Large and Growing Social Inequality in Mortality in Norway: The Combined Importance of Marital Status and Own and Spouse's Education

IN ALL COUNTRIES, age-specific death probabilities vary between sociodemographic groups. Numerous studies have documented large mortality differences betweenmarried and non-married persons (Roelfs et al. 2011; Shor et al. 2012a, 2012b), and mortality is associated with a person’s number of children, which in turn is related to marital status (Grundy and Kravdal 2010). The evidence for differences between educational groups is large as well (Elo 2009), and there is growing interest in the association between mortality and spousal education (Brown et al. 2014; Kravdal 2008; Skalická and Kunst 2008). Furthermore, several studies have shown differences in mortality by income, occupation, or ethnicity (Bævre and Kravdal 2014; Harper, Rushani, and Kaufman 2012; Tarkiainen et al. 2015; Wada et al. 2012), and there are differences between geographic regions of a country that are probably not fully explained by differences in socioeconomic composition (Kravdal et al. 2015). These mortality differences reflect the importance of social support and control, knowledge, purchasing power, and various other factors, including selective influences. When studying differences in mortality between sociodemographic groups, it has been common to focus on only one variable, but many investigators have taken a broader perspective and shown and discussed the main effects of a number of variables. Some have even taken into account interactions between variables (Kohler et al. 2008; Smith and Waitzman 1994). However, estimates frommultivariable studies have rarely been used to predict differences between sociodemographic groups defined by combinations of the considered variables. This means that we have an inadequate impression of how much variation exists in the population—which may have implications for discussions about the need for policy interventions.