U.S. flag

An official website of the United States government

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

National Collaborating Centre for Women's and Children's Health (UK). Diarrhoea and Vomiting Caused by Gastroenteritis: Diagnosis, Assessment and Management in Children Younger than 5 Years. London: RCOG Press; 2009 Apr. (NICE Clinical Guidelines, No. 84.)

Cover
  • This publication is provided for historical reference only and the information may be out of date.

This publication is provided for historical reference only and the information may be out of date.

Cover of Diarrhoea and Vomiting Caused by Gastroenteritis

Diarrhoea and Vomiting Caused by Gastroenteritis: Diagnosis, Assessment and Management in Children Younger than 5 Years.

Show details

8Other therapies

Introduction

A range of other therapies have been proposed for use in gastroenteritis. These have focused on alleviating vomiting and diarrhoea. They have included a range of anti-emetic and antidiarrhoeal agents. Benefits have also been attributed to certain micronutrients and dietary fibre. In recent years there has been considerable interest in the possible role of probiotics.

Clinical question

Which interventions (other than fluid therapy and antibiotic treatment) are effective and safe?

Various medical interventions were considered under the following category headings:

  1. anti-emetics
  2. antidiarrhoeals
  3. micronutrients and fibre
  4. alternative and complementary therapies
  5. probiotics.

A search was performed with no restrictions on date and this returned 1245 references. The titles and abstracts of these were appraised and 163 papers were identified as potentially relevant to the guideline and were obtained in full copy. Of these, 33 were relevant and were included in this chapter. A further two papers were identified from updating searches.160,161

8.1. Anti-emetics

Many children with gastroenteritis experience vomiting, particularly in the early phase of the illness. This is a distressing symptom. Importantly, it is a major factor in leading to failure of ORT. If vomiting could be treated effectively then there might be a reduction in the use of IVT. Various anti-emetic agents have been used to prevent or reduce vomiting in children with gastroenteritis.

The phenothiazines are dopamine antagonists and act centrally by blocking the chemoreceptor trigger zone. They are used to prevent or treat vomiting associated with drugs such as opioids, general anaesthetics and cytotoxics. Unfortunately, severe dystonic reactions sometimes occur with phenothiazines, especially in children. Metoclopramide is an effective anti-emetic and its activity closely resembles that of the phenothiazines. Metoclopramide also acts directly on the gastrointestinal tract and it may be more effective than the phenothiazines for vomiting associated with gastroduodenal disease. 162 As with the phenothiazines, metoclopramide can induce acute dystonic reactions involving facial and skeletal muscle spasms and oculogyric crises. These dystonic effects are more common in the young. Ondansetron is a specific 5HT3 antagonist which blocks 5HT3 receptors in the gastrointestinal tract and in the central nervous system, It has been shown to be effective in the treatment of vomiting in patients receiving cytotoxic agents for cancer. Dexamethasone also has anti-emetic effects and is used to prevent vomiting associated with cancer chemotherapy. In this context it may be used alone or with other anti-emetics such as metoclopramide or a 5HT3 antagonist.

Evidence overview

Five trials were identified as relevant to this review.160,161,163–165

Four of these were conducted in the USA160,161,163,164 and one in Venezuela.165 Two trials had three treatment arms161,165 and the rest had two. Across the five studies, data from 639 children (aged 6 months to 12 years) suffering from gastroenteritis were collected. The following comparisons were investigated:

  • oral ondansetron versus placebo
  • IV ondansetron versus placebo
  • IV metoclopramide versus placebo
  • IV dexamethasone versus placebo.

The outcomes considered were duration of the disease (vomiting and diarrhoea outcomes), tolerance of ORT, need for IVT, dehydration status and hospitalisation. Follow-up, treatment protocol and definition criteria for inclusion of the children with gastroenteritis varied between the three studies.

The first of the trials conducted in the USA164 had two treatment arms and recruited children aged 6 months to 12 years with gastroenteritis presenting to a paediatric emergency department with at least five episodes of vomiting in the preceding 24 hours and who had not received an anti-emetic (n = 145). Those with an underlying chronic condition, possible appendicitis, a urinary tract infection or who had severe gastroenteritis requiring immediate IV fluids were excluded. Children were randomised to treatment with three doses (appropriate for age group) of oral ondansetron (n = 74) or placebo (n = 71) per day. The methods of randomisation were adequate and patients and outcome assessors were blind to treatment allocation. The power calculation presented estimated that for a treatment success rate of 80% in the ondansetron group and 60% in the placebo group, a sample size of 91 per group would be required. Outcomes measured were cessation of vomiting during the stay in the emergency department, the need for IV rehydration, hospitalisation and diarrhoeal episodes during follow-up. [EL = 1+]

The groups were similar at baseline for age, sex and severity of illness. No children were lost to follow-up in the emergency department stay, but by 48 hours 32/145 had been lost. More children in the ondansetron group (64/74) stopped vomiting in the first few hours after treatment in the emergency department compared with those who received placebo (46/71). This difference was statistically significant (RR 1.33; 95% CI 1.10 to 1.62). Fewer children treated with oral ondansetron required IVT (8% estimated from histogram) or were admitted to hospital (2/74) compared with those treated with placebo (22.5% and 11/71, respectively) (RR 0.42; 95% CI 0.17 to 1.00 and RR 0.15; 95% CI 0.03 to 0.71, respectively). The mean number of diarrhoeal episodes while undergoing rehydration (mean length of stay in emergency department 106 minutes ondansetron group versus 120 minutes placebo group) was statistically significantly higher in children who had received ondansetron (mean 1.40) compared with the placebo group (mean 0.50) (P < 0.001) even after adjustment for number of episodes prior to admission.

The second US trial163 recruited children aged 6 months to 10 years presenting to an emergency department with gastroenteritis (defined as at least one episode of vomiting within the 4 hours preceding triage, at least one episode of diarrhoea and mild to moderate dehydration) (n = 215). Those with a body weight less than 8 kg, severe dehydration, an underlying disease that could affect the assessment of dehydration, a history of abdominal surgery or hypersensitivity to ondansetron were excluded. Participants were randomised to treatment with oral ondansetron (dose appropriate for age group) (n = 108) or placebo (n = 107). The methods of randomisation were adequate and treatments were given by a bedside nurse such that the patient, caregivers and outcome assessors were blinded to allocation. The proportion of patients randomised but lost to follow-up was less than 20%. The power calculation presented estimated that to provide the study with a statistical power of 90% to detect a change from 35% in the placebo group to 15% in the treatment group (in the children who vomited during ORT) (type I error 0.05), 215 children would need to be recruited. The primary outcome was the proportion of children who vomited while receiving ORT and the secondary outcomes were episodes of vomiting during ORT, rate of IV rehydration, rate of hospitalisation and diarrhoeal episodes during follow-up (on days 3 and 7 after randomisation). [EL = 1+]

The two groups were comparable at baseline for sex, age, weight, heart rate, dehydration score, the number of vomiting and diarrhoeal episodes prior to presentation, and serum values. Statistically significantly more children in the ondansetron group (92/107) stopped vomiting in the first few hours after treatment compared with those who received placebo (70/107) (RR 1.31; 95% CI 1.12 to 1.54). Fewer of the children receiving ondansetron required IVT (15/107) than those treated with placebo (33/107), the difference being statistically significant (RR 0.45; 95% CI 0.26 to 0.79). There were no statistically significant differences between groups in the numbers of children admitted to hospital or in episodes of diarrhoea while undergoing rehydration (mean length of stay in emergency department 2 hours ondansetron group versus 3 hours placebo group). However, over the next 48 hours, children receiving ondansetron had statistically significantly more diarrhoea than those receiving placebo. In the first 24 hour period, the mean number of diarrhoeal episodes in the ondansetron group (n = 64) was 4.70 compared with 1.37 in the placebo group (n = 54) (P = 0.002) and in the second 24 hours was 2.98 episodes (n = 62) compared with 0.96 episodes (n = 51) (P = 0.015).

The third trial from the USA160 included children aged 1–10 years with acute gastritis or gastroenteritis and mild to moderate dehydration who failed ORT in the emergency department (n = 106). Exclusion criteria were receipt of anti-emetics in the 6 hours prior to enrolment, underlying chronic illness, shock state requiring immediate IV fluids, severe (10% or more) dehydration, and known sensitivity to 5HT3 antagonists. Participants were randomised to treatment with oral ondansetron (dose appropriate for weight) (n = 51) or placebo (n = 55). The methods of randomisation were adequate and the study was conducted double-blind. With estimated failure rates of 30% in the ondansetron group and 60% in the placebo group to achieve a power of 80% with a significance level of 0.05, 48 participants were required in each group. Outcomes were measured daily for up to 6 days or until symptoms resolved if sooner, and were cessation of vomiting, IV fluid administration, hospitalisation and frequency of diarrhoea. [EL = 1+]

Baseline comparability of the groups was similar except that statistically significantly more children in the ondansetron group were ‘moderately’ dehydrated. Hence more children were mildly dehydrated in the placebo group but this was not statistically significant. Nine percent of participants did not participate in follow-up telephone interviews and three patients in the ondansetron group were incorrectly diagnosed. The investigators reported that 93% of patients who had received ondansetron had fewer than three episodes of vomiting during a 6 day follow-up, compared with 88% of patients in the placebo group. Insufficient data were available to establish the statistical significance of this difference or of the reported mean number of vomiting episodes between the groups. Fewer of the children receiving ondansetron required IVT (9/48) than those treated with placebo (30/55), the difference being statistically significant (RR 0.34; 95% CI 0.18 to 0.65). There was no statistically significant difference between groups in the numbers of children admitted to hospital. Nineth-three percent of patients who had received placebo had fewer than three episodes of diarrhoea during a 6 day follow-up, compared with 80% in the ondansetron group. Insufficient data were available to establish the statistical significance of this difference or of the reported mean number of diarrhoeal episodes between the groups.

Data on cessation of vomiting was extracted from two trials163,164 and pooled in a meta-analysis. The results showed that more children in the ondansetron groups stopped vomiting in the first few hours after treatment (146/181) compared with those who received placebo (116/178). This difference was statistically significant (RR 1.32; 95% CI 1.17 to 1.49) (Figure 8.1).

Figure 8.1. Comparison of the effect of ondansetron versus placebo on cessation of vomiting within the first few hours of treatment.

Figure 8.1

Comparison of the effect of ondansetron versus placebo on cessation of vomiting within the first few hours of treatment.

All three trials160,163,164 compared the effects of oral ondansetron with placebo on IV hydration. These findings were pooled in a meta-analysis which showed that fewer of the children receiving ondansetron required IVT than those treated with placebo (79/233), with the difference being statistically significant (RR 0.41; 95% CI 0.28 to 0.59) (Figure 8.2).

Figure 8.2. Comparison of the effect of ondansetron versus placebo on the need for IVT.

Figure 8.2

Comparison of the effect of ondansetron versus placebo on the need for IVT.

The data from the trials160,163,164 were pooled for the number of patients admitted to hospital after the emergency department stay. The findings showed that statistically significantly fewer children given ondansetron were admitted to hospital (9/232) compared with those given placebo (23/233) (RR 0.37; 95% CI 0.17 to 0.82) (Figure 8.3).

Figure 8.3. Comparison of the effect of ondansetron versus placebo on the need for hospital readmission.

Figure 8.3

Comparison of the effect of ondansetron versus placebo on the need for hospital readmission.

The fourth US study161 recruited children aged 6 months to 12 years presenting to an emergency department with paediatrician-identified acute gastritis or gastroenteritis (defined as more than three vomits over previous 24 hours) with mild or moderate dehydration who had failed ORT. Those with current chronic medical disease (excluding asthma), with a history of abdominal surgery, requiring chronic medications, with inhaled or other corticosteroid or anti-emetic use in the previous fortnight or with diagnostic findings inconsistent with isolation acute viral gastritis were excluded. The trial had three arms and participants were randomised to a 10 minute infusion with 0.15 mg/kg IV ondansetron (n = 46), 1 mg/kg IV dexamethasone (n = 55) or a 10 ml bolus of normal saline (placebo) (n = 44). The methods of randomisation were adequate and patients and outcome assessors were blinded to treatment allocation. An a priori power calculation was done to estimate the sample size required in each group. The main outcomes measured were need for hospitalisation, tolerance of oral hydration and dehydration status at 2 and 4 hours post treatment, and patients were followed up to 72 hours. [EL = 1−]

The study was terminated early and the reported findings were for approximately half the number of expected participants (137/270). The groups were similar at study entry for age, sex, blood glucose, dehydration status and number of vomiting episodes. The trial found that statistically significantly more children who received placebo (9/44) required admission to hospital compared with those who had received IV ondansetron (2/46) (RR 0.21; 95% CI 0.05 to 0.93). Statistically significantly more children who received IV ondansetron tolerated oral rehydration 2 hours after treatment (39/45) than those who received placebo (29/43) (RR 1.28; 95% 1.02 to 1.68). However, results taken at 4 hours post treatment were not statistically significant. Results for mean IV fluids administered and dehydration status were also similar between groups at 2 and 4 hours post treatment. There were no statistically significant differences between IV dexamethasone and placebo groups in hospital admission rates, the numbers of children tolerating oral rehydration, mean IV fluids administered or dehydration status at 2 or 4 hours.

