Contact With Mental Health Services Prior to Suicide: A Systematic Review and Meta-Analysis
Abstract
Objective:
Access to mental health care is regarded as a central suicide prevention strategy. This is the first systematic review and meta-analysis of the prevalence of contact with mental health services preceding suicide.
Methods:
A systematic search for articles reporting prevalence of contact with mental health services before suicide was conducted in MEDLINE and PsycINFO, restricted to studies published from January 1, 2000, to January 12, 2017. A random-effects meta-analysis with double arcsine transformations was conducted, with meta-regression used to explore heterogeneity.
Results:
Thirty-five studies were included in the systematic review, and 20 were included in the meta-analysis. Among suicide decedents in the population, 3.7% (95% confidence interval [CI]=2.6%−4.8%) were inpatients at the time of death. In the year before death, 18.3% (CI=14.6%−22.4%) of suicide decedents had contact with inpatient mental health services, 26.1% (CI=16.5%−37.0%) had contact with outpatient mental health services, and 25.7% (CI=22.7%−28.9%) had contact with inpatient or outpatient mental health services. Meta-regression showed that women had significantly higher levels of contact compared with men and that the prevalence of contact with inpatient or outpatient services increased according to the sample year.
Conclusions:
Contact with services prior to suicide was found to be common and contact with inpatient or outpatient mental health services before suicide seems to be increasing. However, the reviewed studies were mainly conducted in Western European and North American countries, and most studies focused on psychiatric hospitalization, which resulted in limited data on contact with outpatient services. Better monitoring and data on suicides that occur during and after treatment seem warranted.
Despite long-standing efforts to decrease suicide rates, they remain quite stable internationally (1) and have increased considerably in the United States (2,3). The general lack of success in reducing suicide rates internationally is clearly related to the fact that suicide is a complex and multifaceted phenomenon with a low base rate at the individual level (4). Although causes and risk factors are far from completely understood, psychiatric disorders are an essential risk factor for suicidal behavior, and more than 90% of persons who die by suicide have a psychiatric disorder (5). One meta-analysis of data from psychiatric inpatients found a suicide rate of 147 per 100,000 inpatient years, with higher rates in more contemporary samples (6). Another recent meta-analysis found that the suicide rate among discharged psychiatric patients was 484 per 100,000 person-years—or 44 times the global suicide rate in 2012 (7).
These studies illustrate the well-known elevated risk of suicide among psychiatric patients, which has had a significant impact on the recommended approaches to suicide prevention. Several agencies, including the World Health Organization (WHO) (8), the Office of the U.S. Surgeon General, the Institute of Medicine (9), and the Department of Health in England (10), have highlighted increased access to health care and improved quality of care as essential strategies to prevent suicide. To explore the potential population effect of increased contact with mental health services as a suicide prevention strategy, accurate estimates of the current prevalence of contact with mental health services preceding suicide are needed.
We located only two reviews on the prevalence of contact with mental health care prior to suicide, and both are dated. One concluded that contact with inpatient psychiatric care in the year before suicide may have been as high as 41% (11). However, this estimate was based on a single study, and the authors found much lower rates (4%−11%) of contact with community-based psychiatric services in the year before suicide. The other review found an average contact rate of 32% in the year before death, but it did not weight the included studies (12). These authors also found variation between age groups and genders in rates of contact. A limitation of the two reviews was the lack of a valid quantitative synthesis of the evidence. Furthermore, not only does mental health care vary greatly between countries, it has also undergone dramatic changes over the past several decades, including downsizing of inpatient facilities and expansion of outpatient and community services (13). As a result, much of the old literature regarding contact with services before suicide now seems obsolete.
Our aim was to conduct a systematic review—and to our knowledge the first meta-analysis—of the prevalence and time of contact with mental health services prior to suicide, restricted to studies published after January 1, 2000. We predicted that differences in contact rates would be observed between treatment settings, genders, sample years, age groups, and regions.
Methods
Protocol and Registration
The review protocol was based on PRISMA-P guidelines (14) and was registered at PROSPERO (CRD42017057797). Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) (15) and Meta-analyses Of Observational Studies in Epidemiology (MOOSE) (16) quality standards were used.
