Updating the association between socioeconomic status and obesity in low‐income and lower‐middle‐income sub‐Saharan African countries: A literature review

Globally, the literature tends to emphasize negative associations between socioeconomic status (SES) and bodyweight in countries improving their economic development. However, little is known about the social distribution of obesity in sub‐Saharan Africa (SSA) where economic growth has been highly heterogeneous the last decades. This paper reviews an exhaustive set of recent empirical studies examining its association in low‐income and lower‐middle‐income countries in SSA. Although there is evidence of a positive association between SES and obesity in low‐income countries, we found mixed associations in lower‐middle‐income countries, potentially providing evidence of a social reversal of the obesity burden.


| INTRODUCTION
In recent decades, excess weight and obesity have continually increased around the world, contributing to the rise in related chronic diseases such as diabetes, cancers, and cardiovascular diseases. By 2025, a global obesity prevalence of 18% among men and 21% among women is predicted, with one in every two adults being classified as overweight. 1 In parallel with the globalized weight gain, the worldwide nutritional panorama remains contrasted, with regional disparities.
Whereas high-income and higher middle-income countries, mostly located in Europe, Asia, and America, are facing a dramatic overweight and obesity epidemic, low-income and lower-middle-income countries, most of which are in sub-Saharan Africa (SSA), are in an unprecedented situation. Indeed, in SSA, the increase in obesity coexists alongside a persistence of hunger and starvation. In West, Central, and East Africa, for instance, the prevalence of female overweight increased significantly between 1975 and 2014 and currently ranges from 30% to 40%, whereas female underweight prevalence remains high (from 10% to 15% depending on the region). 1 There is also evidence that excess weight prevalence reaches higher levels in urban areas and among women compared with rural areas and men. 2 This coexistence of high rates of both underweight and excess weight in the same region or country is referred to in the literature as the double burden of malnutrition 3 and constitutes a major public health concern, with significant policy implications. 4 Investigating the socioeconomic determinants of weight gain in SSA thus calls for greater attention among scholars.
The social and economic drivers of individual bodyweight, such as education, income, wealth, and occupation, may strongly depend on the national level of economic development. 5 Previous literature surveys have contrasted the situation in high-income and low-income countries 5-7 with a negative relationship being noted in richer countries (i.e., excess weight affects the lowest socioeconomic groups in particular) versus a positive relationship in poorer countries (i.e., excess weight affects the highest socioeconomic groups in particular). Nowadays, the worldwide nutritional panorama has changed and does not seem so dualistic. In SSAn countries namely, whereas certain authors find a positive association between SES and body mass indicators, others find no association and, sometimes, negative associations. Even if the literature is still unclear about the association between SES and excess weight in SSA, there is emerging, albeit limited, evidence of a shift in the distribution of overweight and obesity across SES groups, in line with the diffusion theory established by Agyemang et al. 8 Inspired by Monteiro et al., 9 this theory states that obesity and related chronic diseases spread from the rich to the poor as the living standards of the poorest improve. Empirical evidence confirms this theory in the case of upper-middle-income countries such as China, 10,11 Mexico, 12,13 Indonesia, and South Africa. 14 The existence of a shift in the SES-excess weight association is closely linked to complex economic, social, and cultural dynamics embedded in the nutrition transition process. Theorized by Popkin,15 nutrition transition refers to changes in the composition of diet, resulting in an increase in calorie intake accompanied by a reduction in energy expenditure due to the adoption of more sedentary lifestyles. People tend to gain weight because of this increasing calorie imbalance, and such nutritional changes generally occur with economic development, urbanization, technological change, and globalization. 16 Many SSAn countries are experiencing a rapid and intense nutrition transition, 17 and two main factors may account for the potential shift in the SES-excess weight association in this specific context. First, the environmental and economic conditions undoubtedly matter. Influenced by rapid urbanization driven by rural-to-urban migration and the improvement in lifestyle conditions linked to economic development, abundant food has become more widely available, explaining why excess weight gradually affects lower-SES groups. Second, according to Agyemang et al., 8 the literature has also alluded to the role of nutritional and weight perception issues in the specific context of SSA. Lastly, the influence of socio-cultural beliefs linking excess weight to prosperity, power, or good health, as noted by several scholars (e.g., Renzaho 18 ), appears to have lost ground, especially among higher SES groups. The spread of westernized ideals of thinness together with healthy dietary recommendations through an improved access to high schools and universities may explain why higher SES groups are increasingly rejecting stoutness and adopting healthier behaviors. 19 The purpose of this literature review is to provide a detailed overview of the complex shift in the SES-bodyweight association that is taking place during the ongoing nutrition transition in SSA. Although several literature surveys have already been conducted on the topic, 8,[20][21][22] this review brings two main contributions. First, it provides insights based on a comparison of low-income and lowermiddle-income SSAn countries regarding the SES-bodyweight association. Based on a survey of recently published studies (i.e., between 2010 and 2020), our comparative approach emphasizes different associations between low-income and lower-middle-income SSAn countries. Whereas there is evidence of a positive association between SES and excess weight in low-income SSAn countries, the associations are mixed in lower-middle-income countries. A second contribution of this review is to highlight comparative evidence from rare empirical and analytical studies reporting U-inverted trends between SES and body mass index (BMI) in samples of middle-income countries. [23][24][25] This suggests a social reversal of the obesity burden along with the process of economic development and nutrition transition, confirming the diffusion theory.
2 | METHOD 2.1 | Search strategy and inclusion/exclusion criteria A comprehensive search through the existing literature was carried out on Google Scholar and PubMed databases using specific keyword combinations associating two-by-two weight-based and SES-based terms. We systematically associated "obesity," "overweight," "excess weight," "bodyweight," and "body mass index" with "socioeconomic status", "social class," "household income," "wealth," or "education" (i.e., a total of 25 combinations). Next, to minimize exclusion errors, we also used an alternative search approach whereby articles cited within another article were reviewed. Thomas, 14 these richer countries were excluded from the sample as they have already reached a more advanced stage of nutrition transition. Moreover, upper-middle-income and high-income countries in SSA have already been the object of numerous investigations into the social and economic determinants of obesity and non-communicable diseases. 29 Hence, in this review, we decided to focus on low-income and lower-middle income countries for which retrospective research is missing. All in all, we were able to collect 52 articles that examine 131 associations. Among these 131 associations, 78 concern lowermiddle-income countries and 53 concern low-income countries.

