Prevalence and factors associated with overweight and obesity among rural and urban women in Burkina Faso

Introduction Low- and middle-income countries, including Burkina Faso, are facing increasing urbanization with health challenges related to nutrition transition that impact body weight change. This study reported the prevalence and factors associated with overweight/obesity among women living in rural and urban Burkina Faso. Methods We conducted a secondary analysis using data from the Burkina Faso 2013 WHO STEPwise survey. Data included socio-demographic, clinical (anthropometric, systolic/diastolic blood pressure (SBP/DBP), oral/dental symptoms), biological (total and high-density lipoprotein cholesterol and fasting blood sugar), and alcohol and tobacco consumption data. A total of 2191 participants with complete data were considered in the analysis. We categorized the 13 Burkinabe regions by urbanization rate quartiles. We then performed Student's t, chi-squared, and Fisher's exact tests and backward stepwise regressions. Results The overall prevalence of overweight/obesity was 19.6% (13.1% and 44% in rural and urban women respectively, p=0.0001). Common factors positively associated with overweight/obesity in both rural and urban women were being a resident of a region in the highest urbanization rate quartile, having a high level of total cholesterol (alone or via an interaction with age) and having a high DBP. In urban women only, overweight/obesity was also associated with a high SBP. Conclusion The prevalence of overweight/obesity in urban women in Burkina was among the highest levels in urban sub-Saharan Africa and roughly mimicked the urbanization profile of the country. In overweight/obesity conditions, cardiovascular concerns, such as increase in total cholesterol and blood pressure, were objective, and the blood pressure increase was more severe in urban women than in rural women.


Introduction
In low-and middle-income countries (LMICs), the nutrition transition process is accompanied by body weight changes [1,2]. Between 1980 and 2008, the weight gain (in body mass index (BMI), kg/m ² ) per decade was estimated at 0.6 (95% CI: 0.0-1.1) and 0.9 (95% CI: 0.4-1.3) in West African men and women aged >20 years, respectively [3]. Overweight/obesity is known to be associated with increased cardiovascular risk [4,5]. During the same period , the increase in fasting blood sugar (mmol/l) per decade was estimated at 0,05 (95% CI: -0. 15 This study aimed to assess the prevalence of overweight/obesity and the associated factors among rural and urban women in Burkina Faso by using nationally representative data.

Methods
We used data from the Burkina Faso first national survey conducted in 2013, based on the WHO STEPwise approach to Surveillance (STEPS) [12]. The sample size calculation and the data collection process used throughout the country have been reported in two previous papers [13,14]. Complete data from 2438 participating women were available and included data on nutritional status, biological features, and alcohol and tobacco consumption. We conducted the analysis using the data for 2191 women, after excluding those who were pregnant.

Variables of interest
Participant demographic information included age (ranging from 25 to 64 years), marital status (groups: i. married or cohabitating, ii. never married, iii. divorced or separated, iv. widowed), residence environment (i. urban, ii. rural), education levels (groups: i. no formal schooling, ii. primary or more), occupation (groups: i. public or private formal employment or self-employed, ii. student or homemaker or retired or unemployed or volunteer). Anthropometric data were weight (kg), height (m), BMI (weight/height², kg/m²) and waist circumference (cm). We defined overweight/obesity as BMI ≥25 kg/m². Biological data included total cholesterol (mmol/l), high density lipoprotein cholesterol (HDL-cholesterol in mmol/l, a cut-off of >1.2 mmol/l was defined as a high level) and fasting blood sugar (mmol/l, having a level ≥6.1 mmol/l was defined as high blood sugar). Blood pressure (systolic and diastolic values in mmHg) was measured three times, and we used the mean of the three values for each indicator.
High blood pressure was defined as a mean value of SBP/DBP ≥140/90 mmHg or actively undergoing anti-hypertension treatment.
Data on current (past month) alcohol use were recorded by the selfreported alcohol consumption technique. Current (past year) smoking tobacco use considered manufactured cigarette smoking, hand-rolled tobacco smoking, and pipe smoking. Urbanization was characterized using urbanization rates for the 13 Burkinabe regions provided by the "Institut National des Statistiques et de la Démographie (INSD)" in 2006. The urbanization rate of the region was calculated as the proportion of inhabitants living in urban areas in the region. The national mean rate of urbanization was 23.33% (minimum = 6.6%, maximum = 85.4%), and the quartile cutoffs were 8.1, 11.8 and 19.3.
Four regions were included in the first quartile (Q1) or in the second quartile (Q2), three in the third quartile (Q3) and two (of the "Centre" and "Hauts-Bassins" regions) in the fourth quartile (Q4) (Figure 1).
The two regions in this last quartile include the political capital Ouagadougou (located in the "Centre" region) and the economic capital Bobo-Dioulasso (in the "Hauts-Bassins" region) and were the living areas for approximately 62% of the urban dwellers of the country (46.4% for Ouagadougou, 15.4% for Bobo-Dioulasso). These two regions are densely urbanized (Figure 1).

