J Urban Health (2022) 99:1091–1103 https://doi.org/10.1007/s11524-022-00689-5 ORIGINAL ARTICLE Are Women’s Empowerment and Income Inequality Associated with Excess Weight in Latin American Cities? Natalia Tumas · Santiago Rodríguez López · Mónica Mazariegos · Ana Ortigoza · Cecilia Anza Ramírez · Carolina Pérez Ferrer · Kari Moore · Goro Yamada · Mariana Carvalho Menezes · Olga L. Sarmiento · Juan M. Pericàs · Francesc Belvis Costes · Mariana Lazo · Joan Benach Published online: 10 November 2022 © The Author(s) 2022 Abstract While income gradients and gender ine- individuals (58.0% women), living in 538 sub-cities, qualities in excess weight have been noted elsewhere, 187 cities, and 8 countries. We used gender-stratified data from Latin American cities is lacking. We ana- Poisson multilevel models to estimate the Prevalence lyzed gender-specific associations between city-level Rate Ratios (PRR) for overweight/obesity (body mass women’s empowerment and income inequality with index ≥ 25 kg/m2) per a unit change in city-level wom- individual-level overweight/obesity, assessing how en’s empowerment (proxied by a score that measures these associations vary by individual education or liv- gender inequalities in employment and education) ing conditions within cities in Latin America. Data and income inequality (proxied by income-based came from national surveys and censuses, and was Gini coefficient). We also tested whether individual compiled by the SALURBAL project (Urban Health education or sub-city living conditions modified such in Latin America). The sample included 79,422 associations. Higher city labor women’s empower- ment (in women) and higher city Gini coefficient (in men) were associated with a lower prevalence of Mariana Lazo and Joan Benach are senior authors. overweight/obesity (PRR = 0.97 (95%CI 0.94, 0.99) Supplementary Information The online version and PRR = 0.94 (95%CI 0.90, 0.97), respectively). contains supplementary material available at https://​doi.​ The associations varied by individual education org/​10.​1007/​s11524-​022-​00689-5. N. Tumas (*) · J. M. Pericàs · F. Belvis Costes · J. Benach S. Rodríguez López Department of Political and Social Sciences, Research Facultad de Ciencias Exactas, Físicas y Naturales, Group on Health Inequalities, Environment, Employment Universidad Nacional de Córdoba, Córdoba, Argentina Conditions Knowledge Network (GREDS‑EMCONET), Universitat Pompeu Fabra, Barcelona, Spain M. Mazariegos e-mail: natalia.tumas@upf.edu INCAP Research Center for the Prevention of Chronic Diseases (CIIPEC), Institute of Nutrition of Central N. Tumas · J. M. Pericàs · F. Belvis Costes · J. Benach America and Panama (INCAP), Guatemala City, Johns Hopkins University ‑ Pompeu Fabra University Guatemala Public Policy Center (UPF‑BSM), Universitat Pompeu Fabra, Barcelona, Spain A. Ortigoza · K. Moore · G. Yamada · M. Lazo Urban Health Collaborative, Dornsife School of Public N. Tumas · S. Rodríguez López Health, Drexel University, Philadelphia, USA Centro de Investigaciones y Estudios sobre Cultura y Sociedad, Consejo Nacional de Investigaciones Científicas C. Anza Ramírez y Técnicas (CONICET) y Universidad Nacional de CRONICAS Centre of Excellence in Chronic Diseases, Córdoba, Córdoba, Argentina Universidad Peruana Cayetano Heredia, Lima, Perú Vol.: (0123456789) 13 1092 Tumas N. et al. and sub-city living conditions. For labor women’s gender inequalities in overweight/obesity have been empowerment, we observed weakened associations noted [7, 8]. In addition, obesity and social inequali- towards the null effect in women with lower educa- ties are higher in large urbanized areas [4, 9, 10]. tion and in residents of sub-cities with worse living Gender-based inequalities in overweight/obesity conditions (men and women). For the Gini coeffi- are an important concern in improving gender equity cient, the association was stronger among men with and population health [11]. Women’s empowerment primary education, and a negative association was can be defined as efforts to address gender inequali- observed in women with primary education. Our find- ties in education and employment which then lead ings highlight the need for promoting equity-based to improved access and use of resources that allow policies and interventions to tackle the high preva- women to develop their potential [12]. The role of lence of excess weight in Latin American cities. women’s empowerment to tackle malnutrition and improving nutritional outcomes has been reported Keywords Women’s empowerment · Income [12, 13], but the evidence on overweight/obesity is inequality · Overweight · Obesity · Cities · Latin inconclusive, and this relationship is not fully under- America stood. On the one hand, a study including several countries showed that women’s empowerment was associated with a lower obesity prevalence [14] and Introduction a higher dietary diversity among women in Ghana (but not with body mass index (BMI)) [15]. On the Excess weight (overweight/obesity), urbanization, other hand, evidence from global data showed that and social and gender inequalities are closely inter- increasing women’s empowerment appeared to be related growing global issues. Approximately half of associated with increases in BMI among women [16]. the world’s population lives in urban areas [1], and Women empowerment might reduce the social pres- about 39% of the world’s adult population was liv- sure to comply with body-image stereotypes, as well ing with overweight or obesity in 2015 [2]. In par- as reduce home cooking—traditionally carried out allel, social and gender inequalities are one of the by women—easing a greater consumption of ultra- most important determinants of population health processed high-calorie products [11, 17, 18]. How- [3–5]. Indeed, excess weight is more prevalent among ever, women’s empowerment might also contribute more unequal societies [6], and income gradients and to a lower probability of overweight/obesity through improvements in nutritional education [14] and lower interest in maternity, as pregnancies are associ- C. Pérez Ferrer National Institute of Public Health, Cuernavaca, Mexico ated with increments in adiposity in women [19]. A potential explanation for the conflicting results is that M. C. Menezes women’s empowerment is context-specific and its Universidade Federal de Ouro Preto, Ouro Preto, Brazil operationalization varies widely, making comparisons O. L. Sarmiento across studies challenging. School of Medicine, Universidad de los Andes, Bogotá, Evidence of the association between area-income Colombia inequality and excess weight is also inconsistent. Previous research reported that income inequality is