Dr H. Szajewska, The 2nd Department of Pediatrics, The Medical University of Warsaw, 01-184 Warsaw, Dzialdowska 1, Poland. E-mail: email@example.com
Background Uncertainty exists regarding the use of zinc in the treatment of acute gastroenteritis in children living in Europe, where zinc deficiency is rare.
Aim To review evidence for the effectiveness of zinc in treating acute gastroenteritis in children, with special emphasis on data from developed countries.
Methods MEDLINE, EMBASE, and the Cochrane Library were searched through November 2007 for randomized controlled trials (RCTs) relevant to acute gastroenteritis in children younger than 5 years of age and zinc; additional references were obtained from the reviewed articles.
Results Eighteen RCTs (11 180 participants, mainly from developing countries) met the inclusion criteria. Use of zinc was associated with a significant reduction in diarrhoea duration (13 RCTs, 5643 infants, weighted mean difference −0.69 day, 95% CI −0.97 to −0.40) and the risk of diarrhoea lasting longer than 7 days [eight RCTs, n = 5769, relative risk (RR) 0.71, 95% CI 0.53–0.96]. No significant reduction in stool volume was observed for those receiving zinc compared with placebo (three RCTs, n = 606, standardized mean difference, −0.38, 95% CI −1.04 to 0.27). Combined data from five RCTs (n = 3156) showed that zinc significantly increased the chance of vomiting compared to the control agent (RR 1.2, 95% CI 1.05–1.4).
Conclusions These data confirm that zinc supplementation can be useful for treating acute gastroenteritis in children, particularly those from developing countries. However, the role of zinc supplements in treating children with acute gastroenteritis in developed countries needs further evaluation.
Acute gastroenteritis (AGE) remains one of the most common diseases in childhood. Despite an improvement in diagnostic and treatment approaches, morbidity and mortality due to acute diarrhoea are significant among children younger than 5 years of age, especially those in developing countries.1 The overall burden of AGE in developed countries has not been as well studied as that in developing countries. The actual prevalence is hard to estimate, as many patients do not seek medical advice from a physician; even if they do contact their physicians, it is not recorded.2
Most cases of AGE are self-limited and do not require antimicrobial therapy. The treatment consists of prevention of dehydration from occurring and correction of it, if it does occur. To achieve such treatment, oral rehydration solutions or intravenous fluids are used (depending on the degree of dehydration). This approach has been shown to reduce morbidity and mortality substantially. However, better therapy is still required to improve the efficacy of treatment. One promising option is the administration of zinc. Possible mechanisms for the beneficial antidiarrhoeal effect of zinc are discussed in more detail elsewhere.3 In brief, plausible mechanisms include improved absorption of water and electrolytes by the intestine; faster regeneration of gut epithelium; increased levels of enterocyte brush-border enzymes and an enhanced immune response.
Numerous clinical trials and two subsequent meta-analyses4, 5 have shown an antidiarrhoeal effect of zinc (i.e. a reduction in diarrhoea duration and severity) in children younger than 5 years of age. Also, a recently published meta-analysis on the prevention of childhood diarrhoea and respiratory illnesses indicates that zinc supplementation in healthy children leads to a significant reduction in the frequency of these illnesses.6 However, almost all of these studies were performed in developing countries, where malnutrition and zinc deficiency are common. Therefore, uncertainty exists regarding the use of zinc in the treatment of AGE in children living in developed countries, where zinc deficiency is rare.
This review was initiated as part of the development of the European guidelines for the management of AGE in children7 to update evidence for the effectiveness of zinc in treating AGE in children, with special emphasis on data from developed countries. While we were working on the review, a meta-analysis undertaken by Lukacik et al.5 was published. Although it addressed some of our initial objectives, there are important differences in our approach and some findings (e.g. effect on stool frequency and stool output). Accordingly, we thought it important to report the results of our systematic review and meta-analysis and discuss discrepancies.
