The double burden of malnutrition in South Africa: risk factors for undernutrition and overnutrition and the development of a novel mid-upper arm circumference (MUAC) screening tool in children

McLaren, Shawn (2022) The double burden of malnutrition in South Africa: risk factors for undernutrition and overnutrition and the development of a novel mid-upper arm circumference (MUAC) screening tool in children. Doctoral thesis, London Metropolitan University.

Abstract

The World Health Organisation has produced recommendations for screening for malnutrition in childhood. However, evidence is emerging that the performance of these tools is region-specific. The aim of this set of six studies were to identify contextual factors which contribute to childhood malnutrition and assess the performance of a malnutrition screening tool in a South African population.

The first study was conducted on 400 infants and children younger than 24 months of age from primary health facilities in the Eastern Cape, South Africa. The aim was to assess whether household food security, social welfare, and birth weight were associated with childhood nutritional status. Low birth weight was associated with height-for-age, and that this effect was still applicable even when controlling for food security and social welfare (p<0.05). Low birth weight could therefore be considered as an indicator from birth to triage children as high risk for stunting.

The second study was conducted on 1496 infants and young children between birth and 59 months of age, living in the Eastern Cape, South Africa. This study aimed to identify whether vitamin A supplementation, anti-helminth chemotherapy, and vaccination coverage, are associated with childhood nutritional status. We found that coverage of vitamin A supplementation (60.4%) and deworming (61.5%) were poor among this sample, with one third of participants having missed their most recent dose, however vaccination coverage was stronger (86.2%). Having a history of delayed vitamin A, deworming or vaccinations was not associated with anthropometric indicators of nutritional status among these children (Χ2 =23.79, df=19, n=840, p=0.204). However, a significant relationship was observed between delayed vitamin A supplementation and age category (Χ2 =32.105, df=19, n=836, p=0.03) as well as between delayed deworming and age category (Χ2 =45.257, df=17, n=558, p<0.01) and older children were more likely to have missed doses. We found that low birth weight was associated with a greater risk of stunting in this population (p<0.05).

As birth weight was associated with indicators of nutritional status among children, the next study aimed to examine the nutritional status of women of childbearing age in South Africa. This study made use of DHS data on 2640 women between 15 and 49 years of age, obtained using a stratified, nationally representative sample in 2016. It was found that anaemia was common (28.9%) among South African women, and that haemoglobin level is higher among women with a higher body mass index (Independent Kruskall-Wallis test= 27.014; df=5; p<0.001), access to improved sanitation (Mann-Whitney U test p=0.017; n=2690), and increased wealth status (Kruskall-Wallis =29.568; df=4; P<0.01).

The fourth study assessed MUAC’s performance in identifying cases of acute malnutrition in the South African population of infants and young children aged 6 to 24 months (n=400). It was found that the cut-off values currently recommended by the WHO perform poorly in this context. The sensitivity of the current MUAC for moderate acute malnutrition was 0%, and the specificity was 99.6% Correlation analysis for the relationship between WHZ and MUAC suggested a strong, positive, dependent relationship between these two indicators (r=0.78). The least squares regression formula (Y=15.409+0.803x (males); Y=15.13+0.83x (females)) was then used to predict where WHZ=-2 is most likely to correspond with a MUAC value in cm. The new predicted MUAC values of 13.8cm (males) and 13.5cm (females) were subsequently tested for sensitivity and specificity. The sample was too small to calculate sensitivity for both sexes, but 96.4% specificity was achieved for females. A sensitivity of 100% and specificity of 94.5% was achieved with the male MUAC cut- off. Thus, the proposed cut-offs identified all the wasted children (WHZ<-2), while identifying an acceptably low number of false positives.

The fifth study investigated the potential of a MUAC tape for identifying overweight and obesity among this population. The area under the curve (AUC) for identifying overweight males 0-6 months old (n=58) was 0.766. The MUAC cut-off value at 14.5 cm had a sensitivity of 88.9% and specificity of 63.3% (J=0.542). Female children 0-6 months old had an AUC of 0.788 for overweight. The MUAC cut-off with the highest J-value (J=0.585) was 14.2 cm (100% sensitivity, and 58.5% specificity). Data obtained from males aged six to 24 months (n=139) generated ROC curves with AUC of 0.821 for overweight (+2<WLZ<+3) and 0.905 for obesity (WLZ>+3). The MUAC cut-off value of 16.5 cm had the highest J-value (0.589) and a sensitivity of 85% and specificity of 71.4% for identifying overweight. The optimum MUAC cutoff value for identifying obesity among males six to 24 months old was 17.2 cm (88.9% sensitivity, 80.8% specificity, J=0.697). A MUAC cut-off for identifying overweight female children aged six to 24 months (n=130) was determined at 16.5 cm (AUC=0.938). This cut-off value had a sensitivity of 100% and specificity of 76.7% (J=0.767). The optimum MUAC cutoff value for identifying obesity was 17.0 cm (J=0.758). This novel application of MUAC resulted in new cut-off values for identifying overweight and obesity among South African children.

The final study aimed to validate the new cut-off values for under and overnutrition. A validation dataset of 247 infants and young children from the Eastern Cape, South Africa. It was found that the new MUAC values performed well for identifying acute malnutrition (100% sensitivity, 99.1% specificity, PPV=0.25, NPV=1.0) and obesity (90.1% sensitivity, 80.1% specificity, PPV=0.235, NPV=0.992) but performed poorly in identifying overweight (11.1% sensitivity, 86.8% specificity, PPV=0.111, NPV=0.868).

In summary, these studies, which have been published in peer-reviewed scientific journals, have provided information on the context of childhood malnutrition in South Africa. It was found that birth weight and maternal nutritional status were important indicators in determining anthropometric status among South African children. This research has also led to the development of a novel MUAC screening tool which can identify the full spectrum of anthropometric malnutrition in these children.

Documents
8401:43310
[img]
Preview
McLaren-Shawn.pdf - Published Version

Download (4MB) | Preview
Details
Record
Statistics

Downloads

Downloads per month over past year



Downloads each year

View Item View Item