Specific Probiotics for the Treatment of Pediatric Acute Gastroenteritis in India: A Systematic Review and Meta-Analysis

Supplemental Digital Content is available in the text.


INTRODUCTION
Pediatric acute gastroenteritis (PAGE) is a leading cause of morbidity and mortality in children under 5 years old (1)(2)(3). Globally, 1.7 billion cases of PAGE occur each year, with 90% of cases occurring in developing countries (1,4). Deaths due to PAGE in children under 5 years were found to vary from country-to-country in a survey done in 2017: Africa (3%-13%), Asia (1%-9%), with the highest rates found in Syria (20%) (2). Factors influencing the incidence and severity of PAGE in different geographic areas include water sanitation, degree of malnourishment, diet, lifestyle factors, and type of diarrhea etiologies (3,5). It is difficult to account for all these factors when assessing efficacy of new treatments; thus, we focused our review on trials done in 1 country (India).
In India, improvements in water quality and use of oral rehydration therapy (ORT) has resulted in a 40% reduction of

What Is Known
• Pediatric acute gastroenteritis (PAGE) is a significant cause of morbidity and mortality globally, especially in children under 5 years old and is more severe in developing countries. • Most clinical trials studying the efficacy of probiotics for PAGE have been done in developed countries. • The choice of an appropriate probiotic is strain-specific, but it is not known if efficacy differs depending upon study population's country.

What Is New
• Only 3 types of probiotics had sufficient clinical trials (total 17 randomized controlled trials) in India to be included in a meta-analysis. • Two probiotics (Saccharomyces boulardii CNCM I- 745 and Lactobacillus rhamnosus GG) significantly reduced both the duration of diarrhea (1-2 d) and length of hospitalization (1-2 d) compared with controls in children living in India.

Translational Impact
• Two types of probiotics added to standard treatments for PAGE were found effective and safe in clinical trials done in India, but other probiotics required further trials.
PAGE-associated mortality since 2001, but PAGE continues to be a significant cause of morbidity, mortality, increased hospital admissions, and economic burden (4)(5)(6)(7)(8). Etiologies of PAGE in India may also differ from European countries. In Europe, the most common etiologies were found to be Campylobacter or Salmonella (9). In 1 study of outpatient children under 5 years in Odisha, India, the most common etiologies for PAGE were E. coli (30%), rotavirus (26%), or Shigella (24%), and concurrent infection with multiple etiologies was frequent (34%) (10). Shifts in the normally protective intestinal microbiome are found when diarrhea is present (11,12) and use of some strains of living probiotics has been found to be useful in restoring the normal microbiome (13,14). The choice of an appropriate probiotic is challenging due to the diverse types of probiotics available, inconsistent reports of efficacies and strength of available evidence (15). As the efficacy of probiotics has been demonstrated to be both diseasespecific and strain-specific, it is important to assess efficacy only within the identical type of probiotic strain or strains within multistrain mixtures for 1 type of disease (16,17). Many reviews and metaanalyses have either incorrectly pooled different strains of probiotics together (18) or limited their review to 1 strain of probiotic, but then pooled studies done in developed and developing countries (19)(20)(21)(22)(23). The pooled estimates of efficacy may be biased due to heterogeneity related to different probiotic strains or by different factors (microbiome profiles, socioeconomic factors, etc.) related to geographic area. The recent American Gastroenterology Association guidelines recommended against probiotics for PAGE in the United States and Canada, as most randomized controlled trials (RCTs) were done outside of these 2 countries, thus highlighting the importance of country-to-country variations (24).
This review focuses on randomized clinical trials only done in 1 country (India) in an effort to reduce these diverse sources of heterogeneity. Our aim is to determine which types of probiotics are safe and clinically effective for management of PAGE in the Indian population.

METHODS
Primary outcomes included differences in the mean duration of acute diarrhea from enrollment to resolution or the end of the study in subgroups of probiotics with the same strain or mixture of strains compared with the duration in the control groups. The other primary outcomes were the frequency of children with diarrhea resolution by day 3 or 5 ("cured") of treatment in subgroups of probiotics with the same strain or mixture of strains compared with children in the control group and the rapidity of response (the mean number of stools/day from day 1, to day 3, 4, or 5) for the probiotic and control groups. Secondary outcomes included differences in length of hospitalization and differences in the frequency of adverse events between the study groups.

