Introduction
Suicide, being the act of intentionally ending one's own life, has been a major public health challenge worldwide. Suicide-related behaviours consist of suicidal ideation (SI), suicide plan (SP), suicide attempt (SA) and completed suicide (CS) (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; Scocco et al. Reference Scocco, de Girolamo, Vilagut and Alonso2008). While SI refers to thoughts and fantasies or a wish to die, SP is defined as the plan how to end one's own life and SA refers to a self-destructive act with an intent to end one's life. CS is the act of suicide resulting in death (Ran et al. Reference Ran, Wu, Conwell, Chen and Chan2004; Suominen et al. Reference Suominen, Isometsä, Suokas, Haukka, Achte and Lönnqvist2004; Lee et al. Reference Lee, Fung, Tsang, Liu, Huang, He, Zhang, Shen, Nock and Kessler2007; Scocco et al. Reference Scocco, de Girolamo, Vilagut and Alonso2008; Kao et al. Reference Kao, Liu, Cheng and Chou2012). Suicide-related behaviours are significantly associated with suicide (Harkavy-Friedman et al. Reference Harkavy-Friedman, Restifo, Malaspina, Kaufmann, Amador, Yale and Gorman1999; WHO, 2012). Understanding the patterns of suicide-related behaviours is important to develop and implement effective measures to reduce the risk of suicide. A large number of studies have been conducted in general populations all over the world. For example, a survey of 84 850 adults in 17 countries found that lifetime prevalence of SI, SP and SA was 9.25, 3.1, 2.7% in the general population, respectively (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). In the USA, the prevalence of SI, SP and SA was 13.5, 3.9 and 4.6% respectively; 34% of persons with SI proceeded to a SP, and 72% of those with a SP attempted suicide (Kessler et al. Reference Kessler, Borges and Walters1999).
Schizophrenia is a severe psychiatric disorder characterised by a variety of psychotic symptoms coupled with cognitive impairment and behavioural dysfunction. Schizophrenia causes immeasurable suffering for the individual and poses a significant psychosocial and economic burden to families and societies (van Os & Kapur, Reference van Os and Kapur2009). Schizophrenia is significantly associated with suicide-related behaviours (Pompili et al. Reference Pompili, Amador, Girardi, Harkavy-Friedman, Harrow, Kaplan, Krausz, Lester, Meltzer, Modestin, Montross, Mortensen, Munk-Jorgensen, Nielsen, Nordentoft, Saarinen, Zisook, Wilson and Tatarelli2007; Hor & Taylor, Reference Hor and Taylor2010), and suicide is a primary cause of death in young patients with schizophrenia (De Hert et al. Reference De Hert, McKenzie and Peuskens2001). A meta-analysis found that 4.9% of schizophrenia patients committed suicide during their lifetime. However, a major limitation was that most data included in this meta-analysis came from Western countries (Palmer et al. Reference Palmer, Pankratz and Bostwick2005).
China has the largest schizophrenia population (7.16 million) in the world (Phillips et al. Reference Phillips, Zhang, Shi, Song, Ding, Pang, Li, Zhang and Wang2009; Chan et al. Reference Chan, Zhao, Meng, Demaio, Reed, Theodoratou, Campbell, Wang and Rudan2015). There is compelling evidence that sociocultural factors have a strong influence on suicide patterns; therefore, the findings reported in the West may not apply to China (Ran et al. Reference Ran, Xiang, Mao, Hou, Tang, Chen, Chan, Yip and Conwell2005; Zhong et al. Reference Zhong, Chiu and Conwell2016). In the past decade, several studies have examined the prevalence of suicide-related behaviours in Chinese patients with schizophrenia yielding inconsistent results. For example, the lifetime prevalence of SI varied between 7.4% (Zheng et al. Reference Zheng, Tang, Weng, Zhao, Wang, Ma and Xiang2015) and 57.6% (Yan & Wei, Reference Yan and Wei2012), and the figures of SA varied between 3.2% (Xue & Zhang, Reference Xue and Zhang2006) and 51.0% (Kao et al. Reference Kao, Liu, Cheng and Chou2012). One study (Phillips et al. Reference Phillips, Yang, Li and Li2004) found that 10.1% of schizophrenia patients died of suicide in China but in another study, the figure was 4.2% (Ran et al. Reference Ran, Chen, Conwell, Chan, Yip, Xiang and Caine2007). In addition, most studies were done in a single hospital or a particular province; therefore, the findings do not necessarily reflect the nationwide patterns of suicide-related behaviours in schizophrenia patients.
