Introduction

It is estimated that 34 million children under the age of 5 each year are affected by severe acute malnutrition (SAM), a condition associated with significant increased risks of mortality and morbidity [1, 2]. In 2007, a joint United Nations statement endorsed a new model for the management of SAM that combines outpatient treatment with ready-to-use therapeutic foods (RUTF) for uncomplicated cases and inpatient treatment for complicated cases [3]. This model has been shown to be both effective [4, 5] and cost-effective [6–8], with the potential to bring life-saving treatment to millions of children.

Since 2009, the World Health Organization (WHO) has recommended using weight-for-height Z-score (WHZ) < -3 and/or mid-upper arm circumference (MUAC) < 115 mm as anthropometric criteria for admission to therapeutic feeding programs [9, 10]. In recent years, however, the use of MUAC alone for admission has been increasingly discussed [11–13]. Many benefits of using MUAC exist: MUAC is predictive of death, easy to use, acceptable, and linked to community-based screening methods [14, 15]. However, it is known that MUAC and WHZ select different children for treatment [16, 17], complicating a shift from programs currently admitting children using MUAC < 115 mm and/or WHZ < -3 to a new model admitting children using MUAC < 115 mm only. Depending on context, up to 63–79% of children currently recommended for therapeutic feeding with WHZ < -3 and/or MUAC < 115 mm would not be eligible if using MUAC < 115 mm alone for admission [10, 17, 18].

To inform decision making regarding the use of MUAC as a standalone admission criterion in nutrition programming, more information on the programmatic implications of using MUAC alone is urgently needed. Preliminary reports demonstrate demographic and anthropometric differences among children identified by WHZ and MUAC: MUAC is more likely to identify children that are younger, female and more stunted [19–21]. These data have been used to suggest a role for MUAC to identify children that are potentially more vulnerable or at a higher risk of death, supporting the transition to a MUAC-only based admission criterion. Published evidence is, however, scarce and limited in breadth, due to the narrow scope of routine program data often available for analysis. Important parameters, including the clinical profile and treatment response, of children who are currently eligible for therapeutic feeding but would not be recommended for treatment using MUAC < 115 mm as a single anthropometric criterion for treatment remain poorly documented.

To broaden our understanding of how children identified for therapeutic feeding by MUAC and/or WHZ may differ, and to inform future program planning if MUAC-based programming is to occur, we analyzed data from a randomized trial of children aged 6 to 59 months with uncomplicated SAM in Niger. The objective of this analysis was to compare children that would be admitted for outpatient SAM treatment in a MUAC-based program vs. children that are admitted under current admission criteria of low WHZ and/or low MUAC but would be excluded if using MUAC-based admission criteria. We aimed to specifically investigate differences in demographic and anthropometric characteristics between groups previously reported, as well as extend our understanding on other potential differences in vulnerability in terms of clinical and laboratory characteristics and treatment response. These characteristics were also used to explore possible variability within a potential MUAC-only program, comparing children admitted by MUAC < 115 mm admission criteria, with vs. without WHZ < -3. These findings will be useful to describe the potential implications of MUAC-based programming in terms of the clinical needs of the expected patient population and can inform planning if such programs are to be implemented.

Materials and Methods

Results

2,399 children aged 6–59 months with MUAC < 115 mm and/or WHZ < -3 were included in the analysis. Overall, children had a median age of 15 months (IQR: 10–22 months), and there was a nearly equal proportion of girls and boys. The distribution of children across anthropometric groups was as follows: 929 (39%) children with MUAC < 115 mm only, 530 (22%) children with WHZ < -3 only, and 940 (39%) children with both MUAC < 115 mm and WHZ < -3. One (<0.1%) child was confirmed to be HIV-positive.

To describe potential differences among children who would be included in a MUAC < 115 mm program vs. those who are currently eligible but who would be excluded using MUAC < 115 mm alone for admission, the first comparison presents children with MUAC < 115 mm (n = 1869) vs. children who would be excluded from a MUAC-based program having MUAC ≥ 115 mm and WHZ < -3(n = 530) (Table 1). Children with MUAC < 115 mm were more likely to be younger, female (57% vs. 25%) and stunted (81% vs. 73%), compared to children with MUAC ≥ 115 mm and WHZ < -3. Clinical characteristics on admission and program outcomes did not significantly differ between groups. The prevalence of bacterial infection was low in this population and did not differ between groups. Immediate response to treatment, as evidenced by the rate of weight gain from admission to week 1 and week 2, was greater among children with MUAC ≥ 115 mm and WHZ < -3, compared to MUAC < 115 mm children though ultimately there was no difference in length of stay between groups. Results remained largely consistent with adjustment for age, sex and stunting.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographic and clinical characteristics on admission and program outcomes of children with SAM, by MUAC- and WHZ-defined admission criteria.

