Early predictors of adverse short-term outcomes following therapeutic hypothermia for hypoxic–ischaemic encephalopathy in a South African tertiary neonatal unit: developing a multivariable model

Introduction Therapeutic hypothermia (TH) can increase disability-free survival after moderate-to-severe hypoxic–ischaemic encephalopathy (HIE). However, up to 50% of cooled infants may still experience adverse outcomes. Early, robust predictors are needed to optimise patient selection. Our primary objective was to determine the association between pre-cooling data and the combined adverse outcome of death before discharge or suppressed amplitude-integrated electroencephalography (aEEG) at 48 h, in babies with moderate-to-severe HIE treated with TH. The secondary objectives were to develop a predictive model and describe morbidity and short-term outcomes. Methods This retrospective cohort study used data from all babies treated with TH for HIE at Groote Schuur Hospital between June 2019 and June 2022. The most significant pre-cooling variables from bivariate analysis per variable grouping were entered into four multivariable logistic regression models with different encephalopathy assessment methods. After dichotomising continuous variables, predictive models with 0, any 1, any 2, or all 3 of the variables were assessed using receiver operating characteristic curve analysis, and the most predictive model was internally validated. Results An adverse outcome occurred in 38 of 78 babies. The following pre-cooling indicators independently predicted adverse outcomes in multivariable logistic regression: pH ≤ 6.8 at ≤ 1 h, age at TH initiation ≥ 5.16 h, and early severe HIE defined as modified Sarnat grade III, Thompson score ≥ 12, or very suppressed aEEG. A stepwise probability model including the three predictors had an area under the curve of 0.88, with the probability of an adverse outcome when any two or three of the predictors were present being 80.4% and 98%, respectively ( p < 0.001). Accuracy and calibration were confirmed with a Brier score of 0.1229 and Spiegelhalter's z-statistic of −0.3040 ( p = 0.6194). A Thompson score ≥ 16 had a 95% probability of increasing the odds of an adverse outcome by 9.32 on post hoc Bayesian analysis. Conclusions A pre-cooling multivariable predictive model and a Thompson encephalopathy score ≥ 16 both independently identified babies at particularly high risk of adverse outcomes despite TH for moderate-to-severe HIE. Future studies should externally validate these predictors.

Authors

Institutions

Publication Details

Journal
Frontiers in Pediatrics
Published
2026-11-18
DOI
https://doi.org/10.3389/fped.2026.1958655
Primary Topic
Neonatal and fetal brain pathology
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Early predictors of adverse short-term outcomes following therapeutic hypothermia for hypoxic–ischaemic encephalopathy in a South African tertiary neonatal unit: developing a multivariable model

Shakti Pillay, Alan R. Horn, Thomas Butler
4 links
Frontiers in Pediatrics
Neonatal and fetal brain pathology
article

Early predictors of adverse short-term outcomes following therapeutic hypothermia for hypoxic–ischaemic encephalopathy in a South African tertiary neonatal unit: developing a multivariable model

Shakti Pillay, Alan R. Horn, Thomas Butler
article en

Abstract

Introduction Therapeutic hypothermia (TH) can increase disability-free survival after moderate-to-severe hypoxic–ischaemic encephalopathy (HIE). However, up to 50% of cooled infants may still experience adverse outcomes. Early, robust predictors are needed to optimise patient selection. Our primary objective was to determine the association between pre-cooling data and the combined adverse outcome of death before discharge or suppressed amplitude-integrated electroencephalography (aEEG) at 48 h, in babies with moderate-to-severe HIE treated with TH. The secondary objectives were to develop a predictive model and describe morbidity and short-term outcomes. Methods This retrospective cohort study used data from all babies treated with TH for HIE at Groote Schuur Hospital between June 2019 and June 2022. The most significant pre-cooling variables from bivariate analysis per variable grouping were entered into four multivariable logistic regression models with different encephalopathy assessment methods. After dichotomising continuous variables, predictive models with 0, any 1, any 2, or all 3 of the variables were assessed using receiver operating characteristic curve analysis, and the most predictive model was internally validated. Results An adverse outcome occurred in 38 of 78 babies. The following pre-cooling indicators independently predicted adverse outcomes in multivariable logistic regression: pH ≤ 6.8 at ≤ 1 h, age at TH initiation ≥ 5.16 h, and early severe HIE defined as modified Sarnat grade III, Thompson score ≥ 12, or very suppressed aEEG. A stepwise probability model including the three predictors had an area under the curve of 0.88, with the probability of an adverse outcome when any two or three of the predictors were present being 80.4% and 98%, respectively ( p < 0.001). Accuracy and calibration were confirmed with a Brier score of 0.1229 and Spiegelhalter's z-statistic of −0.3040 ( p = 0.6194). A Thompson score ≥ 16 had a 95% probability of increasing the odds of an adverse outcome by 9.32 on post hoc Bayesian analysis. Conclusions A pre-cooling multivariable predictive model and a Thompson encephalopathy score ≥ 16 both independently identified babies at particularly high risk of adverse outcomes despite TH for moderate-to-severe HIE. Future studies should externally validate these predictors.

Frontiers in PediatricsVol. 14
University of Cape Town (ZA)
University of Cape Town
Good health and well-being
Neonatal and fetal brain pathology
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon