We have presented how educational inequalities in health behavior may influence life expectancy. Our analysis shows how a clear educational gradient in health behavior translates into life years lost. The percentage of respondents losing more than five years of life was largest among the lowest educated and decreased with higher level of education. The proportion of respondents that are expected to lose 1 year of life expectancy or less was highest among those with the highest level of education and lowest between those with the lowest level of education. Our findings are in line with previous studies where educational differences in health behavior were also found (12, 37, 38). The paths linking socioeconomic status to better lifestyle choices and health are well documented (11, 13, 39–41). In this study, the respondents with the higher levels of education share the characteristics through which some of the mechanisms operate, such as good childhood circumstances (42), living with a partner or spouse and their high education (43). Despite these factors being known, our study adds an illustration of the potential life lost due to unhealthy behavior across educational levels.
The present study has aimed to present a conceptual approach that may be further explored to find more exact estimates of the impact of social determinants on life expectancy. Each part of this study has limitations, some of which may be further explored in future research. First, the sum of years of life lost associated with each behavior is based on the assumption that single health behaviors contribute independently to life expectancy. It is important however, to acknowledge that interactions exist among behaviors and that their combined effects may not be additive (44). The years of life lost attributed to a health behavior reported in the literature may reflect a joint effect rather than a single effect. Some behaviors tend to cluster (i.e. physical inactivity and obesity) and few studies in our review had the necessary data to account for the influence of other health behaviors or other indicators. Although overlooking both potential interactions and confounders influences the estimate of the magnitude of the years of life lost, this would likely have less impact on their distribution across education levels, which was the primary aim of this paper. Secondly, the estimates extracted through our narrative review are mere approximations based on reviewing a vast literature base. Obtaining an exact estimate of the total years of life lost due to unhealthy lifestyle would be close to impossible based on a review of current available research, due to very different reporting. In addition, the literature is seldom published with the categories that are provided through the Tromsø Study, limiting the possibility of linkage. Third, in the assessment of the educational inequalities, the use of self-reported measures and the inherent drawbacks of the cross-sectional design are clear limitations. Moreover, all respondents with information on health behavior were included, regardless of whether they were already experiencing a chronic disease. A further limitation is that respondents that reported to have smoked daily previously at the time of the survey were no longer considered as daily smokers. Similarly, physical activity guidelines depend on age and intensity, and this was not taken into account. Alcohol consumption guidelines in the Nordic countries are different according to gender, where men are recommended not to exceed 14 units of alcohol per week and women should not exceed 7 units of alcohol per week. We, however, used the same cut off point of 14 units of alcohol per week for both men and women to facilitate linkage to published estimates of life year loss. Healthy eating was not included among the health behaviors assessed in this study due to the challenges in measuring diet. We did, however, include BMI which captures some of the effects of nutritional status. The introduction of mobile applications may improve the quality of measurement of diet for future epidemiological studies, such as in the plan for the eight wave of the Tromsø Study. Similarly, wrist watches and other mobile phone applications are being increasingly used in the measurement of physical activity, which may also improve future measurements of this behavior (45).
This study has provided insight into the potential lifetime losses of adopting an unhealthy lifestyle and the social inequality embedded in it. Additional strengths of this study include the sample size and the assessment of lifestyle risk factors that are responsible for approximately 60% of premature deaths (15). The findings allow the identification of groups at higher risk (i.e. those that would experience the highest loss of life expectancy by engaging in multiple unhealthy behaviors), which supports the prioritization of public health policies. Future research along the lines presented here could include exploring long-term patterns of health behavior and investigate the role of lifestyle as a mediating variable in the association between socioeconomic position and healthy life and life expectancy.