Mortality change roils period rates

Mieke Reuser, Max Planck Institute for Demographic Research
Jutta Gampe, Max Planck Institute for Demographic Research
James W. Vaupel, Max Planck Institute for Demographic Research

When a life is saved under a more favourable mortality regime, how long is death averted? How do the number of lives saved and the lifespan increments affect period lifetables? It turns out that a change in mortality regime produces a wake of turbulence that disturbs period rates as well as life expectancy. The conventional life table approach supposes that when a life is saved under new mortality conditions that the individual gains, on average, the remaining life expectancy at that age. It is however much more likely that individuals that were saved, will die sooner. This means that period mortality rates and life expectancy are overestimating the actual mortality conditions directly after a change to a healthier regime. Until the last person dies who was born under the old regime but whose life is saved under the new, life expectancy as conventionally calculated does not equal the life expectancy of a cohort living under the new regime. How large the distortion of the conventional life expectancy calculation is when death rates are declining, depends on how long death is postponed. The Max Planck Institute for Demographic Research houses a lightbulb laboratory in which large numbers of small bulbs can be lit at either 5 or 6 volts. A population of lightbulbs can be kept under harsh 6 volt conditions at first and switched to less stressful 5 volts thereafter. Data from lightbulb experiments can provide empirical evidence on the nature of mortality turbulence after a regime change. Preliminary results of these experiments show that mortality rates do indeed suffer from historical conditions and the model describing the lifesaving seems to be a combination of frailty and an accelerated failure time model. Current and upcoming experiments will give more insight in how many ‘lives’ are saved and for how long.