- 2025
- Epidemiology and psychiatric sciences 34
- Xuefei Tao
- Zhihui Zhang
- Li Liang
- Shen Xu
- Xiayu Du
- Zhihong Ren
- Xianglian Yu
Study Design
- Type
- Meta-Analysis
- Population
- global populations from observational studies reporting suicidal ideation prevalence
- Methods
- Systematic search in Web of Science, PubMed, PsycINFO and ProQuest up to August 2025; included cross-sectional and longitudinal observational studies
Aims
Although extensive research has been conducted on the impact of the COVID-19 pandemic on global mental health, a systematic synthesis of the cross-time dynamics of suicidal ideation (SI) remains lacking. This study aims to systematically synthesise the global aggregated prevalence of SI before and after the pandemic, investigate the potential association between pandemic exposure and the SI risk through meta-regression analysis of longitudinal studies, and explore key moderating factors.Methods
A systematic search was conducted in Web of Science, PubMed, PsycINFO and ProQuest databases up to August 2025. Observational studies were included if they employed cross-sectional or longitudinal designs and reported the prevalence of SI before and after the pandemic across global regions.Results
The analysis included 354 cross-sectional studies (N = 8,247,875) and 27 longitudinal studies. In cross-sectional studies, the pooled prevalence of SI was 13.20% [95% CI 12.06%-14.42%]. Pre-pandemic prevalence was 12.52% [95% CI 8.46%-18.14%], and post-pandemic prevalence was 13.24% [95% CI 12.07%-14.50%], with no significant difference. Meta-regression analysis identified three moderators. Specifically, larger sample sizes (n) were associated with lower prevalence (β = -0.232, P < 0.0001); higher study quality predicted lower prevalence (β = -0.278, P < 0.001); and studies on adults reported significantly lower prevalence than adolescents (β = -0.366, P < 0.05). Conversely, time progression during the pandemic, development level, geographical area, gender and measurement method did not show significant independent effects. Interaction analyses also found no significant moderating effect of economic development level or geographical area on the temporal trend of SI prevalence. Longitudinal analysis found no significant increase in prevalence from the pre-pandemic to the post-pandemic period (P = 0.101). However, a small but significant increase occurred between early and late stages within the pandemic (β = 0.265, P = 0.021). Subgroup analyses showed no significant moderation of these temporal changes.Conclusions
The COVID-19 pandemic's impact on SI was dynamic. While no significant prevalence change was found between pre- and post-pandemic periods, a significant increase occurred as the crisis progressed. This deteriorating trend was more pronounced in adolescents, identifying them as a key vulnerable group. Methodologically, findings were moderated by the measurement instrument, study quality and sample size, with evidence suggesting potential small-study effects. These findings underscore the need for robust mental health surveillance and targeted interventions for at-risk populations during prolonged public health crises.The protocol was registered on PROSPERO (CRD42024603151).