Hostname: page-component-745bb68f8f-v2bm5 Total loading time: 0 Render date: 2025-02-05T19:24:29.395Z Has data issue: false hasContentIssue false
  • English
  • Français

Suicidal ideation and subsequent completed suicide in both psychiatric and non-psychiatric populations: a meta-analysis

Published online by Cambridge University Press:  19 December 2016

A. A. M. Hubers ,
S. Moaddine ,
S. H. M. Peersmann ,
T. Stijnen ,
E. van Duijn ,
R. C. van der Mast ,
O. M. Dekkers  and
E. J. Giltay
A. A. M. Hubers*
Affiliation:
Department of Psychiatry, Leiden University Medical Center, Leiden, The Netherlands
S. Moaddine
Affiliation:
Department of Psychiatry, Leiden University Medical Center, Leiden, The Netherlands
S. H. M. Peersmann
Affiliation:
Department of Psychiatry, Leiden University Medical Center, Leiden, The Netherlands
T. Stijnen
Affiliation:
Department of Medical Statistics & Bioinformatics, Leiden University Medical Center, Leiden, The Netherlands
E. van Duijn
Affiliation:
Department of Psychiatry, Leiden University Medical Center, Leiden, The Netherlands Center for Mental Health Care Delfland, Delft, The Netherlands
R. C. van der Mast
Affiliation:
Department of Psychiatry, Leiden University Medical Center, Leiden, The Netherlands Faculty of Medicine, Collaborative Antwerp Psychiatric Research Institute (CAPRI), University of Antwerp, Antwerp, Belgium
O. M. Dekkers
Affiliation:
Department of Epidemiology, Leiden University Medical Center, Leiden, The Netherlands Department of Endocrinology, Leiden University Medical Center, Leiden, The Netherlands Department of Clinical Epidemiology, Aarhus University Hospital, Aarhus, Denmark
E. J. Giltay
Affiliation:
Department of Psychiatry, Leiden University Medical Center, Leiden, The Netherlands
*
*Address for correspondence: A. A. M. Hubers, Department of Psychiatry, Leiden University Medical Center, Leiden, The Netherlands. (Email: a.a.m.hubers@lumc.nl)
Rights & Permissions [Opens in a new window]

Abstract

Aims.

Several authors claimed that expression of suicidal ideation is one of the most important predictors of completed suicide. However, the strength of the association between suicidal ideation and subsequent completed suicide has not been firmly established in different populations. Furthermore, the absolute suicide risk after expression of suicidal ideation is unknown. In this meta-analysis, we examined whether the expression of suicidal ideation predicted subsequent completed suicide in various populations, including both psychiatric and non-psychiatric populations.

Methods.

A meta-analysis of cohort and case–control studies that assessed suicidal ideation as determinant for completed suicide in adults. Two independent reviewers screened 5726 articles for eligibility and extracted data of the 81 included studies. Pooled risk ratios were estimated in a random effects model stratified for different populations. Meta-regression analysis was used to determine suicide risk during the first year of follow-up.

Results.

The risk for completed suicide was clearly higher in people who had expressed suicidal ideation compared with people who had not, with substantial variation between the different populations: risk ratio ranging from 2.35 (95% confidence interval (CI) 1.43–3.87) in affective disorder populations to 8.00 (95% CI 5.46–11.7) in non-psychiatric populations. In contrast, the suicide risk after expression of suicidal ideation in the first year of follow-up was higher in psychiatric patients (risk 1.40%, 95% CI 0.74–2.64) than in non-psychiatric participants (risk 0.23%, 95% CI 0.10–0.54). Past suicide attempt-adjusted risk ratios were not pooled due to large underreporting.

Conclusions.

Assessment of suicidal ideation is of priority in psychiatric patients. Expression of suicidal ideation in psychiatric patients should prompt secondary prevention strategies to reduce their substantial increased risk of suicide.

Keywords

Epidemiologyinpatient psychiatryoutpatient psychiatrysuicidesystematic reviews
Type
Original Articles
Information
Epidemiology and Psychiatric Sciences , Volume 27 , Issue 2 , April 2018 , pp. 186 - 198
Copyright
Copyright © Cambridge University Press 2016 

Introduction

Almost 90% of those who died by suicide contacted a health care professional in the 3 months prior to their death (De Leo et al. Reference De Leo, Draper, Snowdon and Kolves2013a ). Of those who had contact with a health care professional in the four weeks before their death, 22% expressed suicidal intent (Isometsa et al. Reference Isometsa, Heikkinen, Marttunen, Henriksson, Aro and Lonnqvist1995). Several authors claim that expression of suicidal ideation is one of the most important predictors for completed suicide (van Heeringen & Hengeveld, Reference van Heeringen, Hengeveld, Hengeveld and van Balkom2009; Posner et al. Reference Posner, Brown, Stanley, Brent, Yershova, Oquendo, Currier, Melvin, Greenhill, Shen and Mann2011; Batterham et al. Reference Batterham, Calear, Mackinnon and Christensen2013), but this association has not been firmly established (Large & Nielssen, Reference Large and Nielssen2012) with inconsistent previous reports. Most importantly, the risk of dying by suicide after expression of suicidal ideation is currently unknown.

Some of the previous studies reported a strong association between suicidal ideation and completed suicide (Taiminen et al. Reference Taiminen, Huttunen, Heilä, Henriksson, Isometsä, Kähkönen, Tuominen, Lönnqvist, Addington and Helenius2001; McGirr et al. Reference McGirr, Renaud, Seguin, Alda, Benkelfat, Lesage and Turecki2007; De Leo et al. Reference De Leo, Draper, Snowdon and Kolves2013b ), whereas others found no association (Bradvik &  Berglund, Reference Bradvik and Berglund1993; Bertelsen et al. Reference Bertelsen, Jeppesen, Petersen, Thorup, Øhlenschlaeger, le Quach, Christensen, Krarup, Jørgensen and Nordentoft2007; Jollant et al. Reference Jollant, Malafosse, Docto and Macdonald2014). Previous meta-analyses (Large et al., Reference Large, Smith, Sharma, Nielssen and Singh2011a , Reference Large, Sharma, Cannon, Ryan and Nielssen b ; Hawton et al., Reference Hawton, Sutton, Haw, Sinclair and Deeks2005a , Reference Hawton, Sutton, Haw, Sinclair and Harriss b ; Reference Hawton, Casanas, Haw and Saunders2013; Chapman et al., Reference Chapman, Mullin, Ryan, Kuffel, Nielssen and Large2015) that assessed this association also reported inconsistent results, with odds ratios ranging from 1.5 in bipolar disorder (Hawton et al. Reference Hawton, Sutton, Haw, Sinclair and Harriss2005b ) to 29.8 in schizophrenia (Hawton et al. Reference Hawton, Sutton, Haw, Sinclair and Deeks2005a ). These meta-analyses only investigated the effect of suicidal ideation within specific populations, without comparing effects between different populations or exploring other sources of clinical heterogeneity. Furthermore, the effects in the general population and several other psychiatric populations remain unknown (Runeson, Reference Runeson2002; Conwell, Reference Conwell2009).

Suicidal ideation is quite common, with a lifetime prevalence of 10% in the general population (Nock et al. Reference Nock, Borges, Bromet, Alonso, Angermeyer, Beautrais, Bruffaerts, Chiu, de Girolamo, Gluzman, de Graaf, Gureje, Haro, Huang, Karam, Kessler, Lepine, Levinson, Medina-Mora, Ono, Posada-Villa and Williams2008). Clinicians can only use suicidal ideation in their clinical risk assessment when they know the absolute risks of suicide for those who do and those who do not express suicidal ideation (Grobbee & Hoes, Reference Grobbee, Hoes, Grobbee and Hoes2009). However, such summary statistics have not been provided by the aforementioned meta-analyses (Large et al., Reference Large, Smith, Sharma, Nielssen and Singh2011a , Reference Large, Sharma, Cannon, Ryan and Nielssen b ; Hawton et al., Reference Hawton, Sutton, Haw, Sinclair and Deeks2005a , Reference Hawton, Sutton, Haw, Sinclair and Harriss b , Reference Hawton, Casanas, Haw and Saunders2013; Chapman et al., Reference Chapman, Mullin, Ryan, Kuffel, Nielssen and Large2015). Furthermore, suicidal ideation is strongly correlated with other predictors of completed suicide like previous suicide attempts (Harris & Barraclough, Reference Harris and Barraclough1997; ten Have et al. Reference ten Have, de Graaf, van Dorsselaer, Verdurmen, van ‘t Land, Vollebergh and Beekman2009) and it is not known whether it independently predicts completed suicide (Large & Nielssen, Reference Large and Nielssen2012).

