Inpatient management of children with severe acute malnutrition: a review of WHO guidelines

Abstract Objective To understand how the World Health Organization’s (WHO’s) guidelines on the inpatient care of children with complicated severe acute malnutrition may be strengthened to improve outcomes. Methods In December 2015, we searched Google scholar and WHO’s website for WHO recommendations on severe acute malnutrition management and evaluated the history and cited evidence behind these recommendations. We systematically searched WHO International Clinical Trials Registry Platform, clinicaltrials.gov and the Controlled Trials metaRegister until 10 August 2015 for recently completed, ongoing, or pending trials. Findings WHO’s guidelines provide 33 recommendations on the topic. However, 16 (48.5%) of these recommendations were based solely on expert opinion – unsupported by published evidence. Another 11 (33.3%) of the recommendations were supported by the results of directly relevant research – i.e. either randomized trials (8) or observational studies (3). The other six recommendations (18.2%) were based on studies that were not conducted among children with complicated severe malnutrition or studies of treatment that were not identical to the recommended intervention. Trials registries included 20 studies related to the topic, including nine trials of alternative feeding regimens. Acute medical management and follow-up care studies were minimally represented. Conclusion WHO’s guidelines on the topic have a weak evidence base and have undergone limited substantive adjustments over the past decades. More trials are needed to make that evidence base more robust. If the mortality associated with severe malnutrition is to be reduced, inpatient and post-discharge management trials, supported by studies on the causes of mortality, are needed.


Introduction
Each year, severe acute malnutrition -defined as a weight-forheight z-score of less than −3 or a mid-upper arm circumference of less than 115 mm -is the direct cause of an estimated 540 000 child deaths and an important underlying contributor to many other child deaths, especially those due to pneumonia and diarrhoea. 1,2 The prevalence of -and case fatality rate for -malnutrition are particularly high in infants. 3,4 Severe acute malnutrition without medical complications can now be effectively managed in the community, with ready-to-use therapeutic foods. 5 The presence of complications such as anorexia, infections or metabolic dysfunction still warrants inpatient management. The World Health Organization (WHO) indicates that, by following its inpatient management guidelines, less than 10% of children with complicated severe acute malnutrition should die. 2 However, despite reported compliance with these guidelines, health centres in sub-Saharan Africa have reported mortality rates of 10-40% among severely malnourished hospitalized children. 3,6 The corresponding published rates in Asia tend to be lower, possibly because: Asian health systems are generally stronger than their African counterparts; therapeutic innovations have been introduced in some Asian facilities that have not yet been used in Africa; and children with uncomplicated severe malnutrition are sometimes admitted to Asian health facilities -but not, generally, to African health facilities. 7,8 In sub-Saharan Africa, human immunodeficiency virus (HIV) is believed to contribute greatly to malnutritionrelated mortality -although, a recent meta-analysis demonstrated an overall case fatality rate of 15% among paediatric inpatients with severe malnutrition without HIV infection. 6 It is unlikely that severe acute malnutrition will be eliminated in the foreseeable future, as preventative interventions would have to reach the 52 million children who have moderate acute malnutrition. 1 Optimizing the management of complicated severe malnutrition therefore remains an important strategy for reducing malnutrition-related mortality.
WHO's first guidelines on the management of malnutrition -published in 1981 and focused on protein-energy malnutrition 9 -were replaced in 1999 by guidelines on the management of severe acute malnutrition. 10 These two documents summarized decades of clinical experience and described the achievement of low malnutrition-related case fatality rates in some specific settings. [9][10][11] Further guideline revisions were made in 2003 2 and 2013. 4 Relevant joint statements from WHO and other United Nations agencies were issued in 2007 5 and 2009. 12 The combination of these joint statements, the 1999 guidelines and the revisions of 2003 and 2013 constitutes the current WHO severe acute malnutrition guidelines and underpins WHO's related training material. 13 Although weak health systems and the inadequate implementation of guidelines undoubtedly contribute to the high number of preventable deaths attributed to complicated severe acute malnutrition, 14 this condition causes high case fatality even in relatively well resourced centres that report full implementation of the WHO guidelines. This review attempts to identify evidence gaps within the guidelines on inpatient management of severe acute malnutrition that, if filled, may help reduce mortality below the levels that can be accomplished solely by adherence to existing guidelines. We reviewed each individual recommendation contained within current WHO guidelines -including those relating to the post-discharge Objective To understand how the World Health Organization's (WHO's) guidelines on the inpatient care of children with complicated severe acute malnutrition may be strengthened to improve outcomes. Methods In December 2015, we searched Google scholar and WHO's website for WHO recommendations on severe acute malnutrition management and evaluated the history and cited evidence behind these recommendations. We systematically searched WHO International Clinical Trials Registry Platform, clinicaltrials.gov and the Controlled Trials metaRegister until 10 August 2015 for recently completed, ongoing, or pending trials. Findings WHO's guidelines provide 33 recommendations on the topic. However, 16 (48.5%) of these recommendations were based solely on expert opinion -unsupported by published evidence. Another 11 (33.3%) of the recommendations were supported by the results of directly relevant research -i.e. either randomized trials (8) or observational studies (3). The other six recommendations (18.2%) were based on studies that were not conducted among children with complicated severe malnutrition or studies of treatment that were not identical to the recommended intervention. Trials registries included 20 studies related to the topic, including nine trials of alternative feeding regimens. Acute medical management and follow-up care studies were minimally represented. Conclusion WHO's guidelines on the topic have a weak evidence base and have undergone limited substantive adjustments over the past decades. More trials are needed to make that evidence base more robust. If the mortality associated with severe malnutrition is to be reduced, inpatient and post-discharge management trials, supported by studies on the causes of mortality, are needed.
care that forms an integral extension of hospital management. We traced the lineage and quantified the evidence cited in support of each recommendation. We also searched trials registries systematically, to determine which evidence gaps may be closed by the results of ongoing or recently completed trials.

