Ya-Ting Li, Hong Xu, Jian-Zhong Ye, Wen-Rui Wu, Ding Shi, Dai-Qiong Fang, Yang Liu, Lan-Juan Li

Abstract

Key words:Lactobacillus rhamnosus GG; Acute diarrhea; Children; Rotavirus; Probiotics;Systematic review; Meta-analysis

INTRODUCTION

The World Health Organization and United Nations International Children's Emergency Fund define diarrhea as more than three loose or watery stools during a 24-h period. A duration of 14 days is the proposed criterion for acute diarrhea or persistent diarrhea. Diarrhea is a major infectious cause of childhood morbidity and mortality worldwide, especially in developing countries[1]. As the second most common cause of death among children under 5 years of age[2], the frequency of acute diarrhea in one year is approximately two to three episodes per child[1]. Previous data showed that the incidence of diarrhea was 6 to 12 episodes in 12 months per child in developing countries[3].

The goals of treatment are prevention or resolution of dehydration and reduction of the diarrhea duration and infectious period[4]. Oral rehydration, gut motility inhibitors, and antibiotics are used to treat acute gastroenteritis[4]. Oral rehydration contributes to a reduced likelihood of dehydration but has no appreciable effects on bowel movements or the duration of diarrhea and is not utilized to its full extent[5].Antibiotics should be considered if pathogenic bacteria are detected. Smectite and zinc remain under-utilized as adjuvant therapies[6,7].

Probiotic supplements have gained considerable popularity in the global market and are predicted to generate 64 billion United States dollars in revenue by 2023[8].Probiotics have health benefits for hosts[9]and have been evaluated in the treatment of diarrhea, and multiple mechanisms of diarrhea improvement have been identified.Probiotics modulate the host immune response[10]. Furthermore, colonic bacterial metabolites such as short-chain fatty acids increase colonic Na and fluid absorption through a cyclic adenosine monophosphate-independent mechanism[5]. In clinical trials, the well-known probiotics Saccharomyces boulardii, Lactobacillus reuteri DSM 17938, and Lactobacillus rhamnosus GG ATCC 53013 (LGG) have been used to treat diarrhea[2,4]. Previously, rotavirus-induced diarrhea was considered an adaptation disease associated with LGG treatment[11]. Wolvers D revealed that the probiotic dose mediated the effectiveness of treatment, and 1010-1011CFU per day was recommended[12]. In addition, a greater effect was observed in the early stage of illness, and a poorer effect on invasive bacterial diarrhea versus watery diarrhea was observed. LGG treatment has been endorsed by leading experts[13-15]. However, most recent randomized controlled trials (RCTs) conducted by Schnadower et al[8]yielded no evidence of a beneficial effect of LGG treatment. Therefore, we conducted a metaanalysis to evaluate the available validated data and update existing knowledge and thus provide guidance to patients.

MATERIALS AND METHODS

Literature search

Relevant studies published before April 2019 were retrieved from the EMBASE,MEDLINE, PubMed, Web of Science databases, and the Cochrane Central Register of Controlled Trials (CENTRAL, the Cochrane Library). The search strategy was conducted with medical subject headings and the search terms “diarrhoea, diarrhea,diarrh＊, gastroenteritis, probiotic＊, Lactobacillus rhamnosus GG, Lactobacillus GG, and LGG”. No language restrictions were applied. Additional studies were identified by manually searching review articles.

Study selection

Nineteen RCTs describing LGG interventions for acute diarrhea were included. The PRISMA statement and the guidelines from the Cochrane Collaboration were followed for this evidence-based medicine study[16,17]. The participants were children aged less than 18 years. The dose of LGG was provided in various forms at different times. Antibiotic-associated diarrhea and persistent diarrhea were excluded. Other applications of LGG, such as preventive strategies, were not included. Some particular article types without complete data were excluded, such as abstracts and letters. We also excluded studies using mixtures of more than one probiotic strain. The primary outcomes were directly related to the development of persistent diarrhea, including the duration of diarrhea and diarrhea lasting ≥ 3 and ≥ 4 d. Secondary outcomes included the hospital stay duration, stool frequency, and improvement in stool consistency and vomiting.