The objective of this meta-analysis was to examine the pooled prevalence of suicide-related behaviours in schizophrenia patients in China and to explore their mediating factors. We hypothesised that schizophrenia patients would have a higher rate of suicide-related behaviours compared with the general population in China.
Methods
Search strategy and selection criteria
This meta-analysis was conducted and reported according to the MOOSE recommendations (Stroup et al. Reference Stroup, Berlin, Morton, Olkin, Williamson, Rennie, Moher, Becker, Sipe and Thacker2000). The relevant literature was searched systematically via the following electronic databases, including PubMed, Embase, PsycINFO, Chinese National Knowledge Infrastructure (CNKI), Wanfang Databases and Chinese Biological Medical Literature Database from their inception until 14 September 2016. Articles were identified using the following search terms: (‘suicide*’ or ‘self-injurious behavior’ or ‘self-mutilation’ or ‘self-immolation’ or ‘self-harm’ or ‘self-inflicted’ or ‘self-injury’ or ‘self-slaughter’ or ‘self-destruction’) and (‘schizophrenia’ or ‘psychotic disorder’ or ‘psychosis’) and (‘epidemiology’ or ‘cross-sectional study’ or ‘prevalence’ or ‘rate’ or ‘risk factor’ or ‘cohort study’ or ‘observational study’) and (‘China’ or ‘Chinese’). The titles and abstracts were screened independently by two investigators (MD and SBW) who both had at least 5 years experiences in clinical research. The full texts of potentially eligible articles were downloaded for further screening. Any disagreement was resolved by discussion with a third investigator to reach a consensus.
Inclusion and exclusion criteria
The articles included in study fulfilled the following criteria: (1) reporting on inpatients and outpatients with schizophrenia by any diagnostic criteria; (2) cross-sectional or cohort study conducted in China; (3) data reported on prevalence of suicide-related behaviours including SI, SP, SA or CS; (4) suicide-related behaviours assessed with standardised questionnaires or questions; and (5) were published in Chinese or English. If more than one article was based on the same dataset, only the publication with the largest sample size was selected. Studies conducted in specific settings (e.g., prison or nursing home) and retrospective surveys were excluded.
Data extraction
Two investigators extracted information with a standardised form that included the authors’ name, publication and survey year, study site (province), type of region (urban/rural), source of sample (inpatients or outpatients), sampling method, sample size, mean age of the patients, proportion of men, screening method and tools and timeframe (i.e., point, 12-month and lifetime prevalence).
Quality assessment
The methodological quality of the included studies was evaluated using the 22-item Strengthening the Reporting of Observational Studies in Epidemiology (von Elm et al. Reference von Elm, Altman, Egger, Pocock, Gotzsche and Vandenbroucke2007). Study quality was defined as low quality when the total score was ⩽11; high quality was defined as the total score >11 (Cao et al. Reference Cao, Zhong, Xiang, Ungvari, Lai, Chiu and Caine2015).
Statistical analysis
Data were analysed using the Comprehensive Meta-Analysis (CMA), Version 2.0 (Biosta, Inc. Englewood, New Jersey, USA). The pooled prevalence estimates and their 95% confidence intervals (95% CI) were conducted using the random-effects models. The I 2 statistic was calculated to measure heterogeneity (Higgins et al. Reference Higgins, Thompson, Deeks and Altman2003). When heterogeneity was present (I 2 > 50%), sensitivity and subgroup analyses were used to explore the reasons for heterogeneity. Publication bias was assessed by the funnel plot and Egger's test (Egger et al. Reference Egger, Davey Smith, Schneider and Minder1997). Significance level was set at 0.05 (two-sided).
Results
Search results
A total of 1136 studies were initially identified (Fig. 1). After excluding the duplications, 670 articles were reviewed by title and abstracts. Finally, 19 articles, which met the inclusion criteria, were included for analyses.
Fig. 1. Flowchart for study selection.
Study characteristics and quality assessment
Table 1 shows the general characteristics of the studies. Sample sizes ranged from 42 to 1655 and the mean age of patients ranged from 31.0 to 43.9 years. Consecutive sampling was used in ten studies. The total STROBE scores ranged between 8 and 20; 16 articles were rated as high quality and three were low quality.