Methods
We identified WHO's recommendations for severe acute malnutrition management by searching Google Scholarusing "severe acute malnutrition" and "author:WHO" as the search terms -and by downloading the publications on the WHO nutrition website in December 2015. 15 Full texts were reviewed if they represented a relevant guideline -as classified by WHO's guideline review committee -or guideline update. 16 Documents specific to humanitarian crises and those without inpatient or post-discharge management recommendations -e.g. the 2007 and 2009 joint statements 5,12 -were excluded. Each included guideline was parsed into individual recommendations. We excluded diagnosis and admission criteria and care principles that are applicable to all hospitalized children -e.g. the monitoring of blood glucose after treatment of hypoglycaemia.
We traced each recommendation's evolution through the development of the guidelines and noted any modifications and references cited in support of the recommendation. The full texts of all potentially relevant citations were reviewed. To determine the origins of each recommendation further, we reviewed three documents predating the current guidelines: WHO's 1981 severe protein-energy malnutrition recommendations, 9 and two textbooks commonly used before the publication of WHO's first severe malnutrition guidelines in 1999. 17,18 According to formal GRADE (grading of recommendations, assessment, development and evaluation) assessment, each of the recommendations we investigated was of low, very low or unclassifiable quality. We evaluated each recommendation using GRADE's directness assessment, as this provided meaningfully differentiated categories of evidence quality. 16 Recommendations that were not supported by any cited evidence were considered to be based entirely on expert opinion. Recom-mendations were defined as indirectly supported if all of the cited studies were either among populations other than children with complicated severe acute malnutrition -e.g. HIV care guidance derived from studies of HIV-infected children without concurrent malnutrition -or only based on a treatment that was similar, but not identical, to the WHO recommended treatment -e.g. commercial ready-to-use therapeutic foods recommended on the basis of trials of locally produced versions of such foods. If at least one study concerning the endorsed intervention in a population of children with complicated severe acute malnutrition was cited in support of a recommendation, then that recommendation was considered to be directly supported. Direct evidence was further categorized as an observational study or a randomized trial.
To determine the aims and extent of any recently completed, ongoing, or pending trials relevant to the management of complicated severe acute malnutrition, we searched the WHO International Clinical Trials Registry Platform, the United States National Institutes of Health's clinicaltrials. gov database and the Controlled Tri-als metaRegister systematically, using the search terms "malnutrition" and "wasting". We completed this search on 10 August 2015. We fully reviewed all records with relevant or non-specific titles and included interventional trials among children with complicated severe acute malnutrition. We excluded trials already cited in WHO guidelines and those stopped before subject enrolment. We investigated the publication status and results of relevant trials by searching PubMed for the corresponding registration numbers.