Data extraction

Two investigators (Li YT and Xu H) independently identified eligible articles and extracted applicable data following the inclusion criteria. Quality control was assessed by another reviewer (Wu WR). The data set included the baseline characteristics of the participants, the duration of diarrhea, the hospital stay duration, the time to improvement in stool consistency, the mean number of stools per day during diarrhea episodes, the proportion of patients with vomiting, the duration of vomiting, stool frequency on days 2 and 3 after treatment, and the number of patients with diarrhea lasting ≥ 3 or 4 d. Cochrane Review Manager (Version 5.1. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2011) and STATA version 12.0(StataCorp LP, College Station, TX, United States) were used for data analyses. Any discrepancies were resolved by discussion.

Risk of bias

All included trials were evaluated following the Cochrane Collaboration's risk of bias tool. Seven domains were examined to identify the bias risk: selection bias, including random sequence generation and allocation concealment, performance bias, including blinding of participants and personnel, detection bias, including blinding of outcome assessments, attrition bias, including incomplete outcome data, reporting bias,including selective reporting, and other bias. Adequate allocation concealment was implemented to ensure blinding of the participants and investigators to avoid influences on the measures. Randomization was performed based on confirmed allocation concealment. Unclear allocation concealment was noted when no method was mentioned. The integrity of the data was evaluated, including the proportion of excluded participants (http://www.cochrane-handbook.org).

Statistical analysis

The Cochrane Review Manager was used to analyze the relevant data. The mean differences (MDs) in continuous data under LGG or placebo treatment were measured. Dichotomous results are pooled and presented as risk ratios. Additionally,95% confidence intervals (CIs) are reported for all types of outcomes. I2and χ2values were calculated to quantify and reflect heterogeneity. A P-value ＜ 0.05 indicates that heterogeneity should not be ignored; thus, a random-effects model was used. A fixedeffects model was employed when no statistically significant inconsistency was detected. Publication bias was assessed by funnel plot asymmetry[18]. Sensitivity analyses were conducted to detect the robustness of results by assessing randomization, missing data, blinding, and allocation concealment. Each individual study was systematically removed from the meta-analysis, and the effect was recalculated and estimated from the remaining studies (Supporting information Figure S1). Regression analysis was conducted, and the relationships between the duration of diarrhea and other covariates, including publication year, participant age,the duration of diarrhea before study enrollment, and the LGG dosage, were examined. Subgroup analyses were performed to diminish significant inconsistency.Preplanned subgroup analyses were performed according to the following clinical characteristics and results from sensitivity or regression analysis: (1) The dosage of LGG per day. A dosage of 1010CFU/day was observed to be a critical element of effective treatment in the study by Szajewska et al[13]. In addition, a larger dose was suggested in other studies[19,20]; (2) The etiology of diarrhea. Diarrhea mortality and severe diarrhea were most frequently caused by rotavirus in children[21]. Compared to control children, several rotavirus-positive children with watery stools in a probiotic group were reported to exhibit a marked reduction in diarrhea symptoms after 24 h[22].A meta-analysis performed by Szajewska et al[23]in 2007 concluded that the duration of rotavirus-induced diarrhea was significantly attenuated by LGG supplementation;(3) The site of treatment (inpatient vs outpatient); (4) Vaccination status; (5)Geography of the clinical trials. The location of the study affected the sanitary habits,exposure to various pathogens, and nutrient status of the participants. All studied environmental factors contribute to various outcomes; (6) Early probiotic administration. A beneficial effect of probiotics was reported in the course of disease when initiated early[12]; and (7) Publication date.

