Background
Suicide, a leading cause of death among men who have sex with men (MSM), has become a public health issue [
1]. In Canada, more MSM are estimated to have died from suicide than from HIV-related illness in 2011 [
1], and in China, a rapid increase of suicide-related behaviors among MSM has been frequently reported [
2‐
4]. Suicidal ideation, defined as thinking about, considering or planning for suicide [
5], may lead to higher risk of suicide [
6,
7]. In a review by David Klonsky et al., it has been suggested that an important reason for the limited achievement in reducing suicide is inadequate knowledge, especially about why and when suicidal ideation progresses to potentially suicidal attempts [
6]. Additionally, the risk for suicide is significantly higher in those with suicidal ideation than those without suicidal ideation [
7]. Furthermore, compared with heterosexual males, MSM are at higher risk for suicidal ideation [
8,
9], which may be associated with the discrimination and stigma brought by their sexual orientation [
8,
10,
11]. For example, in a worldwide population-based study, Mathy found that MSM in South America, North America and Asia were at 7.5, 2.1 and 2.9 times higher risk for suicidal ideation than heterosexual males, respectively [
12]. Based on these findings, the assessment of suicidal ideation is imperative among MSM.
Numerous studies have estimated the lifetime prevalence of suicidal ideation among MSM. However, the reported prevalence varied greatly across studies, ranging from 10.6 to 55.3% [
13,
14], indicating the importance of a reliable estimation of the pooled lifetime prevalence of suicidal ideation among MSM, which could help the service providers to identify the accurate amount of those with suicidal ideation and hence implement appropriate interventions to reduce suicide. Literature search showed that there was one published meta-analysis which estimated the pooled lifetime prevalence of suicidal ideation among MSM [
15]. However, the meta-analysis included exclusively studies with a concurrent heterosexual comparison group, and only 2 studies were included when estimating the pooled lifetime prevalence, which significantly lowered the generalizability of their findings. Furthermore, the meta-analysis was conducted in 2008 and, subsequently, there was a rapid increase of publications about lifetime suicidal ideation among MSM, suggesting that an update on the pooled lifetime prevalence of suicidal ideation among MSM was necessary. Therefore, this study aimed to update the estimation of the pooled lifetime prevalence of suicidal ideation among MSM.
In addition, previous studies have indicated that suicide-related behaviors among sexual minorities differed with different sample sources and HIV statuses [
11,
16]. For example, in a study by Hottes et al., it was found that the pooled lifetime prevalence of suicidal attempt in lesbian, gay and bisexual (LGB) individuals for population-based and community-based surveys was 11 and 20%, respectively [
16], and HIV-positive MSM were at higher risk of suicidal ideation than HIV-negative MSM [
11]. Also, the prevalence of suicidal ideation among MSM may differ by geographic area, which could be explained by the differences in socio-demographic factors and economical levels, as well as the cross-cultural differences in the attitude towards sexual minorities [
12]. For example, it has been estimated that the lifetime prevalence of suicidal ideation among MSM in North America and South America was 22.6 and 34.6%, respectively [
12]. Therefore, the pooled lifetime prevalence stratified by geographic area, sample source and HIV status was also explored in this study.
Discussion
This meta-analysis provided an update on the pooled lifetime prevalence of suicidal ideation among MSM. Nineteen eligible studies with a total of 26,667 participants were involved, of which 9374 were identified with suicidal ideation. The reported lifetime prevalence of suicidal ideation ranged from 13.18% [
10] to 55.80% [
34] among the eligible studies, and the pooled lifetime prevalence of suicidal ideation among MSM by a random effects model was 34.97% (95% CI: 28.35%–41.90%).
The pooled lifetime prevalence of suicidal ideation among MSM found in this study (34.97%) was not only much higher than that among general population found in previous meta-analyses (3.9–11.5%) [
35‐
38], but also higher than the lifetime prevalence of suicidal ideation among lesbians found in some previous studies. For example, in a general population-based sample from the Netherlands, de Graaf et al. found that the lifetime prevalence of suicidal ideation among lesbians was 23.3% [
8], and in a Canadian population-based study by Steele et al., this rate was 29.5% [
39]. The high pooled lifetime prevalence of suicidal ideation among MSM found in this study, which may be associated with the discrimination, stigma, prejudice and isolation brought by their sexual orientation [
8,
10,
11], significantly underscores the importance of early assessment of suicidal ideation among MSM, as well as the need for strengthening the psychological interventions.
