Oral rehydration of malnourished children with diarrhoea and dehydration: A systematic review

Background: Diarrhoea complicates over half of admissions to hospital with severe acute malnutrition (SAM). World Health Organization (WHO) guidelines for the management of dehydration recommend the use of oral rehydration with ReSoMal (an oral rehydration solution (ORS) for SAM), which has lower sodium (45mmols/l) and higher potassium (40mmols/l) content than old WHO ORS. The composition of ReSoMal was designed specifically to address theoretical risks of sodium overload and potential under-treatment of severe hypokalaemia with rehydration using standard ORS. In African children, severe hyponatraemia at admission is a major risk factor for poor outcome in children with SAM complicated by diarrhoea. We therefore reviewed the evidence for oral rehydration therapy in children with SAM. Methods: We conducted a systematic review of randomised controlled trials (RCTs) on 18 th July 2017 comparing different oral rehydration solutions in severely malnourished children with diarrhoea and dehydration, using standard search terms. The author assessed papers for inclusion. The primary endpoint was frequency of hyponatraemia during rehydration. Results: Six RCTs were identified, all published in English and conducted in low resource settings in Asia. A range of ORS were evaluated in these studies, including old WHO ORS, standard hypo-osmolar WHO ORS and ReSoMal. Hyponatraemia was observed in two trials evaluating ReSoMal, three children developed severe hyponatraemia with one experiencing convulsions. Hypo-osmolar ORS was found to have benefits in time to rehydration, reduction of stool output and duration of diarrhoea. No trials reported over-hydration or fatalities. Conclusions: Current WHO guidelines strongly recommend the use of ReSoMal based on low quality of evidence. Studies indicate a significant risk of hyponatraemia on ReSoMal in Asian children, none have been conducted in Africa, where SAM mortality remains high. Further research should be conducted in Africa to evaluate optimal ORS for children with SAM and to generate evidence based, practical guidelines


Introduction
In Africa, diarrhoea has been reported to complicate 49% of admissions to hospital of children with severe acute malnutrition (SAM), and a further 16% develop diarrhoea within 48 hours of admission. The in-hospital case fatality in children with SAM admitted with diarrhoea is high, 19%, versus 9% in those without diarrhoea, (χ 2 = 17.6 p<0.001) and no prospect of improvement has been demonstrated over the last decade [1][2][3][4] .
Management of children with SAM complicated by diarrhoea focuses on exclusive oral or nasogastric (NG) rehydration, and limits intravenous rehydration to those complicated by advanced hypovolaemic shock or for those with severe dehydration who are unable to take or tolerate oral fluid 5,6 . The World Health Organization (WHO) guidelines are used widely in low resource settings as the standard of care, recommending that oral or NG rehydration fluids can be commenced for any child with SAM and diarrhoea (defined as three or more loose, watery stools) ( Table 1). The guidelines do not allow for an assessment of severity of dehydration in children with SAM, indicating that dehydration is often difficult to diagnose in malnourished children because the clinical signs usually relied on to diagnose dehydration are similar to those found in severe wasting without dehydration 5 . However, the available evidence contradicts this contention. In a prospective study involving 920 unselected Kenyan children admitted to hospital with SAM, sepsis, signs of severe dehydration (secondary to diarrhoea) and hypovolaemic shock were common complications and were triage features associated with high early fatality (>20% mortality) 7 . Another prospective observational study conducted at the same centre examined in more detail diarrhoea in malnutrition, and multivariate analysis identified bacteraemia (odds ratio 6.7 (95% confidence interval 2.5-17.8 p<0.001) and hyponatraemia (odds ratio 4.9 (95% CI 2.2-11.1 p<0.001) as key risk factors for mortality 1 .
Only a very small number of children with signs of advanced shock are recommended to receive intravenous (IV) fluids, 15ml/Kg of hypotonic fluid, followed by a blood transfusion if there is no improvement ( Recommendations suggest avoiding IV fluids in children with SAM due to concerns about the ability of these children to handle significant volume loads and potential susceptibility to fluid overload and cardiac failure. However, available evidence suggests that the perturbations of myocardial function are related to complications of sepsis, shock and severe dehydration and not due to 'heart failure' 9,10 . A recent publication by Obonyo et al. 2017 demonstrated 'fluid responsive' myocardial indices following rehydration in children with SAM and hypovolaemic shock 11 .