RESULTS

Study selection

A total of 349 potentially relevant studies were identified. The process of screening was carried out according to the flow diagram shown in Figure S2 (Supporting information). The characteristics of each included study are summarized in Table 1.With 988 participants in a 2007 meta-analysis and 2683 participants in a 2013 metaanalysis, a total of 4073 participants in 19 RCTs were identified in the literature. Two experimental arms in the study of Basu et al[24]were listed separately to exhibit different doses of probiotics, which were marked as Basu 2009a and Basu 2009b.Therefore, the figures, tables, and full texts of 18 articles were reviewed[8,24-40]. A large number of trials were conducted in Europe and Asia. Patients were recruited from outpatient, inpatient, and emergency departments. Inconsistency existed in the daily doses and routes of LGG supplementation during the treatment period. Different criteria were used to define diarrhea in the included studies. Diarrhea resolution was commonly defined as passage of the first normal stool or the last watery stool.

Antibiotic treatment before recruitment was assessed, and different studies varied regarding the use of antibiotics. Similarly, the duration of treatment varied. Studies of moderate to high quality were adequately assessed and are summarized in Figure S3(Supporting information).

Evaluation before enrollment (days)

Before enrollment, age was assessed in 16 studies, and the duration of diarrhea was reported in nine studies (Supporting information Figures S4 and S5). No obvious difference in age was found. The statistical differences and high heterogeneity resulting from the duration of diarrhea [MD -6.21 h, 95%CI (-9.04, -3.38)] could be reduced by subgrouping according to the outcomes of the sensitivity analysis(Supporting information Figure S1). The subgroup excluding the study of Ritchie et al[37]performed in 2010 showed acceptable heterogeneity, and no statistical significance was observed for the duration of diarrhea before study enrollment [MD -0.9 h, 95%CI (-4.02, 2.22)] (I2= 10%). Sensitivity analysis revealed differences in the duration of diarrhea before study enrollment between the two groups in the study of Ritchie et al[37], which recruited aboriginal children in the Northern Territory of Australia. Social disadvantages and poverty contributed to malnutrition in these children[4]. However, no significant differences in the primary and secondary outcomes were found by sensitivity analysis, which is inconsistent with the findings reported in previous meta-analyses[4,13](Supporting information Figure S1).

