Optimized search for parameters useful in the interpretation of fetal heart rate data

A comprehensive computerized perinatal data management system was used to acquire and process FHR data and to link this information to a set of control and outcome variable for 1000 deliveries. Multivariate correlation analyses and stepwise linear regression techniques were used to study the interrelationships between risk factors of pregnancy and labor, monitored data, obstetric interventions, and outcome. It was found that while some FHR characteristics, such as moderate-to-severe bradycardia, lack of beat-to-beat variability, and severe variable decelerations were significantly associated with poor outcome, others, such as accelerations, early decelerations, mild bradycardia, and mild-to-moderate tachycardia, were positively correlated with good outcome. Several statistical models for outcome prediction were developed using FHR data alone and also in conjunction with small subsets of control variables.