Table 1. Summary of the studies included in the meta-analysis
BSI-CV, Chinese version of the Beck Scale for Suicide Ideation; BSS, Beck Scale for Suicide Ideation; DH, during hospitalisation; HAMD, Hamilton Depression Rating Scale for Depression; LF, lifetime; NA, not applicable; PFS, Patients Follow-up Schedule; SA, suicide attempts; SI, suicidal ideation; SP, suicide plan; SSI, Scale for Suicide Ideation; 1 M, 1 month.
Seven studies reported lifetime prevalence of SI, three studies reported 1-month prevalence and one study reported the prevalence during hospitalisation. Only one study reported the lifetime prevalence of SP. Thirteen studies reported the lifetime prevalence of SA, while one study reported the 1-month prevalence and another study reported the prevalence during hospitalisation.
Pooled prevalence of suicide-related behaviours
The lifetime prevalence of SI in seven studies with a total 1017 patients was 25.8% (95% CI 14.7–41.1%, I 2 = 95.49%) (Fig. 2a). The 1-month prevalence of SI in three studies with 642 patients was 22.0% (95% CI 18.2–26.4%, I 2 = 12.57%) (Fig. 2b). The lifetime prevalence of SA in 13 studies with 5098 patients was 14.6% (95% CI 9.1–22.8%, I 2 = 97.02%) (Fig. 2c).
Fig. 2. Forest plot of suicide-related behaviours. (a) Lifetime prevalence of suicidal ideation. (b) One-month prevalence of suicidal ideation. (c) Lifetime prevalence of suicide attempts.
Some results could not be pooled because they were reported only in one study: Zhang (Reference Zhang1998) and Wu (Reference Wu2002) reported that the prevalence of SI and SA during hospitalisation was 40.8 and 10.5%, respectively. One study (Deng, Reference Deng2000) found that the 1-month prevalence of SA was 11.7%; another study (Zhang & Xiao, Reference Zhang and Xiao2002) found that the lifetime prevalence of SP was 16.7%.
Sensitivity analysis and publication bias
In sensitivity analyses, after each study was sequentially excluded, the recalculated results did not change significantly indicating that none of the studies significantly influenced the overall results of the meta-analysis. The publication bias concerning the lifetime prevalence of SA was tested with the visual inspection of the funnel plot graphic indicating slight asymmetry (Supplementary Fig. S1), but the Egger's test did not reveal any publication bias (t = 0.25, p = 0.81).
Subgroup analysis
Table 2 shows the results of the subgroup analyses. The lifetime prevalence of SI in men and women was 29.6% (95% CI 12.8–54.7%) and 24.1% (95% CI 9.6–48.9%), respectively. The lifetime prevalence of SA in men was 13.0% (95% CI 5.9–25.9%) and the figure in women was 13.8% (95% CI 6.3–27.3%). Using the median splitting method, the lifetime prevalence of SI was 17.8% (95% CI 7.4–37.0%) in studies with the mean age ⩽35.8 years, while the corresponding figure was 28.8% (95% CI 12.9–52.7%) in studies with the mean age >35.8 years. The lifetime prevalence of SA was 11.0% (95% CI 3.9–27.2%) in studies with the mean age ⩽38 years, while the corresponding figure was 16.2% (95% CI 5.5–39.5%) in studies with the mean age >38 years.
Table 2. Overall and subgroup prevalence of suicidal ideation and suicide attemptsa
a Suicide plan was not included in subgroup analyses due to limited number.
b Median splitting method were used to establish the cutoff values.
Using the median splitting method, the lifetime prevalence of SI was 17.0% (95% CI 8.3–31.7%) in studies with a sample size ⩽145, while the corresponding figure was 40.5% (95% CI 21.6–62.7%) in studies with a sample size >145. The lifetime prevalence of SA was 20.8% (95% CI 11.7–34.2%) in studies with a sample size ⩽194, and the corresponding figure was 9.8% (95% CI 5.0–18.1%) in studies with a sample size >194.