| Indicators
In the empirical literature reviewed in the study, several indicators were used to quantify bodyweight among individuals, specified as a dependent variable in statistical models. BMI was the most common indicator, defined as weight (kg) divided by squared height (m 2 ). Using  Tables 1 and 2). Different indicators were also used to capture household or individual SES, generally specified as the explanatory variable of interest in the empirical papers reviewed. SES is a multidimensional concept that can be defined as an individual's or household's position within a hierarchical social structure. 31 In the empirical literature, SES is generally measured by household income, household assets, individual occupation, and/or individual education attainment. Due to data availability issues, the papers tend to focus on individual education attainment and household wealth or assets (respectively, 61 and 34 associations). However, some studies use household income, individual employment, or household-based composite indices as SES indicators.

| RESULTS
We list the results of the papers that focus on low-income and lower-middle-income SSAn countries in Tables 1 and 2, respectively. A total of 52 empirical studies focusing on a single or several countries in SSA were reviewed, with 131 associations examined. To increase the results' readability, Tables S1 and S2 also present the nature of associations found according to the SES indicator and the sub-period (2010-2015 and 2016-2020) considered. Table S3 lists the studies reviewed with information on the bodyweight indicator(s) selected, the number of associations highlighted, and the type of sample considered (women only or both sexes).