Results
Eight documents containing 33 current recommendations met our inclusion criteria (Fig. 1). 2,9,10,[17][18][19][20] The lineage of the 33 recommendations is summarized in Table 1. Expert opinion, in the absence of published evidence, was the basis for 16 (48.5%) of the recommendations. Three (9.1%) and six (18.2%) of the recommendations were drawn from direct observational or indirect evidence, respectively. The remaining eight recommendations (24.2%) were each supported by at least one direct randomized trial.  Twenty-three (69.7%) recommendations had been added or revised since the original guideline published in 1999. 10 Only six (26.1%) of these 23 were supported by a directly relevant randomized trial. Three (13.0%) and six (26.1%) were supported by at least one direct observational or indirect study, respectively, while no references were cited in support of the remaining eight (34.8%) recommendations. The 1999 guidelines 10 presented a 10-step management protocol -as originally proposed in the article Ten steps to recovery that was published in 1996. 11 Five (15.2%) of the 33 current recommendations are identical to -or slight modifications of -the recommendations first proposed in this 1996 article. Seven (21.2%) of the current recommendations originated before 1996 -although five of these have since been slightly revised.

Recommendation age and quality
The age of the recommendation and quality of supporting evidence varied according to the involved clinical area.

Micronutrients
Micronutrient recommendations were largely based on expert opinion, although three randomized trials [23][24][25] directly supported two of the recommendations made in the 2013 update: low-dose vitamin A administration, reserving high-dose vitamin A for those with eye signs of deficiency or measles. 4 Collectively, the trials demonstrated that either dose of vitamin A was superior to placebo, and that high-dose vitamin A offered no benefit compared with low-dose and might be associated with nosocomial diarrhoea and pneumonia. The 2013 update also recommended that HIV-infected children receive the same zinc and vitamin A doses as uninfected peers. 4 This recommendation was supported by a systematic review of studies among HIV-infected children and adults without malnutrition, which indicated that HIV infection should not alter zinc requirements. 26 Specific recommendations on the broader micronutrient package, which have remained constant for over 20 years, are all based on expert opinion.

Feeding
Indirectly related studies were the predominant reference type cited in support Recommendations for the use of therapeutic milk feeds -i.e. F-75 and F-100 -and the criteria for transition to ready-to-use therapeutic foods were last updated in 2003 2 and were based on the results of six studies. Five of these studies demonstrated an association between refeeding syndrome and death among adolescents with eating disorders, children with neurological dysphagia, children with parent-imposed starvation, and critically ill adults in high-income settings. [27][28][29][30][31] The 2013 update 4 advised against use of undiluted F-100 among young infants, based on a direct study that indicated a possible connection between undiluted F-100 and renal solute overload, hypernatraemia and death. 32 Specific advice on breastfeeding has remained largely unchanged for almost half a century. 18

Fluid management
Three of the six recommendations on fluid management -including the specification of low-osmolarity salts for cholera -had been revised in the 2013 update. 4 Recommendations for the treatment of shock or severe dehydration underwent a relatively minor re-ordering in the preference of intravenous fluids, based on a direct randomized trial of 62 children, in which Ringer's lactate solution with 5% dextrose was compared with half-strength Darrow's solution with 5% dextrose. Neither of these fluids was found to correct shock sufficiently and the choice of fluid had no significant effect on mortality. 33 Finally, the study that was cited in support of limiting the timing, indications and infusion rates for transfusions demonstrated a strong association between mortality and transfusion -although adjustment for confounding by indication may have been insufficient. 34

Antiretroviral treatment
Although three recommendations on antiretroviral treatment were added in the 2013 update, 4 none was supported by direct evidence. Antiretroviral initiation recommendations referenced WHO's guidelines on the management of childhood HIV infection. 21,22 The advice to initiate antiretrovirals after clinical stabilization cited two pharmacokinetic studies among children with varying degrees of malnutrition 35,36 and one retrospective study that demonstrated faster recovery when antiretroviral treatment was initiated within 21 days of the diagnosis of uncomplicated severe malnutrition. 37

Other clinical problems
Recommendations on the management of hypoglycaemia, hypothermia and acute infections -including specifics related to antimicrobial treatmentwere made in the Ten steps to recovery article. 11 They remain unchanged and are not supported by any cited evidence.