J. M. Pericàs Liver Unit, Internal Medicine Department, Vall d’Hebron related to obesity, as inequality may determine obe- Institute for Research, Vall d’Hebron University Hospital, sogenic environments that influence individuals’ Barcelona, Spain choices towards less healthy dietary patterns and life- styles [4, 14], and the psychosocial impacts of living J. M. Pericàs Centro de Investigación Biomédica en Red de in a more hierarchical society might lead to excess Enfermedades Hepáticas y Digestivas (CIBERehd), weight [20]. However, some evidence showed that Madrid, Spain income inequality was also negatively associated with excess weight [21, 22], although the mechanisms J. Benach Ecological Humanities Research Group (GHECO), are less clear. Research from high-income countries Universidad Autónoma, Madrid, Spain showed that higher income inequality was positively Vol:. (1234567890) 13 Social and Gender Inequalities in Excess Weight in Latin America 1093 correlated with BMI or obesity [8, 20], but others social and gender inequalities in overweight/obesity. suggested an association with lower obesity in men To the best of our knowledge, no prior study has [23]. In low- and middle-income countries (LMICs), assessed the association between women’s empower- the evidence is more limited. Low-income inequality ment and income inequality with overweight/obesity was linked to faster increments of overweight rates in in Latin American cities. Thus, the aims of this study women from lower socioeconomic groups [21]. Par- are to analyze the association between (1) city-level ticularly in Latin America, it was described a negative women’s empowerment and (2) income inequality association with BMI and overweight/obesity among with overweight/obesity by gender and (3) to assess Mexican women [22]. how these associations vary by individual educa- The rising prevalence of overweight/obesity and tion or by the living conditions within cities in Latin related-risk behaviors has not been similar between America. We hypothesize that individuals living in and within regions and countries [24, 25]. In high- cities with greater women’s empowerment have a income countries, the prevalence began to attenu- lower prevalence of overweight/obesity compared to ate in the past decade, while rising trends are likely those from cities with lower women’s empowerment, to continue in LMICs, where more than half of peo- through improved nutritional education and reduced ple with obesity live [25, 26]. Furthermore, research interest in maternity. Furthermore, individuals living shows that as a country’s development increases, the in cities with greater income inequality have a higher burden of obesity shifts toward the lower socioeco- prevalence of overweight/obesity compared to those nomic groups, and that this pattern is more noticeable from less unequal cities, mainly due to the greater among women [25, 26]. Less is known about how exposure to obesogenic environments in highly une- social and gender inequalities shape the pattern of qual contexts. We expect these associations to be overweight/obesity between and within cities, espe- stronger among women. Last, these associations are cially in LMICs. Particularly in Latin America, excess modified by individual education and by the living weight affects approximately 50% of the adult popu- conditions within the cities. lation, and increasing trends are projected, mainly among women [27, 28]. Additionally, Latin America is among the most urbanized regions worldwide, Methods with fast-growing rates: the urban population grew from 40% in 1950 to 80% in 2015 [1]. At the same Sample time, this is one of the most unequal regions, includ- ing more than half of the most unequal cities [29, This cross-sectional study was developed using 30]. Gender inequalities also characterize the region: data compiled by the SALURBAL project (Urban poverty, difficulties in schooling, and low levels of Health in Latin America; “Salud Urbana en Amé- participation in the labor force, among others, affect rica Latina”) [36], a collaborative multinational women predominantly [31, 32]. In women, there are project that examines drivers of health in Latin large educational inequalities in obesity independent American cities of more than 100,000 inhabitants of city socioeconomic development, while in men, [30]. We analyzed harmonized data from health education gradients were observed only in cities with surveys and population censuses ranging from 2002 lower socioeconomic development [33]. Social con- to 2017 in: Argentina, Brazil, Chile, Colombia, El texts within Latin American cities are heterogeneous Salvador, Guatemala, Mexico, and Peru (for details [30, 34] as they vary in socioeconomic conditions, on the years of surveys and census in each country, the degree of poverty and social and gender inequali- see Table SI, Supplementary material). The ana- ties, ethnic composition, among others, and may play lytical sample included a total of 79,422 individuals a role in the social patterning of overweight/obesity (58.0% women), living in 187 cities and 538 sub- as well [33, 35]. cities (administrative units within cities). Sub-cities Given the concomitant features of high urbaniza- were defined as municipios, comunas, departamen- tion rates, high levels of social and gender inequali- tos, or similar units depending on the country. There ties, and high prevalence of overweight/obesity, Latin was a median of 708 individuals per city (interquar- America is a unique scenario for studying the urban tile range (IQR), 422–1,114), 474 individuals per Vol.: (0123456789) 13 1094 Tumas N. et al. sub-city (IQR, 135–819), and 5 sub-cities per city Effect Modifiers (IQR, 2–35). We evaluated individual education and sub-city living conditions as potential effect modifiers. Individual edu- Outcome cational attainment was categorized as: less than pri- mary, primary, secondary, and university and obtained We included excess weight (which encompasses from the same health survey data used to define the out- overweight/obesity, BMI ≥ 25 kg/m2) as the out- come. Sub-city living conditions were measured using a come at the individual level. For all countries except composite index based on census indicators of general Argentina, BMI was calculated based on standard- living conditions, including the percentage of house- ized measures of weight and height. For Argentina, holds with piped water inside the dwelling, the percent- we used self-reported weight and height, as the sur- age of households with overcrowding (3 or more people vey did not include measured weight and height. per room, inverted), and the percentage of the popula- tion aged 15–17 