The lifetime prevalence of SI was 32.4% (95% CI 13.3–59.8%) in studies conducted during or before May 2002, and 13.1% (95% CI 5.4–28.5%) in those conducted after May 2002. The lifetime prevalence of SA was 12.0% (95% CI 5.5–24.2%) in studies conducted during or before May 2002 and 11.5% (95% CI 5.7–21.5%) in studies conducted after May 2002. The lifetime prevalence of SI in the east region (27.3%) was slightly higher than in the middle-west region of China (23.6%), although the difference was not significant. The pooled lifetime prevalence of SA in the east region (14.7%) was similar to that in the middle-west region (14.5%). Lifetime prevalence of SI was higher in outpatients (37.4%) than in inpatients (18.5%), but the lifetime prevalence of SA in inpatients (17.6%) was higher than in outpatients (11.5%).
The lifetime prevalence of SI was 26.8% (95% CI 14.4–44.3%) in cross-sectional studies, and 20.1% (95% CI 3.7–62.1%) in cohort studies. The lifetime prevalence of SA was 12.9% (95% CI 4.4–32.2%) in cross-sectional studies and 15.2% (95% CI 8.6–25.4%) in cohort studies. The lifetime prevalence of SI and SA was 44.9% (95% CI 19.4–73.3%) and 13.6% (95% CI 7.3–23.9%), respectively, in studies using standardised questionnaires, while the corresponding figures were 19.7% (95% CI 10.0–34.9%) and 22.1% (95% CI 5.8–56.7%), in studies using self-designed questionnaires.
Discussion
To the best of our knowledge, this was the first meta-analysis on the pooled prevalence of suicide-related behaviours in Chinese patients with schizophrenia. The lifetime prevalence (25.8%) and 1-month prevalence (22.0%) of SI in Chinese schizophrenia patients were lower than their Western counterparts (40%) (Fenton et al. Reference Fenton, McGlashan, Victor and Blyler1997). Similarly, the lifetime prevalence of SA in schizophrenia in China (14.6%) was much lower than in India (23.3%), USA (48.3%) and Norway (30.5%) (Bhatia et al. Reference Bhatia, Thomas, Semwal, Thelma, Nimgaonkar and Deshpande2006; Barrett et al. Reference Barrett, Sundet, Simonsen, Agartz, Lorentzen, Mehlum, Mork, Andreassen and Melle2011). The discrepancy in the results across studies could be due to differences in demographic, socioeconomic and cultural factors, sampling methods, and criteria for defining suicide-related behaviours. For example, compared with Western countries, under-reporting of suicide-related behaviour is more common in China due to fear of stigmatisation and discrimination associated with suicide in traditional Chinese culture (Ma et al. Reference Ma, Xiang, Cai, Li, Xiang, Guo, Hou, Li, Li, Tao, Dang, Wu, Deng, Chan, Ungvari and Chiu2009).
The risk factors of suicide-related behaviours in the general population include male gender, being single, young age, having high education level, thoughts of hopelessness and substance abuse. Clinical risk factors of suicide-related behaviours in schizophrenia include insomnia, chronicity of illness, poor treatment adherence, late onset of illness and frequent admissions (Hawton et al. Reference Hawton, Sutton, Haw, Sinclair and Deeks2005; Pompili et al. Reference Pompili, Lester, Grispini, Innamorati, Calandro, Iliceto, De Pisa, Tatarelli and Girardi2009; Popovic et al. Reference Popovic, Benabarre, Crespo, Goikolea, Gonzalez-Pinto, Gutierrez-Rojas, Montes and Vieta2014). Unlike in Western countries, the prevalence of suicide is higher in female Chinese schizophrenia patients than men, and in rural China than in urban areas (Phillips et al. Reference Phillips, Yang, Li and Li2004). Possible reasons may include the heavy personal and financial impact for women with several mental illness and the easy access to lethal agents, particularly pesticides, in rural areas (Law & Liu, Reference Law and Liu2008). In this study however, the lifetime prevalence of SI in women was lower than in men (24.1 v. 29.6%), which is inconsistent with a previous study in China (Phillips et al. Reference Phillips, Li and Zhang2002). Men may be more likely to be subjected to stigmatisation related to schizophrenia than women (Crisp et al. Reference Crisp, Gelder, Goddard and Meltzer2005), which may account for the higher risk. In contrast, the lifetime prevalence of SA was similar between the both genders (13.8% in women v. 13.0% in men). Although previous studies found that CS was higher in men than in women with schizophrenia (Ran et al. Reference Ran, Chen, Conwell, Chan, Yip, Xiang and Caine2007; Hor & Taylor, Reference Hor and Taylor2010), this could not be examined due to the small number of studies with relevant data.
