Meta-analysis of risk factors for suicide after psychiatric discharge and meta-regression of the duration of follow-up

Abstract

Background:

Rates of suicide following discharge from psychiatric hospitals are extraordinarily high in the first week post-discharge and then decline steeply over time. The aim of this meta-analysis is to evaluate the strength of risk factors for suicide after psychiatric discharge and to investigate the association between the strength of risk factors and duration of study follow-up.

Methods:

A PROSPERO-registered meta-analysis of observational studies was performed in accordance with PRISMA guidelines. Post-discharge suicide risk factors reported five or more times were synthesised using a random-effects model. Mixed-effects meta-regression was used to examine whether the strength of suicide risk factors could be explained by duration of study follow-up.

Results:

Searches located 83 primary studies. From this, 63 risk estimates were meta-analysed. The strongest risk factors were previous self-harm (odds ratio = 2.75, 95% confidence interval = [2.37, 3.19]), suicidal ideation (odds ratio = 2.15, 95% confidence interval = [1.73, 2.68]), depressive symptoms (odds ratio = 1.84, 95% confidence interval = [1.48, 2.30]), and high-risk categorisation (odds ratio = 7.65, 95% confidence interval = [5.48, 10.67]). Significantly protective factors included age ⩽30, age ⩾65, post-traumatic stress disorder, and dementia. The effect sizes for the strongest post-discharge suicide risk factors did not decline over longer periods of follow-up.

Conclusion:

The effect sizes of post-discharge suicide risk factors were generally modest, suggesting that clinical risk factors may have limited value in distinguishing between high-risk and low-risk groups. The highly elevated rates of suicide immediately after discharge and their subsequent decline remain unexplained.

Keywords

Suicide risk assessment mental health discharge

Among the many established and putative suicide risk factors, previous psychiatric care, particularly psychiatric hospitalisation, is emerging as one of the most significant statistical risk factors for later suicide. In a meta-analysis of 50 years of longitudinal research on risk factors for suicidal thoughts and behaviours, Franklin and associates (Franklin et al., 2017) found that a history of prior psychiatric hospitalisation was the strongest risk factor for suicide. This was also the finding of a Danish population study examining hierarchical levels of psychiatric treatment (Hjorthøj et al., 2014) and is consistent with very high rates of post-discharge suicide demonstrated in two meta-analyses that respectively examined suicide rates after longer and shorter periods of follow-up (Chung et al., 2017, 2019). When the results of these two meta-analyses of suicide rates after discharge are combined, a picture emerges of very high suicide mortality rates in the first week of follow-up = 2950 suicides per 100,000 person years (24 studies, 95% confidence interval [CI] = [1740, 5000]), that subsequently decline over time by a factor of 10, as shown by rates at periods of follow-up of up to 1 month = 2060 (29 studies, CI = [1300, 3280]), 3 months of follow-up = 1132 (18 studies, CI = [874, 1467]), 3 to 12 months of follow-up = 654 (43 studies, CI = [533, 802]), 1 and 5 years of follow-up = 494 (44 studies, CI = [354, 688]), 5 and 10 years of follow-up = 366 (44 studies, CI = [283, 472]), and over 10 years = 277 (34 studies, CI = [218, 352]).

Potential explanations for the very high initial post-discharge suicide rates and their subsequent decline over time are under-researched, and no consistent pattern of temporal changes in the strength of post-discharge suicide risk factors have emerged. In 1993, Goldacre and associates examined a limited set of risk factors over 1 month and 1 year of follow-up after discharge from a group of English psychiatric hospitals. No clear explanation for the decline in suicide rates over longer periods of follow-up was found. Goldacre hypothesised that discharge itself was a suicide risk factor, possibly because of a loss of hospital care, reduced supervision, and psychiatric relapse due to renewed exposure to problems in the home environment (Goldacre et al., 1993). Two further studies examining the strength of post-discharge suicide risk factors found no significant temporal changes (Iliachenko et al., 2020; Kessler et al., 2020). When suicide risk factors have been found to change in strength over time after psychiatric discharge, the results have not been replicated. In 2009, Hunt and associates found that male sex, a history or self-harm, a history of violence, self-initiated discharge, and failure of follow-up were more strongly associated with earlier rather than later suicide (Hunt et al., 2009). In 2021, Britton and associates found that that young age was a stronger risk factor for suicide over a longer period of follow-up and that suicidal ideation became a weaker risk factor over time (Britton et al., 2021). More recently, Aaltonen and associates found that the risk factors of more severe depression and poor global functioning declined in strength over longer periods of follow-up, while involuntary treatment, living alone, and substance use, increased in strength over time (Aaltonen et al., 2024).

To date, there is one meta-analysis of suicide risk factors after psychiatric hospital discharge (Large et al., 2011). In this 2011 study, the strongest associations with post-discharge suicide were a prior history of self-harm or a suicide attempt (odds ratio [OR] = 3.15), depressive symptoms (OR = 2.70), suicidal ideation (OR = 2.47), and a high-risk categorisation (OR = 3.94) as derived from sample of 13 primary studies reporting suicides in the first year post-discharge (Large et al., 2011). In this meta-analysis, numerous widely accepted suicide risk factors in the general community, including older age, single marital status, living alone, unemployment, substance use and physical illness, were not significantly associated with suicide during the post-discharge period. Disconcertingly, patients who had more post-discharge psychiatric care were significantly more likely to suicide than those who received less care. The meta-analysis did not examine changes in the strength of suicide risk factors according to duration of study follow-up.

Aims and hypothesis

The first aim was to use meta-analysis to re-examine the strength of associations between suicide risk factors and suicide over all periods of follow-up, serving as an update and extension of the 2011 meta-analysis (Large et al., 2011). The second aim was to determine whether between-study heterogeneity in the strength of suicide risk factors (effect size) could be explained by the mean duration of primary study follow-up.

We hypothesised that the effect size of the stronger risk factors for post-discharge suicide, as determined by this meta-analysis, would decline over longer periods of time, and/or that the effect size of protective factors would increase over time, thus explaining the decline in post-discharge suicide rates.

Methods

The meta-analysis was performed according to the Meta-analysis of Observation Studies in Epidemiology (MOOSE) (Stroup et al., 2000) and Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) (Moher et al., 2009) guidelines (see Supplemental Material 1., PRISMA checklist). The meta-analysis was registered with PROSPERO: registration CRD42023449429. The study was an extension and update of the 2011 meta-analysis (Large et al., 2011) and a protocol was not published.

Search strategy

A.T. and M.L. independently searched MedLINE, PsycINFO and Embase for relevant articles published in the English language from inception until 11 April 2024 (see Figure 2. Flow chart of search strategy) using the search terms “suicid*.ti. and ((discharge* or hospital* or admit* or admis* or in-patient or inpatient).ab. or (discharge* or hospital* or admit* or admis* or in-patient or inpatient).ti.) and ((psych* or mental or affective or schizophreni* or bipolar or depress*).ab. or (psych* or mental or affective or schizophreni* or bipolar or depress*).ti.)”. A.T. and M.L. independently performed searches and pared down by titles and then abstracts. Papers chosen for full-text examination after examination of abstracts by either author were considered for inclusion and exclusion.

Inclusion and exclusion criteria

Cohort and case-control studies were included if they reported on the clinical characteristics of groups of patients who died by suicide or survived after discharge from a psychiatric inpatient unit.

Studies were excluded if they (1) did not describe the characteristics of non-suicide patients (i.e. had no control or comparator group), (2) included patients discharged from non-psychiatric settings, (3) included suicides among currently admitted patients, (4) included patients who were not admitted or discharged, (5) used controls who died from causes other than suicide, or (6) reported on populations of patients who were unrepresentative of psychiatric inpatients (for example, from forensic hospitals). The inclusion and exclusion criteria were applied by A.T and M.L independently and then by consensus. Where two or more studies reported on overlapping samples, the study with the largest sample size was included.

Data extraction

A.T. and M.L. independently extracted counts (suicide & risk factor, total suicide, non-suicide & risk factor, total non-suicide) and other unadjusted effect size data when counts were not available. The mean and standard deviation of continuous risk variables (such as age, duration of hospitalisation and clinical rating scales) were similarly extracted. Discrepant data points were examined by H.P. and A.B. and resolved by consensus with A.T. and M.L.