Duration of diarrhea

ygoloitE fo noitaruD noitnevretni;dedulcxe aehrraid lairetcaB d 7 aehrraid decudni-surivatoR%8.57;dedulcxe aehrraid lairetcaB litnu ro d 7%1.75 aehrraid surivatoR aehrraid deppots;dedulcxe aehrraid lairetcaB litnu ro d 7 aehrraid decudni-surivatoR aehrraid%60.65 deppots wef a ylno ni erutluc lootS d 5 atad on dna stnapicitrap detneserp aehrraid decudni-surivatoR raelcnU aehrraid ydoolB ;%05 dedulcxe;dedulcxe aehrraid ydoolB raelcnU aehrraid decudni-surivatoR%85 lortnoC puorg FRO FRO FRO sliated oN nevig 023 nilunI yad/gm raelcnU GGL airetirc noisulcxE)egasod(6;01 × 021 naht rehto ssenlli cimetsyS yad/UFC cimetsys ;noissimda no aehrraid aehrraid fo snoitacilpmoc eruliaf ;noitazilatipsoh gnirud tnesnoc demrofni edivorp ot 01;rehto ssenlli fo smotpmyS 01 × 2 yad/UFC fo tnempoleved ;aehrraid naht fo noitacilpmoc cimetsys yna;noitazilatipsoh gnirud aehrraid demrofni edivorp ot eruliaf tnesnoc 21;rehto ssenlli fo smotpmyS 01 × 2 yad/UFC fo tnempoleved ;aehrraid naht fo noitacilpmoc cimetsys yna;noitazilatipsoh gnirud aehrraid demrofni edivorp ot eruliaf tnesnoc 9;01 × 21 ereves ;noitirtunlaM yad/UFC gnitsixeoc ;noitardyhed;ssenlli cimetsys etuca gniylrednu ;ycneicifedonummi citsyc ;esaesid cinorhc ereves rehto ro ygrella doof ;sisorbif fo esu ;sesaesid IG cinorhc 3 suoiverp eht ni scitoiborp rehto yna ro scitoibitna ;kw eht ni noitacidem laehrraiditna ecnailpmoc roop ;kw 3 suoiverp 01/UFC ;01 gniriuqer snoitcefni cimetsyS yad noitirtunlam ereves ;scitoibitna SHCN fo %56 ＜ ega rof thgiew(aehrraid ydoolb ;sdradnats/gk/lm 05 esaesid gnitsixeoc ;sloots ydoolB yad fo esruoc eht ecneulfni yam taht aehrraid airetirc noisulcnI elbisiv tuohtiw yad/sloots yretaw 3 ≥/sllec doolb etihw 01＜ ;sucum ro doolb sucum ,sllec der on dna dleif rewop-hgih;ypocsorcim loots no airetcab ro ,sekalf;noitaraperp pord gnignah evitagen erutluc loots lairetcab evitagen tuohtiw yad/sloots yretaw 3 ≥etihw 01＜ ,sucum ro doolb cipocsorcam on dna ,dleif rewop-hgih/sllec doolb airetcab ro ,sekalf sucum ,sllec doolb der gnignah evitagen ;ypocsorcim loots no loots lairetcab evitagen ;noitaraperp pord erutluc tuohtiw yad/sloots yretaw 3 ≥etihw 01＜ ,sucum ro doolb cipocsorcam on dna ,dleif rewop-hgih/sllec doolb airetcab ro ,sekalf sucum ,sllec doolb der gnignah evitagen ;ypocsorcim loots no loots lairetcab evitagen ;noitaraperp pord erutluc h 84＜ rof yad/sloots diuqil ro esool 2 ＞2 esool ro yretaw erom ro 3( aehrraid etucA h-42 eno tsael ta gnirud h 42 rep sloots htiw )noissimda erofeb h 27 eht ni doirep retfa noitardyhed ereves ro etaredom sdiulf VI dipar htiw noitcerroc suoitcefni etuca htiw nerdlihc dna stnafnI noitardyher laro deliaf dna aehrraid/pxe( n tneitaP)lortnoc ecruos 323/323 stneitapnI 581/881 stneitapnI 581/681 stneitapnI 29/001 stneitaptuO 36/16 stneitapnI 05/05 stneitapnI slairt dedulcni eht fo scitsiretcarahC 1 elbaT yrtnuoC puorg egA elcitra fo epyT elcitrA]52[aidnInerdlihC;retnec 1 ;TCR,la te usaB ry 1 :noitaruD 7002]42[aidnInerdlihC;retnec 1 ;TCR,la te usaB ry 1 :noitaruD a9002]42[aidnInerdlihC;retnec 1 ;TCRla te usaB ry 1 :noitaruD b9002 ylatI om 63-3;sretnec 6 ;TCR te inanaC]62[om 21 :noitaruD7002 ,la]72[lizarB42-1 ,syoB retnec 1 ;TCRla te atsoC om 3002 dnaloP raelcnU retnec 1 ;TCR-aknoiwrezC te aksralfazS]82[9002 ,la

IIG ro IG surivoroN;%7.71 surivatoR ;%6.91;%1.9 surivonedA;%4.7 eliciffid muidirtsolC%0.5 allegihS airetcaB ;%53 surivatoR oN ;%5.4 setisaraP ;%42 ydoolB ;%5.43 snegohtap%7.8 aehrraid decudni-surivatoR%16 aehrraid decudni-surivatoR%001 aehrraid;%86 snegohtap lairetcaB setisarap ;%0.04 surivatoR snegohtap oN ;%6 puorg citoiborp :deifitnedi%52 ydoolB ;%6.52 surivatoR etih W ;dedulcxe aehrraid;%3.41 sloots ni sllec doolb%7.4 aehrraid lairetcaB raelcnU;%3.33 sloots ydoolB d 5 detarelot sA,h 6-4 rof da neht mutibil d 5 ≤d 5 raelcnU raelcnU d 5 d 2 gnihctaM obecalp FRO FRO citoiborp oN on htiw SRO GGL enillatsyrC orcim esolullec nilunI obecalP 01 UFC ;01 × 1 ro ,siseme suoilib ,sititaercnaP tuohtiw ro htiw ,yad rep sloots yretaw 3 ≥ 884/384 yliad eciwt ygrella nwonk a ;aizehcotameh d 7 naht rewef rof ,gnitimov ot ro GG susonmahr .L ot a ro esolullec enillatsyrcorcim,nicymorhtyre ot ygrella nwonk matcal-ateb dna ,nicymadnilc did revigerac ;stnega citoibitna;hsinapS ro hsilgnE kaeps ton scitoibitna gniviecer nerdlihc 9;01 × 01 ≥;egasu citoiborp suoiverP ro diuqil 4 ＞ htiw nerdlihc dna stnafnI 041/741 052/UFC gniylrednu detaertnu htiw yad/Lm;esaesid lewob llams cinorhc FRO;esaesid lewob yrotammalfni esaesid cinorhc gniylrednu yna esaesid evisserppusonummi ro tnemtaert ro 9/UFC ;01 × 6 tsal eht ni tnemtaert citoibitnA d 5 ot 1 rof yad/sloots diuqilimes 84/25 FRO htiw yad:thgiew a ;gnideeftsaerb ,kw 3 yretaw 3 ≥ htiw nerdlihc dna stnafnI h 84 ＜ rof yad/sloots elitnecrep ht5 eht ＜ oitar thgieh 01;yretaw 3 ＞ htiw nerdlihc dna stnafnI 12/12 01 × 2 detats toN yad/UFC evitisop sloots dna d 7 ＜ rof yad/sloots fo noitardyhed egareva ;surivator rof spuorg htob ni %5 yletamixorppa ro diuqil 1 ro h 21 ni sloots yretaw 3 ＞25/54 GGL + SROcitoiborp ro citoibitnA 01 yad /UFC 01 cinorhc ;d 5 tsal eht ni esu;doolb ro ,sup ,sucum htiw loots diuqilimes egral ro llams eht fo sesaesid d 5 ＜;noisserppusonummi ;enitsetni airunoteklynehp 9-6/UFC 01 × 1;tnesnoc desufer stneraP eht gnimussa ro yretaw( yad rep sloots 3 ＞ 501/501 yad eht edistuo gnivil nerdlihc)reniatnoc eht fo epahs ydoolb ;aera lapicinum;noitardyhed ereves ;aehrraid dna ekat ot ytilibani ,kcohs detcepsus ;sdoof laro niater noitcefni cimetsys redwop GGL lariv-non rof srotcaf ksiR h 42 tsal eht ni sloots esool 2 naht eroM 87/77 yliad eciwt aehrraid degnolorp( aehrraid ssorg ,d 7 naht erom gnitsal ro ,erusopxe citoibitna ,doolb;)esaesid lewob yrotammalfni ksir ;esimorpmoc enummi detaicossa-citoiborp rof srotcaf ot ygrella na ro ssenlli cimetsys stcudorp klim 01-9 UFC 01;gnideeftsaerb evisulcxE sloots yretaw 3 ＞ htiw nerdlihc dna stnafnI 91/02 yliad eciwt aimecitpes d 41 ＜ rof aehrraid dna h 42 tsal ni-ytisrevinU detinU detailiffa setatS DEP stneitapnI sa detsiL dna swollof stneitaptuo stneitaptuO ylatI stneitapnI dnalniF stneitapnI acirfA dna stneitaptuo stneitapnI tpygE eporuE aciremA aidnI DEP detinU setatS stneitapnI dnaliahT om 84-3 om 63-1 om 63-3 om 63 ≤om 63-1 om 63 ≤om 27-6 om 42-1 TCR rewodanhcS]8[8102 la te;TCR te iniladnauG]92[;retnecitlum 0002 ,la ry 1 :noitaruD retnec 1 ;TCR te onirauG]03[om 3 :noitaruD7991 ,la retnec 1 ;TCR te irualosI]13[ton :noitaruD4991 ,la detats;sretnec 21 ;TCR te iksnisaJ]23[ton :noitaruD2002 ,la detats]33[;retnec 1 ;TCR,la te arsiM ton :noitaruD 9002 detats]43[TCR,la te noxiN 2102]53[;retnec 1 ;TCR,la te tnaP kw 6 :noitaruD 6991