The RCT conducted in Venezuela165 recruited children aged 6 months to 8 years with gastroenteritis with emesis who had vomited twice within 1 hour (n = 36). Patients were hospitalised for a minimum of 24 hours. Exclusion criteria were severe dehydration, seizures, rectal temperature of 39 °C or more, parenteral anti-emetic medication in the 6 hours prior to the start of the study or parasite-induced gastroenteritis. The trial had three arms and participants were randomised to a 10 minute infusion with 0.3 mg/kg IV ondansetron (n = 12), 0.3 mg/kg IV metoclopramide (n = 12) or a 15 ml bolus of normal saline (placebo) (n = 12).

The methods of randomisation were not reported although it was stated that pharmacy-controlled medication preparation permitted the patients and outcome assessors to be blind to treatment allocation. The proportion of patients randomised but lost to follow-up was less than 20%. No sample size power calculation was provided.

The groups were different at baseline for age, height, weight and degree of hydration, with comparability on gender and food intake only. There were no statistically significant differences between the IV ondansetron and placebo groups in vomiting or diarrhoea episodes in the first 24 hours. Ten of the 12 children receiving IV metoclopramide had more than four episodes of diarrhoea during the first 24 hours compared with 4/12 children in the placebo group. The difference was statistically significant (RR 2.50; 95% CI 1.08 to 5.79). However, no statistically significant difference was found for emetic episodes between the two groups in the first 24 hours.

Evidence summary

There was evidence from three RCTs [EL = 1+] that supported the effectiveness of oral ondansetron in the treatment of gastroenteritis in children. The meta-analysis performed after extracting the data from two RCTs showed that children with gastroenteritis and receiving oral ondansetron along with rehydration solution were more likely to stop vomiting. Pooled data from three trials demonstrated that the ondansetron group were less likely to require further IVT and less likely to be hospitalised compared with children who had received only rehydration solutions and placebo. No consistent results were found for diarrhoea outcomes. Two of the three trials reported statistically significant results to show that children receiving ondansetron seemed to experience more episodes of diarrhoea.

There was a lack of high-quality evidence for the effectiveness of IV ondansetron, IV metoclopramide and IV dexamethasone in the treatment of children with gastroenteritis. A small RCT [EL = 1−] showed no difference in the cessation of vomiting during the first 24 hours following treatment in children receiving IV ondansetron or IV metoclopramide compared with children treated with placebo. The risk of having more than four diarrhoeal episodes was higher in both the treatment groups (IV ondansetron group and IV metoclopramide group) compared with the placebo group, but the difference was statistically significant only for the IV metoclopramide group.

However, a second underpowered trial did show more children given ondansetron did not require hospitalisation and tolerated ORT more quickly than those given placebo. No statistically significant differences were found between the IV dexamethasone and placebo groups for hospitalisation rates or ORT tolerance.

Cost-effectiveness evidence

A simple economic model was also developed which demonstrates potential economic advantages of ondansetron if given to children with persistent vomiting in whom IV fluids are being considered; further details are provided in Appendix B. Owing to limited evidence for the efficacy of IV ondansetron, the economic analysis only considers the use of oral ondansetron.

The use of anti-emetics such as ondansetron may be effective in the cessation of vomiting and may in turn help with the successful delivery of ORT, thereby reducing the need to treat with IVT. This would have cost-saving implications for the NHS through fewer admissions for IVT. The analysis uses a meta-analysis of three RCTs160,163,164 to estimate the difference in effect between placebo and ondansetron for three outcomes: cessation of vomiting, hospitalisation and the need for IVT. Savings in downstream costs were calculated based on the difference in effect obtained with ondansetron. The net cost was calculated by offsetting the cost of ondansetron against the saving achieved with reduced downstream resource use. The analysis suggests that ondansetron, when compared with placebo, is dominant owing to an increase in the cessation of vomiting and net savings due to a reduction in IVT and lower rates of hospitalisation. A probabilistic sensitivity analysis (PSA) was carried out which showed that the results were not sensitive to parameter uncertainty within the economic model, with ondansetron dominant in 99.96% of simulations. Although some trials reported changes in diarrhoea in response to ondansetron, this was not included in the economic model mainly because of the uncertainty of the clinical significance of this and whether or not increased diarrhoea would lead to increased use of NHS resources. Limitations of the model, particularly with regard to diarrhoeal outcomes, need to be addressed by further research in order to make firm conclusions regarding the cost-effectiveness of ondansetron.

GDG translation from evidence to recommendation

Although many children vomit during ORT, this is usually not so severe as to prevent oral rehydration. Occasionally, vomiting is frequent and persistent. In such cases, a decision might be made to administer ORS solution by nasogastric tube or to change to IVT. The availability of an effective anti-emetic could therefore be very valuable. The GDG considered that evidence from randomised controlled trials indicated that oral ondansetron could increase the success rate with ORT. The GDG was concerned that ondansetron might have adverse effects such as worsening diarrhoea. There was no evidence to support other agents, including metoclopramide and dexamethasone. The GDG concluded that administration of anti-emetics could not currently be recommended. However, the GDG did consider that further research on the use of ondansetron was needed, focusing particularly on the possible risk of worsened diarrhoea.

Research recommendation

In children with persistent vomiting caused by gastroenteritis, is oral ondansetron cost-effective and safe compared with placebo therapy?

Why this is important

Several randomised controlled trials have shown that in children with persistent vomiting during oral rehydration therapy, administration of oral ondansetron, an anti-emetic agent, can increase the likelihood of successful oral rehydration. However, in two of these there was evidence suggesting that diarrhoea was more pronounced in those given ondansetron than in those in the placebo groups. In one, in children given ondansetron, the number of stools passed during the rehydration phase was significantly greater, and in the other the number of stools passed in the first and second 24 hour period after rehydration was significantly greater. In those studies, diarrhoea was not a primary outcome, and it was reported as an adverse event. The reliability of the finding was therefore somewhat uncertain. If ondansetron does worsen diarrhoea it would be crucially important to determine the clinical significance of this effect, for example in relation to the risk of dehydration recurring or re-admission to hospital. If ondansetron is shown to be both effective and safe in secondary care then studies should also be undertaken to evaluate its use in primary care.

8.2. Antidiarrhoeal agents

A range of drugs have been used as antidiarrhoeal agents in patients with gastroenteritis and other disorders. Adsorbent agents such as clay minerals (kaolin or smectite) and charcoal have been employed. Antisecretory drugs such as racecadotril (a peripherally acting enkephalinase inhibitor) reduce intestinal water and electrolyte secretion. Bismuth subsalicylate (BSS) has a number of properties that may be important in reducing diarrhoea, including inhibition of intestinal fluid secretion, suppression of intestinal inflammation and a bactericidal action. Anti-motility agents such as loperamide may reduce diarrhoea by lengthening intestinal transit time and hence absorption.

Nowadays it is generally advised that these medicines should be avoided in the treatment of children with gastroenteritis. Nevertheless, it was considered important to review the available evidence in relation to the use of these agents.

8.2.1. Adsorbent agents

8.2.1.1. Kaolin

Evidence overview

A trial166 conducted in the Gambia included children with diarrhoea aged between 3 and 18 months (n = 97). Those requiring antibiotic therapy or with positive diagnosis for malaria were excluded. Participants were allocated to treatment with kaolin and ORS solution (n = 45) or to administration of ORS solution alone (n = 52). The method of randomisation was inadequate (birth order allocation) and allocation concealment and follow-up were not reported. Compliance with the doses of kaolin was poor in 33% of the participants. The outcomes considered were duration of diarrhoea and the number of stools per day in both groups. [EL = 1−]

No statistically significant differences were found between the kaolin and the ORS solution groups in the mean duration of diarrhoeal episodes or in the mean number of stools per day.

Evidence summary

There was a lack of high-quality evidence for the effectiveness of kaolin in the treatment of acute diarrhoea in children. A quasi-randomised controlled trial [EL = 1−] showed no differences in the duration of acute diarrhoea or in the number of stools per day between children receiving kaolin with rehydration therapy and children receiving rehydration therapy alone.

8.2.1.2. Activated charcoal

Evidence overview

One RCT was identified that included 39 children aged between 1½ months and 10 years with acute gastroenteritis and severe dehydration (n = 39).167 [EL = 1−] Children with acute gastroenteritis due to Entamoeba histolytica were excluded. Participants were allocated to treatment with activated charcoal with oral and IV rehydration solution (n = 16) or oral and IV rehydration alone (n = 23). The method of randomisation, allocation concealment, follow-up and baseline comparability of the two groups were poorly reported. The outcomes considered were duration of diarrhoea and fluid therapy administered.

The study found that the group receiving the activated charcoal had a statistically significantly shorter mean duration of diarrhoea (mean 2.12 days) than the control group (mean 3.00 days) (WMD −0.88 days; 95% CI −1.50 to −0.26 days). Children in the activated charcoal group required statistically significantly less ORS solution (mean 3.25 packs) than the control group (mean 5.43 packs), (WMD −2.18 packs; 95% CI −3.84 to −0.52 packs). There was no statistically significant difference in the mean IVT (Ringer’s lactate solution) required between groups.

Evidence summary

There was some evidence from a poorly reported RCT suggesting that the additional use of activated charcoal in the treatment of children with gastroenteritis shortened the duration of the diarrhoea and reduced the amount of ORT required when compared with the standard rehydration therapy alone. On the other hand, the same trial showed no difference in the amount of IV rehydration therapy required between the children receiving activated charcoal and the children who did not received the adsorbent agent.

8.2.1.3. Smectite

Evidence overview

One relevant systematic review of trials comparing smectite with placebo or no treatment for diarrhoea in children was identified.168 This well-conducted systematic review included nine RCTs published between 1986 and 2002. Two trials were conducted in France, one in Italy, one in Lithuania, two in Thailand, one in Egypt and one in China. The nine studies included data from 1238 participants: 622 received smectite and 616 received placebo or no additional treatment. Even though it was reported that the doses of smectite were similar across the studies, the duration of the intervention varied from 2 to 6 days among six studies, and the remaining three did not report such information. Participants were children aged between 1 and 60 months, and were inpatients and/or outpatients. Definitions for the outcome measures and the resolution of diarrhoea were different among the studies. The methodological quality and conduct of the studies were not uniform. An adequate generation of the allocation sequence was reported only for three trials and the allocation concealment was appropriate only in one. Three trials were double-blinded and in only five was an intention-to-treat analysis performed. Data were extracted for the following outcomes: duration of diarrhoea, frequency of stools, vomiting and adverse events. The review also reported the proportion of patients without diarrhoea by days 3 and 5 as well as the proportion of those presenting diarrhoea for more than 7 days. All the trials had better than 80% follow-up. [EL = 1+]

Data from six trials (1076 patients) were pooled in a meta-analysis that showed a statistically significant reduction in the duration of diarrhoea when smectite was administered (WMD −22.7 hours; 95% CI −24.8 to −20.6 hours). Two studies that provided data on the number of stools were pooled. No statistically significant difference was found between the intervention and the control group in the number of stools during the first 24 hours of follow-up. However, children treated with smectite had a statistically significant reduction in the number of stools from the first 24 hour period onwards: at 24 to 48 hours WMD −0.62; 95% CI −1.0 to −0.2), and at 48 to 72 hours WMD −0.58; 95% CI −0.9 to −0.3). The reviewers pooled the data from four trials together, finding that by day 3 the proportion of children treated with smectite and without diarrhoea was statistically significantly higher than the proportion of children that were not treated with smectite and were without diarrhoea (RR 1.64; 95% CI 1.36 to 31.98). When results were pooled using a fixed effects model, statistically significantly more children treated with smectite (122/128) were cured by day 5 compared with those receiving placebo (106/126) (RR 1.24 (95% CI 1.08 to 1.42). However, as significant heterogeneity was observed (χ2 = 8.01; P = 0.02; I2 = 75%), the reviewers also presented results using a random effects model, where the difference was not statistically significant (RR 1.19; 95% CI 0.93 to 1.53). A funnel plot and regression asymmetry test (P = 0.23 and the 95% CI included 0) did not show any publication or other small-sample bias.