attending school [34]. Higher score values correspond to better living conditions. Exposures Covariates We included two exposures at the city level: labor women’s empowerment and the Gini coefficient. We adjusted the analyses by age (quadratic term, given In brief, labor women’s empowerment measures the non-linear association between age and overweight/ account for gender inequalities in education and obesity) [39], city population size (because previous stud- employment [37, 38]. We operationalized labor ies showed links between population size and overweight/ women’s empowerment by means of a score devel- obesity) [40], city GDP per capita (in order to test to what oped using national census data on employment extent labor women’s empowerment and Gini coefficient and education disaggregated by cities and sex and were associated with overweight/obesity beyond GDP of derived by principal component analysis [38]. The cities) [21], and country (in order to control for unmeasured data used to develop the score included: the ratio confounding effects). We stratified the analysis by gender of the female-to-male proportion of the popula- (using self-reported sex female/male), considering that the tion aged 25 or older who completed secondary effect of social indicators on overweight/obesity differed by education or above, the ratio of the female-to-male gender [33, 41]. For details on the study variables and lev- proportion of the population aged 25 or older who els of analysis, see Table SII, Supplementary material. completed university or above, the labor force par- ticipation rate among the female population 15 years Statistical Analysis or above, and the ratio of the labor force participa- tion rate among females to the labor force participa- We described the distribution of individual-, sub- tion rate among males (15 years or older). Higher city-, and city-level characteristics by presence or score values indicate greater participation of women absence of overweight/obesity. We estimated age- in the labor force and greater education of women adjusted percentages of overweight/obesity (through relative to men. We proxied income inequality by direct standardization, using the pooled sample age the income-based Gini coefficient, calculated based distribution) by quartiles of labor women’s empower- on census information or national household sur- ment and Gini coefficient and by gender. veys. The Gini coefficient is a well-known summary We estimated gender-specific associations between measure of the distribution of per-capita household overweight/obesity and the exposures of interest, income in the city, with higher values indicating using three-level Poisson models with robust vari- greater income inequality. We limited the sample to ance estimation and random intercepts for sub-cit- the cities with available data on income-based Gini ies and cities. We used two models: (i) the adjusted coefficient (187 of 232 of the SALURBAL cities model included individual age (quadratic), individual included in this study). education (categories), sub-city living conditions (z scores), city population size (z scores), city GDP Vol:. (1234567890) 13 Social and Gender Inequalities in Excess Weight in Latin America 1095 per-capita (z scores), labor women’s empowerment (z with overweight/obesity were older compared to scores) (adjusted model a)/Gini coefficient (z scores) those without this condition (Table 1). Overweight/ (adjusted model b)—separately—and countries (cat- obesity was more frequent in individuals with lower egories) as fixed effects; (ii) the interaction model educational level and in sub-cities with worse liv- included interaction terms (labor women’s empow- ing conditions. In addition, respondents with over- erment/Gini coefficient*individual education; labor weight/obesity were more likely to live in cities with women’s empowerment/Gini coefficient*sub-city higher GDP and with lower labor women’s empow- living conditions) and all the adjustment variables. erment. The city Gini coefficient was similar among Continuous variables were standardized in all mod- the individuals with and without overweight/obesity, els, using the pooled sample (men and women). The although differences were statistically significant. z-score has a mean of zero and a standard deviation Overweight/obesity was patterned by city-level of one and indicates how many standard deviations the labor women’s empowerment, with a lower propor- variable is from the mean. It is obtained by subtracting tion of overweight/obesity in cities with higher labor the mean from each observation and then dividing the women’s empowerment (more evident in women) difference by the standard deviation. Statistical analy- (Fig. 1). Patterns by the Gini coefficient were less sis was performed with Stata 15 [42] and R 4.1.0 [43]. clear, although the age-adjusted percentage of over- weight/obesity in men and women was lowest at the highest city income inequality quartile (Fig. 1). Results Table 2 shows the adjusted prevalence rate ratios (PRRs) of overweight/obesity associated with city A higher percentage of men (63.8%) had overweight/ labor women’s empowerment (adjusted model a) and obesity compared to women (60.4%), and individuals city Gini coefficient (adjusted model b) for women Table 1  Study characteristics by overweight/obesity status in 187 Latin American cities. SALURBAL Study Overall sample Overweight/obesity (BMI ≥ 25 kg/m2) Yes No p-value* (n = 79,422) (n = 49,085) (n = 30,337) Gender, % Women 58.0 60.4 39.6 < 0.01 Men 42.0 63.8 36.2 Age, mean (SD) 42.8 (16.9) 44.9 (15.9) 39.3 (17.8) < 0.01 Individual education, % University 13.4 12.4 15.0 < 0.01 Secondary 34.2 31.0 37.9 Primary 35.2 37.0 32.3 Less than primary 17.2 18.6 14.8 Sub-city living conditions, mean (SE) 0.72 (1.96) 0.59 (2.00) 0.91 (1.87) < 0.01 City labor women’s empowerment, mean (SD) 0.65 (3.03) 0.34 (3.09) 1.15 (2.85) < 0.01 City Gini coefficient, mean (SD) 0.49 (0.10) 0.49 (0.10) 0.50 (0.10) < 0.01 City GDP per-capita (USD, thousands), mean (SD) 18.2 (13.6) 18.7 (14.6) 17.5 (11.8) < 0.01 City population size (millions), mean (SD) 3.7 (5.6) 3.7 (5.7) 3.7 (5.6) 0.93 SD, standard deviation; GDP, gross domestic product. Sub-city living conditions range from − 9.92 to 3.90 with higher scores cor- responding to better conditions; labor women’s empowerment ranges from − 8.20 to 6.28 with higher scores corresponding to higher labor women’s empowerment; Gini coefficient ranges from 0.29 to 0.68 with higher scores corresponding to higher income inequal- ity. *p-values test the null hypothesis that there is no difference in the proportion (χ2) or mean (t-test) of the characteristic between individuals with and without overweight/obesity Vol.: (0123456789) 13 1096 Tumas N. et al. Fig. 1  Age-adjusted percentage of overweight/ obesity by quartiles of city. Gini coefficient and labor women’s empowerment and gender in 187 Latin American cities. SALUR- BAL study. 