Search and Information Sources
We searched the databases MEDLINE and PsycINFO through Ovid. The primary search term was “suicide” as a keyword, combined with “mental health services,” “inpatients,” “outpatients,” “hospitalization,” or “patient discharge” in both keywords and truncated free text. The search was limited to articles published from January 1, 2000, to January 12, 2017, when the search was conducted. We included peer-reviewed articles that had been published electronically (including preprints) or in print. In MEDLINE, the “review” filter was applied in order to identify and remove review articles from the search. In PsycINFO, the “all journals” filter was applied to exclude books and book chapters. Reference lists of included articles were screened.
Eligibility Criteria
Studies were included that reported completed suicide (including deaths of undetermined cause or open verdicts) as an outcome combined with a frequency, proportion, or rate of contact with mental health services within one, three, and six months; one year; or lifetime prior to the suicide. Title and abstract had to be available in English. We included only studies that had samples drawn from the general population and excluded review articles, articles that reported on samples selected according to specific subgroups, and clinical follow-up studies.
Study Selection
The search was deduplicated twice. References were handled in EndNote×7 throughout the review. One reviewer (MØM) screened titles and abstracts (N=3,742) for eligibility, which resulted in 315 records. Inclusion criteria were pilot-tested and refined until a Cohen’s kappa of .68 was reached. After the criteria were revised, records were screened a second time by two independent reviewers (MØM and ATK). When disagreement occurred, records were kept for full-text screening. A total of 181 articles were retrieved in full text and assessed for eligibility by two independent reviewers (MØM and ATK). Disagreements were resolved through discussion. When the reviewers were uncertain about whether to include a study or if nonresolvable disagreements occurred, a third reviewer (FAW) made the decision.
When several records reported from the same population and time period, the record with the largest sample was kept. This criterion resulted in the exclusion of 20 studies. Records reporting on identical cohorts were collapsed into one record. One record (17) reporting on two different time periods was divided into two different studies. We used consultants to interpret two papers—one in German and one in Italian. One study that fulfilled the inclusion criteria was discovered during preparation of the manuscript and included in the review (18).
Data Items and Data Extraction
Two independent reviewers (MØM and ATK) extracted data from published reports with a form and coding instructions that had been pilot-tested. Disagreement was resolved through discussion. If no agreement was reached, the authors of the article were contacted. Data were extracted on contact rates or proportions, study design, country, time frame of included suicides, and subgroups (gender and type of services). When the outcome was reported as a proportion, two reviewers independently converted the proportion into number of cases. Measures of contact were grouped by setting: inpatient, defined as a psychiatric hospitalization; outpatient, comprising all psychiatric treatment not based on a hospitalization; and inpatient or outpatient services, a broader category that includes both the mentioned categories. This broader category was used when studies reported the prevalence of contact with mental health services without differentiating between inpatient or outpatient treatment or without specifying the level of care this treatment setting included.
Risk of Bias
Risk of bias was assessed by using an eight-item form based on previously published quality criteria for systematic reviews of studies of prevalence (19,20) and adapted for this review. The items assessed whether the study had a representative national sample, whether it was a true or close representation of the sampling frame, whether random selection or census was used, whether the study had a reliable and valid data collection method, whether there was an acceptable definition of suicide that included established death codes, and whether an appropriate fraction to estimate prevalence was used. The last item was an overall assessment of the study’s risk of bias.
Bias was assessed by two independent reviewers (MØM and ATK). Items were rated as low or high, and the overall rating was rated as low, moderate, or high risk of bias. When information was missing, the item was rated as high risk of bias, thus overestimating the risk of bias. Disagreement was resolved through discussion. Pilot-testing was used to train the two independent reviewers.
Data Analysis and Synthesis
To reduce heterogeneity, studies of specific age groups were included only in the qualitative review and were not included in the meta-analysis. The meta-analysis was conducted for the following time periods: current and one year. Data were analyzed with the metafor package in R, version 1.9–9 (21). The metaprop function was used to provide the pooled estimates, forest plots, and meta-regression. The double arcsine transformation was used to stabilize confidence intervals (22). A random-effects model stratified by type of setting was used to conduct the meta-analysis, because of a priori assumed heterogeneity. The meta-analysis used inverse variance weighting. Heterogeneity was measured with I2 statistics, where I2 above 75% was considered to be high (23). Lines indicating overall prevalence were not printed in the forest plots because of high heterogeneity.
Heterogeneity was explored by meta-regression. Gender was prespecified as a trial-level covariate. In addition, the midyear of the individual studies’ sampling period was calculated to assess the recency of the various samples. Analyses with regions and age groups as covariates were initially planned, but low power and unequal distribution of data did not allow the analyses to be conducted. Publication bias was assessed through visual inspection of funnel plots.