Therefore, we conducted a systematic review and meta-analysis to assess whether the expression of suicidal ideation predicted subsequent completed suicide in various populations, including non-psychiatric and psychiatric populations. Secondly, we aimed to estimate the absolute risks of suicide after expression of suicidal ideation in these populations and to investigate whether the expression of suicidal ideation predicted subsequent completed suicide independent of the presence of past suicide attempts.

Method

Search strategy

Ten electronic databases (PubMed, Embase, Web of science, PsycINFO, PsycARTICLES, Psychology and behavioural sciences collection, Cochrane, CINAHL, Academic search premier and ScienceDirect) were searched until February 5, 2016 without language restrictions. A medical librarian was involved in formulating the search string (Supplement S1).

Eligibility criteria

Only journal articles fulfilling the following inclusion criteria were eligible for inclusion: (1) assessment of presence or absence of suicidal ideation as a distinct determinant (i.e. not combined with suicidal behaviour). Suicidal ideation was considered present when any form of ideation, ranging from death wish to suicide plans or threats, was expressed; (2) assessment of completed suicide (which could include open verdicts) as a distinct outcome measure; (3) comparison of suicidal ideation v. no suicidal ideation with respect to risk of subsequent completed suicide (4) cohort or case–control study design; and (5) mean age of the study population ≥18 years.

Next, the following exclusion criteria were applied: (1) presence of suicidal ideation was assessed after a suicide attempt; (2) comparison of suicidal ideation v. suicide attempt as determinant in a cohort study; or (3) comparison of those who died by suicide and those who survived an attempt as outcome in a case–control study.

For the assessment of the absolute risks of completed suicide, we included only cohort studies and nested case–control studies (with the size of the source population specified and random selection of controls from the source population) in which the number of suicides in the suicidal ideation group and the exposed person time could be extracted or estimated.

To determine whether suicidal ideation predicted subsequent completed suicide independent of the presence of past suicide attempts, articles that assessed the effect of suicidal ideation on subsequent completed suicide adjusted for previous suicide attempts were selected.

Study selection

All retrieved articles in the original search were screened independently by two of the three reviewers (A.A.M.H., S.M. and S.H.M.P.), first on title, then abstract and subsequently full-text evaluation to consider final eligibility. Disagreements with regard to final eligibility were discussed to reach consensus or, if necessary, another independent reviewer (E.J.G.) got involved. In addition, the reference lists of eligible articles and relevant review articles identified by the search strategy were examined by one of the reviewers to search for eligible studies. When multiple publications used (partially) overlapping study populations only the largest study or, when similar, the most recent study was included.

Data extraction

For each eligible article, two of the three reviewers (A.A.M.H. and S.M. or S.H.M.P.) independently extracted data using a standardised form. Disagreements were discussed or another independent reviewer (E.J.G.) got involved if needed. When information necessary to compute the effect size for the primary aim was missing, a request for the missing numbers was emailed to the corresponding author. In case of no response, the study was not included. Authors of eleven studies were emailed, two of them provided additional data and could be included in the meta-analysis (Appleby et al. Reference Appleby, Cooper, Amos and Faragher1999b ; Dutta et al. Reference Dutta, Murray, Allardyce, Jones and Boydell2011).

Risk of bias assessment

Two independent reviewers (A.A.M.H. and S.M. or S.H.M.P.) assessed four risk of bias aspects (Supplement S2), judged on the basis of adapted items from the Newcastle–Ottawa scale (Wells et al. Reference Wells, Shea, O'Connol, Peterson, Welch, Losos and Tugwelln.d.) and Altman (Reference Altman2001).

Statistical analyses

A study protocol was written a priori (Supplement S1), but was not published or registered. The primary outcome of this meta-analysis was the pooled unadjusted risk ratio for the association between suicidal ideation and completed suicide in a random effects model according to the method of DerSimonian and Laird (Borenstein et al. Reference Borenstein, Hedges, Higgins, Rothstein and Sharples2009a ). Due to the varying absolute risks of suicide among different populations (Nordentoft et al. Reference Nordentoft, Mortensen and Pedersen2011), the pooling of risk ratios was stratified for the following populations: affective disorders (including both in- and outpatients), (former) psychiatric inpatients (mixed diagnoses), schizophrenic patients (including both in- and outpatients), other mixed psychiatric populations (including people with substance use disorders, patients with borderline personality disorder and mixed diagnosis psychiatric outpatients [sometimes combined with inpatients]), general population and a residual category of non-psychiatric study populations that could not be grouped in one of the other categories. Studies in each subgroup were combined using a random effects model with separate estimates of tau-squared (Borenstein et al. Reference Borenstein, Hedges, Higgins, Rothstein and Sharples2009b ). To pool results from different studies, odds ratios, risk ratios and incidence rate ratios were considered to approach the same value, which is reasonable given the low risk of completed suicide. When only χ 2 values or p-values were given in combination with a direction of the effect, these were used to estimate risk ratios. If the given p-value was <0.05, we assumed a p-value of 0.049. There were no eligible studies only reporting a p-value >0.05. When articles reported on multiple suicidal ideation determinants (e.g. thoughts and plans separately) we computed a combined effect across these different determinants (Borenstein et al. Reference Borenstein, Hedges, Higgins, Rothstein and Sharples2009c ).

With regard to the risk of suicide after expression of suicidal ideation we were most interested in the suicide risk during the first year of follow-up. Since studies had varying follow-up times, and we did not expect a constant suicide rate over time, we conducted maximum-likelihood meta-regression analyses with mean study follow-up time as determinant and loge-transformed rates of completed suicide as outcome. Given the limited number of studies per population subgroup, analyses were stratified for psychiatric and non-psychiatric populations only. When one article assessed suicidal ideation at multiple time points, we only included the determinant that assessed suicidal ideation closest to baseline.

In order to assess whether suicidal ideation predicted completed suicide independently of previous attempts the past suicide attempt-adjusted risk ratios were extracted and used for estimation.

To check the robustness of the results, we restricted the analyses to: (1) studies with a low risk of bias (Supplement S2), (2) cohort studies (for the primary research aim only) and (3) studies that assessed suicidal ideation at baseline or in the preceding month (for the secondary research aim only). Only this last analysis was not pre-specified in the study protocol, but was conducted because of the large variation in suicidal ideation assessment periods, which could have influenced suicide risk. In addition, we stratified analyses of the primary outcome for studies with short (≤1-year) v. long (>1-year) timeframe between expression of suicidal ideation and completed suicide and suicidal ideation assessment method. Mean effects within subgroups were compared by testing for total between groups heterogeneity. Maximum likelihood meta-regression was applied to examine the effects of mean age and gender in the study populations.

Heterogeneity was assessed using the I 2 statistic, small study bias by inspecting the funnel plot and Egger's test for funnel plot asymmetry. A p-value <0.05 was considered statistically significant. Post-hoc, we changed the significance level for Egger's test for funnel plot asymmetry in the different subgroups to 0.10, given the limited number of studies that were included in these analyses and the relatively low power of the test. Statistical analyses were performed using Comprehensive Meta-analysis software version 2.0.064 (www.meta-analysis.com).

Results

Literature search and study characteristics

The literature search identified 5726 unique articles: 80 articles were finally included, which presented results of 81 studies (Fig. 1). The majority of the articles was published in English (n = 75), others in German (n = 2), Spanish (n = 1), French (n = 1) or South African (n = 1). Together these studies included 4 601 378 participants (median 216; range 14–4 045 993), with a total of 7729 completed suicides (median 60; range 3–1429). Suicidal ideation was assessed retrospectively by interviewing a next of kin or clinician of the deceased (n = 19), was extracted from medical records (n = 45) or was determined by asking the patient him/herself (n = 16) (missing: n = 1) (Supplement S3).

Fig. 1. Flowchart of the study selection process. Si indicates suicidal ideation; cs, completed suicide; sa, suicide attempts. Mixed psychiatric populations: n = 9; including three studies on people with substance use disorders, two on patients with borderline personality disorder, two on mixed diagnosis psychiatric in & outpatients, and two on mixed diagnosis psychiatric outpatients. Non-psychiatric residual category: n = 8; including two studies on military, one on veterans, one on stalkers, one on HIV infected males, one on survivors of childhood cancer, one on prisoners, and one on emergency department visitors.

The majority of included studies were (nested) case–control studies (n = 51, 63.0%). In only 26 studies the absolute risk of completed suicide could be extracted. None of the 81 eligible studies assessed the effect of suicidal ideation on completed suicide with sole adjustment for previous suicide attempts. However, there were 46 studies that applied a multivariable model, but only 28 adjusted for previous attempts as a separate determinant in this model.