One trial showed a reduction in the risk of having diarrhoea for more than 7 days for those children receiving smectite when compared with the control group (RR 0.60; 95% CI 0.42 to 0.85). No statistically significant difference in the number of vomiting episodes between the two groups was found when the results of two studies were combined. One RCT reported the duration of vomiting and showed no statistically significant difference between the two groups. Another trial found no statistically significant differences between the treatment groups for the incidence of vomiting on day 1 and or on day 3 of the intervention. No statistically significant differences in adverse events associated with treatments were identified and three RCTs reported no adverse events associated with short-term treatment with smectite.

Evidence summary

There was evidence from a well-conducted systematic review [EL = 1+] for the effectiveness of smectite in the treatment of children with diarrhoea. The review showed that children receiving smectite had a reduction in the frequency and duration of diarrhoea, experienced a higher resolution of diarrhoea by day 3 and were less likely to have diarrhoea for more than 7 days. On the other hand, the review showed no statistically significant differences between the smectite group and the control group in the number of episodes or duration of vomiting or in the resolution of diarrhoea by day 5. Statistically significant differences in adverse effects between the smectite and control groups were not reported and some trials reported that no adverse events occurred.

8.2.2. Antisecretory agents

Evidence overview

Two randomised placebo-controlled trials of racecadotril169,170 and three trials of BSS were identified.171–173

8.2.2.1. Racecadotril

Two trials compared the effect of racecadotril with placebo.169,170 Both trials had two treatment arms and, in total, data from 307 participants (aged 3 months to 4 years) were collected.

One trial (n = 135) conducted in Peru169 examined the effect of racecadotril compared with placebo in boys aged between 3 and 35 months admitted to hospital for dehydration. The included participants had passed watery diarrhoea for 5 days or less, had passed three or more diarrhoeic stools in the 24 hours prior to admission and had passed one diarrhoeic stool within 4–6 hours post admission. Those with blood in the stool, severe dehydration (inability to drink because of drowsiness) or any serious concomitant illness were excluded. Racecadotril (1.5 mg/kg) (n = 68) or placebo (n = 67) was randomly administered as a powder every 8 hours for 5 days or until diarrhoea stopped if earlier. Oral rehydration was given as needed to all participants. No details were provided about sequence generation, allocation concealment or blinding of assessors. The recorded outcomes included stool output, duration of diarrhoea and overall cure rate measured at 5 days. [EL = 1+]

Both groups were broadly comparable at baseline. Seventeen percent of all participants were lost to follow-up. A statistically significant difference was reported in the mean 48 hour stool output favouring racecadotril over placebo for all participants and for both the rotavirus positive and negative groups. For all 135 participants, the mean stool output in the first 48 hours was 92 ± 12 g/kg in the racecadotril group (n = 68) compared with 170 ± 15 g/kg in the placebo group (n = 67) (P < 0.001). A statistically significant reduction in mean stool output was also observed in the rotavirus-positive participants (n = 73). In the racecadotril group (n = 34), the mean stool output in the first 48 hours was 105 ± 17 g/kg and 195 ± 20 g/kg in the placebo group (n = 39) (P < 0.001). The authors stated that in the rotavirus-negative subgroup (n = 62) there was a statistically significant reduction in the mean 48 hour stool output in participants receiving racecadotril compared with those receiving placebo (31% lower in the racecadotril group; 95% CI 16% to 46%; P < 0.001). For all participants, the mean hourly rate of stool production in first 48 hours was statistically significantly lower in the racecadotril group (1.8 ± 0.2 g/kg/hour) compared with the placebo group (3.1 ± 0.3 g/kg/hour) (P < 0.001). The mean total stool output before recovery was statistically significantly lower in the racecadotril group (157 ± 27 g/kg) compared with the placebo group (331 ± 39 g/kg) (P < 0.001). This was also observed in the rotavirus-positive participants (n = 73): in the racecadotril group (n = 34) the mean stool output was 174 ± 36 g/kg and in the placebo group (n = 39) it was 397 ± 37 g/kg (P < 0.001). Although no further details were provided, the authors stated that in the rotavirus-negative subgroup (n = 62), there was a statistically significant reduction in the mean stool output before recovery in participants receiving racecadotril compared with those receiving placebo (37% lower in the racecadotril group; 95% CI 20% to 56%; P < 0.001). The median duration of diarrhoea was shorter for the racecadotril group than the placebo group in both the rotavirus subgroups. In both subgroups, the median duration of diarrhoea was 28 hours for the racecadotril group. However, in the placebo group the rotavirus-positive participants had a median duration of diarrhoea of 72 hours compared with 52 hours in the rotavirus-negative participants (P < 0.001). In all participants, at 5 days, 57 of the racecadotril group (n = 68) were cured of diarrhoea (passing of two consecutive formed stools or not having passed a stool for 12 hours) compared with 44 cured participants in the placebo group (n = 67). This difference was statistically significant (RR 1.28; 95% CI 1.04 to 1.56; P = 0.02).

A multicentre RCT (n = 172),170 conducted in 13 centres in France, examined the effect of racecadotril compared with placebo in children of both sexes aged between 3 months and 4 years hospitalised for severe acute diarrhoea. Included participants had passed watery diarrhoea (three watery stools per day or more) for less than 72 hours’ duration and had passed one watery stool post admission to hospital. Children were excluded if they had chronic diarrhoea, a weight-for-age deficit of 20% or more of the US National Center for Health Statistics (NCHS) standard, a systemic illness or any antibiotic, antidiarrhoeal drug or acetylsalicylic acid usage in the preceding 48 hours. Racecadotril (1.5 mg/kg) (n = 89) or placebo (n = 83) was randomly administered as a powder three times daily for 5 days or until diarrhoea stopped if earlier. Rehydration was administered orally or by gastric tube without restriction. No details were provided about sequence generation, allocation concealment or blinding of assessors and there were considerable losses in data collection and follow-up (28%). Measures of stool output were presented as both a full data set (n = 168) and per-protocol results (n = 121). [EL = 1−]

Treatment groups were broadly comparable at baseline. Using the full data set (n = 167), the mean hourly rate of stool production in the first 24 hours was statistically significantly lower in the racecadotril group (n = 85) (10.5 g/hour) compared with the placebo group (n = 82) (16 g/hour) (read from graph). The authors estimated that the treatment difference showed that stool output was approximately 65% of that with placebo (95% CI 44% to 95%; P = 0.025). The mean hourly rate of stool production in the first 24 hours was also found to be lower in the racecadotril group (n = 58) (11 g/hour) compared with the placebo group (n = 63) (17.5 g/hour) in the per-protocol population (read from graph). The authors estimated that the treatment difference showed that stool output was approximately 65% of that with placebo (95% CI 36% to 90%; P = 0.015). In the full data set analysis, mean stool output in the first 48 hours was shown to be statistically significantly reduced in the racecadotril group (n = 84) (9 g/hour) compared with the placebo group (n = 82) (15 g/hour) (read from graph). The authors estimated that the treatment difference showed a 60% reduction in stool output in those receiving racecadotril compared with placebo (95% CI 43% to 88%; P = 0.009). Using covariate analysis in the per-protocol population, the mean hourly rate of stool production in the first 48 hours was found to be statistically significantly lower in the racecadotril group (n = 53) (8 g/hour) compared with the placebo group (n = 63) (16 g/hour) (read from graph). The authors estimated that this indicated a treatment difference of a 50% reduction in stool output (95% CI 33% to 75%; P = 0.001). The authors stated that this effect was independent of rotavirus status (P = 0.5 for interaction) and that racecadotril was similarly statistically significantly effective in the rotavirus-positive (8 g/hour versus placebo 19 g/hour) and rotavirus-negative (6 g/hour versus placebo 13 g/hour) groups (P = 0.001).

Evidence summary

There was evidence from two randomised placebo-controlled trials for the effectiveness of racecadotril in the treatment of diarrhoea in gastroenteritis. One trial conducted in Peru [EL = 1+] and one poorly reported European multicentre trial [EL = 1−] found that children under 4 years given racecadotril (1.5 mg/kg three times daily) and oral rehydration had a reduced total and average hourly stool output 48 hours after starting treatment compared with children given placebo and oral rehydration. The effect on total stool output was independent of rotavirus status. One trial also reported that the rate of stool output was reduced at 24 hours. [EL = 1−] The average stool output before recovery was found to be reduced, irrespective of the child’s rotavirus status in one of the trials, which also reported a higher diarrhoeal cure rate at 5 days for children given racecadotril. [EL = 1+] A ‘cured’ child had passed two consecutive formed stools or no passage of stool for 12 hours.

8.2.2.2. Bismuth subsalicylate

Three RCTs investigating BSS were identified from the searches.171–173 One was conducted in Bangladesh,171 one in Chile173 and one in Peru.172 Two trials had two treatment arms171,173 and the third had three. Data were collected from 808 children across the three trials although the outcomes considered varied and were as follows: onset of persistent diarrhoea, duration of diarrhoea, intake of oral or IV rehydration and total administration of rehydration solutions, total stool output, total volume of vomitus and duration of hospitalisation.

The largest RCT (n = 451)171 compared the effects of administering BSS (100 mg/kg per day) to children presenting with acute diarrhoea with the administration of placebo. The participants were children aged between 4 and 36 months with a history of acute watery diarrhoea. The exclusion criteria were use of antimicrobials within the previous 48 hours, blood in the stool, severe malnutrition, other systemic illness, salicylates intake in the previous 24 hours, allergy to salicylates, or varicella or measles in the previous 3 months. The methods of randomisation and allocation concealment were adequate. Children were followed up for the duration of the hospitalisation and another 4 days. The outcomes measured were onset of persistent diarrhoea, duration of acute diarrhoea (median), total intake of ORS solution and total stool and urine output. [EL = 1+]

The groups were similar at baseline and the loss to follow-up was 8%. The difference in the proportions of children who developed persistent diarrhoea between treatment groups was not statistically significant. The duration of diarrhoea was also found to be similar in children receiving BSS and placebo. However, when considering those children positive to rotavirus, the authors found a statistically significant reduction in the duration of diarrhoea among the children treated with BSS (median 56 hours) when compared with the children receiving placebo (median 72 hours) (P = 0.03). Children treated with BSS had a statistically significant lower output of stool plus urine (mean 386 g/kg, SD 248 g/kg) than the children receiving placebo (mean 438 g/kg, SD 272 g/kg), (WMD −52 g/kg; 95% CI −100 to −4 g/kg). There was no statistically significant difference between the study groups in the mean total intake of ORS solution. While no defined adverse reactions were observed, it was reported that two children had ‘black tongue’ during treatment.

The second RCT (n = 142)173 examined the effects of treating children aged between 4 and 36 months and suffering from diarrhoea and dehydration with BSS. The study compared the effects of administering BSS (100 mg/kg) for 5 days (n = 72) with placebo (n = 70). The exclusion criteria were symptoms that had lasted for more than 72 hours, blood in stools, severe malnutrition, antibiotic use in the previous 48 hours, a salicylate intake greater than 20 mg/kg in the previous 12 hours, allergy to bismuth or to salicylate or acute illness inconsistent with a diarrhoeal state. The method of randomisation was not reported but allocation concealment was adequate. Patients were monitored in hospital for at least 5 days and then were followed for 3 more days (whether they remained in hospital or were discharged). Outcomes measured were disease duration (hours), time to last loose/watery stool, time until last unformed stool, duration of hospital stay and IV fluid intake (ml/kg). [EL = 1+]

The groups were similar at baseline and the loss to follow-up was 13.4%. The use of BSS demonstrated statistically significant benefits compared with placebo in terms of shortening the duration of diarrhoea. The mean time to last loose/watery stool was shorter in the group treated with BSS (mean 73.4 hours) compared with the group receiving placebo (mean 107.5 hours) (P < 0.02). The mean time until last unformed stool was also statistically significantly shorter in the group treated with BSS (mean 130.4 hours) compared with the group receiving placebo (mean 170 hours) (P < 0.01). The need for IV rehydration therapy was lower in the group treated with BSS than in the placebo group. The difference was reported as statistically significant (data for this outcome have been extracted from a histogram and therefore are estimates). At day 3, the intervention group received on average 30 ml/kg and the control group 45 ml/kg. At day 5, the intervention group intake was on average 20 ml/kg and the control group 42 ml/kg. There was a statistically significant reduction in the duration of hospitalisation among the BSS group compared with the placebo group. The mean of hospital stay for the intervention group was 6.9 days, while for the control group it was 8.5 days (P = 0.01). No adverse reactions were observed.

