A METHOD OF ANALYSIS AND INTERPRETATION OF CONCURRENT CAUSE-SPECIFIC DEATH RATES

All individuals during their lifetimes are exposed, to mortality from different causes, which are competing risks. Several statistical models have attempted to deal with this problem, but most start with the easier hypothesis that the risks are independent of each other. Such an hypothesis is harder to satisfy in the case of chronic pathological processes, which are strongly correlated with predisposing factors. In such cases, it preferable to introduce individual factors into the analysis (e.g. bio-physiological characteristics or lifestyles). These factors can be used to predict overall mortality rates or those for some major causes of death. A logistic model was applied to individual data obtenaid from a long-term longitudinal study. A vector of probabilities (one for every possible cause) was estimated for each individual. Various issues relating to competing mortality risks from different causes may be approched from a new angle.