Frequency of duodenal anatomical variants in neonatal and pediatric upper gastrointestinal tract series (UGI) and the influence of exam quality on diagnostic reporting of these.

Aim: To determine frequency of duodenal anatomical variants on clinically indicated pediatric UGI examinations and determine the influence of these variants and exam quality on the reliability of diagnosis.
Materials and Methods: Two-pediatric radiologists retrospectively reviewed 100-UGI exams performed on children ≤18-years. Exams were considered diagnostic if the duodenojejunal (DJ) flexure was identified. For diagnostic exams, readers categorized the duodenal location and shape as: normal, normal variant, or abnormal. Exam quality was assessed according to duodenal visualization, number of boluses required, and patient positioning.
Results: Reader 1: 90/100 exams diagnostic -77% normal duodenum, 20% normal variant, and 3% abnormal. Reader 2: 97/100 exams diagnostic - 88% normal, 8% normal variant, and 4% abnormal. Original reports: 99/100 exams diagnostic - 92% normal, 3% normal variant, and 5% abnormal. 42% of exams were "high-quality" and 58% were "low-quality". The number of abnormal was the same between readers in "high-quality" studies. In "low-quality" studies reader 1 and the original read diagnosed 1 further case as non-rotation which was diagnosed as a normal variant by reader 2. Two further cases were reported as non-rotation by the original reader. Inter-rater reliability was significantly higher among each pair of raters in "high quality" exams (κ 0.3; p ≤ 0.05) compared to "low quality" exams (k < 0.1 - p > 0.05).
Conclusion: Duodenal variants were diagnosed in 8-20% of UGI. Compromised exam quality contributes to poor interrater reliability and may result in diagnostic errors of normal variant duodenums, posing a risk for unnecessary intervention and/or delayed treatment.
(Copyright © 2022 Elsevier Inc. All rights reserved.)