner TE. Mental disorders and thwarted belongingness, perceived burdensomeness, and acquired capability for suicide. Psychiatry Research. 2015; 226:316–327. [PubMed: 25650048]

Šimundić AM. Measures of diagnostic accuracy: Basic definitions. Medical & Biological Sciences. 2008; 22(4):61–65.

Thibodeau MA, Welch PG, Sareen J, Asmundson GJ. Anxiety disorders are independently associated with suicide ideation and attempts: Propensity score matching in two epidemiological samples. Depression and Anxiety. 2013; 30(10):947–954. [PubMed: 24108489]

Van Orden KA, Witte TK, Cukrowicz KC, Braithwaite S, Selby EA, Joiner TE. The interpersonal theory of suicide. Psychological Review. 2010; 117(2):575–600. [PubMed: 20438238]

Warshaw MG, Massion AO, Peterson LG, Pratt LA, Keller MB. Suicidal behavior in patients with panic disorder: Retrospective and prospective data. Journal of Affective Disorders. 1995; 34(3): 235–247. [PubMed: 7560552]

Way B, Miraglia R, Sawyer D, Beer R, Eddy J. Factors related to suicide in New York state prisons. International Journal of Law and Psychiatry. 2005; 28:207–221. http://dx.doi.org/10.1016/j.ijlp. 2004.09.003. [PubMed: 15950281]

Weissman MM, Klerman GL, Markowitz JS, Ouellette R. Suicide ideation and suicide attempts in panic disorder and attacks. New England Journal of Medicine. 1989; 321:1209–1214. [PubMed: 2797086]

Wenzel A, Beck AT. A cognitive model of suicidal behavior: Theory and treatment. Applied and Preventive Psychology. 2008; 12:189–201. http://dx.doi.org/10.1016/j.appsy.2008.05.001.

Wenzel, A.; Brown, GK.; Beck, AT. Cognitive therapy for suicidal patients: Scientific and clinical applications. Washington, DC: APA Books; 2008. in press

Bentley et al. Page 22

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Wilcox HC, Storr CL, Breslau N. Posttraumatic stress disorder and suicide attempts in a community sample of urban American young adults. Archives of General Psychiatry. 2009; 66:305–311. [PubMed: 19255380]

World Health Organization. Public health action for the prevention of suicide: A framework. Geneva, Switzerland: WHO Press; 2012.

Zamora, J.; Muriel, A.; Abraira, V. Meta-DiSc: Meta-analysis of diagnostic and screening tests, version 1.4. Madrid, Spain: 2014.

Bentley et al. Page 23

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Highlights

• We conducted a meta-analysis of anxiety predicting suicidal thoughts and behaviors.

• Anxiety was a significant risk factor for suicide ideation and attempts.

• As it has traditional been studied, anxiety is a weak predictor of suicidality.

• Future research must approximate current theories about risk factors for suicide.

Bentley et al. Page 24

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Fig. 1. PRISMA diagram.

Bentley et al. Page 25

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Fig. 2. Overall odds ratios (ORs) and corresponding 95% confidence intervals for any anxiety

construct, any anxiety symptoms, and any anxiety diagnosis predicting suicide ideation,

suicide attempt, and suicide death. An OR must exceed 1.0 to be statistically significant.

Bentley et al. Page 26

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Fig. 3. Funnel plots for anxiety constructs predicting suicide-related outcomes (ORs). Open circles

indicate values captured in the meta-analysis. Shaded circles indicate imputed values

missing to the left of the mean due to publication bias. Open diamonds indicate weighted

mean ORs before adjusting for publication bias. Shaded diamonds indicates weighted mean

ORs after adjusting for publication bias.

Bentley et al. Page 27

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Fig. 4. ROC curves for anxiety predicting suicide attempt and death.

Bentley et al. Page 28

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Fig. 5. Moderation effect for length of follow-up period on suicide ideation and attempt.

Bentley et al. Page 29

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Bentley et al. Page 30

T a b

le 1

D es

cr ip

ti on

o f

in cl

ud ed

s tu

di es

.

