Comparative effectiveness and safety of interventions for acute diarrhea and gastroenteritis in children: A systematic review and network meta-analysis

Abstract

Background: Many interventions have shown effectiveness in reducing the duration of acute diarrhea and gastroenteritis (ADG) in children. Yet, there is lack of comparative efficacy of interventions that seem to be better than placebo among which, the clinicians must choose. Our aim was to determine the comparative effectiveness and safety of the pharmacological and nutritional interventions for reducing the duration of ADG in children.

Methods: Data sources included Medline, Embase, CENTRAL, CINAHL, LILACS, and Global-Health up to May 2017. Eligible trials compared zinc (ZN), vitamin A, micronutrients (MN), probiotics, prebiotics, symbiotics, racecadotril, smectite(SM), loperamide, diluted milk, lactose-free formula(LCF), or their combinations, to placebo or standard treatment (STND), or among them. Two reviewers independently performed screening, review, study selection and extraction. The primary outcome was diarrhea duration. Secondary outcomes were stool frequency at day 2, diarrhea at day 3, vomiting and side effects. We performed a random effects Bayesian network meta-analysis to combine the direct and indirect evidence for each outcome. Mean differences and odds ratio with their credible intervals(CrI) were calculated. Coherence and transitivity assumptions were assessed. Meta-regression, subgroups and sensitivity analyses were conducted to explore the impact of effect modifiers. Summary under the cumulative curve (SUCRA) values with their CrI were calculated. We assessed the evidence quality and classified the best interventions using the Grading of Recommendations, Assessment, Development & Evaluation (GRADE) approach for each paired comparison.

Results: A total of 174 studies (32,430 children) proved eligible. Studies were conducted in 42 countries of which most were low-and middle-income countries (LMIC). Interventions were grouped in 27 categories. Most interventions were better than STND. Reduction of diarrhea varied from 12.5 to 51.1 hours. The combinations Saccharomyces boulardii (SB)+ZN, and SM+ZN were considered the best interventions (i.e., GRADE quality of evidence: moderate to high, substantial superiority to STND, reduction in duration of 35 to 40 hours, and large SUCRA values), while symbiotics (combination of probiotics+prebiotics), ZN, loperamide and combinations ZN+MN and ZN+LCF were considered inferior to the best and better than STND [Quality: moderate to high, superior to STND, and reduction of 17 to 25 hours]. In subgroups analyses, effect of ZN was higher in LMIC and was not present in high-income countries (HIC). Vitamin A, MN, prebiotics, kaolin-pectin, and diluted milk were similar to STND [Quality: moderate to high]. The remainder of the interventions had low to very-low evidence quality. Loperamide was the only intervention with more side effects than STND [Quality: moderate].

Discussion/conclusion: Most interventions analyzed (except vitamin A, micronutrients, prebiotics, and kaolin-pectin) showed evidence of superiority to placebo in reducing the diarrhea. With moderate-to high-quality of evidence, SB+ZN and SM+ZN, demonstrated the best combination of evidence quality and magnitude of effect while symbiotics, loperamide and zinc proved being the best single interventions, and loperamide was the most unsafe. Nonetheless, the effect of zinc, SB+ZN and SM+ZN might only be applied to children in LMIC. Results suggest no further role for studies comparing interventions against no treatment or placebo, or studies testing loperamide, MN, kaolin-pectin, vitamin A, prebiotics and diluted milk.

Prospero registration: CRD42015023778.

Fig 1. Article selection flowchart.

Fig 2. Network meta-analysis plots.