Discharge and follow-up
Six recommendations on discharge from hospital and outpatient care were added in the 2013 update 4 and were almost exclusively drawn from expert opinion. Supporting citations were limited to two indirect retrospective studies demonstrating that mid-upper arm circumference was an adequate measure of outpatient progress. 38,39 The results of these studies led to the recommendation to eliminate percentage weight gain as a criterion for discharge from outpatient follow-up.

Ongoing or recent trials
Our search of trials registries yielded the full records of 58 trials -after review of trial titles (Fig. 2). Twenty of these trials met our inclusion criteria ( Table 2). Fifteen of the 20 trials had been completed -and the results of four had been published -by the time of our search. [40][41][42][43] Two had reported statistically significant results; one demonstrated that community follow-up increased linear growth and clinic attendance 43 and the other that long-chain n-3 polyunsaturated fatty acid in erythrocytes increased among severely malnourished children who were given readyto-use therapeutic food enriched with polyunsaturated fatty acid. 42 The other two published trials, which detected no significant differences, compared alternative formulations of ready-to-use therapeutic food with standard formulations. Of the 16 unpublished trials, nine and two had been designed to investigate alternative feeding regimens and the use of probiotics, respectively. One each had been designed to investigate pancreatic enzyme replacement, antioxidants, intravenous rehydration, stool output assessment, and antiretroviral pharmacokinetics. Three unpublished antibiotic trials -completed between 2008 and 2014 -examined ciprofloxacin pharmacokinetics, ceftriaxone for concurrent pneumonia, and post-discharge prophylaxis with co-trimoxazole (Table 2).