Types of Oral Rehydration Solutions
The original or 'old' oral rehydration solution (ORS) recommended by WHO was designed largely to treat children with cholera and thus had a high sodium content (since cholera is a secretory diarrhoea with large losses of both sodium and water). At that time WHO guidance advised that children with SAM should be given a modified version of oral rehydration solution (ORS) called ReSoMal (rehydration solution for malnutrition), which has lower sodium, higher potassium and glucose and lower osmolarity than Old WHO ORS (Table 2) 5,6 . This was due to concerns that 'children with bilateral pitting oedema typically have high intracellular sodium and are therefore inclined to retain fluids' and 'are prone to fluid retention and susceptible to fluid changes', thereby predisposing the child to fluid overload and heart failure 6 . Whilst this was suggested by Wharton et al., who

Amendments from Version 2
One of the reviewers noted an error in Table 3 and the references in the Discussion section; we have revised the manuscript to correct these errors.

REVISED
reported an excess of heart failure in children receiving a high energy milk to which sodium was added 13 , apparently to improve acceptability in all children (including those without diarrhoea) the sodium content was probably higher and the milk given over a longer period of time than they would have received if they were only being rehydrated. Thus, this cannot be extrapolated to inform management of rehydration in children with SAM. Furthermore, an observational study has reported that in SAM children receiving liberal ReSoMal leads to excess mortality due to heart failure due to excessive sodium intake as a reference source to support these recommendations. However, this does not qualify as sufficient evidence to inform the management of African children who have diarrhoea 14 . Other than these observations we are have found no physiological data published to support this contentious opinion that children with severe malnutrition are prone to sodium overload, even for the sub group of greatest concern, kwashiorkor 15,16 .
The current standard (hypo-osmolar) WHO ORS, with lower sodium and glucose content, was developed in order to reduce the intensity of diarrhoea in children. A meta-analysis of 8 trials, showed that reduced osmolarity ORS was associated with fewer unscheduled intravenous fluid infusions (the primary endpoint) compared with WHO standard ORS and stool output, reported in eleven trials, was less in the reduced osmolarity ORS group 17 . Hypo-osmolar ORS has now been adopted into practice and has largely replaced the old ORS formulation and is recommended in current WHO paediatric management guidelines 12 . The Standard (hyposmolar) WHO ORS recommended for non-SAM children is therefore closer in composition to ReSoMal with respect to sodium content but has a much lower osmolarity (due to the lower glucose content) than ReSoMal which means theoretically that it may have less potential to exacerbate stool volume and diarrhoea in children with SAM.
The 'strong' recommendations for rehydration of children with SAM are informed by a nutritional specialist group for the WHO and are based on expert opinion, since the review of the data indicated low quality of evidence 6 . The most recent updates to WHO guidelines in 2013 did not revise any of their recommendations, with the exception of the addition of a single 15ml/kg bolus of hypotonic intravenous fluid for severely dehydrated children unable to tolerate oral rehydration (as per shock management) 6 . No further IV rehydration beyond this was considered, with most rehydration strategies focused on oral rehydration. Owing to the poor outcomes recognised in African children with SAM complicated by diarrhoea, we therefore conducted a systematic review of the current available evidence underlying oral rehydration solutions for children with dehydration and severe acute malnutrition.

Objectives
To conduct a critical appraisal of available evidence evaluating the use of ReSoMal and hypo-osmolar ORS in the treatment of dehydration in children with SAM.

Methods
We did not publish a protocol prior to conducting this review. A search of online literature was performed. There were predetermined criteria, as detailed below for eligibility of studies, data outcomes, and an assessment of risk of bias and study method quality in each of the identified studies.

