Global differences between women and men in the prevalence of obesity:
Is there an association with genderinequality?
FGarawi, K Devries, N Thorogood, R Uauy
Supplementary Information
Table S1. Sample Characteristics
Variable / N / Mean / Std. Dev. / Median / IQR / Min / MaxOutcome
Sex difference in obesity (F-M, %) / 151 / 8.89 / 9.40 / 6.45 / 12.00 / -4.14 / 40.64
Male obese (%) / 151 / 12.40 / 12.77 / 9.82 / 16.55 / 0.02 / 84.59
Female obese (%) / 151 / 21.29 / 16.47 / 18.00 / 22.73 / 0.03 / 80.46
Explanatory variables(1)
GII 2010 / 122 / 0.53 / 0.18 / 0.56 / 0.34 / 0.17 / 0.85
GGG 2010 / 119 / 0.68 / 0.06 / 0.68 / 0.07 / 0.49 / 0.85
SIGI 2009 / 85 / 0.13 / 0.11 / 0.11 / 0.18 / 0.003 / 0.58
Gini 2010 / 123 / 40.10 / 9.08 / 39.20 / 11.2 / 16.80 / 74.30
GNI per capita (ln) / 145 / 8.83 / 1.28 / 9.02 / 2.24 / 5.17 / 10.98
Urbanization (%) / 149 / 56.21 / 23.70 / 58.24 / 38.90 / 12.53 / 100.00
Total Fertility Rate / 145 / 2.65 / 1.28 / 2.20 / 1.41 / 1.01 / 6.86
Total Energy (Availability, kcal/day) / 138 / 2734.59 / 511.86 / 2745.50 / 851.00 / 1631.00 / 3799.00
Female - Male smoking difference / 123 / -20.01 / 14.25 / -18 / 19 / -58.00 / 8.00
(1)In these summary measures we only include those countries for which obesity data was available
Figure S1: Vast sex differences in obesity prevalence by region
Each boxplot summarizes the distribution of the female-male differences in the prevalence of obesity in different regions of the world. Median is indicated by line in middle of box. Values below 0 indicate prevalence is higher among males. While the median difference is less than 5% for North America, Western Europe, and S/SE Asia regions, it is greater than 10% for the Middle East and North Africa (MENA), Latin America and the Caribbean, and the Pacific regions. The outliers above sub-Saharan Africa and South Asia are countries in which the mean prevalence of obesity was higher relative to most of the other countries within the respective region (more so in the case of Maldives).
Tables S2-S4. The following 3 tables present the complete models of the multiple regression analyses that examine the association between sex differences in obesity and gender inequality adjusting for various factors.
Across models the estimated effect on the outcome was larger for GII than GGG. After accounting for wealth (GNI), within-country income inequality (Gini), and other factors associated with obesity, this association attenuated more for the GGG than for the GII. The smoking variable was negatively associated with the sex differences in obesity meaning that as the sex gap in smoking narrows (i.e., less difference between men and women smoking prevalence), the sex disparities in obesity decrease. The association between smoking and body weight has been previously documented.1 Smoking has appetite-suppressing effects,2 and thus smokersmay consume less energy (and those who do not smoke or have recently quit may consume more.) In much of the developingworld there is still strong social disapproval of female smoking, and rates are much lower for females than males, and thus itcould be that sex differences in smoking rates proxy other aspects ofgender inequality.3
- Gruber, J. and M. Frakes, Does falling smoking lead to rising obesity? Journal ofHealth Economics, 2006. 25(2): p. 183-197.
- Seeley, R.J. and D.A. Sandoval, Neuroscience: Weight loss through smoking.Nature, 2011. 475(7355): p. 176-177.
- Hitchman, S.C. and G.T. Fong, Gender empowerment and female-to-male smokingprevalence ratios. Bull World Health Organ, 2011. 89(3): p. 195-202.
