• Citer

Analysis of variation of mid-upper arm circumference and weight-for-height in children for the assessment of malnutrition in populations and individuals

par Grellety Bosviel, Emmanuel
Président du jury Castetbon, Katia
Promoteur Donnen, Philippe
Co-Promoteur Dramaix Wilmet, Michèle
Publication Non publié, 2019-02-06

Thèse de doctorat

• Accès en ligne
• Détails
• Contenu
• Statistiques

Résumé :

Death from hunger and starvation can be avoided with appropriate diagnosis and treatment if the necessary knowledge and resources are available. The current definitions of acute malnutrition are based either upon a weight-for-height Z-score (WHZ) below -2 standard deviations of the international reference population (World Health Organization 2006 Growth Standards) or a mid-upper arm circumference (MUAC) lower than 125 mm. These indicators are used independently to define the sum of moderate and severe acute malnutrition, commonly referred to as global acute malnutrition (GAM). Severe acute malnutrition (SAM) is defined as the children with WHZ < -3 SD or MUAC <115 mm. These cut-off points are used both to estimate the prevalence of malnutrition and also to identify those children who should be admitted for individual treatment of their acute malnutrition.However, the ramifications of the new WHO standards and the introduction of the absolute MUAC as an additional criterion have not been sufficiently explored. There appears to have been little detailed analysis of the variation of MUAC in children using these new standards. Thus, there is insufficient information available for predicting changes in patient load due to the addition of an absolute MUAC cut-off, the degree of overlap between the criteria and the factors that affect the selection of malnourished children using the two criteria, WHZ and MUACNevertheless, because of the ease of use of MUAC and strong advocacy based mainly the relative sensitivity and specificity of WHZ and MUAC in predicting long-term all-cause mortality in the community, many organizations and some governments are now moving from using MUAC to screen children in the community and elsewhere to MUAC-only programs with abandonment of using WHZ altogether.A better understanding of the relationship between these measures is important as differences can have significant implications on the decision to intervene in a nutritional crises, assessment of potential program size, resource requirements and outcome, selection of children admitted for treatment and the strategy which will have the greatest influence upon mortality and the other poor outcomes of being malnourished. To this end my thesis included the following studies which explored the variation of MUAC and WHZ in children for the assessment of malnutrition.In my first study, I examined the direction and degree of discrepancy between MUAC and WHZ of children aged 6-59 months in 1,832 anthropometric surveys from 47 countries, mainly in Africa. The results show that using MUAC or WHZ, 16.3% of children were identified with GAM and 3.5% with SAM. The proportion of overlap between the two indicators was 28.2% for GAM (15-38.5%) and 16.5 % for SAM (6.1-29.8%). Overlap for individual countries was especially low for SAM. The numbers of children diagnosed by either criterion varied dramatically by country: the difference between the relative case-load using WHZ and MUAC for GAM varied from minus 57% to plus 72%. For SAM, in four of the 38 countries, less than 25% of severely malnourished children would be identified and admitted for treatment if a MUAC-only admission policy were being used. For all countries examined, the discrepancies were not adequately explained by any single hypothesis. My second study was in three parts. Each part examined the veracity of the assertion that MUAC is a better indicator of mortality than WHZ. 1) I analysed individual data from 76,887 children admitted to a range of treatment programmes to determine the mortality rates associated with SAM. 2) I conducted an exhaustive search of the literature to identify reports of children diagnosed by WHZ or MUAC with their respective mortality rates.3) I analysed the effect of case load using the prevalence data published in the first study with Case Fatality Rates (CFRs) derived from the empirical data, the literature data and theoretical simulations. We found that mathematical coupling caused a reversal of significance generating Simpson’s paradox so that the interpretation of the relative mortality rates of WHZ and MUAC is unsafe when children with both criteria are included in each group being compared. The analysis suggests that children with SAM identified by WHZ <-3 and admitted for treatment are at as least as high a risk of death as children in treatment with MUAC<115cm and probably at higher risk. Review of 21 datasets that compared WHZ and MUAC mortality rates show problems with interpretation of the reported CFRs in each of the studies; inconsistencies greatly limit analysis, comparability and interpretation. Caseload is a more important determinant of the number of SAM related child deaths than the relative CFR to give the number of SAM attributable deaths. Where most of the children are identified as SAM using WHZ, rather than MUAC, it is estimated that fewer than half of all SAM related deaths will be identified using a MUAC-only programme.In my third study, I have conducted a Monte Carlo simulation of anthropometric surveys and imposed random errors of measurement on the data in order to examine the effect of measurement error. The results show that there is an increase in the standard deviation with each of the errors, that the spread becomes exponentially greater with the magnitude of the sort of error that occur in real life situations and that the effect of an increase in standard deviation (SD) that appears to be fairly trivial has a major effect upon the reported prevalence of the condition. I show that even within quite a narrow range of SDs (from 0.8 to 1.2) the proportion of children <-2 WHZ can increase from 6% to 15% - which would move the population from one of “acceptable” prevalence to an acute emergency situation. The corresponding SAM would increase from about 1% to nearly 5%. If one was to use such a survey to estimate the current case-load of SAM children the difference would be five-fold. However, this range of SDs is generally thought to represent a “good survey”. When larger and more complex surveys are considered, for example those included in the WHO database or the DHS surveys, the SD is frequently greater than 1.2 leading to give a higher reported prevalence of malnutrition than may be the actual prevalence. In my fourth study, I performed a secondary analysis of the surveys collected in my first study in order to examine the change in reliability of such surveys over time. I analysed the statistical distributions of the derived anthropometric parameters from 1,843 surveys conducted by 19 agencies between 1986 and 2015. The results show that with the introduction of standardised guidelines and software by 2003 and their more general application from 2007 the mean standard deviation, kurtosis and skewness of the parameters used to assess nutritional status have each moved to now approximate the distribution of the WHO standards when the exclusion of outliers from analysis is based upon the SMART flagging procedure. Where WHO flags, that only exclude data incompatible with life, are used the quality of anthropometric surveys has improved and the results now approach those seen with SMART flags and the WHO standards distribution. Agencies vary in their uptake and adherence to standard guidelines. Those agencies that fully implement the guidelines achieve the most consistently reliable results.In conclusion, well-defined and internationally accepted criteria to assess anthropometric survey quality should be universally applied and reported if the surveys are to be reliable, credible and form the basis for appropriate intervention. Using WHZ-only or MUAC-only estimates of prevalence will underestimate the burden of acute malnutrition. Such a program policy would result in between 300,000 and 600,000 SAM deaths occurring in children each year who have no possibility of being treated. WHZ and MUAC are complementary indicators, it is only by using both criteria to identify SAM and admit children for treatment that we will move towards reducing avoidable SAM- related mortality in most countries. This will only be realised when we can conveniently identify children with a low WHZ in community screening programs.