Selection criteria
Population. Children aged 0 to 12 years with SAM requiring oral rehydration solution for management of dehydration secondary to gastroenteritis. We used the WHO definitions for malnutrition (Weight-for-height Z score (WHZ) <-3, mid-upper arm circumference (MUAC) <115mm or oedema consistent with kwashiorkor), gastroenteritis (dehydrating diarrhoea, >3 loose stools per day) and for dehydration. We excluded studies with chronic or persistent diarrhoea lasting ≥ 14 days.

Intervention and comparison.
All studies that compare two or more different ORS were included. Studies were excluded if they considered rehydration in children without severe malnutrition, only considered rehydration via the IV route or only included patients with congenital heart disease, trauma, or diabetic ketoacidosis.
Outcome. Clinical trials that reported on any outcomes were included. The primary outcome for this review was frequency of hyponatraemia (sodium concentration <135mmol/L) during and after rehydration therapy. Secondary outcomes were all cause mortality, time to rehydration, stool output, frequency of fluid overload and frequency of oral rehydration failure.

Study design.
Only randomised-controlled trials (RCTs) were included.

Search methods
Online database search. A comprehensive literature search of the following databases was conducted on the 18 th July 2017 using the English search terms 'malnutrition' AND 'children' AND 'rehydration' AND 'oral': • PubMed/ Medline • Global Health Library (Virtual Health Library) • Cochrane Database of Systematic Reviews  WHO guidelines Children with SAM should be rehydrated slowly, either orally or by nasogastric tube, using WHO standard oral rehydration solution (ORS), 5ml/Kg every 30 minutes for the first 2 hours and then 5-10ml/Kg/hour up to a maximum of 10 hours.
ReSoMal should not be given if the child has suspected cholera or has profuse watery diarrhoea.
Intravenous fluid should not be used unless the child has shock and cannot be rehydrated orally or by nasogastric tube.
Evidence for oral rehydration  19 showed no significant differences in hyponatraemia between patients with cholera treated with ReSoMal or standard ORS. Notably, the patient who developed hyponatraemic seizures did not have cholera.

Implications in practice
There appears to be no additional significant benefit to using hypo-osmolar ORS or ReSoMal in comparison with standard WHO ORS.
Guidelines for children with SAM and suspected or confirmed cholera are identical to those with non-cholera diarrhoea other than the ORS used. Cholera is a secretory diarrhoea with high stool volume output and the current restrictive guidelines may therefore result in under treatment of children with dehydration. The authors screened the results of the literature search for studies that met the inclusion criteria as determined by the PICOS outline.

Study selection
The search produced 432 studies ( Figure 1). After screening and evaluation, six studies were identified that investigated ORS in children with SAM complicated by dehydration, incorporating a total of 686 children. All six of these studies were conducted in Asia, four of which were conducted at the International Centre for Diarrhoeal Disease Research, Bangladesh (ICDDR, B) 18-21 and two in India (New Delhi 22 and Calcutta 23 ). One study included children with and without SAM 18 , but reported independently on outcomes for children with SAM. One study included children with cholera only 20 (see Box 1 for further details). There was moderate heterogeneity in the population eligibility criteria, sample size, and methods employed by each study, and in their results. Table 3 and Table 4 show the setting, methodology and   Table 5 shows the formulations of ORS used in the studies.

Risk of bias
The quality of each of the included studies was assessed for risk of bias using the Cochrane collaboration's tool in order to evaluate validity. Four studies had a low risk of bias 19-22 and two had lowmoderate risk of bias 18,23 due to lack of pre-determined outcomes in the methods.    21 . The hypothesis being tested in these trials was whether these fermentable carbohydrates, that form short chain fatty acids in the colon, could improve gut barrier function by providing energy to the colon and improving overall metabolism and reduce the duration of diarrhoea. These have the potential to reduce overall stool volume and recovery time and could be considered as candidates for future trials in African children with mortality as a key endpoint.