Results
Identification and Description of Studies
The inclusion criteria were met by 59 studies. [A PRISMA flow diagram of the systematic search is included in an online supplement to this article.] Twenty studies were excluded because of overlap, and four were merged with another study (Figures 1 and 2). As a result, 35 studies were included in the qualitative review (17,18,24–60). Five authors were contacted for additional information.
FIGURE 1. Prevalence of contact with inpatient services at time of death among suicide decedents in the general populationa
aLines beside or within the shaded areas indicate 95% confidence intervals. The solid diamond indicates the overall measure. Heterogeneity: I2=98.6%, τ2=.0018, p<.001
FIGURE 2. Prevalence of contact with mental health services in the year before death among suicide decedents in the general population, by type of servicea
aLines beside or within the shaded areas indicate 95% confidence intervals. Solid diamonds indicate overall measures.
bHeterogeneity: I2=99.5%, τ2=.008, p<.001
cHeterogeneity: I2=99.3%, τ2=.0174, p<.001
dHeterogeneity: I2=98.7%, τ2=.0021, p<.001
The included studies were from 19 primarily Western European and Northern American countries [see table in online supplement]. The studies used linkage of national registers (N=9), psychological autopsies (N=10), national clinical surveys (N=4), and record reviews (N=13) and included samples from 1980 to 2015 [see supplement]. The median midyear of the samples was 2000. Eleven studies reported the main outcome across genders in all settings, and seven of these reported on contact with inpatient treatment. Eight studies included open verdicts or sudden unexplained deaths in their definition of suicide (35,40,41,43,50,55,56,59).
Of the studies that included all age groups, 18 reported on contact with inpatient services, seven on contact with outpatient treatment, and 13 on contact with inpatient or outpatient mental health services prior to suicide [see table in supplement]. The time points used to report the proportion of contact varied between studies, with most studies reporting on contact within one year. Eleven studies reported on specific age groups: six on contact with mental health services by persons age 25 and under who died by suicide and five on contact by persons over age 50. There were large variations between these studies in the time points and settings used to report the outcome variable. There was a slight tendency of lower contact rates in the studies reporting on suicides of persons under age 25. No clear trend could be observed in the studies of persons over age 50.
Meta-Analysis
Twenty individual studies reporting on 21 samples were included in the meta-analysis. Of these, nine studies reported prevalence of suicides among persons who were in contact with inpatient services at the time of death (Figure 1), giving a pooled prevalence of 3.7% (95% confidence interval [CI]=2.6%−4.8%). Eighteen studies reported on suicides among persons in contact with mental health services within the year prior to death, stratified by setting (Figure 2). The pooled prevalence for contact with inpatient services was 18.3% (CI=14.6%−22.4%), for outpatient services it was 26.1% (CI=16.5%−37.0%), and for inpatient or outpatient mental health services it was 25.7% (95% CI=22.7%−28.9%).
Heterogeneity was large and significant in all strata. As noted above, 11 studies reported the prevalence of contact with services by gender. In the seven studies that reported contact rates for inpatient treatment within one year, gender was a significant covariate (p≤.001, τ2=.003), with contact being more common for women (35.0%, CI=30.1%−40.1%) than for men (19.7%, CI=17.3%−22.1%).
The midyear of the individual studies’ sampling period was a significant moderator between studies of contact with inpatient or outpatient mental health services within one year of death (p=.003, τ2=.002), with higher contact estimates in the most recent samples. The results of the meta-regression were not significant for current inpatients or those who had inpatient contact within one year.
Post hoc meta-regressions were conducted for the studies in which deaths of undetermined cause were included in the definition of suicide as a trial-level covariate. The meta-regression was significant for current inpatients (p=.041, τ2=.001) but not significant for contact with inpatient services within one year or for contact with inpatient or outpatient services within one year. We further examined the borderline significant finding by conducting a subgroup analysis; the results were not significant for all strata and showed that studies that included open verdicts and sudden unexplained deaths in the definition of suicide had a lower contact prevalence than studies that reported only on suicides.
Funnel plots of the studies included in the meta-analysis showed no systematic skewness or asymmetry for contact with mental health services overall or with inpatient services within one year, indicating no publication bias (data not shown).
