Maternal nutrition in rural Kenya: Health and socio-demographic determinants and its association with child nutrition

High levels of food insecurity and human immunodeficiency virus (HIV) infection place most breastfeeding mothers in Kenya at high risk of malnutrition. We examined the role of selected socio-economic, demographic and health factors as determinants of nutritional status among HIV-infected and HIV-uninfected mothers in rural Kenya and further examined the interrelationship between maternal nutritional and child nutritional status within this population. A cross-sectional design was used to collect data from non-pregnant mothers with children ages 4-24 months in Kisumu District, Kenya. Over 80% of the mothers were breastfeeding at the time of the study. Mean maternal body mass index (BMI) (21.60±3.15) and percent body fat (22.29±4.86) values were lower than among lactating mothers in other Sub-Sahara African countries. Maternal HIV status was not significantly associated with any of the maternal nutritional indicators assessed in the study. Breastfeeding, recent severe illness and having multiple children below 2 years of age were negatively associated with maternal nutritional status, while higher maternal age, socio-economic status and household food security were each positively associated with maternal nutritional status. Significant positive association was reported between maternal weight, height, BMI, mid-upper arm circumference (MUAC), body fat and fat-free mass estimates, and children’s height-for-age, weight-for-age, weight-for-height and MUAC-for-age z-score. This analysis identifies determinants of maternal nutritional status in rural Kenya and highlights the importance of interventions that address malnutrition in both HIV-infected and HIV-uninfected mothers in rural Kenya. Significant association between maternal and child nutritional status stresses the importance of addressing maternal and young child nutritional status as interrelated factors. © 2011 Blackwell Publishing Ltd.

Mother–child (singleton) pairs were recruited through two post‐natal Maternal & Child Health (MCH) clinics in the rural parts of Winam Division located in Kisumu East District in Kenya. Kisumu East District reports relatively high levels of poor health outcomes including high infant, under‐five and mortality and high levels of child malnutrition and food insecurity [UN‐HABITAT 2006; Ministry of Public Health & Sanitation (MOPHS 2009)]. The study area is predominantly inhabited by the Luo tribe and is served by two main Catholic mission area hospitals that are located to serve populations in the upper zone and lower zone of the division. The study inclusion criteria were non‐pregnant mothers with children ages 4–24 months who resided within the study location. Maternal HIV status was noted from the clinic records, and separate sampling frames were created for HIV‐infected and HIV‐uninfected mothers. Because of their low numbers, all eligible HIV‐infected mothers were approached and requested to participate in the study. A computer‐generated simple random sample of eligible HIV‐uninfected mothers was created using SAS Version 9.1 (SAS Institute, Cary, NC, USA). All HIV‐infected and randomly selected HIV‐uninfected mothers were visited at their home and requested to participate in the study. Community health workers (CHWs), with a minimum of high school level of education, were trained to recruit, take body measurements, conduct interviews and record data as per the study protocol. A total of 53 HIV‐infected and 115 HIV‐uninfected mothers were recruited through the upper zone clinic, and a total of 33 HIV‐infected and 145 HIV‐uninfected mothers were recruited through the lower zone clinic bringing the numbers to 86 and 260 for HIV‐infected and HIV‐uninfected mothers, respectively. Participation rates varied by maternal HIV status (66% among HIV‐positive mothers and 89% among HIV‐negative mothers) and by zones (80% in the upper zone and 85% in the lower zone) giving an overall participation rate of 82%. Data collection began in July 2009 and lasted 2 months. All communication with the respondents was carried out in the local Dholuo language. Human subjects approval was obtained for the research study from the George Mason University and Kenya Medical Research Institute. Health officers and the local community leaders were informed in detail about the aim and procedures of the study. Informed written and verbal consent and assent by mothers of study children was obtained before the study. Anthropometric measurements were assessed at the research office based within the clinic precincts. Mothers were asked to remove their shoes and to change into standard lightweight clothes (sleeveless