Association between community medicine rotation duration and general medicine in-training examination scores among second-year resident physicians in Japan: a nationwide cross-sectional study

Background Although community medicine is a required component of postgraduate clinical training, its actual duration and relation to knowledge-based performance remain unclear. This study examined the national distribution of community medicine rotation duration and its association with General Medicine In-Training Examination (GM-ITE) performance. Methods We conducted a nationwide multicenter cross-sectional study using GM-ITE 2024 data from second-year resident physicians. Outcomes were compared between those with 1 month and those with 2–3 months of community medicine training. The primary outcome was the GM-ITE total score rate (%), and the prespecified secondary outcome was the correct rate (%) in the general medicine category. Other domain- and specialty-specific correct rates were examined descriptively. Multivariable linear regression was used to estimate the associations between community medicine rotation duration and total score rate. Results The study included 2,834 second-year residents. Community medicine rotation duration was 1 month for 2,243 residents (79.1%), 2 months for 435 (15.3%), 3 months for 67 (2.4%), 4 months or more for 21 (0.7%), and missing for 68 (2.4%). Total score rate was similar between the one-month and 2–3-month groups (58.36±10.11 vs 58.50±10.23; p=0.785; Cohen’s d=0.01). The general medicine correct rate was nominally higher in the 2–3-month group than in the one-month group (54.43±18.31 vs 52.49±17.37; p =0.031; Cohen’s d=0.11), but this difference did not remain statistically significant after false discovery rate adjustment (q=0.122). In the multivariable analysis restricted to residents with 1–3 months of community medicine rotation and complete covariate data, a two-month rotation was not significantly associated with total score rate (adjusted β: 0.44; 95% CI: −0.54 to 1.42; p=0.382), whereas a small adjusted association was observed for a three-month rotation (adjusted β: 2.33; 95% CI: 0.29 to 4.37; p =0.025). Conclusions One month was the predominant duration of community medicine training in Japan. A small, adjusted association with the total GM-ITE score rate was observed for a three-month rotation but not for a two-month rotation, and it should be interpreted with caution. The general medicine finding was small, exploratory, and not statistically significant after adjustment for false discovery rate.