Association between suicidal ideation and subsequent completed suicide

For all population subgroups investigated in this meta-analysis, the pooled risk of suicide was significantly higher in study participants who had expressed suicidal ideation compared with study participants who had not. Whereas the overall risk ratio (RR) was 4.17 (95% confidence interval (CI) 3.29–5.27), the risk ratios per population subgroup varied substantially. The risk ratio was highest in the non-psychiatric residual subgroup, followed by schizophrenia, mixed psychiatric patients, the general population, psychiatric inpatients and affective disorder (Fig. 2). Overall, the risk ratio was significantly higher in non-psychiatric populations (RR 6.61; 95% CI 4.62–9.47) than in psychiatric populations (RR 3.53; 95% CI 2.81–4.44) (p-value = 0.004) (Fig. 3).

Fig. 2. Forest plot showing the association between suicidal ideation and subsequent completed suicide. CI, confidence interval. aAuthors provided additional data to compute effect size.

Fig. 3. Bar chart showing suicide risk ratio and the suicide risk (%) during first year of follow-up. Lines indicate 95% confidence interval. Numbers of studies included in (A) and (B) differ because in only 26 of the 81 studies the absolute risk of completed suicide could be extracted. A. Risk ratio for suicide after expression of suicidal ideation, stratified for psychiatric and non-psychiatric populations. B. Results specified separately for study participants who had expressed suicidal ideation and study participants who had not, stratified for psychiatric and non-psychiatric populations. Percentages calculated using maximum likelihood meta-regression analyses with mean study follow-up time as determinant and loge-transformed rates of completed suicide as outcome (see Fig. 4).

Fig. 4. Meta-regression of mean study follow-up time on suicide rate. Maximum likelihood meta-regression analyses with mean study follow-up time as determinant and loge-transformed rates of completed suicide as outcome presented on logarithmic scales. The bubble size is proportional to the study's weight. Vertical line indicates 1 year follow-up. The suicide risk can be calculated using the following formulas: suicide risk (during first year of follow-up) = 1 – e(-rate per person-year at year 1). Figure part A, B: Psychiatric subgroup included: Dingman & McGlashan, Reference Dingman and McGlashan1986; Drake & Cotton, Reference Drake and Cotton1986; Allebeck et al. Reference Allebeck, Varla, Kristjansson and Wistedt1987; Goldstein et al. Reference Goldstein, Black, Nasrallah and Winokur1991; Bradvik & Berglund, Reference Bradvik and Berglund1993; Stephens et al. Reference Stephens, Richard and McHugh1999; Bioulac et al. Reference Bioulac, Bourgeois, Ekouevi, Bonnin, Gonzales and Castello2000; de Hert et al. Reference De Hert, McKenzie and Peuskens2001; Spiessl et al. Reference Spiessl, Hubner-Liebermann and Cording2002; Coryell & Young, Reference Coryell and Young2005; Dutta et al. Reference Dutta, Boydell, Kennedy, van Os, Fearon and Murray2007; Berg, Reference Berg2010; Dutta et al. Reference Dutta, Murray, Allardyce, Jones and Boydell2011; Kuo et al. Reference Kuo, Tsai, Liao, Conwell, Lin, Chang, Chen and Chen2011; Sani et al. Reference Sani, Tondo, Koukopoulos, Reginaldi, Kotzalidis, Koukopoulos, Manfredi, Mazzarini, Pacchiarotti, Simonetti, Ambrosi, Angeletti, Girardi and Tatarelli2011; Park et al. Reference Park, Choi, Kyoung Yi and Hong2013; Simon et al. Reference Simon, Rutter, Peterson, Oliver, Whiteside, Operskalski and Ludman2013. Lin et al. Reference Lin, Hung, Liao, Lee, Tsai, Chen and Kuo2014; Figure part C, D: Non-psychiatric subgroup included: Crandall et al. Reference Crandall, Fullerton-Gleason, Aguero and LaValley2006; Khang et al. Reference Khang, Kim and Cho2010; McEwan et al. Reference McEwan, Mullen and MacKenzie2010; Hyman, Reference Hyman2012; Brinkman et al. Reference Brinkman, Zhang, Recklitis, Kimberg, Zeltzer, Muriel, Stovall, Srivastava, Sklar, Robison and Krull2014; Kleiman & Liu, Reference Kleiman and Liu2014; Al-Sayegh et al. Reference Al-Sayegh, Lowry, Polur, Hines, Liu and Zhang2015; Werbeloff et al. Reference Werbeloff, Dohrenwend, Levav, Haklai, Yoffe, Large, Davidson and Weiser2015.

Risk of completed suicide

Although the psychiatric subgroups had the lowest risk ratios, meta-regression analyses showed the suicide risk was highest in the psychiatric subgroups who had expressed suicidal ideation (risk during first year of follow-up 1.40%; 95% CI 0.74–2.64). This was clearly higher than the suicide risk in psychiatric patients who had not expressed suicidal ideation (risk during first year of follow-up 0.41%; 95% CI 0.20–0.84). In non-psychiatric study participants the suicide risk during the first year of follow-up was 0.23% (95% CI 0.10–0.54) in participants who had expressed suicidal ideation and 0.04% (95% CI 0.01–0.13) in participants who had not (Fig. 3 and Fig. 4). When grouping all studies, the suicide risk during the first year of follow-up after expression of suicidal ideation was 1.03% (95% CI 0.43–2.44) (data not shown).

Risk of bias assessment and sensitivity analyses

When restricting the analyses to studies that scored ‘adequate’ on all four risk of bias items (n = 34, 42.0%), the pooled risk ratio decreased slightly to 3.41 (95% CI 2.67–4.35) and the overall suicide risk during the first year of follow-up also decreased slightly (risk during first year of follow-up 0.77%; 95% CI 0.27–2.16). In addition, restricting the primary analysis to cohort studies resulted in a slight decrease in the risk ratio (RR 3.10; 95% CI 2.22–4.32). Restricting the secondary analysis to studies that assessed suicidal ideation in the month around baseline resulted in a similar risk (risk during first year of follow-up 0.99%; 95% CI 0.39–2.51).

Is suicidal ideation an independent predictor for completed suicide?

Of all studies that used a multivariable model for the prediction of suicide and adjusted for (among other predictors) the presence of a previous suicide attempt (n = 28), only 14 reported the adjusted effect size of suicidal ideation. In 11 of these studies, suicidal ideation was a significant independent predictor of completed suicide. The other 14 studies that included previous attempts in their multivariable model either did not include suicidal ideation as a separate predictor (n = 6), or did include suicidal ideation but did not report the adjusted effect size (n = 8), which was non-significant in seven of these studies. While in at least 11 studies suicidal ideation was a significant independent predictor of completed suicide, the 14 adjusted effect sizes were not pooled due to large underreporting of mainly non-significant results.

Subgroup analyses

There was considerable heterogeneity (I 2 for overall risk ratio = 84.7%, Q-value = 523), even when stratified for specific populations (I 2 range 39–79%, Q-values 12.7–65.6). The results of subgroup analyses can be found in Supplement S4.

Publication bias

The overall funnel plot (Supplement S5) for the primary research question showed a rather symmetrical funnel plot. However, visual inspection of some of the funnel plots of the subgroups showed there might be missing studies with negative effects. The overall Egger's test indicated no funnel plot asymmetry (two-tailed p-value = 0.94). Only in the general population subgroup, the Egger's test for funnel plot asymmetry was significant (two-tailed p-value = 0.04).

Discussion

Meta-analysing 81 eligible studies showed that overall, people expressing suicidal ideation are four times more likely to die by suicide than people not expressing suicidal ideation, with the highest relative risks in non-psychiatric populations. However, absolute suicide risks, rather than relative risks, are essential in suicide prediction, and were highest in psychiatric populations, with a suicide risk during the first year of follow-up of 1.4% compared with 0.23% in non-psychiatric populations.

The results of this meta-analysis show an overall modest relationship, with moderate risk ratios for psychiatric subgroups and strong risk ratios for non-psychiatric subgroups (Rosenthal, Reference Rosenthal1996). The results are in line with previous meta-analyses on suicidal ideation and subsequent completed suicide (Large et al., Reference Large, Smith, Sharma, Nielssen and Singh2011a , Reference Large, Sharma, Cannon, Ryan and Nielssen b ; Hawton et al., Reference Hawton, Sutton, Haw, Sinclair and Deeks2005a , Reference Hawton, Casanas, Haw and Saunders2013; Chapman et al., Reference Chapman, Mullin, Ryan, Kuffel, Nielssen and Large2015), which only assessed psychiatric populations. The highest relative risk was found in a meta-analysis restricted to schizophrenic patients (Hawton et al. Reference Hawton, Sutton, Haw, Sinclair and Deeks2005a ; Chapman et al. Reference Chapman, Mullin, Ryan, Kuffel, Nielssen and Large2015) followed by meta-analyses restricted to psychiatric inpatients (Large et al., Reference Large, Smith, Sharma, Nielssen and Singh2011a , Reference Large, Sharma, Cannon, Ryan and Nielssen b ) and depressed patients (Hawton et al. Reference Hawton, Casanas, Haw and Saunders2013; Chapman et al. Reference Chapman, Mullin, Ryan, Kuffel, Nielssen and Large2015). Remarkably, in contrast to our study, one recent meta-analysis (Chapman et al. Reference Chapman, Mullin, Ryan, Kuffel, Nielssen and Large2015) reported that the association between suicidal ideation and completed suicide was not significant in patients with a mood disorder, probably explained by the broader definition of suicidal ideation in our study and the fact that we excluded studies that compared expression of suicidal ideation in suicide completers v. attempters. Unfortunately, none of the previous meta-analyses (Hawton et al., Reference Hawton, Sutton, Haw, Sinclair and Deeks2005a , Reference Hawton, Sutton, Haw, Sinclair and Harriss b ; Reference Hawton, Casanas, Haw and Saunders2013; Large et al., Reference Large, Smith, Sharma, Nielssen and Singh2011a , Reference Large, Sharma, Cannon, Ryan and Nielssen b ; Chapman et al., Reference Chapman, Mullin, Ryan, Kuffel, Nielssen and Large2015) assessed the absolute risks of suicide.

