Effect of death education interventions on death attitudes in medical students: systematic review and meta-analysis

Background The quality of palliative care depends on medical practitioners’ death cognition and care ability, however, medical students worldwide generally face problems such as death anxiety and insufficient palliative communication skills. Existing studies have not yet formed systematic conclusions on the consistency of the effects of death education interventions and the optimal intervention model. Methods Following PRISMA, 5 databases (PubMed, Web of Science, Cochrane Library, CNKI, Wanfang Data) were searched (2010.1–2025.10), including 12 non-randomized intervention studies (1,111 students). Quality was assessed using the Methodological Index for Non-Randomized Studies Scale (MINORS). Stata 18.0 conducted meta-analysis, with subgroup (intervention method, duration, culture), sensitivity, and publication bias (funnel plot, Egger’s test) analyses. Results Meta-analysis showed that death education interventions (based on non-randomized intervention studies evidence) had potential positive effects: it reduced fear of death (SMD = −0.38, 95%CI [−0.59, −0.17]), death avoidance (SMD = −0.40, 95%CI [−0.64, −0.17]), and escape acceptance (SMD = −0.32, 95%CI [−0.58, −0.06]); meanwhile, it enhanced neutral acceptance (SMD = 0.43, 95%CI [0.15, 0.72]) and approach acceptance (SMD = 0.29, 95%CI [0.01, 0.58]). Additionally, preliminary results showed that death education interventions improved attitudes toward palliative care. Exploratory subgroup analysis revealed that: blended learning had the promising effect; interventions with a duration of ≥16 class hours showed better improvement; Chinese students exhibited significant improvement in neutral acceptance, while others performed better in approach acceptance. It should be noted that several subgroup analyses include small numbers of studies and are explicitly labeled as exploratory. Sensitivity analysis indicated that the pooled effect size had a small fluctuation range. The funnel plot and Egger’s test suggested no significant publication bias. Pooled analyses of all DAP-R dimensions show very high heterogeneity (I 2 > 80%), which may affect the reliability and generalizability of the pooled effect sizes. Conclusion Caution is warranted in causal interpretation given the non-randomized nature of included studies. Death education interventions shows potential improving medical students’ death attitudes and enhancing their palliative care literacy. Blended learning and ≥16 class hours might be more effective. It is suggested that medical colleges integrate it into core curricula, optimize design per culture, and build long-term follow-up systems to achieve “humanity-technology integration.”