95% confidence intervals are shown in verti- cal lines and men separately. After adjusting for individual The results of the interaction analyses are shown characteristics, city GDP per-capita, city popula- in Figs. 2a and b and 3a and b. Among women, the tion size, and country, both higher city labor wom- association between labor women’s empowerment en’s empowerment (in women) and higher city Gini and overweight/obesity was different across individ- coefficient (in men) were associated with a lower ual education (global p-value for interaction < 0.01). prevalence of overweight/obesity (PRR = 0.97 (95% The protective association with increased labor CI 0.94, 0.99) and PRR = 0.94 (95% CI 0.90, 0.97), women’s empowerment was null for women with respectively). less than primary education (Fig. 2a). Similarly, the Table 2  Prevalence rate PRR (95% CI) ratios (95% confidence intervals) of overweight/ Women (n = 46,093) Adjusted model a Adjusted model b obesity associated with City labor women’s empowerment, z score 0.97 (0.94, 0.99) - city characteristics in 187 Latin American cities. City Gini coefficient, z score - 1.01 (0.98, 1.04) SALURBAL study Educational level (ref. University) 1.00 1.00 Secondary 1.16 (1.12, 1.20) 1.16 (1.12, 1.20) Primary 1.30 (1.25, 1.35) 1.30 (1.25, 1.35) Adjusted model a Less than primary 1.23 (1.19, 1.27) 1.23 (1.20, 1.27) includes labor women’s Sub-city living conditions, z score 0.98 (0.97, 1.00) 0.98 (0.97, 1.00) empowerment; adjusted model b includes Gini City GDP per capita (USD), z score 1.01 (1.00, 1.02) 1.01 (1.00, 1.02) coefficient; PRRs, City population size, z score 0.98 (0.97, 0.99) 0.98 (0.97, 0.99) prevalence rate ratios for Men (n = 33,329) Adjusted model a Adjusted model b each SD higher value City labor women’s empowerment, z score 0.98 (0.95, 1.01) - of the predictor; GDP, gross domestic product. Gini coefficient, z score - 0.94 (0.90, 0.97) Multilevel structure: Educational level (ref. University) 1.00 1.00 individuals nested within Secondary 1.00 (0.98, 1.03) 1.00 (0.98, 1.03) sub-cities, nested within Primary 0.94 (0.91, 0.97) 0.94 (0.91, 0.97) cities. Analyses are adjusted by age, individual Less than primary 0.88 (0.85, 0.91) 0.88 (0.85, 0.91) education, sub-city living Sub-city living conditions, z score 0.99 (0.97, 1.00) 0.99 (0.98, 1.01) conditions, city GDP per City GDP per capita (USD), z score 1.02 (1.01, 1.02) 1.02 (1.01, 1.02) capita, city population size, City population size, z score 0.99 (0.98, 1.00) 1.00 (0.99, 1.01) and country Vol:. (1234567890) 13 Social and Gender Inequalities in Excess Weight in Latin America 1097 Fig. 2  Adjusted asso- ciations between labor women’s empowerment and overweight/obesity, by a individual education and b sub-city living condi- tions in men and women in 187 Latin American cities. SALURBAL study. Significant global p-values for interactions: < 0.01 for individual education in women; < 0.05 for sub-city living conditions in men protective association with increased labor women’s associations were similar across tertiles of sub-city empowerment was weaker among women living in living conditions (Fig. 3b). sub-cities with worse living conditions (Fig. 2b). In men, a similar pattern was observed for sub-city liv- ing conditions (global p-value for interaction < 0.05) Discussion (Fig. 2b). The results of the interaction analyses of the association between the Gini coefficient and over- We analyzed associations between city-level income weight/obesity are shown in Fig. 3a and b. Among inequality and labor women’s empowerment with men, the negative association between the Gini coef- overweight/obesity by gender in a large sample of ficient and overweight/obesity was different across Latin American cities and examined how these asso- individual education (global p-value for interac- ciations vary by individual education or by sub-city tion < 0.05), being stronger among men with lower living conditions. We identified a lower prevalence levels of education, compared to those with univer- of overweight/obesity associated with higher city sity (Fig. 3a). In women, a negative association was labor women’s empowerment in women and with only observed among those with primary education higher city income inequality in men, after adjust- (global p-value for interaction < 0.01) (Fig. 3a). Last, ing for individual-, sub-city-, and city-level factors. Furthermore, the negative association between labor Vol.: (0123456789) 13 1098 Tumas N. et al. Fig. 3  Adjusted asso- ciations between Gini coefficient and overweight/ obesity, by a individual education and b sub-city living conditions in men and women in 187 Latin American cities. SALUR- BAL study. Significant global p-values for interac- tions: < 0.05 for individual education in men and < 0.01 in women women’s empowerment and overweight/obesity in an inverse association between women’s empower- women varied by individual education and by sub- ment and the prevalence of overweight/obesity, only city living conditions, with weakened associations among women. Consistent with our results, a study towards the null effect in those with the lowest educa- including almost 160 countries described a lower tional level and in residents of sub-cities with worse prevalence of obesity associated with higher women’s living conditions. A similar pattern was observed empowerment [14]. Contrary to our findings, another for sub-city living conditions in men. We also found study including 190 countries showed that women’s that individual education modified the association empowerment was associated with increases in BMI between income inequality and overweight/obesity over time among women [18], and a recent study in men, such that the association was stronger among including eastern African countries also reported a those with primary education. In women, a negative positive association with women’s BMI [16]. The association was observed among those with primary discrepancy between these findings could be partially education. explained by other broader regional and country con- While there is a relative consensus that women’s textual factors related to what women’s empowerment empowerment might prevent undernutrition [12], implies for factors linked to overweight/obesity (such the evidence is scarce and less consistent for excess as diet and physical activity) and also to differences weight. Aligned with our first hypothesis, we found in social