The expression of suicidal ideation was reported as a significant independent predictor of suicide more often than expected by chance. However, underreporting of non-significant effects refrained us from concluding to what extend suicidal ideation is an independent predictor of completed suicide. Many studies included in our meta-analysis did develop a prediction model for suicide, although it is widely accepted that it is very difficult to accurately predict suicide for an individual patient (van Hemert et al. Reference van Hemert, Kerkhof, de Keijser, Verwey, van Boven, Hummelen, de Groot, Lucassen, Meerdinkveldboom, Steendam, Stringer, Verlinde and van de Glind2012), and there is a lot of debate on whether or not clinical risk categorisation is helpful in preventing completed suicide (Large et al. Reference Large, Ryan and Nielssen2011c ; O'Connor et al. Reference O'Connor, Allan and Scott2014). The low baseline rate of suicide results in low positive predictive values and in the majority of suicides occurring in the group that is classified as low-risk according to the prediction model (Large et al. Reference Large, Sharma, Cannon, Ryan and Nielssen2011b ; Madsen & Nordentoft, Reference Madsen and Nordentoft2012; Paton et al. Reference Paton, Large and Ryan2014). It was striking that only 60% of the studies that developed a prediction model for suicide included previous suicide attempts as a predictor, whereas this has been reported as one of the most important independent predictors of completed suicide (Large et al. Reference Large, Ryan and Nielssen2011c ). Future studies on suicide risk assessment should therefore include known predictors from the literature (e.g. previous suicide attempts) rather than only selecting predictors by univariable screening in the developmental dataset (Steyerberg, Reference Steyerberg2009).

Strengths and limitations

A major strength of our paper was the attempt to provide risk estimates for different subgroups. Moreover, we strictly excluded studies that might have combined suicidal ideation with behaviour, and included many studies for the primary research aim. There are also limitations that need to be considered. First, the majority of studies only provided long-term follow-up results, while the short-term risk has most clinical value. Although the provided suicide risks should be interpreted with caution due to the declining rates over time and the limited number of studies, especially with short-term follow-up, that could be included in the meta-regression analyses, results clearly show the risk of suicide is higher after expression of suicidal ideation, especially in psychiatric patients. Future studies should focus on short-term suicide risk in the different psychiatric populations. Second, only part of the large amount of heterogeneity could be explained for. Included studies often did not provide several patient characteristics that would be interesting in exploring/explaining this heterogeneity, like duration of suicidal ideation (ten Have et al. Reference ten Have, van Dorsselaer and de Graaf2013) or behavioural traits (Turecki, Reference Turecki2014). We did not find a significant difference between suicidal ideation assessment subgroups, possible due to varying or lacking definitions of suicidal ideation and the large variation in suicidal ideation assessment methods within the subgroups. Unfortunately we could not further explore this variation due to insufficient reporting in the individual studies and study level confounding. Clear definitions of suicidal ideation and its assessment method are necessary (Silverman et al. Reference Silverman, Berman, Sanddal, O'Carroll and Joiner2007) and validated and standardised assessment methods should be used. Even within studies that used standardised assessment methods, the instruments used varied widely with only a maximum of two studies using the same instrument. This can also be a source of heterogeneity as different validated suicidal ideation assessment methods (Yigletu et al. Reference Yigletu, Tucker, Harris and Hatlevig2004; Vuorilehto et al. Reference Vuorilehto, Valtonen, Melartin, Sokero, Suominen and Isometsa2014) can result in different suicide ideation prevalences, even in the same population at the same time point. The considerable heterogeneity that still exists between studies within the different population subgroups, but that supposedly also exists within individual studies, limits the generalisability of our findings to individual patients and will result in higher suicide risks in some patients and lower suicide risks in others. We recommend future studies to explore sources of clinical heterogeneity in more detail, especially with regard to suicidal ideation assessment method. Third, study level confounding (Hingorani et al. Reference Hingorani, Windt, Riley, Abrams, Moons, Steyerberg, Schroter, Sauerbrei, Altman and Hemingway2013) might have influenced the results of our subgroup analyses, e.g. resulted in a significant effect of gender on the association between suicidal ideation and completed suicide, as the percentage of females was lower in certain high-risk population subgroups like schizophrenia. Fourth, some required study parameters could not be directly extracted from the original articles. For example the exposed person-years were conservatively estimated when possible, which could have resulted in an underestimation of the suicide risk. Fifth, reporting bias cannot be ruled out. The power of Egger's test is relatively low (Sterne et al. Reference Sterne, Egger, Moher, Higgins and Green2011) and effects in the different population subgroups might have been underestimated as we did not attempt to access unpublished data or grey literature and due to underreporting of negative and non-significant effects. Sixth, studies that were published after February 5, 2016 were not included in this meta-analysis. Our meta-analysis especially has limited power on analyses regarding absolute suicide risk in psychiatric and non-psychiatric populations and the independent predictive effect of suicidal ideation. A future update of this meta-analysis should focus on studies that address these research questions.

Implications for clinical practice

For clinical psychiatric care, it is important to inquire about suicidal ideation in psychiatric patients given the substantial risk after expression of suicidal ideation, especially in the first period after expression. Although the relative risk was highest in non-psychiatric populations, the absolute suicide risk after expression of suicidal ideation in non-psychiatric populations was lower than in psychiatric populations, given their lower baseline suicide risk. However, even in psychiatric patients, suicidal ideation assessment should be placed in perspective as the suicide risk during the first year of follow-up among psychiatric patients who had not expressed suicidal ideation was 0.41%, which was higher than the suicide risk in non-psychiatric persons who had expressed suicidal ideation. Expression of suicidal ideation is one of the factors that clinicians take into account when assessing suicide risk in an individual patient (Popovic et al. Reference Popovic, Benabarre, Crespo, Goikolea, González-Pinto, Gutiérrez-Rojas, Montes and Vieta2014), which should of course not merely be based on a ‘tick box’ approach that classifies patients as either at high or low risk on the basis of a list of risk factors (van Hemert et al. Reference van Hemert, Kerkhof, de Keijser, Verwey, van Boven, Hummelen, de Groot, Lucassen, Meerdinkveldboom, Steendam, Stringer, Verlinde and van de Glind2012; O'Connor et al. Reference O'Connor, Allan and Scott2014). The suicide risks provided in this meta-analysis can help clinicians in their clinical suicide risk assessment and stepped care approach, for example to decide whether burdensome secondary prevention measures such as (compulsory) admission are proportional to the risk of suicide.

In a patient who expresses suicidal ideation, follow-up inquiries are needed about the nature of these thoughts and the assessment of other known risk factors for suicide (Hawton & van Heeringen, Reference Hawton and van Heeringen2009). As patients will not always express suicidal ideation spontaneously (van Hemert et al. Reference van Hemert, Kerkhof, de Keijser, Verwey, van Boven, Hummelen, de Groot, Lucassen, Meerdinkveldboom, Steendam, Stringer, Verlinde and van de Glind2012), it is important to ask them about suicidal thoughts. Actually, asking patients about suicidal ideation can reduce future suicidal ideation and follow-up inquiries can benefit long-term mental health (Dazzi et al. Reference Dazzi, Gribble, Wessely and Fear2014). Patients often feel relieved by having an opportunity to talk about their suicidal ideas (Gliatto & Rai, Reference Gliatto and Rai1999), especially if clinicians take time, show empathy, and acknowledge the suicidal feelings (Talseth et al. Reference Talseth, Lindseth, Jacobsson and Norberg1999). Remarkably, in several of the included studies that assessed the presence of suicidal ideation in medical records, this information was missing. This means that it was unavailable to clinicians and appropriate care might have been unnecessarily withheld from these patients (King et al. Reference King, Baldwin, Sinclair, Baker, Campbell and Thompson2001b ).