Statistical analysis
We used Stata statistical software for Windows (Version 12.0, College Station, StataCorp, Texas, United States) for the analysis. First, we described the socio-demographic characteristics of the participants for the whole sample and for the rural and urban subgroups. Similarly, we described the clinical, nutritional and biological features based on the urbanicity status and overweight/obesity status. After that, we performed a multivariate logistic regression on the overall sample to estimate the odds ratios of factors associated with overweight/obesity in women. We included in the final model all variables with a p-value <0.20 in the univariate analyses, except for multinomial variables, where the overall p-value in the univariate analysis was considered. A backward stepwise approach was used to construct the final model, and interactions were tested in the overall sample and subgroup analyses (residence environment). For all statistical analyses, a pvalue <0.05 was considered significant.

Ethical considerations
The protocol of the STEPS survey was approved by the Ethics
Compared to rural women, urban women had significantly higher BMI, waist circumference, SBP, DBP, glycemia and total blood cholesterol and frequently had HBP (Table 2). Compared to those without overweight/obesity, overnourished women showed a similar trend (Table 2). Using a logistic regression model, we showed in Table   3 that a rural environment (adjusted odds ratio (aOR)=0.45; 95% CI: 0.33-0.62), low quartiles of urbanization rate (global p-value < 0.0001) and no education (aOR=0.68; 95% CI: 0.50-0.91) were protective factors and independently associated with overweight/obesity in the overall sample of women (Table 3). In addition, overweight/obesity was associated with increased DBP levels (aOR=1.04; 95% CI: 1.03-1.05) and total cholesterol (aOR=1.29; 95% CI: 1.14-1.46) for the whole sample. A marginally significant and positive interaction between age and glycemia was found (p=0.089).
In the subgroup (rural or urban) analyses (Table 4), there was a similar association between urbanization rate and overweight/obesity, with roughly the same strength. The odds ratio for overweight/obesity was associated with increased total cholesterol levels (aOR=1.26; 95% CI: 1.07-1.48) in rural women, while the interaction effect between total cholesterol and age category was significantly associated with overweight/obesity in urban women (overall p=0.08). p=0.062) in urban areas. In rural areas, the presence of dental symptoms was negatively associated with overweight/obesity in rural women. Among socio-demographic factors, education and marital status and age were associated with overweight/obesity in rural and urban women, respectively. Rural women with no education were less overweight/obese than single urban women or urban women aged 34-49 years. When testing the interaction between DBP and SBP in rural and urban women, a significant association was observed in the subgroup of urban women (p=0.042).

Discussion
Overall, in this study, we found that approximately one out of five women was overweight/obese, with a large difference according to living area (13% and 44% in rural and urban women, respectively).

Overweight/obesity and cardiovascular risk factors in the overall sample
The findings in this study were consistent with the results reported in a previous study among peri-urban and urban residents in Overweight/obesity was associated with cardiovascular disease risks in the overall sample, as it was in each subgroup of rural and urban women. The first concern was the risk of increased DBP, with an aOR=1.04 in women overall and 1.02 and 1.15 in rural and urban women, respectively (Table 3, Table 4). Only urban women with overweight/obesity had an additional risk of increased SBP, with an aOR=1.08 (Table 4). The prevalence of HBP was greater in urban women than in rural women (31.2% vs 13.5%, respectively; p=0.0001) ( Table 2). The second serious cardiovascular concern was the increased total cholesterol level, with an aOR=1.29 in women overall, 1.26 in rural women and 1.73 in urban women (via an interaction in the 35-49 years subgroup). These concerns were in line with those of the WHO in LMICs and reported by some authors in SSA [16,17] as was cardiovascular risk worsening in urban areas [18].

Rural women
The overweight/obesity prevalence in rural Burkinabe women ( [28], and alcohol has an acute inhibitory effect on human ghrelin secretion [29]. Rural areas in Burkina Faso usually have a low food availability, and rural women adjust to hunger through alcohol consumption, which suggests an undernourished situation.
The presence of oral or dental symptoms was also negatively associated with the presence of overweight/obesity (aOR=0.61, 95% CI: 0.42-0.89) ( Table 4). These symptoms might be related to eating difficulties/swallowing impairment resulting in insufficient food intake, causing undernourishment. The hypothesis that alcohol consumption, or the presence of oral or dental symptoms, is positively associated with undernourishment in rural women needs to be tested.

Urban women
The  In overweight/obesity conditions, cardiovascular concerns, such as increase in total cholesterol and blood pressure, were objective and more severe in urban women.

Competing interests
The authors declare no competing interests.

Acknowledgments
The authors thank the Ministry of Health for providing them with the STEPS survey database, and Kaboré Ilyasse for figure management.