Refining the biophysical profile with a risk-related evaluation of test performance

Objective: Our objective was to determine the ability of biophysical profile variables to predict bad perinatal outcome in high-risk third-trimester pregnancy. Study Design: The outcomes of 1146 fetuses were correlated with abnormal single or multiple variables occurring in biophysical profile done within 72 hours of delivery. Theoretic risks of poor outcomes for different combinations of abnormal variables, calculated with Shortliffe's formula, were compared with the actual risks observed (χ² tests). Results: Two hundred forty-six fetuses had at least one abnormal biophysical profile variable with the risk of bad outcome, for a single abnormal variable, ranging from 8% (body movements) to 100% (tone) and increasing from 14% (any variable abnormal) to 63% (all variables abnormal). In most (57%) observed combinations of biophysical profile variables, significant differences between theoretic and actual risks of bad outcomes were found. By stepwise logistic regression the best predictive model contained all variables except fetal movement. Conclusions: Dynamic biophysical profile variables appear to be interdependent. Not all combinations of abnormal variables occur and specific combinations improve prediction of poor outcome. Risk-related scales for biophysical profile outcomes might prove superior to more conventional scoring systems.