Fortunately, the majority of patients with suicidal ideation do not die by suicide. Nevertheless, assessment of suicidal ideation is of priority in people with psychiatric illnesses and when a patient expresses suicidal ideation, prompt secondary prevention strategies are necessary to reduce their substantial increased risk of suicide. Future suicide prediction studies should focus on absolute suicide risks and the independent predictive effect of suicidal ideation.

Supplementary Material

The supplementary material for this article can be found at https://doi.org/10.1017/S2045796016001049

Acknowledgements

We would like to thank medical librarian J.W. Schoones, Walaeus Library, Leiden University Medical Center for his help in conducting the literature search. In addition, we would like to thank the authors that provided additional information on their studies.

Funding

This research received no specific grant from any funding agency, commercial or not-for-profit sectors.

Conflict of Interest

None.

Availability of Data and Materials

Data supporting our findings can be found in Fig. 2 and Supplement S3.

References

Allebeck, P, Varla, A, Kristjansson, E, Wistedt, B (1987). Risk factors for suicide among patients with schizophrenia. Acta Psychiatrica Scandinavica 76, 414–419.Google Scholar
Al-Sayegh, H, Lowry, J, Polur, RN, Hines, RB, Liu, F, Zhang, J (2015). Suicide history and mortality: a follow-up of a national cohort in the United States. Archives of Suicide Research 19, 35–47.CrossRefGoogle ScholarPubMed
Altman, DG (2001). Systematic reviews of evaluations of prognostic variables. British Medical Journal 323, 224–228.CrossRefGoogle ScholarPubMed
Appleby, L, Dennehy, JA, Thomas, CS, Faragher, EB, Lewis, G (1999 a). Aftercare and clinical characteristics of people with mental illness who commit suicide: a case-control study. Lancet 353, 1397–1400.CrossRefGoogle ScholarPubMed
Appleby, L, Cooper, J, Amos, T, Faragher, B (1999 b). Psychological autopsy study of suicides by people aged under 35. British Journal of Psychiatry 175, 168–174.CrossRefGoogle ScholarPubMed
Baader-Matthei, T, Richter, P, Mundt, C (2004). Suicides of psychiatric hospitalized patients and their risk factors. A case control study. Revista Chili Neuro-Psiquiatria 42, 293–316.Google Scholar
Batterham, PJ, Calear, AL, Mackinnon, AJ, Christensen, H (2013). The association between suicidal ideation and increased mortality from natural causes. Journal of Affective Disorders 150, 855–860.Google Scholar
Berg, JE (2010). Death by suicide long after electroconvulsive therapy. Is the sense of coherence test of Antonovsky a predictor of mortality from depression? Mental Illness 2, 11–15.CrossRefGoogle Scholar
Bertelsen, M, Jeppesen, P, Petersen, L, Thorup, A, Øhlenschlaeger, J, le Quach, P, Christensen, TØ, Krarup, G, Jørgensen, P, Nordentoft, M (2007). Suicidal behaviour and mortality in first-episode psychosis: the OPUS trial. British Journal of Psychiatry Suppl 51, s140–s146.Google Scholar
Bickley, H, Hunt, IM, Windfuhr, K, Shaw, J, Appleby, L, Kapur, N (2013). Suicide within two weeks of discharge from psychiatric inpatient care: a case-control study. Psychiatric Services 64, 653–659.Google Scholar
Bioulac, S, Bourgeois, M, Ekouevi, DK, Bonnin, JM, Gonzales, B, Castello, MF (2000). [Predictive factors of suicide? an 8-year-long prospective longitudinal study of 200 psychiatric inpatients]. Encephale 26, 1–7.Google Scholar
Borenstein, M, Hedges, LV, Higgins, JPT, Rothstein, HR (2009 a). Chapter 12. Random-effects model. In Introduction to Meta-Analysis (ed. Sharples, K), pp. 69–75. John Wiley & Sons, Ltd.: Chichester, UK.CrossRefGoogle Scholar
Borenstein, M, Hedges, LV, Higgins, JPT, Rothstein, HR (2009 b). Chapter 19. Subgroep analyses. In Introduction to Meta-Analysis (ed. Sharples, K), pp. 149–186. John Wiley & Sons, Ltd.: Chichester, UK.Google Scholar
Borenstein, M, Hedges, LV, Higgins, JPT, Rothstein, HR (2009 c). Chapter 24. Mutliple outcomes or time-points within a study. In Introduction to Meta-Analysis (ed. Sharples, K), pp. 225–238. John Wiley & Sons, Ltd.: Chichester, UK.Google Scholar
Bradvik, L, Berglund, M (1993). Risk factors for suicide in melancholia. A case-record evaluation of 89 suicides and their controls. Acta Psychiatrica Scandinavica 87, 306–311.CrossRefGoogle ScholarPubMed
Brinkman, TM, Zhang, N, Recklitis, CJ, Kimberg, C, Zeltzer, LK, Muriel, AC, Stovall, M, Srivastava, DK, Sklar, CA, Robison, LL, Krull, KR (2014). Suicide ideation and associated mortality in adult survivors of childhood cancer. Cancer 120, 271–277.Google Scholar
Brown, GK, Beck, AT, Steer, RA, Grisham, JR (2000). Risk factors for suicide in psychiatric outpatients: a 20-year prospective study. Journal of Consulting and Clinical Psychology 68, 371–377.Google Scholar
Bukstein, OG, Brent, DA, Perper, JA, Moritz, G, Baugher, M, Schweers, J, Roth, C, Balach, L (1993). Risk factors for completed suicide among adolescents with a lifetime history of substance abuse: a case-control study. Acta Psychiatrica Scandinavica 88, 403–408.CrossRefGoogle ScholarPubMed
Castle, K, Duberstein, PR, Meldrum, S, Conner, KR, Conwell, Y (2004). Risk factors for suicide in blacks and whites: an analysis of data from the 1993 National Mortality Followback Survey. American Journal of Psychiatry 161, 452–458.Google Scholar
Chapman, CL, Mullin, K, Ryan, CJ, Kuffel, A, Nielssen, O, Large, MM (2015). Meta-analysis of the association between suicidal ideation and later suicide among patients with either a schizophrenia spectrum psychosis or a mood disorder. Acta Psychiatrica Scandinavica 131, 162–173.CrossRefGoogle ScholarPubMed
Cheng, KK, Leung, CM, Lo, WH, Lam, TH (1990). Risk factors of suicide among schizophrenics. Acta Psychiatrica Scandinavica 81, 220–224.CrossRefGoogle ScholarPubMed
Colman, I, Kingsbury, M, Sareen, J, Bolton, J, van Walraven, C (2016). Migraine headache and risk of self-harm and suicide: a population-based study in Ontario, Canada. Headache 56, 132–140.Google Scholar
Conlon, L, Garland, M, Prescott, P, Mannion, L, Leonard, M, Fahy, TJ (2007). Psychiatric aftercare and suicide risk: a case-control study using blind rating. Archives of Suicide Research 11, 291–295.Google Scholar
Conwell, Y (2009). Suicide prevention in later life: a glass half full, or half empty? American Journal of Psychiatry 166, 845–848.CrossRefGoogle ScholarPubMed
Coryell, W, Young, EA (2005). Clinical predictors of suicide in primary major depressive disorder. Journal of Clinical Psychiatry 66, 412–417.Google Scholar
Coryell, W, Kriener, A, Butcher, B, Nurnberger, J, McMahon, F, Berrettini, W, Fiedorowicz, J (2016). Risk factors for suicide in bipolar I disorder in two prospectively studied cohorts. Journal of Affective Disorders 190, 1–5.Google Scholar
Crandall, C, Fullerton-Gleason, L, Aguero, R, LaValley, J (2006). Subsequent suicide mortality among emergency department patients seen for suicidal behavior. Academic Emergency Medicine 13, 435–442.Google Scholar
Dazzi, T, Gribble, R, Wessely, S, Fear, NT (2014). Does asking about suicide and related behaviours induce suicidal ideation? What is the evidence? Psychological Medicine 44, 3361–3363.CrossRefGoogle ScholarPubMed
De Hert, M, McKenzie, K, Peuskens, J (2001). Risk factors for suicide in young people suffering from schizophrenia: a long-term follow-up study. Schizophrenia Research 47, 127–134.Google Scholar
De Leo, D, Draper, BM, Snowdon, J, Kolves, K (2013 a). Contacts with health professionals before suicide: missed opportunities for prevention? Comprehensive Psychiatry 54, 1117–1123.CrossRefGoogle ScholarPubMed
De Leo, D, Draper, BM, Snowdon, J, Kolves, K (2013 b). Suicides in older adults: a case-control psychological autopsy study in Australia. Journal of Psychiatric Research 47, 980–988.Google Scholar
Didham, R, Dovey, S, Reith, D (2006). Characteristics of general practitioner consultations prior to suicide: a nested case-control study in New Zealand. New Zealand Medical Journal 119, U2358.Google Scholar
Dingman, CW, McGlashan, TH (1986). Discriminating characteristics of suicides. Chestnut Lodge follow-up sample including patients with affective disorder, schizophrenia and schizoaffective disorder. Acta Psychiatrica Scandinavica 74, 91–97.CrossRefGoogle ScholarPubMed
Dong, JY, Ho, TP, Kan, CK (2005). A case-control study of 92 cases of in-patient suicides. Journal of Affective Disorders 87, 91–99.CrossRefGoogle ScholarPubMed
Drake, RE, Cotton, PG (1986). Depression, hopelessness and suicide in chronic schizophrenia. British Journal of Psychiatry 148, 554–559.CrossRefGoogle ScholarPubMed
Dutta, R, Boydell, J, Kennedy, N, van Os, J, Fearon, P, Murray, RM (2007). Suicide and other causes of mortality in bipolar disorder: a longitudinal study. Psychological Medicine 37, 839–847.Google Scholar
Dutta, R, Murray, RM, Allardyce, J, Jones, PB, Boydell, J (2011). Early risk factors for suicide in an epidemiological first episode psychosis cohort. Schizophrenia Research 126, 11–19.Google Scholar
Farberow, NL, Shneidman, ES, Neuringer, C (1966). Case history and hospitalization factors in suicides of neuropsychiatric hospital patients. Journal of Nervous and Mental Disease 142, 32–44.Google Scholar
Fawcett, J, Scheftner, WA, Fogg, L, Clark, DC, Young, MA, Hedeker, D, Gibbons, R (1990). Time-related predictors of suicide in major affective disorder. American Journal of Psychiatry 147, 1189–1194.Google ScholarPubMed
