This study examined changes in old age LE by educational attainment in the Nordic countries and decomposed the increases in LE by educational structure and education-specific mortality. Our focus was to study older populations, which increasingly contribute to overall LE. We used national register data comprising total 65 + populations in four Nordic countries, and extended our analyses to more recent years than those covered in earlier studies. Our main findings show that (1) LE increased at all ages and in all educational groups across the four countries and the gap in LE between women and men decreased; (2) inequality in LE between educational groups persisted and increased over time; and (3) most of the gains in LE65 could be attributed to the decrease in mortality among those with low education, yet the proportion of these gains explained by educational expansion was as high as 37% among women in Sweden.
Total LE65 increased in all four Nordic countries between 2001 and 2015. Danish women and men, starting from initially lower levels, experienced the highest gains in LE, decreasing the differences in LE with the other Nordic countries. Previous studies (Jørgensen et al. 2018) have demonstrated differences in LE between the Nordic countries, and recently more than 60% of the disadvantage among the Danes and Finnish men has been attributed to alcohol- and smoking-related mortality (Östergren et al. 2019).
Women had higher LEs than men, but the pace of the increase was greater among men, resulting in reduced sex differences in LE. Sex differences in LE are typically attributed to combinations of biological factors (e.g. genetic and hormonal), lifestyle and behaviours (e.g. smoking and accidents) and social roles (e.g. occupation and care-seeking) (Oksuzyan et al. 2008). A recent study shows that the major lifestyle-attributable risk factors, smoking, alcohol use and obesity, contributed to the decreasing sex gap in LE as the lifestyle-attributable fraction declined among men but increased among women between 1990 and 2014 (Janssen et al. 2021). Another recent study into causes of death in Sweden showed that the sharper decreases in mortality from ischemic heart diseases and circulatory diseases among men than women accounted for 70% of the reduced sex gap in LE (Sundberg et al. 2018). Hence, it seems that the changes in health behaviours are more favourable among men than women, even though the absolute mortality risks are still higher among men.
Our study showed that in all countries, LE was consistently lowest for those with low education, and there was a pronounced advantage in LE for those with the highest education. Most prior research is based on data limited to working ages or to age groups under 80. In line with earlier studies focusing on older populations (Martelin et al. 1998; Deboosere et al. 2009; Kinge et al. 2015), this study showed persisting and increasing educational gaps in LE also in the oldest age groups. As Rehnberg et al. (2019) show, the divergence or convergence of mortality inequalities vary by age and measure used. In this study, absolute educational inequalities decreased sharply with age but in relation to the remaining LE, inequalities were of the same magnitude or even greater at advanced ages.
The observed changes in LE in the Nordic countries coincided with a substantial compositional change in the educational structures. The decomposition analysis revealed that the change in educational structure contributed to 10–20% of the increase in LE65, but the figure was markedly higher among women in Norway (30%) and Sweden (37%). Overall, the contribution of change in educational structure translates into 0.2–0.4 years increases in LE65 between 2001 and 2015. Earlier research has shown rather similar relative contributions of educational expansion to the increase in LE and also a greater advantage among women compared with men, albeit mostly in different countries and at younger ages (Luy et al. 2019).
The four Nordic countries started with different educational distributions, LEs and educational inequalities in LEs, and experienced different developments in them over the study period. Since LE systematically increased in all educational groups across all countries, the change in the educational gap was related to the different pace of gains in LE. For instance, women in Finland and Denmark had large proportions of people with low education who had relatively large increases in LE, which contributed greatly to the total increase in LE. However, women in Norway and Sweden had lower proportion of people with low education who exhibited modest increase in LE, which contributed less to the total increase in LE. As a consequence of decreasing proportion of people with low education, those remaining in the group may have become more negatively selected over time in terms of health and personal characteristics (Mackenbach 2010; Dowd and Hamoudi 2014). In such a case, mortality risk in this group would be increasingly driven by people with health problems and the progress in LE would be harder to achieve. In this study, educational expansion concerned especially the youngest part of the population who have the lowest mortality risk. Thus, the impacts of the greatest reductions in the group with low education are yet to be seen, as the younger people who have experienced educational expansion have not yet reached old age.