Risk of Bias
An interobserver agreement of κ=.609 was reached for the assessment of risk of bias. Of the 35 included studies, 29 had a representative sampling frame, 12 had a representative national sample and 19 studies used census or random sampling. The method for data collection was assessed as low risk of bias in 23 studies. Sixteen studies had an acceptable definition of suicide based on established death codes. Twenty-four of the 35 studies reported numerators and denominators that were directly and clearly defined. Overall risk of bias was rated as high in nine, moderate in 14, and low in 12 studies [see table in supplement].
Discussion
This systematic review and meta-analysis of mental health service contact prior to suicide found that within the prior year, 18.3% of persons who died by suicide had contact with inpatient mental health services, 26.1% had contact with outpatient mental health services, and 25.7% had contact with inpatient or outpatient mental health services. These rates are remarkably lower than those in the previous review by Pirkis and Burgess (11), who reported that up to 41% of persons who died by suicide were in contact with inpatient services in the year before death. The rates are also lower than those in the review by Luoma and colleagues (12), who reported that approximately 32% of persons who died by suicide were in contact with mental health services in the year before death. The lower estimates found in this review are probably due to the use of formal meta-analytic approaches with weighting, which are less influenced by small studies with outlying estimates, as well as by the inclusion of several recent large registry studies. In addition, the overall number of suicides included in this review is much larger than in the previous reviews.
As expected, a great degree of heterogeneity was found in all subgroups. Meta-regression showed that contact with mental health services was significantly more common among females than among males and that the recency of the sample was a significant trial-level covariate for increased contact with inpatient or outpatient mental health services but not for inpatient services exclusively.
The overall prevalence of persons admitted to a psychiatric hospital at the time of their death by suicide death was 3.7%. Despite large heterogeneity, the confidence interval for the overall estimate was narrow. The overall sample was large and represented by several extensive, national clinical surveys or registry studies in Western European countries. All the studies except those by Qin and colleagues (52–54) lay within the confidence interval of the overall effect. The slightly higher prevalence in the studies by Qin and colleagues could be explained by the broad time span of the sample, ranging from 1981 to 1997.
The overall prevalence of contact with inpatient or outpatient mental health services in the year preceding suicide was 25.7%. Even though heterogeneity was high, the confidence interval was quite narrow, and most studies lay within the boundaries of the confidence interval. The exceptions were the study by Bakken and colleagues (25), who found a higher prevalence, and the study by Kung and colleagues (44), who found a lower prevalence. These two studies probably represent two opposite types of health care organization—the comprehensive, publicly financed health care system in Norway and the private insurance–based systems in the United States.
Only five of the included studies reported exclusively on the prevalence of contact with outpatient services in the year before death. An overall estimate of 26.1% of suicide decedents were found to have been in contact with outpatient services in the prior year, which is similar to the estimate for the combined inpatient and outpatient mental health services category. This surprising result is due to great variation in the contact prevalence between the few studies reporting on outpatients. It is also likely that organization of and referral to outpatient services might be subject to greater variation between countries, compared with inpatient services. The sparse data on contact with outpatient services was pointed out 20 years ago by Pirkis and Burgess (11) and is even more striking now given the past decades’ increased focus on outpatient treatment and community services worldwide (13).
As expected, we found significant differences in contact rates between genders; admission to a psychiatric hospital in the year before death was more common among females than among males. This finding implies that use of such services is less prevalent among men (61,62). Because of the lack of data, it was not possible to analyze gender differences regarding use of outpatient services or mental health services in general (inpatient or outpatient); this is an area requiring future research.
Sample year was a significant covariable for contact with inpatient or outpatient mental health services in the year before death but not for contact with inpatient services only. This finding indicates that contact with the wider range of mental health services became more common during the study period, and the differences between treatment settings could be a result of the general increased focus on outpatient services. As noted above, there might also be greater variation in the way outpatient services are organized in various countries, compared with inpatient services, which should also be taken into account. However, the magnitude of increase in the prevalence of contacts with mental health services during the study period was modest, which limits the current potential population effect of increased contact with mental health services as a suicide prevention strategy.
Post hoc analysis of the relation between suicide definitions and contact indicated a lower prevalence of contact in studies that included deaths of undetermined cause in their definition. Contrary to expectations, one of these meta-regressions was borderline significant, but the findings were not replicated in subgroup analyses. Meta-regressions are sensitive to false positives, and one should be especially careful when conducting post hoc analyses (63). When all results were taken together, we found no evidence of higher levels of contact in studies that included deaths of undetermined cause in their suicide definition, and we believe that this lends support to the practice of including undetermined deaths in countries where the use of such verdicts is common in suicide research.
