This is the authors' abstract. We don't add key points for this paper.
Objective To compare bedside ultrasonography (US) and abdominal radiography (AXR) within the same cohort for their ability to discriminate the need for surgical intervention in neonatal necrotising enterocolitis (NEC), and to evaluate the value of combining the two modalities. Methods A total of 509 neonates admitted to our neonatal intensive care unit from January 2015 to December 2023 were included; all had a clinical diagnosis of NEC and underwent abdominal radiography (a supine cot-side view in all infants, with an additional erect view obtained in the radiology department only when the infant was stable enough to be transported) and bedside US within 24 h. They were divided into a surgical group ( n = 88) and a conservative group ( n = 421) by clinical outcome. Eight signs from each of AXR and US were recorded as binary variables. After univariate screening by the χ 2 or Fisher’s exact test, an AXR model, a US model, and a combined model were fitted by multivariable logistic regression. Model discrimination was assessed by the area under the receiver operating characteristic curve (AUC); AUCs were compared by the DeLong test, and their 95% confidence intervals (CIs) were estimated from 1,000 bootstrap resamples. The added value of the combined model was measured by the net reclassification index (NRI) and the integrated discrimination improvement (IDI). Calibration was assessed by the Hosmer–Lemeshow (HL) test, the Brier score, and calibration curves. Firth’s penalised logistic regression was used as a sensitivity analysis for signs with low cell counts. For neonates diagnosed with NEC during the same period but without US, the US signs were imputed by multiple imputation with chained equations ( m = 20) and the models refitted, to assess the potential selection bias of the complete-case analysis. Results The surgical group had a lower gestational age [33.0(30.0, 36.0) week] and birth weight [1,730(1,235, 2,205) g] than the conservative group (35.0[32.0, 36.0] week; 2,200[1,727.5, 2,732.5] g; both P < 0.001). On multivariable analysis, five signs were independent predictors of surgical intervention: AXR portal venous gas [odds ratio (OR) = 32.76, 95% CI 7.11–150.94], AXR pneumoperitoneum (OR = 11.91, 95% CI 1.81–78.58), US complex peritoneal effusion (OR = 16.94, 95% CI 3.65–78.76), US peritoneal effusion (OR = 7.53, 95% CI 4.04–14.03), and US portal venous gas (OR = 2.71, 95% CI 1.33–5.51). The AUCs of the AXR, US, and combined models were 0.696, 0.842, and 0.862. On the DeLong test, both the US and combined models were superior to the AXR model (both P < 0.001), whereas the difference between them was not significant ( P = 0.067). The NRI and IDI of the combined model were 1.074 and 0.285 relative to the AXR model and 0.823 and 0.106 relative to the US model. All three models were well calibrated (HL test P > 0.05; Brier scores 0.116, 0.091, and 0.075). Conclusion For neonates with a clinical diagnosis of NEC, bedside US was superior to AXR in the overall discrimination of the need for surgical intervention. AXR portal venous gas and pneumoperitoneum were complementary to the US findings of complex peritoneal effusion and peritoneal effusion; although adding AXR to US did not raise the AUC over US alone, it still added value in reclassifying borderline cases. US and AXR may serve as complementary modalities in the routine imaging work-up of NEC.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Huang et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: