Assessing the use of acute malnutrition indicators for nutrition surveillance: Results from 682 283 child observations in 27 low- and middle-income countries

Abstract

Surveillance with anthropometric indicators is crucial to detect any deterioration in the nutritional status of a population as it provides information on trends to monitor progress and effectiveness of interventions and facilitates geographical and contextual situation analysis which informs prioritization of actions and allocation of resources. For this purposes it is essential that the indicators used to monitor the situation yield comparable results. However, the two indicators most widely used to identify children´s acute malnutrition (AM), the weight-for-height/length Z-score (WHZ) and the absolute value of mid-upper arm circumference (MUAC) provide discrepant results when applied to the same populations. The aim of this report is to shed light on the relationships between WHZ and MUAC in identifying possible population level patterns of acute malnutrition, and explore how they relate to individual characteristics such as sex, age or stunting status, in order to guide their interpretation and use to inform nutrition interventions. The MUAC for age (MUACZ) is also assessed to explore further possibilities of using this indicator as part of population based surveillance taking into account the age bias that exists when assessing children for acute malnutrition using absolute MUAC measurement only. The JRC-UNICEF collaboration was set up to collate, harmonise and analyse a large dataset composed of surveys from 19 West and 7 East African countries and Yemen. In total, 135 national and subnational representative surveys containing 682,283 child observations from 27countries (2011-2018) were collated. We use descriptive statistics and regression analyses to analyse these data. The findings show that WHZ and MUAC measurements identify different manifestations of acute malnutrition and are thus complementary and additive, rather than alternative or exchangeable. Overall and in most of the countries the global acute malnutrition prevalence was lower when using MUAC as compared to WHZ or MUACZ. However, results at country and regional level differed from findings described in other multi-country studies, suggesting that the relationship between the indicators doesn´t follow a geographical pattern (no regional or country pattern can be identified), but rather depend on the sample characteristics of the population surveyed. Importantly, sex, age, and stunting status were confirmed to impact how children are diagnosed as acutely malnourished by the different indicators. Whereas absolute MUAC consistently identifies more acutely malnourished children in younger age groups (below 2 years), MUAC for age (MUACZ) identifies more acute malnutrition in older children. And in relation to the sex of the child, depending on the indicator, the prevalence of acute malnutrition is higher among girls (MUAC) or among boys (WHZ and MUACZ). Conversely, acute malnutrition was consistently higher among stunted children (compared to non-stunted children) across the three indicators, although MUACZ invariably identified the highest number of AM children within the stunted children, as compared to MUAC and WHZ. Finally, these discrepancies can result in discordant situation analysis if the same severity thresholds are applied to all AM population estimates, independently of the indicator used. At the time being, the only global thresholds prescribed to categorize the severity of AM within populations are the ones defined by WHO for wasting based on WHZ. In conclusion, the recommendation is to always specify the indicator used to diagnose acute malnutrition when reporting nutrition outcomes, as well as to disaggregate results by sex, age (below and at/above 24 months) and stunting status for better interpretation. The use of MUAC for age showed potential to improve estimation of acute malnutrition for surveillance but requires additional research. Also, further investigations are needed to define global thresholds to describe severity of acute malnutrition at population level when using the different indicators. Alternatively, to reconsider the age targeting of surveys to 0-23 months, in line with 1000 days programming, and to develop population threholds specific of this age group. Meanwhile, the WHO population based thresholds to interpret the severity of global acute malnutrition for children under five years should be used exclusively for WHZ, and for the acute malnutrition derived from absolute MUAC we recommend to use alternative methods such as the one developed by the Integrated Food Security Phase Classification initiative.