It was not possible to conduct a meta-analysis of studies on specific age groups, because these studies reported the outcome variable in different settings and time points. With one exception (49), studies of suicide decedents under age 25 generally found lower rates of contact with mental health services, compared with the general population. Studies of suicide decedents over age 50 mainly reported on service contact during the individuals’ lifetime and consequently found a higher prevalence of contacts than in the younger and general population.
The risk of bias in individual studies was generally rated as being moderate or low. Of the studies included in the meta-analysis, only one small study was rated as having a high risk of bias (39), but this had little weight in the meta-analysis. Our review included several national registry studies, which is likely to have improved the overall quality of our findings, because the samples in these studies were larger and more representative of the general population and because record linkage is effective for identification of cases in this line of research.
The high degree of heterogeneity and the wide distribution of data across settings and time-periods restricted our opportunities to conduct planned subgroup analyses and meta-regressions as specified in the protocol. We did, however, find gender to be a significant predictor of contact with services, and we also found somewhat lower rates of contact among suicide decedents under age 25, as discussed above. Nevertheless, such individual characteristics accounted only for a limited amount of the observed heterogeneity, as expected. Furthermore, although there is some variation in suicide rates between countries and regions (1), it is difficult to see how such differences could explain the large variation in contact with services among suicide decedents. The large variation is thus more likely to stem from differences on a system level, particularly regarding how mental health care is organized in various countries, and issues related to the availability of services in various countries or health care systems.
Previous studies have found large differences between high- and low-income countries in the degree of contact with mental health services among persons with anxiety, mood, and substance use disorders (64). In one study of European countries, the rate of contact with mental health services for mental health reasons varied between 9.7% in Italy and 29.9% in Netherlands—both well-organized, high-income countries (65). Differences in these general contact rates with mental health care might thus be attributable to determining factors related to national political priorities, such as the amount of resources allocated to health care (64), availability of mental health services (65), and other factors related to the dimensionality of mental health services in various geographical areas or nations.
Factors related to the organization of the mental health care system might also be of great importance. For instance, patients with a particular degree of depression might receive treatment in primary care in some regions and in specialized care in other areas. This difference would in turn directly influence contact with services before suicide. Several studies have demonstrated that the ways in which primary and specialized care interact might also be important in accounting for the variability in contact with various services (65,66). Both these issues, along with the issue of capacity, may explain our finding that the variation between single studies was most pronounced for outpatient services, compared with both inpatient services and combined inpatient or outpatient services. Finally, even in Norway, where contact rates with mental health services were found to be above 40% in the year before suicide, a recent study found that most people with mental health conditions do not receive any treatment (67). The pathways to specialized care are thus somewhat coincidental and characterized by large variations, and thus they are difficult to disentangle on the basis of findings from a systematic review.
In general, existing studies lacked clinical information on contact with mental health care before suicide, which precluded any attempt to include such covariates in the meta-analysis. On the basis of our systematic review and meta-analysis, we recommend that in addition to a better formal surveillance of suicide among patients, future studies systematically report on three time points—current, three months, and one year—when possible, provide comprehensive descriptions of the settings included, differentiate between inpatient and outpatient treatment, and report the prevalence of contact for each gender.
The results of our review might inform suicide prevention interventions and research in several ways. First, although prevalence rates of contact were found to vary between countries and health care systems, a substantial proportion of suicide decedents were in contact with mental health care services in every study included. However, even in Norway, the country with the highest contact rates, not more than four in ten suicide decedents had been in contact. This finding means that if the WHO strategy (8) of reducing the number of suicides by improving access to care is to be more effective, actions must be taken on both a health administrative level and a political level to enable services to reach more individuals. Second, this finding might also imply that other parts of the health care system, in particular primary care services, should be actively involved in recognition and treatment of people with mental disorders.
In addition to the general implications for suicide prevention highlighted above, the findings of this review might also have implications for health care systems individually. A high rate of contact with services is not necessarily evidence of efficient suicide prevention, because it can also indicate poor service quality. In the same way, low rates of contact can result from effective identification of people at risk and effective treatment. Nevertheless, the results of this review can serve as a basis for comparing various service systems or nations. In the development of suicide prevention strategies, policy makers and health authorities should evaluate the specific health care system in order to decide where to best place the effort. In systems with a low prevalence of contact before suicide, increasing access to care would probably be the most appropriate place to start. In systems with a high level of contact, it might be more effective to assess and improve the performance of the system, including the quality of care. Finally, the proportions of persons in contact with services identified in this review once again point to the dramatically increased risk of suicide both during and after contact with mental health services. Although there is some evidence of the effect of suicide prevention interventions at the system level (68,69), our findings clearly highlight the urgent need for development of better measures of suicide prevention both during an episode of inpatient or outpatient treatment and in the first year following treatment for this high-risk group.
