Purpose or Case Report: After-hours upper gastrointestinal series (UGI) are performed by residents with indirect supervision. This study evaluates resident performance after hours with indirect supervision versus normal business hours with direct supervision and identifies areas of improvement.
Methods & Materials: From 2/4/14-9/8/25, 1560 UGI were performed on children 2 years old or less. Studies were categorized as within (n=1340) or outside (n=220) normal business hours. Imaging impression was compared to surgical outcome. False positive and negative cases were reviewed for educational insights. Microsoft Copilot was used to automate data processing and statistical analysis. Copilot classified outcomes, calculated sensitivity, specificity, accuracy, PPV, and NPV, and performed significance testing. It also produced visualizations and trend analyses to identify performance variations over time.
Results: Sensitivity, specificity, and accuracy, with indirect (77.3%, 99.0%, 96.8%) versus direct (93.2%, 99.7%, 99.5%) supervision were not significantly different except for accuracy which was better during the day (p=0.0005). PPV/NPV were 89.5%/97.5% after hours and 91.1%/99.8% during the day. Although not statistically significant, the sensitivity was lower after hours, and when evaluated over time this mostly occurred in the first two years. Six false positive cases (2 after-hours) showed a low duodenojejunal junction (DJJ) interpreted as abnormal rotation. Five of eight false negatives were after hours. Missed volvulus and obstruction cases included interpretation errors (overlooking duodenal dilation and right sided jejunum), technical error (dilute iohexol limiting density), and 2 cases with no imaging abnormality despite appropriate technique suggesting intermittent volvulus.
Conclusions: Radiology residents can safely and effectively perform emergent upper GI studies after hours with indirect supervision. Sensitivity and specificity were not statistically significantly different than studies performed during normal business hours with direct supervision, though accuracy was better during the day. Reinforcing a 5–10-minute delayed image to locate jejunum may reduce false positive and false negative studies. Requesting a 2-3 hour delayed radiograph to assess cecal location if there is a low DJJ may reduce false positive studies. Avoiding dilute contrast and reinforcing assessment of duodenal caliber may reduce false negative studies.