Specialty-associated differences in intensive care unit rotation exit examination performance: a single-center retrospective cross-sectional study.

BACKGROUND

Intensive care unit (ICU) rotations bring together residents from multiple specialties. Comparisons of exit performance may be distorted when departments differ in postgraduate-year (PGY) composition and when non-equivalent examination forms are used across PGY levels.

METHODS

We conducted a retrospective cross-sectional analysis of ICU exit examination records from 412 residents at a tertiary teaching hospital in China between November 2025 and April 2026. Three 100-item, 60-min examinations containing four question types were administered according to PGY. Departmental differences were evaluated using four complementary strategies: a same-examination subgroup of 227 PGY-2 residents, within-examination standardization, analysis of covariance adjusting for the combined PGY/form grouping, and hierarchical regression. Item characteristics, version-specific reliability, and exploratory differential item functioning were assessed retrospectively.

RESULTS

In the same-examination subgroup, the total-score effect size was η 2 = 0.098 (p < 0.001), with surgery having the lowest mean score (75.9; SD 9.1; n = 77; 95% CI 73.9-78.0). Department explained an additional 7.6% of total-score variance after adjustment for PGY/form. The descriptive attenuation ratio was 21.3% for total scores, ranging from 7.3% for case-series items to 80.2% for basic-knowledge items. These ratios do not identify separate training-year or examination-version effects. The largest negative case-analysis gaps occurred in surgery and anesthesiology.

CONCLUSION

Specialty-associated differences in examination performance were observed within this single-center educational setting and using this assessment instrument. The absence of baseline scores and incomplete evidence of cross-form comparability preclude causal or inherent-specialty interpretations; prospective baseline-controlled studies are needed before specialty-tailored interventions are adopted.