An RCT with three treatment arms (n = 215)172 compared the effects of administering BSS to treat diarrhoea with that of placebo. Participants were boys aged between 6 and 59 months who had acute diarrhoea (three or more watery stools in the preceding 24 hours). The exclusion criteria were presence of blood in the stools, diarrhoea for more than 5 days, use of antibiotics, antidiarrhoeal medication or any treatment with acetylsalicylic acid in the 72 hours before admission, clinical evidence of another illness requiring antibiotic therapy, severe malnutrition, allergy to salicylate or to bismuth or exclusive breastfeeding. The boys were randomised to treatment with BSS 100 mg/kg (n = 108), BSS 150 mg/kg (n = 108) or placebo (n = 107) every 4 hours for 5 days or until the diarrhoea stopped. The methods of randomisation and allocation concealment were adequate. The participants were followed up while in hospital and the outcomes assessed were duration of diarrhoea (proportion of patients with diarrhoea by day 5), total stool output (ml/kg), total volume of vomitus (ml/kg), total intake of rehydration (ml/kg) and duration of hospital stay (days). [EL = 1+]

The groups were similar at baseline and the loss to follow-up was 8% of 275 patients initially enrolled. Diarrhoea stopped within 5 days of admission in 76/85 (89%) children treated with 100 mg/kg BSS, in 73/83 (88%) children treated with 150 mg/kg BSS and in 62/84 (74%) children receiving placebo. By day 5, diarrhoea had ceased in statistically significantly more children in each of the two intervention groups compared with the control group (100 mg/kg BSS RR 1.21; 95% CI 1.05 to 1.40, and 150 mg/kg BSS RR 1.19; 95% CI 1.03 to 1.38). The mean total stool output was 182 ml/kg (SD 197 ml/kg) in children treated with 100 mg/kg BSS (n = 85), 174 ml/kg (SD 159 ml/kg) in children treated with 150 mg/kg BSS (n = 83) and 260 ml/kg (SD 254 ml/kg) in children receiving placebo (n = 84). The mean total stool output was statistically significantly reduced in each of the two intervention groups compared with the control group (100 mg/kg BSS WMD −78 ml/kg; 95% CI −147 to −9 ml/kg, and 150 mg/kg BSS WMD −86 ml/kg; 95% CI −150 to −22 ml/kg). The mean total volume of vomitus was 11.6 ml/kg (SD 19.6 ml/kg) in children treated with 100 mg/kg BSS (n = 85), 8.7 ml/kg (SD 18.3 ml/kg) in children treated with 150 mg/kg BSS (n = 83) and 16.2 ml/kg (SD 27.0 ml/kg) in children receiving placebo (n = 84). The mean total volume of vomitus was statistically significantly reduced in the group receiving 150 mg/kg BSS compared with the control group (WMD −7.5 ml/kg; 95% CI −14.5 to −0.5 ml/kg). No statistically significant difference in the mean total volume of vomitus was found between the group receiving 100 mg/kg BSS and the placebo group (WMD −4.6 ml/kg; 95% CI −11.7 to 2.5 ml/kg). The mean total intake of rehydration solutions was 239 ml/kg (SD 177 ml/kg) in children treated with 100 mg/kg BSS (n = 85), 236 ml/kg (SD 152 ml/kg) in children treated with 150 mg/kg BSS (n = 83) and 314 ml/kg (SD 234 ml/kg) in children receiving placebo (n = 84). The mean total intake of rehydration solutions was statistically significantly reduced in each of the two intervention groups compared with the control group (100 mg/kg BSS WMD −75 ml/kg; 95% CI −138 to −12 ml/kg, and 150 mg/kg BSS WMD −78 ml/kg; 95% CI −138 to −18 ml/kg). The mean length of hospital stay was 3.3 days (SD 1.5 days) in children treated with 100 mg/kg BSS (n = 85), 4.1 days (SD 2.1 days) in children treated with 150 mg/kg BSS (n = 83) and 3.4 days (SD 1.5 days) in children receiving placebo (n = 84). The mean length of hospital stay was statistically significantly reduced in each of the two intervention groups compared with the control group (100 mg/kg BSS WMD −8.0 days; 95% CI −1.4 to −0.3 days, and 150 mg/kg BSS WMD −0.7 days; 95% CI −1.3 to −0.2 days). The effects of administering BSS 100 mg/kg compared with BSS 150 mg/kg were also reported. No statistically significant differences were observed between the treatment groups for the duration of diarrhoea, total stool output, total volume of vomitus, total intake of rehydration or duration of hospitalisation. No adverse reactions were observed.

Evidence summary

Three RCTs with EL = 1+ were identified for the effectiveness of BSS in the treatment of diarrhoea. Meta-analysis could not be performed because of the variability of the outcomes. Data from two RCTs with small sample sizes showed that children with acute diarrhoea and receiving BSS with ORS solution had a significantly reduced duration of diarrhoea, duration of hospital stay and need for fluid therapy. However, results from the third RCT, which had a large sample size, did not show a statistically significant reduction in the duration of diarrhoea, in the incidence of persistent diarrhoea or in the total intake of ORS solution in the group of children treated with BSS compared with the placebo group. In this trial, a statistically significant reduction in the duration of diarrhoea was seen only for the subgroup of children who were positive for rotavirus. Two trials showed a reduction in the stool output and one trial showed a statistically significant reduction in vomiting in patients receiving a higher dose of BSS. No evidence of benefit was found between the administration per day of 100 mg/kg versus 150 mg/kg of BSS in the treatment of diarrhoea in children for this outcome or any other outcome investigated. No adverse events were identified although two incidents of ‘black tongue’ during treatment were reported in one trial.

8.2.3. Antimotility agents

Evidence overview

One relevant systematic review of trials comparing loperamide with placebo for diarrhoea in children was identified.174 [EL = 1+]

8.2.3.1. Loperamide

This well-conducted systematic review174 included 13 RCTs published between 1980 and 1999. Four of the 13 studies were conducted in European countries, two in South Africa, two in Mexico, one in Costa Rica, one in India, one in Saudi Arabia and two in North Africa (Egypt and Libya). Across the 13 trials, data from 1788 children was collected, 975 were assigned to the intervention group (loperamide) and 813 to the control group. The methodology and conduction of the 13 RCTS was heterogeneous: method of randomisation and allocation concealment were only reported in six and seven studies, respectively, nine trials were double-blinded and two trials did not report inclusion of more than 90% of all randomised participants. Definitions for diarrhoeal resolution, rehydration protocols administered and doses of the study medication varied across the studies. Children recruited in the trials presented with different grades of dehydration and duration of diarrhoea prior to enrolment. However, in most of the trials, participants were mildly dehydrated and had had diarrhoea for less than 3 days prior to inclusion in the studies). Data were extracted and meta-analysis was performed for the following outcomes: diarrhoea continuing at 24 hours and at 48 hours, reduction in diarrhoea duration, diarrhoea counts for the first 24 hours and adverse events.

In the four trials reporting diarrhoea at 24 hours, the prevalence of diarrhoea among the group treated with loperamide was statistically significantly lower than in the control group (RR 0.66; 95% CI 0.57 to 0.78). When combining the data from the three trials defining the resolution of diarrhoea as the last unformed stool, the findings were not statistically significantly different. Data on the proportion of patients with diarrhoea at 48 hours were available in four studies. The meta-analysis performed showed that the loperamide group had a statistically significantly lower proportion of patients with diarrhoea when compared with the control group (RR 0.59; 95% CI 0.45 to 0.78). The mean duration of diarrhoea was obtained combining the data from six trials. It was found that the group receiving loperamide had a statistically significantly shorter duration of diarrhoea when compared with the control group (WMD −0.80 days; 95% CI −0.87 to −0.74 days). When restricting the analysis to those five studies administering a loperamide dose of up to 0.25 ml/kg per day, the findings were not statistically significantly different. Four studies were included in the meta-analysis for the number of stools at 24 hours. The group treated with loperamide showed a statistically significantly lower mean number of stools than the control group (count ratio = 0.84; 95% CI 0.77 to 0.92). Twelve RCTs reported information on serious adverse events (defined as presence of ileus, lethargy or death). When pooling the data together, it was found that eight participants out of 927 in the intervention group and none out of 764 in the control group underwent some serious adverse event. When abdominal distension and sleepiness were also included among the adverse events, it was found that in total 21 children out of 927 in the intervention group and four out of 764 in the placebo group suffered some kind of adverse event. These last findings were statistically significant.

Evidence summary

There was evidence from a well-conducted systematic review [EL = 1+] for the effectiveness of loperamide in the treatment of diarrhoea in children. Meta-analysis performed in the review showed that children receiving loperamide experienced less stool output and had a reduction of the duration of diarrhoea when compared with children that did not receive the drug. Serious adverse events only occurred in the children receiving loperamide and these participants also had statistically significantly more total adverse events than the children in the control groups.

GDG translation from evidence to recommendation

Diarrhoea is the predominant clinical symptom in gastroenteritis and a major cause of dehydration. It also causes concern to parents who may understandably ask whether there is a treatment available to alleviate it. Various antidiarrhoeal agents have been proposed and some have been widely used. However, as they have been considered relatively ineffective, unnecessary and potentially harmful, their use is avoided.

The GDG considered the evidence available regarding several adsorbent agents (kaolin, charcoal and smectite), an antisecretory agent (racecadotril), BSS and an antimotility agent (loperamide), and drew the following conclusions.

There was no evidence to support the use of kaolin. There was some evidence of possible benefit from activated charcoal but this came from one small study. Young children would probably find this agent unpalatable and adherence would be poor. There was evidence suggesting that smectite was an effective antidiarrhoeal, seemingly without adverse effects, at least in the short term. However, further research would be necessary to examine its potential clinical and health economic benefits in the UK.

There was evidence that racecadotril had an antidiarrhoeal effect but further research is required to examine the possible clinical and health economic benefits that might be associated with its use in the UK.

Studies on BSS had yielded inconsistent results in children and it was thought that any possible benefit was likely to be small.

Clinical trials on the antimotility agent loperamide had demonstrated a statistically significant antidiarrhoeal effect in children with gastroenteritis. Loperamide is not licensed for use in young children with acute diarrhoea in the UK. For that reason, but also given the reported adverse effects such as drowsiness, abdominal distension and ileus, its use was not recommended.

Recommendation on antidiarrhoeal agents

Do not use antidiarrhoeal medications.

Research recommendation

Is racecadotril (an enkephalinase inhibitor) cost-effective and safe in the treatment of gastroenteritis in children compared with a placebo?

8.3. Micronutrients and fibre

Zinc is an important trace element in gastrointestinal structure and function. It is involved in epithelial barrier integrity, tissue repair and immune function. Diarrhoea is associated with significant zinc loss.175 In developing countries, zinc deficiency may be common owing to inadequate food intake, reduced availability in animal food sources, and high phytate content in the diet resulting in impaired absorption.176 In 1995, a study from India first reported significant clinical benefit from zinc therapy in gastroenteritis.177 The WHO has recommended zinc supplementation in children with gastroenteritis. Vitamin A deficiency has been associated with an increased mortality rate in patients with gastroenteritis in developing countries. It has therefore been postulated that supplementation during the acute illness could be of clinical benefit. Glutamine is an amino acid that acts as an important substrate for rapidly dividing cells, including enterocytes and lymphocytes. It has therefore been postulated that glutamine supplementation might be of benefit in gastroenteritis.178 Folic acid is a water-soluble vitamin with a crucial role in DNA synthesis. It has been proposed that its administration might be beneficial in gastroenteritis, promoting mucosal repair and regeneration and enhancing the immune response.179 Dietary fibre (soy polysaccharide) supplementation during the illness has been proposed as an effective adjunctive treatment for the reduction of diarrhoea in children with gastroenteritis.

8.3.1. Zinc

Evidence overview

A Cochrane review180 was identified that included 18 trials in total. Eight of these trials were relevant here175,177,181–186 with the remainder being excluded as they had participants who were malnourished or who had persistent diarrhoea.

Three of the studies were conducted in India,177,182,185 two in Bangladesh,175,184 one in Brazil181 and one in Nepal.186 The remaining study was a multicentre trial carried out in Pakistan, India and Ethiopia.183 Seven studies had two treatment arms and one had three arms.175 Across the eight studies, 5155 participants suffering from acute diarrhoea (children aged 1–60 months) were enrolled. Four studies were hospital-based trials,175,181,182,185 three were community-based trials177,177,183 and one study included outpatient participants as well as inpatient.184

All eight trials included in this review were RCTs. In one study185 the method used to generate the allocation sequence was unclear but in the other seven trials the method of randomisation was considered adequate. Five studies assured adequate allocation concealment,177,182–184,186 but in the remaining three this was unclear.175,181,185 The outcome assessor was blinded in seven trials but allocation concealment was unclear in one.185 Only five trials reported a sample size power calculation.175,177,182–184 Seven trials included more than 90% of the randomised participants in the analysis. In one study the number of participants lost to follow-up was unclear.185 Definition criteria for inclusion of the participants varied between the studies. Dose and duration of the supplementation, formulation and type of zinc salts administered and follow-up were not uniform across all the studies.

Six trials administered zinc alone175,181,183–186 and two trials administered zinc with a multivitamin preparation.177,182 In one trial181 the placebo group received vitamin C while the intervention group did not. The outcomes measured were duration of diarrhoea, stool output, hospitalisation, adverse events (vomiting) and death. Not all of the studies included reported all the outcomes.

One RCT (n = 81)181 was aimed at determining the effect of oral zinc supplementation on the duration of acute diarrhoea in young children. Participants were children aged from 3 to 60 months suffering from diarrhoea for less than 7 days and with, at least, mild dehydration. Children were hospitalised and followed for 5 days or until resolution of diarrhoea. [EL = 1−]

Another RCT (n = 287)182 examined the clinical benefit of zinc supplementation as an adjunct to ORT on stool output and duration of diarrhoea in children with non-cholera diarrhoea. Participants were outpatients aged between 3 and 36 months, with diarrhoea for less than 72 hours and mild dehydration. [EL = 1+] The findings showed that there was no statistically significant difference in the proportion of children with diarrhoea by the 5th day from the start of the study in each of the study groups. The geometric mean of the total stool output in the intervention group was 111 g/kg (95% CI 86 to 147 g/kg) and in the placebo group was 148 g/kg (95% CI 116 to 190 g/kg). The effect size favoured the group receiving the zinc supplementation (ratio of geometric means = 0.69; 95% CI 0.48 to 0.99).

Another RCT (n = 275)175 compared two different doses of zinc supplementation (5 mg and 20 mg of zinc acetate) with placebo on the duration of diarrhoea and stool output in infants with acute diarrhoea. Participants were male infants aged between 1 and 6 months, hospitalised and with diarrhoea for less than 3 days. [EL = 1+] The trial reported the geometric mean and confidence intervals for stool frequency and total stool output. There were no statistically significant differences in either outcome between the groups receiving 5 mg zinc, 20 mg zinc or for those receiving placebo.

A multicentre randomised controlled trial (n = 1110)183 assessed safety and therapeutic effects of providing zinc supplementation to children suffering from diarrhoea. It compared the impact of administering 10 mg of zinc sulfate per day for 14 days with that of placebo for the treatment of acute diarrhoea. The participants were outpatients aged 1–5 months. [EL = 1+]

Another RCT (n = 1067)184 was aimed at determining whether daily zinc supplementation was associated with an increase risk of vomiting in children with diarrhoea. The study compared 20 mg of zinc supplementation per day for 10 days with placebo. Participants were young children aged between 3 and 59 months with diarrhoea and admitted in hospital or in the outpatient clinic. [EL = 1+]

Another RCT (n = 50)185 assessed the impact of zinc supplementation on the duration of diarrhoea and stool frequency in children with acute dehydrating diarrhoea. It compared the administration of 40 mg of elemental zinc with the administration of placebo in children aged 6–18 months. Participants were hospitalised and followed until recovery. [EL = 1−] It was reported that no vomiting occurred during the follow-up in either group.

Another RCT (n = 947)177 was aimed at evaluating the effects of daily supplementation with 20 mg of elemental zinc on the severity of acute diarrhoea. The study compared the administration of zinc with the administration of placebo. Participants were children aged from 6 to 35 months and who had diarrhoea for less than 7 days. All participants enrolled received multivitamin supplementation. [EL = 1+]

Another study (n = 899)186 was designed to measure the impact of daily zinc supplementation on the duration and severity of acute diarrhoea in children. The trial also assessed the effect of administering zinc with vitamin A but only the zinc-supplemented group and the placebo group have been considered here. Participants were children aged between 6 and 35 months who presented with diarrhoea for less than 96 hours. [EL = 1+] The proportion of children with diarrhoea by the 3rd day of follow-up was reported. It showed a statistically significant difference favouring the group that received the zinc supplementation (27% of children with diarrhoea) when compared with the group receiving the placebo (35% of children with diarrhoea) (RR 0.75; 95% CI 0.61 to 0.91). When the two treatment groups were compared, the ratio of the geometric means of the number of stools in the first 4 days of follow-up showed a statistically significant difference that favoured the group receiving the zinc supplementation (RR 0.91; 95% CI 0.85 to 0.97). Vomiting was found to be significantly higher in the zinc supplementation group (16%) compared with the placebo group (8.7%) (RR 1.7; 95% CI 1.4 to 2.2), although vomiting events were equal among the two groups.

Five trials reported the mean duration of diarrhoea. The results of these trials are presented according to participants’ age (Figure 8.4). One small trial181 reported a statistically significant reduction in the mean duration of diarrhoea for children given zinc compared with those given placebo (WMD −31.2 hours; 95% CI −46.4 to −16.0 hours). Meta-analysis of all five trials using the random effects model found no statistically significant difference in mean duration of diarrhoea between those receiving zinc (n = 903) or placebo (n = 821) (WMD −4.4 hours; 95% CI −15.9 to 7.2 hours).

Figure 8.4. Comparison of the effect of zinc versus placebo on the mean duration of diarrhoea.

Figure 8.4

Comparison of the effect of zinc versus placebo on the mean duration of diarrhoea.

Four trials (n = 3168) reported the proportion of children with diarrhoea by day 7. Results are presented according to participants’ age in Figure 8.5. Two trials182,186 reported that statistically significantly fewer children given zinc had diarrhoea by day 7 compared with those given placebo. The first reported this finding in children aged between 6 and 35 months (RR 0.58; 95% CI 0.38 to 0.87) and the second in children aged between 3 and 36 months (RR 0.11; 95% CI 0.01 to 0.88). However, when the results of the four trials were pooled together, there was no statistically significant difference in the proportion of children with diarrhoea by day 7 between those receiving zinc (n = 1568) or placebo (n = 1600) (RR 0.90; 95% CI 0.64 to 1.27).

Figure 8.5. Comparison of the effect of zinc versus placebo on the proportion of children with diarrhoea by day 7.

Figure 8.5

Comparison of the effect of zinc versus placebo on the proportion of children with diarrhoea by day 7.

Six trials reported outcomes for stool frequency. Four trials (n = 2135) reported the mean number of stools per day. Results are presented according to participants’ age in Figure 8.6. Two trials177,181 reported that children given zinc had statistically significantly lower stool frequency than those given placebo. The first trial made this finding in children aged between 6 and 35 months (WMD −2.00; 95% CI −3.61 to −0.39) and the second in children aged between 3 and 36 months (WMD −5.20; 95% CI −8.52 to −1.88). Pooled results of the four trials found no statistically significant difference in the mean stool frequency between those receiving zinc (n = 1056) or placebo (n = 1079) (WMD −0.32; 95% CI −0.8 to 0.17). However a statistically significant reduction in stool frequency was seen in children aged over 6 months (two RCTs, WMD −1.90; 95% CI −3.22 to −0.58).

Figure 8.6. Comparison of the effect of zinc versus placebo on stool frequency.

Figure 8.6

Comparison of the effect of zinc versus placebo on stool frequency.

Six trials reported results for vomiting. Four trials (n = 2475) reported the proportion of children who vomited. Results are presented according to participants’ age in Figure 8.7. One trial184 reported that statistically significantly more children given zinc had vomited compared with those given placebo. This finding was reported for children seen in inpatients (RR 1.95; 95% CI 1.64 to 2.32) and outpatients (RR 2.53; 95% CI 2.04 to 3.13). The data from all four trials (n = 2475) were combined in a meta-analysis that showed a statistically significant increase of vomiting in children receiving zinc supplementations when compared with children receiving placebo (RR 1.63; 95% CI 1.11 to 2.40).

Figure 8.7. Comparison of the effect of zinc versus placebo on the proportion of children who vomited.

Figure 8.7

Comparison of the effect of zinc versus placebo on the proportion of children who vomited.

Evidence summary

Eight RCTs, none of them located in European countries, were identified for the effectiveness of zinc supplementation in the treatment of acute diarrhoea in children. Although one small trial demonstrated a statistically significant reduction in duration of diarrhoea in zinc-supplemented children when compared with the control group, results from a meta-analysis (five RCTs) did not find a statistically significant reduction. One study reported that statistically significantly more children who had received zinc did not have diarrhoea at day 3 compared with placebo. Meta-analysis performed regarding the proportion of children with diarrhoea by day 7 (four RCTs) and stool frequency (four RCTs) showed no statistically significant differences between the zinc and placebo groups, although two studies presenting results for each of these outcomes did report statistically significant improvement with zinc administration. When data from four trials were combined this showed a statistically significantly increased number of vomiting events among children supplemented with zinc when compared with children in the control group. There was significant heterogeneity in the meta-analyses undertaken in this review. Adjustment for nutritional, geographical and treatment differences did not affect the significance of the findings, suggesting that no one single variable explained the heterogeneity found alone. The applicability of these results might therefore be limited to those settings with similar population characteristics.

8.3.2. Vitamin A

Evidence overview

Three trials were identified for this review.187–189 Two of these were conducted in Bangladesh187,188 and one in Turkey.189 All three had two treatment arms. In total, data from 286 children (aged 6 months to 7 years) suffering from acute diarrhoea were collected across the three studies. Follow-up, treatment protocol and definition criteria for inclusion of the children with acute diarrhoea varied between the three studies. The outcomes considered were duration of the disease, stool output, vomiting, clinical cure, bacteriological cure and treatment failure.

Two studies compared the effects of 200 000 iu vitamin A with placebo in children with diarrhoea.187,188

One RCT (n = 83)187 examined the therapeutic effect of vitamin A in children suffering from acute diarrhoea. The participants were boys aged between 1 and 5 years presenting with watery non-cholera diarrhoea of less than 48 hours’ duration. The study compared the administration of a single oral dose of 200 000 iu vitamin A plus 25 iu vitamin E with the administration of placebo, which consisted on 25 iu vitamin E. The methods of randomisation were adequate. Allocation concealment details were not given, although the patients and outcome assessors were stated to be blinded to treatment allocation. Comparability of the groups at study entry was adequate in all the studies and the proportion of patients randomised but lost to follow-up was under 20%. [EL = 1+]

There were no statistically significant differences between the two groups for the total duration of the diarrhoeal episode, for the proportion of children with a diarrhoeal episode lasting more than 10 days, for the total stool output or for the stool output in the first 24 hours after the start of the intervention. The study found no statistically significant differences between study groups for the volume vomited per day or for the rate of treatment failure (defined by the researchers as the need for IVT after initial oral rehydration).

Another RCT (n = 90)188 examined the effects of a single administration of 200 000 iu vitamin A in children with shigella infection compared with the administration of a placebo. The patients were children aged between 1 and 7 years who had presented to the outpatient department with diarrhoeal stools. The methods of randomisation and allocation concealment were adequate and a sample size power calculation was given. The patients and outcome assessors were blind to treatment allocation. The outcomes measured in the trial were achievement of clinical and bacteriological cure on study day 5. [EL = 1+]

Comparability of the groups at study entry was adequate and the proportion of patients randomised but lost to follow-up was under 20%. ‘Clinical cure’ was defined as three or fewer formed stools in a day without any visible blood or mucus and absence of fever and abdominal pain. The authors found that 45% of children in the intervention group (19/42) and 20% in the control group (8/41) were clinically cured by day 5. The difference was statistically significant (RR 2.32; 95% CI 1.15 to 4.69). No statistically significant difference in bacteriological cure (defined as the continuous absence of shigella in both stool and rectal swab samples, from study day 3 onwards) was found between the two groups,

A quasi-randomised controlled trial189 compared the therapeutic value of a single dose of 100 000 iu oral vitamin A in outpatient children with acute diarrhoea against the administration of placebo. The 120 children enrolled in the study were infants aged between 6 and 12 months suffering from acute diarrhoea for less than 5 days. Those with malnutrition, dysentery or a chronic condition were excluded. The method of randomisation was based on the patients’ file numbers (odd or even). The patients and outcome assessors were blind to treatment allocation and a sample size power calculation was given. [EL = 1−]

The groups were comparable at study entry. The proportion of patients randomised but lost to follow-up was reported as under 20%. No statistically significant differences in the treatment groups were reported for the mean total duration of diarrhoea or for persistent diarrhoea.

Evidence summary

Two RCTs located in Bangladesh [EL = 1+] and one quasi-RCT from Turkey [EL = 1−] were identified for the effectiveness of vitamin A supplementation in the management of acute diarrhoea in children. Meta-analysis could not be performed because of the variability of the studies. Data from one of the RCTs showed no evidence of benefit in the duration of the diarrhoeal episode, in the stool output, vomiting or number of treatment failures when comparing children that received 200 000 iu of vitamin A with children receiving placebo. The other RCT suggested that children with shigella infection supplemented with 200 000 iu of vitamin A were more likely to have fewer formed stools with absence of fever and abdominal pain by day 5 than the placebo children. The trial did not show a statistically significant difference between the two groups when considering bacteriological cure. The quasi-RCT showed no statistically significant differences in duration of diarrhoea or the onset of persistent diarrhoea between children receiving 100 000 iu of vitamin A and children receiving placebo.

8.3.3. Glutamine

Evidence overview

A single-centre study undertaken in Turkey was identified.178

The study was a quasi-randomised controlled trial in which a total of 159 infants were enrolled. The participants were children aged 6–24 months with diarrhoea of less than 10 days’ duration. Those children with chronic conditions, severe malnutrition, associated infectious diseases or having been under antibiotic or antidiarrhoeal therapy were excluded from the trial. Eligible cases in the study were divided into two groups according to their hospital file number on admission, and thus allocation concealment was inadequate. The trial assessed the effect of administering 0.3 g/kg per day of glutamine in the treatment of acute diarrhoea in children. Mothers administered the supplement (glutamine or placebo) in three daily doses for 7 days. The authors reported that patients (and carers) and assessors were blinded to the treatment received and the sample size was calculated based on a preliminary study that looked at duration of diarrhoea in 15 subjects. The outcome considered was duration of the diarrhoeal episode. [EL = 1−]

Loss to follow-up was nearly 20%. The comparability between the two groups at the start of the trial was adequate. The mean duration of diarrhoea in the group receiving glutamine (n = 63, mean 3.40 days, SD 1.96 days) was shorter than in the control group (n = 65, mean 4.57 days, SD 2.48 days). This finding was statistically significant (WMD −1.17 days; 95% CI −1.94 to −0.40 days). Data relating to duration of diarrhoea after treatment were stratified by stool frequency on admission. Children in the glutamine group with a stool frequency below eight per day (n = 46) had a mean duration of diarrhoea of 3.30 days (SD 1.96 days) compared with 4.68 days (SD 2.60 days) in those receiving the placebo (n = 41). This finding was statistically significant (WMD −1.38 days; 95% CI −2.36 to −0.40 days). When the data collected from the children with high stool frequency, eight or more stools per day, were analysed, no statistically significant difference in mean duration of diarrhoea was observed between the groups. No statistically significant differences in the proportion with persistent diarrhoea or vomiting were found.

Evidence summary

There was a lack of high-quality evidence for the effectiveness of glutamine supplementation in the treatment of acute diarrhoea in children. A quasi-randomised controlled trial conducted in Turkey [EL = 1−] showed no difference in the onset of persistent diarrhoea or vomiting. On the other hand, the study suggested that glutamine supplementation shortens the duration of diarrhoea by 1 day.

8.3.4. Folic acid

Evidence overview

A single study located in Bangladesh was identified as relevant and included in the review.179 This RCT recruited 106 male children aged between 6 and 23 months who presented with watery diarrhoea (of less than 72 hours’ duration) and some degree of dehydration. They were enrolled after admission into the study ward. The participants were randomised into two treatment arms: those receiving 5 mg folate (n = 54) and those receiving placebo (n = 52) every 8 hours for 5 days. The method of randomisation and allocation concealment were not reported by the authors. Subjects and investigators were blinded to the treatment administered. A power calculation was performed. The outcomes measured were course of illness, stool output, ORS solution intake and IV fluids received. [EL = 1+]

Baseline comparability between the two groups was adequate but loss to follow-up was not reported. No statistically significant differences were observed between the two groups for the mean duration of diarrhoea, the proportion having diarrhoea beyond 5 days, the mean total stool output, the mean total intake of ORS solution or the proportion of all children receiving IVT. Data were also analysed for rotavirus-positive children (n = 63) and the results did not show any statistically significant differences in the intake, output or duration of diarrhoea between the folic acid group (n = 34) and the placebo group (n = 29).

Evidence summary

Evidence from an RCT [EL = 1+] located in Bangladesh suggested that 5 days of folate supplementation for the treatment of children with acute diarrhoea did not reduce the duration of diarrhoea, the stool output, the ORS solution intake or the number of children requiring IVT.

8.3.5. Fibre

Evidence overview

Two RCTs were included in this review.190,191 One of the studies was conducted in Peru190 and the other in the USA.191 Both trials had two treatment arms. In total, they recruited 91 children with acute diarrhoea.

One hospital-based RCT (n = 34)190 examined the therapeutic effect of dietary fibre on the severity and duration of acute watery diarrhoea. The study compared the effects of the administration of a soy protein lactose-free formula with added fibre (0.6 g/ml) with the administration of a soy protein formula without fibre in children. The participants were hospitalised male children aged between 2 and 24 months suffering from acute diarrhoea for less than 96 hours. Although the reported methods of randomisation were adequate, the allocation concealment was unclear. A sample size power calculation was not given. The outcomes considered were duration of diarrhoea (defined as the number of hours post admission until excretion of the last liquid or semi-liquid stool not followed by another abnormal stool within 24 hours) and treatment failure (defined as recurring dehydration above 5%, electrolyte disorders after initial rehydration or important faecal output during intervention). [EL = 1−]

Comparability of the groups at study entry was adequate and the proportion of participants lost to follow-up was 15%. The median duration of diarrhoea after hospitalisation was estimated at 43 hours in the intervention group and 163 hours in the control group. The authors reported the difference as statistically significant (P < 0.003). The difference between the two groups for treatment failure was not statistically significant.

One community-based RCT (n = 55)191 assessed the efficacy of dietary fibre in reducing the duration of watery diarrhoea in middle-class American children. The trial compared the effects of feeding children aged under 24 months with a soy fibre supplemented infant formula for 10 days against a standard soy formula in shortening the duration of acute diarrhoea. Although the reported methods of randomisation were adequate, the allocation concealment was unclear. A sample size power calculation was not given. The main outcome measured was mean duration of diarrhoea. [EL = 1−]

Comparability of the groups at study entry was adequate and the proportion of participants lost to follow-up was 25%. Results were stratified by age over or under 6 months. When comparing the two study groups in children over 6 months in age, the authors found a statistically significant difference that favoured the administration of formula with added fibre in shortening the duration of the diarrhoeal episode. The mean duration of diarrhoea was 9.7 hours in the intervention group and 23.1 hours in the control group (P < 0.05). The difference was not statistically significant when they compared the mean duration of diarrhoea in infants younger than 6 months. The authors reported for all children that there was no statistically significant difference in the duration of diarrhoea between the children formula-fed with added dietary fibre (12.2 hours) and the ones fed with no added fibre (16.9 hours).

Evidence summary

There was a lack of high-quality evidence on the clinical effectiveness of administering soy protein formula with added fibre in children with acute diarrhoea. Data from one RCT located in Peru [EL = 1−] showed a statistically significant reduction in the duration of diarrhoea but no difference in the number of treatment failures between the group receiving the fibre supplementation and the control group. Another RCT conducted in the USA [EL = 1−] suggested that, in children aged over 6 months, supplementation with fibre shortened the duration of diarrhoea by more than 12 hours. However, no statistically significant effect was observed in the combined group (under and over 6 months of age) or in children younger than 6 months.

GDG translation from evidence to recommendation

Many of the studies on the effectiveness of micronutrients and vitamins in the treatment of gastroenteritis were carried out in populations at risk of nutritional deficiency.

The GDG was aware that a recent Cochrane review had concluded that zinc supplementation could be effective in the treatment of diarrhoea and vomiting in children with gastroenteritis in areas where diarrhoea was an important cause of child mortality. The studies judged relevant to this guideline demonstrated some benefit from zinc in reducing stool frequency but not the mean duration of diarrhoea. There was some evidence that zinc treatment is associated with increased vomiting. The GDG therefore concluded that there was insufficient evidence to justify recommending zinc supplementation for well-nourished children with gastroenteritis.

There was no research evidence that vitamin A administration had a beneficial effect in children with gastroenteritis (with the possible exception of those with shigella), despite the fact that most of the trials took place in settings where malnutrition might be expected. There was little evidence to support a beneficial effect from glutamine supplementation in the treatment of gastroenteritis. There was no evidence to support the use of folic acid therapy, with no benefit being seen in a study carried out in a population that might have been at risk of malnutrition. Although there was some evidence suggesting possible benefit from the use of fibre-supplemented milk formulas in reducing the duration of diarrhoea, the trials were not of high quality.

8.4. Alternative and complementary therapies

Evidence regarding the possible value of alternative or complementary therapies in the treatment of gastroenteritis was sought. Only two studies of homeopathy remedies were identified and examined.

Evidence overview

A systematic review and an RCT were identified that examined the effects of homeopathy compared with placebo for the treatment of acute diarrhoea in children.

The systematic review192 examined the effectiveness of individualised homeopathy therapy (with one of 19 possible prescriptions) compared with placebo for the treatment of acute diarrhoea in children. [EL = 1−] The results of three RCTs were included and meta-analysed in the review. The trials had been conducted by the review’s first author in Nicaragua (two RCTs: pilot study (n = 33) and main study (n = 81)) and Nepal (one RCT: n = 116) applying similar methodology and design.

The review included results from 247 children aged 6 months to 5 years with a history of diarrhoea (defined as three or more unformed stools per day) for no more than 7 days (Nicaragua) or 5 days (Nepal). Children were excluded if they had received antidiarrhoeal treatment within 24–48 hours prior to enrolment or if they had severe diarrhoea requiring hospitalisation or IV hydration. Children were assessed on enrolment for baseline characteristics, given ORT as necessary according to WHO recommendations and then interviewed by a homeopathic practitioner who used a computer program to prescribe the appropriate homeopathic treatment from a choice of 19 possible remedies. These had been prepared previously in the USA by a homeopathic pharmacist to a liquid homeopathic dilution in the 30C potency. Randomisation was performed by sequential assignment of children to pre-randomised and coded tubes of either placebo or homeopathic therapy. Parents were instructed to give their child 1 tablet from the prescribed tube after each unformed stool, to be dissolved in the mouth. Follow-up was performed by parents and auxiliary nurses for 5 days (for 6 days in the pilot Nicaraguan trial) after the initial visit. All study staff and the patients were blind to treatment allocation. The primary outcome measures were duration of diarrhoea (defined as the number of days until there were two consecutive days with less than three unformed stools per day) and mean number of stools per day.

When data were pooled, children were similar at baseline for sex and diarrhoeal outcomes prior to study enrolment. However, children in the placebo group were significantly younger, shorter and lighter than those receiving homeopathy treatments. This reflected a discrepancy in randomisation in the Nepali study, as groups were essentially similar at baseline in the Nicaraguan studies. This bias would tend to overestimate any difference in treatment effect seen between the two groups.

Combined results suggested that children receiving homeopathy had a statistically significantly shorter duration of diarrhoea (3.1 ± 2.0 days versus 3.8 ± 1.9 days; P = 0.008) and statistically significantly fewer number of stools per day (2.7 ± 2.0 versus 3.4 ± 2.0 stools per day; P = 0.004) compared with those receiving placebo. However, as overall children in the placebo group were younger and smaller than those receiving homeopathy, the reliability of the meta-analysis results are questionable.

