Parameters of mortality in human populations with widely varying life spans

Abstract

A three‐component, competing‐risk mortality model, developed for animal survival data, fits human life table data for all ages over a range of mean life spans from 16 to 74 years. The competing risks are a novel exponentially‐decreasing hazard, dominant during immaturity; a constant hazard, dominant during adulthood; and an exponentially increasing Gompertzian hazard, dominant during senescence. By fitting the model to a specific life table using non‐linear techniques, estimates of the five model parameters and their standard errors obtain; the proportion of deaths expected from each hazard alone may then be calculated. Preliminary analysis of 13 life tables indicates that for human populations under heavy stress, with very short mean life spans of about 20 years, the three hazard components account for roughly equal numbers of deaths; for modern populations, with mean life spans of about 75 years, nearly all deaths are due to the hazard of senescence. Factor analysis of the correlation matrix of parameter values for the 13 populations shows a two‐factor structure. One factor involves only the multiplicative constants (initial values) of the three hazards, but not the hazard rates of change; the second factor involves only the parameters of the immaturity hazard and the rate of acceleration of the senescence hazard, but not the constant hazard nor the multiplicative constant (initial value) of the senescence hazard.