STUDENT RESEARCH

Summary: The aim of this study was to determine if increasing the capacity of nurses and health volunteers, by providing them with appropriate tools and nutrition knowledge to identify low weight for length Z-score and offer effective nutrition counselling would affect the prevalence of wasting in children less than 2 years of age. Using a cross-sectional survey of households with young children in the Upper Manya Krobo district, an algorithm was developed based on individual and household-level predictors of low WLZ. Then a cluster-randomized intervention was carried out in rural GMP outreach services to compare three treatments: (i) control group, use of the mandated standard-of-care (SOC) weight measurements, weight-for-age z-score (WAZ) chart and the GHS counselling cards; (ii) algorithm group, use of a newly-developed algorithm and SOC; and (iii) length group, measurement of length, use of WLZ chart and SOC. Thirty clusters of community-based GMP clinics were randomly assigned to the three treatment groups (n=10 per group).

Using two cross-sectional surveys at baseline (n=661) and one year later (n=666), we tested differences in prevalence of low WLZ in children 0 to 24 months. At both rounds of survey, children’s anthropometric and dietary data, caregivers’ young child feeding and child care practices, and household assets, demographics, and food security were assessed. Community- and child-level analyses were by intention-to-treat, adjusting for clustering effect, child age, sex, diet, and health factors. The results show that the algorithm was unable to accurately discriminate between children with low WLZ and those with higher WLZ. The indicator WAZ < -1 alone was moderately accurate in identifying low WLZ in children. The intervention had no effect on prevalence of low WLZ at the community level (p = 0.67) or low wasting status at the individual level (p = 0.71).

However, a significantly higher number of scheduled GMP sessions was held in the algorithm and length groups compared to control (mean difference: 2.7; 95% CI: 0.01 to 5.40, and 2.93; 95% CI: 0.55 to 5.31, respectively). A higher proportion of caregivers were provided with individual counseling in the algorithm and control group compared to length (mean difference: 0.24; 95% CI: 0.17 to 0.31, and 0.23; 95% CI: 0.15 to 0.31, respectively). The inter-technical error of measurement of both nurses and health volunteers were unacceptably high. Pre- and in-service training for health personnel are important for improving and maintaining reliable length measurements. In future, there is a need to evaluate rural GMP programs in Ghana to identify bottlenecks and how these can be addressed to improve program effectiveness.

Matilda E. Laar | Doctoral, McGill University, Canada