shirt and skirt) provided by the research team before any anthropometric assessments were taken. All measurements were taken to ensure privacy of the study participants. Two separate measures were taken by a pair of trained CHWs. A second set of measures were taken if the difference between the first set of measures were beyond set points: 0.1 kg for weight, 0.5 cm for height, 0.2 cm for mid‐upper arm circumference (MUAC) and 2 mm for skinfolds. All anthropometric measurements followed the procedures described in Lohman et al. (1988). Body weight was measured using a portable electronic scale (Pelstar LLC, Bridgeview, IL, USA). Mother’s height was measured using a portable adult/infant measuring unit (Perspective Enterprises, Portage, MI, USA). The height measurement was read to the nearest 0.1 cm once correct positioning was confirmed. Maternal body mass index (BMI) was computed as weight in kilograms divided by the square of height in meters. The MUAC was measured on the left arm to the nearest 0.1 cm using a MUAC insertion tape (Abbott Laboratories Inc., Columbus, OH, USA). The ‘Lange’ brand calipers (Beta Technology, Santa Cruz, CA, USA) was used to measure skinfold thickness to the nearest millimetre. Skinfold measurements included triceps, biceps, subscapular suprailiac and abdomen. Triceps, biceps, subscapular and suprailiac skinfolds were measured on the left side of the body. Maternal body density was estimated from the sum of skinfolds at the triceps, biceps, subscapular and suprailiac and by using the age‐appropriate Durnin & Womersley equations, and percent body fat was estimated using the Siri equation (Durnin & Womersley 1974; Gibson 2005). Children’s clothes and shoes were removed before any anthropometric measurements were taken. Mothers were asked to hold the children while standing on the electronic weighing scale, and the children’s body weight was taken by the difference method. Their length was measured using a horizontally placed portable adult/infant measuring unit. Head circumference, MUAC and skinfold thickness were measured with the child seated and supported in an upright position on the mother’s lap. Both head circumference and MUAC were measured to the nearest 0.1 cm using an insertion tape. Skinfold thickness assessment included measurements at the triceps, biceps and subscapular locations. MUAC and skinfold measurements were made on the left side of the body. The World Health Organization (WHO) 2006 growth reference standards, which uses the WHO Multicenter Growth Reference Study population, was used to transform children’s measurements into sex‐ and age‐specific z‐scores: length‐for‐age z‐score (LAZ), weight‐for‐age z‐score (WAZ), weight‐for‐length z‐score (WLZ), MUAC‐for‐age z‐score (MCAZ) and head circumference‐for‐age z‐score (HCAZ) (WHO 2007). Stunting was defined as LAZ below −2SD, underweight was defined as WAZ below −2SD, wasting was defined as WHZ below −2SD, low MUAC was defined as MCAZ below −2SD while low head circumference was defined as HCAZ below −2SD. Household food insecurity was assessed using the household food insecurity access scale (HFIAS) version 3 (Coates et al. 2007). The scale has been shown to be a valid and reliable tool in measuring household food insecurity among poor households in rural Tanzania (Knueppel et al. 2010). The HFIAS questions focus on the household food security situation in the previous 4 weeks and places households into four ordinal levels showing increasing household food insecurity at successive levels: food secure, mildly food insecure access, moderately food insecure access and severely food insecure access. Food secure and mildly food insecure access categories were merged to create ‘food secure’ category, while ‘moderately food insecure access and severely food insecure access categories were merged to create ‘food insecure’ category. Child’s breastfeeding status at the time of study was noted as part of a questionnaire on child‐feeding practices. Breastfeeding duration was later calculated from the information provided. Mothers were asked about illness/morbidity experience in the last 7 days using open‐ended and probing questions and a structured questionnaire. The questionnaire was based on a similar questionnaire previously used in Kenya and included a list of illnesses/diseases commonly found in the study area and population (Neumann et al. 2003). The questionnaire was administered by trained CHWs during the mothers’ visit to the research offices. Signs, symptoms, changes in activity and food intake were ascertained by specific probing questions. Signs and symptoms in the questionnaire were organized by general, non‐specific and specific categories, which comprised a diagnosis or illness