A.T. and M.L. standardised terminology of risk factors by consensus. Similar terms in the primary research were aggregated (for example, depressive symptoms, depressed mood and depressed). The continuous variable of age and the categories of older age (defined as >65 years) and younger age (defined as <30 years) were examined.

The pooled effect size for each post-discharge suicide risk factor (defined here to include both risk and protective factors) was expressed as an odds ratio (OR) in line with common reporting practices of primary literature and similar meta-analyses (Grover et al., 2023; Large et al., 2017). The moderator variable of mean study duration of follow-up was extracted and where study follow-up was not reported directly, it was estimated by the interval between the mid-point of data collection and the end-point of follow-up. A.T., H.P. and A.B. reconciled any disputed data points through further examination of studies.

Strength of reporting scale

The strength of reporting of the included primary studies was assessed using a 0 to 6 point scale derived from the Newcastle-Ottawa scale for assessing the quality of non-randomised studies in meta-analysis (Wells et al., 2021). One point was awarded if the study included a (1) sample of patients that were unselected by diagnosis, (2) employed a cohort study design, (3) included a recognised diagnostic system, (4) identified suicides by using coronial records or other external mortality data (as opposed to hospital records), (5) employed a register of discharges, and (6) ascertained research data on suicide risk factors at the time of discharge rather than by retrospective examination of patient notes. The impact of the strength of reporting on the strength of suicide risk factors was examined by comparing groups of studies categorised as having stronger or less strong reporting based on the strength of reporting score dichotomised using the median score.

Statistical analysis

The pooled strength of the effect size of suicide risk factors were estimated by meta-analysis when a risk factor was reported five or more times. This threshold of five was consistent with other large-scale studies of suicide risk factors among psychiatric inpatients (Large et al., 2017) and was chosen to increase the accuracy of the pooled estimates, reduce the impact of possible publication bias, and allow more meaningful assessments of between-study heterogeneity and tests of publication bias. Some studies reported data points at various follow-up periods and subgroups according to gender; in these studies, the pooled mean OR of the data points or subgroups was used as the unit of analysis. Count data and where possible, other effect size data, were converted to OR using Comprehensive Meta-Analysis Version 2 (CMA V2) (Borenstein et al., 2022). Effect sizes that are not able to be converted to OR (for instance, hazard ratios) were recorded, but ultimately were too few for a separate meta-analysis. Pooled OR and 95% confidence intervals were calculated using a random-effects model. A random-effects model was selected a priori due to the variation in study design, hospital settings, clinical populations, countries and year of study. The strength of the pooled effect size was classified as weak if the OR was between 1.0 and 2.0, moderate if the OR was between 2.0 and 3.0 and strong if the OR was above 3.0, in line with the qualitative descriptors of Rosenthal (Rosenthal, 1996). Between-study heterogeneity in effect size was assessed by examination of funnel plots and with Q-value and I² statistics. Publication bias was assessed using examination of the funnel plot and Egger’s regression. The extent of possible publication bias was assessed using the Tweedie and Duval trim and fill method. The sensitivity and specificity were separately calculated for high-risk categorisation using a random-effects model.

Random-effects (method of moments) meta-regression was used to examine the explanatory value of the moderator variable of the duration of follow-up on the observed heterogeneity in effect sizes reported in 10 or more data points. The threshold of 10 was chosen for consistency with previous research and to ensure the reliability of the results (Grover et al., 2023). Multiple data-points and subgroups within the same study (according to the length of follow-up and gender) were regarded as independent for the purpose of the meta-regressions.

Sensitivity analyses were performed according to the study method (cohort vs control), the diagnostic mix of the patient population (unselected, mood disorders, schizophrenia spectrum disorders), and dichotomised strength of reporting scores, when more than five data points were available in each group. Between-group differences were examined using a mixed-effects model. A sensitivity analysis of the meta-regressions was performed using naturally transformed (ln) follow-up duration.

CMA V2 was used for meta-analyses and meta-regressions (Borenstein et al., 2022).

Results

Search results

The electronic searches identified a total of 83 studies that met the inclusion criteria (see Figure 1. Flow chart of searches: 1: Supplemental Material 2. Table of Included Studies: Aaltonen et al., 2019, 2024; Agerbo et al., 2001; Allgulander et al., 1992; Appleby et al., 1999; Axelsson and Lagerkvist-Briggs, 1992; Beck et al., 1985, 1989; Berglund et al., 1987; Bickley et al., 2013; Bjorkenstam et al., 2014; Black et al., 1985a, 1985b; Bradvik and Berglund, 1993; Britton et al., 2017, 2021; Carlborg et al., 2009, 2010; Che et al., 2023; Chiu et al., 2023; Choi et al., 2019, 2020; Clapham et al., 2019; De Hert et al., 2001; Deisenhammer et al., 2019; Dennehy et al., 1996; Desai et al., 2005; Enns et al., 2022; Farberow and MacKinnon, 1974; Fazel et al., 2006; Flood and Seager, 1968; Fowler et al., 1979; Fredriksen et al., 2022; Geddes and Juszczak, 1995; Geddes et al., 1997; Goldacre et al., 1993; Goldstein et al., 1991; Guimaraes et al., 2022; Høyer et al., 2004; Haglund et al., 2019; Heu et al., 2018; Ho, 2003, 2006; Hoffmann and Modestin, 1987; Hogan and Awad, 1983; Hong et al., 2024; Hunt et al., 2009; Iliachenko et al., 2020; Isometsa et al., 2014; Jiang et al., 2021; Kan et al., 2007; Kaster et al., 2022; Kessler et al., 2014, 2020; King, 2001; Kuo et al., 2005, 2010; Lee and Lin, 2009; Lin et al., 2008; Luxton et al., 2013; McKenzie and Wurr, 2001; Madsen et al., 2020; Modestin and Schwarzenbach, 1992; Motto and Bostrom, 1990; Motto et al., 1985; Nielsen et al., 2023; Nordström et al., 1995; Olfson et al., 2016; Park et al., 2013; Pirkola et al., 2005; Plans et al., 2019; Pokorny, 1983; Pokorny and Kaplan, 1976; Prestmo et al., 2023; Reutfors et al., 2009; Riblet et al., 2018; Sani et al., 2011; Shaffer et al., 1974; Stephens et al., 1999; Tsai et al., 2002; Tseng et al., 2020; Yim et al., 2004).

Figure 1.

Flow chart of searches.

The primary studies included 54 cohort studies and 29 case-control or nested case-control studies. 82 studies were conducted in high-income countries: 10 from the UK, 25 from the United States and Canada, 32 from mainland Europe and 15 from high-income Asian countries. One paper was from a South American upper middle-income country. There were no papers from low-or middle-income settings as defined by the World Bank LAMI classification system (Metreau et al., 2024). The mean year of publication was 2006, and the median was 2009.

The included studies reported on a total of 9,452,741 discharged patients, of whom 39,190 (0.41%) died by suicide. There was a mean of 119,655 subjects/study (SD = 378,479), a mean of 496.1 suicides/study (SD = 990.1 suicides/study), and a median 1846 subjects/study and 91 suicides/study.

15 studies or study data points reported follow-up periods of 1 month or less, 14 study data points reported follow-up periods between 1 and 3 months, 19 study data points reported follow-up periods between 3 months and a year, 18 study data points reported follow-up periods between a year and 3 years, and 28 study data points reported follow-up periods greater than 3 years.

The mean of the strength of the reporting scale was 4 (SD = 1.7) and the median was also 4 (interquartile range = 3.5) (see Supplemental Material 3. Strength of Reporting Scores).

The 83 studies reported 1698 effect sizes after the exclusion of a small number of study-specific variables (such as the place and year of suicide). Of the 1698 effect sizes, 255 were excluded: 209 were reported four or fewer times (reported in 1 study = 134, 2 studies = 29, 3 studies = 32, 4 studies = 14) and 46 were residual categories of uncertain meaning. By virtue of being reported in five or more data points, 1443 effect sizes were included in a meta-analysis of 63 risk factors (62 individual risk factors and high-risk categorisation). All variables, including those reported in four or fewer data points, can be found in Supplemental Material 4. Data Used in the Meta-analysis.