Table S2. Complete results of multiple regression model using Gender Inequality Index (GII) as exposure.a
Model 1 / Model 2 / Model 3 / Model 4Mean Prev. / b / 0.409 / 0.405 / 0.382 / 0.376
95% CI / (0.238, 0.580)** / (0.223, 0.587)** / (0.193, 0.571)** / (0.181, 0.570)**
P-value / [0.00] / [0.00] / [0.00] / [0.00]
GII / 4.323 / 3.836 / 5.170 / 3.825
(3.126, 5.520)** / (0.734, 6.939)* / (2.597, 7.743)** / (0.644, 7.007)*
[0.00] / [0.02] / [0.00] / [0.02]
Gini / 0.050 / 0.122
(-0.118, 0.218) / (-0.049, 0.293)
[0.56] / [0.16]
GNI (ln) / -0.228 / -1.059
(-3.030, 2.575) / (-3.435, 1.317)
[0.87] / [0.38]
Smoking / -0.129 / -0.141
(-0.237, -0.021)* / (-0.249, -0.034)*
[0.02] / [0.01]
Urbanization / 0.031 / 0.037
(-0.062, 0.124) / (-0.063, 0.137)
[0.51] / [0.46]
TFR / -0.863 / -0.975
(-2.687, 0.961) / (-2.773, 0.822)
[0.35] / [0.28]
TEA / 0.084 / 0.147
(-0.385, 0.553) / (-0.345, 0.640)
[0.72] / [0.55]
Adjusted R2 / 0.38 / 0.37 / 0.44 / 0.44
N / 89 / 89 / 89 / 89
* p<0.05; ** p<0.01
aIn all 4 models, the outcome is sex differences in obesity prevalence, computed as female – male prevalence; Mean prevalence is the simple mean of female and male prevalence; GII – Gender Inequality Index; Gini – Coefficient of income inequality; GNI (ln) – Log transformed gross national income; Smoking – difference between female and male prevalence expressed as percentage; Urbanization – level of urbanization expressed as percentage; TFR – Total Fertility Rate; TEA – Total Energy Availability. N – sample size. Model 1 is unadjusted, Model 2 adjusts for GNI and Gini coefficient, Model 3 adjusts for smoking, total fertility rate, urbanization, and total energy availability. Model 4 includes all variables (full model). We report regression with robust standard errors. For the primary exposure, GII, we report standardized regression coefficient.
Table S3. Complete results of multiple regression model using Global Gender Gap (GGG) as exposure.a
Model 1 / Model 2 / Model 3 / Model 4Mean Prev. / b / 0.358 / 0.422 / 0.477 / 0.409
95% CI / (0.211, 0.505)** / (0.241, 0.604)** / (0.276, 0.678)** / (0.206, 0.612)**
P-value / [0.00] / [0.00] / [0.00] / [0.00]
GGG / -3.136 / -2.013 / -1.884 / -1.663
(-4.351, -1.920)** / (-3.417, -0.609)** / (-3.360, -0.407)* / (-2.957, -0.369)*
[0.00] / [0.01] / [0.01] / [0.01]
Gini / 0.183 / 0.218
(-0.004 - 0.370) / (0.038, 0.398)*
[0.06] / [0.02]
GNI (ln) / -1.529 / -2.180
(-2.918, -0.140)* / (-3.959, -0.401)*
[0.03] / [0.02]
Smoking / -0.167 / -0.149
(-0.280, -0.054)** / (-0.252, -0.046)**
[0.00] / [0.01]
Urbanization / -0.014 / 0.030
(-0.111, 0.083) / (-0.078, 0.139)
[0.78] / [0.58]
TFR / 0.654 / -0.632
(-0.863, 2.171) / (-2.142, 0.877)
[0.39] / [0.41]
TEA / -0.236 / 0.038
(-0.720, 0.248) / (-0.455, 0.532)
[0.34] / [0.88]
Adjusted R2 / 0.28 / 0.37 / 0.38 / 0.44
N / 90 / 90 / 90 / 90
* p<0.05; ** p<0.01
aIn all 4 models, the outcome is sex differences in obesity prevalence, computed as female – male prevalence; Mean prevalence is the simple mean of female and male prevalence; GGG – Global Gender Gap Index; Gini – Coefficient of income inequality; GNI (ln) – Log transformed gross national income; Smoking – difference between female and male prevalence expressed as percentage; Urbanization – level of urbanization expressed as percentage; TFR – Total Fertility Rate; TEA – Total Energy Availability. N – sample size. Model 1 is unadjusted, Model 2 adjusts for GNI and Gini coefficient, Model 3 adjusts for smoking, total fertility rate, urbanization, and total energy availability. Model 4 includes all variables (full model).We report regression with robust standard errors. For the primary exposure, GGG, we report standardized regression coefficient.