Primary outcome
Unlike the studies reported in Asia, diarrhoea comorbidity in African children hospitalised with SAM has a poor prognosis, with a case fatality rate of 18-20% 1 . A large prospective study investigating risk factors for mortality in 1206 Kenyan children with SAM and diarrhoea at admission to hospital (≥3 watery stools/day) showed that both hyponatraemia and hypokalaemia were associated with a greater risk of mortality: hyponatraemia odds ratio 4.6 (95% CI 2.0,10.6, p<0.001) and hypokalaemia odds ratio 2.5 (95% CI 1.3, 4.6, p<0.004) 1 . Hyponatraemia has nearly twice the impact on risk of mortality when compared with hypokalaemia; therefore, it would be prudent to place the importance of sodium status ahead of potassium. By this deduction, there is no clear advantage of ReSoMal over standard or hypo-osmolar rehydration solutions in terms of sodium status. Conversely, there are risks of serious harm through development of symptomatic hyponatraemia.
These findings highlight the lack of compelling evidence to support the current rehydration guidelines for management of children with SAM complicated by diarrhoea. Hypo-osmolar solutions have no apparent benefit on sodium status, i.e. there is no significant difference in numbers with hyponatraemia after rehydration, but do have significant improvements on reduction of stool volume and frequency, and on duration of diarrhoea. It is unclear, however, whether this translates to a survival advantage. Just two of the studies reported on mortality, and no deaths were observed in either. The mortality rate in children with SAM and diarrhoea in African children is substantially higher than reported from the studies included in this review. It would therefore be useful to conduct similar trials in children in the African continent.

Reappraisal of current guidelines
The WHO reviewed the guidelines for management of SAM in 2013 and identified five papers (four of which have been included in this review). The review discusses evidence presented by the papers and, despite universally 'low quality of evidence', continues to make 'strong recommendations': choosing only to amend the SAM management guidelines to allow the use of hypo-osmolar ORS and advise that 'either ReSoMal OR half strength standard WHO low-osmolarity oral rehydration solution with added potassium and glucose should be given' to children with some or severe dehydration. The review also emphasises that ReSoMal 'should not be given if children are suspected of having cholera or profuse watery diarrhoea'.

Application of guidelines
In practice, according to the WHO guidelines for management of children with SAM and diarrhoea, after the first two hours ReSoMal should be alternated hourly with F75, a specialised feeding formula for children with SAM (see Table 1). These are largely based on expert opinion and not on experimental evidence. The rehydration guidelines do not offer a pragmatic or evidence based approach to management of children with SAM complicated by dehydration and they are open to wide interpretation and misuse. Furthermore, for undernourished children with severe dehydration (equivalent of 10% or more loss of body weight), up to 20% of children hospitalised with gastroenteritis fulfil SAM anthropometric criteria for SAM (MUAC <11.5cm or WHZ <-3SD), but following rehydration they are reclassified as undernourished. Thus, the current recommendations have much wider implications with many ineligible children receiving potentially harmful low sodium rehydration solutions 24 .

Conclusions
The available evidence for management of children with SAM and dehydration is limited and does not lend support to the WHO guidelines. There are arguments to support the use of hypoosmolar ORS in children with SAM, but the currently recommended ReSoMal exposes children with SAM to risk of severe hyponatraemia. Further research should evaluate use of standard hypo-osmolar ORS in children with SAM, and assess optimal rates of rehydration in order to construct evidence based pragmatic guidelines that are designed for the context in which they will be used. In particular, it would be useful to conduct research in sub-Saharan Africa, given that none of the available evidence relates to this population.

Competing interests
No competing interest were declared. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Comments to the authors:
This is an welcome, interesting and well-written review illustrating the sparse evidence underlying the current WHO recommendations for rehydration in children with SAM (as other treatments for SAM!), and points to important future studies to be done.

