The PREM score: a graphical tool for predicting survival in very preterm births

Objective: To develop a tool for predicting survival to term in babies born more than 8 weeks early using only information available at or before birth. Design: 1456 non-malformed very preterm babies of 22-31 weeks' gestation born in 2000-3 in the north of England and 3382 births of 23-31 weeks born in 2000-4 in Trent. Outcome: Survival to term, predicted from information available at birth, and at the onset of labour or delivery. Method: Development of a logistic regression model (the prematurity risk evaluation measure or PREM score) based on gestation, birth weight for gestation and base deficit from umbilical cord blood. Results: Gestation was by far the most powerful predictor of survival to term, and as few as 5 extra days can double the chance of survival. Weight for gestation also had a powerful but non-linear effect on survival, with weight between the median and 85th centile predicting the highest survival. Using this information survival can be predicted almost as accurately before birth as after, although base deficit further improves the prediction. A simple graph is described that shows how the two main variables gestation and weight for gestation interact to predict the chance of survival. Conclusion: The PREM score can be used to predict the chance of survival at or before birth almost as accurately as existing measures influenced by post-delivery condition, to balance risk at entry into a controlled trial and to adjust for differences in "case mix" when assessing the quality of perinatal care.