Prediction of peri-operative mortality in care of preterm children in non-cardiac surgery

Abstract

Background: The aim of this study was to develop a risk calculation model for peri-operative 30-day-mortality in preterm infants in non-cardiac surgery.

Methods: Retrospective monocentric follow-up cohort-study of 27,453 pediatric anesthesias at a German university hospital and level one perinatal center between 2008 and 2021 for non-cardiac surgeries. Inclusion criteria were age < 37 post-menstrual weeks at the time of surgery. The primary endpoint was 30-day-mortality after surgery. For statistical analysis, stepwise backwards logistic regressions were performed to identify predictors for 30-day mortality after surgery.

Results: Between 2007 and 2021, 278 preterm infants underwent surgery. The 30-day-mortality was 8.6% (24/278; CI95%:5.6–12.6). A preselection of potential risk factors was based primarily on prior knowledge available from the literature and the results of previously published studies. The final prediction model using a multivariable logistic regression revealed lower post-menstrual age (odds-ratio(OR): 0.67; CI95%: 0.54–0.83) and lower body weight at the time of surgery for extremely preterm infants (OR: 0.024; CI95%: 0.003–0.22), administration of dopamine or norepinephrine or epinephrine (OR: 11.6; CI95%: 3.58–37.7), and life-threatening emergencies between 10pm-7am (OR: 10.1; CI95%: 2.36–43.5) as significant independent risk factors for 30-day-mortality. The Area-Under-The-Receiver-Operating-Characteristic-Curve (0.90; CI95%: 0.85–0.96) showed a good discrimination of the final model. The investigation of the calibration curve (p = 0.99, Spiegelhalter test) and the goodness of fit test (p = 0.85, Hosmer-Lemeshow test) indicated no significant discrepancies between estimated and observed probabilities for the peri-operative 30-day mortality.

Conclusions: Peri-operative 30-day-mortality of preterm infants during non-cardiac surgery is high. The prediction model with easily ascertainable factors as described could be a valuable tool for estimating 30-day-mortality in preterm infants and should be validated in larger populations.

Supplementary Information: The online version contains supplementary material available at 10.1186/s12871-025-03168-x.

Keywords: Child; Congenital; Death; Paediatric; Prematurity.

Fig. 1

Impact of weight at the time of surgery and prematurity differentiated by perterm, very and extremely preterm on 30-day-mortality (body weight at time of surgery was missing in n=3)

Fig. 2

Receiver-Operating-Curve (ROC) (final model) [Area-under under the ROC = 0.90 (CI95%:0.84-0.96)]

Fig. 3

Relationship between risk predictor “RP” and the predicted probability of 30-day-mortality

Fig. 4

Probability for 30-day mortality