Development and internal validation of a prediction model to predict survival for congenital diaphragmatic hernia in the early postnatal period.

PURPOSE

To develop an easily applied predictive model to predict survival rate for infants with congenital diaphragmatic hernia (CDH) in the early postnatal period according to the Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD) guideline.

METHODS

The retrospective study was conducted including 225 neonates with prenatal or postnatal diagnosed CDH between 2001 and 2018. Patients did not receive the therapy of fetal endoscopic tracheal occlusion and extracorporeal membrane oxygenation. The study took into consideration these variables that are easily available in most centers within the first 1 h after admission. A multivariable prediction model to predict the survival rate for CDH was generated and its performance was analyzed.

RESULTS

The multiple logistic regression analysis was generated using five clinical variables that are routinely available in most centers, including birth weight, 1-min Apgar score, side of hernia, presence of liver herniation, and PaCO in the admission arterial blood analysis. The area under the receiver operating characteristic curve value for this model was 0.912, which was greater than that of a single biomarker in predicting the survival rate of CDH. This model had a sensitivity of 90.6% and a specificity of 74.6%. This model demonstrated good calibration (Hosmer-Lemeshow goodness-of-fit test,  = .410). Besides, the model had a better discriminative ability compared to the previously established predictive models of CDH.

CONCLUSIONS

The simple and generalizable model was developed by five predictors for CDH in the early period using the TRIPOD checklist. It demonstrated good performance in predicting the survival rate of infants with CDH, holding promise for future clinical application.