The overall pooled effect size for all risk factors was OR = 1.22 (95% CI = [1.19, 1.25]) with a Z value = 14.31 and p = < 0.05. The overall heterogeneity was very high Q = 21673 [df = 1697] and I² = 92). The Funnel Plot (Figure 2) appeared symmetrical while an Egger’s regression intercept was statistically significant. Correction with Duval and Tweedie’s trim and fill revealed slight publication bias towards studies that reported a lower OR. The Tweedie and Duval method estimated a corrected OR of 1.29 (95% CI = [1.25, 1.33]) by trimming 74 of the 1698 risk factors. The duration of study follow-up was not significantly associated with between-study heterogeneity in effect sizes.

Figure 2.

Funnel of 1698 effect sizes of risk factors for post-discharge suicide.

Meta-analysis of post-discharge suicide risk factors

The strength of post-discharge suicide risk factors that were reported five or more times are shown in Table 1.

Table 1.

Meta-analysis of risk factors for suicide after psychiatric discharge.

	Counts					Effect size
	Data points (N)	RF suicide	Total suicide	RF non-suicide	Total non-suicide	OR	Lower limit	Upper limit	Z-score	p value
Demographic Factors
Male	48	17997	29419	3350756	7218351	1.48	1.34	1.63	7.781	< 0.001
Age	44	6328	16135	2118946	5179464	0.92	0.85	0.98	-2.55	0.01
Age (<30)	15	1506	11589	574615	3770886	0.73	0.58	0.91	-2.780	0.005
Age (>65)	18	780	7183	435911	2575815	0.62	0.47	0.83	-3.265	0.001
Employed	6	1460	3262	36203	94351	0.91	0.58	1.44	-0.392	0.70
Unemployed	14	659	1366	2802	10306	1.07	0.86	1.33	0.577	0.56
Single	19	867	1915	75560	202501	0.99	0.76	1.28	-0.109	0.91
Married	22	2174	5718	113259	334095	1.04	0.89	1.22	0.516	0.61
Widowed or divorced	13	326	1322	57351	141992	1.30	0.99	1.72	1.858	0.06
Urban residence	5	2577	3452	1026835	1252465	0.90	0.67	1.22	-0.671	0.50
Rural residence	5	974	2545	380978	969271	1.01	0.74	1.37	0.067	0.95
Living alone	10	411	1131	502	1619	1.18	1.00	1.38	2.011	0.04
Living with family	6	351	456	449	596	1.05	0.77	1.43	0.314	0.75
Homeless	10	364	3174	201790	952414	0.85	0.60	1.20	-0.925	0.36
High socioeconomic status	13	1077	4443	33995	145733	1.25	1.11	1.42	3.657	< 0.001
Low socioeconomic status	7	314	603	7648	27373	1.05	0.82	1.33	0.358	0.72
More education	16	966	3562	13908	99780	1.24	1.09	1.41	3.247	0.001
Not white	8	254	2382	125806	645324	0.68	0.44	1.07	-1.664	0.10
Social and Family History
Adverse childhood experiences	7	98	779	3000	40544	1.06	0.79	1.43	0.374	0.71
Criminal/forensic history	6	156	529	12482	29738	1.19	0.73	1.94	0.711	0.48
Family history of mental illness	13	388	1497	6742	31714	1.54	1.16	2.06	2.950	0.003
Family history of suicide	10	75	602	335	3821	1.68	1.04	2.71	2.124	0.03
Adverse life events	23	719	5391	131549	2958835	1.85	1.31	2.63	3.454	< 0.001
Social support	7	160	464	631	3166	1.19	0.91	1.57	1.284	0.20
Psychiatric Symptoms
Anxiety symptoms	15	813	7742	214190	3370091	1.12	0.86	1.45	0.85	0.39
Depressive symptoms	28	669	2082	14160	104784	1.84	1.48	2.30	5.45	< 0.001
Hopelessness/helplessness/worthlessness/guilt	8	174	686	999	3882	1.83	1.06	3.16	2.18	0.03
Mania	8	396	4008	32581	192301	0.96	0.64	1.43	-0.20	0.84
Psychomotor changes	5	136	550	2704	19834	1.17	0.81	1.69	0.84	0.40
Aggression/violence	8	163	1028	1344	9423	1.06	0.69	1.63	0.28	0.78
Delusions	10	245	2169	6142	91447	1.26	0.81	1.96	1.03	0.31
Psychosis	10	632	3722	10105	163415	1.77	1.36	2.32	4.18	< 0.001
Sleep disturbance	6	134	1190	9698	328096	1.01	0.78	1.29	0.04	0.97
Social impairment	7	207	896	40058	139087	0.83	0.45	1.56	-0.57	0.57
Suicidal ideation	25	1650	4882	198404	676403	2.15	1.73	2.68	6.88	< 0.001
Previous self-harm	40	1983	7047	17397	789808	2.75	2.37	3.19	13.45	<0.001
Longer duration psychiatric of illness	13	878	1319	21167	30450	0.96	0.70	1.31	-0.28	0.78
Psychiatric Diagnosis
Substance use disorder	42	3395	28238	748811	8887044	1.15	0.96	1.37	1.511	0.13
Alcohol use disorder	33	4443	19995	1739577	5770281	0.87	0.73	1.04	-1.512	0.13
Affective disorder	40	8647	32549	1625897	10275636	1.72	1.53	1.92	9.311	< 0.001
Severe depressive disorders (severe/melancholic/psychotic)	12	1079	5634	253687	1256019	1.34	0.86	2.05	1.31	0.19
Bipolar disorder	26	2424	18662	599792	7591749	1.22	1.05	1.43	2.519	0.01
Schizophrenia spectrum disorder	45	5412	29366	1816613	9608269	0.94	0.83	1.06	-1.055	0.29
Post-traumatic stress disorder	5	590	3071	302142	1141103	0.66	0.53	0.82	-3.685	< 0.001
Neurotic disorder	9	1090	9091	101278	1379214	1.10	0.80	1.50	0.575	0.57
Psychosomatic disorder	5	51	430	939	8418	1.01	0.48	2.09	0.015	0.98
Personality disorder	26	1222	18971	242208	5571718	1.51	1.29	1.77	5.049	< 0.001
Organic mental disorder	15	78	2727	4026	117943	0.64	0.46	0.88	-2.205	0.007
Dementias	5	38	4257	31992	1030492	0.17	0.06	0.48	-3.361	< 0.001
Physical comorbidity	13	725	3438	350346	1068059	0.68	0.45	1.02	-1.852	0.064
Psychiatric comorbidity	8	399	790	473	1125	1.54	1.05	2.24	2.236	0.03
Psychiatric Treatment
Longer length of hospital stay	19	1490	3194	434110	847185	0.99	0.80	1.22	-0.127	0.90
More previous admissions	26	2765	8452	610468	1537051	0.97	0.53	1.80	-0.092	0.93
Involuntary mental health treatment	12	258	1026	233	1151	1.36	0.97	1.91	1.758	0.08
Medication (antidepressants)	13	255	818	9212	19931	1.32	0.99	1.76	1.865	0.06
Medication (antipsychotics)	10	288	1131	11571	20244	0.76	0.52	1.12	-1.378	0.17
Medication (mood stabiliser)	6	111	458	5647	20071	1.00	0.51	1.97	0.002	0.99
Electroconvulsive therapy	10	157	856	8059	79302	0.80	0.59	1.07	-1.525	0.13
Discharged against medical advice	15	671	5870	54391	1442296	1.78	1.43	2.21	5.206	<0.001
Non-adherence to psychiatric treatment	8	211	755	224	877	1.57	0.76	3.27	1.213	0.23
More psychiatric follow-up after discharge	13	2465	4986	155945	248611	0.94	0.73	1.21	-0.476	0.63
Problems with follow-up	5	284	945	227	1189	1.94	1.39	2.70	3.924	<0.001
Risk Assessment
High-risk categorisation	18	1030	2765	38099	273728	7.65	5.48	10.67	11.960	<0.001

Most of the identified risk factors were only weakly associated with post-discharge suicide. There were 15 suicide risk factors with an OR > 1.5: high-risk categorisation, previous self-harm, suicidal ideation, depressive symptoms, problems with follow-up, adverse life event, hopelessness/helplessness/worthlessness/guilt, discharged against medical advice, psychosis, affective disorder, family history of suicide, non-adherence to psychiatric treatment, family history of mental illness and psychiatric comorbidity (see Figure 3. Forrest plot of the top 15 effect sizes of risk factors for post-discharge suicide).

