The purpose of this review was to evaluate the effectiveness of approaches to treating SAM, both the WHO protocol for inpatient management and community-based management using RUTF, as well as the effectiveness of approaches to managing MAM. In all cases we found fewer high quality studies than expected. We were unable to conduct a pooled analysis comparing the impact of the WHO protocol vs. standard care for the treatment of SAM due to the type of studies available. We conducted meta-analyses for community-based management of SAM as well as management of MAM; however, for the MAM analysis, the data available only allowed us to pool studies comparing two food commodities. Thus, we were unable to adequately evaluate the intervention effects separate from product effectiveness. While there are limitations to both the review and Delphi process that will be discussed subsequently, the estimates generated from the literature review and subsequently vetted through the Delphi process represent the next step in modeling interventions to address SAM and MAM in LiST.
The WHO protocol for the inpatient management of SAM is substantiated through considerable evidence, based both in research and expert opinion [1, 54]. Several studies have demonstrated that it is possible to attain low CFRs. However, as illustrated by the lack of high quality intervention studies, lack of adjustment for confounding variables in observational studies, and absence of key details in many publications, there is a clear need for further research to improve our understanding of how to consistently achieve low CFRs across varying resource-constrained settings.
The shift to outpatient care for the treatment of uncomplicated SAM represents a major turning point in the management of severe acute malnutrition, as is has facilitated improved coverage and lower opportunity costs to caregivers . Community-based management of severe acute malnutrition is backed by a wealth of observational and programmatic data [55–57], yet we found fewer impact studies than expected. While no significant difference in mortality was found in our meta-analysis, children given RUTF were 51% more likely to achieve nutritional recovery in a timely manner, though there was substantial heterogeneity. The differences in anthropometric outcomes, while statistically significant, were small and may not be clinically significant.
It should be noted that these pooled estimates were based on two cohorts of children, both in Malawi, and thus may not be generalizable. Additionally, HIV is an important factor to consider given that the HIV prevalence rate of children with SAM in Sub-Saharan Africa is high . Unfortunately we were unable to disaggregate the meta-analysis as only one trial tested for HIV. A 2009 review that included children with SAM concluded that HIV-infected children are significantly more likely to die than HIV-uninfected children, but used a broader definition of acute malnutrition . Much remains unclear about how to care for HIV-uninfected children with SAM .
The results of our meta-analysis on community-based treatment of MAM demonstrate that RUSF is slightly more beneficial than CSB. Although statistically significant, the higher rate of weight in the RUSF group is small and may not be clinically important. Children in the RUSF group were significantly more likely to recover and less likely to be non-responders. However, these estimates contained considerable levels of heterogeneity, both in terms of study design and in terms of intervention quality, which is poorly captured by most studies. Furthermore, several individual studies that we were unable to pool in our meta-analysis report modest or no statistically significant difference in key nutritional outcomes when comparing products [60–62]. There are several dozen ongoing or planned studies focused on demonstrating efficacy or effectiveness between or among a range of possible food products and nutrient supplements in the context of the management of MAM, most of which will have reports in the upcoming few years (personal communication CMAM Forum, 2012).
There are several limitations of this analysis. As some of the participants in our Delphi process indicated, outcomes of treatment programs are highly context specific and depend on background rates of HIV, seasonal fluctuations in food availability, and many other context-specific variables. Additionally, the outcomes of the programs depend not only on the products used, but the general quality of the program design and implementation, as has been noted by several researchers [16, 63–65]. Despite the importance of context, intervention quality, and the linking of inpatient and outpatient treatment programs along with preventive strategies, it was not possible to undertake a disaggregated analysis by context, due to the limited number of trials available, the lack of detail given on the interventions and analysis in many studies, and the requirement for a single effect estimate in LiST.
Further to the issues inherent in the analysis, there are issues with individual studies that warrant discussion. The diets given to children were often not described in detail, and the amounts of CSB given to the comparison group varied, sometimes including enough to share with family members. Thus dietary intake of study participants is not clear in all cases. Furthermore, all but one of the studies in the meta-analysis were conducted in Africa, with a bias towards Malawi (see additional file 2), thus limiting the generalizability of the results. Additionally, all studies passively recruited participants who were brought to treatment facilities. This may introduce bias if there are systematic differences between caregivers who are more likely, and those who are less likely, to bring their children to facilities for treatment.