https://doi.org/10.1371/journal.pone.0137606.t001

To describe potential heterogeneity within children who would be included in a MUAC < 115 mm program, the second analysis compared children with both MUAC < 115 mm and WHZ < -3 (n = 940) vs. children with MUAC < 115 mm and WHZ ≥ -3 (n = 929) (Table 2). Children with MUAC < 115 mm and WHZ < -3 were older and male; both groups were similar in terms of clinical characteristics on admission, except for a greater prevalence of diarrhea (34% vs. 29%) and bacteremia (5% vs. 2%) among children with MUAC < 115 mm and WHZ < -3. Compared to children with MUAC < 115 mm and WHZ ≥ -3, children with both MUAC < 115 mm and WHZ < -3 were less likely to recover, particularly children < 24 months of age (71% vs. 53%), and were more likely to be transferred to inpatient care due to lack of weight gain or weight loss. Immediate weight gain was greater among children with both MUAC < 115 mm and WHZ < -3 compared to children with MUAC < 115 mm and WHZ ≥ -3, though length of stay was greater.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Demographic and clinical characteristics on admission and program outcomes of children eligible for nutritional treatment using a MUAC-only admission criterion, by MUAC- and WHZ-defined admission criteria.

https://doi.org/10.1371/journal.pone.0137606.t002

Discussion

Our results show that children identified for therapeutic feeding with MUAC < 115 mm were clinically similar on admission to children currently eligible for treatment but excluded using the MUAC-based admission criterion: specifically, we found no difference between groups in terms of clinical and laboratory characteristics and program outcomes evaluated in this study population. Among children with MUAC < 115 mm, we found that those with both MUAC < 115 mm and WHZ < -3 presented with a more severe clinical condition, demonstrated by increased prevalence of diarrhea and bacteremia, and did not respond as well to treatment, evidenced by a lower risk of nutritional recovery and an increased risk of transfer to inpatient care, compared to children admitted with MUAC < 115 mm and WHZ ≥ -3.

Debate and interest surrounding the use of MUAC-only in nutritional programming continues, but a shift from therapeutic feeding programs admitting children using MUAC < 115 mm and/or WHZ < -3 to a new model admitting children using MUAC only remains complicated by limited information regarding the clinical profile and response to treatment of children selected by MUAC vs. WHZ. It has been shown that a single MUAC-based cut-off (e.g. independent of age and sex) is more likely to identify SAM children who are younger, female and stunted [19–21]. It has been suggested that such a demographic and anthropometric shift in the patient profile selected with a single MUAC-based admission criterion may appropriately identify children that are more vulnerable or in need of nutritional rehabilitation. We compared children who would be included in a MUAC-based program (e.g. MUAC < 115 mm as the single anthropometric criterion) with children who are currently eligible for treatment but would be excluded from a MUAC-based program (MUAC ≥ 115 mm and WHZ < -3) in terms of a broad range of clinical, and laboratory characteristics, as well as program outcomes, to better understand the operational implications of applying a MUAC-only anthropometric criterion for admission to therapeutic feeding in rural Niger. Our findings suggest the single criterion of MUAC < 115 mm for admission to therapeutic feeding identified few important differences between groups in this context. However, among those children who would be accepted into a MUAC-only program, we found children with both MUAC < 115 mm and WHZ < -3 were more likely to be clinically unwell at admission, indicated by increased diarrhea and bacteremia, and achieved poor program outcomes. These findings suggest that where MUAC < 115 mm is used as the single anthropometric criterion for admission to therapeutic feeding, the combination of anthropometric indicators may help to further target children at increased risk or need of additional follow-up. An upward adjustment of a single MUAC threshold above 115 mm could also be applied to increase sensitivity to select more children at high risk, but such adjustments would have important programmatic implications, including increased workload and costs, and should be context-specific.

We noted significantly greater initial weight gain (e.g. at week 1 and week 2) among children with MUAC ≥ 115 mm and WHZ <- 3 children compared to children with MUAC < 115 mm and considered several potential mechanisms or explanations of this finding. First, greater initial weight gain may reflect a greater need for nutritional therapy among children with MUAC ≥ 115 mm and WHZ <- 3. The rate of weight gains observed in this group is not inconsistent with mean weight gain observed during recovery of severely malnourished children in inpatient settings (7.8 to 10 g/kg/day, [26]) and early community-based therapeutic programs (3 to 6.8 g/kg/day, [27]). In this study population, MUAC ≥ 115 mm and WHZ <- 3 may characterize a sub-group of children with a pathophysiological condition rapidly corrected and responsive to rapid weight gain, or conversely, MUAC may characterize a sub-group of children suffering from more severe metabolic impairments that limit rapid tissue restoration. Second, it is possible the greater initial weight gain among children with MUAC ≥ 115 mm and WHZ <- 3 reflects rapid water repletion associated with correction of moderate dehydration not measured in this study. The greater effect of dehydration on WHZ assessment, compared to MUAC, and the inadequacy of WHZ to assess nutritional status of children with diarrhea or dehydration was been noted elsewhere [28, 29].