Figure 3.

Forrest plot of the top 15 effect sizes of risk factors for post-discharge suicide.

Demographic factors

The demographic risk factors that were significantly associated with post-discharge suicide included male sex (OR = 1.48, 95% confidence interval [CI] = [1.34, 1.63]), living alone (OR = 1.18, 95% CI = [1.00, 1.38]), high socioeconomic status (OR = 1.25, 95% CI = [1.11, 1.42]), and more education (OR = 1.24, 95% CI = [1.09, 1.41]). Both older age (OR = 0.62, 95% CI = [0.47, 0.83]), and younger age (OR = 0.73, 95% CI = [0.58, 0.91]) were protective. Demographic factors not associated with post-discharge suicide included employment status, marital status (single, married or widowed/divorced), urban residence, rural residence, living with family, homelessness, low-socioeconomic status, and ‘non-white’ (noting that the primary research did not clarify whether this referred to race or skin colour).

Social and family history

Having a family history of mental illness (OR = 1.54, 95% CI = [1.16, 2.06]), family history of suicide (OR = 1.68, 95% CI = [1.04, 2.71]), or adverse life events (OR = 1.85, 95% CI = [1.31, 2.63]), were weakly associated with post-discharge suicide. Adverse childhood experience, criminal history, and more social support, were not associated with post-discharge suicide.

Psychiatric signs and symptoms

Previous self-harm (OR = 2.75, 95% CI = [2.37, 3.19]), and suicidal ideation (OR = 2.15, 95% CI = [1.73, 2.68]) were moderately associated with post-discharge suicide. Depressive symptoms (OR = 1.84, 95% CI = [1.48, 2.30]), the specific depressive symptoms of hopelessness/helplessness/worthlessness/guilt (OR = 1.83, 95% CI = [1.06, 3.16]), and psychosis (OR = 1.77, 95% CI = [1.36, 2.32]) showed weak associations. Psychiatric symptoms not associated with post-discharge suicide were anxiety symptoms, mania, psychomotor changes (including agitation and retardation), aggression/violence, delusions, sleep disturbance, social impairment, and longer duration of illness.

Psychiatric diagnosis

Affective disorders (OR = 1.72, 95% CI = [1.53, 1.92]), bipolar disorders (OR = 1.22, 95% CI = [1.05, 1.43]), personality disorders (OR = 1.51, 95% CI = [1.29, 1.77]), psychiatric comorbidities (OR = 1.54, 95% CI = [1.05, 2.24]) were weakly associated with post-discharge suicide. Psychiatric diagnoses that were not associated with post-discharge suicide included substance use disorder, alcohol use disorder, schizophrenia spectrum disorder, ‘neurotic disorder’, and physical comorbidity. Post-traumatic stress disorder (OR = 0.66, 95% CI = [0.53, 0.82]), organic mental disorders (OR = 0.64, 95% CI = [0.46, 0.88]), and dementias (OR = 0.17, 95% CI = [0.06, 0.48]) were protective against post-discharge suicide.

Psychiatric treatment factors

Psychiatric treatment factors that demonstrated weak associations included discharge against medical advice (OR = 1.78, 95% CI = [1.43, 2.21]), and problems with follow-up (OR = 1.94, 95% CI = [1.39, 2.70]). Psychiatric treatment factors not associated with post-discharge suicide include longer length of hospital stay (OR = 0.99, 95% CI = [0.80, 1.22]), more previous admissions, involuntary mental health treatment, antidepressants, antipsychotics, mood stabilisers, electroconvulsive disorders, non-adherence to psychiatric treatment, and more psychiatric follow-up after discharge (OR = 0.94, 95% CI = [0.73, 1.21]).

High-risk categorisation

16 studies reported 18 high-risk categorisation methods. A variety of methods were used in the primary studies to distinguish between high- and low-risk patients. Three studies utilised clinical judgement, five utilised the number of co-occurring risk factors, four used statistical analysis (logistic regression) and four utilised machine learning. The most frequently included suicide risk factors were previous self-harm and depressive symptoms. High-risk categorisation was strongly associated with post-discharge suicide (OR = 7.65, 95% CI = [5.48, 10.67]). The meta-analytic sensitivity of high-risk categorisation was 45% (CI = [36%, 55%]) and specificity was 92% (CI = [88%, 95%]).

Heterogeneity

8 risk factors had low between-study heterogeneity (I² < 25%), 10 had moderate heterogeneity (25% ⩽ I² < 50%), 19 had high heterogeneity (50% ⩽ I² < 75%), and 26 had very high heterogeneity (I² ⩾ 75%) (see Supplemental Material 5. Assessment of Heterogeneity and Publication Bias.).

Statistical evidence of publication bias

Evidence of publication bias was found among eight out of the 63 suicide risk factors (see Supplemental Material 5. Assessment of Heterogeneity and Publication Bias). Egger’s test was significant for younger age (intercept = −3.33, p = 0.01), married (intercept = −1.23, p = 0.03), homelessness (intercept = 2.07, p = 0.02), more education (intercept = −1.17, p = 0.03), depressive symptoms (intercept = −2.30, p = 0.01), suicidal ideation (intercept = 1.73, p = 0.01), longer hospital stay (intercept = 2.46, p = 0.01) and discharge against medical advice (intercept = 1.65, p = 0.01). More education and discharge against medical advice were the only risk factors that had a corrected OR that differed by ⩾5% from the uncorrected value, as estimated using Duval and Tweedie’s trim and fill method.

Strength of reporting

The strength of reporting had minimal impact on the effect size of suicide risk factors. The only statistically significant finding was that the OR for male sex was 19% higher in the group of studies with stronger reporting when compared with the group of studies with less strong reporting (Supplemental Material 6. Sensitivity analysis according to dichotomised strength of reporting score).

Meta-regression

A longer mean duration of follow-up was associated with a lower OR (lower suicide risk) in the meta-regression of the variables of being married and antipsychotic medication (see Table 2. Meta-regression using the moderator of length of follow-up). A longer duration of follow-up was associated with a higher OR (increased suicide risk) in the meta-regression of the variables of male, living alone, longer duration of illness, neurotic disorders, personality disorders and involuntary mental health treatment. None of the risk factors that were most strongly associated with suicide (including high-risk categorisation, previous self-harm, suicidal ideation, depressive symptoms and adverse life events) had effect sizes that were significantly lower in studies with longer follow-up. No significantly protective factor was reported in a sufficient number of primary studies for the meta-regression analysis. An analysis using natural logarithm transformed length of stay data yielded similar results to untransformed length of stay data.

Table 2.

Meta-regression using the moderator of length of follow-up.

Risk factor	Point estimate of slope	Standard error	p value
Demographic Factors
Male	<-0.001	<0.001	0.01
Age	<0.001	<0.001	0.83
Age (<30)	<-0.001	< 0.001	0.39
Age (>65)	<-0.001	<0.001	0.35
Unemployed	<0.001	0.001	0.76
Single	<0.001	<0.001	0.4
Married	<-0.001	<0.001	0.03
Widowed or divorced	<0.001	<0.001	0.19
Living alone	0.002	0.001	0.02
Homeless	<0.001	<0.001	0.05
High socioeconomic status	<-0.001	<0.001	0.73
More education	<-0.001	<0.001	0.98
Not white	<-0.001	0.001	0.94
Social and Family History
Family history of mental illness	<-0.001	<0.001	0.29
Family history of suicide	<0.001	<0.001	0.34
Adverse life events	<-0.001	<0.001	0.30
Psychiatric Symptoms
Anxiety symptoms	<-0.001	<0.001	0.62
Depressive symptoms	<-0.001	<0.001	0.06
Delusions	<-0.001	<0.001	0.87
Suicidal ideation	<0.001	<0.001	0.65
Previous self-harm	<0.001	<0.001	0.76
Longer duration of psychiatric illness	0.001	<0.001	0.001
Psychiatric Diagnosis
Substance use disorder	<0.001	<0.001	0.84
Alcohol use disorder	<0.001	<0.001	0.10
Affective disorder	<-0.001	0.000	0.79
Severe depressive disorders (severe/melancholic/psychotic)	0.001	<0.001	0.05
Bipolar disorder	<0.001	<0.001	0.25
Schizophrenia spectrum disorder	<0.001	<0.001	0.06
Neurotic disorder	0.002	<0.001	0.01
Personality disorder	<0.001	<0.001	0.001
Organic mental disorder	<-0.001	<0.001	0.63
Physical comorbidity	<-0.001	<0.001	0.24
Psychiatric comorbidity	<-0.001	<0.001	0.58
Psychiatric Treatment
Longer length of hospital stay	<-0.001	<0.001	0.23
More previous admissions	<0.001	0.001	0.56
Involuntary mental health treatment	0.002	<0.001	0.01
Medication (antidepressants)	<0.001	<0.001	0.73
Medication (antipsychotics)	-0.001	<0.001	0.02
Electroconvulsive therapy	<0.001	<0.001	0.24
Discharged against medical advice	<0.001	<0.001	0.69
Non-adherence to psychiatric treatment
More psychiatric follow-up after discharge	0.000	<0.001	0.67
Risk Assessment
High-risk categorisation	0.002	0.001	0.24

