Patient Outcomes in Teaching Versus Nonteaching General Internal Medicine Services

PURPOSE: Patient care quality appears to be similar when delivered by trainee and attending physicians. The authors conducted a systematic review and meta-analysis to examine whether outcomes differ for general internal medicine (GIM) patients admitted to teaching versus nonteaching services. METHOD: The authors searched Medline, EMBASE, and Cochrane Library databases in May 2012 to identify peer-reviewed, English-language studies with contemporaneous controls comparing inpatient mortality, 30-day readmission rate, and/or length of stay (LOS) for inpatients admitted to teaching or nonteaching GIM services. RESULTS: The 15 included studies (1 randomized controlled trial, 14 observational) included 108,570 patients admitted to U.S. hospitals during 1987-2011. Inpatient mortality did not differ between teaching and nonteaching services (13 studies, 108,015 patients; 2.5% versus 2.8%; OR, 1.07; 95% CI, 0.87-1.32; I = 82%); results were consistent in risk-adjusted studies (adjusted OR, 0.91; 95% CI, 0.76-1.08) and higher-quality studies (OR, 0.94; 95% CI, 0.73-1.21). There were no differences in 30-day readmission rates (11 studies, 106,021 patients; 15.1% versus 13.1%; OR, 1.05; 95% CI, 0.93-1.18). Patients on teaching services appeared to have longer LOS (11 studies, 82,352 patients; unadjusted mean difference, 0.40 days; 95% CI, 0.04-0.77 days), but there was substantial heterogeneity (I = 95%). Differences disappeared in risk-adjusted studies (mean difference: -0.09 days; 95% CI, -0.24 to 0.06 days) and in higher-quality studies (mean difference: -0.05 days; 95% CI, -0.37 to 0.28 days). CONCLUSIONS: There was no convincing evidence that outcomes differed substantively for patients admitted to teaching or nonteaching GIM services.