In this population-based sample of older adults, we found education-related differences in balance and walking speed, even after adjustment for chronic diseases and lifestyle-related factors. However, the differences in grip strength and chair stands were attenuated by this adjustment. Analyses stratified by age group revealed that the differences were less evident in advanced age (80+ years). Gender-stratified analyses revealed that women with university education had significantly better grip strength, balance and walking speed compared to women with elementary school education and men with university education had significantly better chair stands and walking speed compared to men with elementary school education, in the fully adjusted models. Further analyses showed that the difference in grip strength was only evident in female manual workers, while the difference in balance and walking speed was only evident among female and male non-manual workers, respectively.
Earlier reports on age variations in social inequalities in health are contradictory [8–12]. Variations between measures of health used in the different studies have been suggested to contribute to the inconsistent results . To our knowledge, this is the first study showing age variations in education-related differences in objectively measured physical performance. The education-related differences were smaller in people 80+ years compared to people <80 years in all four dimensions of physical performance in the fully adjusted model. This finding may have several possible explanations. First, exposure to psychosocial risk factors among persons with lower education or the impact of these factors on health may be greater among people in middle age and early old age compared to the oldest-old . Secondly, the relative importance of education on physical performance may be less at older ages due to frailty . Thirdly, the age differences may also reflect differences between birth cohorts. Finally, this finding could be explained by differential survival effects, since the oldest persons with worst physical function and lower education may be less likely to be included in the study due to death. However, this hypothesis was not supported by a study examining mortality selection in old age . Since the interactions between education and age group were only significant for balance, borderline significant for walking speed and non-significant for grip strength and chair stands, no ultimate conclusions can however be drawn on the age differences.
We found that women with university education had significantly better grip strength, balance and walking speed compared to women with elementary school education and men with university education had significantly better chair stands and walking speed compared to men with elementary school education. Our results suggest that there are different patterns of education-related differences in physical performance in older men and women. The gender differences may be explained by differences in the types and durations of exposures in men and women with various levels of education, respectively, such as work-related exposures. However, since the interactions between education and gender were found to lack statistical significance for all outcomes, the results on the gender differences have to be interpreted with caution.
Stratified analysis by gender and occupational class showed that the education-related difference in grip strength was only evident among female manual workers. This may be explained by the fact that heavy manual work may maintain physical capacity, specific to the task that had been performed during work, in aging . However, this pattern was not seen in men, perhaps because of differences in types and amounts of work-related tasks in male versus female manual workers. Previous reports on social gradients in grip strength are contradictory. One study found low income but not low education to be associated with poor grip strength . Manual work has been associated with worse grip strength among people aged 80+ years , while others have reported no such association among younger populations . The contradictory findings in the literature may be explained by differences between the cohorts, such as differences in age, countries, and education of the study populations [4, 37]. Another explanation may be the modifying effect of occupational class on education-related differences in grip strength among women, as found in the present study.
Further stratified analysis by occupational class suggested that the education-related differences in balance in women and in walking speed in men were only evident among non-manual workers. One possible explanation may be that strenuousness of manual work contributes to poor performance in balance and walking speed independent of educational level. In addition, non-manual workers with higher education may be exposed to more favourable both work-related and non-work related factors compared with lower educated non-manual workers. However, it must be pointed out that since manual workers were fewer than non-manual workers, the lack of significant education-related differences among the manual workers in balance and walking speed may partly be explained by low statistical power. For the same reason, it is however less likely that the lack of education-related differences among female non-manual workers in grip strength is due to low statistical power.
Chronic diseases explained only a small part of the differences in physical performance in our findings. By contrast, a previous study reported that chronic diseases explained almost 40% of the education-related differences in mobility . A possible explanation for the contradictory results may be the differences in the type of chronic diseases that were included. The previous study included a pre-selected list of common chronic diseases, while we included all prevalent chronic diseases. Moreover, other studies investigating the mediating effect of lifestyle on education-related differences in physical performance showed conflicting results [5–7, 21]. We found that physical activity explained some of the education-related differences in physical performance, suggesting that promoting physical exercise may help to reduce some of the education-related differences in physical performance in old age. Since people in more socially disadvantaged groups may participate less frequently in leisure-time exercise , promoting participation in these groups may be an important public health target. However, the differences between results from various statistical models were not statistically significant for any of the outcomes.
A major strength of this study is the large sample of people, living both at home and in institutions, providing a complete picture of the general older population in our geographic location. Moreover, we employed objective testing of physical performance instead of subjective self-reported measures, and we used different sources of medical diagnoses, including direct clinical examination, thus limiting potential biases. However, the results for walking speed may have been biased by the different distances covered. The longer walk may better reflect a person’s normal walking speed than the shorter walk. However, it was not feasible to perform the longer walk at home visits due to restricted space. Excluding this group would have led to greater bias in the results, since the subjects assessed on home visits constitute a more disabled group in comparison with the entire sample. In addition, data from the National Health and Nutrition Survey (NHANES) showed that walking speed measured over the distances 2.4 and 6 meters are comparable . Moreover, tests for walking speed are generally considered highly reliable, regardless of the distance [29, 39, 40].
Another limitation is the missing data on physical performance, especially on grip strength, which was missing for all subjects, assessed at home visits. We chose to impute the missing data on physical performance. Imputation of large amounts of missing data may lead to biased results, therefore the results for grip strength should be considered with caution. However, excluding this group probably would have led to greater bias in the results since the subjects with imputed data had lower education and worse physical performance compared with the entire sample. Other limitations of the study include the cross-sectional design, which restricts the ability to determine the direction of the observed associations. However, educational attainment was achieved long before the assessment of the outcomes. It is unlikely that directionality is questionable. Moreover, possible selective survival bias due to the cross-sectional design may have led to underestimations of the education-related differences, since persons with low education who also had worse physical performance may die earlier than highly educated persons . Finally, health inequalities in Sweden, a country with a large nationalized system of public services, such as education, health care and social benefits, and extensive welfare , may be smaller than in some other countries.