Table S4. Complete results of multiple regression model using Social Institutions and Gender Index (SIGI)a
Model 1 / Model 2 / Model 3 / Model 4Mean Prev. / b / 0.503 / 0.561 / 0.614 / 0.670
95% CI / (0.241, 0.765)** / (0.282, 0.840)** / (0.374, 0.855)** / (0.431, 0.910)**
P-value / [0.00] / [0.00] / [0.00] / [0.00]
SIGI / 0.841 / 0.507 / 0.935 / 1.670
(-0.750, 2.432) / (-1.449, 2.463) / (-1.093, 2.964) / (-0.539, 3.880)
[0.29] / [0.61] / [0.36] / [0.13]
Gini / 0.099 / 0.282
(-0.090, 0.287) / (0.091, 0.474)**
[0.30] / [0.00]
GNI (ln) / -1.322 / -2.529
(-3.447, 0.802) / (-4.165, -0.892)**
[0.22] / [0.00]
Smoking / -0.154 / -0.225
(-0.322, 0.015) / (-0.383,-0.067)**
[0.07] / [0.01]
Urbanization / -0.053 / -0.036
(-0.143, 0.036) / (-0.118, 0.045)
[0.24] / [0.37]
TFR / 1.084 / 0.208
(-0.975, 3.142) / (-1.405, 1.821)
[0.30] / [0.80]
TEA / 0.355 / 0.500
(-0.286, 0.997) / (-0.132, 1.132)
[0.27] / [0.12]
Adjusted R2 / 0.35 / 0.35 / 0.37 / 0.45
N / 55 / 55 / 55 / 55
* p<0.05; ** p<0.01
aIn all 4 models, the outcome is sex differences in obesity prevalence, computed as female – male prevalence; Mean prevalence is the simple mean of female and male prevalence; SIGI – Social Institutions and Gender Index; Gini – Coefficient of income inequality; GNI (ln) – Log transformed gross national income; Smoking – difference between female and male prevalence expressed as percentage; Urbanization – level of urbanization expressed as percentage; TFR – Total Fertility Rate; TEA – Total Energy Availability. N – sample size. Model 1 is unadjusted, Model 2 adjusts for GNI and Gini coefficient, Model 3 adjusts for smoking, total fertility rate, urbanization, and total energy availability. Model 4 includes all variables (full model).We report regression with robust standard errors. For the primary exposure, SIGI, we report standardized regression coefficient.
Figure S2-S4: Magnitude of sex differences in obesity prevalence by index of gender inequality and national income (log-scale)
Size of the symbol proportional to magnitude of the sex differences. Triangles indicate countries in which obesity prevalence is greater in men. Circles indicate countries in which there is a female excess in obesity.
S2.
S3.
S4.
Table S5. List of 151 countries with BMI data by income group.1
Low income / Lower middle income / Upper middle income / High income: OECD / High income: nonOECDAfghanistan / Armenia / Albania / Australia / Bahrain
Bangladesh / Bolivia (Plurinational State of) / Algeria / Austria / Barbados
Benin / Cameroon / Argentina / Belgium / Croatia
Burkina Faso / Congo / Azerbaijan / Canada / Cyprus
Cambodia / Côte d'Ivoire / Bosnia and Herzegovina / Czech Republic / Kuwait
Central African Republic / Egypt / Brazil / Denmark / Malta
Chad / Fiji / Bulgaria / Estonia / Oman
Comoros / Georgia / Chile / Finland / Saudi Arabia
Eritrea / Ghana / China / France / Singapore
Ethiopia / Guatemala / Colombia / Germany / Trinidad and Tobago
Gambia / Guyana / Costa Rica / Greece / United Arab Emirates
Guinea / India / Cuba / Hungary
Haiti / Indonesia / Dominica / Iceland
Kenya / Kiribati / Dominican Republic / Ireland / Not Ranked2
Kyrgyzstan / Lao People's Democratic Republic / Gabon / Israel / Cook Islands
Madagascar / Lesotho / Iran (Islamic Republic of) / Italy / Nauru
Malawi / Mauritania / Jamaica / Japan / Niue
Mali / Mongolia / Jordan / Korea (Republic of)
Mozambique / Morocco / Kazakhstan / Luxembourg
Myanmar / Nicaragua / Latvia / Netherlands
Nepal / Nigeria / Lebanon / New Zealand
Niger / Pakistan / Libyan Arab Jamahiriya / Norway
Rwanda / Papua New Guinea / Lithuania / Poland
Sierra Leone / Philippines / Malaysia / Portugal
Tanzania (United Republic of) / Samoa / Maldives / Slovakia
Togo / Senegal / Mauritius / Spain
Uganda / Solomon Islands / Mexico / Sweden
Zimbabwe / Sri Lanka / Namibia / Switzerland
Swaziland / Palau / United Kingdom
Turkmenistan / Panama / United States
Ukraine / Peru
Uzbekistan / Romania
Vanuatu / Russian Federation
Viet Nam / Saint Lucia
Yemen / Serbia
Zambia / Seychelles
South Africa
Thailand
The FYR of Macedonia
Tunisia
Turkey
Uruguay
Venezuela (Bolivarian Republic of)
1Income ranking based on World Bank Classification.
2Income ranking not available for Cook Islands, Nauru and Niue.