Sensitivity analyses

Grouping by diagnosis impacted risk factor strength (that is, produced a significant p-value) for male, the continuous variable of age, younger age and older age categories (see Supplemental Material 7. Sensitivity analysis according to study diagnostic group). Study design (cohort vs case-control) impacted on the strength of the risk factors of male, age, and widowed/divorced (Supplemental Material 8. Sensitivity analysis according to cohort or control design).

Discussion

This meta-analysis synthesises data from 83 primary studies of risk factors for suicide after psychiatric discharge. The meta-analysis produced 62 estimates of the strength of easily assessed suicide risk factors, as well as an estimate of the potential strength of risk categorisation. A high-risk categorisation, based on the presence of two or more suicide risk factors, was strongly associated with post-discharge suicide. However, the limited sensitivity of the high-risk categorisation suggested that fewer than half of all future suicides can be identified by combining risk factors. Disappointingly, the strength of individual suicide risk factors was generally weak, with only two risk factors having an OR over 2 (suicidal ideation and previous self-harm) and none having an OR over 3. Contrary to our hypothesis, between-study heterogeneity among the strongest risk factors for post-discharge suicide (including high-risk categorisation) was not explained by a declining strength of association with suicide over longer periods of follow-up. There was insufficient data to examine whether any factor accrued stronger protection against post-discharge suicide over time.

The strengths of this study include the number of included primary studies, the large number of total subjects and the comprehensive set of meta-analytic estimates of common suicide risk factors. The study has several limitations. First, the absence of studies from low- and middle-income settings means that our findings cannot be generalised outside of discharges occurring in high-income countries. Second, most of the variables we reported had a high degree of between-study heterogeneity that was not explained by the moderator variable of the length of follow-up, reporting strength, study design, or diagnostic mix. This suggests that there might be contextual factors affecting the strength of suicide risk factors that we did not capture in the meta-analysis. Third, the meta-analysis focused on associations between clinical risk factors and suicide, and potential interactions between risk factors were not examined excepting the indirect analysis in the analysis of high-risk categorisation. Given that suicide is widely believed to be the result of complex interactions of underlying factors, the merits of examination with individual subject-level data, rather than meta-analysis, could be considered. Fourth, this meta-analysis (and the meta-analyses of rates of post-discharge suicide (Chung et al., 2017, 2019) relied on estimates of cumulative mortality at various durations of follow-up. It is likely that cumulative mortality is less sensitive to change over time than mortality rates at specific intervals, and it is therefore possible that a meta-analysis based on OR calculated at specific intervals might be more sensitive to changes over time. Fifth, some important risk factors, such as specific personality disorders, were reported fewer than five times and were therefore not included in this meta-analysis. No protective factor was reported by enough primary studies to examine changes in their protective strength over time. Finally, the associations that we demonstrated cannot be assumed to be causal because of the cross-sectional way in which the risk factors were ascertained in both cohort and controlled primary studies.

This study builds on the earlier meta-analysis of risk factors for post-discharge suicide (Large et al., 2011) and was able to examine a greater number of suicide risk factors because of new primary literature published in the intervening years and the decision to include studies reporting suicides over more than 1-year post-discharge. Similar to the earlier study (Large et al., 2011), high-risk categorisation, previous self-harm, suicidal ideation and depressive symptoms were the strongest risk factors for post-discharge suicide. Previous self-harm and depressive symptoms were also the most commonly reported risk factors used in high-risk categorisation in a meta-analysis of inpatient suicides (Large et al., 2017) suggesting the salience of these risk factors among psychiatrically admitted patients. Both this and the earlier meta-analyses of post-discharge suicide found that male sex, depressive disorder/affective disorder, symptoms of hopelessness/helplessness/worthlessness/guilt, recent social difficulties/adverse life event and unplanned discharge/discharge against medical advice were significantly associated with post-discharge suicide (Large et al., 2011). The current meta-analysis identified a range of significant post-discharge suicide risk factors that did not reach significance in the 2011 meta-analysis. This suggests that the earlier meta-analysis had type II errors due to the smaller sample size of primary studies. These factors included the protective factor of older age, and the risk factors of living alone, more education, family history of mental illness, psychiatric comorbidity, and bipolar disorder.

A notable finding of the Large and associates 2011 meta-analysis was that less psychiatric follow-up (among seven studies) was a protective factor against suicide (Large et al., 2011). Reassuringly, this earlier finding seems to have been a Type I error. In the current meta-analysis, more psychiatric follow-up was not a risk factor for suicide. However, ‘problems with follow-up’ was associated with post-discharge suicide suggesting that the nature of the follow-up is more important than its duration.

Risk factors identified in this meta-analysis but not examined previously by Large et al. include personality disorders and problems with follow-up (Large et al., 2011). Protective factors identified in this meta-analysis but not previously examined include younger age, dementias, and PTSD (Large et al., 2011).

The finding that PTSD was protective against post-discharge suicide was surprising because PTSD appears to be a risk factor for suicide in the general community (Fox et al., 2021.). However, our result only suggests that a primary diagnosis of PTSD is significantly less associated with suicide than other psychiatric diagnoses in populations of people who were so unwell as to require psychiatric admission. People admitted to a psychiatric hospital exhibit not only mental illness but often pose a risk to themself or others, and the admission may reflect a failure of community care. The nature of the psychiatrically discharged patients studied in this meta-analysis might also explain why several factors that are usually considered to be risk factors for suicide in the community were not significantly significant in this study. These include being widowed/divorced, low-socioeconomic status, and substance use disorder.

The overall strength of risk factors in the current meta-analysis was generally lower than we had expected, with only two of 62 individual suicide risk factors having an OR > 2. These modest effect size were generally lower than those reported in a recent meta-analysis of risk factors for suicide in the community (Favril et al., 2022), but were similar to those reported in a large-scale meta-analyses of prospective studies unselected by discharge status (Franklin et al., 2017; Ribeiro et al., 2016). The similarity of our findings to that of Franklin and associates is not surprising because most studies included in our meta-analysis were also prospective and none relied on psychological autopsy methods. However, it may also be that suicide risk factors generally have less predictive strength among populations of admitted people – almost all of whom are likely to have a mental disorder.

The strongest risk factors for post-discharge suicide (high-risk categorisation, depressive symptoms and previous self-harm) did not decline with increasing periods of follow-up and the findings of the meta-regression did not provide a good explanation for the declining rates of suicide over longer periods of follow-up described by Chung (Chung et al., 2017, 2019). This is consistent with primary studies that have reported modest and inconsistent differences in the strength of a variety of suicide risk factors across different post-discharge periods (Aaltonen et al., 2024; Britton et al., 2021; Goldacre et al., 1993; Hunt et al., 2009; Iliachenko et al., 2020; Kessler et al., 2020).

