Prediction of Low Birthweight and Prematurity by a Multiple Regression Analysis with Maternal Characteristics Known since the Beginning of the Pregnancy

Abstract

Between 1963 and 1969, the ‘Institut National de la Santé et de la Recherche Médicale’ carried out a prospective survey comprising 20,000 pregnancies, in 12 obstetric departments of public hospitals in Paris. A preliminary analysis has been done on 9,500 cases. For these cases, the authors have studied some maternal characteristics known since the beginning of the pregnancy, and even before, and have examined how low birthweight and prematurity were distributed according to these variables. Separately, these characteristics are not predictive enough to isolate a group of high risk women. With a function calculated by a multiple regression with the whole set of the characteristics, it is possible to improve the prediction of the considered risks, and, at the same time, to point out the most predictive factors of these risks, factors which are different for low birthweight and for prematurity. The multiple binary regression enables us to classify each woman in a high risk group or in a low risk group for low birth-weight, or prematurity according to the values of risk functions Υ_A and Υ_B for this woman, nevertheless with many errors of prediction; but this is only a first prediction, possible ‘before’ the beginning of the pregnancy, and has to be corrected and improved during successive examinations, by taking into account risks appearing during the course of the pregnancy.