Trial Characteristics
The study participant characteristics and safety data are described supplementary files (see Text, Supplemental Digital Content 2, Results, http://links.lww.com/PG9/A31 and in Table, Supplemental Digital Content 3, Study Population, http://links.lww.com/ PG9/A42). Most trials provided complete descriptions of the probiotic intervention (Table 1), but 10 trials did not provide the manufacturer or brand name (see Table, Supplemental Digital Content 4, Probiotic Description, http://links.lww.com/PG9/A43).
Of the 17 RCTs with ≥2 RCTs/probiotic type, 59% had an overall low risk of bias, while 41% had an overall high risk of bias ( Table 1). Most of the trials had a low risk of bias for randomization method (59%), attrition (88%), open/placebo controls (53%), and reporting post hoc outcomes (94%) (see Figure, Supplemental Digital Content 2, Risk of Bias, http://links.lww.com/PG9/A33). Ten (59%) of the 17 RCTs were double-blinded using placebos, while 7 (41%) used open controls (standard treatments only). There was no significant impact of blinding on the estimated treatment effects; thus, these types of controls were combined. For example, in trials with S. boulardii, the mean duration of diarrhea for unblinded (open) studies was 2.3 days for S. boulardii versus 3.6 days for open controls and comparable data were observed (2.7 d for S. boulardii versus 3.6 d for placebo controls), and results were similar for L. rhamnosus GG trials, while all B. clausii trials used open controls. Trials ranked as high risk typically did not describe the method of randomization, whether the study allocation was blinded, or did not report if the outcome was determined by blinded study staff. The risk of bias had an inconsistent impact on outcome measures by the type of probiotic: trials with S. boulardii showed the greatest reduction in duration of diarrhea in low bias risk trials (see Table Supplemental Digital Content 5, Subgroup Analysis, http://links.lww.com/PG9/A44), while the greatest reduction was seen in high-risk trials with L. rhamnosus GG and 3 of 4 trials with B. clausii mix were of high risk.

Duration of Diarrhea
Of the 17 RCTs (20 treatment arms), most (15, 88%) reported the mean duration of diarrhea as the primary outcome ( Table 2). As shown in Figure 1 Subgroup analyses found several factors influenced the effect of probiotics on the mean duration of diarrhea (see Table, Supplemental Digital Content 5, Subgroups, http://links.lww.com/PG9/ A44). In 4 RCTs, when S. boulardii CNCM I-745 was added to ORT and zinc, there was a significant reduction of the duration of diarrhea (SMD, -2.05 d; 95 CI, -3.4 to -0.75 d; P = 0.002). In 1 study, no zinc was given (50) and 4 RCTs did not report if zinc was given or not (43,45,47,49). Trials using S. boulardii with a low risk of bias showed a greater reduction in diarrhea duration, but the difference was not significant (Cochrane's Q = 2.1; P = 0.15). No other subgroups significantly influenced the efficacy of S. boulardii CNCM I-745.
Subgroup analysis for B. clausii trials was limited, as all trials were in inpatients and used open controls and all 4 trials gave ORT to the subjects. Zinc was also given in 3 trials but not reported in 1 trial (43). Subgroups were not significantly different: daily dose (Cochrane's Q = 0.63; P = 0.43) and low versus high risk of bias (Cochrane's Q = 0.63; P = 0.43).
Other subgroup analyses found factors that did not significantly impact outcome measures for PAGE for any of the 3 probiotics included: probiotic formulation, probiotic initiation times, use of ORT, rural versus urban settings, or funding sources (see Table, Supplemental Digital Content 5, Subgroups, http://links.lww.com/ PG9/A44).

Duration of Diarrhea in Rotavirus-Positive Children
Only 2 types of probiotics had sufficient trials to assess rotaviral diarrhea (see Table, Supplemental Digital Content 6, Rotaviral Diarrhea, http://links.lww.com/PG9/A45). Two RCTs using S. boulardii reported outcomes for rotavirus-positive children, but different outcome measures were used. Das et al (45) found S. boulardii significantly reduced diarrhea duration by -1.2 ± 0.1 days and Sirsat and Sankpal (49) found 25% more children were cured by day 3 if given S. boulardii compared with controls. L. rhamnosus GG reported mean duration of diarrhea in 3 trials of rotavirus-positive patients (52,53,57). The reduction in diarrheal days in rotavirus-positive children was not significant for L. rhamnosus GG (SMD, -3.11 d; 95% CI, -9.29 to 3.1 d; P = 0.32; I 2 = 98%). The 4 RCTs with B. clausii mixture did not report data by diarrheal etiology.

Rapidity of Response (Stool Frequency/Day Over Time)
The rapidity of a response was different by the type of probiotic, but not all trials provided data for this outcome (see Table, Supplemental Digital Content 7, Rapidity of Response, http://links.lww.com/PG9/ A46). In 3 RCTs with S. boulardii CNCM I-745 (43,46,48), the number of stools/day were equivalent at day 1 and day 3, but by day 4, the mean number of stools was significantly fewer for S. boulardii compared with controls (SMD, -0.61 stools/day; 95% CI, -1.06 to -0.17; P = 0.007), as shown in Figure

Length of Hospital Stay
Of the 12 RCTs that enrolled inpatient children, 9 trials provided data on the mean length of hospitalization stay (LOS) for the probiotic compared with the control group (see Table,  CI, -2.91 to 0.64; P = 0.21; I 2 = 98%). There was significant heterogeneity found among these studies (97%-98%), but no significant publication bias was noted (Egger's P = 0.16).

