School of Public Health

Colin D. Rehm

Socioeconomic Disparities in Health: The Role of Diet Cost

Numerous studies have linked diet quality to all-cause mortality. Diet cost has been implicated as an important determinant of diet quality and has been linked to many of the dietary patterns and scores related to adverse health outcomes, such as weight gain, type 2 diabetes mellitus (T2DM), cardiovascular disease (CVD) and all-cause mortality. However, few prospective studies have evaluated whether diet cost is associated with these adverse health outcomes. Therefore, this body of work sought to elucidate the relationship between diet cost and adverse health outcomes, while also examining the extent to which diet cost explains the association between socioeconomic status (SES) and health. 

To address these aims, we used data on post-menopausal women (ages 49-64 years) included in the Women’s Health Initiative (WHI).  Participants’ daily diet cost was estimated by linking a national food price database developed by the United States Department of Agriculture was linked to the participants’ food frequency questionnaire. The four outcomes of this study were weight gain, T2DM, CVD and all-cause mortality. Adjusted linear regression models were used to evaluate the association between diet cost and weight change, whereas Cox proportional hazards regression models were used to evaluate the association between diet cost and T2DM, CVD and all-cause mortality. To evaluate the extent by which diet cost explained the socioeconomic (income/education) gradient in outcomes, we evaluated the percent difference in the diet-cost adjusted income/education coefficients to the coefficients from models without the diet cost term.

The association between diet cost and diet cost was evaluated in 10,807 women from the control arm of the Dietary Modification (DM-C) trial. For weight change, a 50% increase in diet costs was associated with excess weight gain of 0.33 kg (95% CI 0.06, 0.59) over up-to seven years of follow-up, though the association was modified by weight change prior to baseline. Among women who previously gained weight or were weight stable there was no significant association between diet cost and weight change. For women who previously lost weight, a 50% increase in diet cost was associated with excess weight gain of 0.87 kg (95% CI 0.34, 1.40). Given the unexpected direction of the association between diet cost and weight change subsequent SES-mediation analyses were not conducted.

Over eight years of follow-up 2,174 new cases of T2DM were observed among 47,683 women from the DM-C and Observational Study (OS). A 50% increase in diet costs was associated with a 14% reduced risk of T2DM (hazard ratio [HR] 0.86; 95% CI 0.78, 0.94). In regression calibration models that incorporated estimated diet costs from the 4DFR, a 50% increase in diet cost was associated with a 22% reduced risk of diabetes (HR 0.78; 95% CI 0.67, 0.90). A strong social gradient in diabetes risk was observed for both education and income, with individuals of lower SES having an elevated risk of being diagnosed with T2DM. In mediation analyses, diet costs explained 15-19% (p<0.05 for all mediation analyses) of the association between income/education and T2DM.

With eight years of follow-up 1,208 cardiovascular events were observed among 42,632 women from the DM-C and OS. A 50% increase in diet costs was associated with a 19% reduced risk of CVD (HR 0.81; 95% CI 0.72, 0.92). In regression calibration models, a 50% increase in energy adjusted diet costs was associated with a 28% reduced risk of CVD (HR 0.72; 95% CI 0.58, 0.88). A strong social gradient in CVD risk was observed for both education and income, whereby individuals of lower SES experienced an elevated risk of CVD. In mediation analyses, diet costs explained 12-19% (p<0.008 for all mediation analyses) of social gradient in CVD.

Over 12 years of follow-up, 2,055 deaths were observed among 49,336 women from the DM-C and OS. Among the entire population, diet cost was not significantly associated with mortality HR for 50% increase diet cost: 0.95; 95% CI 0.87, 1.04). When restricting the analysis to healthy never smokers, a 50% increase in diet costs was associated with a non-significant 15% reduced risk of death (HR 0.85; 95% CI 0.70, 1.03). Given the lack of a main effect between diet cost and mortality, subsequent SES-mediation analyses were not conducted.

This is the first systematic evaluation of the association between diet cost and adverse health outcomes in the United States. Contrary to the original hypothesis, higher diet costs were not associated with decreased weight gain. For T2DM and CVD, a significant inverse association between diet costs and risk of these outcomes was observed, and for mortality, there a suggestion of an association between higher diet costs and reduced mortality risk among healthy never smokers, but this association was not statistically significant. Diet cost accounted for 12-19% of the association between income/education and T2DM and CVD. The positive results observed for T2DM and CVD should be compared to results from other studies. Examining upstream factors associated with adverse health, including diet costs, expands our understanding of socioeconomic disparities in health, while also unpacking the consequences of the contemporary food environment on disease risk.

URI​: http://hdl.handle.net/1773/25427