Overweight and Obesity in the Middle East and North Africa: Trends, Socioeconomic Inequalities and Determinants

Abstract

While there is widespread agreement at least in the scientific community that obesity has become a public health problem also in the developing world, there remains a persistent, and rather fundamental shortage of data to accurately measure the size and recent evolution of the problem, let alone its determinants. In this paper we make use of the Demographic and Health Survey, a high quality survey whose main focus has been on “traditional” health challenges in developing countries, but which has routinely collected information on height and weight, at least for women. We focus on the Middle Eastern and North Africa region, as this is one region that appears to be facing a particularly severe but yet widely under-appreciated and under-researched public health challenge due to obesity. We use data for Egypt (5 rounds from 1992-2008), Jordan (4 rounds from 1990-2007) and Morocco (4 rounds from 1987-2003/4).

We start at the macro level to demonstrate that the prevalence of obesity in those countries is significantly higher than in other countries at comparable per capita income levels. While the exceptionally high level of obesity applies to all socioeconomic groups in the country, it has thus far been the higher educated, the wealthier and the urban residents that display the largest prevalence (especially so in Egypt), unlike most other countries at a similar level of economic development. Recently, however, a trend has emerged towards a reversal in the socioeconomic gradient. The bulk of the increase in obesity appears to have occurred over the 1990ies, flattening during the first years of the new millennium and modestly declining in most recent years, at least in Egypt and Jordan. On the whole, the high levels of obesity in those countries appear to have been driven by the fairly stable, high obesity prevalence in the upper socioeconomic groups, while the observed changes in obesity over time are mostly driven by the changes in obesity within the lower socioeconomic groups. A decomposition analysis of the micro data also reveals that fairly consistently, obesity has been driven by an obesity enhancing age effect, a declining cohort effect, and a highly variable period effect, the combination of which renders the prediction of future trends very difficult. The initial multivariate analysis confirms at first sight the surprising independent, positive association between education, wealth and urban residence (alongside other relevant factors) on one hand and obesity on the other hand. Year-by-year analysis also shows that these effects are declining and that most recently obesity is becoming equally important in people and areas characterised by lower socioeconomic status. Once, however, we control for unobserved heterogeneity at the household level by estimating a within siblings (fixed effect) linear probability model (Griliches, 1979; Ashenfelter and Krugman, 1994), we tend to find that at least in most recent years, compared to people with lower education, college graduates are significantly less likely to be obese. Finally, we study the importance of neighborhood effects in determining obesity by estimating a spatial autoregressive model (Kelejian and Prucha, 1998).

Keywords: DHS, MENA countries, Obesity, socioeconomic inequalities, women