Safety
Of the 25 study arms, 11 (44%) did not collect any adverse reaction data ( Table 2), but 14 (56%) did collect and report adverse event data. Of the 14 with safety data, 10 (71%) reported no adverse events were observed during the study, while 4 (29%) reported at least 1 child with an adverse event, but the frequency was not significantly different for probiotic compared with control groups. None of the different probiotic types were associated with significant adverse events.

DISCUSSION
We conducted a systematic review and meta-analysis of 22 RCTs (with 4059 participants) to estimate the efficacy of probiotics available in India for the treatment of PAGE. Our study is a unique study focusing on clinical trials in 1 developing country (India) in an effort to limit geographic and nutritional factors that may influence rates of pediatric acute diarrhea. We included 17 RCTs in our meta-analysis of 3 different types of probiotics. We found differences in efficacy by type of probiotic: S. boulardii CNCM I-745 had the greatest reduction in mean days of diarrhea and reduction in LOS for hospitalized children and had the most rapid response by number of stools/day (reduced by day 4), whereas L. rhamnosus GG did not show a significant reduction in stool frequency until day 5. Trials using B. clausii did not show a significant difference by day 5. Most trials followed World Health Organization recommendations to use ORT and zinc in India or other countries where zinc deficiency is found (1), but only 41% used zinc as recommended. Zinc was not added to ORT in 24% of the RCTs and 35% of the RCTs failed to report if zinc was added or not.
Guidelines for which probiotics may be recommended for PAGE have been inconsistent. The 2020 American Gastroenterology Association guidelines did not recommend probiotics, citing most trials have been done in other countries besides the United States or Canada (24). In Europe, a recent update of the European Society of Paediatric Gastroenterology, Hepatology and Nutrition had weak recommendations for S. boulardii (based on 22 RCTs), L. rhamnosus GG (16 RCTs), and 2 other probiotics with L. reuteri DSM17938 (6 RCTs), but recommended against the use of B. clausii mix or a mix of L. helveticus R52 and L. rhamnosus R11 (65). We found 7 meta-analyses previously published on probiotics and treatment of PAGE (19-23, 65, 66) and all combined results from trials done in different countries. An updated Cochrane meta-analysis included 79 RCTs in children and adults with acute gastroenteritis in different countries and found a mean of 1-day reduction in diarrhea for each of the following: S. boulardii (11 RCTs), L. rhamnosus GG (14 RCTs), and L. reuteri DSM 17938 (6 RCTs), but did not present pediatric trials separately by country (66).
Two meta-analyses of S. boulardii trials found significant efficacy for PAGE (20,23) Two previous meta-analyses were done including trials only using L. rhamnosus GG and both found a significant mean reduction in duration of diarrhea: in 15 RCTs (SMD, -0.85 d; 95% CI, -1.15 to -0.56) (22) and in 19 RCTs (SMD, -1.0 d; 95% CI, -1.5 to -0.5) (67). Szajewska et al (22) reported L. rhamnosus GG was twice as effective in European countries compared with a mix of non-European countries (South East Asia, South America, United States, and Australia). A meta-analysis by Li et al (67) found doses of L. rhamnosus GG less than 10 10 CFU/d, delayed probiotic initiation (>3 d) and non-Asian/non-European countries resulted in a loss of efficacy for duration of diarrhea by L. rhamnosus GG. Li et al (67) also found L. rhamnosus GG was more effective in rotaviral diarrhea for shortening diarrheal duration (7 RCTs; SMD, -1.3 d; P < 0.001) and reduced LOS by 1.3 days. Our results agree with these 2 meta-analyses done in different countries. In our meta-analysis of 6 RCTs in India, L. rhamnosus GG appeared to be slightly more effective in reducing the duration of diarrhea (SMD, −1.75 d; P < 0.001) compared with the mixed group of different countries in the previous 2 meta-analyses. In our meta-analysis, L. rhamnosus GG also reduced the mean LOS in Indian inpatients (4 RCTs; SMD, -1.13 d; P < 0.001), but not as great as an extent as the previous 2 meta-analyses. In the Indian trials reported initiation times, L. rhamnosus GG was started within 4 days of the onset of diarrhea. Our meta-analysis also confirmed a sufficiently high dose (1 × 10 10 CFU/d) was needed for L. rhamnosus GG to be significantly effective. It is also interesting that the same strain was not found to be effective in another country. A large RCT involving 971 children with PAGE admitted to 10 pediatric emergency departments across the United States failed to find efficacy for L. rhamnosus GG (68).