General comments:
The fact that no children died in the reviewed studies illustrates that the study population of Asian children is very different from the population of interest mentioned in the introduction (children with SAM and diarrhoea in Africa, with high mortality). This issue is discussed, but in addition to the differences already mentioned, it may be relevant to mention that Asian children more frequently have marasmus and less frequently kwashiorkor. Cholera is also likely to be more frequent in Asia (requiring higher Na in rehydration fluids), and HIV is less prevalent. This is likely to make SAM+ diarrhoea a totally different disease complex in African than in Asian children. This does not make the review less relevant, but underlines the conclusion that we know very little about what we do when treating very sick children.
It may be informative to explain the history of development of ReSoMal, "Standard-WHO-ORS" and "low-osmolarity ORS" (now endorsed by WHO, and therefore by most health staff considered Standard ORS), which (at least to me) has been a little confusing: Standard ORS was mainly developed to treat cholera in the first place, and therefore had a relatively high Na content. ReSoMal was developed at a time when this high-Na ORS was used, due to (theoretical) concerns about malnutrition being prone to Na retention, greater K deficit and higher risk of hypoglycaemia in children with SAM (particularly in oedematous children), but as emphasised it was never formally tested whether it was indeed better. Low-osmolar ORS was developed to reduce the intensity of diarrhoea in children, and has been shown to be successful in that respect (see ) (although probably https://www.ncbi.nlm.nih.gov/pubmed/11869639 less appropriate for children with cholera, in whom Na loss is higher), and in most parts of the world low-osmolar ORS is now standard treatment for children with diarrhoea (and no malnutrition). This means that the ORS used for normal children is now closer in composition to ReSoMal regarding Na content than when this was developed. Notably, it also means that the osmolarity of ReSoMal is now than the ORS used for well nourished children,potentially increasing stool output in already very higher 1 than the ORS used for well nourished children,potentially increasing stool output in already very higher vulnerable children.
Another very practical problem with ReSoMal is what to do when ReSoMal is not available, which is the case for many primary care health facilities. Well-trained health staff will have learned that ORS should NEVER be given to malnourished children. Even though the WHO protocol has a recipe for ReSoMal made from ORS (including addition of potassium) my guess is that this "barrier of complication" makes necessary rehydration therapy unavailable for many children with SAM and dehydration seen in primary care.
As highlighted by the authors the evidence base underlying WHOs guidelines for SAM is generally poor, with few RCTs and a lot of expert opinion and pathophysiologic speculations. Some of the least poor evidence is (in my opinion) is a from semi-observational studies where outcomes were compared in hospital facilities before and after implementation of a protocolised treatment package with some elements similar to the WHO guidelines. One of these, a study from Bangladesh compared outcomes before and after implementation of a treatment protocol for children with diarrhoea and SAM including early milk feeding, empiric antibiotics, emphasis on oral rehydration and slower initial rehydration. They found that children given protocolised treatment had almost half the risk of dying compared to children treated before the protocol was implemented. I know that your inclusion criteria was limited to RCTs, and therefore this study is not eligible, but it may be worthwhile to include in the discussion. Worth noting is that the rehydration solution used in both periods contained 90 mmol/l of sodium, higher than currently used low-osmolarity ORS, indication that good improvements in outcome can be obtained by relevant interventions while still giving higher sodium ORS than suggested by WHO guidelines. See . https://www.ncbi.nlm.nih.gov/pubmed/10371570 A quick search for experimental /laboratory/animal study data supporting the "well-established fact" that children with malnutrition have tendency to Na retention. Except for children with kwashiorkor (in whom this tendency is very obvious) I haven't found any, which was a bit of a surprise (although I must admit that my search may not have been extensive enough). I wonder if it would be worthwhile to ask the people who wrote the WHO manual if they knew of laboratory or experimental data supporting the idea of Na and fluid retention (in non-oedematous children)?
I am a bit puzzled about the choice of primary outcome: The most important outcome should be mortality, or something else of direct clinical relevance (like: clinical deterioration, convulsions, failure of oral therapy ect). It is of course also relevant to report other outcomes (like laboratory tests), bearing in mind that these are always surrogate markers for relevant health outcomes. Other similar outcomes to report could be hypo-kalemia, and hypo-glycemia (if these were reported by any studies). Even though they (as argued) may not have equal clinical significance, they should be reported in the result section and not in the discussion (as hypo-kalemia is now).
The studies associating hyponatremia with mortality are from Africa; and the included studies finding development of hyponatremia with rehydration are from Asia, where hyponatremia does not seem be associated with mortality (see for example: Chisti Predictors of death in under-five children with et al: diarrhoea admitted to a critical care ward in an urban hospital in Bangladesh, Acta Paediatrica 2011 ). So the suggested evidence for a causal links between low-sodium rehydration and mortality is indirect and a bit weak weak… The causes for developing hyponatriemia in African children may be different from the causes of hyponatremia in Asian children. Also, I am not sure that hyponatremia was neccesarily a cause of death in the African children, although associated with mortality, but a sign of severe disease. It is possible that correcting their hyponatriemia would not make their prognosis better, but simply be "biochemical make-up". Similarly, iatrogenic hyponatremia (although probably not a good thing) may not 2 3 "biochemical make-up". Similarly, iatrogenic hyponatremia (although probably not a good thing) may not be caused by the same mechanisms as admission hyponatremia. Your concerns about hyponatremia is relevant, but maybe the weakness of surrogate markers should be discussed. The conclusion being be that a randomised trial among those who need better treatment the most is urgently needed. And with the high current mortality it is neccesary and realistic to use mortality as an outcome.

