Since the late 1960s, family arrangements and trajectories have become less predictable and less stable. The second demographic transition brought along various changes in family formation and dissolution, and these interact with health and mortality in a number of ways. For instance, the excess mortality of unmarried persons or single mothers has been repeatedly demonstrated in literature. The new family forms obviously call for a more detailed approach. In particular, the diversity of partnerships and household compositions has to be taken into account. These changes require going beyond cross-sectional snapshots of the population and adopting a longitudinal approach to evaluate how inequalities in health result from the accumulation of advantages or disadvantages across the life course. Despite the existence of longitudinal data in Belgium, there is still very limited research in this area. To fill this gap, three lines of inquiry will be pursued in WP3:
Long-term trends in mortality differentials by marital status from 1970 to 2006
There is accumulating evidence that mortality differentials by marital status have increased since the 1960s in several western countries. Is this pattern observed in Belgium as well? How are differentials related to the share of the population that is unmarried, divorced or widowed? This analysis will be based on data from the censuses from 1970 to 2006, which offer a long reference period.
In-depth analysis of cause-specific mortality differentials by marital status and family arrangements, 1991-2010
The population will be stratified into three broad age categories: (1) children aged less than 20, (2) young and middle-aged adults (20-39; 40-64), and (3) adults after retirement age (65 and over). For each of them, several states will be distinguished (e.g. married, cohabiting, with or without children, living alone, etc. for adults; child of a married couple, a cohabiting couple or a lone parent household etc. for children), using register data. Rate differences in overall mortality will be calculated and will be decomposed in order to determine the share of each cause (group) of death to these inequalities. Cause-specific rate ratios will be estimated from Poisson regressions, and age-adjusted risks ratios will be used to assess the extent of the mortality gradient. Adjustments will be made for socio-economic characteristics to evaluate to what extent mortality differentials arise from selection.
Longitudinal analysis of mortality differentials according to individual life courses
In this part of WP3, the method of sequence analysis will be used in order to identify typical trajectories of family formation and dissolution. It will then be possible to evaluate how the instability of the life course is related to differentials in mortality. Another question is whether individuals following less dominant patterns are more likely to experience premature death.