Development of a prediction model for mortality in infants undergoing therapeutic hypothermia for neonatal encephalopathy.

OBJECTIVE

To develop and internally validate a model predicting neonatal mortality in infants with neonatal encephalopathy requiring therapeutic hypothermia (TH), using national data.

STUDY

DESIGN

Data from 385 infants treated with TH across 19 hospitals (2016-2021) were analysed. Multivariable logistic regression with backward stepwise selection was applied. Discrimination was assessed using the C-statistic, with internal validation by bootstrapping. The THERM (Therapeutic Hypothermia Early Risk Model for Mortality) tool was developed to calculate individualised mortality risk.

RESULTS

Forty-six infants (11.9%) died within 28 days. Four predictors were retained: prelabour Caesarean section, adrenaline use, base excess ≤-22 mmol/L, and seizures during the first day of life. The model demonstrated excellent discrimination [optimism-adjusted C-statistic 0.885 (95% CI: 0.827-0.936)].

CONCLUSIONS

Four routinely collected variables predicted mortality in infants undergoing TH. The THERM tool provides a practical resource for clinicians, enabling personalised risk assessment and supporting parental counselling during the first day of life.