Development and validation of a prediction model for complications in patients with esophageal foreign bodies: A retrospective cohort study

Esophageal foreign body (EFB) impaction is a common emergency, but not all patients develop serious complications. An easy-to-use risk prediction tool is lacking. We retrospectively analyzed 270 hospitalized patients with EFB. Multivariable logistic regression was used to identify independent risk factors for complications. Lasso regression with 10-fold cross-validation was performed for variable selection. Internal validation was conducted using bootstrap resampling. Model performance was assessed by ROC analysis, calibration plot, and decision curve analysis (DCA). An optimal cutoff was derived using the Youden index for risk stratification. Complications occurred in 67 patients (24.8%). Independent predictors included sharp foreign body (adjusted OR = 2.85, 95% CI: 1.13–8.15, P = 0.035), atherosclerosis (adjusted OR = 2.57, 95% CI: 1.17–5.74, P = 0.019), and CRP (per 10 mg/L: adjusted OR = 1.15, 95% CI: 1.05–1.28, P = 0.007). The ROC curve showed an AUC of 0.722 (95% CI: 0.642–0.802, P < 0.001). Bootstrap-corrected AUC was 0.698, and the calibration slope was 0.94, indicating good internal validity. DCA demonstrated positive net benefit across threshold probabilities of 0.10–0.60. Using a predicted probability cutoff of 0.25, the high-risk group had a significantly higher complication rate than the low-risk group (47.8% vs. 17.2%, P < 0.001). Sharp foreign body, atherosclerosis, and elevated CRP independently predict complications in patients with EFB. The proposed prediction model shows moderate to good discriminative ability and can help clinicians perform rapid risk stratification. However, external validation is required before clinical implementation.