Modelling seasonal mortality with individual data

Most studies of seasonal variation in mortality rely on aggregated death counts at population level. In this paper, we use individual data to present a series of models for different aspects of seasonal variation. The models are fitted to a variety of international pensioner data sets and suggest a high degree of commonality across countries with different climates and different health systems. The power of individual life-history survival modelling allows the detection of seasonal patterns in even modest-sized portfolios. We measure the tendency for seasonal fluctuations to increase with age, and we again find strong similarities between geographically distinct populations. We further find that seasonal effects are generally uncorrelated with gender, but that low-income pensioners can suffer greater seasonal swings than high-income ones. Finally, we propose a single-parameter measure for the extent to which winter mortality is a spike and summer mortality is a shallower trough, and show results for a variety of data sets.