Abstract
BACKGROUND: Health care value is a national priority, and there are substantial efforts to reduce overuse of low-value testing. Residency training programs and teaching hospitals have been implicated in excessive testing. We evaluated the impact of surgery residents on the frequency of inpatient testing and investigated potential inter-resident variation. STUDY DESIGN: Inpatient laboratory and imaging orders placed on general surgery services were extracted from an academic institution from 2014 to 2016 and linked to National Surgical Quality Improvement Program data. Using negative binomial mixed effects regression with unstructured covariance, we evaluated the frequency of testing orders compared with median use, accounting for case, patient, and attending-level variables. RESULTS: There were 111,055 laboratory orders and 7,360 imaging orders linked with 2,357 patients. Multivariable analysis demonstrated multiple significant predictors of increased testing including: postoperative complications, medical comorbidities, length of stay, relative value units, attending surgeon, and resident surgeon (95% CIs > 1, p < 0.05). Compared with the median resident physician, 47 residents (37.9%) placed significantly more laboratory orders, and 2 residents (1.6%) placed significantly more imaging orders (95% CI >1, p < 0.05). Resident identification explained 3.5% of the total variation in laboratory ordering and 4.9% in imaging orders. CONCLUSIONS: Individual surgical residents had a significant association with the frequency of inpatient testing after adjusting for attending, case, and patient-level variables. There was greater resident variation in laboratory testing compared with imaging, yet surgical residents had small contributions to the total variation in both laboratory and imaging testing. Our models provide a means of identifying high users and could be used to educate residents on their ordering patterns.