Evidence of a double burden of malnutrition in urban poor settings in Nairobi, Kenya

Background: Many low-and middle-income countries are undergoing a nutrition transition associated with rapid social and economic transitions. We explore the coexistence of over and undernutrition at the neighborhood and household level, in an urban poor setting in Nairobi, Kenya. Methods: Data were collected in 2010 on a cohort of children aged under five years born between 2006 and 2010. Anthropometric measurements of the children and their mothers were taken. Additionally, dietary intake, physical activity, and anthropometric measurements were collected from a stratified random sample of adults aged 18 years and older through a separate cross-sectional study conducted between 2008 and 2009 in the same setting. Proportions of stunting, underweight, wasting and overweight/obesity were dettermined in children, while proportions of underweight and overweight/obesity were determined in adults. Results: Of the 3335 children included in the analyses with a total of 6750 visits, 46% (51% boys, 40% girls) were stunted, 11% (13% boys, 9% girls) were underweight, 2.5% (3% boys, 2% girls) were wasted, while 9% of boys and girls were overweight/obese respectively. Among their mothers, 7.5% were underweight while 32% were overweight/obese. A large proportion (43% and 37%%) of overweight and obese mothers respectively had stunted children. Among the 5190 adults included in the analyses, 9% (6% female, 11% male) were underweight, and 22% (35% female, 13% male) were overweight/obese. Conclusion: The findings confirm an existing double burden of malnutrition in this setting, characterized by a high prevalence of undernutrition particularly stunting early in life, with high levels of overweight/obesity in adulthood, particularly among women. In the context of a rapid increase in urban population, particularly in urban poor settings, this calls for urgent action. Multisectoral action may work best given the complex nature of prevailing circumstances in urban poor settings. Further research is needed to understand the pathways to this coexistence, and to test feasibility and effectiveness of context-specific interventions to curb associated health risks. Copyright: The study was conducted in two urban slums of Nairobi Kenya (Korogocho and Viwandani) where the African Population and Health Research Center (APHRC) runs a health and demographic surveillance system titled the Nairobi Urban Health and Demographic Surveillance System (NUHDSS). The two slums are located about seven kilometers (km) from each other. They occupy a total area of slightly less than one square km and are densely populated with 63,318 and 52,583 inhabitants per square km, respectively. The slums are characterised by high levels of poverty, poor housing, poor infrastructure such as potable water and waste disposal, high levels of violence and insecurity, unemployment, and poor health indicators [23,26,30]. For children under five years, data was collected longitudinally between January 2007 and December 2010 as part of a longitudinal study on maternal and child health, a component of the Urbanization, Poverty and Health Dynamics project. Details of this study are published elsewhere [26,31,32]. his study was designed to investigate growth patterns and its correlates among children. All children born between September 2006 and December 2010 were enrolled in the study along with their mothers. The mother-child dyad was followed up after every four months, collecting and updating health information of both the mother and the child. The height/length of the child was measured using wooden portable measuring board (Model WB-27T) by having the baby lying straight (flat) on the board, legs extended, head and feet flat against the board. The height of the mother was measured using a Seca stadiometer (Model 213). The height of the mother was taken by having her stand barefoot straight against the stadiometer, legs pushed back against the measuring board. The weight of both the mother and the baby was taken using the Seca weighing scale (Model No. 881). The mother was first weighed alone then she was weighed with the baby and the baby’s weight was calculated by subtracting the mother’s weight from the combined mother/baby weight.Self-reported data on feeding practices from the mother was recorded using questionnaires. Additional parameters collected including age, education level, and employment status of the mothers were also collected. It is important to note that some children could not be traced until after several visits due to high population mobility in the study setting. As a result, some children may have more data points than others. For the purpose of this study, we used a total of 3335 children who had an interview visit in 2010, and additional data at different time-points, totalling 6750 observations within the year with 35%, 55% and 11% having three, two and one observation respectively. The annual attrition rate in the study was estimated to be between 20% and 30% [31]. All the children were aged below five years at the time of assessment. For adults, data collected from a separate cross-sectional study on cardiovascular disease and risk factors among adults between May 2008 and April 2009 was used. Details of this study are published elsewhere (Vivjer at al., 2013; Oti et al., 2013; Ettarh et al., 2013). Briefly, the study involved a stratified random sample of adults aged 18+ years. Sampling was done using the NUHDSS sampling frame. The study aimed to examine behavioral and physiological risk factors for cardiovascular diseases in Korogocho and Viwandani. Demographic information, perception and lifestyle regarding cardiovascular risk factors on these adults was recorded, and direct measurements including height, weight, and waist/hip circumference captured based on the WHO STEPwise approach to chronic disease risk factor surveillance (http://www.who.int/chp/steps/instrument/en/). Self reported data on dietary intake and physical activity were also collected. Data collection for both the child and adult studies was done by carefully trained field workers during household visits. The two studies from which data were derived were approved by the Ethical Review Board of the Kenya Medical Research Institute (KEMRI). The field workers were trained in research ethics and obtained written informed consent from all respondents, recorded in a consent form. Proxy written consent for children was obtained from their caregivers, recorded in a consent form. APHRC owns the datasets used in this analysis and has a data sharing policy that enables other researchers to access datasets. APHRC’s data sharing policy is available at: http://aphrc.org/wp-content/uploads/2014/05/GUIDELINES-ON-DATA-ACCESS-AND-SHARING.pdf. Data may be accessed through APHRC’s microdata portal at: http://aphrc.org/catalog/microdata/index.php/catalog Length/Height-for-age, weight-for-age, and weight-for-height categories were generated for children under five based on the World Health Organization growth standards, whereby stunting (low height-for-age), underweight (low weight-for-age), and wasting (low weight-for-height) are defined as z-scores of < -2 standard deviations, while overweight/obesity (high weight-for-height) is defined as z-scores of > +2 standard deviations [33]. Associations between nutritional status and individual level factors, maternal factors, and feeding practices were investigated and the chi-squared test results reported. The STATA command nptrend was used to performs non-parametric test of trend for factors with natural ordering. For adult participants, Body Mass Index (BMI) was calculated from directly measured height and weight. Cut-off points of <18.50, 25.00–29.99, and ≥30 kg/m2 were used for underweight, overweight and obesity respectively [34]. Waist circumference and waist-to-hip ratio (WHR) were used as measures of abdominal obesity, defined as waist circumference of > 80 cm in women and > 94cm in men, and a WHR of > 0.80 in women and > 0.95 in men [35]. Adequate physical activity and sufficient fruit and vegetable consumption were categorized as no/yes. Adequate physical activity was defined as engaging in ≥ 3 days of rigorous activity for at least 20 minutes daily or ≥ 5 days of moderate intensity activity for at least 30 minutes daily [36]. Sufficient fruit and vegetable consumption was defined as consuming > 5 servings of fruits and/or vegetables daily [37]. Sex-stratified relationships between BMI category and these individual level factors were investigated and the chi-squared test results reported. Maternal nutritional status was determined from anthropometric measurements of mothers of children in the child study, post-partum. Data analysis was computed using stata version 13.1. Chi chi-squared test was used to determine differences in proportions. Statistical significance was determined at the 5% level of significance. For the child data, since some children were measured at multiple time-points, corrected weighted pearson chi square statistic was used to get valid p-values by converting the chi statistic to an F statistic. Adult data were adjusted for age distribution.