Conclusions
This systematic review and meta-analysis provides the first pooled prevalence rates of contact with mental health services before suicide. Although our overall estimates are lower than reported in two previous reviews, contact with mental health services is still common prior to suicide. We found substantial variation in the prevalence of contact between various samples and settings. Given the large number of suicide decedents in the general population who had been in contact with mental health services in the year before death, the lack of data from many countries and regions is striking, particularly regarding outpatient services. In addition, information about diagnoses and other clinical variables was absent in most of the studies reviewed, which is unfortunate considering how important this type of knowledge can be in the development of suicide prevention strategies. This review, together with other recent meta-analyses that also emphasize the enormous increased risk of suicide among mental health patients (6,7), points to the importance of improved monitoring of suicides in relation to health services. It further shows the need for interventions aimed at reducing barriers to care for suicidal individuals and the need for development and evaluation of more suicide-specific interventions for the high-risk population already receiving care.
1 Suicide Rates, Crude: Data by Country. Geneva, World Health Organization, 2017Google Scholar
2 Curtin SC, Warner M, Hedegaard H: Increase in Suicide in the United States, 1999–2014. NCHS Data Brief 241. Atlanta, Centers for Disease Control and Prevention National Center for Health Statistics, 2016Google Scholar
3 : Suicide risk after psychiatric hospital discharge. JAMA Psychiatry 74:669–670, 2017Crossref, Medline, Google Scholar
4 : Suicide and suicidal behaviour. Lancet 387:1227–1239, 2016Crossref, Medline, Google Scholar
5 : Psychological autopsy studies of suicide: a systematic review. Psychological Medicine 33:395–405, 2003Crossref, Medline, Google Scholar
6 : Meta-analysis of suicide rates among psychiatric in-patients. Acta Psychiatrica Scandinavica 131:174–184, 2015Crossref, Medline, Google Scholar
7 : Suicide rates after discharge from psychiatric facilities: a systematic review and meta-analysis. JAMA Psychiatry 74:694–702, 2017Crossref, Medline, Google Scholar
8 Preventing Suicide: A Global Imperative. Geneva, World Health Organization, 2014Google Scholar
9 Institute of Medicine: Reducing Suicide: A National Imperative. Washington, DC, National Academy of Sciences, 2002Google Scholar
10 Preventing Suicide in England: A Cross-Government Outcomes Strategy to Save Lives. London, Department of Health, Mental Health and Disability Division, 2012Google Scholar
11 : Suicide and recency of health care contacts: a systematic review. British Journal of Psychiatry 173:462–474, 1998Crossref, Medline, Google Scholar
12 : Contact with mental health and primary care providers before suicide: a review of the evidence. American Journal of Psychiatry 159:909–916, 2002Link, Google Scholar
13 : Components of a modern mental health service: a pragmatic balance of community and hospital care: overview of systematic evidence. British Journal of Psychiatry 185:283–290, 2004Crossref, Medline, Google Scholar
14 : Preferred Reporting Items for Systematic reviews and Meta-Analysis Protocols (PRISMA-P) 2015 statement. Systematic Reviews 4:1, 2015Crossref, Medline, Google Scholar
15 : Preferred Reporting Items for Systematic reviews and Meta-Analyses: the PRISMA statement. PLoS Medicine 6:e1000097, 2009Crossref, Medline, Google Scholar
16 : Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 283:2008–2012, 2000Crossref, Medline, Google Scholar
17 : Reductions in postdischarge suicide after deinstitutionalization and decentralization: a nationwide register study in Finland. Psychiatric Services 58:221–226, 2007Link, Google Scholar
18 : Population-based analysis of health care contacts among suicide decedents: identifying opportunities for more targeted suicide prevention strategies. World Psychiatry 15:135–145, 2016Crossref, Medline, Google Scholar
19 : Assessing risk of bias in prevalence studies: modification of an existing tool and evidence of interrater agreement. Journal of Clinical Epidemiology 65:934–939, 2012Crossref, Medline, Google Scholar
