• Cynthia Kay, Jeffrey L Jackson, and Michael Frank.
• Dr. Kay is instructor/fellow, Department of Medicine, Medical College of Wisconsin, Milwaukee, Wisconsin. Dr. Jackson is professor of medicine, Medical College of Wisconsin, and chief, General Internal Medicine Section, Zablocki VA Medical Center, Milwaukee, Wisconsin. Dr. Frank is professor of medicine and program director, Internal Medicine Residency Program, Medical College of Wisconsin, Milwaukee, Wisconsin.
• Acad Med. 2015 Jan 1; 90 (1): 100-4.

PurposeTo explore the relationship between United States Medical Licensing Examination (USMLE) Step 1 scores, yearly in-service training exam (ITE) scores, and passing the American Board of Internal Medicine certifying examination (ABIM-CE).MethodThe authors conducted a retrospective database review of internal medicine residents from the Medical College of Wisconsin from 2004 through 2012. Residents' USMLE Step 1, ITE, and ABIM-CE scores were extracted. Pearson rho, chi-square, and logistic regression were used to determine whether relationships existed between the scores and if Step 1 and ITE scores correlate with passing the ABIM-CE.ResultsThere were 241 residents, who participated in 728 annual ITEs. There were Step 1 scores for 195 (81%) residents and ABIM-CE scores for 183 (76%). Step 1 and ABIM-CE scores had a modest correlation (rho: 0.59), as did ITE and ABIM-CE scores (rho: 0.48-0.67). Failing Step 1 or being in the bottom ITE quartile during any year of testing markedly increased likelihood of failing the boards (Step 1: relative risk [RR]: 2.4; 95% CI: 1.0-5.9; first-year residents' RR: 1.3; 95% CI: 1.0-1.6; second-year residents' RR: 1.3; 95% CI: 1.1-1.5; third-year residents' RR: 1.3; 95% CI: 1.1-1.5).ConclusionsUSMLE Step 1 and ITE scores have a modest correlation with board scores. Failing Step 1 or scoring in the bottom quartile of the ITE increased the risk of failing the boards. What effective intervention, if any, program directors may use with at-risk residents is a question deserving further research.

18 keywords...

hide…