A u

th or

(s ),

y ea

r T

ot al

N S

am p

le t

yp ea

S am

p le

p op

u la

ti on

b L

en gt

h o

f fo

ll ow

-u p

(i

n

m on

th s)

A n

xi et

y p

re d

ic to

r( s)

S u

ic id

e- re

la te

d o

u tc

om e(

s)

A ll

gu la

nd er

e t

al . (

19 92

) 80

,7 90

S w

ed is

h di

sc ha

rg ed

i np

at ie

nt s

C li

ni ca

l A

do le

sc en

t an

d ad

ul t

16 8

A nx

ie ty

n eu

ro si

sc D

ea th

A lo

ns o

et a

l. (

20 10

) 21

8 S

pa ni

sh p

sy ch

ia tr

ic p

at ie

nt s

C li

ni ca

l A

du lt

12 0

O C

D d

ur at

io n

an d

sy m

pt om

s A

tt em

pt

A ng

st &

C al

yt on

( 19

98 )

62 d

ec ea

se d

S w

is s

m al

es f

ro m

g en

er al

co

m m

un it

y G

en er

al A

du lt

20 4

A ny

a nx

ie ty

d ia

gn os

is D

ea th

B ak

ke n

& V

ag lu

m (

20 06

) 16

0 N

or w

eg ia

n pa

ti en

ts w

it h

su bs

ta nc

e ab

us e

C li

ni ca

l A

du lt

72 A

G , G

A D

, O C

D , P

T S

D ,

S O

C , S

O M

, S P

E C

A tt

em pt

B ea

ut ra

is (

20 04

) 30

2 N

ew Z

ea la

nd p

at ie

nt s

w it

h pr

io r

at te

m pt

H is

to ry

o f

S IT

B s

A do

le sc

en t

an d

ad ul

t 60

A ny

a nx

ie ty

d ia

gn os

is A

tt em

pt , d

ea th

B er

gl un

d (1

98 4)

1, 31

2 S

w ed

is h

in pa

ti en

ts w

it h

al co

ho l

ab us

e C

li ni

ca l

A du

lt 38

4 A

ny a

nx ie

ty d

ia gn

os is

D ea

th

B er

gl un

d &

N il

ss on

( 19

87 )

(a )

70 0

S w

ed is

h fe

m al

e in

pa ti

en ts

w it

h m

oo d

di so

rd er

C li

ni ca

l A

du lt

32 4

A ny

a nx

ie ty

d ia

gn os

is D

ea th

B er

gl un

d &

N il

ss on

( 19

87 )

(b )