Inclusion and exclusion criteria
Only randomized controlled trials (RCTs) that evaluated the effectiveness of zinc in the treatment of AGE in children were included. Studies regarding chronic diarrhoea, prevention of diarrhoea or combined therapy were excluded from analysis. All studies were performed in children up to 5 years of age with AGE. Patients received zinc at any dose and in any form or placebo/no intervention as an adjunct to the treatment of diarrhoea. The primary outcomes measures were the duration of diarrhoea (number of days) and stool output. The secondary outcome measures were stool frequency, diarrhoea lasting longer than 7 days and vomiting.
Studies appropriate for inclusion were identified by searching Medline (PubMed), Embase (Ovid) and The Cochrane Central Register of Controlled Trials (CENTRAL) through November 2007. The search strategy employed a combination of a validated filter for identifying controlled trials with topic-related keywords: ‘zinc’, ‘diarrhoea’, ‘diarrh*’, ‘gastrit*’, ‘gastroenteritis’, ‘vomit*’, ‘child*’, ‘infant*’, ‘toddler*’. The Cochrane Highly Sensitive Search Strategy filter was used to obtain RCTS.8 Additionally, reference lists from original studies and review articles were obtained.
Two independent reviewers (BP and DG) assessed the abstracts from the clinical trials that were identified according to the search strategy and, subsequently, the full texts of the studies that seemed relevant. Both reviewers independently carried out data extraction and entered the data into a computer program, The Cochrane Review Manager [(RevMan) Version 5.0.: The Nordic Cochrane Centre, The Cochrane Collaboration 2007, Copenhagen, Denmark], was used for statistical analysis and to perform the meta-analysis. The differences between the reviewers were resolved by discussion.
The reviewers independently, but without being blind to the authorship or journal, assessed the included trials for allocation concealment; blinding of the investigators, participants, outcome assessors and data analysts; intention-to-treat (ITT) analysis and comprehensive follow-up.8 Generation of allocation sequences was considered adequate if the resulting sequences were unpredictable (e.g. computer-generated random numbers, table of random numbers, drawing lots or envelopes and throwing dice). Conversely, it was considered inadequate if the resulting sequences were predictable (e.g. according to the case record number, date of birth, date of admission and alternation). Allocation concealment was considered adequate when the randomization method used did not allow the investigator or the participant to identify or influence the intervention group before enrolment of eligible participants in the study. However, the quality of the allocation concealment was considered unclear when randomization was used, but no or inadequate information about the method was available and when inappropriate methods of randomization (e.g. alternate medical record numbers, unsealed envelopes and open allocation schedules) were used.
With regard to the ITT analysis, an answer of ‘yes’ meant that the authors had specifically reported undertaking this type of analysis and/or that our own appraisal confirmed this finding. Conversely, a ‘no’ meant that the authors did not report the use of ITT analysis and/or that we could not confirm its use on study assessment. We found it more proper to call it available case analysis when data were analysed for every participant for whom the outcome was obtained. To evaluate the completeness of patient follow-up, we determined the percentage of participants excluded or lost to follow-up.
We used RevMan to perform statistical analysis and meta-analysis. The weighted mean difference (WMD) between the treatment and control groups was selected to represent the difference in continuous outcomes. Missing standard deviations were calculated by multiplying standard errors of means by the square root of the sample size: s.d. = S.E.M. × √N.8 In the case of the stool output outcome measure, we used the standardized mean difference (SMD) as it was measured in a variety of ways. For the dichotomous measure, the relative risk (RR) between the experimental and the control groups with 95% confidence intervals (CIs) was calculated. The weights given to each study are based on the inverse of the variance. For each total, the test for heterogeneity [chi-squared statistic with its degrees of freedom and P-value] and the statistic I2 measuring the extent of inconsistency among results was given. In these cases, when significant heterogeneity (P < 0.1, I2 > 50%) was observed, a random-effects model was used and the sensitivity analysis was conducted. Data from trials designed to compare more than two treatments (zinc at different doses)9 or having two experimental and two control groups10 were pooled and compared collectively with the control group. To investigate whether a review was subject to publication bias, we prepared a ‘funnel plot’ and examined this for signs of asymmetry.