A meta-analysis by Ianiro et al (21) included 6 RCTs using B. clausii probiotics and found a significant reduction in the duration of diarrhea (SMD, -0.4 d; 95% CI, -0.69 to -0.07; P = 0.02) and a reduction in LOS (3 RCTs; SMD, -0.85 d; 95% CI, -1.6 to -0.15; P = 0.03), but did not conduct any subgroup analyses. In a recent review of probiotics for the management of acute gastroenteritis in children, this Bacillus mixture was not recommended to treat PAGE (65). In our meta-analysis, the 4-strain mixture of B. clausii appears was effective for the reduction of diarrheal duration when trials are done in India (SMD, -1.4 d; P < 0.001), but the reduction in LOS was not significantly reduced by the B. clausii mix. Two reasons why our results may differ from the previous reviews is that we only included RCTs done in India and, second, we had rigorous inclusion criteria that excluded unpublished or duplicative trials. These excluded trials typically had nonsignificant efficacy findings. An advantage of using living probiotics is that they may possess one or more multiple mechanisms of action, which may explain why specific probiotics are effective against different etiologies of PAGE (69,70).
Strengths of our study included an exhaustive search of all RCTs in 1 country, including gray literature and meeting abstracts. The inclusion criteria were rigorous, in that only trials with welldescribed probiotic treatments (strains, daily dose, and formulations) were included and only those probiotics with at least 2 RCTs were included. Most trials used a standard definition of diarrhea (≥3 loosewatery stools/day) for the inclusion criteria. However, since there is no consensus for a standard outcome measure, the trials differed in the definition of diarrheal outcome (see Table, Supplemental Digital Content 4, Probiotic Description and Definitions, http://links.lww. com/PG9/A43) and types of outcomes.
Limitations were related to trials using different outcome measures, as we could only compare efficacy when common outcomes were used. In addition, probiotics with just a single RCTs could not be assessed and the proof of efficacy for some probiotic strains must wait until confirmatory trials are published. Factors that may be sources of heterogeneity were not reported in all trials. Reasons for heterogeneity may include differences in clinical baseline characteristics (such as age, degree of malnourishment, urban/rural settings, inpatient/outpatient, breast-fed versus formula-fed infants, etc.) or may due to differences in study design (degree of blinding, attrition, study quality, study size, adjunctive therapies given, etc.) Important confounder factors and influences were often not described in papers (diet, environmental factors, malnourishment status, etiologies of diarrhea, basic health status, mortality data, frequency of breastfeeding versus formula-fed, etc.). We were able to conduct subgroup analysis on most of these factors, but most did not significantly reduce heterogeneity. Daily doses less than 10 10 CFU/d were not effective for L. rhamnosus GG, while trials with S. boulardii all used 2 × 10 10 CFU/d and B. clausii used a lower dose (10 9 CFU/d). Trials differed in quality of reporting (method of randomization not reported, allocation blinding not reported, etc.). About half (48%) used placebo, but 52% of the trials were not double-blinded. However, subgroup analysis by study design bias did not find this influenced efficacy measures. This confirms the finding by Moustgaard et al (71) who reviewed 142 meta-analyses and found no significant effect of blinding or placebo use on estimated treatment effects. When trials were grouped by low versus high risk of bias, trials with low risk had better efficacy with S. boulardii (see Figure, Supplemental Digital Content 8, Risk of Bias, http://links.lww.com/PG9/A39), while the interpretation of high risk is difficult for L. rhamnosus GG, as there is only 1 trial with high risk. This holds true for B. clausii mix trials, as only one trial was of low risk. The type of funding was also assessed (23% by academic grant, 9% funded by Pharma companies, 23% were unfunded) but 45% did not report funding sources. The type of funding did not appear to significantly influence efficacy outcomes.

Generalizability and Future Studies
As the included trials were based upon study patients who lived in India, the results may not be extrapolated to probiotic efficacy in developed countries where the level of water sanitation and hygiene differ and the frequency of breast-feeding is typically lower. In addition, microbiome profiles and etiologies of diarrhea differ geographically and how this may influence probiotic efficacy is unknown. Implications for clinical practice and policy may include a more rational choice of the types of probiotics used for PAGE. Future research should include factors not reported in previous trials (described above) and a search for other clinical probiotic strains or multi-strain mixtures that might be effective for the treatment of PAGE.

CONCLUSIONS
In India, 2 types of probiotics (S. boulardii CNCM I-745 and L. rhamnosus GG) were well tolerated and significantly shortened both the duration of diarrhea and hospitalization stays in pediatric patients with PAGE. This is of important clinical significance. Earlier recovery and reduced hospital stays translate to less morbidity and lower healthcare costs for the system and the carers of these children.