Finland was the only country where the educational gap basically remained the same over time. The change in the educational structure was different in Finland, which started with the lowest level of education. Finnish men and women had the greatest increases in the proportion of people with middle education, which in comparison to other education groups, showed the lowest increase in LE. This could imply that the social composition of the middle education group became increasingly heterogeneous i.e. those with less favourable living conditions and thus higher mortality risk had increased access to higher education. At younger ages, the socioeconomic mortality gap has been large in Finland, especially among men (Mäki et al. 2013). However, in this study among populations aged 65 and over, the educational gap in LE among Finnish men was similar with Swedish and Danish men. While the LE in old age does not significantly differ between Finnish men and men in other Nordic countries, it has been shown that the proportion of the birth cohorts reaching old age is lower among men in Finland than in the other Nordic countries (Jørgensen et al. 2018). Hence, mortality selection into old ages might be more pronounced in Finland, and if this effect is stronger among those with low education it might lead to reduced inequalities. Another potential reason for relatively small inequalities in Finland might stem from the fact that it was impossible to separate those with missing information on education from the low education group. If those with missing information had higher than basic education, that might dilute the differences between those with low education and other educational groups.
Denmark and Norway but also Swedish men had great increases in LE among those with high education, which increased the gap between high and low education groups. Education has become a more important source of social division over time, as it has become increasingly associated with career development and assets, which may in turn lead to an accumulation of resources beneficial to health. It is possible that the improved living conditions together with improved access to higher education increased advantage in this educational group. Overall, Norway showed the most pronounced educational gap in LE65 among both women (3.0 years in the first and 3.9 years in the last period) and men (3.7 and 4.1 years) and the largest increase in the educational gap in LE65 over the study period. Norway is ahead of the other Nordic countries in terms of educational expansion; therefore its Nordic neighbours may well be looking to a future of widening educational inequalities in LE.
A recent stalling of LE progress in the UK has been attributed to cutbacks in funding for health care and social welfare programmes, which has led to an increasing mortality, especially in old age (Hiam et al. 2017). On the other hand, it has been shown that an increase in health care expenditure may contribute to narrowing educational mortality inequalities as it has greater impact on mortality decline among those with low education (Mackenbach et al. 2019). Szebehely and Meagher (2018), and Rostgaard et al. (in this volume) argue that eligibility criteria for access to elder care services such as residential care and home help have tightened in the Nordic countries. Declining coverage of care has in turn disproportionally increased family care among people with fewer resources, while care purchased from the market and paid out-of-pocket by the user has increased among those with more resources. The recent changes in elder care in the Nordic countries and their impact on old age mortality and social inequalities in mortality call for more research.
Strengths and limitations
The strengths of this study include (1) the use of national register data from four Nordic countries; (2) the use of harmonized life tables and decomposition methods; and (3) the focus on the oldest age groups in analysing inequalities in LE. In addition, educational attainment is an important indicator of socioeconomic conditions since it has far reaching implications for future employment and earnings. Several important limitations of our study should be acknowledged. First, while education is often gained early in life, is stable over time and is likely to precede the onset of diseases and disability for most of the population, it cannot be ruled out that part of the association between education and mortality might be due to health-related selection into low educational attainment. Second, educational attainment is a cohort characteristic, and it should be noted that period life table approach used in this and many other studies provides inferences on mortality of a hypothetical life table cohort reflecting current (period) mortality conditions and does not refer to mortality experience of real cohorts. However, the period mortality rates are affected by the different cohort experiences of the age groups constituting the period life table, including differential access to education. Furthermore, applying cohort life table approach is quite limited for studying the most recent period because such life tables rely on the observed death rates for extinct or at least almost extinct cohorts (Barbieri et al. 2015). Third, it was not possible to completely harmonize the educational classifications across the four countries. For Finland we could not distinguish missing information on education from those with low education, and information on education was missing for older people born before 1922 and 1915 in Denmark and Sweden, respectively. This limitation was partially resolved using a redistribution method for deaths and population exposures with unknown education. The method was validated by sensitivity analyses using Danish register data for recent calendar years and by comparing results from the Németh et al. (2021) study with the results without information on education at advanced ages. Although the adjusted estimates for Denmark and Sweden look plausible, education-specific LE estimates for the most advanced ages beyond 80 rely on extrapolation assumptions and should be treated with caution. More efforts using linkages to other data sources on education such as prior enumeration-based censuses are needed to produce more reliable and statistically robust results.
The persistent and even growing educational gaps in LE in older ages coincided with increasing LE in all educational groups. Although rising educational levels in the Nordic countries carry potential for further gains in national LE, educational expansion has contributed to uneven gains in LE between education groups. The pronounced educational gaps observed in LE and unequal pace of improvement in older ages in the Nordic countries should be a particular concern because overall LE progress depends on survival and health status at increasingly older ages. The unfavourable trends in LE development, i.e. the large educational mortality gradient and the slower increase in LE compared with the best-performing high-income countries, call for more in-depth research on the potential impact of social policies on maintaining or even increasing health inequalities in welfare states.