Flood, RA, Seager, CP (1968). A retrospective examination of psychiatric case records of patients who subsequently committed suicide. British Journal of Psychiatry 114, 443–450.CrossRefGoogle ScholarPubMed
Fruehwald, S, Matschnig, T, Koenig, F, Bauer, P, Frottier, P (2004). Suicide in custody: case-control study. British Journal of Psychiatry 185, 494–498.Google Scholar
Funahashi, T, Ibuki, Y, Domon, Y, Nishimura, T, Akehashi, D, Sugiura, H (2000). A clinical study on suicide among schizophrenics. Psychiatry and Clinical Neurosciences 54, 173–179.Google Scholar
Gliatto, MF, Rai, AK (1999). Evaluation and treatment of patients with suicidal ideation. American Family Physician 59, 1500–1506.Google ScholarPubMed
Goldstein, RB, Black, DW, Nasrallah, A, Winokur, G (1991). The prediction of suicide. Sensitivity, specificity, and predictive value of a multivariate model applied to suicide among 1906 patients with affective disorders. Archives General Psychiatry 48, 418–422.CrossRefGoogle ScholarPubMed
Grobbee, DE, Hoes, AW (2009). Diagnostic research. In Clinical Epidemiology. Principles, Methods, and Applications for Clinical Research (ed. Grobbee, DE and Hoes, AW), pp. 58–102. Jones and Bartlett Publishers: Sadbury, Massachusetts.Google Scholar
Harris, EC, Barraclough, B (1997). Suicide as an outcome for mental disorders. A meta-analysis. British Journal of Psychiatry 170, 205–228.Google Scholar
Hawton, K, van Heeringen, K (2009). Suicide. Lancet 373, 1372–1381.Google Scholar
Hawton, K, Sutton, L, Haw, C, Sinclair, J, Deeks, JJ (2005 a). Schizophrenia and suicide: systematic review of risk factors. British Journal of Psychiatry 187, 9–20.CrossRefGoogle ScholarPubMed
Hawton, K, Sutton, L, Haw, C, Sinclair, J, Harriss, L (2005 b). Suicide and attempted suicide in bipolar disorder: a systematic review of risk factors. Journal of Clinical Psychiatry 66, 693–704.CrossRefGoogle ScholarPubMed
Hawton, K, Casanas, IC, Haw, C, Saunders, K (2013). Risk factors for suicide in individuals with depression: a systematic review. Journal of Affective Disorders 147, 17–28.Google Scholar
Hingorani, AD, Windt, DA, Riley, RD, Abrams, K, Moons, KG, Steyerberg, EW, Schroter, S, Sauerbrei, W, Altman, DG, Hemingway, H, PROGRESS Group (2013). Prognosis research strategy (PROGRESS) 4: stratified medicine research. British Medical Journal 346, e5793.Google Scholar
Hoyer, EH, Licht, RW, Mortensen, PB (2009). Risk factors of suicide in inpatients and recently discharged patients with affective disorders. A case-control study. European Psychiatry 24, 317–321.Google Scholar
Hunt, IM, Kapur, N, Webb, R, Robinson, J, Burns, J, Turnbull, P, Shaw, J, Appleby, L (2007). Suicide in current psychiatric in-patients: a case-control study The National Confidential Inquiry into Suicide and Homicide. Psychological Medicine 37, 831–837.Google Scholar
Hunt, IM, Kapur, N, Webb, R, Robinson, J, Burns, J, Shaw, J, Appleby, L (2009). Suicide in recently discharged psychiatric patients: a case-control study. Psychological Medicine 39, 443–449.Google Scholar
Hunt, IM, Bickley, H, Windfuhr, K, Shaw, J, Appleby, L, Kapur, N (2013). Suicide in recently admitted psychiatric in-patients: a case-control study. Journal of Affective Disorders 144, 123–128.Google Scholar
Hyman, JRLL (2012). Suicide incidence and risk factors in an active duty US military population. American Journal of Public Health 102 (Suppl. 1), s138–s146.CrossRefGoogle Scholar
Isometsa, ET, Heikkinen, ME, Marttunen, MJ, Henriksson, MM, Aro, HM, Lonnqvist, JK (1995). The last appointment before suicide: is suicide intent communicated? American Journal of Psychiatry 152, 919–922.Google ScholarPubMed
Jollant, F, Malafosse, A, Docto, R, Macdonald, C (2014). A pocket of very high suicide rates in a non-violent, egalitarian and cooperative population of South-East Asia. Psychological Medicine 44, 2323–2329.Google Scholar
Kasckow, J, Liu, N, Haas, GL, Phillips, MR (2010). Case-control study of the relationship of depressive symptoms to suicide in a community-based sample of individuals with schizophrenia in China. Schizophrenia Research 122, 226–231.CrossRefGoogle Scholar
Kelly, DL, Shim, JC, Feldman, SM, Yu, Y, Conley, RR (2004). Lifetime psychiatric symptoms in persons with schizophrenia who died by suicide compared to other means of death. Journal of Psychiatric Research 38, 531–536.Google Scholar
Kessler, RC, Warner, CH, Ivany, C, Petukhova, MV, Rose, S, Bromet, EJ, Brown, M III, Cai, T, Colpe, LJ, Cox, KL, Fullerton, CS, Gilman, SE, Gruber, MJ, Heeringa, SG, Lewandowski-Romps, L, Li, J, Millikan-Bell, AM, Naifeh, JA, Nock, MK, Rosellini, AJ, Sampson, NA, Schoenbaum, M, Stein, MB, Wessely, S, Zaslavsky, AM, Ursano, RJ, for the Army STARRS Collaborators (2015). Predicting suicides after psychiatric hospitalization in US army soldiers. The army study to assess risk and resilience in servicemembers (army STARRS). JAMA Psychiatry 72, 49–57.CrossRefGoogle ScholarPubMed
Khang, YH, Kim, HR, Cho, SJ (2010). Relationships of suicide ideation with cause-specific mortality in a longitudinal study of South Koreans. Suicide Life Threatening Behaviour 40, 465–475.Google Scholar
Kim, HM, Smith, EG, Ganoczy, D, Walters, H, Stano, CM, Ilgen, MA, Bohnert, AS, Valenstein, M (2012). Predictors of suicide in patient charts among patients with depression in the Veterans Health Administration health system: importance of prescription drug and alcohol abuse. Journal of Clinical Psychiatry 73, e1269–e1275.Google Scholar
King, EA, Baldwin, DS, Sinclair, JM, Campbell, MJ (2001 a). The Wessex Recent In-Patient Suicide Study. 2. Case-control study of 59 in-patient suicides. British Journal of Psychiatry 178, 537–542.Google Scholar
King, EA, Baldwin, DS, Sinclair, JM, Baker, NG, Campbell, MJ, Thompson, C (2001 b). The Wessex Recent In-Patient Suicide Study. 1. Case-control study of 234 recently discharged psychiatric patient suicides. British Journal of Psychiatry 178, 531–536.Google Scholar
Kjelsberg, E, Eikeseth, PH, Dahl, AA (1991). Suicide in borderline patients--predictive factors. Acta Psychiatrica Scandinavica 84, 283–287.CrossRefGoogle ScholarPubMed
Kleiman, EM, Liu, RT (2014). Prospective prediction of suicide in a nationally representative sample: religious service attendance as a protective factor. British Journal of Psychiatry 204, 262–266.CrossRefGoogle Scholar
Kullgren, G (1988). Factors associated with completed suicide in borderline personality disorder. Journal of Nervous and Mental Disease 176, 40–44.CrossRefGoogle ScholarPubMed
Kuo, CJ, Tsai, SY, Liao, YT, Conwell, Y, Lin, SK, Chang, CL, Chen, CC, Chen, WJ (2011). Risk and protective factors for suicide among patients with methamphetamine dependence: a nested case-control study. Journal of Clinical Psychiatry 72, 487–493.CrossRefGoogle ScholarPubMed
Large, M, Smith, G, Sharma, S, Nielssen, O, Singh, SP (2011 a). Systematic review and meta-analysis of the clinical factors associated with the suicide of psychiatric in-patients. Acta Psychiatrica Scandinavica 124, 18–29.Google Scholar
Large, M, Sharma, S, Cannon, E, Ryan, C, Nielssen, O (2011 b). Risk factors for suicide within a year of discharge from psychiatric hospital: a systematic meta-analysis. Australian & New Zealand Journal of Psychiatry 45, 619–628.Google Scholar
Large, MM, Nielssen, O (2012). Suicidal ideation and later suicide (letter). American Journal of Psychiatry 169, 662.Google Scholar
Large, M, Ryan, C, Nielssen, O (2011 c). The validity and utility of risk assessment for inpatient suicide. Australasian Psychiatry 19, 507–512.CrossRefGoogle ScholarPubMed
Li, J, Ran, MS, Hao, Y, Zhao, Z, Guo, Y, Su, J, Lu, H (2008). Inpatient suicide in a Chinese psychiatric hospital. Suicide Life Threatening Behaviour 38, 449–455.CrossRefGoogle Scholar
Lin, SK, Hung, TM, Liao, YT, Lee, WC, Tsai, SY, Chen, CC, Kuo, CJ (2014). Protective and risk factors for inpatient suicides: a nested case-control study. Psychiatry Research 217, 54–59.Google Scholar
Lui, SY (2009). Risk factors for deliberate self-harm and completed suicide in young Chinese people with schizophrenia. Australian & New Zealand Journal of Psychiatry 43, 252–259.CrossRefGoogle ScholarPubMed
Lukaschek, K, Baumert, J, Krawitz, M, Erazo, N, Förstl, H, Ladwig, KH (2014). Determinants of completed railway suicides by psychiatric in-patients: case-control study. The British Journal of Psychiatry 205, 398–406.CrossRefGoogle ScholarPubMed
Madsen, T, Nordentoft, M (2012). Drs Madsen and Nordentoft reply. Journal of Clinical Psychiatry 73, 1034–1035.Google Scholar
Mahon, MJ, Tobin, JP, Cusack, DA, Kelleher, C, Malone, KM (2005). Suicide among regular-duty military personnel: a retrospective case-control study of occupation-specific risk factors for workplace suicide. American Journal of Psychiatry 162, 1688–1696.Google Scholar
McEwan, T, Mullen, P, MacKenzie, R (2010). Suicide among stalkers. Journal of Forensic Psychiatry and Psychology 21, 514–520.CrossRefGoogle Scholar
McGirr, A, Renaud, J, Seguin, M, Alda, M, Benkelfat, C, Lesage, A, Turecki, G (2007). An examination of DSM-IV depressive symptoms and risk for suicide completion in major depressive disorder: a psychological autopsy study. Journal of Affective Disorders 97, 203–209.Google Scholar
