As far as we are aware, this is the first longitudinal study to document a relationship between area-level SEP (measured as education and income) and incident MetS. In this well-defined Australian cohort, incident MetS occurred in 18.7% of men and 13.1% of women after an average of 3.6 years follow up. These data are comparable to similar statistics in other western countries (e.g., United States (US) [61, 62]), but higher than in some Asian populations (e.g., Korean , Japan , or Taiwan ), though incidence estimates are subject to different definitions of MetS . Area-level education was independently and inversely associated with the incident MetS. Men and women living in areas where a greater proportion of the population complete a university education, independent of their own income, education, and health risk behaviours, were significantly less likely to develop MetS than their counterparts in areas where a lower proportion of the population obtain this level of education. The association between area-level income and the incident MetS, on the other hand, was modified by individual-level income in which a statistically significant association was only observed for the high income participants. These component-specific findings highlight the importance of investigating separate effects of specific features of the area-level SEP on the occurrence of MetS.
Our observation of a relationship between area-level education and incident MetS is plausible and consistent with the results of earlier studies investigating cardiometabolic risk factors. Higher neighbourhood education, for example, has in past studies been reported to be significantly associated with lower body mass index (BMI), and a lower prevalence of overweight/obesity , and hypertension . The absence of a statistically significant association with individual-level income and education in multivariate regression models suggests that the ability of area-level education to predict incident MetS was robust, over and above the predictive ability of individual-level SEP. With each percentage change in the proportion of individuals with a university education resulting in a 2% difference in the risk of acquiring MetS, socioeconomic disparities in the development of MetS could be substantial, given the marked differentials in the distribution of population with university education across Suburbs in the study region (i.e., from 0% to 21% of the population).
Several mechanisms can be proposed to explain the protective effect of area-level education on the incidence of MetS. First, a greater proportion of individuals with a higher level of education in communities can be plausibly linked to uptake of rapid dissemination of health education messages regarding cardiometabolic health such as healthy dietary behaviours and physical activity, both protective against metabolic disorders and the MetS. It is likely that such protective behaviours are quickly diffused throughout communities where a large proportion of residents have a higher level of education, leading to the establishment of social norms affecting health behaviours of even the less educated in these communities. Second, it is possible that communities with greater proportions of highly educated individuals are more aware of the impact of the residential environment on their health and thus are able to invest additional resources to establish and/or maintain a healthful living environment. Complementary literature indicates that neighbourhood education is positively associated with greater neighbourhood walkability  and availability of healthy foods [68, 69], which in turn, can encourage physical activity and healthy diet. Third, it has been postulated that highly educated individuals, who often have a high level of health literacy, tend to cluster in areas with advantaged social environments. For instance, earlier studies have reported a positive association between neighbourhood education and greater neighbourhood safety and social cohesion (e.g., ). As a result, the sources of chronic stress (e.g., poor social cohesion, violence, or crime) that induce metabolic abnormalities and MetS through endocrine pathways [46, 70, 71] would be less likely to occur in communities where a greater proportion of local residents achieve a high level of education.
In examining the relationships between area-level income and the incident MetS, no statistically significant association was found in models involving the entire sample. For the highest income participants, however, area-level income was positively, rather than negatively, associated with the occurrence of MetS. Such findings were not supportive of the proposed hypotheses and mirror the current debate on the relationship between income and health, with some arguing that area-level income is not associated with individual health outcomes [72–74]. However the counterintuitive finding as seen in the high-income participants is not without precedent. In a US-based study, for example, a positive association between neighbourhood socioeconomic advantage and a worsening insulin resistance syndrome profile was reported for black men, but not for other ethnic groups . In another study, area-level income was not significantly associated with BMI in black men and black women, while a higher individual-level income was positively associated with increased BMI in white men and black men . Furthermore, it is worth commenting that a greater risk of MetS in low income people with an university education as found in our analysis could also reflect the effect of status discrepancy. In highly educated individuals who earned a low income, there may be inner conflict over a sense of self-worth that drives undue actions and negative biological responses, leading to the increased risk for the MetS. However, this finding should be interpreted with caution given the small sample involved in the analysis and possible misreport of individual-level incomes that often occur in surveys .
The study has several important strengths. First, it is the first study using a longitudinal design to examine the role of area-level SEP in shaping the development of MetS in an Australian population. The observed associations, therefore, were not affected by the influence of prevalence-incidence bias and the potential for reverse causation as experienced by previous studies that relied on cross-sectional data. Second, the study overcomes shortcomings associated with the use of composite measures to rate the area-level SEP, unravelling relationships to assess the independent effect of area-level education and income on the development of MetS. Moreover, the analysis also included evaluation of individual- and area-level SEP interactive effects on the occurrence of MetS. Finally, important confounding factors (e.g., health risk behaviours) were taken into account in the analyses.
Our study is not without limitations. First, 23% loss to follow up might have caused selection bias. However, in a post-hoc analysis, the Wave 2 sample was similar to the baseline sample with respects to frequency distributions of participants’ baseline characteristics: age, sex, household income, education, and behaviours (i.e., smoking, alcohol consumption, and physical activity). Furthermore, the low participation rate among the eligible population may have also engendered selection bias. However, in a published analysis conducted after baseline recruitment to examine the cohort participants in comparison with the eligible population, it was reported that there were no major differences in terms of cardiometabolic behavioural and biological risk factors (i.e., current smoking status, physical activity, BMI, hypertension, blood cholesterol level), though study participants were more likely to be in the middle level of household income and education attainment (i.e., finishing high school) . Second, socioeconomic characteristics under the study were limited to income and education, which may not capture fully the multi-faceted nature of SEP. Third, area- and individual-level socioeconomic characteristics were only measured at baseline, which was not representative of lifetime socioeconomic conditions. Furthermore, possible change to neighbourhood and individual SEP during the follow up period that also influences cardiometabolic health  was not accounted for in analysis. Similarly, area-level SEP measures were not updated for those who moved to a new residential area prior to the second clinical examination. As a result, assessment of area-SEP for these participants may not be entirely accurate, potentially causing some bias in the observed relationships. However, as we found that only 16% of the entire sample had changed their residential location, the effect of bias, if present, is likely to be minimal.
Fourth, as individuals can enter and exit the definition of MetS in a given time period (e.g., one year) depending on changes in levels of its clinical components , there is a possibility that in assessing the incident MetS, the current study has missed participants who developed MetS and then reversed prior to their second clinic visit, particularly those with a long interval between the two visits (e.g., >five years). This shortcoming may have resulted in an underestimate (not an overestimate) of the true MetS incidence and strength of its relationships with area-level SEP characteristics. Fourth, in examining associations between area- level income and the MetS incidence, stratified analysis was subject to small sample sizes and was therefore potentially under-powered to detect a significant association in low and middle income groups, while the presence of a significant association in the high income group may possibly be due to chance. Finally, the findings are potentially susceptible to the modifiable area unit problem, whereby the analytical results are sensitive to the definition of the spatial unit employed . For example in this study, the operationalisation of area-level SEP at the State Suburb and reported associations with MetS incidence may be reflective of the underlying spatial properties (i.e., level of aggregation and configuration of zoning).