50 0

S w

ed is

h m

al e

in pa

ti en

ts w

it h

m oo

d di

so rd

er C

li ni

ca l

A du

lt 32

4 A

nx ie

ty s

ym pt

om s

D ea

th

B lu

m en

th al

e t

al . (

19 89

) 36

G er

m an

p sy

ch ia

tr ic

p at

ie nt

s C

li ni

ca l

A du

lt 60

A ny

a nx

ie ty

d ia

gn os

is D

ea th

B ol

to n

et a

l. (

20 10

) 5,

97 2

ge ne

ra l

co m

m un

it y

G en

er al

A du

lt 36

A ny

a nx

ie ty

d ia

gn os

is , G

A D

, P

D , P

T S

D , S

O C

, S P

E C

A tt

em pt

B or

g &

S tå

hl (

19 82

) 68

S w

ed is

h ps

yc hi

at ri

c pa

ti en

ts G

en er

al A

du lt

24 A

nx ie

ty s

ym pt

om s,

O C

D

sy m

pt om

s D

ea th

B or

ge s

et a

l. (

20 08

) 5,

00 1

ge ne

ra l

co m

m un

it y

G en

er al

, H is

to ry

of

S IT

B s

A do

le sc

en t

an d

ad ul

t 12

0 A

ny a

nx ie

ty d

ia gn

os is

, A G

, G

A D

, P D

, P T

S D

, S O

C ,

S P

E C

Id ea

ti on

, g es

tu re

, p la

n,

at te

m pt

B rå

dv ik

e t

al . (

20 10

) 85

S w

ed is

h ge

ne ra

l co

m m

un it

y G

en er

al A

do le

sc en

t an

d ad

ul t

70 8

A D

J, A

N X

-N O

S , O

C D

, P

D A

, S O

C , S

O M

D ea

th

B re

sl au

e t

al . (

20 12

) 1,

18 6

ge ne

ra l

co m

m un

it y

G en

er al

A du

lt 24

A ny

a nx

ie ty

d ia

gn os

is A

tt em

pt

B ri

tt on

, e t

al . (

20 12

) 38

1 ve

te ra

n su

ic id

e de

sc en

de nt

s H

is to

ry o

f S

IT B

s A

du lt

1 A

nx ie

ty s

ym pt

om s

D ea

th

B ro

w n

et a

l. (

20 00

) 6,

89 1

ps yc

hi at

ri c

ou tp

at ie

nt s

C li

ni ca

l A

du lt

28 8

A nx

ie ty

s ym

pt om

s, P

D D

ea th

B ry

an e

t al

. ( 20

14 )

17 6

ve te

ra n

ou tp

at ie

nt s

H is

to ry

o f

S IT

B s

A du

lt 14

P T

S D

s ym

pt om

s A

tt em

pt

C ed

er ek

e &

Ö je

ha ge

n (2

01 5)

17 8

S w

ed is

h ps

yc hi

at ri

c pa

ti en

ts C

li ni

ca l

A du

lt 11

A D

J A

tt em

pt

C ha

n et

a l.

( 20

14 )

66 i

np at

ie nt

s w

it h

de pr

es si

on C

li ni

ca l

A du

lt 12

A ny

a nx

ie ty

d ia

gn os

is A

tt em

pt

C la

rk e

et a

l. (

20 10

) 1,

29 2

ge ne

ra l

co m

m un

it y

G en

er al

A du

lt 27

6 A

ny a

nx ie

ty d

ia gn

os is

Id ea

ti on

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Bentley et al. Page 31

A u

th or

(s ),

y ea

r T

ot al

N S

am p

le t

yp ea

S am

p le

p op

u la

ti on

b L

en gt

h o

f fo

ll ow

-u p

(i

n

m on

th s)

A n

xi et

y p

re d

ic to

r( s)

S u

ic id

e- re

la te

d o

u tc

om e(

s)

C on

ne r

et a

l. (

20 12

) 30

9, 86

1 U

ni te

d S

ta te

s A

ir F

or ce

m

em be

rs G

en er

al A

du lt

72 A

ny a

nx ie

ty d

ia gn

os is

D ea

th

C ou

gl e

et a

l. (

20 09

) 2,

55 1

fe m

al es

i n

ge ne

ra l

co m

m un

it y

G en

er al

A du

lt 12

P T

S D

Id ea

ti on

C ou

rt et

e t

al . (

20 03

) 76

W es

t E

ur op

ea n

pa ti

en ts

w it

h pr

io r

at te

m pt

H is

to ry

o f

S IT

B s

A du

lt 12

A G

, G A

D , P

D , S

O C

A tt

em pt

C ul

lb er

g et

a l.

( 19

88 )

16 3

S w

ed is

h ge

ne ra

l co

m m

un it

y w

it h

pr io

r at

te m

pt H

is to

ry o

f S

IT B

s A

du lt

12 0

A D

Jd D

ea th

D ar

ke e

t al

. ( 20

05 )

49 5

A us

tr al

ia n

pa ti

en ts

w it

h su

bs ta

nc e

us e

di so

rd er

C li

ni ca

l A

du lt

12 P

T S

D A

tt em

pt

D en

ne hy

e t

al . (

20 11

) 4,

36 0

pa ti

en ts

w it

h bi

po la

r di

so rd

er C

li ni

ca l

A du

lt 16

A ny

a nx

ie ty

d ia

gn os

is , P

D A

tt em

pt , d

ea th

D es

ai e

t al

. ( 20

05 )

12 1,

93 3

D ep

t. o

f V

et er

an s

A ff

ai rs

ps

yc hi

at ri

c pa

ti en

ts C

li ni

ca l

A du

lt 48

P T

S D

D ea

th

F aw

ce tt

e t

al . (

19 90

) 95

4 in

pa ti

en ts

w it

h m

oo d

di so

rd er

C li

ni ca

l A

du lt

12 0

A nx

ie ty

s ym

pt om

s, O

C D

sy

m pt

om s

A tt

em pt

F le

ns bo

rg -M

ad se

n et

a l.