21-11]63[;%9.71 surivatoR d 2 obecalP 01 × 2 aimecitpes ;noitirtunlam ereveS htiw nerdlihc dna stnafni dehsiruonrednU 91/12 stneitapnI natsikaP om 42-1 retnec 1 ;TCR,la te azaR%5.2 surivortsA yad/UFC d 41 ＜ rof h 42 tsal eht ni sloots yretaw 3 ＞om 2 :noitaruD 5991 aehrraid ydoolB noitardyhed etaredom tsael ta dna 9 snegohtap lairetcaB d 3 lacitnedI 01 × 51 ＞eht gnirud deriuqer negyxO aehrraid etuca htiw nerdlihc lanigirobA 13/33 raelcnU ailartsuA om 42-4;retnec 1 ;TCR te eihctiR]73[;%5.8 surivatoR ;%5.21 obecalp yad/UFC,caidrac cinorhc ;doirep yduts h 42 gnirud sloots esool 3 ≥ sa denifed om 12 :noitaruD 0102 ,la%6 setisaraP;esaesid yrotaripser ro ,laner ot elba dna d 7 ＜ rof noitatneserp erofeb lanitsetniortsag suoiverp FRO etarelot esorcus nevorp ;yregrus no detcepsus ;ecnarelotni;ycneicifedonummi nwonk;tnemllorne erofeb esu citoiborp ega fo om 4 naht regnuoy 01 lairetcaB ;%4.72 surivatoRd 5 obecalP yad/UFC01 detats toN dna h 42 tsal eht ni loots yretaw 1 ≥46/95 stneitapnI aissuR om 63-1;retnec 1 ;TCR te avokinrohS]83[%12 aehrraid d 5 ＜ rof aehrraid ry 1 :noitaruD 7991 ,la 01;%4.25 surivatoR kw 4 fo gm 071 dna UFC01 ecneserp eht( snoitcefnioC yretaw esool 3 ≥ sa denifed saw aehrraiD 95/56 raelcnU aidnI om 06-6 TCR te uhdniS fo gm 071]93[seiceps muidiropsotpyrC esolullec dna surivator htob fo doirep h-42 a nihtiw sloots 4102 ,la enillatsyrcorcim%6.74 ereves ;)muidiropsotpyrC esolullec yad/citoiborp ;noitirtunlam gnidecerp eht ni noitpmusnoc;scitoiborp ot ygrella ;htnom sitiloc ro nemodba etuca 01]04[%1.42 surivatoR d 5 cniz dna SRO UFC 01 × 1;yretnesyd ;noitirtunlam ereveS eht ni sloots esool erom ro eerht fo egassaP 001/001 ro DPO aidnI om 06-6;lebal-nepO la te ynnuS d/gm 02 yad rep gnitsixeoc fo ecnedive lacinilc h 42 tsal DEP TCR 4102 lacinilc ;sessenlli cimetsys etuca;esaesid cinorhc fo ecnedive gnidecerp eht ni esu citoiborp esu citoibitna ;skeew eerht]92[ycnegreme cirtaideP :DEP ;lairt dellortnoc dezimodnaR :TCR .lagutroP dna ,modniK detinU eht ,learsI ,eceerG ,sdnalrehteN ,ainevolS ,ylatI ,aitaorC ,tpygE ,natsikaP ,dnaloP ni detcudnoc saw la te iniladnauG fo yduts ehT.GG susonmahr sullicabotcaL :GGL ;tnemtraped tneitaptuO :DPO ;tnemtraped

A reduced duration of diarrhea was found in the LGG group compared to that in the matched group according to 15 RCTs submitted to meta-analysis, which included 3721 participants [MD -24.02 h, 95%CI (-36.58, -11.45)] (Figure 1A). Significantly heterogeneous results were detected among the included trials (I2 = 98%). Our data support the results of the prior meta-analyses[4]indicating that LGG treatment reduced participants' duration of diarrhea.