This meta-analysis suggests that widely accepted risk factors for suicide in the community have limited application in populations of discharged psychiatric patients. Given what is known about the high suicide rate in the immediate post-discharge period, the modest statistical association between risk factors and suicide, and the low sensitivity of high-risk categorisation, clinicians lack the tools to make meaningful distinctions between subgroups of discharged patients according to suicide risk. This suggests that all discharged patients must be regarded as being at some risk of suicide, and efforts should be made to ensure comprehensive follow-up for every patient. Emphasis should be placed on the quality of follow-up, given that the risk variable coded as ‘problems with follow-up’ was associated with suicide, while the duration of the follow-up was not.

In 1993, Goldacre and associates hypothesised that discharge itself was a suicide risk factor (Goldacre et al., 1993). It is disappointing that the same inference might be drawn more than 30 years later. Furthermore, potentially important aspects of the patient’s experience of discharge, such as the loss of the support of the hospital, stigma, loss of social role associated with hospitalisation, and the potential for the renewed exposure to problems in the home environment have rarely been studied. The result is that the very high rate of suicide in the immediate post-discharge period remains unexplained and further research is required, particularly into approaches to prevent post-discharge suicide.

The modest strength of the risk factors for suicide, and low sensitivity of high-risk models, also point to the need for more research using advanced methods. Future primary research using machine learning and artificial intelligence might capture patterns of risk that predict suicide with greater statistical power, and shed light on the complex and hereto unpredictable causes of suicide in this setting.

Supplemental Material

sj-docx-1-anp-10.1177_00048674251348372 – Supplemental material for Meta-analysis of risk factors for suicide after psychiatric discharge and meta-regression of the duration of follow-up

Supplemental material, sj-docx-1-anp-10.1177_00048674251348372 for Meta-analysis of risk factors for suicide after psychiatric discharge and meta-regression of the duration of follow-up by Allyson Tai, Hannah Pincham, Ashna Basu and Matthew Large in Australian & New Zealand Journal of Psychiatry

Supplemental Material

sj-docx-2-anp-10.1177_00048674251348372 – Supplemental material for Meta-analysis of risk factors for suicide after psychiatric discharge and meta-regression of the duration of follow-up

Supplemental material, sj-docx-2-anp-10.1177_00048674251348372 for Meta-analysis of risk factors for suicide after psychiatric discharge and meta-regression of the duration of follow-up by Allyson Tai, Hannah Pincham, Ashna Basu and Matthew Large in Australian & New Zealand Journal of Psychiatry

Supplemental Material

sj-docx-3-anp-10.1177_00048674251348372 – Supplemental material for Meta-analysis of risk factors for suicide after psychiatric discharge and meta-regression of the duration of follow-up

Supplemental material, sj-docx-3-anp-10.1177_00048674251348372 for Meta-analysis of risk factors for suicide after psychiatric discharge and meta-regression of the duration of follow-up by Allyson Tai, Hannah Pincham, Ashna Basu and Matthew Large in Australian & New Zealand Journal of Psychiatry

Supplemental Material

sj-docx-4-anp-10.1177_00048674251348372 – Supplemental material for Meta-analysis of risk factors for suicide after psychiatric discharge and meta-regression of the duration of follow-up

Supplemental material, sj-docx-4-anp-10.1177_00048674251348372 for Meta-analysis of risk factors for suicide after psychiatric discharge and meta-regression of the duration of follow-up by Allyson Tai, Hannah Pincham, Ashna Basu and Matthew Large in Australian & New Zealand Journal of Psychiatry

Supplemental Material

sj-docx-5-anp-10.1177_00048674251348372 – Supplemental material for Meta-analysis of risk factors for suicide after psychiatric discharge and meta-regression of the duration of follow-up

Supplemental material, sj-docx-5-anp-10.1177_00048674251348372 for Meta-analysis of risk factors for suicide after psychiatric discharge and meta-regression of the duration of follow-up by Allyson Tai, Hannah Pincham, Ashna Basu and Matthew Large in Australian & New Zealand Journal of Psychiatry

Supplemental Material

sj-xlsx-6-anp-10.1177_00048674251348372 – Supplemental material for Meta-analysis of risk factors for suicide after psychiatric discharge and meta-regression of the duration of follow-up

Supplemental material, sj-xlsx-6-anp-10.1177_00048674251348372 for Meta-analysis of risk factors for suicide after psychiatric discharge and meta-regression of the duration of follow-up by Allyson Tai, Hannah Pincham, Ashna Basu and Matthew Large in Australian & New Zealand Journal of Psychiatry

Footnotes

Statement of Contribution

A.T. was responsible for the initial draft. The literature searches, identification of papers, and data extraction were conducted independently by A.T. and M.L. A.T., H.P., and A.B. performed the data reconciliation. A.T. and M.L. independently conducted the analysis of results and interpretation. This project was conceived by M.L.

Declaration of Conflicting Interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship and/or publication of this article: M.L. has provided evidence in legal matters following post-discharge suicides.

Funding

The author(s) received no financial support for the research, authorship and/or publication of this article.

ORCID iDs

Allyson Tai

Hannah Pincham

Ashna Basu

Matthew Large

Data Availability Statement

All data used in the meta-analysis are available in .

Supplemental Material

Supplemental material for this article is available online.

References

Aaltonen

Sund

Hakulinen

, et al. (2024) Variations in suicide risk and risk factors after hospitalization for depression in Finland, 1996-2017. JAMA Psychiatry 81(5): 506–515.

Aaltonen

Isometsa

Sund

, et al. (2019) Risk factors for suicide in depression in Finland: First-hospitalized patients followed up to 24 years. Acta Psychiatrica Scandinavica 139(2): 154–163.

Agerbo

Mortensen

Eriksson

, et al. (2001) Risk of suicide in relation to income level in people admitted to hospital with mental illness: Nested case-control study. BMJ: British Medical Journal 322(7282): 334–335.

Allgulander

Allebeck

Przybeck

, et al. (1992) Risk of suicide by psychiatric diagnosis in Stockholm County. A longitudinal study of 80,970 psychiatric inpatients. European Archives of Psychiatry and Clinical Neuroscience 241(5): 323–326.

Appleby

Dennehy

Thomas

, et al. (1999) Aftercare and clinical characteristics of people with mental illness who commit suicide: A case-control study. Lancet 353(9162): 1397–1400.

Axelsson

Lagerkvist-Briggs

(1992) Factors predicting suicide in psychotic patients. European Archives of Psychiatry and Clinical Neuroscience 241(5): 259–266.

Beck

Brown

Steer

(1989) Prediction of eventual suicide in psychiatric inpatients by clinical ratings of hopelessness. Journal of Consulting and Clinical Psychology 57(2): 309–310.

Beck

Steer

Kovacs

, et al. (1985) Hopelessness and eventual suicide: A 10-year prospective study of patients hospitalized with suicidal ideation. American Journal of Psychiatry 142(5): 559–563.

Berglund

Krantz

Lundqvist

, et al. (1987) Suicide in psychiatric patients. A prospective study of 67 cases without initial signs of severe depression or alcoholism. Acta Psychiatrica Scandinavica 76(4): 431–437.

10.

Bickley

Hunt

Windfuhr

, et al. (2013) Suicide within two weeks of discharge from psychiatric inpatient care: A case-control study. Psychiatric Services 64(7): 653–659.

11.

Bjorkenstam

Hjern

, et al. (2014) Suicide in first episode psychosis: A nationwide cohort study. Schizophrenia Research 157(1–3): 1–7.

12.

Black

Warrack

Winokur

(1985a) The Iowa record-linkage study. I. Suicides and accidental deaths among psychiatric patients. Archives of General Psychiatry 42(1): 71–75.

13.

Black

Winokur

Warrack

(1985b) Suicide in schizophrenia: The Iowa record linkage study. Journal of Clinical Psychiatry 46(11 II): 14–17.

14.

Borenstein

Hedges

Higgins

, et al. (2022) Comprehensive meta-analysis version 2 Engelwood. NJ Biostat. Available at: https://meta-analysis.com

15.

Bradvik

Berglund

(1993) Risk factors for suicide in melancholia. A case-record evaluation of 89 suicides and their controls. Acta Psychiatrica Scandinavica 87(5): 306–311.

16.

Britton

Bohnert

Ilgen

, et al. (2017) Suicide mortality among male veterans discharged from Veterans Health Administration acute psychiatric units from 2005 to 2010. Social Psychiatry and Psychiatric Epidemiology 52(9): 1081–1087.

17.