( 20

09 )

18 ,1

46 g

en er

al D

an is

h co

m m

un it

y G

en er

al A

du lt

31 2

A ny

a nx

ie ty

d ia

gn os

is A

tt em

pt

F ri

de ll

e t

al . (

19 96

) 42

S w

ed is

h in

pa ti

en ts

w it

h pr

io r

at te

m pt

H is

to ry

o f

S IT

B s

A du

lt 60

A ny

a nx

ie ty

d ia

gn os

is , A

D J

A tt

em pt

G al

la gh

er e

t al

. ( 20

14 )

14 4

in pa

ti en

ts C

li ni

ca l

A do

le sc

en t

18 S

oc ia

l an

xi et

y sy

m pt

om s

Id ea

ti on

G ol

ds te

in e

t al

. ( 20

11 )

41 3

pa ti

en ts

w it

h bi

po la

r di

so rd

er C

li ni

ca l

A do

le sc

en t

60 A

ny a

nx ie

ty d

ia gn

os is

, P D

A tt

em pt

G ol

ds to

n et

a l.

( 20

09 )

18 0

in pa

ti en

ts C

li ni

ca l

A do

le sc

en t

16 8

G A

D , P

D , S

P E

C A

tt em

pt

H ay

as hi

e t

al . (

20 12

) 10

6 Ja

pa ne

se i

np at

ie nt

s w

it h

pr io

r se

lf -i

nj ur

y H

is to

ry o

f S

IT B

s A

du lt

24 A

ny a

nx ie

ty d

ia gn

os is

A tt

em pt

H ol

m a

et a

l. (

20 10

) 24

9 F

in ni

sh p

at ie

nt s

w it

h M

D D

C li

ni ca

l A

du lt

60 A

ny a

nx ie

ty d

ia gn

os is

, an

xi et

y sy

m pt

om s

A tt

em pt

H un

g et

a l.

( 20

15 )

10 2,

45 4

ge ne

ra l

el de

rl y

C hi

ne se

co

m m

un it

y G

en er

al A

du lt

60 A

nx ie

ty s

ym pt

om s

D ea

th

K ea

ne e

t al

. ( 19

96 )

13 4

A m

er ic

an I

nd ia

n hi

gh s

ch oo

l st

ud en

ts G

en er

al A

do le

sc en

t 6

A nx

ie ty

s ym

pt om

s A

tt em

pt

K ei

lp e

t al

. ( 20

08 )

26 p

at ie

nt s

w it

h M

D D

a nd

p ri

or

at te

m pt

H is

to ry

o f

S IT

B s

A du

lt 24

P T

S D

A tt

em pt

K ov

ac s

et a

l. (

19 93

) 18

3 in

pa ti

en ts

C li

ni ca

l A

do le

sc en

t 14

4 A

D J

Id ea

ti on

, a tt

em pt

L aa

n et

a l.

( 20

11 )

12 9,

78 1

ge ne

ra l

D ut

ch c

om m

un it

y G

en er

al A

do le

sc en

t an

d ad

ul t

12 0

A ny

a nx

ie ty

d ia

gn os

is D

ea th

L ea

dh ol

m e

t al

. ( 20

13 )

34 ,6

71 p

at ie

nt s

w it

h de

pr es

si on

C li

ni ca

l A

du lt

20 4

A ny

a nx

ie ty

d ia

gn os

is D

ea th

L ew

in so

hn e

t al

. ( 20

01 )

(a )

53 9

D an

is h

fe m

al es

i n

ge ne

ra l

co m

m un

it y

G en

er al

A do

le sc

en t

14 4

A ny

a nx

ie ty

d ia

gn os

is A

tt em

pt

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Bentley et al. Page 32

A u

th or

(s ),

y ea

r T

ot al

N S

am p

le t

yp ea

S am

p le

p op

u la

ti on

b L

en gt

h o

f fo

ll ow

-u p

(i

n

m on

th s)