Subgroup analyses were conducted based on clinical features such as age,geographical location, treatment time, outpatient or inpatient settings, the time of enrollment, and literature quality scores. Differences in methodological quality could not explain the statistically significant heterogeneity (Supporting information Figure S6). Regression analysis between the duration of diarrhea and LGG dose revealed that different doses of LGG contributed to the heterogeneity (P = 0.009, adjusted Rsquared = 40.21%), suggesting that subgroups according to a high or low dose of LGG should be assessed. A reduced duration of diarrhea was noted in the studies applying＞ 1010CFU/day of LGG [MD -22.56 h, 95%CI (-36.41, -8.72)] (Figure 1A). In contrast,although only three studies used lower dosages, no statistically significant differences were detected in the groups receiving lower dosages [MD -30.95 h, 95%CI (-83.28, -21.39)] (Figure 1A). A reduced duration of diarrhea was supported in the studies with participants who received LGG treatment before the second day of diarrhea symptoms [MD -1.58 h, 95%CI (-3.05, -0.11)] and during the second to third days of diarrhea symptoms [MD -15.83 h, 95%CI (-20.06, -10.98)] (Figure 1B). However,Ritchie et al[37]enrolled participants with diarrhea for more than 3 d, and no statistically significant differences were found in the duration of diarrhea [MD 1.2 h,95%CI (-21.42, 23.82)] (Figure 1B). A reduced diarrhea duration was reported in studies performed in both Asia and Europe [MD -24.42 h, 95%CI (-47.10, -1.82); MD -32.02 h, 95%CI (-49.26, -14.79), respectively]. Paradoxically, the reduction in the diarrhea duration in other regions was not statistically significant [MD -9.35 h, 95%CI(-20.77, 2.07)] (Figure 1C). In the etiological analysis, the effectiveness of LGG was clearly demonstrated in rotavirus-induced diarrhea cases [seven RCTs; MD -31.05 h,95%CI (-50.31, -11.80)] (Figure 2). Analysis with the studies carried out in the 1990s and 2000s revealed a clear reduction in the diarrhea duration [MD -36.32 h, 95%CI (-62.20, -10.45); MD -29.40 h, 95%CI (-50.56, -8.25), respectively] (Supporting information Figure S7). In contrast, no reduction in the diarrhea duration was observed in the analysis with studies carried out in the 2010s [MD -3.43 h, 95%CI (-13.25, 6.39)] (Supporting information Figure S7). No studies evaluated the effectiveness of LGG in children vaccinated against rotavirus.

Diarrhea ≥ 3 d

A meta-analysis of four RCTs was performed using a fixed-effects model. The risk of experiencing diarrhea for 3 or more days was reduced when patients received LGG[odds ratio (OR) 0.54, 95%CI (0.38, 0.77)] (Figure 3A).

Diarrhea ≥ 4 d

Three studies were pooled (n = 479) and showed a reduction in the risk of diarrhea lasting for 4 or more days for participants treated with LGG [OR 0.58, 95%CI (0.4,0.84)] (Figure 3B).

