Association between wealth, education and obesity in women: evidence from 111 Demographic and Health Surveys in 55 low- and middle-income countriesAssessing the relationship between national poverty rates and health outcomes

From 1980 to 2013, worldwide obesity prevalence rose 27.5% in adults and 47.1% in children, to 2.1 billion obese individuals in 2013. National prevalence of obesity exceeds 50% in some countries. Sub-nationally, it is widely accepted that there is a socio-economic gradient in obesity, particularly for women. Obesity risk shifts from high to lower socio-economic status groups as countries become wealthier.  Recent work suggests education may have a protective effect. However, accurately estimating the effects of wealth and education on obesity risk has been impeded by the use of disparate, non-standardized measures of SES, as well as a focus on high-income countries. We seek to address this gap by analyzing the association between wealth, education, and overweight and obesity in women in low- and middle-income settings. Our aim is to understand whether education is indeed protective, and whether these effects are disparate at different levels of household wealth and country economic growth.

We combine 111 nationally representative and standardized Demographic and Health Surveys (DHS) spanning the period 1991-2012, using the largest sample to date, 857,678 female DHS respondents. We use a novel, cross-country comparable wealth index based on a dichotomous hierarchical probit model to assess household wealth. Controlling for smoking, ownership of bicycles, scooters and cars, urbanicity, occupation, and age, we test multiple variants of multilevel models with country and year fixed-effects to estimate the effect of wealth and education on obesity, overweight, and BMI. These effects were estimated for the entire sample as well as separately by World Bank income groups, GBD regions, and individually for each country.  We used sampling weights and variance inflation factors to adjust standard errors. Likelihood ratio tests confirmed that wealth and education had both separate and interactive effects.

In low-income countries, the probability of being overweight or obese increases with income. The obesogenic effect of an extra dollar per day is highest for women with a primary school level education. The effect is mitigated as years of education increase. In middle income countries, the probability of being overweight or obese increases with income for all women except the most educated, for whom income has a protective effect against obesity. This suggests that there are non-trivial differences in the protective effects of education in different household income groups. Overall, there were higher estimated probabilities of being overweight or obese in richer countries. While no countries have had reductions in national obesity prevalence since 1980, these results suggest that within countries, certain combinations of wealth and education may increase or decrease the risk of obesity, providing valuable insight for targeted public health interventions.