In this study, the lifetime prevalence of SI and SA was higher in older patients, which is inconsistent with previous findings that younger age was associated with a higher suicide risk in schizophrenia (Popovic et al. Reference Popovic, Benabarre, Crespo, Goikolea, Gonzalez-Pinto, Gutierrez-Rojas, Montes and Vieta2014), which could be partly due to different study designs. This meta-analysis included both cross-sectional and cohort studies, while the previous review mainly had case–control and cohort control studies (Popovic et al. Reference Popovic, Benabarre, Crespo, Goikolea, Gonzalez-Pinto, Gutierrez-Rojas, Montes and Vieta2014). In addition, the previous study examined risk factors of completed suicide in schizophrenia, while the present study explored the moderators of SI and SA using subgroup analyses.
With larger sample size studies, the prevalence of SI was higher and the prevalence of SA was lower than in smaller sample size studies, although no publication bias was found. There is no satisfactory explanation for this result except noting that the results of studies with small sample size were relatively unstable. The prevalence of SI in studies published during and before May 2002 was more than twofold higher compared with studies published after that (32.4 v. 13.1%), while the prevalence of SA was similar between the two periods (12.0 v. 11.5%). This is in line with the decreasing trends of suicide rate in both the Chinese general population (Wang et al. Reference Wang, Chan and Yip2014) and schizophrenia (Miles, Reference Miles1977; Caldwell & Gottesman, Reference Caldwell and Gottesman1990; Inskip et al. Reference Inskip, Harris and Barraclough1998; Palmer et al. Reference Palmer, Pankratz and Bostwick2005; Hor & Taylor, Reference Hor and Taylor2010). The falling suicide rate may be related improved socioeconomic conditions (Wang et al. Reference Wang, Chan and Yip2014), and the increased attention to suicide prevention in China.
Suicide-related behaviours are largely influenced by the sociocultural and economic background (Xiang et al. Reference Xiang, Weng, Leung, Tang and Ungvari2008; Ma et al. Reference Ma, Xiang, Cai, Li, Xiang, Guo, Hou, Li, Li, Tao, Dang, Wu, Deng, Chan, Ungvari and Chiu2009); improvement of economic conditions could possibly be associated with a reduction in suicide rates (Yin et al. Reference Yin, Xu, Shao, Li and Wan2016). Certain gene polymorphisms, such as COMT Val158Met, 5HTR2A-T102C and rs6313 (T102C) (Kia-Keating et al. Reference Kia-Keating, Glatt and Tsuang2007; Calati et al. Reference Calati, Porcelli, Giegling, Hartmann, Moller, De Ronchi, Serretti and Rujescu2011; Gonzalez-Castro et al. Reference Gonzalez-Castro, Tovilla-Zarate, Juarez-Rojop, Pool Garcia, Velazquez-Sanchez, Genis, Nicolini and Lopez Narvaez2013), are also associated with the risk of suicide. Hence, variations in ethnic composition, geographical area and economic status may influence suicide patterns. Subgroup analyses between different regions defined by the Chinese economic zone found that the lifetime prevalence of SI and SA in the eastern region is slightly higher than the mid-western part of China (SI: 27.3 v. 23.6%, SA: 14.7 v. 14.5%).
Compared with SI, SA is a more severe suicide-related behaviour that is significantly associated with suicide in schizophrenia (Drake et al. Reference Drake, Gates, Whitaker and Cotton1985). Higher prevalence of SA in inpatient population than in other settings is consistent with the suicide rate with previous studies in inpatient units (Carlborg et al. Reference Carlborg, Winnerbäck, Jönsson, Jokinen and Nordström2010). Severe positive symptoms which often result in inpatient treatment, are risk factors of suicide in schizophrenia (Krupinski et al. Reference Krupinski, Fischer, Grohmann, Engel, Hollweg and Moller2000). In this meta-analysis, only one study reported the lifetime prevalence of SP (16.7%) (Zhang & Xiao, Reference Zhang and Xiao2002), which was much higher compared with the Chinese general population (0.9%) (Lee et al. Reference Lee, Fung, Tsang, Liu, Huang, He, Zhang, Shen, Nock and Kessler2007). This meta-analysis however could not pool the prevalence of CS because no studies met the inclusion criteria for such analysis. In China, the prevalence of CS in schizophrenia over 30 years was 14.3% (Zhao et al. Reference Zhao, Ma, Wang, Guo and Liu1992). Other studies found that the prevalence of CS in schizophrenia during the first 5 years of illness was 2.4% (Gonzalez-Pinto et al. Reference Gonzalez-Pinto, Aldama, Gonzalez, Mosquera, Arrasate and Vieta2007).