A n

xi et

y p

re d

ic to

r( s)

S u

ic id

e- re

la te

d o

u tc

om e(

s)

L ew

in so

hn e

t al

. ( 20

01 )

(b )

40 2

D an

is h

m al

es i

n ge

ne ra

l co

m m

un it

y G

en er

al A

do le

sc en

t 14

4 A

ny a

nx ie

ty d

ia gn

os is

A tt

em pt

L in

ks e

t al

. ( 20

13 )

18 0

C an

ad ia

n pa

ti en

ts w

it h

B P

D C

li ni

ca l

A du

lt 24

P T

S D

A tt

em pt

M ay

e t

al . (

20 12

) 49

p sy

ch ia

tr ic

p at

ie nt

s w

it h

su ic

id e

id ea

ti on

H is

to ry

o f

S IT

B s

A du

lt 12

0 A

ny a

nx ie

ty d

ia gn

os is

A tt

em pt

M on

ni n

et a

l. (

20 11

) 27

3 F

re nc

h ps

yc hi

at ri

c pa

ti en

ts w

it h

pr io

r at

te m

pt H

is to

ry o

f S

IT B

s A

du lt

24 P

T S

D A

tt em

pt

N im

eu s

et a

l. (

20 00

) 19

1 S

w ed

is h

ps yc

hi at

ri c

pa ti

en ts

w it

h pr

io r

at te

m pt

H is

to ry

o f

S IT

B s

A du

lt 12

A D

J D

ea th

N or

ds tr

öm e

t al

. ( 19

96 )

54 S

w ed

is h

in pa

ti en

ts w

it h

pr io

r at

te m

pt H

is to

ry o

f S

IT B

s A

du lt

72 A

nx ie

ty s

ym pt

om s

D ea

th

P re

us s

et a

l. (

20 03

) 1,

23 7

ge ne

ra l

co m

m un

it y

w it

h al

co ho

l ab

us e

or f

am il

y m

em be

r w

it h

al co

ho l

ab us

e

C li

ni ca

l A

du lt

60 P

D A

tt em

pt

R ao

e t

al . (

19 93

) 28

1 pa

ti en

ts w

it h

de pr

es si

on a

nd /o

r an

xi et

y, o

r co

nt ro

ls C

li ni

ca l

A do

le sc

en t

15 6

A ny

a nx

ie ty

d ia

gn os

is D

ea th

R ei

nh er

z et

a l.

( 19

95 )

(a )

19 3

fe m

al es

i n

ge ne

ra l

co m

m un

it y

G en

er al

A do

le sc

en t

16 8

P T

S D

, S O

C , S

P E

C , a

nx ie

ty

sy m

pt om

s Id

ea ti

on

R ei

nh er

z et

a l.

( 19

95 )

(b )

18 5

m al

es i

n ge

ne ra

l co

m m

un it

y G

en er

al A

do le

sc en

t 16

8 P

T S

D , S

O C

, S P

E C

, a nx

ie ty

sy

m pt

om s

Id ea

ti on

R ib

ei ro

e t

al . (

20 12

) 23

9 ps

yc hi

at ri

c ou

tp at

ie nt

s w

it h

su ic

id al

it y

H is

to ry

o f

S IT

B s

A du

lt 1

P T

S D

, a nx

ie ty

s ym

pt om

s Id

ea ti

on , a

tt em

pt

R ii

hi m

ak i

et a

l. (

20 13

) 13

4 F

in ni

sh p

at ie

nt s

w it

h m

oo d

di so

rd er

C li

ni ca

l A

du lt

60 A

ny a

nx ie

ty d

ia gn

os is

, G A

D ,

P D

, S O

C , S

O M

, a nx

ie ty

sy

m pt

om s

A tt

em pt

S an

ch ez

-G is

ta u

et a

l. (

20 13

) 82

S pa

ni sh

p at

ie nt

s w

it h

fi rs

t- ep

is od

e ps

yc ho

si s

C li

ni ca

l A

do le

sc en

t 24

A ny

a nx

ie ty

d ia

gn os

is A

tt em

pt

S an

i et

a l.