One RCT193 recruited 292 children aged between 5 months and 6 years who presented to a municipal acute care clinic in Honduras. [EL = 1+] The children had acute diarrhoea that was confirmed visually by study staff (defined as the passage of three or more unformed stools in the previous 24 hours). Children were excluded if the diarrhoea had lasted more than 4 days, there was visible blood in the stool, they were severely dehydrated or they lived outside the catchment area of the clinic. Once enrolled, the participants’ parents were given bottles of premixed ORT and a vial of tablets. Randomisation was performed by sequential assignment of children to pre-randomised and coded vials of either placebo (n = 134) or homeopathic combination therapy (n = 131) tablets. The combination therapy was prepared in the USA by a homeopathic pharmacist and contained the five most common single remedies – Arsenicum album, Calcarea carbonica, chamomilla, podophyllum and sulphur – in a liquid homeopathic dilution in the 30C potency. Parents were instructed to give their child two tablets after each unformed stool, to be dissolved in the mouth, and to give ORT according to WHO recommendations. Follow-up was performed by parents and auxiliary nurses for 7 days after the initial visit or until symptoms resolved, if sooner (defined as two consecutive days with fewer than three unformed stools each day). The main outcome measures were duration of diarrhoea, mean rate of unformed stool passage per day during follow-up and total number of unformed stools during follow-up. All study staff and the patients were blind to treatment allocation.

Results were presented for 265 children (9% loss to follow-up). The groups were similar for age, sex, height, weight, body temperature, vomiting, dehydration status and duration of diarrhoea and unformed stools prior to study entry. There were no statistically significant differences in the distribution of identified pathogens between the treatment and placebo groups. No statistically significant differences between treatment with homeopathic combination therapy or placebo were reported for the three main outcomes of duration of diarrhoea, mean rate of unformed stool passage per day during follow-up or total number of unformed stools during follow-up. Univariate adjustment for baseline characteristics did not alter these results.

Evidence summary

Evidence from a systematic review and meta-analysis of three RCTs suggested that individualised homeopathy treatment reduced the duration and stool frequency of diarrhoea compared with placebo in young children. [EL = 1+] However, as overall the treatment groups were not similar for age, weight and height at baseline, these findings should be treated cautiously as they may overestimate the true treatment effect size.

Evidence from an RCT examining the effects of a combined homeopathy tablet compared with placebo found that there were no differences in effect on duration of diarrhoea, mean rate of unformed stool passage per day during follow-up or total number of unformed stools during follow-up in young children. [EL = 1+]

GDG translation from evidence to recommendation

The GDG considered that the clinical trials assessing homeopathy had significant methodological limitations. Moreover, there was a lack of consistency in the evidence. Therefore, no recommendation was made for the use of homeopathy.

8.5. Probiotics

In 1985, a lactobacillus was identified through screening of bacteria in fermented milk products that was acid and bile resistant, adhered to intestinal epithelial cells.140,194 This was Lactobacillus rhamnosus GG, a non-pathogenic organism. The current view of probiotic therapy is based on the concept of a normal balanced intestinal microbiota.195 Much research has been directed towards examining the potential benefit of a variety of probiotics, including various lactobacillus species and other non-pathogenic microbes in the treatment of infectious gastroenteritis. The possible mechanisms of action include competition with pathogens for binding sites and substrates, lowering of intestinal luminal pH, production of bacteriocins, promotion of mucin production, upregulation of genes mediating immunity, and production of trophic short-chain fatty acids to promote mucosal cell growth and differentiation.195 The published evidence on the use of probiotics in the treatment of children with gastroenteritis was therefore examined.

Evidence overview

Seven studies were identified for inclusion: three systematic reviews196–198 and four RCTs.40,199–201

8.5.1. Lactobacillus and saccharomyces yeast probiotics

One systematic review,196 published in 2003, was conducted to examine the effectiveness of probiotics compared with control in the treatment of infectious diarrhoea. This well-conducted systematic review included 23 RCTs published between 1981 and 2002. Fourteen studies were carried out in developing countries. The included studies enrolled a total of 1917 participants. Of these, 1449 were children (740 were randomised into the intervention group and 709 into the control group) and 352 were adults (173 were randomised into the intervention group and 179 into the control group). The participants included were inpatients as well as outpatients. Although all the studies enrolled participants with acute diarrhoea, definition criteria were not uniform across the studies. Many studies did not comment on the nutritional status of the participants but those with underlying severe or chronic illnesses were excluded. Two studies recruited specifically malnourished children and a further two studies included malnourished children. The methodology was described in detail and studies were appraised for their quality. The methodology and design of the trials included in the review were not uniform and only three studies reported an adequate method of randomisation, allocation concealment, blinding and loss to follow-up. Different types of probiotic were evaluated and there was a wide range of treatment regimens. The number of organisms administered, duration of treatment, and timing of intervention and means of administration varied across all the studies. Data were extracted and meta-analysis performed for the following outcomes: diarrhoea lasting 3 or more days, diarrhoea lasting 4 or more days, duration of diarrhoea, and mean stool frequency on day 2 and on day 3. Other outcomes reported were need for unscheduled IVT and death. [EL = 1++]

The persistence of diarrhoea on day 3 of the intervention was reported in 15 studies (1341 participants). A meta-analysis was performed that showed that those receiving probiotics were less likely to have diarrhoea lasting 3 or more days (RR 0.66; 95% CI 0.55 to 0.77), but there was heterogeneity (I2 =46.6%) between studies. When data from children were pooled (11 RCTs, n = 1008), the analysis also showed that statistically significantly more of those receiving placebo (265/490) had persistence of diarrhoea on day 3 compared with those receiving probiotics (195/518) (RR 0.68; 95% CI 0.54 to 0.85; I2 = 50.2%) (Figure 8.8).

Figure 8.8. Comparison of the effect of probiotics versus placebo on diarrhoea (persistence of diarrhoea on day 3).

Figure 8.8

Comparison of the effect of probiotics versus placebo on diarrhoea (persistence of diarrhoea on day 3).

Data from 13 studies were pooled in a meta-analysis. The relative risk of diarrhoea lasting 4 or more days in the group treated with probiotics when compared with the control group was 0.31 (95% CI 0.19 to 0.50). However, there was heterogeneity between studies (I2 = 72.9%). A meta-analysis pooling together the data from the nine RCTs that recruited children (n = 895) showed that more of those receiving placebo (168/436) had persistence of diarrhoea on day 4 or beyond compared with those receiving probiotics (79/459) (RR 0.26; 95% CI 0.13 to 0.52; I2 = 62%) (Figure 8.9).

Figure 8.9. Comparison of the effect of probiotics versus placebo on diarrhoea (persistence of diarrhoea on day 4).

Figure 8.9

Comparison of the effect of probiotics versus placebo on diarrhoea (persistence of diarrhoea on day 4).

The mean duration of diarrhoea was measured in 12 studies that included only children. Those children receiving the probiotic agent had a statistically significantly shorter duration of diarrhoea when compared with the children in the control group (WMD −30.5 hours; 95% CI −42.5 to −18.5 hours; random effects model; I2 = 85.6%) (Figure 8.10).

Figure 8.10. Comparison of the effect of probiotics versus placebo on the duration of diarrhoea.

Figure 8.10

Comparison of the effect of probiotics versus placebo on the duration of diarrhoea.

The mean stool frequency on day 2 was reported in five trials (n = 417) and on day 3 it was reported in four trials (n = 447). Participants in the probiotic group had statistically significantly fewer stools on day 2 of intervention (WMD −1.51; 95% CI −1.85 to −1.17) and on day 3 of intervention (WMD −1.31; 95% CI −1.56 to −1.07). Data extracted for the stool frequency from the studies that recruited children were included in a meta-analysis. The analysis showed that on day 2 (four RCTS, n = 232), children treated with probiotics passed statistically significantly fewer stools than those receiving placebo (WMD −1.01; 95% CI −1.66 to −0.36). On day 3 of intervention (two RCTs, n = 170), children in the probiotic group passed statistically significantly fewer stools than children in the control group (WMD −1.12; 95% CI −1.79 to −0.46). The review reported that occasionally children developed severe dehydration requiring parenteral fluid therapy but in none of the studies was this attributable to the administration of the probiotic agent causing an adverse event. No death events were recorded among the included studies.

Evidence summary

There was evidence from a well-conducted systematic review examining the effectiveness of probiotics compared with control in the treatment of acute diarrhoea in children. The review showed that children receiving a probiotic had a reduction in the duration of diarrhoea and in the stool frequency. However, there was evidence of significant statistical heterogeneity and there was variation across the included studies regarding the specific probiotic employed, the therapeutic regimen used, the methodology and the population included.

8.5.2. Lactobacillus rhamnosus GG

One systematic review197 evaluated the effectiveness of Lactobacillus rhamnosus GG in the treatment of acute infectious diarrhoea in children. This well-conducted systematic review included eight RCTs involving 988 participants, 494 in the intervention group and 494 in the control group. Of the eight studies, four were located in European countries, three in South America and one in Pakistan. Participants were children aged between 1 and 36 months with acute diarrhoea, and were inpatients as well as outpatients. They presented with different grades of dehydration and duration of diarrhoea prior to enrolment. However, in most of the trials, participants were mildly dehydrated and had diarrhoea for less than 3 days prior to inclusion in the studies. The trial conducted in Pakistan included undernourished children. Definition criteria for diarrhoeal resolution, rehydration protocols administered and doses of the study medication varied across the studies. The methodology and conduct of the eight RCTs was heterogeneous: method of randomisation and allocation concealment were unclear or inadequate in four and six studies, respectively, two trials were not blinded and three trials did not include an intention-to-treat analysis. In one study, 43% of the participants enrolled did not complete the follow-up. Data were extracted and meta-analysis was performed for the following outcomes: duration of diarrhoea, diarrhoea on day 3, stool output and hospitalisation. In addition, the review looked at vomiting and treatment failure. [EL = 1+]

Seven trials measured the duration of diarrhoea (n = 876). The authors of the review performed a meta-analysis (high heterogeneity, with I2 = 97.4%) showing that children receiving the Lactobacillus rhamnosus GG, compared with children in the control group, experienced a statistically significant reduction of 1 day in the duration of the diarrhoeic episode (WMD −1.08 days; 95% CI −1.87 to −0.28 days) (Figure 8.11).

Figure 8.11. Comparison of the effect of Lactobacillus rhamnosus GG versus placebo on the duration of diarrhoea.

Figure 8.11

Comparison of the effect of Lactobacillus rhamnosus GG versus placebo on the duration of diarrhoea.

Three trials reported the duration of rotavirus diarrhoea in children (n = 201). The data were pooled, despite high heterogeneity (I2 = 94%), showing that children in the intervention group had a statistically significantly shorter duration of diarrhoea than those children in the control group (WMD −2.08 days; 95% CI −3.55 to −0.60 days). Two RCTs (n = 303) showed no statistically significant difference between the children treated with the probiotic and the children in the control group. The results of three studies were combined (I2 = 86.4%) to show that the mean hospital stay was not statistically significantly different among children receiving the Lactobacillus rhamnosus GG and the control group. One trial (n = 36) measured the number of emetic episodes on day 1 of the study and on day 2. The trial found no statistically significant difference between the two groups in the frequency of vomiting on day 1. On day 2, the difference was statistically significant favouring the intervention group (WMD −2.95; 95% CI −3.4 to −0.6). One trial reported two cases of myoclonic jerks (one case in each study group) and another trial reported no adverse events.

Evidence summary

There was evidence from a well-conducted systematic review [EL = 1+] for the effectiveness of Lactobacillus rhamnosus GG in children suffering from acute diarrhoea. Data from seven trials were combined, showing that children treated with Lactobacillus rhamnosus GG experienced a 1 day reduction in the duration of diarrhoea but there was significant statistical heterogeneity. Analysis performed on total stool output and hospitalisation showed no statistically significant difference between children receiving Lactobacillus rhamnosus GG and children that did not receive the probiotic.

8.5.3. Lactobacillus acidophilus LB

Evidence overview

One RCT was identified for inclusion.201 This study, conducted in Peru, randomised a total of 80 children with acute diarrhoea presumed to be of infectious origin into two treatment arms (40 participants in each group). Children with signs of dehydration requiring hospitalisation and those with illnesses other than the acute diarrhoeal episode were excluded. Lactobacillus acidophilus LB administration over 4.5 days was compared with placebo. Methods of randomisation and allocation concealment were unclear although the study was stated to be double-blind. [EL = 1+].