and cultural norms associated with women Vol:. (1234567890) 13 Social and Gender Inequalities in Excess Weight in Latin America 1099 being more empowered. Also, the different meth- consumption of high-calorie ultra-processed products odological approaches and the scales of the women’s [17]. It should be noted that the dominant models of empowerment measures used in each case might con- production, the shifts in the agricultural system, and tribute to the divergence in the results [7]. Indeed, the change in technology, mass media, and urbaniza- there are important gaps and heterogeneities in the tion are important drivers of the nutrition transitions measurement of women’s empowerment, and the [52]. Another proposed mechanism suggests that application of existing measures is still limited [12]. women’s empowerment might reduce the social pres- We are not aware of any previous research studying sure that traditionally faces women to align to body the relationship between women’s empowerment and size preferences and patterns [11]. excess weight in Latin America, but prior evidence in We also found an inverse association between cities of the region showed that women’s labor force city income inequality and the prevalence of over- participation (as a proxy of women’s empowerment) weight/obesity in men, contrary to what we hypoth- was negatively associated with infant mortality rate, esized. Similar to our study, greater income inequality another key population health indicator [38]. was linked to lower obesity rates across counties of Several mechanisms might underlie the inverse New York only in men [23]. Evidence from Mexico associations between women’s empowerment and also revealed an inverse association with BMI and overweight/obesity. On the one hand, an indirect path- overweight/obesity among women [22]. Moreover, way through employment and higher income, as well Jones-Smith et al. [21] in a sample of women from as a greater education level, that accompany women’s 37 LMICs found that lower income inequality was empowerment [11]. Growing evidence has docu- associated with faster increments of overweight in mented that empowerment leads to better employ- the lower wealth groups from countries with a higher ment conditions, education, political participation, GDP. In contrast, in a study including 127 low-, mid- and health [11, 44] and to improved engagement in dle-, and high-income countries, De Vogli et al. [53] the prevention of non-communicable diseases, includ- found a positive association between the Gini coef- ing weight loss [11, 45, 46]. The smoking habit might ficient and BMI. In another study using country- also play a role, given its association with women aggregated data of OECD countries, higher income empowerment [47] and weight loss [48]. In addi- inequality was linked to a faster increase in obesity tion, lower fertility rates at higher levels of education prevalence [54]. Evidence from high-income coun- might influence these associations, as pregnancy‐ tries showed a positive association between income related weight gain has emerged as a potential cause inequality and the percentage of obesity [20] and of excess adiposity [49]. The latter could be of impor- BMI [8]. Relative to this, it was suggested that high- tance in Latin America because of the elevated ado- income inequality leads to increases in most of the lescent fertility, which is strongly linked with long- health conditions associated with low social posi- term increments in adiposity [19]. We hypothesize tion, including overweight/obesity [55]. Inequality that women’s empowerment might decrease over- is a scale-specific contextual indicator and it is dif- weight/obesity indirectly by reducing the interest in ferently related to social conditions. Thus, different maternity and therefore cumulative weight gain asso- geographic areas considered could contribute to dif- ciated with parity. However, we did not have informa- ferences in the associations [23]. tion about parity to further explore this hypothesis in The evidence about the gender-specific impacts our study. On the other hand, research suggested that of income inequality on overweight/obesity and the socioeconomic development and women’s changing underlying mechanisms is still scarce. The existing roles in societies may affect the excess weight under- literature showed that income inequality has detri- lying behavioral mechanisms [18]. The “moderniza- mental effects on social cohesion, and also that fast- tion theory” states that socioeconomic development food outing could be considered a proxy of social leads to domestic nutrition transitions (from low to capital [22, 56]. Prior research argued that since men high caloric diet) and relates the greater unhealthy tended to have greater non-kin ties in their personal food intake to increasing income and higher partici- networks than women [57], significant effects of pation of women in the labor force [18, 50, 51]. The income inequality could be observed in men but not latter usually implies less homemade food and greater in women. In addition, Prince et al. [58] found higher Vol.: (0123456789) 13 1100 Tumas N. et al. probability of overweight/obesity associated with Our findings have policy implications. We pro- a stronger sense of community belonging in men, vide evidence suggesting that increasing education but no significant impact was evidenced on women. and labor force participation among women might be Based on the above, we hypothesize that in contexts relevant for addressing overweight/obesity in Latin of high inequality, men could be more vulnerable America. This should be acknowledged together than women to the impact of lack of social cohesion with policies aimed to promote healthier lifestyles. on outing unhealthy food intake. However, the path- Public policies and interventions at the population ways to explain the gender differences in these asso- level should be oriented to improve living and work- ciations are unclear and require further research. ing environments (e.g., through improved access to Some explanations have been proposed for the link healthy food) and promote healthy macro policies between higher income inequality and a lower preva- (regulations and fiscal measures, such as taxes on lence of overweight/obesity. Clement et al. [22] dem- high-fat and high-sugar food). Also, interventions onstrated that the main pathways proposed to explain targeting less educated groups and focusing on socio- the consequences of income inequality on health economically disadvantaged areas of cities are recom- outcomes [54] also explain the negative associations mended to reduce disparities in overweight/obesity. between