Infant feeding, growth monitoring and the double burden of malnutrition among children aged 6 months and their mothers in KwaZulu-Natal, South Africa

South Africa has a documented high prevalence of stunting and increasing obesity in children as well as obesity in adults. The double burden of malnutrition, which can be on an individual-, household- or population level, has implications for both health and the economic development of a community and country. This paper describes a large-scale survey (N = 774) of infant feeding, growth monitoring and anthropometry among mother and child pairs aged 6 months of age in KwaZulu-Natal (KZN), South Africa, conducted between January and August 2017. Among children, a large increase in the prevalence of stunting and obesity was seen between birth and 6 months of age increasing from 9.3% to 21.7% and 4.0% to 21.0%, respectively. 32.1% of the mothers were overweight [body mass index (BMI): 25.0–29.9] and 28.4% had obesity grade 1 (BMI: 30–<40). Although most mothers (93%; 563/605) initiated breastfeeding, the introduction of other foods started early with 17.6% (56/319) of the mothers having started giving other fluids or food to their child within the first month. At 6 months 70.6% (427/605) children were still breastfed and 23.5% were exclusively breastfed. In addition, we found that length measurements were done less frequently than weight measurements between birth and 6 months, on average 2.2 (SD: 1.3) versus 5.8 (SD: 1.5) times. Moreover, there is a need for improvement of health worker training and understanding regarding anthropometric measurements when assessing malnutrition in children in the clinics. Early detection and improved infant feeding practices are key in preventing both stunting and obesity in children. This study forms part of a larger study undertaken to estimate exclusive breastfeeding rates among 14 weeks old infants in KZN at two time points, known as KIBS1 (KwaZulu‐Natal Initiative for Breastfeeding support) and KIBS2 (Horwood et al., 2018, 2020). In this paper, we present the findings of a cross‐sectional survey conducted among mothers and caregivers of children aged 6 months (25‐31 weeks), which aimed to explore growth monitoring practices, anthropometry and feeding practices among 6‐month‐old children, and was conducted alongside the KIBS2 breastfeeding survey between January and August 2017. The study was undertaken in primary health care (PHC) clinics in KZN, one of the largest provinces in South Africa, with a population of over 11 million people (Stats SA, 2019). Free health care services are provided to mothers and children attending public health facilities in South Africa. PHC clinics provide the initial point of contact where maternal and child health services are provided, including antenatal, post‐natal and child health, nutrition, immunisation, and curative services. A comprehensive schedule of immunisations is provided to all children in South Africa, including the first dose of the measles vaccine at 6 months. In addition, mothers are advised to bring their infants monthly for growth monitoring for the first 2 years of life (National Department of Health South Africa, 2019). In KZN over 80% of children are fully immunised at 1 year. However, severe malnutrition in children under 5 years of age remains high at 5.3/1000 children, and 29% of children are stunted. At the time of the study, infant mortality was estimated at 35 per 1000 live births in South Africa [National Department of Health (NDoH) et al., 2019]. The sample size was calculated based on obtaining valid estimates for breastfeeding rates among children at 14 weeks for the KIBS2 study. Thirty clinics were randomly sampled, and the sample included clinics in all districts of the province. This survey was conducted alongside KIBS2 and caregivers attending with 6‐month‐old children were recruited for the duration of the KIBS2 study period but were not part of the KIBS2 study. All mothers or caregivers aged 15 years or above who attended the participating clinics with a child aged 6 months (25–31 weeks) were eligible to participate in the study. The 6 months age was chosen to coincide with the time when children attend for measles immunisation, which presented an opportunity to reach children in a narrow age band. Non‐maternal caregivers answered a subset of relevant questions. Exit interviews were conducted after completion of the clinic visit by trained fieldworkers in the local language (English or isiZulu) using structured questionnaires (Supporting Information File 1). Background data, such as age, education level and household setting, including access to water and electricity were asked. Mothers and non‐maternal caregivers were asked questions about feeding practices since birth and other feeding practices such as whether any other food or fluids were given to the child together with, or instead of breastmilk. Current feeding practices were assessed using a 24‐h food and fluids recall. Moreover, mothers and non‐maternal caregivers were asked about their knowledge and attitudes towards breastfeeding with statements and questions. The questions were a series of true/false (T/F) questions constructed in collaboration with the Nutrition Directorate, Department of Health in KZN. These included the following statements: breastfed babies have less diarrhoea (T); a mother who feels the baby is not getting enough breastmilk should top up with formula milk (F); infant formula contains all the ingredients found in breastmilk (F). Patient‐held records for the children [Road to Health Card (RTHC)] were reviewed by fieldworkers and all anthropometric data (length and weight measurements) recorded on the RTHC since birth until the day of data collection were captured, together with the date of recording. The mother's current height and weight were measured and recorded at the site. Ethical approval was obtained from the Biomedical Research Ethics Committee at the University of KwaZulu‐Natal (BE064/14) and from the KZN Department of Health. All participants provided written informed consent. Confidentiality and anonymity were assured through the allocation of study numbers. To ensure all mothers of young children were able to participate, ethical approval explicitly allowed the inclusion of younger mothers aged 15–17 years. Permission to undertake the study was obtained from the KZN Department of Health, district managers in all districts, and facility managers in participating clinics. Data was captured on handheld android tablets and uploaded to a central server in real time. Extensive quality control checks were carried out by trained study staff. Data were cleaned and analysed in Stata 16.0 (StataCorp, 2019). Anthropometric data was cleaned in two stages. First, as anthropometric data was captured from the RTHC, inter‐ and intra‐rater reliability could not be assessed. Therefore, if errors in the recording of the data of children were identified this data was removed from the data set. Errors of recording occurred when the value recorded was incompatible with a child's length or weight. The following numbers of children were removed: seven for birthweight, two for birth length, 14 for weight at 6 months and 45 for lengths at 6 months. Second, the anthropometric data were cleaned based on attained z‐scores from the WHO Child Growth Standards. Statistical analysis was undertaken using the Stata command ‘zscore06’ to calculate the different z‐scores; Length‐for‐age z‐score (LAZ) and weight‐for‐length z‐score (WLZ). Measurements were flagged at the following criteria Measurements were set to missing if one or more of these extreme values existed after individually assessing them. The following numbers of children with extreme values were excluded: two for LAZ at birth; seven for LAZ at 6 months; 22 for WLZ at birth; 17 for WLZ at 6 months. There was a wide variation in the quality and number of measurements across clinic visits from birth to 6 months. The study team's presence at the site is a likely reason for an increased number of weight and length measurements performed at the time of the interview. However, to display the difference in weight and length outcomes, all recorded measurements were included. Therefore, this resulted in different denominators for calculations regarding length and weight, such as LAZ and WLZ. Descriptive statistical analyses were undertaken to describe the characteristics and distribution of the population. Categorical data are presented as percentages while continuous data are presented as means with standard deviations and confidence intervals. Multi‐variable analysis was used to investigate potential risk factors with the dependent variables LAZ and WLZ with cut‐offs at <−2 and >2 z‐scores, respectively. LAZ < −2 indicates stunting and WLZ > 2 indicates overweight. Selected variables were based on the UNICEF Conceptual framework on young child malnutrition from 1991 (United Nations Children’s Fund, 1991). The selected variables were gender, birthweight, household information, reported breastfeeding practices for the first 6 months and current breastfeeding practices (assessed through 24 h recall), mother’s age, mother returning to school, mother’s height and HIV status. These were all included in the final model because of potential confounding factors. Only the adjusted OR with 95% CI analysis is presented in the results.