Britton

Ilgen

Bohnert

, et al. (2021) Shorter- and longer-term risk for suicide among male us military veterans in the year after discharge from psychiatric hospitalization. Journal of Clinical Psychiatry 82(2): 19M13228.

18.

Carlborg

Jokinen

Nordstrom

, et al. (2009) CSF 5-HIAA, attempted suicide and suicide risk in schizophrenia spectrum psychosis. Schizophrenia Research 112(1–3): 80–85.

19.

Carlborg

Jokinen

Nordstrom

, et al. (2010) Attempted suicide predicts suicide risk in schizophrenia spectrum psychosis. Nordic Journal of Psychiatry 64(1): 68–72.

20.

Che

Gwon

Kim

(2023) Follow-up timing after discharge and suicide risk among patients hospitalized with psychiatric illness. JAMA Network Open 6(10):e2336767.

21.

Chiu

Liu

, et al. (2023) Incidence, risk and protective factors for suicide mortality among patients with major depressive disorder. Asian Journal of Psychiatry 80: 103399.

22.

Choi

Lee

Kim

, et al. (2019) Suicide risk after discharge from psychiatric care in South Korea. Journal of Affective Disorders 251: 287–292.

23.

Choi

Nam

Lee

, et al. (2020) Association of continuity of care with readmission, mortality and suicide after hospital discharge among psychiatric patients. International Journal for Quality in Health Care 32(9): 569–576.

24.

Chung

Hadzi-Pavlovic

Wang

, et al. (2019) Meta-analysis of suicide rates in the first week and the first month after psychiatric hospitalisation. BMJ Open 9(3): e023883.

25.

Chung

Ryan

Hadzi-Pavlovic

, et al. (2017) Suicide rates after discharge from psychiatric facilities: A systematic review and meta-analysis. JAMA Psychiatry 74(7): 694–702.

26.

Clapham

Boden

Brandt

, et al. (2019) Suicide ideation and behavior as risk factors for subsequent suicide in schizophrenia: A nested case-control study. Suicide & Life-Threatening Behavior 49(4): 996–1005.

27.

De Hert

McKenzie

Peuskens

(2001) Risk factors for suicide in young people suffering from schizophrenia: A long-term follow-up study. Schizophrenia Research 47(2–3): 127–134.

28.

Deisenhammer

Behrndt

Kemmler

, et al. (2019) Suicide risk factors in patients recently discharged from a psychiatric hospital: A case-control study. Journal of Clinical Psychiatry 80(5): 18m12702.

29.

Dennehy

Appleby

Thomas

, et al. (1996) Case-control study of suicide by discharged psychiatric patients. British Medical Journal 312(7046): 1580.

30.

Desai

Dausey

Rosenheck

(2005) Mental health service delivery and suicide risk: The role of individual patient and facility factors. American Journal of Psychiatry 162(2): 311–318.

31.

Enns

Mota

Bolton

, et al. (2022) Incidence and predictors of suicide attempts and suicide deaths among individuals recently hospitalized for a mental disorder: A population-based study. Journal of Clinical Psychiatry 83(4): E1–E6.

32.

Farberow

MacKinnon

(1974) A suicide prediction schedule for neuropsychiatric hospital patients. The Journal of Nervous and Mental Disease 158(6): 408–419.

33.

Favril

Uyar

, et al. (2022) Risk factors for suicide in adults: Systematic review and meta-analysis of psychological autopsy studies. BMJ Mental Health 25(4): 148–155.

34.

Fazel

Grann

Goodwin

(2006) Suicide trends in discharged patients with mood disorders: Associations with selective serotonin uptake inhibitors and comorbid substance misuse. International Clinical Psychopharmacology 21(2): 111–115.

35.

Flood

Seager

(1968) A retrospective examination of psychiatric case records of patients who subsequently committed suicide. The British Journal of Psychiatry 114(509): 443–450.

36.

Fowler

Tsuang

Kronfol

(1979) Communication of suicidal intent and suicide in unipolar depression: A forty year follow-up. Journal of Affective Disorders 1(3): 219–225.

37.

Fox

Dalman

Dal

, et al. (2021) Suicide risk in people with post-traumatic stress disorder: A cohort study of 3.1 million people in Sweden. Journal of Affective Disorders 279: 609–616.

38.

Franklin

Ribeiro

Fox

, et al. (2017) Risk factors for suicidal thoughts and behaviors: A meta-analysis of 50 years of research. Psychological Bulletin 143(2): 187–232.

39.

Fredriksen

Gjestad

Walby

, et al. (2022) High scores on the Montgomery-Asberg Depression Rating Scale and psychotic symptoms predict suicide: A prospective cohort study of psychiatric acute ward patients. Journal of Clinical Psychiatry 83(5): 21m14018.

40.

Geddes

Juszczak

(1995) Period trends in rate of suicide in first 28 days after discharge from psychiatric hospital in Scotland, 1968-92. British Medical Journal 311(7001): 357–360.

41.

Geddes

Juszczak

O’Brien

, et al. (1997) Suicide in the 12 months after discharge from psychiatric inpatient care, Scotland 1968-92. Journal of Epidemiology and Community Health 51(4): 430–434.

42.

Goldacre

Seagroatt

Hawton

(1993) Suicide after discharge from psychiatric inpatient care. The Lancet 342(8866): 283–286.

43.

Goldstein

Black

Nasrallah

, et al. (1991) The prediction of suicide. Sensitivity, specificity, and predictive value of a multivariate model applied to suicide among 1906 patients with affective disorders. Journal Archives of General Psychiatry 48(5): 418–422.

44.

Grover

Huber

Brewer

, et al. (2023) Meta-analysis of clinical risk factors for suicide among people presenting to emergency departments and general hospitals with suicidal thoughts and behaviours. Acta Psychiatrica Scandinavica 148(6): 491–524.

45.

Guimaraes

Cherchiglia

da Rocha

, et al. (2022) Factors associated with risk of death by suicide after psychiatric hospitalization by the Unified Health System in Brazil (2002-2015). General Hospital Psychiatry 77: 69–76.

46.

Høyer

Olesen

Mortensen

(2004) Suicide risk in patients hospitalised because of an affective disorder: A follow-up study, 1973–1993. Journal of Affective Disorders 78(3): 209–217.

47.

Haglund

Lysell

Larsson

, et al. (2019) Suicide immediately after discharge from psychiatric inpatient care: A cohort study of nearly 2.9 million discharges. Journal of Clinical Psychiatry 80(2): 18m12172.

48.

Heu

Bogren

Wang

, et al. (2018) Aspects of additional psychiatric disorders in severe depression/melancholia: A comparison between suicides and controls and general pattern. International Journal of Environmental Research and Public Health 15(7): 1299.

49.

Hjorthøj

Madsen

Agerbo

, et al. (2014) Risk of suicide according to level of psychiatric treatment: A nationwide nested case-control study. Social Psychiatry and Psychiatric Epidemiology 49(9): 1357–1365.

50.

(2003) The suicide risk of discharged psychiatric patients. Journal of Clinical Psychiatry 64(6): 702–707.

51.

(2006) Duration of hospitalization and post discharge suicide. Suicide and Life-Threatening Behavior 36(6): 682–686.

52.

Hoffmann

Modestin

(1987) Completed suicide in discharged psychiatric inpatients. Social Psychiatry 22(2): 93–98.

53.

Hogan

Awad

(1983) Pharmacotherapy and suicide risk in schizophrenia. The Canadian Journal of Psychiatry 28(4): 277–281.

54.

Hong

Lee

Han

, et al. (2024) Suicide death and other-cause mortality in psychiatric patients: A South Korean study using nationwide claims data. Journal of Affective Disorders 352: 288–295.

55.

Hunt

Kapur

Webb

, et al. (2009) Suicide in recently discharged psychiatric patients: A case-control study. Psychological Medicine 39(3): 443–449.

56.

Iliachenko

Ragazan

Eberhard

, et al. (2020) Suicide mortality after discharge from inpatient care for bipolar disorder: A 14-year Swedish national registry study. Journal of Psychiatric Research 127: 20–27.

57.

Isometsa

Sund

Pirkola

(2014) Post-discharge suicides of inpatients with bipolar disorder in Finland. Bipolar Disorders 16(8): 867–874.