(A) Diarrhea duration, (B) Stool Frequency at day 2, (C) Diarrhea at day 3, (D) Vomiting and (E) Side effects. Nodes are proportional to the number of patients included in the corresponding treatments, and edges are weighted according to the number of studies included in the respective comparisons. Interventions' abbreviations: ALL-PRB: All probiotics, except LGG and SB; CAO: Kaolin-Pectin; DM: Diluted Milk; LCF: lactose-free formula; LCF+PRB: Lactose Free Formula + Probiotics; LGG: L. rhamnosus-GG; LGG+SM: LGG + Smectite; LOP: Loperamide; MN: Micronutrients; PRE: prebiotics; RC: Racecadotril; SB: Saccharomyces boulardii; SB+LCF: S. boulardii + Lactose-Free Formula; SB+ZN: S. boulardii + Zinc; SB+ZN+LCF: S. boulardii + Zinc + Lactose free Formula; SM: Smectite; SM+ZN: Smectite + Zinc; STND: Standard treatment or Placebo; SYM: Symbiotics; SYM+LCF: Symbiotics + Lactose-Free Formula; VA: Vitamin A; YOG: Yogurt; YOG+PRB+ZN: Yogurt + Probiotics + Zinc; ZN: Zinc; ZN+LCF: Zinc + Lactose Free Formula; ZN+MN: Zinc + micronutrients; ZN+PRB: Zinc + Probiotics.

Fig 3. League table with NMA estimates for diarrhea duration and stool frequency at day 2.

Comparisons should be read from left to right. The effectiveness estimate is located at the intersection of the column- defining treatment and the row-defining treatment. Diarrhea duration (bottom part of the table) effect estimate is presented in mean difference in hours with the 95% CrI, an MD below 0 favors the column- defining treatment (fewer hours of diarrhea). Stool frequency at day 2 (upper part of the table) effects estimate is presented in mean difference in the number of stools with the 95% CrI, an MD below 0 favours the column-defining treatment (fewer stools per day). To obtain MDs for comparisons in the opposing direction, negative values should be converted into positive values and vice versa. Significant results are in bold and underlined. Cells filling represent the GRADE quality of evidence assessment: Green: High quality; Light Green: Moderate quality; Light Orange: Low; Darker Orange: Very Low. Blank cells: not available interventions and comparisons MD: Mean difference; 95CrI%: 95% Credible Intervals; NMA: Network Meta- analysis. Interventions' abbreviations are described in Table 1.

Fig 4. Forest plot of subgroup analyses by country income classification.

All interventions are compared to STND. Subgroup analyses were performed based on the country income classification (based on the World Bank classification at 2016) Comparisons are highlighted in Dark Grey if the GRADE Quality was Low or Very Low, and in Light grey, when Quality was High or Moderate. When an intervention has no effect estimate in HIC means that there were no studies in HIC for the mentioned interventions. MD: Mean difference; 95CrI%: 95% Credible Intervals; LMIC: Low- and Middle-Income countries; HIC: High-income countries. Interventions' abbreviations are described in Table 1.

Fig 5. League table with NMA estimates for diarrhea at day 3 and vomiting.

Comparisons should be read from left to right. The effectiveness estimate is located at the intersection of the column-defining treatment and the row-defining treatment. Diarrhea at day 3 (bottom part of the table) effect estimates are presented in odds ratio (OR) with the 95% CrI. A OR below 1.0 favours the column- defining treatment (less presence of diarrhea at day 3). Vomiting (upper part of the table) effect estimates are presented in Odds ratio (OR) with the 95% CrI, an OR below 1.0 favors the column-defining treatment (less presence of vomiting). To obtain ORs for comparisons in the opposing direction, reciprocals should be taken. Significant results are in bold and underlined font. Cells filling represent the GRADE quality of evidence assessment: Green: High quality; Light Green: Moderate quality; Light Orange: Low; Darker Orange: Very Low. Blank cells: not available interventions and comparisons. OR: Odds Ratio; 95CrI%: 95% Credible Intervals; NMA: Network Meta-analysis. Interventions' abbreviations are described in Table 1.

Fig 6. League table with NMA estimates for side effects.

Comparisons should be read from left to right. The effectiveness estimate is located at the intersection of the column-defining treatment and the row-defining treatment. Side effects estimates are presented in odds ratio (OR) with the 95% CrI, an OR below 1.0 favors the row-defining treatment (less presence of side effects). To obtain ORs for comparisons in the opposing direction, reciprocals should be taken. Significant results are in bold and underlined. Cells filling represent the GRADE quality of evidence assessment: Green: High quality; Light Green: Moderate quality; Light Orange: Low; Darker Orange: Very Low. OR: Odds Ratio; 95CrI%: 95% Credible Intervals. Interventions' abbreviations are described in Table 1.