Published before-and-after controlled trial studies on death education in Chinese or English; reviews, case reports, and studies with only qualitative results were excluded.

Full-time medical students (including nursing, clinical medicine, pharmacy, rehabilitation, etc.); participants with mental illness or major bereavement experience in the past 6 months were excluded.

Participants received explicit death education interventions, including but not limited to: palliative care courses, death experience activities (e.g., simulated death, funeral home visits), and online death education courses.

(a) Studies involving non-medical students (e.g., psychology, sociology majors); (b) Duplicate publications, literatures with incomplete data or inability to extract effect sizes (e.g., only p-values reported without means and standard deviations); (c) Studies that include death education in interventions but cannot separately analyze its effect (e.g., mixed multiple interventions with inability to separate the effect of death education).

Two researchers independently conducted literature screening and data extraction, with discrepancies resolved through consultation with a third researcher. The extracted content included: author, year, country/region, study type, sample size, intervention protocol, course duration, outcome indicators (means and standard deviations of each DAP-R dimension and secondary outcomes), and quality-related information (random sequence generation method, allocation concealment measures, blinding implementation).

The Methodological Index for Non-Randomized Studies (MINORS) scale was used for quality assessment (18). The scale includes 12 items, scored as 0 points (not reported), 1 point (reported but insufficient), or 2 points (reported and sufficient), with a total score of 24. Studies with a score ≥16 were considered high-quality. It must be emphasized that high MINORS scores do not equate to high-level evidence comparable to randomized trials, as the inherent limitations of non-randomized studies (such as potential selection bias) cannot be completely eliminated by this scale.

All 12 studies reported DAP-R scale data. The heterogeneity test showed I² > 50% (83.12% ~ 90.79%), so a random-effects model was used for analysis. A forest plot for the fear of death dimension is shown in Figure 2, and the pooled effect sizes for each dimension are presented in Table 2.

Three studies reported data on attitudes toward palliative care. However, the scoring method of (3) was consistent with the standard FATCOD scale, but the final score exceeded the theoretical maximum score of 150, which may be due to data entry errors or other measurement biases. To maintain consistency in data synthesis, only the pooled effect size of Subgroup A was analyzed (0.39 [0.14, 0.65]). As shown in Subgroup A of Figure 3, death education interventions significantly improved students’ positive attitudes toward palliative care (strong effect), such as proactive participation in communicating with terminally ill patients and emphasis on grief support for family members. Given the very small number of studies contributing to this analysis (n = 2), this finding must be regarded as preliminary and hypothesis-generating.

Intervention methods were categorized into 3 types (offline, online, blended). Table 3 presents the corresponding effect sizes from subgroup analyses grouped by intervention type, and Figure 4 shows the forest plot for the effect of death education interventions on the fear of death dimension. The results (Table 3) showed that blended learning had the optimal effect in improving death attitudes (fear of death: SMD = −0.84, 95%CI [−1.22, −0.47], p < 0.001, medium-large effect; Figure 4), followed by offline learning (fear of death: SMD = −0.20, 95%CI [−0.34, −0.06], p = 0.01, small effect), and online learning had the weakest effect (fear of death: SMD = −0.11, 95%CI [−0.47, 0.25], p = 0.55, no effect). This suggests that courses should integrate online and offline blended learning. These subgroup findings are exploratory due to the small number of studies in some categories (e.g., only 1 online learning study) and should be interpreted as hypothesis-generating rather than definitive evidence.

Courses were divided into two groups: “<16 class hours” and “≥16 class hours.” Table 4 presents the effect sizes from subgroup analyses grouped by course duration. The results showed that interventions with ≥16 class hours had better improvement effects than those with <16 class hours in the dimensions of fear of death (−0.41 vs. −0.21), death avoidance (−0.45 vs. −0.16), and escape acceptance (−0.36 vs. −0.10), indicating that sufficient course duration is crucial for in-depth changes in death attitudes. It should be noted that the “<16 class hours” subgroup only includes 2 studies, so the stability of the conclusion needs further verification.

Studies were divided into two groups: “China” and “Others”. Table 5 presents the effect sizes from subgroup analyses grouped by cultural background. The “China” group demonstrated greater improvement in neutral acceptance compared to other countries (SMD = 0.29 vs. 0.09), which may be associated with cultural factors (3, 14). These exploratory cultural comparisons require validation through intentionally designed cross-cultural studies before firm conclusions can be drawn.