With the hypothesis that differences among children selected by MUAC < 115 mm vs. MUAC ≥ 115 mm and WHZ <- 3 may further differ by age, sex and stunting, we investigated potential effect modification by these factors. A recent study from McDonald et al. importantly highlighted an increased risk of mortality with multiple anthropometric deficits (WHZ < -2, HAZ < -2 and weight-for-age Z <-2), compared to single or no anthropometric deficits [30]. The contribution of anthropometric deficit defined by MUAC < 115 mm, however, was not considered in this study. In the inpatient study from Kenya, Berkley et al. found no difference in the risk of death between children identified as SAM by MUAC only vs. those identified as SAM by WHZ only, but those children identified by both criteria had an increased risk of mortality [19]. We, however, found no evidence for worse treatment outcomes or response among children with HAZ < -2 and WHZ < -3. Our results do suggest that other clinical differences in children selected by low MUAC and/or low WHZ may be more pronounced in particular subgroups defined by age, sex or stunting. Identification of such subgroups can provide a more precise understanding of how anthropometric criteria may be selecting different profiles of vulnerable children.

Published data on potential variability among children selected by different anthropometric criteria remains very limited to date. The most detailed study available was done in 1999–2002 among 8,190 children hospitalized in rural Kenya [19]. This study similarly suggested that MUAC was more likely to identify severely malnourished children that were younger, female, and stunted. It also showed that mortality risk was similar in children with MUAC < 115 mm and children with MUAC ≥ 115 mm and WHZ <- 3, yet higher in children presenting both criteria at admission. Our results similarly suggest a higher vulnerability in children with both criteria compared with children with MUAC < 115 mm and WHZ ≥ -3. In comparing the clinical features of the children identified by MUAC < 115 mm only and by WHZ < -3 only, this study in Kenya also reported higher prevalence of seizures during current illness and impaired consciousness in low WHZ-only children but fewer other clinical signs (including cough, diarrhea, sub-costal indrawing and anemia). The comparison of these findings with the findings from our outpatient population in Niger is, however limited, given the specific definition of anthropometric groups for comparison and differences between the inpatient vs. outpatient study populations.

Routine data collected from two Médecins Sans Frontières (MSF) therapeutic feeding programs have recently been analyzed to explore this subject. In Burkina Faso, Goossens et al. analyzed data from 24,792 children admitted to a therapeutic feeding program using MUAC ≤ 118 mm as the sole anthropometric admission criterion. This study showed greater weight gain associated with lower MUAC at admission [12]. Program records from 2,205 children admitted to a therapeutic feeding program in South Sudan were also analyzed to compare children who would have been admitted to a MUAC-only program to those who would have been excluded from a MUAC only program but are included under current WHO guidelines. Like in our study, MUAC-eligible children were younger and more likely to be female. However, they were more severely malnourished and at a greater risk of death, but more responsive to treatment than those in the low WHZ-only group.(24) Although the tall and thin body shape of some pastoralist populations, as found in the latter South Sudan program, is known to influence the performance of different anthropometric indicators, the comparison of demographic characteristics from Niger are consistent with those reported by Grellety et al. from South Sudan.

This study is the first to date to compare the clinical and laboratory characteristics associated with the type of anthropometric diagnosis among a large population of children with uncomplicated SAM children admitted to an outpatient nutritional program. Characterization of individual vulnerability was informed by the reliable assessment of clinical and laboratory characteristics at admission, as well as through a careful follow-up over the course of the treatment. This study offers new information to extend current evidence on the potential implications of moving towards MUAC as the sole anthropometric admission criterion for therapeutic feeding.

This study does, however, have some limitations. In particular, laboratory analyses did not include parasitological or virological exam or confirmation of malaria infection. The prevalence of bacteriological infection and mortality were low, limiting the interpretation of these results. Additionally, this analysis draws from data collected during a randomized trial among children presenting to a health facility with SAM and receiving treatment. The study population does not reflect an untreated or general population, or a nutritional program that draws its participants from community-based screening and referral. These data therefore do not allow a clear conclusion regarding the use of using MUAC-only based criterion among untreated children in the community.

Therapeutic feeding programs must aim to target and treat malnourished children that are the most vulnerable and at the highest risk of death, but programs are often challenged to do so in highly resource-constrained settings. It was believed that a single MUAC-based anthropometric admission criterion, in addition to its operational advantages, would identify the most vulnerable children in need of treatment. Our study suggests that a single anthropometric admission criterion of MUAC < 115 mm identified younger and more female children but did not well differentiate children in terms of other measures of vulnerability. Children with MUAC < 115 mm and children with MUAC ≥ 115 mm and WHZ < -3 in rural Niger displayed similar characteristics in terms of their clinical and laboratory profile on admission. An upward adjustment of the MUAC threshold could yield a more sensitive criterion for admission to select children at high risk but would have important programmatic implications. For MUAC to be considered as a single anthropometric admission criterion for therapeutic feeding, experience with various MUAC-based thresholds and supporting evidence on the clinical characteristics and needs of children selected with each should be further evaluated in a broad range of contexts.

Supporting Information