( 20

11 )

4, 44

1 It

al ia

n ps

yc hi

at ri

c pa

ti en

ts C

li ni

ca l

A du

lt 42

0 G

A D

, O C

D , P

D D

ea th

S ch

ne id

er e

t al

. ( 20

01 )

27 8

G er

m an

p at

ie nt

s w

it h

M D

D C

li ni

ca l

A du

lt 60

A ny

a nx

ie ty

d ia

gn os

is , P

D ,

an xi

et y

sy m

pt om

s D

ea th

S ok

er o

et a

l. (

20 05

) 19

8 F

in ni

sh p

at ie

nt s

w it

h M

D D

C li

ni ca

l A

du lt

18 A

ny a

nx ie

ty d

ia gn

os is

A G

, G

A D

, O C

D , P

D , P

T S

D ,

S O

C , S

P E

C , a

nx ie

ty

sy m

pt om

s

A tt

em pt

T id

em al

m e

t al

. ( 20

14 )

(a )

3, 67

8 S

w ed

is h

fe m

al es

w it

h bi

po la

r di

so rd

er C

li ni

ca l

A du

lt 84

A ny

a nx

ie ty

d is

or de

r A

tt em

pt

T id

em al

m e

t al

. ( 20

14 )

(b )

2, 40

8 S

w ed

is h

m al

es w

it h

bi po

la r

di so

rd er

C li

ni ca

l A

du lt

84 A

ny a

nx ie

ty d

is or

de r

A tt

em pt

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Bentley et al. Page 33

A u

th or

(s ),

y ea

r T

ot al

N S

am p

le t

yp ea

S am

p le

p op

u la

ti on

b L

en gt

h o

f fo

ll ow

-u p

(i

n

m on

th s)

A n

xi et

y p

re d

ic to

r( s)

S u

ic id

e- re

la te

d o

u tc

om e(

s)

T ej

ed or

e t

al . (

19 99

) 15

0 S

pa ni

sh p

at ie

nt s

w it

h pr

io r

at te

m pt

H is

to ry

o f

S IT

B s

A du

lt 12

0 A

ny a

nx ie

ty d

ia gn

os is

, A D

J,

an xi

et y

sy m

pt om

s A

tt em

pt , d

ea th

T ui

ks u

et a

l. (

20 14

) 13

7 F

in ni

sh p

at ie

nt s

w it

h m

oo d

di so

rd er

C li

ni ca

l A

do le

sc en

t 84

A nx

ie ty

s ym

pt om

s A

tt em

pt

V al

to ne

n et

a l.

( 20

06 )

16 0

F in

ni sh

p at

ie nt

s w

it h

bi po

la r

di so

rd er

C li

ni ca

l A

du lt

18 A

ny a

nx ie

ty d

ia gn

os is

, an

xi et

y sy

m pt

om s

A tt

em pt

V al

to ne

n et

a l.

( 20

08 )

17 6

F in

ni sh

p at

ie nt

s w

it h

bi po

la r

di so

rd er

C li

ni ca

l A

du lt

18 A

ny a

nx ie

ty d

ia gn

os is

A tt

em pt

W ed

ig e

t al

. ( 20

12 )

23 1

pa ti

en ts

w it

h B

P D

C li

ni ca

l A

du lt

19 2

P T

S D

A tt

em pt

W en

ze l

et a

l. (

20 11

) 29

7 in

pa ti

en ts

w it

h su

ic id

al it

y H

is to

ry o

f S

IT B

s A

du lt

36 0

A ny

a nx

ie ty

d ia

gn os

is D

ea th

W ol

k, W

ei ss

m an

( 19

96 )