20 : Development quality criteria to evaluate nontherapeutic studies of incidence, prevalence, or risk factors of chronic diseases: pilot study of new checklists. Journal of Clinical Epidemiology 64:637–657, 2011Crossref, Medline, Google Scholar
21 : Conducting meta-analyses in R with the metafor package. Journal of Statistical Software 36:1–48, 2010Crossref, Google Scholar
22 : Meta-analysis of prevalence. Journal of Epidemiology and Community Health 67:974–978, 2013Crossref, Medline, Google Scholar
23 : Measuring inconsistency in meta-analyses. BMJ 327:557–560, 2003Crossref, Medline, Google Scholar
24 : Health care contacts in the year before suicide death. Journal of General Internal Medicine 29:870–877, 2014Crossref, Medline, Google Scholar
25 : Comparison of data from the Cause of Death Registry and the Norwegian Patient Register. Tidsskrift for Den Norske Lægeforening 135:1949–1953, 2015Crossref, Medline, Google Scholar
26 : Characteristics of suicide completers with a psychiatric diagnosis before death: a postmortem study of 98 cases. Psychiatry Research 220:556–563, 2014Crossref, Medline, Google Scholar
27 : Suicides and serious suicide attempts: two populations or one? Psychological Medicine 31:837–845, 2001Crossref, Medline, Google Scholar
28 : Suicidal behavior in the North Varese province: an epidemiological analysis [in Italian]. Epidemiologia e Psichiatria Sociale 10:180–185, 2001Crossref, Medline, Google Scholar
29 : Elderly suicide in Hong Kong: a case-controlled psychological autopsy study. Acta Psychiatrica Scandinavica 109:299–305, 2004Crossref, Medline, Google Scholar
30 : Patterns of health care usage in the year before suicide: a population-based case-control study. Mayo Clinic Proceedings 90:1475–1481, 2015Crossref, Medline, Google Scholar
31 : Suicide in a health maintenance organization population. JAMA Psychiatry 72:294–296, 2015Crossref, Medline, Google Scholar
32 : Health status and suicide in the second half of life. International Journal of Geriatric Psychiatry 25:371–379, 2010Crossref, Medline, Google Scholar
33 : A comparison of suicides in psychiatric in-patients, after discharge and in not recently hospitalized individuals. Comprehensive Psychiatry 69:100–105, 2016Crossref, Medline, Google Scholar
34 : Psychiatric hospitalizations during the last 12 months before suicide. General Hospital Psychiatry 29:63–65, 2007Crossref, Medline, Google Scholar
35 : Deaths by suicide and their relationship with general and psychiatric hospital discharge: 30-year record linkage study. British Journal of Psychiatry 204:267–273, 2014Crossref, Medline, Google Scholar
36 : Adolescent suicide in Quebec and prior utilization of medical services. Canadian Journal of Public Health 95:357–360, 2004Crossref, Medline, Google Scholar
37 : Gender differences in youth suicide and healthcare service use. Crisis 37:290–298, 2016Crossref, Medline, Google Scholar
38 : Psychiatric care of suicides in Hong Kong. Journal of Affective Disorders 76:137–142, 2003Crossref, Medline, Google Scholar
39 : Suicide after discharge from psychiatric hospital [in German]. Psychiatrische Praxis 29:22–24, 2002Crossref, Medline, Google Scholar
40 : Suicide in young people aged 15–24: a psychological autopsy study. Journal of Affective Disorders 63:159–170, 2001Crossref, Medline, Google Scholar
41 : Safety of patients under the care of crisis resolution home treatment services in England: a retrospective analysis of suicide trends from 2003 to 2011. Lancet. Psychiatry 1:135–141, 2014Crossref, Medline, Google Scholar
42 : Suicide within 12 months of mental health service contact in different age and diagnostic groups: national clinical survey. British Journal of Psychiatry 188:135–142, 2006Crossref, Medline, Google Scholar
43 : 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 5:111–118, 2001Crossref, Medline, Google Scholar
44 : Substance use, firearm availability, depressive symptoms, and mental health service utilization among white and African-American suicide decedents aged 15 to 64 years. Annals of Epidemiology 15:614–621, 2005Crossref, Medline, Google Scholar
45 : Risk factors for male and female suicide decedents ages 15–64 in the United States: results from the 1993 National Mortality Followback Survey. Social Psychiatry and Psychiatric Epidemiology 38:419–426, 2003Crossref, Medline, Google Scholar
46 : Suicide and attempted suicide among older adults in Western Australia. Psychological Medicine 30:813–821, 2000Crossref, Medline, Google Scholar