Specific points:
Page 3, first paragraph: It is not clear where the exact numbers come from: 49% with diarrhoea, 16% developing diarrhoea when admitted, mortality of 9% and 19%. It looks like they are derived from a specific study, but which one? Or are they averages from the four studies referenced? In any case I would not be so specific, but rather state that "..diarrhoea complicates around half of admissions for SAM…" It would be helpful if the sodium contents of the different solutions could be incorporated into table 3 (although they are given in table 5), to appreciate the comparison.
Fluid therapy seems particularly tricky in children with oedema, and clinicians have varying approaches to this. In spite of this, the WHO protocol does not differentiate fluid therapy to oedematous and non-oedematous children, and is not very clear about diagnosing dehydration in these (overhydrated) children. It seems plausible that oedematous and non-oedematous children could require different strategies, but of course there is no evidence for this. In order to make this more clear future studies should report results separately for oedematous and non-oedematous children, and test any effect for interaction with oedema. Although I am aware that your included studies did not do this, it would be helpful if table 3 could specify how many children had oedematous malnutrition in each study? electrolyte composition and discuss the evidence suggesting that some soluble fibres should be added as ingredient to the oral rehydration solutions.
The statement that there is no data on the risk of sodium overload in children suffering from severe malnutrition needs qualification. The current concern is based on an old paper from Uganda reporting an excess of heart failure in children receiving a high energy milk to which sodium was added, apparently to improve acceptability (See: Wharton 1967 ). It can be argued that the quantity of sodium given to et al., these children was presumably higher than those receiving low osmolarity ORS, also that these children received this high sodium milk for much longer that ORS is given, and also they had a high sodium intake in absence of diarrhoea, which means these findings cannot be extrapolated to children with diarrhoea, but this paper should be quoted. Also, there are observations suggesting that even Resomal given liberally to children in absence of diarrhoea can lead to an excess mortality due to heart failure which could be due to excessive sodium intake (See: Grellety Y, 2000 ). Arguably, this is again low quality evidence in favour of using low sodium rehydration solution in case of diarrhoea, but this should be mentioned as well.

Minor comments
The term "standard ORS" to refer to the old WHO solution (with 90 mmoles Na/L) is confusing, as this solution is no more recommended and the "low osmolarity ORS (Na 75 mmoles/L) is now the WHO standard. Choose another term avoiding the confusion, maybe just "old WHO ORS with 90 mmoles/L Na".
3/ Added and discussed the two references suggested by Professor Briend to the additional paper/thesis he suggested (Wharton and Grellety) 4/ Included a section in the paper that discussed the two trials that were included on the overall review, that highlights that we included in the systematic review two trials that compared compare oral rehydration solutions with identical electrolyte content (thus do not directly address the question of electrolyte/osmolarity) but are worthy of specific mention. And detail the two trials and conclude that these have the potential to reduce overall stool volume and recovery time and could be considered as candidates for future trials in African children with mortality as a key endpoint. 5/ We chose to use sodium and time to rehydration/length of diarrhoea as our major endpoints since none of the Asian studies reported any mortalities. We were unsure of how many children had kwashiorkor as these were not reported in most of the trials.
We have made the corrections noted by Prof Briend and included relevant references where missing or more specific reference at the beginning reporting mortality (from one paper).
We have no competing interests Competing Interests: