Consistent with previous literature [6, 8, 11], education was a significant predictor of healthy aging in models that combined men and women. As educational attainment increased, so did the likelihood of healthy aging. The goal of this study was to examine the role of income and occupation as possible mediators of the relationship between educational attainment and healthy aging. In models that combined men and women, perceived income adequacy was not significantly related to healthy aging. Among men, however, there was a negative gradient between perceived income adequacy and healthy aging when the effects of income and occupation were considered together in the presence of education: men who thought their income met their needs “very well” were, unexpectedly, less likely to meet criteria for healthy aging compared to those who were at least somewhat dissatisfied with how their income met their needs. Unlike the perceived income adequacy measure, there was a significant positive gradient between life satisfaction with finances and healthy aging in models of men and women separately as well as combined (Tables 3, 4 and 5, Models 4 and 5).
The difference in significant predictors for models with or without education indicates that while education explains the influence of occupation on healthy aging for men, it does not fully explain the influence of perceived income adequacy and life satisfaction with finances on healthy aging, leaving the possibility open that these variables could act as mediators. Additionally, the findings suggest that among men, perceived income adequacy and life satisfaction with finances are more important predictors of healthy aging than education. For women, however, education remains the more influential predictor of healthy aging.
Overall, the study provides evidence that income may mediate the relationship between education and healthy aging among men, but not women, consistent with previous research indicating that, among men, wealth was a more important predictor of physical function than education [20]. There was little evidence, however, that occupation mediates this relationship for either men or women. This conclusion was drawn based on criteria [36] of eliminating or at least reducing the association between education and healthy aging. Together, perceived income adequacy and life satisfaction with finances eliminated the significant association between education and healthy aging among men. In addition, the effects of occupation on healthy aging disappeared for men after education was considered, providing support for a causal model where education influences occupation, which in turn influences financial resources [15]. In contrast, for women, our findings of a strong and persistent influence of education on healthy aging supports the pre-eminence of education over income and occupation as predictors of healthy aging for women.
The differences in our findings by gender may relate to differences in labour force participation and domestic responsibilities between men and women [37], especially among this older cohort, and may not hold for younger cohorts. A substantial proportion of women in this study were homemakers, and men may have been more likely to provide the primary source of income for their families. Subsequently, the men may have placed more importance on finances and their health may have been more affected by financial stress than women. Indeed, previous research has shown that men have significantly higher levels of financial stress than women [22].
Based on the negative gradient observed between perceived income adequacy and healthy aging, one might speculate that men who believe their income meets their needs may have had higher-paying jobs that negatively affected their health. Alternatively, one could argue that individuals who are healthier may have greater financial needs (e.g., travel) than those less healthy in a sample who all had health insurance coverage; those who are less healthy may, in comparison, be more easily satisfied with how their finances meet their needs.
The difference in the direction of the gradient observed for perceived income adequacy and life satisfaction with finances was an unexpected finding, and highlights that these measures have some important conceptual differences. Although these measures are both based on subjective income assessments, perceived income adequacy reflects how their income meets their needs, while life satisfaction with finances measures more general satisfaction with finances. Future research could examine the reasons for these differences, such as psychological and lifestyle-related factors.
Perceived income adequacy and life satisfaction with finances were used as proxy measures of income because a substantial proportion of participants (15.9 %) did not provide information on household income. This level of non-response is similar to that found in other studies [38]. In addition, response rates may differ by income level, indicating the potential for further bias [38]. Use of the household income variable in our study would have created a biased analytic sample as individuals who provided household income differed significantly by healthy aging status. Those who met criteria for healthy aging were more likely to have provided household income information than those who did not meet criteria for healthy aging. Thus, use of alternate measures of income (perceived income adequacy and life satisfaction with finances) was necessary. Although perceived income adequacy and life satisfaction with finances are conceptually different from income, these subjective income measures may be important in providing a perspective independent of income in dollars. This is supported by our finding of a significant association of healthy aging with life satisfaction with finances and perceived income adequacy for men. In addition, sensitivity analyses adding household income to the full model produced the same pattern of results as the analysis without income (Table 3, Model 9), and income was not a statistically significant predictor once education was included in the model (refer to Additional file 5, Model 3 and Additional file 6, Model 3). Our results also indicate that the measure of income that is used can influence the observed association with healthy aging.
Whereas education and income reflect resources, occupational prestige is a direct measure of social standing. It has been associated with measures such as self-reported health, with higher prestige associated with reduced likelihood of poor or fair self-reported health [39]. Limitations related to the occupational prestige measure, however, should be noted. In this older cohort, women were less likely to work outside the home and employment may have been more stable. These changes may decrease the generalizability of our findings to future older adults. The occupation classification system was last updated in the 1980s while the data were collected in the 1990s. Although the levels of prestige associated with certain occupations may have changed over that decade, most participants were already retired and thus earlier classification systems may be most relevant.
Another potential limitation is that healthy aging was defined as a dichotomous variable (presence vs. absence). Although commonly done, alternate definitions of healthy aging using multiple levels or as a continuous index in future studies may provide a more detailed picture of this construct and its predictors.
The results may not be fully generalizable to all populations. This cohort of older adults spent their childhood and early adulthood during the Depression and the Second World War. The education they received and their early careers may have been shaped by those experiences. These formative experiences may have been quite different for other groups of people in other places and other times. Similarly, the gender differences we observed may have been unique to this group of people.
A substantial portion of the baseline sample was deceased, institutionalized or otherwise not eligible to participate at follow-up. Attrition through death is inevitable in longitudinal research, particularly in the aging field, but is important to note as it has the potential to produce bias given that mortality and institutionalization do not occur randomly. For example, those who were dead or institutionalized had lower levels of education than others in the sample (see Additional file 2). These systematic differences may lead to problems generalizing the findings to a broader population of older adults, but they do reflect the selective sample of older adults able to remain the community.
This study was based on data from a population-based, prospective cohort study with a high response rate. Additional study strengths included using two different proxy measures of income (life satisfaction with finances and perceived income adequacy) as well as comparisons of these proxy measures with household income and sensitivity analyses. The study used a multidimensional definition of healthy aging, incorporating physical, cognitive, social, and psychological domains of health. It is generally recognized that absence of disease and good mental health are critical aspects of successful aging, but that overall life satisfaction, social participation, personal growth, and factors identified as important by the general population should also be considered [40–42]. This study utilized a definition of healthy aging that comprised a wide breadth of measures and thus captured healthy aging in a manner consistent with this broader definition.
Future research directions include examining other potential mechanisms by which education may be associated with healthy aging, especially in women. Such mediating factors could include greater engagement in productive activity, perceived control, social support, health beliefs, health service use and cognitive reserve. This area of inquiry is complex. Associations between childhood opportunities, educational attainment, occupation, and income may all be important and inter-related. Furthermore, they may operate at different stages of life. Finally, different cohorts may experience these effects differently. Access to education, job security, and income can change considerably over time, as can their associations with gender. Further study should consider a life course approach following large populations over long time frames.