category. Medications and visits to a health facility were verified by clinic or hospital cards when available. A morbidity score was defined to identify those with mild or severe forms of illness. Mild illness included fever without chills, chills without fever, cold/sore throat, ear problems, eye problems, mouth sore or toothache, skin rash/sores, diarrhoea, painful urination and trouble with arms or legs. Severe illness included malaria; tuberculosis; typhoid; pneumonia; asthma; meningitis; epilepsy; measles; whooping cough; jaundice; fever and chills; combination of bed ridden, poor appetite and any of the mild conditions; combination of bed ridden and swollen/painful joint(s); combination of bed ridden and accident/injury; poor appetite and mouth sores; bloody diarrhoea and painful urination with bloody urine. Child’s weight at birth was either recorded from the child health cards or self‐reported by the mother. In addition, mothers were asked to indicate the child’s size at birth ranging from the following values: ‘very small’, ‘smaller than average’, ‘average’, ‘larger than average’ and ‘very large’. ‘Very small’ and ‘smaller than average’ were merged to form ‘smaller’ category during analysis, while ‘larger than average’ and ‘very large’ were merged to form ‘larger’ category. About 20% of the children had missing information on their birthweights. Information on household membership was collected using questionnaires with the mother being the respondent. Information included birth dates, marital status, religion, tribe and gender of household members, and highest class attained. Household size and number of ‘under‐twos’ (mother’s biological children under 2 years of age) within the household, maternal education level and maternal age were defined from the census data. Education levels included primary school, secondary school and post‐secondary. Because of the low numbers of mothers in the post‐secondary category, secondary and post‐secondary education categories were merged into one ‘post‐primary school’ category. Information on the household SES was collected through interviews with the mothers in each household. The SES questionnaire, which has been previously used among populations in rural Kenya (Neumann et al. 2003), included both social and economic factors and accounted for employment, income, land ownership and usage, education and literacy, household possessions and expenditures, types of houses, and involvement of parents in leadership and community positions. All variables were weighted, based on ranking by community leaders, and a composite SES score was developed by adding up the points, whereby a higher score represents a higher level of SES. SAS Version 9.1 was used for data analysis. Twenty‐five per cent of the mothers were HIV‐infected. The t‐test and chi‐square procedures were utilized to make comparisons across maternal HIV status categories. The association between potential determinants and each of the maternal nutritional status indicators (BMI, MUAC, abdominal skinfold and body fat, and fat‐free mass estimates) was analysed using simple regression analysis. Determinants that showed significant association with at least one of the nutritional status indicators were included in the final multiple linear regression model. Predictor variables in the multiple linear regression models included maternal HIV status indicator, maternal age and morbidity status, household SES score, multiple ‘under‐twos’ indicator, HFIAS, breastfeeding status and child’s age. All models were assessed for goodness‐of‐fit, violation of regression assumptions and for presence of multicollinearity. Multiple linear regression analysis was also used to examine the relationship between each maternal nutritional status (BMI, MUAC, abdominal skinfold and body fat, and fat‐free mass estimates) and child nutritional status while adjusting for maternal HIV status and age, household SES and children’s age and sex. Each reported beta value came from separate regression models between a pair of each maternal nutritional status indicator and each child nutritional status indicator. In addition, multiple logistic regression analysis was used to examine the relationship between each maternal nutritional status and the odds of a child being underweight, wasted, stunted and having low MCAZ and low HCAZ. These sets of analysis adjusted for maternal HIV status and age, household SES and children’s age and sex as well. All models were assessed for goodness‐of‐fit and violation of regression assumptions. Each reported odds ratio (OR) value came from separate regression models between a pair of each maternal nutritional status indicator and each child nutritional status indicator.

N/A

• Dima Health