McManus, H, Petoumenos, K, Franic, T, Kelly, MD, Watson, J, O'Connor, CC, Jeanes, M, Hoy, J, Cooper, DA, Law, MG; Australian, HIV, Observational Database (2014). Determinants of suicide and accidental or violent death in the Australian HIV Observational Database. PLoS ONE 9, e89089.Google Scholar
Mock, CN, Grossman, DC, Mulder, D, Stewart, C, Koepsell, TS (1996). Health care utilization as a marker for suicidal behavior on an American Indian Reservation. Journal of General Internal Medicine 11, 519–524.Google Scholar
Murphy, GE, Wetzel, RD, Robins, E, McEvoy, L (1992). Multiple risk factors predict suicide in alcoholism. Archives General Psychiatry 49, 459–463.Google Scholar
Nock, MK, Borges, G, Bromet, EJ, Alonso, J, Angermeyer, M, Beautrais, A, Bruffaerts, R, Chiu, WT, de Girolamo, G, Gluzman, S, de Graaf, R, Gureje, O, Haro, JM, Huang, Y, Karam, E, Kessler, RC, Lepine, JP, Levinson, D, Medina-Mora, ME, Ono, Y, Posada-Villa, J, Williams, D (2008). Cross-national prevalence and risk factors for suicidal ideation, plans and attempts. British Journal of Psychiatry 192, 98–105.Google Scholar
Nordentoft, M, Mortensen, PB, Pedersen, CB (2011). Absolute risk of suicide after first hospital contact in mental disorder. Archives of General Psychiatry 68, 1058–1064.CrossRefGoogle ScholarPubMed
Nyman, AK, Jonsson, H (1986). Patterns of self-destructive behaviour in schizophrenia. Acta Psychiatrica Scandinavica 73, 252–262.Google Scholar
O'Connor, N, Allan, J, Scott, C (2014). Debate: clinical risk categorization is valuable in the prevention of suicide and severe violence? Yes. Australasian Psychiatry 22, 7–9.Google Scholar
Palacio, C, Garcia, J, Diago, J, Zapata, C, Lopez, G, Ortiz, J, Lopez, M (2007). Identification of suicide risk factors in Medellin, Colombia: a case-control study of psychological autopsy in a developing country. Archives of Suicide Research 11, 297–308.CrossRefGoogle ScholarPubMed
Park, S, Choi, JW, Kyoung Yi, K, Hong, JP (2013). Suicide mortality and risk factors in the 12 months after discharge from psychiatric inpatient care in Korea: 1989–2006. Psychiatry Research 208, 145–150.Google Scholar
Paton, MB, Large, MM, Ryan, CJ (2014). Debate: clinical risk categorisation is valuable in the prevention of suicide and severe violence–no. Australas Psychiatry 22, 10–12.Google Scholar
Popovic, D, Benabarre, A, Crespo, JM, Goikolea, JM, González-Pinto, A, Gutiérrez-Rojas, L, Montes, JM, Vieta, E (2014). Risk factors for suicide in schizophrenia: systematic review and clinical recommendations. Acta Psychiatrica Scandinavica 130, 418–426.Google Scholar
Posner, K, Brown, GK, Stanley, B, Brent, DA, Yershova, KV, Oquendo, MA, Currier, GW, Melvin, GA, Greenhill, L, Shen, S, Mann, JJ (2011). The Columbia-Suicide Severity Rating Scale: initial validity and internal consistency findings from three multisite studies with adolescents and adults. American Journal of Psychiatry 168, 1266–1277.Google Scholar
Powell, J, Geddes, J, Deeks, J, Goldacre, M, Hawton, K (2000). Suicide in psychiatric hospital in-patients. Risk factors and their predictive power. British Journal of Psychiatry 176, 266–272.CrossRefGoogle ScholarPubMed
Roos, JL, Boraine, H, Bodemer, W (1992). [Suicide in schizophrenic patients]. South African Medical Journal 81, 365–369.Google Scholar
Rosenthal, JA (1996). Qualitative descriptors of strength of association and effect size. Journal of Social Service Research 21, 37–59.CrossRefGoogle Scholar
Rowe, JL, Bruce, ML, Conwell, Y (2006). Correlates of suicide among home health care utilizers who died by suicide and community controls. Suicide Life Threatening Behaviour 36, 65–75.Google Scholar
Runeson, BS (2002). Suicide after parasuicide (editorial). British Medical Journal 325, 1125–1126.Google Scholar
Salama, AA (1988). Depression and suicide in schizophrenic patients. Suicide Life Threatening Behaviour 18, 379–384.Google Scholar
Sani, G, Tondo, L, Koukopoulos, A, Reginaldi, D, Kotzalidis, GD, Koukopoulos, AE, Manfredi, G, Mazzarini, L, Pacchiarotti, I, Simonetti, A, Ambrosi, E, Angeletti, G, Girardi, P, Tatarelli, R (2011). Suicide in a large population of former psychiatric inpatients. Psychiatry and Clinical Neurosciences 65, 286–295.Google Scholar
Shah, AK, Ganesvaran, T (1997). Inpatient suicides in an Australian mental hospital. Australian & New Zealand Journal of Psychiatry 31, 291–298.CrossRefGoogle Scholar
Silverman, MM, Berman, AL, Sanddal, ND, O'Carroll, PW, Joiner, TE (2007). Rebuilding the tower of Babel: a revised nomenclature for the study of suicide and suicidal behaviors. Part 1: Background, rationale, and methodology. Suicide Life Threatening Behaviour 37, 248–263.Google Scholar
Simon, GE, Rutter, CM, Peterson, D, Oliver, M, Whiteside, U, Operskalski, B, Ludman, EJ (2013). Does response on the PHQ-9 Depression Questionnaire predict subsequent suicide attempt or suicide death? Psychiatric Services 64, 1195–1202.Google Scholar
Spiessl, H, Hubner-Liebermann, B, Cording, C (2002). Suicidal behaviour of psychiatric in-patients. Acta Psychiatrica Scandinavica 106, 134–138.Google Scholar
Stephens, JH, Richard, P, McHugh, PR (1999). Suicide in patients hospitalized for schizophrenia: 1913–1940. Journal of Nervous and Mental Disease 187, 10–14.CrossRefGoogle ScholarPubMed
Sterne, JAC, Egger, M, Moher, D (2011). Chapter 10. Addressing reporting biases. In Cochrane Handbook for Systematic Reviews of Interventions version 5.1.0 (ed. Higgins, JPT and Green, S). The Cochrane collaboration. Available from www.handbook.cochrane.org Google Scholar
Steyerberg, EW (2009). Restrictions on candidate predictors. In Clinical Prediction Models, pp. 175–189. Springer: New York, USA.Google Scholar
Taiminen, T, Huttunen, J, Heilä, H, Henriksson, M, Isometsä, E, Kähkönen, J, Tuominen, K, Lönnqvist, J, Addington, D, Helenius, H (2001). The Schizophrenia Suicide Risk Scale (SSRS): development and initial validation. Schizophrenia Research 47, 199–213.Google Scholar
Talseth, AG, Lindseth, A, Jacobsson, L, Norberg, A (1999). The meaning of suicidal psychiatric in-patients’ experiences of being cared for by mental health nurses. Journal of Advanced Nursing 29, 1034–1041.Google Scholar
ten Have, M, de Graaf, R, van Dorsselaer, S, Verdurmen, J, van ‘t Land, H, Vollebergh, W, Beekman, A (2009). Incidence and course of suicidal ideation and suicide attempts in the general population. Canadian Journal of Psychiatry 54, 824–833.Google Scholar
ten Have, M, van Dorsselaer, S, de Graaf, R (2013). Prevalence and risk factors for first onset of suicidal behaviors in the Netherlands Mental Health Survey and Incidence Study-2. Journal of Affective Disorders 147, 205–211.Google Scholar
Thong, JY, Su, AH, Chan, YH, Chia, BH (2008). Suicide in psychiatric patients: case-control study in Singapore. Australian & New Zealand Journal of Psychiatry 42, 509–519.CrossRefGoogle ScholarPubMed
Thoresen, S, Mehlum, L (2004). Risk factors for fatal accidents and suicides in peacekeepers: is there an overlap? Military Medicine 169, 988–993.Google Scholar
Turecki, G (2014). The molecular basis of suicide. Nature Reviews Neuroscience 15, 802–816.Google Scholar
van Heeringen, K, Hengeveld, MW (2009). Suïcidaal gedrag. In Leerboek Psychiatrie (ed. Hengeveld, MW and van Balkom, AJLM), pp. 493–501. De Tijdstroom: Utrecht, The Netherlands.Google Scholar
van Hemert, AM, Kerkhof, AJ, de Keijser, J, Verwey, B, van Boven, C, Hummelen, JW, de Groot, MH, Lucassen, P, Meerdinkveldboom, J, Steendam, M, Stringer, B, Verlinde, AA, van de Glind, G (2012). Multidisciplinaire richtlijn diagnostiek en behandeling van suïcidaal gedrag. Tijdstroom: Utrecht, The Netherlands.Google Scholar
Vuorilehto, M, Valtonen, HM, Melartin, T, Sokero, P, Suominen, K, Isometsa, ET (2014). Method of assessment determines prevalence of suicidal ideation among patients with depression. European Psychiatry 29, 338–344.Google Scholar
Wells, GA, Shea, B, O'Connol, D, Peterson, J, Welch, V, Losos, M, Tugwell, P (n.d.). The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomized studies in meta-analysis. Retrieved 7 July 2015 from http://www.ohri.ca/programs/clinical_epidemiology/nosgen.pdf Google Scholar
Werbeloff, N, Dohrenwend, BP, Levav, I, Haklai, Z, Yoffe, R, Large, M, Davidson, M, Weiser, M (2015). Demographic, behavioural, and psychiatric risk factors for suicide: a 25-year longitudinal cohort study. Crisis epub ahead of print. DOI: https://doi.org/10.1027/0227-5910/a000359 Google Scholar
Wolfersdorf, M, Klinkisch, M, Franke, C, Keller, F, Wurst, FM, Dobmeier, M (2003 a). In-patient suicide – a comparison of psychiatric in-patient suicides v. a control group of inpatients from the same treatment era without suicide. Psychiatrische Praxis 30, 14–20.Google Scholar
Wolfersdorf, M, Neher, F, Arbeitsgemeinschaft Suizidalität und Psychiatrisches Krankenhaus (2003 b). [Schizophrenia and suicide – results of a control group comparison of schizophrenic inpatient suicides with schizophrenic inpatients without suicide]. Psychiatrische Praxis 30, 272–278.Google Scholar
Yigletu, H, Tucker, S, Harris, M, Hatlevig, J (2004). Assessing suicide ideation: comparing self-report versus clinician report. Journal of the American Psychiatric Nurses Association 10, 9–15.Google Scholar
Zonda, T (2006). One-hundred cases of suicide in Budapest: a case-controlled psychological autopsy study. Crisis 27, 125–129.Google Scholar
Figure 0