income inequality and excess weight. They Strengths of our study include the use of a multi- showed that income inequality might limit weight level harmonized database of national health surveys gain through a worse quality of sleep (psychological combined with contextual data on city social character- pathway), less fast-food consumption (social capital istics, accounting for the heterogeneity within cities in pathway), and limited access to piped water—through several Latin American countries. To our knowledge, weight loss associated with increased occurrence of this is the first study considering women’s empower- infectious diseases and increased physical activity to ment and income inequality indicators at the city level, fetch and carry water (public policy pathway). Addi- coupled with information about the sub-city social tional research is needed to elucidate the mechanisms features, providing a singular perspective on the inter- underlying these associations. play across urban social gradients in excess weight in Overall, we also found that the associations Latin America. Overall, our study sheds light on the between overweight/obesity and women’s empower- linkages between important issues that, to date, remain ment and income inequality were modified by indi- scarcely studied. Future work should further explore vidual education and to a lesser extent by sub-cities the mechanisms behind the associations we observed, living conditions, which is most aligned with our and longitudinal research will also contribute to a bet- third hypothesis. Specifically, women with less than ter understanding of the social and gendered patterning primary education and individuals living in sub-cities of overweight/obesity in Latin America. Our study has with worse living conditions were less or not ben- some limitations. We acknowledge that the score used efited by the potential protective effect of women’s to proxy women’s empowerment was based on census empowerment on overweight/obesity. These findings data and built upon employment and education indi- could be linked to the relative deprivation hypothesis, cators, not reflecting other dimensions of the empow- which posits that better area-level social conditions erment of women such as decision-making power and may not imply garnering health benefits and even gender-based violence, among others [38, 61]. We did could be unfavorable for those individuals with lower not include in the analysis the total fertility rate due to social positions [55]. We also found that men with a lack of data; however, prior research has argued that primary education living in cities with high levels of obesity due to multiparity in women is, at least in part, income inequality seemed to have even lower preva- driven by the social context in which the reproductive lence of overweight/obesity. In line with that, Bjorn- rights of women are situated [7]. Prioritizing compari- strom [59] and Fan et al. [60] in the USA showed that sons across countries, we assumed that women from 15 the lowest social position groups from unequal locali- to 18 years within the work market would represent a ties can incorporate some health-related behaviors higher women’s empowerment, which is arguable. Nev- from the high social position groups, leading to better ertheless, we expect that the number of women in the nutritional outcomes for the former. workforce within this age range would be rather low. The countries in our study are heterogeneous in the Vol:. (1234567890) 13 Social and Gender Inequalities in Excess Weight in Latin America 1101 number of native ethnic communities in cities, which https://​drexel.​edu/​lac/​data-​evide​nce for a complete list of data may have a higher risk of overweight/obesity because, sources. The findings of this study and their interpretation are the responsibility of the authors and do not represent the views usually, they are socially excluded, have lower income, or interpretations of the institutions or groups that compiled, and are less empowered [62]. We did not account for collected, or provided the data. The use of data from these insti- this heterogeneity within cities, which may play a role tutions does not claim or imply that they have participated in, in our findings. Furthermore, city characteristics were approved, endorsed, or otherwise supported the development of this publication. They are not liable for any errors, omissions or retrieved from censuses developed at different years other defect, or for any actions taken in reliance thereon. The in the countries, frequently not aligned with the health funder of the study had no role in the study design, data collec- survey years from which we obtained the individual tion, data analysis, data interpretation, or writing of this article. data. This forced us to assume that city characteristics were relatively stable across the years examined. Data Author Contribution NT, ML, and JB conceived the study. NT, SRL, and GY did the statistical analyses. NT, SRL, ML, on income inequality (income-based Gini coefficient) and JB contributed to drafting the first version of the manu- was not available for all cities, and, e.g., only a single script. ML, KM, AO, and GY participated in or supported data city for Guatemala and Peru could be included. Other collection. All authors participated in the interpretation of the limitations were the cross-sectional design, which pre- results, reviewed critically the manuscript, and approved the final version. vented us from assuming any type of causality, and the use of self-reported weight and height in Argentina, Funding Open Access funding provided thanks to the although evidence suggested good agreement between CRUE-CSIC agreement with Springer Nature. objective and self-reported weight and height [63]. Data Availability When the data can be made public with- out violating confidentiality—once the SALURBAL project ends—it will be placed in a public repository as required by the Conclusion funder (Wellcome Trust). In summary, we found that the prevalence of over- Declarations weight/obesity was negatively associated with wom- en’s empowerment in women and income inequality Ethics Approval and Consent to Participate The SALUR- in men. Our results also suggested that city inequalities BAL study protocol was approved by the Drexel University Institutional Review Board with ID #1612005035 and by appro- differ across educational backgrounds and—to a lesser priate site-specific IRBs. extent—across the sub-cities living conditions. Over- all, these findings highlight the need for promoting Open Access This article is licensed under a Creative Com- equity-based policies and interventions to tackle excess mons Attribution 4.0 International License, which permits weight in Latin America. Improving women’s empow- use, sharing, adaptation, distribution and reproduction in any erment is especially recommended to address the gen- medium or format, as long as you give appropriate credit to the dered overweight/obesity pandemic in the region. original author(s) and the source, provide a link to the Crea- tive Commons licence, and indicate if changes were made. The images or other third party material in this article are included Acknowledgements The authors acknowledge the contribu- in the article’s Creative Commons licence, unless indicated tion of all SALURBAL project team members. For more infor- otherwise in a credit line to the material. If material is not mation on SALURBAL and to see a fullist of investigators, included in the article’s Creative Commons licence and your see https://​drexel.