Associations between Age and Education Differences within and between Couples, and Health

Adriana Santacroce, Sapienza Università di Roma

The positive educational gradient in health is well established: highly educated are less likely to report bad symptoms. Also, differences in terms of age between the two partners are found to affect both health and mortality. In this paper, educational and age-based inequalities were studied at the European level by taking into account the clustered nature of individuals within the couples. A sample of 37,858 men and women (18,929 couples) was derived from the Survey of Health, Ageing and Retirement in Europe (SHARE, wave 5). A two-level linear random effect regression model for binary outcomes has been performed to distinguish the individual-based and couple-based contribution in reporting two or more chronic diseases. Preliminary results show how within-couple differences in education are not statistically significant, differently from the between-couple differences. Results suggest that the protective role given by a high educational does extend to the couple, regardless differences in terms of years of education within the couple.

Introduction The positive educational gradient is well established: highly educated are less likely to report bad symptoms (Marmot and Wilkinson 2005). Also, differences in terms of age between the two partners are found to affect both health (Meyler, Stimpson, and Peek 2007) and mortality (Drefahl 2010). Following a previous study on education-based health inequalities (Nilsen et al. 2012), we study educational and age-based inequalities from the perspective of couples at the European level.

Data and methods We used survey data from the wave 5 (2013) of the Survey of Health, Ageing and Retirement in Europe (SHARE) (Börsch-Supan and Jürges 2005) for individuals aged 50 and over in the following 15 countries: Austria, Germany, Sweden, the Netherlands, Spain, Italy, France, Denmark, Switzerland, Belgium, Israel, Czech Republic, Luxembourg, Slovenia and Estonia. The couple has been identified as the adult who completed the cover-screen (CV) section of the questionnaire, together with his/her partner, regardless of his/her age. A two-level linear random effect regression model for binary response has been performed to distinguish the individual-based and couple-based contribution in the health outcomes. In particular, the methodological part consists in disentangling the variables referring to both partners and couples into the couple mean level, in order to give estimates of the between-couple component; and into the deviation from the mean of the couple, in order to give estimates of the within-couple component. Moreover, we specified that the standard errors allow for intra-group correlation for the individuals living in the same country. Dependent variable has been defined as having or not two or more chronic diseases. The main covariates in the analysis were age, gender, years of education, an instrumental variable indicating which member of the couple replied first, place of residence, having material deprivation, and country of residence. Both age and years of education were disentangled into the between-couple and the within-couple dimension, to test potential differences across and within the couple.

Preliminary results

In our sample of 18,929 European couples, the proportion of men in couple with a younger women declines for younger cohorts: 52% of men aged 80+ are in couple with a women aged at least three or more years less than him; the level lowers to 28% for men aged 50-59, whilst the proportion of same age couples is slightly higher than for younger cohorts than older cohorts: percentages decline from 11% for men aged 50-59 married to same age women until 6% for men aged 80+ married to same age women.

Women have a higher likelihood than men (OR: 1.12; 95% CI: 1.01-1.24) to report two or more diseases. By increasing the age difference between the two partners there is a detrimental effect on reporting chronic disease whilst the education difference between the two partners is not statistically significant. The educational gradient at the couple level is significant, with a lower likelihood of reporting a detrimental effect by increasing the years of education. Countries that report a higher likelihood of having two or more chronic diseases compared with the reference category (Austria) are Germany, France, and Belgium; the Netherlands, Luxemburg and Israel; the Scandinavian Sweden and Denmark, the southern Spain, and the Eastern Czech Republic, Slovenia, and Estonia. By contrast, a lower risk is reported by Switzerland and Italy.

Preliminary conclusion

The findings of this study indicate that couple level explains most of the heterogeneity in having two or more chronic diseases. Moreover, it suggests that the protective role given by a high educational level does extend to the entire couple, regardless differences in terms of years of education within the couple.

References

Börsch-Supan, Axel and Hendrik Jürges. 2005. The Survey of Health, Ageing and Retirement in Europe – Methodology.

Drefahl, Sven. 2010. “How Does the Age Gap between Partners Affect Their Survival?” Demography 47(2):313–26.

Marmot, Michael and Richard Wilkinson. 2005. “Social Determinants of Health at Older Ages.” in Social Determinants of Health. Oxford Scholarship Online.

Meyler, Deanna, Jim P. Stimpson, and M. Kristen Peek. 2007. “Health Concordance within Couples: A Systematic Review.” Social Science and Medicine 64(11):2297–2310.

Nilsen, Sara Marie, Johan Håkon Bjørngaard, Linda Ernstsen, Steinar Krokstad, and Steinar Westin. 2012. “Education-Based Health Inequalities in 18,000 Norwegian Couples: The Nord-Trøndelag Health Study (HUNT).” BMC public health 12:998.

Presented in Session 1170: Health, Wellbeing, and Morbidity