Compared with the general population, schizophrenia patients tend to use more violent and lethal methods to commit suicide (Harkavy-Friedman et al. Reference Harkavy-Friedman, Restifo, Malaspina, Kaufmann, Amador, Yale and Gorman1999; Hunt et al. Reference Hunt, Kapur, Windfuhr, Robinson, Bickley, Flynn, Parsons, Burns, Shaw and Appleby2006). SI and SA are important predictors for CS (De Leo et al. Reference De Leo, Cerin, Spathonis and Burgis2005; Kessler et al. Reference Kessler, Berglund, Borges, Nock and Wang2005). The progression from SI, SP and SA into CS requires access to the means for suicide. In China, pesticide indigestion, hanging, cutting wrist, jumping from height and drowning have been the common ways of suicide for schizophrenia patients (Jiang et al. Reference Jiang, Gao, Liu and Su1998; Xue, Reference Xue2010). Around 60% of patients with SI made their first SA during the subsequent year (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). History of SA and frequent psychiatric admissions were major risk factors of suicide in schizophrenia (Popovic et al. Reference Popovic, Benabarre, Crespo, Goikolea, Gonzalez-Pinto, Gutierrez-Rojas, Montes and Vieta2014). According to the results of the subgroup analysis on genders, regions and settings, more attention should be paid to settings in eastern areas, male gender and inpatients with SI and SA.
There are several limitations to this study. First, the pooled point prevalence of SA, and the prevalence of SP and CS could not be analysed due to the small number of studies. Publication bias was not assessed for the lifetime and 1-month prevalence of SI as there were <10 studies with relevant data (Wan et al. Reference Wan, Hu, Li, Jiang, Du, Feng, Wong and Li2013). Second, important variables related to suicide behaviours, such as economic conditions, place of residence, medical comorbidities, and the level of family and social support, were not reported in most studies. Third, heterogeneity is difficult to avoid in meta-analysis of epidemiological surveys (Winsper et al. Reference Winsper, Ganapathy, Marwaha, Large, Birchwood and Singh2013; Long et al. Reference Long, Huang, Liang, Liang, Chen, Xie, Jiang and Su2014), which remains a major obstacle in interpreting data. Heterogeneity of results was probably due to the discrepancy in measurements on suicide-related behaviours, sampling and the stage of schizophrenia. In addition, relative small sample size could be another reason for heterogeneity (Yan et al. Reference Yan, Xiang, Hou, Ungvari, Dixon, Chan, Lee, Li, Li and Zhu2013). Future studies with standardised assessments on suicide, larger sample sizes and multi-centre design should be conducted to minimise the heterogeneity. Fourth, irrespective of the lack of pre-study registration, the study was conducted strictly according to the PRISMA and MOOSE recommendations to avoid risk of selective bias and incomplete reporting. Finally, recall bias of suicide-related behaviours may exist in the cross-sectional and cohort studies. Different study periods in cohort studies may result in differing prevalence rates for suicide-related behaviours (Wang et al. Reference Wang, Shi and Yuan2006).
In conclusion, suicide-related behaviours are common in Chinese patients with schizophrenia. Given the significant association between suicide-related behaviours and premature death in schizophrenia, regular screening for such behaviours and effective suicide prevention programmes should be implemented.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S2045796017000476.
Acknowledgements
The research protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO: CRD42017069621).
Financial Support
The study was supported by the University of Macau (SRG2014-00019-FHS; MYRG2015-00230-FHS; MYRG2016-00005-FHS).
Conflict of Interest
The authors have no conflict of interest.
Ethical Standard
Not applicable.
Availability of Data and Materials
All the data used in this manuscript have been included in the tables and figures.