​edu/​lac/​salur​bal/​team/. The Salud Urbana en intended use is not permitted by statutory regulation or exceeds América Latina (SALURBAL)/Urban Health in Latin America the permitted use, you will need to obtain permission directly project is funded by the Wellcome Trust (205177/Z/16/Z). For from the copyright holder. To view a copy of this licence, visit the purpose of open access, the author has applied a CC-BY http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. public copyright license to any Author Accepted Manuscript version arising from this submission. This work received funding from the European Union’s Horizon 2020 research and innovation program under the Marie Sklodowska-Curie grant agreement Nº 89102 (Dr. Tumas was supported by this References grant). Joan Benach gratefully acknowledges the financial support by ICREA under the ICREA Academia programme. 1. Population Division of the Department of Economic and SALURBAL acknowledges the contributions of many differ- Social Affairs. In: Revision of world population prospects. ent agencies in generating, processing, facilitating access to New York: United Nations; 2017. data, or assisting with other aspects of the project. Please visit Vol.: (0123456789) 13 1102 Tumas N. et al. 2. Chooi YC, Ding C, Magkos F. The epidemiology of obe- 20. Pickett KE, Kelly S, Brunner E, Lobstein T, Wilkinson sity. Metabolism: clinical and experimental. W.B. Saun- RG. Wider income gaps, wider waistbands? An ecologi- ders; 2019;92:6–10. cal study of obesity and income inequality. J Epidemiol 3. Barreto ML. Health inequalities: a global perspective. Community Health. 2005;59:670–4. Cien Saude Colet. 2017;22:2097–108. 21. Jones-Smith JC, Gordon-Larsen P, Siddiqi A, Pop- 4. Wilkinson R, Pickett K. The spirit level: why equality is kin BM. Cross-national comparisons of time trends in better for everyone. London, England: Penguin; 2010. overweight inequality by socioeconomic status among 5. Benach J, Vergara M, Muntaner C. Desigualdad en salud: women using repeated cross-sectional surveys from la mayor epidemia del siglo XXI. Papeles de Relaciones 37 developing countries, 1989–2007. Am J Epidemiol. Ecosociales y Cambio Global. 2008;13:29–40. 2011;173:667–75. 6. Eckersley R. Beyond inequality: acknowledging the com- 22. Clément M, Levasseur P, Seetahul S, Piaser L. Does ine- plexity of social determinants of health. Soc Sci Med. quality have a silver lining? Municipal income inequality 2015;147:121–5. and obesity in Mexico. Soc Sci Med. 2021;(1982):272. 7. Garawi F, Devries K, Thorogood N, Uauy R. Global dif- 23. Kim D, Wang F, Arcan C. Geographic association ferences between women and men in the prevalence of between income inequality and obesity among adults in obesity: is there an association with gender inequality? New York state. Prev Chronic Dis. 2018;15:E123. Eur J Clin Nutr. 2014;68:1101–6. 24. Allen L, Williams J, Townsend N, Mikkelsen B, Rob- 8. Diez-Roux AV, Link BG, Northridge ME. A multilevel erts N, Foster C, et al. Socioeconomic status and non- analysis of income inequality and cardiovascular disease communicable disease behavioural risk factors in risk factors. Soc Sci Med. 2000;50:673–87. low-income and lower-middle-income countries: a sys- 9. Jiwani SS, Carrillo-Larco RM, Hernández-Vásquez A, tematic review. Lancet Glob Health. 2017;5:e277–89. Barrientos-Gutiérrez T, Basto-Abreu A, Gutierrez L, 25. Ng M, Fleming T, Robinson M, Thomson B, Graetz et al. The shift of obesity burden by socioeconomic sta- N, Margono C, et al. Global, regional, and national tus between 1998 and 2017 in Latin America and the prevalence of overweight and obesity in children and Caribbean: a cross-sectional series study. Lancet Glob adults during 1980–2013: a systematic analysis for Health Elsevier Ltd. 2019;7:e1644–54. the Global Burden of Disease Study 2013. Lancet. 10. Liddle B. Urbanization and inequality/poverty. Urban 2014;384:766–81. Science. 2017;1:35. 26. Mitchell S, Shaw D. The worldwide epidemic of 11. Costa-Font J, Györi M. The weight of patriar- female obesity. Best Pract Res Clin Obstet Gynaecol. chy? Gender obesity gaps in the Middle East and 2015;29:289–99. North Africa (MENA). London, England. Soc Sci 27. Popkin BM, Reardon T. Obesity and the food sys- Med. 2020;(1982):266. tem transformation in Latin America. Obes Rev. 12. Madzorera I, Fawzi W. Women empowerment is central 2018;19:1028–64. to addressing the double burden of malnutrition. Lon- 28. Aschner P. Obesity in Latin America. In: Ahima R, edi- don, England. EClinicalMedicine. 2020;20:100286. tor. Metabolic syndrome. 1st ed. Springer International 13. Wells JC, Sawaya AL, Wibaek R, Mwangome M, Poul- Springer International Publishing; 2016. https://​doi.​org/​ las MS, Yajnik CS, et al. The double burden of mal- 10.​1007/​978-3-​319-​11251-0_4. nutrition: aetiological pathways and consequences for 29. Geraldi J. UN HABITAT, State of the world’s cities health. Lancet. 2020;395(10217):75–88. 2010/2011: cities for all, bridging the urban divide 14. de Soysa I, Lewin OL. Gender empowerment, inequali- [Internet]. London: UN-HABITAT; 2010. Available ties and the prevalence of adult female obesity: an from: https://​unhab​itat.​org/​state-​of-​the-​worlds-​cities-​ empirical analysis using new data, 1990–2013. Scand J 20102 ​ 0 11- ​ c ities- ​ for- ​ a ll- ​ b ridg ​ i ng- ​ t he- ​ u rban- ​ d ivide. Public Health. 2019;47:796–807. Accessed 22 Apr 2022. 15. Malapit HJL, Quisumbing AR. What dimensions of 30. Diez Roux A v., Slesinski SC, Alazraqui M, Caiaffa WT, women’s empowerment in agriculture matter for nutri- Frenz P, Jordán Fuchs R, et al. A novel international part- tion in Ghana? Food Policy. 