Discussion
The 2013 update stated that "major research gaps were identified in each of the sections covered". 4 Our analysis shows that such gaps persist and extend across the entire spectrum of guidance on the management of complicated severe acute malnutrition. The absence of relevant published data has forced a reliance on expert opinion. The evidence that was cited in support of many recommendations was of very low quality and often did not specifically pertain to the recommended treatment. These deficits demonstrate that guideline reforms have been driven by an overwhelming clinical need -rather than by a body of compelling evidence. This is not criticism of WHO or the guidelines' authors, who should be commended for creating pragmatic management documents by threading together the little solid evidence available and expert opinion. It should be noted that recommendations supported by weak evidence or expert opinion are not necessarily incorrect. Many of the recommendations are grounded in the results of basic science research and careful clinical observations, much of which was made before the 1996 seminal Ten steps to recovery article. However, the population of paediatric inpatients with severe malnutrition has dramatically changed in the last 10-20 years. Over that period, HIV has emerged as an important contributing problem, younger infants have come to represent an increasing proportion of malnourished children and, for cases without complications, outpatient care has eclipsed hospital management. 3,4 Data from previous eras may therefore not be generalizable to the modern child with complicated severe acute malnutrition.
In some areas the absence of clinical data is particularly concerning. For example, given that the largest burden of mortality from malnutrition is in sub-Saharan Africa, where the prevalence of HIV infection is relatively high, the lack of evidence to guide the management of HIV-infected children with malnutrition is worrying. 1 A 2009 meta-analysis found that, among severely malnourished children, HIV infection was associated with a threefold increased risk of mortality. 6 In a cohort study of severely malnourished children admitted to Queen Elizabeth Hospital in Malawi, HIV-infected children represented 64% of the deaths. The same study found that 67% of infants died. 3 In the absence of data addressing these two populations -i.e. young infants and HIV-infected children with complicated severe malnutrition -the guidelines' authors have been forced to generalize the management practices from other populations, without evidence that this is optimal or even appropriate. 4 Furthermore, in the Malawian study, 25% of the children who were discharged died in the following 12 months and these deaths represented 44% of the total recorded mortality. 3 Post-discharge mortality rates are high and their causes are poorly understood. This knowledge gap warrants urgent attention.
Antimicrobial therapy for severely malnourished inpatients represents another conspicuous knowledge gap. Empiric antibiotics have been recommended since at least 1969 18 and the currently endorsed regimen has remained unchanged since it was standardized to ampicillin and gentamicin in 1996. 11 A 1996 trial demonstrated the superiority of ampicillin and gentamicin compared with previously endorsed protocols relying on co-trimoxazole or penicillin and gentamicin. 44 We are not aware of any subsequent studies comparing the currently recommended regimen with other antimicrobials. In the care of severe acute malnutrition, fluid man- For ethical or practical reasons, some guidance areas are simply not amenable to clinical trials. Beyond vitamin A and zinc, the composition of the recommended micronutrient package has remained constant for decades and is underpinned by estimated daily requirements and observed micronutrient deficiencies in severe malnutrition. It would be impractical to run factorial trials for all micronutrients. However, this does not preclude the improvement of micronutrient packages via observational studies and targeted clinical trials, particularly given the etiological, environmental and social changes that have occurred in the decades since the foundational micronutrient research was conducted.
Our quantitative gap analysis builds on other reviews, 45,46 including a 2012 systematic review 46 -conducted in preparation for WHO's 2013 updatethat concluded: "For many of the most highly ranked questions evidence was lacking or inconclusive". 4 Our review of trials registries revealed that most of the upcoming, ongoing or recently completed relevant trials were focused on feeding and nutrition -areas not identified as priorities in the 2013 update 4 or supporting systematic review. 46 Furthermore, the areas in highest need of evidence -e.g. intravenous fluids, antimicrobials, the treatment of infants younger than six months and HIVinfected children, and post-discharge management -were very modestly represented. Therefore, it is unlikely that currently registered trials will address many of the critical knowledge gaps related to the inpatient management of severe acute malnutrition.
The paucity of relevant research may arise from a misconception that further work is unnecessary because adequate guidelines exist. As we have described, however, the underlying evidence for most management areas is weak. The current global undernutrition research agenda is largely focused on reducing the burden of stunting and moderate acute malnutrition -important and justifiable areas of concern. However, we should not neglect the half a million children who die from severe acute malnutrition annually. 1 Epidemiological studies that define the etiologies of illness and mortality form an important foundation for interventional trials. Recent such studies have challenged long-standing beliefs about the relative importance of specific pathogens in pneumonia and diarrhoeal disease. For example, a South African study found cytomegalovirus, Mycobacterium tuberculosis and Pneumocystis jiroveci to be frequent causes of treatment failure among children with severe pneumonia 47 while the Global Enteric Multicentre Study demonstrated Cryptosporidium, enterotoxigenic Escherichia coli, rotavirus and Shigella to be leading causes of childhood diarrhoeal death. 48 These findings have spurred interventional trials that hopefully will improve management and save lives. A few studies have evaluated the causes of infection or death among children admitted for severe acute malnutrition. 49,50 However, we are aware of no recent or robustly sampled investigation of the causes -including non-infectious etiologies -of mortality among such children during hospitalization or postdischarge. The failure of many trials to find statistically significant results may stem from a superficial understanding of the contemporary etiologies of such mortality. For example, if children with environmental enteric dysfunction require specific treatment more than children who are affected by food insecurity alone, then including both groups of children in trials of readyto-use therapeutic foods could lead to attenuated estimates of efficacy and reduced statistical power. An improved understanding of the epidemiology of complicated severe acute malnutrition will facilitate the efficient design of clinical trials and catalyse the discovery of new and effective interventions.
This paper is not a detailed systematic review but rather a tracing of the lineage of each recommendation and its supporting citations. We did not review evidence that was not referenced in the relevant WHO guidelines. Early guidelines were published at a time when evidence citation was uncommon. Any relevant evidence available to these guidelines' authors will not have been captured by our review unless cited in subsequent updates. Additionally, it is impossible to quantify the cumulative clinical experience of the many experts who have contributed to the guidelines. This paper does not address why severely malnourished children admitted to Asian hospitals seem to experience different mortality rates to their counterparts in African facilities. It would be useful to determine, in various settings, the proportion of cases of severe malnutrition that present with complications.