The baseline comparability of the two groups at the start of the study was adequate. Three participants out of 80 were lost to follow-up. No statistically significant differences were observed between the two groups for any of the outcomes considered, which were duration of diarrhoea, proportion of children with diarrhoea at the end of the study, total ORS solution intake and vomiting. One child in the probiotic group had severe dehydration and was withdrawn from the study and another child from the placebo group developed an itchy rash.

Evidence summary

An RCT located in Peru [EL = 1+] examined the effectiveness of Lactobacillus acidophilus LB in the treatment of acute diarrhoea in children. It showed no statistically significant differences between the probiotic group and the placebo group when considering duration of diarrhoea, ORS solution total intake, vomiting, adverse events and proportion of children without diarrhoea by the end of the study.

8.5.4. Lactobacillus paracasei strain ST11

Evidence overview

One RCT conducted in Bangladesh was identified.200 In total, 230 male infants and young children aged 4–24 months in the course of diarrhoea of less than 48 hours’ duration were randomly divided into two treatment arms to evaluate the therapeutic benefit of administering lyophilised Lactobacillus paracasei strain ST11 (n = 115) compared with placebo (n = 115) for 5 days. Children with bloody diarrhoea, with severe malnutrition or requiring antibiotic therapy were excluded. Children whose stool sample resulted positive for Vibrio cholerae were also excluded. The method of randomisation and allocation concealment were adequate. The trial was performed double-blind. The outcomes measured were mean duration of diarrhoea, cessation of diarrhoea, total stool output and total ORS solution intake. Participants were followed in hospital for 6 days or until cessation of diarrhoea. [EL = 1+]

The baseline comparability of the two groups under study was adequate. and the loss to follow-up was 11.8%. There were no statistically significant differences in the findings for the two groups for the mean duration of diarrhoea, the proportion of children without diarrhoea by the end of the 6th day, the mean total stool output and the mean total ORS solution. Findings for children requiring IV fluids were not statistically significantly different overall and for rotavirus-infected (n = 130) children. However, when non-rotavirus-infected children were considered (n = 63), the probiotic showed a statistically significant positive effect in reducing the total ORS solution intake and the stool output, and in increasing the number of patients with resolution of diarrhoea by the end of follow-up.

Evidence summary

An RCT conducted in Bangladesh [EL = 1+] evaluated the effectiveness of Lactobacillus paracasei strain ST11 in the management of acute diarrhoea in children. The study found no statistically significant differences between children treated with the probiotic and children that received placebo regarding the duration of diarrhoea, the total stool output, the total ORS solution intake, the number of children without diarrhoea by the end of the study or the number of children requiring IV fluids. However, they found the probiotic to have a clinical benefit when considering data from non-rotavirus-infected children only.

8.5.5. Lactobacillus rhamnosus strains 573L/1, 573L/2 and 573L/3

Evidence overview

One RCT was identified for inclusion.40 This trial, conducted in Poland, randomised a total of 93 children aged between 2 months and 6 years with acute diarrhoea into two treatment arms to assess the effectiveness of administering Lactobacillus rhamnosus strains 573L/1, 573L/2 and 573L/3 for 5 days compared with placebo. Patients with chronic diseases, immunosuppressive conditions or exclusively breastfed were excluded. Methods of randomisation and allocation concealment were adequate. Investigators and patients were blinded to the treatment. The outcomes considered were duration of diarrhoea, diarrhoea lasting more than 7 days, IV therapy and adverse events. [EL = 1+]

The baseline comparability of the two groups was adequate and more than 90% of the enrolled participants were included in the study analysis. The study found that the duration of diarrhoea was not statistically significantly different between the treatment (n = 46) and placebo (n = 41) groups. In rotavirus-infected patients (n = 39), children treated with the probiotic had a statistically significant shorter duration of diarrhoea (n = 22, mean 77.5 hours, SD 35.4 hours) than children that received placebo (n = 17, mean 115.0 hours, SD 66.9 hours) (WMD −37.5 hours; 95% CI −72.6 to −2.4 hours).

The difference in the proportions of cases of diarrhoea lasting more than 7 days in each group was not statistically significant. On admission, children were rehydrated per os or intravenously according to ESPGHAN recommendations. The authors reported that the mean duration of parenteral rehydration required was not statistically significantly different between the two treatment groups. The duration of parenteral rehydration required was statistically significantly shorter among children treated with the probiotic (n = 22, mean 14.9 hours, SD 13.7 hours) when compared with children receiving placebo (n = 17, mean 37.7 hours, SD 32.9 hours). This difference between the two groups was statistically significant (WMD −22.8 hours; 95% CI −39.5 to −6.2 hours). No adverse events were reported.

Evidence summary

An RCT located in Poland showed no statistically significant differences between the children receiving a probiotic preparation of Lactobacillus rhamnosus strains 573L/1, 573L/2 and 573L/3 and the children receiving placebo. When only rotavirus-infected children were considered, the trial showed a statistically significant clinical benefit of the probiotic in reducing the duration of diarrhoea and the duration of the IVT required.

8.5.6. Saccharomyces boulardii

One systematic review198 evaluated the therapeutic effects of Saccharomyces boulardii in the treatment of acute diarrhoea in children. The review included five RCTs involving a total of 619 participants. Of these RCTs, two were located in Pakistan, one in Mexico, one in Turkey and one in Argentina. The participants were children aged between 2 months and 12 years suffering from diarrhoea. The systematic review was well conducted. However, all the studies included had methodological limitations: only two trials reported an adequate method of randomisation, only one had adequate allocation concealment, two were not blinded and three did not apply an intention-to-treat analysis. Meta-analyses were performed for duration of diarrhoea and for number of stools on day 3 and on day 4. Other outcomes measured were resolution of diarrhoea on days 2 and 8, presence of diarrhoea at several time intervals, hospitalisation and vomiting. [EL = 1+]

The meta-analysis performed (four RCTs, n = 473) showed that children receiving the probiotic had a statistically significantly shorter duration of the diarrhoeic episode (WMD −1.08 days; 95% CI −1.3 to −0.83 days) (Figure 8.12).

Figure 8.12. Comparison of the effect of Saccharomyces boulardii versus placebo on the duration of diarrhoea.

Figure 8.12

Comparison of the effect of Saccharomyces boulardii versus placebo on the duration of diarrhoea.

Data from three RCTs (n = 331) were combined in a meta-analysis. This showed that children receiving the probiotic had statistically significantly fewer stools on day 3 when compared with the control group (WMD −1.3; 95% CI −1.9 to −0.6). The results of two RCTs (n = 218) were pooled and also showed a statistically significant reduction in the number of stools on day 4 in those children receiving the probiotic compared with placebo (WMD −1.1; 95% CI −1.6 to −0.6). One RCT (n = 130) measured the proportion of children without diarrhoea on day 2 and on day 8 from the start of the study. It found that statistically significantly more children in the intervention group experienced cessation of diarrhoea on both study days, when compared with the control group (RR 4; 95% CI 1.8 to 9.1, and RR 1.9; 95% CI 1.4 to 2.8, respectively). The proportion of children with diarrhoea on days 3 and 6 was reported in one study (n = 101), which found that the children in the intervention group were more likely to be diarrhoea-free by day 3 and by day 6 when compared with the children in the control group (RR 0.71; 95% CI 0.56 to 0.90, and RR 0.49; 95% CI 0.24 to 0.99, respectively). Another study (n = 88) found no statistically significant difference between the two groups for the presence of diarrhoea on day 4. The same study (n = 88) measured the proportion of children with diarrhoea on days 6 and 7, as well as the proportion of children with diarrhoea lasting more than a week. It found that the children in the intervention group were statistically significantly more likely to be diarrhoea-free on days 6 and 7 when compared with the control group (RR 0.49; 95% CI 0.24 to 0.99, and RR 0.39; 95% CI 0.20 to 0.75, respectively). In addition, the number of children with diarrhoea lasting more than a week was statistically significantly higher in the control group (RR 0.25; 95% CI 0.08 to 0.83). One RCT (n = 200) found that children who were treated with Saccharomyces boulardii had a statistically significantly shorter stay in hospital than those children who did not receive the probiotic (WMD −1 day; 95% CI −1.4 to −0.62 days) but that there was no statistically significant difference between the two groups for the duration of vomiting. Adverse events associated with the administration of Saccharomyces boulardii were not reported in any of the trials included in the review.

Evidence summary

There was evidence from a well-conducted systematic review [EL = 1+] for the clinical benefit of Saccharomyces boulardii in the management of acute diarrhoea in children. Meta-analysis performed showed that the administration of the probiotic shortened the duration of diarrhoea by 1 day and resulted in fewer stools on days 3 and 4, but the authors reported methodological limitations in the included trials.

8.5.7. Escherichia coli Nissle 1917

Evidence overview

One multicentre RCT199 was identified. The study was carried out in 11 paediatric outpatient centres across Ukraine, Germany and Russia. In total, 113 children aged between 2 and 47 months with non-bloody acute diarrhoea were randomised into two treatment arms (58 patients in the probiotic E. coli Nissle 1917 group and 55 in the placebo group) and followed up until treatment response or to a maximum of 10 days. The participants were allocated into the two groups based on random numbers. Allocation was such that participants and outcome assessors were unaware of the treatment given. The authors used an intention-to-treat analysis. The outcomes considered were duration of diarrhoea, proportion of patients without diarrhoea within the 10th day of follow-up and adverse events. [EL = 1+]

The baseline comparability between the two groups at the start of the study was adequate, and 12.3% of participants were lost during follow-up. The treatment response was defined as a reduction in stool frequency to three or fewer watery or loose stools in 24 hours over a period of at least two consecutive days. When comparing the median duration of diarrhoea between the two groups, the authors reported that E. coli Nissle 1917 statistically significantly shortened the duration of diarrhoea by 2.3 days (median duration of diarrhoea in the intervention group = 2.5 days, median duration of diarrhoea in the control group = 4.8 days). Diarrhoea stopped within 10 days for 52/55 children in the intervention group and for 39/58 children in the control group. This difference was statistically significant (RR 1.41; 95% CI 1.16 to 1.70). Two children in the intervention group (n = 55) suffered from rhinitis and abdominal pain while two children in the control group (n = 58) had acute otitis media.

Evidence summary

There was evidence form one multicentre RCT conducted in Russia, Ukraine and Germany for the effectiveness of E. coli Nissle 1917 in the treatment of acute diarrhoea in children. The study reported that those children receiving E. coli Nissle 1917 probiotic responded to treatment in a statistically significantly shorter time than those receiving placebo. Diarrhoea had stopped by day 10 in statistically significantly more children receiving E. coli Nissle 1917 probiotic compared with the placebo.

GDG translation from evidence to recommendation

There were many studies examining the efficacy of probiotic therapy in children with gastroenteritis. There was evidence from a high-quality systematic review suggesting that probiotic treatment had a beneficial effect – shortening the duration of diarrhoea and reducing the stool frequency. However, the available studies varied in quality, in the specific probiotics studied, in the treatment regimens used and in the outcomes examined. Therefore, despite some evidence of possible clinical benefit, the GDG did not consider it appropriate to recommend the use of a probiotic at this time. This was, however, considered to be an important field for further research. Licensed preparations of probiotics are not currently available in the NHS.

Research recommendation

Are probiotics effective and safe compared with a placebo in the treatment of children with gastroenteritis in the UK? Which specific probiotic is most effective and in what specific treatment regimen?

Why this is important

The available studies of probiotic therapy frequently report benefits, particularly in terms of reduced duration of diarrhoea or stool frequency. However, most of the published studies have methodological limitations. Moreover, there is great variation in the specific probiotics evaluated and in the treatment regimens used. Many of these studies were conducted in developing countries where the response to probiotic therapy may differ. Good-quality randomised controlled trials should be conducted in the UK to evaluate the effectiveness and safety of specific probiotics, using clearly defined treatment regimens and outcome measures.

Copyright © 2009, National Collaborating Centre for Women’s and Children’s Health.

No part of this publication may be reproduced, stored or transmitted in any form or by any means, without the prior written permission of the publisher or, in the case of reprographic reproduction, in accordance with the terms of licences issued by the Copyright Licensing Agency in the UK [www.cla.co.uk]. Enquiries concerning reproduction outside the terms stated here should be sent to the publisher at the UK address printed on this page.

The use of registered names, trademarks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant laws and regulations and therefore for general use.

Bookshelf ID: NBK63850
Contents

Views

  • PubReader
  • Print View
  • Cite this Page
  • PDF version of this title (3.1M)

In this Page

Other titles in this collection

Related NICE guidance and evidence

Recent Activity

ClearTurn OffTurn On

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...