47 : Contact of mental and nonmental health care providers prior to suicide in Taiwan: a population-based study. Canadian Journal of Psychiatry 53:377–383, 2008Crossref, Medline, Google Scholar
48 : Systematic services audit of consecutive suicides in New Brunswick: the case for coordinating specialist mental health and addiction services. Canadian Journal of Psychiatry 53:671–678, 2008Crossref, Medline, Google Scholar
49 : Suicide in young Singaporeans aged 10–24 years between 2000 to 2004. Archives of Suicide Research 16:174–182, 2012Crossref, Medline, Google Scholar
50 : Suicide in mental health in-patients and within 3 months of discharge. National clinical survey. British Journal of Psychiatry 188:129–134, 2006Crossref, Medline, Google Scholar
51 : Late onset suicide: distinction between “young-old” vs “old-old” suicide victims. How different populations are they? Archives of Gerontology and Geriatrics 54:136–139, 2012Crossref, Medline, Google Scholar
52 : Trends in suicide risk associated with hospitalized psychiatric illness: a case-control study based on Danish longitudinal registers. Journal of Clinical Psychiatry 67:1936–1941, 2006Crossref, Medline, Google Scholar
53 : Suicide risk in relation to psychiatric hospitalization: evidence based on longitudinal registers. Archives of General Psychiatry 62:427–432, 2005Crossref, Medline, Google Scholar
54 : Suicide risk in relation to socioeconomic, demographic, psychiatric, and familial factors: a national register-based study of all suicides in Denmark, 1981–1997. American Journal of Psychiatry 160:765–772, 2003Link, Google Scholar
55 : Suicide and hospitalization for mental disorders in Sweden: a population-based case-control study. Journal of Psychiatric Research 44:741–747, 2010Crossref, Medline, Google Scholar
56 : Suicide in children and young people in England: a consecutive case series. Lancet. Psychiatry 3:751–759, 2016Crossref, Medline, Google Scholar
57 : Contacts with mental health services before suicide: a comparison of indigenous with non-indigenous Australians. General Hospital Psychiatry 34:185–191, 2012Crossref, Medline, Google Scholar
58 : An exploration of human services system contacts prior to suicide in South Carolina: an expansion of the South Carolina Violent Death Reporting System. Injury Prevention 12(suppl 2):ii17–ii21, 2006Crossref, Medline, Google Scholar
59 : Suicide in juveniles and adolescents in the United Kingdom. Journal of Child Psychology and Psychiatry, and Allied Disciplines 49:1155–1165, 2008Crossref, Medline, Google Scholar
60 : Understanding the unique characteristics of suicide in China: national psychological autopsy study. Biomedical and Environmental Sciences 18:379–389, 2005Medline, Google Scholar
61 : Barriers to help-seeking by men: a review of sociocultural and clinical literature with particular reference to depression. Journal of Affective Disorders 71:1–9, 2002Crossref, Medline, Google Scholar
62 : Help-seeking behaviour in men and women with common mental health problems: cross-sectional study. British Journal of Psychiatry 186:297–301, 2005Crossref, Medline, Google Scholar
63 : How should meta-regression analyses be undertaken and interpreted? Statistics in Medicine 21:1559–1573, 2002Crossref, Medline, Google Scholar
64 : Use of mental health services for anxiety, mood, and substance disorders in 17 countries in the WHO world mental health surveys. Lancet 370:841–850, 2007Crossref, Medline, Google Scholar
65 : Differences in lifetime use of services for mental health problems in six European countries. Psychiatric Services 58:213–220, 2007Link, Google Scholar
66 : Use of general practitioners versus mental health professionals in six European countries: the decisive role of the organization of mental health-care systems. Social Psychiatry and Psychiatric Epidemiology 48:137–149, 2013Crossref, Medline, Google Scholar
67 : Diagnostic and genetic overlap of three common mental disorders in structured interviews and health registries. Acta Psychiatrica Scandinavica 137:54–64, 2018Crossref, Medline, Google Scholar
68 : Mental health service changes, organisational factors, and patient suicide in England in 1997–2012: a before-and-after study. Lancet. Psychiatry 3:526–534, 2016Crossref, Medline, Google Scholar
69 : Implementation of mental health service recommendations in England and Wales and suicide rates, 1997–2006: a cross-sectional and before-and-after observational study. Lancet 379:1005–1012, 2012Crossref, Medline, Google Scholar
