Optimal and discriminating birth weights in human populations.

A new model for estimating the optimal birth weight of a sample of infants from a human population, based on conditional survival probability, is developed. Also, a method is developed for estimating the 'lower discriminating birth weight' (LDBW), the weight below which an infant has a greater probability of dying than surviving, and the 'upper discriminating birth weight' (UDBW), the weight above which an infant has a greater probability of dying than surviving. These models are tested using previously published human birth weight data from British, German, Indian, Italian, and United States samples. The relationship between LDBW and 'prematurity' is briefly discussed.