Fig. 1. Flowchart of the study selection process. Si indicates suicidal ideation; cs, completed suicide; sa, suicide attempts. Mixed psychiatric populations: n = 9; including three studies on people with substance use disorders, two on patients with borderline personality disorder, two on mixed diagnosis psychiatric in & outpatients, and two on mixed diagnosis psychiatric outpatients. Non-psychiatric residual category: n = 8; including two studies on military, one on veterans, one on stalkers, one on HIV infected males, one on survivors of childhood cancer, one on prisoners, and one on emergency department visitors.

Figure 1

Fig. 2. Forest plot showing the association between suicidal ideation and subsequent completed suicide. CI, confidence interval. aAuthors provided additional data to compute effect size.

Figure 2

Fig. 3. Bar chart showing suicide risk ratio and the suicide risk (%) during first year of follow-up. Lines indicate 95% confidence interval. Numbers of studies included in (A) and (B) differ because in only 26 of the 81 studies the absolute risk of completed suicide could be extracted. A. Risk ratio for suicide after expression of suicidal ideation, stratified for psychiatric and non-psychiatric populations. B. Results specified separately for study participants who had expressed suicidal ideation and study participants who had not, stratified for psychiatric and non-psychiatric populations. Percentages calculated using maximum likelihood meta-regression analyses with mean study follow-up time as determinant and loge-transformed rates of completed suicide as outcome (see Fig. 4).

Figure 3

Fig. 4. Meta-regression of mean study follow-up time on suicide rate. Maximum likelihood meta-regression analyses with mean study follow-up time as determinant and loge-transformed rates of completed suicide as outcome presented on logarithmic scales. The bubble size is proportional to the study's weight. Vertical line indicates 1 year follow-up. The suicide risk can be calculated using the following formulas: suicide risk (during first year of follow-up) = 1 – e(-rate per person-year at year 1). Figure part A, B: Psychiatric subgroup included: Dingman & McGlashan, 1986; Drake & Cotton, 1986; Allebeck et al.1987; Goldstein et al.1991; Bradvik & Berglund, 1993; Stephens et al.1999; Bioulac et al.2000; de Hert et al.2001; Spiessl et al.2002; Coryell & Young, 2005; Dutta et al.2007; Berg, 2010; Dutta et al.2011; Kuo et al.2011; Sani et al.2011; Park et al.2013; Simon et al.2013. Lin et al.2014; Figure part C, D: Non-psychiatric subgroup included: Crandall et al. 2006; Khang et al. 2010; McEwan et al. 2010; Hyman, 2012; Brinkman et al.2014; Kleiman & Liu, 2014; Al-Sayegh et al.2015; Werbeloff et al.2015.

Supplementary material: File

Hubers supplementary material

Hubers supplementary material 1

Download Hubers supplementary material(File)
File 1 MB