2015;52:54–63. nership for actionable evidence on urban health in Latin 16. Jones RE, Haardörfer R, Ramakrishnan U, Yount KM, America: LAC‐urban health and SALURBAL. Global Miedema SS, Roach TD, et al. Intrinsic and instrumen- Chall. 2019;3:1800013. tal agency associated with nutritional status of East 31. Fisher B, Naidoo R. The geography of gender inequality. African women. Soc Sci Med. 2020;247:112803. PLoS One. 2016;11:e0145778. 17. Chou SY, Grossman M, Saffer H. An economic analysis 32. Duryea S, Galiani S, Nopo H, Piras C. The education gender gap of adult obesity: results from the behavioral risk factor in Latin America and the Caribbean. In: Inter-American Devel- surveillance system. J Health Econ. 2004;23:565–87. opment Bank Research Department Working Paper 600, Wash- 18. Fox A, Feng W, Asal V. What is driving global obesity ington, DC. 2007. trends? Globalization or “modernization”? Glob Health. 33. Mazariegos M, Auchincloss AH, Braverman-Bronstein A, 2019;15:15–32. Kroker-Lobos MF, Ramírez-Zea M, Hessel P, et al. Edu- 19. Câmara SMA, McGurk MD, Gigante D, Lima MDA, cational inequalities in obesity: a multilevel analysis of Shalaby AK, Sentell T, et al. Intersections between ado- survey data from cities in Latin America. Public Health lescent fertility and obesity—pathways and research Nutr. 2021;25(7):1–9. gaps focusing on Latin American populations. Ann N Y 34. Ortigoza AF, Tapia Granados JA, Jaime Miranda J, Alaz- Acad Sci. 2022. https://​doi.​org/​10.​1111/​nyas.​14854. raqui M, Higuera D, Villamonte G, et al. Characterising Vol:. (1234567890) 13 Social and Gender Inequalities in Excess Weight in Latin America 1103 variability and predictors of infant mortality in urban set- 49. Mazariegos M, Ortiz-Panozo E, González de Cosío T, tings: findings from 286 Latin American cities. J Epide- Lajous M, López-Ridaura R. Parity, lactation, and long- miol Community Health. 2021;75:264–70. term weight change in Mexican women. Matern Child 35. Tumas N, Rodríguez López S, Bilal U, Ortigoza A, Diez Nutr. 2020;16(3):e12988. Roux AV. Urban social determinants of non-communi- 50. Popkin BM. Urbanization, lifestyle changes and the nutri- cable diseases risk factors in Argentina. Health Place. tion transition. World Dev. 1999;27:1905–16. 2022;77:102611. 51. Popkin BM. The world is fat: the fads, trends, policies, 36. Quistberg DA, Diez Roux AV, Bilal U, Moore K, Ortigoza and products that are fattening the human race. New A, Rodriguez DA, et al. Building a data platform for York, NY: Penguin Books; 2009. cross-country urban health studies: the SALURBAL 52. Popkin BM. Nutrition transition and the global diabetes study. J Urban Health. 2019;96(2):311–37. epidemic. Curr Diab Rep. 2015;15(9):64. 37. Chant S. Cities through a “gender lens”: a golden “urban 53. de Vogli R, Kouvonen A, Elovainio M, Marmot M. Eco- age” for women in the global south? Environ Urban. nomic globalization, inequality and body mass index: 2013;25:9–29. a cross-national analysis of 127 countries. Crit Public 38. Ortigoza A, Braverman A, Hessel P, di Cecco V, Friche Health. 2014;24:7–21. AA, Teixeira Caiaffa W, et al. Women’s empowerment 54. Su D, Esqueda OA, Li L, Pagán JA. Income inequality and infant mortality in Latin America: evidence from 286 and obesity prevalence among OECD countries. J Biosoc cities. Cities Health. 2021;1–9. https://​doi.​org/​10.​1080/​ Sci. 2012;44:417–32. 23748​834.​2021.​19087​94. 55. Pickett KE, Wilkinson RG. Income inequality and health: 39. Frasca D, Blomberg BB, Paganelli R. Aging, obesity, a causal review. Soc Sci Med. 2015;128:316–26. and inflammatory age-related diseases. Front Immunol. 56. Leyden KM. Social capital and the built environment: the 2017;8:1745. importance of walkable neighborhoods. Am Public Health 40. Flies EJ, Mavoa S, Zosky GR, Mantzioris E, Williams C, Assoc. 2011;93(9):1546–155. Eri R, et al. Urban-associated diseases: candidate diseases, 57. Zheng H, Rising US. income inequality, gender and indi- environmental risk factors, and a path forward. Environ vidual self-rated health, 1972–2004. Soc Sci Med Perga- Int 2019;133:105187. mon. 2009;69:1333–42. 41. Ameye H, Swinnen J. Obesity, income and gender: 58. Prince SA, Kristjansson EA, Russell K, Billette JM, the changing global relationship. Global Food Secur. Sawada M, Ali A, et al. A multilevel analysis of neigh- 2019;23:267–81. bourhood built and social environments and adult 42. StataCorp. Stata statistical software: release 15. Stata- self-reported physical activity and body mass index Corp LLC, College Station, TX. 2017. in Ottawa, Canada. Int J Environ Res Public Health. 43. R Core Team. R: a language and environment for statis- 2011;8(10):3953–78. tical computing. The R Foundation; 2020. https://​www.r-​ 59. Bjornstrom EES. An examination of the relation- proje​ct.​org/. Accessed 8 Feb 2021. ship between neighborhood income inequality, social 44. World Development Report 2012. Gender equality and resources, and obesity in Los Angeles county. Am J development. World Bank, Washington D.C.; 2011. Health Promot. 2011;26:109–15. 45. Tucker CM, Lopez MT, Campbell K, Marsiske M, Daly 60. Fan JX, Wen M, Kowaleski-Jones L. Tract- and county- K, Nghiem K, et al. The effects of a culturally sensitive, level income inequality and individual risk of obesity in empowerment-focused, community-based health promo- the United States. Soc Sci Res. 2016;55:75–82. tion program on health outcomes of adults with type 2 61. Buvinic M, et al. Measuring women’s economic empow- diabetes. J Health Care Poor Underserved. 2014;25:292. erment. In: A compendium of selected tools. Washington, 46. Wong CKH, Wong WCW, Wan YF, Chan AKC, Chan DC: Data2X & Center For Global Development; 2020. FWK, Lam CLK. Effect of a structured diabetes educa- 62. Ingaramo RA. Obesity, diabetes, and other cardiovascular tion programme in primary care on hospitalizations and risk factors in native populations of South America. Curr emergency department visits among people with type 2 Hypertens Rep. 2016;18(1):9. diabetes mellitus: results from the Patient Empowerment 63. Hodge JM, Shah R, McCullough ML, Gapstur SM, Program. Diabet Med. 2016;33:1427–36. Patel AV. Validation of self-reported height and weight 47. Hitchman SC, Fong GT. Gender empowerment and in a large, nationwide cohort of U.S. adults. PLoS One. female-to-male smoking prevalence ratios. Bull World 2020;15(4):e0231229. Health Organ. 2011;89:195–202. 48. Piirtola M, Jelenkovic A, Latvala A, Sund R, Honda C, Publisher’s Note Springer Nature remains neutral with regard Inui F, et al. Association of current and former smoking to jurisdictional claims in published maps and institutional with body mass index: a study of smoking discordant twin affiliations. pairs from 21 twin cohorts. PLoS One. 2018;13:e0200140. Vol.: (0123456789) 13