Validated analysis of mortality rates demonstrates distinct genetic mechanisms that influence lifespan.

A key goal of gerontology is to discover the factors that influence the rate of senescence, which in this context refers to the age-dependent acceleration of mortality, inversely related to the morality rate doubling time. In contrast factors that influence only initial mortality rate are thought to be less relevant to the fundamental processes of aging. To resolve these two determinants of mortality rate and lifespan, initial morality rate and rate of senescence are calculated using the Gompertz equation. Despite theoretical and empirical evidence that the Gompertz parameters are most consistently and reliably estimated by maximum-likelihood techniques, and somewhat less so by non-linear regression, many researchers continue to use linear regression on the log-transformed hazard rate. The present study compares these three methods in the analysis of several published studies. Estimates of the mortality rate parameters were then used to compare the theoretical values to the actual values of the following parameters: maximal lifespan, 50% survival times, variance in control groups and agreement with the distribution of deaths. These comparisons indicate that maximum-likelihood and non-linear regression estimates provide better estimates of mortality rate parameters than log-linear regression. Of particular interest, the improved estimates indicate that most genetic manipulations in mice that increase lifespan do so by decreasing initial mortality rate, not rate of senescence, whereas most genetic manipulations that decrease lifespan surprisingly do so by increasing the rate of senescence, not initial mortality rate.