Background: Obesity and overweight are rising worldwide while underweight rates persist in low-income countries. The aim of this study was to examine changes in the prevalence of underweight and overweight/obesity among non-pregnant women aged 15-49 years, and its socio-demographic correlates in Addis Ababa, Ethiopia. Methods: The data are from 2000, 2005 and 2011 nationally representative Ethiopian Demographic and Health Surveys in Addis Ababa. The dependent variable was women’s nutritional status measured in terms of body mass index coded in binary outcomes to examine risk of being underweight (25 kg/m2 vs. ≤25 kg/m2). Logistic regression models were used to estimate the strength of associations. Results: The prevalence of overweight/obesity increased significantly by 28%; while underweight decreased by 21% between 2000 and 2011. Specifically, the prevalence of urban obesity increased by 43.3% i.e., from 3.0% to 4.3% in about 15 years. Overall, more than one-third (34.7%) of women in Addis Ababa were either under or overweight. Women’s age and proxies for high socio-economic status (i.e. household wealth quintile, educational attainment, access to improved source of drinking water, and television watching) were positively associated with being overweight. The correlates of underweight were young age and proxies for low socio-economic status (i.e. low wealth quintile, limited access to improved source of water or toilet facility). Conclusions: There is a need for policies to recognize the simultaneous public health problems of under and overnutrition, and for programs to target the distinct populations that suffer from these nutrition problems in this urban area.
We draw from United Nations Children’s Fund (UNICEF’s) nutrition conceptual framework to identify potential correlates of nutritional status (overweight and underweight) [26]. The framework highlights that immediate causes of nutritional status are diet intake and health status. The underlying causes of diet intake and health status in turn rest on three pillars: (1) household food security (i.e. income to buy food, access to foods); (2) maternal and caregiver practices (i.e. maternal education, inadequate or inappropriate information or education breastfeeding practices, etc.) and (3) health services and the environment (i.e., access to maternal healthcare services, exposure to media (which has been associated with being sedentary [10, 11], dwelling characteristics, access to water and sanitation). These underlying causes are in turn determined by basic societal causes, including cultural or socio-political characteristics that may be dictated by ethnicity, or religion, working and marital status, woman’s relationship to head of household) and economic structures (i.e. wealth or socioeconomic status). These basic societal factors may shape community and individual resources and behaviours [26]. For example, weight gain increases with parity especially in urban areas women with lower parity are more likely to have lower BMI levels [27]. In Ethiopia, health and religious beliefs have strong link [28, 29]. An in-depth analysis shows that Muslim women show better decision making power on their own health care as compared to other religious groups [30]. Another study shows that in Ethiopia, women’s decision-making autonomy has positive effect on their nutritional status [31]. The data are from 2000, 2005 and 2011 nationally representative Ethiopian Demographic and Health Surveys (EDHS). The survey was implemented by the Ethiopian Central Statistical Agency (CSA) with the technical assistance of Inner City Fund (ICF) International through the USAID-supported MEASURE DHS project. The survey inquires about household members’ and individual characteristics using household questionnaire, woman’s questionnaire and man’s questionnaire. Individual women of reproductive aged 15-49 years were interviewed face-to-face on their background characteristics and height and weight measurements were carried out on women aged 15-49 years. This study focused on Addis Ababa the capital of Ethiopia. EDHS employed two stage cluster sampling technique. Census enumeration areas were the sampling units for the first stage while households comprised the second stage of sampling. A fixed number of 30 households were selected for each enumeration areas. For this study, variables were obtained from the individual women’s and household questionnaires. The women’s questionnaire provided information on the characteristics of the individual woman while the household questionnaire provided information on household possessions and amenities such as source of drinking water, toilet facilities and dwelling characteristics [32–34]. The dependent variable in this study is women’s nutritional status measured by their BMI. A cut-off point of 18.5 is used to define underweight and a BMI of 25 or above usually indicates overweight or obesity according to the WHO Expert Committee on Physical Growth [4]. Pregnant women were excluded from the study. From the EDHS database, the following variables were identified: Underlying determinants: woman’s educational attainment (no education, primary, and secondary or higher education), women’s decision-making autonomy on own healthcare, large household purchases and visits to relatives, partner’s educational status, antenatal visit and place of delivery. UNICEF’s multiple indicator cluster survey was used to define source of water and sanitation categories [35]. Source of drinking water was categorized as improved for those who have piped water source, public tap or standpipe, tube well or borehole, protected well or spring and rain water; and unimproved for those with access to water piped outside of the compound, unprotected well, unprotected spring, well or borehole, bottled water, river/dam/lake/pond/stream/canal/irrigation channel, or tanker truck. As to type of toilet facility, flush toilets, ventilated pit latrine and pit latrine with slabs were categorized as improved and all the rest, including pit latrine without slab, open field, composting toilet and others were grouped into ‘unimproved’. Exposure to media was assessed in terms of exposure to newspaper/magazine and television. Hence, each of these variables was categorized as ‘yes’ if the respondent reads newspaper/magazine or watches television regardless of the frequency; as ‘no’ if the respondent doesn’t read newspaper/magazine or does not watch television at all. Basic determinants: women’s working status (yes/no), age (years), marital status (never married, currently married, divorced, widowed, living together), parity (number of children ever born), woman’s relationship to head of household (head, wife, daughter, etc.), sex of household head (male/female), age of household head (years) and religion (Orthodox Christian, Muslim, Catholic, Protestant, Traditional). Wealth index was also examined in the following way: our preliminary analysis showed that at national level the majority of the study participants in Addis Ababa (94% and 98% 2005 and 2011, respectively) belonged to the highest category of the wealth quintile. However, for the purpose of this study, we developed a wealth index factor score using the principal component analysis method to regroup the study population into the wealth quintile specific to Addis Ababa. In the grouping of the wealth status, after obtaining the wealth quintiles, the 815 and 1648 sample size for 2005 and 2011 EDHS data were classified into five categories of approximately equal numbers ranging from the least advantaged (first quartile or lowest class) to the most advantaged (fifth quintile highest class). Wealth index was not included in the 2000 EDHS and hence all the analyses related to wealth quintiles in this study refer only to the 2005 and 2011 data. All analyses were conducted using the Statistical Package for Social Sciences (SPSS) version 17.0. Individuals with missing values for BMI (n = 91) or any of the other covariates were excluded in the analyses. Variables including women’s decision-making autonomy on own healthcare, large household purchases and visits to relatives, partner’s educational status antenatal visit and place of delivery were excluded from the analyses for having large (50-80%) missing values. Due to the non-proportional allocation of the sample to the different regions and to their urban and rural areas during stratification, EDHS recommends sampling weights for any analysis using EDHS data to ensure representativeness of the survey results at the national and regional level. However, in order for the survey precision in urban areas to be comparable with that in rural areas, urban areas were oversampled. The DHS also advises against use of sample weights for oversampled areas as it drastically overestimates sampling variances and confidence intervals. Since the current study was entirely based on samples from urban Addis Ababa without comparisons with other regions in the country, sample weighting was not needed in the estimation of means, proportions or ratios. Covariates were cross-tabulated by BMI categories, and Chi-square values were used to test for significant associations. Multivariate logistic regression models were fitted for each outcome for each one of the EDHS data (six in total for years 2000, 2005, and 2011). The multivariate models included variables that statistically significantly associated with BMI levels (p-value 10 value of variance inflation factor were not included in the final model.
N/A