| Low-income countries
As shown in Table 1 and Burkina Faso (e.g., meat and poultry, dairy products, sweets, and sweet drinks), these energy-dense foods remain relatively expensive and cannot be consumed by the poorest members of society. [33][34][35] Finally, for several authors, the positive link between SES and excess weight in low-income countries is also due to macroeconomic drivers such as urbanization, tertiarization, and globalization. 34,35,40,43,52 In short, Western lifestyles have gained a foothold among the most affluent members of the poorest SSAn countries as the latter have better access to obesogenic goods (i.e., ultra-high calorie meals and sedentary free time such as watching TV) and disproportionally live in cities and occupy the highest (non-physically intensive) job positions. 44,45,48,49 Notwithstanding the large number of positive and significant associations, around 40% of the associations examined are positive and non-significant (21/53). These non-significant associations among low-income countries in SSA temper previous observations.
Another corpus of studies argues that the so-called double burden of malnutrition simultaneously and disproportionally affects lowersocioeconomic groups in SSA, 3 even within the same household. 24 For example, in Burkina Faso, an adult without formal education and with a low income has a high risk of being affected by overweight or obesity, in addition to suffering from serious micronutrient deficiencies. 27 Therefore, we can assume that the coexistence of malnutrition and overweight within poor households, and within poor individuals, contributes to making the association between SES and bodyweight indicators unclear.
T A B L E 1 List of reviewed articles focusing on low-income countries in SSA. Note: "Nb" refers to the number of associations found in the sample of reviewed articles. "#" refers to the number of times a given bodyweight indicator is used in a given association.
"♀" refers to the number of associations focusing exclusively on female population. Bodyweight indicators are body mass index (BMI), waist circumference (WC), waist-to-height ratio (WHtR), waist-to-hip ratio (WHR), percentage of body fat (%BF), and sum of the three skinfolds (STS). References are listed in Table   S3.
Source: Authors. Note: "Nb" refers to the number of associations found in the sample of reviewed articles. "#" refers to the number of times a given bodyweight indicator is used in a given association.  Table   S3.
T A B L E 2 List of reviewed articles focusing on lower-middle-income countries in SSA.   Table   S3.

| Lower-middle-income countries
As shown in Table 2 the BMI classification), other country-specific cut-offs should be adopted in order to take ethnic variations and major differences in morphology into account. 47 In Asia for instance, a growing number of studies use country-specific cut-offs to classify individuals affected by overweight and obesity. In contrast, most of the studies focusing on SSA use the WHO-based universal BMI cut-offs, even though these cut-offs were initially defined from and for Caucasian populations. 54 Estimating optimal BMI cut-offs for the Ethiopian population, Sinaga et al. 55 argue that suitable overweight and obesity thresholds should be far below the commonly-used WHO-based universal thresholds. some higher middle-income countries around the world. 10,11,13,14 This social reversal could be driven by changes in environmental conditions (i.e., access to food), living conditions, weight perceptions, and antiobesity public policies. More specifically, given the higher exposure to thinness ideals and increased aversion to obesity-related health problems, it could be argued that the most affluent people tend to adopt strategies to avoid weight gain and related morbidity. In contrast, while the poorest value stoutness as a symbol of health and success, they do not have the economic resources or environmental conditions that allow them to eat sufficiently and thus to be exposed to weight gain. Although we did not find evidence of such earlier reversal in the context of lower-middle-income and low-income countries in SSA, the literature focusing on richer SSAn countries tends to confirm this idea. 14,51,58,59 Cohen et al. 60 give a convincing explanation of the ambivalent situation that African women face in cities. On the one hand, women from lower-socioeconomic groups (often emigrating from rural areas) gain substantial weight by eating high-calorie industrial food in order to feel part of and enjoy the Western lifestyle. On the other hand, women from upper-socioeconomic groups are more receptive to Western influences and the thinness ideals conveyed by education and the media and so increasingly reject stoutness. 61 Second, the influence of cultural factors on the relationship between BMI and SES should also be explored in more depth, taking into account the high cultural heterogeneity of African countries. which is associated in people's minds with social success and Western culture, and makes them feel part of the modern lifestyle. 60,64 The phenomenon of time-inconsistencies refers to irrational nutritional behavior among vulnerable socioeconomic groups because of a preference for present satisfaction. 65 Given their limited (and potentially unstable) income, middle social groups emerging from poverty may prefer to maximize their present satisfaction by consuming high-fat foods perceived as palatable (qualified as "short-term low-risk strategies") rather than invest in a future and uncertain health-based satisfaction through suitable food intake restrictions and regular physical exercise (qualified as "long-term high-risk strategies"). 66 In line with these assumptions, we found some studies emphasizing new food