Behavior characteristics of peer reviewers in medical journals: a survey from China

Peer review is fundamental to quality scientific communication, yet reviewer behavior remain underexplored. The impact of emerging large language models (LLMs) on peer review practices is similarly understudied. We aim to characterize behavioral traits of Chinese medical journal peer reviewers and identify evidence-based recommendations to optimize review willingness, efficiency and quality.

An online questionnaire survey was distributed to 532 medical researchers in China through the Wenjuanxing platform in February 2025. The questionnaire (38 questions) assessed four domains: basic information, peer review model and efficiency, peer review quality, and reviewer motivations. Statistical analysis included descriptive statistics, Spearman correlations, Kruskal-Wallis tests, etc. RESULTS: The response rate was 51.9% (276/532, 95% confidence interval (CI): 47.6%-56.1%). The valid questionnaires were 275: 91.6% male; 64.7% of 41-55 years old. Double-blind was supported by 80.7% of respondents, exceeding international prevalence. Reviewers exhibited social desirability bias in self-reported review turnaround time: 92.7% reported completing reviews within 15 d, whereas the actual recent 3-year administrative data was only 69.7% (P < 0.001, Cramér's V = 0.303). Reviewers expected their submissions to finish review in 15 d and at most 60 d, which was very pressurized for the editorial office. Efficient reviewers expected their manuscripts to be reviewed faster (ρ = 0.551, 95% CI: 0.460-0.630). Reviewers weighted scientificity and novelty most heavily (30% each), followed by clinical feasibility (20%). Review quality showed heterogeneity: 17.8% of respondents (49/275) reported < 50% agreement with feedback received on their submissions vs. 47.6% (131/275) reporting ≥ 70% agreement. Only 24.7% respondents used LLM for peer review assistance, yet 91.2% of users reported a positive impact. The most frequently used LLMs in China were DeepSeek, Doubao, ChatGPT and Kimi in sequence. Compared with males, female reviewers were more likely to use LLM for peer review assistance (43.5% vs. 23.0%, P = 0.029, Cohen's h = 0.439), but the difference was only significant in DeepSeek (P = 0.005). LLM use did not significantly alter main peer review characteristics (all P > 0.05). Regarding motivation, recognition and acknowledgment ranked first (74.9%), followed uniquely in China by requests for priority handling of their submissions (64.4%) and recommended submissions (63.6%), reflecting publication-pressure contexts.

Misalignments exist between reviewer expectations and editorial capacity regarding review efficiency. System-level improvements in manuscript handling system and implementation of standardized review templates and training may improve review quality. Formal recognition of review contributions and fast handling of reviewer's submissions could enhance motivation while addressing unique pressures in Chinese academic journals.