22 6

pa ti

en ts

w it

h de

pr es

si on

a nd

/o r

an xi

et y,

o r

co nt

ro ls

G en

er al

A do

le sc

en t

20 4

A ny

a nx

ie ty

d ia

gn os

is A

tt em

pt , d

ea th

Y as

ee n

et a

l. (

20 12

) 2,

86 4

pa ti

en ts

w it

h pa

st y

ea r

m aj

or

de pr

es si

ve e

pi so

de C

li ni

ca l

A du

lt 36

G A

D , S

O C

, S P

E C

, p an

ic

at ta

ck Id

ea ti

on , a

tt em

pt

Y en

e t

al . (

20 03

) 57

8 pa

ti en

ts w

it h

pe rs

on al

it y

di so

rd er

C li

ni ca

l A

du lt

24 P

D , P

T S

D A

tt em

pt

N o te

. A D

J= ad

ju st

m en

t di

so rd

er ;A

G =

ag or

ap ho

bi a;

A N

X -N

O S

= an

xi et

y di

so rd

er n

ot o

th er

w is

e sp

ec if

ie d;

G A

D =

ge ne

ra li

ze d

an xi

et y

di so

rd er

; O

C D

= ob

se ss

iv e–

co m

pu ls

iv e

di so

rd er

; P

D =

pa ni

c di

so rd

er ;

P D

A =

pa ni

c di

so rd

er w

it h

ag or

ap ho

bi a;

P T

S D

= po

st tr

au m

at ic

s tr

es s

di so

rd er

; S

O C

= so

ci al

a nx

ie ty

d is

or de

r; S

P E

C =

sp ec

if ic

p ho

bi a;

S O

M =

so m

at ic

s ym

pt om

d is

or de

r.

a G

en er

al =

ge ne

ra l

co m

m un

it y;

C li

ni ca

l= ou

tp at

ie nt

o r

in pa

ti en

t sa

m pl

e; H

is to

ry o

f S

IT B

s= hi

st or

y of

s ui

ci da

l or

n on

su ic

id al

s el

f- in

ju ri

ou s

th ou

gh ts

a nd

/o r

be ha

vi or

s.

b R

ef er

s to

a ge

o f

sa m

pl e

at b

as el

in e.

c A lt

ho ug

h no

e xp

li ci

t de

fi ni

ti on

w as

p ro

vi de

d by

s tu

dy a

ut ho

rs , a

nx ie

ty n

eu ro

se s

w er

e di

st in

gu is

he d

fr om

o th

er r

el at

ed d

is or

de r

ca te

go ri

es (

e. g.

, a ff

ec ti

ve d

is or

de rs

, s ub

st an

ce u

se d

is or

de rs

).

d D

ef in

ed b

y st

ud y

au th

or s

as a

cu te

c ri

si s

re ac

ti on

.

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Bentley et al. Page 34

T a b

le 2

O ve

ra ll

e ff

ec t

si ze

s fo

r an

y an

xi et

y co

ns tr

uc t

(d ia

gn os

is a

nd s

ym pt

om s)

p re

di ct

in g

su ic

id e-

re la

te d

ou tc

om es

.

# of

i n

cl u

d ed

O R

c as

es O

ve ra

ll O

R (

95 %

C I)

p va

lu e

z I2

( %

) #

of i

n cl

u d

ed H

R c

as es

O ve

ra ll

H R

( 95

% C

I) p

va lu

e z

I2 (

% )

Id ea

ti on

26 1.

49 (

1. 18

, 1 .8

8) <

.0 1

3. 39

85 .5

7 –

– –

–

A tt

em pt

80 1.

64 (1

.4 7,

1 .8

3) <

.0 01

8. 90

76 .9

7 10

1. 04

( 0.

96 , 1

.1 1)

.3 6

0. 93

71 .2

0

D ea

th 39

1. 01

( 0.

87 , 1

.1 8)

.8 8

0. 16

46 .9

5 10

1. 52

( 1.

28 , 1

.8 1)

b. 00

1 4.

79 92

.9 4

N o te

. T he

re w

er e

no H

R p

re di

ct io

n ca

se s

fo r

su ic

id e

id ea

ti on

.

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Bentley et al. Page 35

T a b

le 3

O ve

ra ll

O dd

s ra

ti os

f or

a nx

ie ty

s ym

pt om

s pr

ed ic

ti ng

s ui

ci de

-r el

at ed

o ut

co m

es .

