Evaluating Resident On-Call Performance: Does Volume Affect Discrepancy Rate?

Abstract

To examinehow study volume affects discrepancy rates for on-call radiology residents. Inparticular, we studied how both total shift volume and volume at a particularpoint in time might effect performance. Weretrospectively analyzed 518 weekend call shifts at our institution. The totalnumber of computed tomography (CT) studies per shift was recorded. For everyabdomen-pelvis (AP) or chest-abdomen-pelvis (CAP) CT, preliminary and finalreports were compared for possible discrepancy and rated (by effect on short-termmanagement). We also developed "peristudy volume," defined as CTs read within ±30minutes of a given CT, an estimate of how busy a resident might be at a giventime. We performed logistic regressions to determine whether overall shiftvolume or peristudy volume were predictors of discrepancies. CTvolume/day increased from 58.1 ± 10.1 in 2011 to 75.3 ± 12.5 in 2015(p<0.001). 4695 AP (or CAP) CTs were reviewed, with 145 discrepancies thatcould affect short-term management (3.1%). When reading a study during a shift with≥51 total CTs,residents had increased odds of an error compared to reading a study during ashift with ≤30 studies (OR: 2.97, CI: 1.19-6.46) (p=0.01). When reading a CTwith a peristudy volume of ≥6, residents had increased odds of an error comparedto reading a study with ≤5 peristudy CTs (OR: 1.6, CI: 1.1-2.3) (p=0.01). Whenon-call residents interpret AP CT during high volume shiftsor during busy time-points, odds of discrepancies increase. Awareness of thesedata may inform residency programs in staffing decisions.