The impact of advances in medicine on the biometric analysis of infant mortality.

Bourgeois-Pichat's biometric model was developed as a method to decompose infant mortality into endogenous and exogenous components. The model assumes that no endogenous deaths occur after the first month of life. This paper uses data for the United States to examine the hypothesis that recent advances in medicine extend endogenous mortality past the first month of life. The biometric model is found consistently to underestimate the endogenous infant mortality rate and to overestimate the exogenous infant mortality rate relative to cause of death analysis. Direct examination of the age distribution of infant mortality shows that the proportion of all infant mortality that occurs in the first month of life declined from 75.37 per cent in 1970 to 65.40 per cent in 1985, and that a significant and increasing proportion of the mortality occurring after the first month of life is due to endogenous causes. The development of new empirical models is suggested.