Pediatric intensivists (PIs) are often responsible for immediate interpretations of radiographs in the absence of board-certified radiologists—after hours or early in the morning, for example—but those physicians could be acting on interpretations a pediatric radiologist (PR) might not agree with, a team of New York researchers reported in the Journal of the American College of Radiology this month.

At Children’s Hospital of Montefiore Medical Center in the Bronx, New York, lead author Adam Z. Fink, MD, and colleagues rely on emergency room general radiologists and on-call radiology residents for quick interpretation of radiographs in both the pediatric intensive care unit (PICU) and neonatal intensive care unit (NICU). These clinicians typically review portable chest radiographs (CXRs) and abdominal radiographs (AXRs) overnight before a certified radiologist can make an official call in the morning.

“Thus, pediatric and neonatal intensivists frequently review these radiographs independently, and may act on them before an official interpretation is made by a radiologist,” Fink et al. wrote.

Two previous trials have suggested interpretation discordance between intensivists and experts, but little has been reported about agreement rates between PIs and PRs, the authors said. In the past, studies have reported an up to 90 percent concordance rate between pediatric emergency medicine physicians and pediatric radiologists.

Fink and his team designed a multi-institutional prospective study, which spanned two PICUs and three NICUs and included three PRs and PIs each. Physicians examined a total of 960 CXRs and AXRs and recorded interpretations via an online form.

The total discrepancy rate in the trial was relatively high, the authors reported, but most interpretations were deemed nonactionable—while the total discrepancy rate reached 34.7 percent, just 7.9 percent were considered actionable. The most common actionable discrepancies were linked to line and tube positions, and identification and interpretation of parenchymal opacities in the lungs. Following suite were identification of air leaks in the PICU, which accounted for 11 percent of all discrepancies, and differentiation of normal from abnormal bowel gas patterns in the NICU.

“Although the total discrepancy rate in CXR and AXR interpretation between PRs and PICU and NICU PIs was high, most discrepancies were not actionable,” Fink and co-authors said. “Common nonactionable discrepancies that may serve as areas for discussion and focused teaching with our PI colleagues include retrocardiac atelectasis in PICU patients and the differentiation of radiographic patterns of neonatal lung disease in NICU patients.”

The authors said addressing these issues, using simulation-based education efforts to assess real-time PI interpretations and looking into virtual radiology rounds could all help lower discrepancy rates.