Meta-analysis shows that death education interventions reduces medical students’ fear of death scores, which aligns with the findings of the four-stage “listening-seeing-touching-transcending” death model (25). Notably, after receiving systematic scientific training, medical students are more inclined to view death as a “natural physiological process” (neutral acceptance). For instance, a Turkish study integrated religious and cultural content, yet medical students’ approach acceptance scores did not increase significantly (15). This indicates that scientific and rational cognition dominates medical students’ death attitudes. Constructivist courses break the “passive acceptance” limitation of traditional education through group sharing and role-playing exercises. This helps internalize the perception that “death is a natural law” into a stable attitude. Furthermore, “physical immersion (lying in a simulated coffin) + emotional resonance (reading simulated elegies)” activates students’ deep emotional experiences. This “embodied cognition” mechanism effectively mitigates the fear caused by death taboos, corroborating the conclusion that “spiritual care enhances death acceptance” (32).

Death education interventions improves medical students’ attitudes toward palliative care, which can be explained by the “situated learning theory.” Real scenarios such as hospice ward internships and nursing home services allow medical students to directly observe patient suffering and family grief. This direct observation triggers “emotional resonance” (31). It is important to note that mastering knowledge alone is insufficient to translate into clinical behavior; instead, “case discussions + simulated practice” must be integrated. This suggests that death education interventions should avoid the tendency of “valuing knowledge over application” (33). Combining standardized patient simulation for communication training with clinical internships helps students master the skills of “breaking bad news.” This approach yields significant intervention effects. Additionally, incorporating “death event reflection sessions” during internships enables students to convert classroom knowledge into practical competence (10). This “practice-reflection-repractice” cycle aligns with the survey result that “students most require ‘real-scenario learning’” (34). Moreover, interprofessional collaborative learning is critical for the future work of multidisciplinary palliative care teams. Examples include joint participation of medical and nursing students in death case discussions (35).

Blended learning (online + offline) showed the associations with improved outcomes in our exploratory analysis. Another study reported a 92.4% satisfaction rate for this blended model, significantly higher than single theoretical teaching (68.3%) (27). Structured courses also demonstrate advantages (36): by “identifying knowledge gaps through pre-tests and consolidating effects through post-tests.” These advantages are reflected in two aspects:

Subgroup analysis revealed that interventions lasting ≥16 class hours might be more effective than those with <16 class hours. A promising direction could be to offer death education as a “semester-long course” (e.g., 8 weeks, 2 class hours per week) rather than short-term workshops. The reason is that changing medical students’ emotional attitudes toward death requires a gradual “cognition-emotion-behavior” process. Sufficient time also avoids “information overload” and consolidates effects through “reflection diaries” and “group discussions.”

Differentiated design by target group: surveys show that junior medical students require more “basic death cognition” education, like the physiological process of death. Senior students, in contrast, focus more on “clinical death response” such as family comfort (37). Nursing students show significantly stronger willingness to learn “palliative care skills” than students majoring in religious education. This difference stems from their professional needs (38). Additionally, male students tend to have lower knowledge scores (2) Therefore, more interactive and practical interventions should be designed for them.

Among the included studies, 66.7% of courses were led by nursing/clinical teachers, but only 33.3% of these teachers had received systematic death education training (29, 31). This leads to a tendency of “emphasizing teaching over clinical practice” in courses. One potential strategy to consider would be establishing a “dual-teacher faculty team”: co-taught by “clinical teachers (responsible for care skills)” and “specialized teachers (responsible for emotional counseling).” Meanwhile, an “internship-education linkage mechanism” should be built. This would involve setting up “death event reflection sessions” during clinical internships, where mentors guide students to discuss real death cases and convert classroom knowledge into practical abilities. Promote the “simulation-reality transition model”: first train communication skills through standardized patients, then gradually involve students in real palliative care to reduce psychological impact. Additionally, death education can be included as a reference indicator in medical students’ professional qualification examinations. This would urge institutions to prioritize education quality, ultimately ensuring that “every medical student possesses basic death response and palliative care capabilities.”

Few studies conducted follow-ups exceeding 3 months. None tracked students’ practical behaviors after entering clinical practice (e.g., whether they proactively participate in palliative care) (41). Furthermore, although attempts were made to explore the impact of education on patient care, evaluations relied solely on students’ self-reports. This approach lacked objective assessments from patients or their families. It is recommended to establish a “three-level tracking system (curriculum-internship-work)” and adopt a mixed evaluation method combining “student behavior observation + patient satisfaction survey.” This approach would include: (1) Assess attitudes and knowledge upon course completion; (2) Evaluate “stress responses when facing patient death” 6 months after clinical internships; (3) Measure “palliative care practice behaviors” 1 year after employment. Long-term data will verify the sustainability of death education effects.