Description of studies
Out of 43 clinical trials that were primarily identified as assessing zinc supplementation in patients with acute diarrhoea, 18 of them9–26 met the inclusion criteria (Table 1). The remaining 25 clinical trials were considered ineligible for inclusion for various reasons. We excluded one cluster-randomized trial27 to avoid ‘unit of analysis error’.8 We also excluded a study by Valery et al.28 because of inconsistency in the participants’ ages (up to 11 years of age). Additionally, studies regarding chronic diarrhoea, prevention of diarrhoea or combined therapy were excluded from analysis. Two included trials12, 21 did not assess any of the primary outcomes, but provided data about vomiting.
Table 1. Characteristics of included studies
A, adequate; AC, allocation concealment; ACA, available case analysis; B, blinding; BW, body weight; DB, double blind; FU, completeness to follow-up; ITT, intention-to-treat analysis; MA, meta-analysis; MD, mean difference; NA, not applicable; ND, not described; NCHS, National Center for Health Statistics; R, randomization; RCT, randomized controlled trial.
Moderately malnourished only (weight/age 61–75% of NCHS median) Excluded if severe malnutrition
N = 28 20 mg/day added to multivitamin syrup for 14 day
N = 28 Placebo
Eighteen studies that met the inclusion criteria recruited a total of 11 180 participants (6109 assigned to the experimental group and 5071 assigned to the control group) younger than 5 years of age with acute diarrhoea. In a majority of the studies, acute diarrhoea was defined as the passage of ≥3 watery stools in the previous 24 h or ≥1 loose stools containing blood. The pre-enrolment duration of diarrhoea was <7 days. Definitions of diarrhoea resolution were not homogeneous across the studies. In addition, clinical heterogeneity was found among the studies in the nutritional status of participants (in different studies, malnutrition was either an inclusion or exclusion criterion), aetiology of the diarrhoea and settings (in-patients,9, 9, 14, 16, 17, 21–23 community or out-patients13, 15, 18, 19, 25, 26). Most of the studies were conducted in countries with a medium Human Development Index (HDI);29 however, one was conducted in Brazil with a high HDI, and one was conducted in Ethiopia with a low HDI. Among countries with a medium HDI, all of them were Asian, except one Eurasian country (Turkey). HDI implies a determination of whether a country is developed, developing or underdeveloped. All developed countries are countries with a high HDI. This index is considered to be a standard means of measuring well-being, especially child welfare. Therefore, also AGE in countries with a medium HDI is characteristic of developing countries. Zinc was administered as different types of salts (sulphate,10, 12, 14–16, 19, 21, 24 gluconate,13, 25, 26 acetate9, 17, 20), in various forms (syrup,9, 10, 13–17, 20, 22, 23, 25, 26 or tablets9, 12, 18, 19, 21) and at dosages ranging from 5 to 45 mg daily. The duration of the supplementation was 5–15 days or depended on the diarrhoea duration. The methodological quality of the trials varied, as shown in Table 1.
Data regarding diarrhoea duration were available from 16 studies. Only 13 of them provided a measure of variance. A meta-analysis of 13 RCTs9, 13–17, 19, 20, 22, 24, 25 (5643 participants) showed a significantly lower average duration of diarrhoea for those treated with zinc compared with placebo (WMD −0.69, 95% CI −0.97 to −0.40; Figure 1). The included trials were significantly heterogeneous (χ2 = 55.22, P < 0.00001; I2 = 78%). To investigate whether a review was subject to publication bias, we prepared a ‘funnel plot’ (Figure 2). It was asymmetrical, lacking smaller studies with statistically significant effect of zinc supplementation. As it was to the disadvantage of intervention, it could not arouse suspicion of publication bias.
Data from two RCTs18, 23 were presented as median values and from one study,26 as the hazard ratio. Therefore, these data could not be pooled in our meta-analysis. A reduction in diarrhoea duration for children receiving zinc compared with controls was observed in two studies.23, 26 In one study, the difference between the experimental and control groups was not statistically significant.
Subgroup analysis. We performed a post hoc subgroup analysis for subsets of studies according to the nutritional status of the participants (Figure 3). The results of three RCTs10, 16, 22 suggest that the effect of zinc supplementation on the duration of diarrhoea was greater in malnourished patients than in groups without severe malnutrition (eight RCTs9, 9, 14, 15, 17, 19, 24, 25) or with no malnutrition at all (one RCT15). However, these results should be observed with caution because of the limitations of this type of analysis (as with any observational investigation).
Four studies provided data regarding the stool output.9, 14, 16, 22 On account of the use of different units, we used the SMD method to pool data from three RCTs (n = 606). The random-effects pooled estimate of the SMD for this outcome was −0.38 (95% CI −1.04 to 0.27), which indicated no significant reduction in stool volume for those receiving zinc compared with placebo. In one RTC22 that presented results for this outcome as median values, total diarrhoeal stool output (g/kg/day) was 28% less in the zinc-supplemented group than in the control group, but this difference was not statistically significant (P = 0.06).
Diarrhoea lasting longer than 7 days
A meta-analysis of eight RCTs10, 17, 18, 20, 22, 23, 25, 26 (n = 5769) showed a statistically significant reduction in the number of episodes of diarrhoea lasting longer than 7 days after enrolment in the zinc-supplemented group compared with the placebo group (RR 0.71, 95% CI 0.53–0.96), with an estimated risk difference (RD) of −0.04 (95% CI −0.08 to 0.00).
Pooled results from three trials9, 18, 24 (n = 1384) showed no significant reduction in stool frequency (number of stools per day) in the zinc-supplemented group compared with the placebo group (WMD −0.02, 95% CI −0.29 to 0.25).
Five RCTs9, 13, 14, 18, 25 (n = 3156) provided data regarding the number of patients who vomited during the study. More children vomited in the zinc-supplemented group than in the control group (RR 1.22, 95% CI 1.05–1.43), with an estimated RD of 0.04 (95% CI −0.03 to 0.10).
Results of meta-analyses of all other end points are presented in Table 2.
Table 2. Meta-analysis of tertiary outcomes
Comparison or outcome
Effect size (95% CI)
CI, confidence interval; RR, relative risk; WMD, weighted mean difference.
0.84 (0.67 to 1.05)
0.62 (0.44 to 0.87)
0.76 (0.64 to 0.90)
0.68 (0.11 to 4.31)
0.36 (0.07 to 1.85)
0.32 (0.01 to 7.75)
0.57 (0.28 to 1.17)
0.66 (0.24 to 1.85)
Diarrhoea on days 3–5
0.85 (0.73 to 0.99)
Total number of stools
−3.44 (−7.72 to 0.84)
0.70 (−0.19 to 1.59)
−0.52 (−0.74 to −0.31)
0.20 (−0.03 to 0.43)
−0.44 (−0.74 to −0.14)
Duration of watery diarrhoea
−0.90 (−1.42 to −0.38)
Total number of watery stools
−5.20 (−8.52 to −1.88)
Number of watery stools per day
−2.00 (−3.62 to −0.38)
0.86 (0.77 to 0.97)
1.58 (1.00 to 2.47)
1.28 (0.94 to 1.73)
1.96 (0.80 to 4.85)
Vomiting or regurgitation
4.89 (1.47 to 16.32)
2.45 (0.49 to 12.28)
Regurgitation on enrolment day
2.87 (2.15 to 3.83)
Vomiting episodes per days of diarrhoea
0.09 (−0.06 to 0.24)
Weight gain on recovery (% of admission weight)
0.70 (−0.84 to 2.24)
Number of days until mother perceived the episode had resolved
−1.10 (−1.66 to −0.54)
Taken to a physician
0.74 (0.54 to 1.01)
Our meta-analysis shows that zinc therapy in young children with acute infectious gastroenteritis has a moderate, beneficial effect on diarrhoea duration. Compared with patients given a placebo, patients who received zinc were less likely to have diarrhoea lasting longer than 7 days. We did not observe this beneficial effect of zinc on the severity of diarrhoea. We found no significant reduction in stool frequency and stool output in the zinc-supplemented group compared with the placebo group. Zinc therapy was correlated with a slightly increased number of children who vomited.
The results of our systematic review and meta-analysis are consistent with those from previous meta-analyses4, 5 when diarrhoea duration and vomiting are considered. However, in contrast to the meta-analysis by Lukacik et al.,5 we did not observe an advantageous effect of zinc therapy on diarrhoea severity (stool frequency and stool output). There are several possible mechanisms to explain this discrepancy. First, we did not include in the meta-analysis a cluster-randomized trial by Baqui et al.27 According to Cochrane Handbook, cluster-randomized trials cannot be directly combined with individually randomized trials. If the clustering is ignored and cluster trials are analysed as if individuals had been randomized, the study has overly narrow CI and receives more weight than is appropriate in a meta-analysis. Consequently, we have false-positive conclusions that the intervention had an effect. The incorporation of this study into the meta-analysis by Lukacik et al. is the main reason for the difference in the number of participants between meta-analyses (n = 15 231 Lukacik et al. vs. n = 11 180 this analysis). Secondly, we identified one additional study by Boran et al.15 (published in 2006) and two more RCTs published during a period not covered by the search strategy time limits (through 2006) used by Lukacik et al. Thirdly, our inclusion criteria were not equivalent: (i) we excluded every trial in which zinc was given with a co-intervention (also an RCT by Patel et al.30); (ii) we included only those studies with participants aged <5 years; (iii) although we assessed the methodological quality of the included trials, inadequate concealment of allocation was not an exclusion criterion for us. Other likely sources of observed differences are various key words and language limitations (lack of information). The above-mentioned distinctions pertain only to the part of the meta-analysis by Lukacik et al. that concerns acute diarrhoea.
Our analysis has several strengths. First, it included only randomized trials of which a large number were of high quality and had large sample sizes. Second, we only included trials assessing zinc therapy without co-intervention to avoid an overestimation of the antidiarrhoeal effect of zinc. Third, the ages of our participants were homogeneous. Additionally, in case of heterogeneity, we used a more robust random-effects model to pool data from the studies. We also prepared a ‘funnel plot’ and examined this for signs of asymmetry to investigate whether a review was subject to publication bias. Although we observed the signs of asymmetry, it was to the disadvantage of intervention and so could not arouse suspicion of publication bias. We are aware of some limitations of our analysis. One of them is the presence of significant heterogeneity in the results (diarrhoea duration and stool output) of various trials. Possible sources of its existence are clinical diversity (variability in the participants’ nutritional status; aetiology of the diarrhoea; dose, form and duration of the interventions) and variability in trial design and quality. However, we did not determine the causes of the observed heterogeneity in several subgroup analyses. A second important limitation of our meta-analysis is a lack of consistency in outcome measures. Therefore, we were not able to combine some results.
Our meta-analysis confirmed that zinc supplementation for the treatment of AGE in young children from developing countries results in a clinically important reduction in the duration of diarrhoea. Zinc deficiency, which is common in these populations, is associated with impaired water and electrolyte absorption, decreased brush-border enzymes and impaired cellular and humoral immunity. However, there is no proven benefit for its use in European children or generally in children with AGE who live in developed countries. In an open-label RCT15 conducted in well-nourished Turkish children, zinc therapy (15–30 mg/day) increased zinc levels, but it did not change either the duration or severity of the diarrhoea. UNICEF and WHO recommend zinc supplementation (10 mg for infants younger than 6 months of age and 20 mg for older infants and children for 10–14 days) as a universal treatment for all children with diarrhoea. In previous recommendations, The Centers for Disease Control and Prevention and the American Academy of Pediatrics31 stated that a number of trials have supported zinc supplementation as an effective agent in treating and preventing diarrhoeal disease; however, further research is needed to identify the mechanism of action of zinc and to determine its optimal delivery to the neediest populations. The role of zinc supplements in developed countries needs further evaluation. Finally, the European Society for Paediatric Gastroenterology, Hepatology and Nutrition and the European Society of Pediatric Infectious Diseases interpret the WHO recommendation as an endorsement to give zinc to children in developing countries and have formulated their recommendation accordingly.7 As zinc excess should be avoided and dosages in children without malnutrition have not been defined, further work is needed to establish whether zinc supplementation will also be of benefit to all children, malnourished and well nourished alike. In summary, although zinc administration appears to be a promising option for the treatment of children’s diarrhoea, there is no enough evidence to support its routine use in children from developed countries.
Declaration of personal interests: None. Declaration of funding interests: This study was funded in part by the Medical University of Warsaw and the Nutricia Research Foundation (research grant 5FNUT).