Subgroup analysis found that the lifetime prevalence of suicidal ideation in high-income countries was higher than that found in low- and middle- income countries. This tendency was contradictory to a study by Mathy, which showed that the lifetime prevalence of suicidal ideation among MSM was 22.6% in North America and 34.6% in South America [
12]. The high heterogeneity observed when estimating the pooled lifetime prevalence stratified by geographic area (
I
2 =98.8% for studies involved in high-income countries and
I
2 =98.3% for studies involved in low- and middle- income countries) may lead to the contradictory results. More cross-cultural studies are warranted to better clarify the difference in the lifetime prevalence of suicidal ideation observed in different geographic areas.
Subgroup analyses also indicated that the lifetime prevalence of suicidal ideation differed significantly between different sample sources. Specifically, MSM from community-based surveys exhibited higher lifetime prevalence of suicidal ideation, in comparison with those from population-based surveys. This finding is consistent with a previous meta-analysis [
16]. By pooling 30 cross-sectional studies conducted in North America and Western Europe, Hottes et al. found that the pooled lifetime prevalence of suicidal attempt was 11 and 20% among LGB individuals from population-based and community-based surveys, respectively [
16]. Venue-based LGB community surveys may induce selection bias and tended to overrepresent gay or lesbian-identified, urban, and high-income sexual minorities [
40], thus contributing to a higher lifetime prevalence of suicide-related behaviors.
Additionally, this meta-analysis found that the lifetime prevalence of suicidal ideation among HIV-positive MSM was much higher than that among HIV-negative MSM, indicating more attention should be paid to HIV-positive MSM. Suicidal ideation may arise as the infected MSM learned of their HIV status [
41]. This was mainly due to the fact that compared with HIV-negative MSM, HIV-positive MSM may experience higher levels of stigma associated with increased risk for suicidal ideation [
11].
It is noteworthy that none of the eligible studies were identified with high quality according to the AHRQ scale. This was specifically because most eligible studies did not indicate how confounding was assessed and did not summarize the completeness of data collection. Furthermore, none of the eligible studies indicated if evaluators of subjective components of study were masked to other aspects of the status of the participants, and explained how missing data were handled in the analysis. A low level of quality of reporting did not necessarily imply a low level of quality of methodology [
42]. Additionally, levels of quality of observational studies may differ by different types of assessment scales [
43]. Quality is an amorphous concept and it is important to distinguish between the quality of reporting and the quality of methodology, which reflects the susceptibility to bias and assesses the validity [
43]. Nonetheless, reporting detailed information on the methods may help identify the appropriateness of methodology. Thus, it is imperative for the researchers to report their findings of observational studies in accordance with the appropriate guidelines, such as the STROBE checklist.
Certain limitations should be acknowledged in this meta-analysis. First, recall bias might exist when collecting information on the lifetime prevalence of suicidal ideation, as a consequence of which, the misclassification of suicidal ideation among the included studies remains a possibility. Second, the heterogeneity in the whole sample and most subgroups was high. The eligible studies included in this meta-analysis came from a varied range of geo-political background with diverse social and environmental factors, which may lead to the high heterogeneity. Additionally, though numerous studies have indicated that psychological stress (i.e. discrimination, stigma and stressful life events) and mental health problems (i.e. depression and anxiety) may affect the lifetime prevalence of suicidal ideation among sexual minorities [
44‐
46], subgroup analyses stratified by these factors were unable to conduct since quite few eligible studies have reported related information among MSM. Future research should, therefore, explore more factors associated with the lifetime prevalence of suicidal ideation among MSM, especially factors related to psychological stress and adverse mental health outcomes. Besides, though some studies have indicated that age may be associated with suicide-related behaviors [
47,
48], subgroup analysis stratified by age was unable to perform due to the inconsistent cutoff points.
Despite the preceding limitations, this study has quite a few strengths. First, this study provided an update on the pooled lifetime prevalence of suicidal ideation among MSM. Nineteen eligible studies conducted globally were involved, which significantly increased the generalizability of our findings. Second, this meta-analysis provided the first quantitatively pooled lifetime prevalence of suicidal ideation among MSM stratified by the subgroups: geographic area, sample source and HIV status. The different prevalence found within each subgroup would be helpful in identifying the factors associated with suicidal ideation. Finally, low sensitivity and low possibility of publication bias significantly increased the reliability of the findings of this study.