# of

i n

cl u

d ed

c as

es O

ve ra

ll O

R (

95 %

C I)

p va

lu e

z I2

( %

)

Id ea

ti on

7 1.

54 (

1. 29

, 1 .8

4) <

.0 01

4. 72

28 .3

8

A tt

em pt

9 1.

39 (

1. 17

, 1 .6

5) <

.0 01

3. 80

83 .1

2

D ea

th 12

1. 20

( 0.

95 , 1

.5 2)

.1 3

1. 51

6. 40

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Bentley et al. Page 36

T a b

le 4

O ve

ra ll

o dd

s ra

ti os

f or

a nx

ie ty

d ia

gn os

es p

re di

ct in

g su

ic id

e- re

la te

d ou

tc om

es .

# of

c as

es O

ve ra

ll O

R (

95 %

C I)

p va

lu e

z I2

( %

)

A n

y ty

p e

o f

a n xi

et y

d ia

g n o si

s

Id ea

ti on

19 1.

45 (

1. 05

, 2 .0

0) <

.0 5

2. 26

89 .0

0

A tt

em pt

71 1.

58 (

1. 36

, 1 .8

4) <

.0 01

5. 90

69 .7

7

D ea

th 27

0. 93

( 0.

77 , 1

.1 3)

.4 8

− 0.

70 54

.4 4

A d

ju st

m en

t d is

o rd

er

A tt

em pt

4 0.

33 (

0. 16

, 0 .6

9) <

.0 1

− 2.

94 0.

00

D ea

th 5

0. 24

( 0.

09 , 0

.6 7)

< .0

1 −

2. 74

0. 00

A g

o ra

p h o b ia

A tt

em pt

4 0.

99 (

0. 55

, 1 .7

8) .9

8 −

0. 03

0. 00

G A

D

Id ea

ti on

2 1.

70 (

1. 18

, 2 .4

6) <

.0 1

2. 84

53 .8

4

A tt

em pt

7 1.

27 (

0. 78

, 2 .0

8) .3

4 0.

96 77

.9 4

O C

D

A tt

em pt

2 0.

95 (

0. 28

, 3 .1

8) .9

3 −

0. 08

0. 00

D ea

th 2

0. 23

( 0.

03 , 1

.6 9)

.1 5

− 1.

44 0.

00

P a

n ic

d is

o rd

er

A tt

em pt

9 1.

96 (

1. 38

, 2 .7

9) <

.0 01

3. 76

41 .5

5

D ea

th 4

1. 09

( 0.

39 , 3

.0 4)

.8 7

0. 16

0. 00

P T

S D

Id ea

ti on

4 2.

25 (

1. 46

, 3 .4

7) <

.0 01

3. 67

50 .6

2

A tt

em pt

10 2.

07 (

1. 49

, 2 .8

8) <

.0 01

4. 31

56 .8

7

S o ci

a l

a n xi

et y

d is

o rd

er

Id ea

ti on

4 1.

38 (

1. 10

, 1 .7

2) <

.0 1

2. 83

0. 00

A tt

em pt

7 1.

67 (

1. 25

, 2 .2

3) <

.0 1

3. 45

0. 00

S o m

a ti

c sy

m p to

m s

d is

o rd

er

A tt

em pt

2 1.

85 (

0. 88

, 3 .9

0) .1

1 1.

62 0.

00

S p ec

if ic

p h o b ia

Id ea

ti on

4 1.

45 (

1. 16

, 1 .8

1) <

.0 1

3. 26

0. 39

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Bentley et al. Page 37

# of

c as

es O

ve ra

ll O

R (

95 %

C I)

p va

lu e

z I2

( %

)

A tt

em pt

5 1.

43 (

0. 91

, 2 .2

4) .1

2 1.

55 50

.4 9

N o te

. G A

D =

ge ne

ra li

ze d

an xi

et y

di so

rd er

; O

C D

= ob

se ss

iv e–

co m

pu ls

iv e

di so

rd er

; P

T S

D =

po st

tr au

m at

ic s

tr es

s di

so rd

er .

Clin Psychol Rev. Author manuscript; available in PMC 2016 February 29.