58.

Jiang

Rosellini

Horvath-Puho

, et al. (2021) Using machine learning to predict suicide in the 30 days after discharge from psychiatric hospital in Denmark. British Journal of Psychiatry 219(2): 440–447.

59.

Kan

Dong

JYS

, et al. (2007) Risk factors for suicide in the immediate post-discharge period. Social Psychiatry and Psychiatric Epidemiology 42(3): 208–214.

60.

Kaster

Blumberger

Gomes

, et al. (2022) Risk of suicide death following electroconvulsive therapy treatment for depression: A propensity score-weighted, retrospective cohort study in Canada. The Lancet Psychiatry 9(6): 435–446.

61.

Kessler

Bauer

Bishop

, et al. (2020) Using administrative data to predict suicide after psychiatric hospitalization in the veterans health administration system. Frontiers in Psychiatry 11: 390.

62.

Kessler

Warner

Ivany

, et al. (2014) Predicting suicides after psychiatric hospitalization in US army soldiers: The Army Study to Assess Risk and Resilience in Servicemembers (Army STARRS). JAMA Psychiatry 72(1): 49–57.

63.

King

(2001) The Wessex suicide audit 1988-1993: A study of 1457 suicides with and without a recent psychiatric contact. International Journal of Psychiatry in Clinical Practice 5(2): 111–118.

64.

Kuo

C-J

Tsai

S-Y

Liao

Y-T

, et al. (2010) Psychiatric discharge against medical advice is a risk factor for suicide but not for other causes of death. The Journal of Clinical Psychiatry 71(6): 808–809.

65.

Kuo

C-J

Tsai

, et al. (2005) Risk factors for completed suicide in schizophrenia. Journal of Clinical Psychiatry 66(5): 579–585.

66.

Large

Myles

, et al. (2017) Suicide risk assessment among psychiatric inpatients: A systematic review and meta-analysis of high-risk categories. Psychological Medicine 48(7): 1119–1127.

67.

Large

Sharma

Cannon

, et al. (2011) Risk factors for suicide within a year of discharge from psychiatric hospital: A systematic meta-analysis. Australian and New Zealand Journal of Psychiatry 45(8): 619–628.

68.

Lee

Lin

(2009) Are psychiatrist characteristics associated with postdischarge suicide of schizophrenia patients? Schizophrenia Bulletin 35(4): 760–765.

69.

Lin

Lee

Kuo

, et al. (2008) Hospital characteristics associated with post-discharge suicide of severely depressed patients. Journal of Affective Disorders 110(3): 215–221.

70.

Luxton

Trofimovich

Clark

(2013) Suicide risk among U.S. service members after psychiatric hospitalization, 2001-2011. Psychiatric Services 64(7): 626–629.

71.

McKenzie

Wurr

(2001) Early suicide following discharge from a psychiatric hospital. Suicide and Life-Threatening Behavior 31(3): 358–363.

72.

Madsen

Erlangsen

Hjorthøj

, et al. (2020) High suicide rates during psychiatric inpatient stay and shortly after discharge. Acta Psychiatrica Scandinavica 142(5): 355–365.

73.

Metreau

Young

Eapen

(2024) World Bank Country Classifications by Income Level for 2024-2025. Available at: https://blogs.worldbank.org/en/opendata/world-bank-country-classifications-by-income-level-for-2024-2025 (accessed 1 December 2024).

74.

Modestin

Schwarzenbach

(1992) Effect of psychopharmacotherapy on suicide risk in discharged psychiatric inpatients. Acta Psychiatrica Scandinavica 85(2): 173–175.

75.

Moher

Liberati

Tetzlaff

, et al. (2009) Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement. Journal of Clinical Epidemiology 62(10): 1006–1012.

76.

Motto

Heilbron

Juster

(1985) Development of a clinical instrument to estimate suicide risk. American Journal of Psychiatry 142(6): 680–686.

77.

Motto

Bostrom

(1990) Empirical indicators of near-term suicide risk. Crisis 11(1): 52–59.

78.

Nielsen

Christensen

RHB

Madsen

, et al. (2023) Prediction models of suicide and non-fatal suicide attempt after discharge from a psychiatric inpatient stay: A machine learning approach on nationwide Danish registers. Acta Psychiatrica Scandinavica 148(6): 525–537.

79.

Nordström

Asberg

Aberg-Wistedt

, et al. (1995) Attempted suicide predicts suicide risk in mood disorders. Acta Psychiatrica Scandinavica 92(5): 345–350.

80.

Olfson

Wall

Wang

, et al. (2016) Short-term suicide risk after psychiatric hospital discharge. JAMA Psychiatry 73(11): 1119–1126.

81.

Park

Choi

Kyoung Yi

, et al. (2013) Suicide mortality and risk factors in the 12 months after discharge from psychiatric inpatient care in Korea: 1989-2006. Psychiatry Research 208(2): 145–150.

82.

Pirkola

Sohlman

Wahlbeck

(2005) The characteristics of suicides within a week of discharge after psychiatric hospitalisation – A nationwide register study. BMC Psychiatry 5: 32.

83.

Plans

Nieto

Benabarre

, et al. (2019) Completed suicide in bipolar disorder patients: A cohort study after first hospitalization. Journal of Affective Disorders 257: 340–344.

84.

Pokorny

(1983) Prediction of suicide in psychiatric patients. Report of a prospective study. Journal Archives of General Psychiatry 40(3): 249–257.

85.

Pokorny

Kaplan

(1976) Suicide following psychiatric hospitalization. Journal of Nervous and Mental Disease 162(2): 119–125.

86.

Prestmo

Hoyen

Vaaler

, et al. (2023) Post-discharge suicide among high-risk psychiatric inpatients: Risk factors and warnings signs. Journal of Affective Disorders Reports 12: 100506.

87.

Reutfors

Brandt

Jonsson

, et al. (2009) Risk factors for suicide in schizophrenia: Findings from a Swedish population-based case-control study. Schizophrenia Research 108(1-3): 231–237.

88.

Ribeiro

Franklin

Fox

, et al. (2016) Self-injurious thoughts and behaviors as risk factors for future suicide ideation, attempts, and death: A meta-analysis of longitudinal studies. Psychological Medicine 46(2): 225–236.

89.

Riblet

Richardson

Shiner

, et al. (2018) Death by suicide in the first year after irregular discharge from inpatient hospitalization. Psychiatric Services 69(9): 1032–1035.

90.

Rosenthal

(1996) Qualitative descriptors of strength of association and effect size. Journal of Social Service Research 21(4): 37–59.

91.

Sani

Tondo

Koukopoulos

, et al. (2011) Suicide in a large population of former psychiatric inpatients. Psychiatry and Clinical Neurosciences 65(3): 286–295.

92.

Shaffer

Perlin

Schmidt

Jr , et al. (1974) The prediction of suicide in schizophrenia. The Journal of Nervous and Mental Disease 159(5): 349–355.

93.

Stephens

Richard

McHugh

(1999) Suicide in patients hospitalized for schizophrenia: 1913-1940. Journal of Nervous and Mental Disease 187(1): 10–14.

94.

Stroup

Berlin

Morton

, et al. (2000) Meta-analysis of observational studies in epidemiology: A proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. Jama 283(15): 2008–2012.

95.

Tsai

SYM

Kuo

Chen

, et al. (2002) Risk factors for completed suicide in bipolar disorder. Journal of Clinical Psychiatry 63(6): 469–476.

96.

Tseng

MCM

Chang

Liao

, et al. (2020) Length of stay in relation to the risk of inpatient and post-discharge suicides: A national health insurance claim data study. Journal of Affective Disorders 266: 528–533.

97.

Wells

Shea

O’Connell

, et al. (2021) The Newcastle-Ottawa Scale (NOS) for Assessing the Quality of Nonrandomised Studies in Meta-Analyses. Available at: www.ohri.ca/programs/clinical_epidemiology/oxford.asp

98.

Yim

PHW

Yip

PSF

RHY

, et al. (2004) Suicide after discharge from psychiatric inpatient care: A case-control study in Hong Kong. Australian and New Zealand Journal of Psychiatry 38(1–2): 65–72.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.03 MB

0.18 MB

0.16 MB

0.04 MB

0.03 MB

0.17 MB