Stool number and consistency

Stool number and consistency were evaluated in most trials. Eight trials reported the mean number of stools in one day during diarrhea episodes. A notable decrease in the stool number per day was noted in the LGG group [MD -0.9, 95%CI (-1.56, -0.23)](Figure 4A). However, a significantly reduced stool number was observed in the highdose LGG groups receiving no less than 1010CFU/day [MD -1.08, 95%CI (-1.87, -0.28)], while the lower-dose groups showed no significant reduction [MD -0.25 d,95%CI (-1.43, 0.93)] (Figure 4A). After the intervention, stool frequency was evaluated on days 2 and 3. Seven trials provided data on day 2, and the overall effect did not differ between the two groups [MD -0.46, 95%CI (-1.06, 0.15)] (Figure 4B). In addition,similar frequencies were observed in the two groups on day 3, with no differences between them [MD 0.34, 95%CI (-0.29, 0.97)] (Figure 4C). Three trials calculated the mean time to improvement in stool consistency, and an obvious reduction was reported [MD -5.65, 95%CI (-7.49, -3.80)] (Figure 4D).

Hospital stay duration

A total of 1823 participants from six RCTs were analyzed. Due to statistically significant heterogeneity, a random-effects model was used, which revealed a significant reduction in the hospital stay duration in the two groups [MD -39.16 h,95%CI (-72.24, -6.07)] (Figure 5A). A reduction in the hospital stay duration was found in rotavirus-positive children [MD -21.12 h, 95%CI (-26.96, -15.28)] (Figure 5B).

Vomiting

Vomiting in different trials was reported as the number of participants with vomiting[number (%)] or as the duration of vomiting (hours). Compared with the placebo group, no difference in the risk of vomiting was reported in the experimental group[OR 1.11, 95%CI (0.59, 2.12)] (Figure 6A). Furthermore, no reduction in the duration of vomiting was noted with LGG treatment [MD -2.02 h, 95%CI (-4.24, 0.21)] (Figure 6B).

Adverse effects

Probiotics have been proposed to be well-tolerated and safe therapeutic agents. Most authors did not report adverse effects. Raza et al[36]reported one case of myoclonic jerks in their trial. Lower rates of respiratory infection, wheezing, and even vulvar abscess were noted in Schnadower's trial[8,39], but these effects were not thought to be correlated with LGG use[40]. Aggarwal et al[40]reported no adverse effects, and a metaanalysis performed in 2013 showed comparable rates of adverse effects among study groups[13]. In our study, eight studies effectively evaluated the safety of LGG. Adverse effects were reported on a coded reporting form or during daily telephone calls[26,34]. In Schnadower's study, the caregivers completed a daily diary that was collected by telephone or through email[8]. However, the reporting methods were unclear in five articles[24,37,39-41]. In general, no adverse effects or similar rates of side effects were documented in the LGG and placebo groups.

Risk of bias in the included studies

The risk of bias in 18 articles was assessed according to the Cochrane Handbook for Systematic Reviews of Interventions. One trial employed alternating group allocation,and the random sequence generation method was not reported in five trials. Other RCTs provided detailed randomization methods, which mainly included computergenerated strategies, resulting in a low risk of selection bias. Allocation concealment was not applied in two trials and was not mentioned in seven. Nine trials used the sealed envelope technique for allocation concealment. Double blinding was strictly executed in 12 trials, while four trials allowed openness to patients or doctors, and

Figure 2 Lactobacillus GG vs control with regard to mean duration of diarrhea (hours) in children with rotavirus diarrhea. LGG: Lactobacillus rhamnosus GG;CI: Confidence interval; SD: Standard deviation.

two trials did not report a detailed blinding method. For detection bias, investigators were blinded to the group assignments in ten trials, while blinding assessments were not performed in three trials. Most trials provided complete data with a loss to followup rate less than 10%, although one trial had an unknown risk of incomplete outcome data, reflecting a low risk of attrition bias (Supporting information Figure S3).

Publication bias

According to Egger's[18]regression asymmetry test, no small sample or publication bias was found in a funnel plot [P = 0.10, 95%CI (-11.33, -1.14)] (Supporting information Figure S8).

DISCUSSION

Findings and agreement or disagreement with other studies

Nineteen trials comparing a control group with an experimental group treated with LGG were identified in this meta-analysis. In summary, the analysis revealed that treatment with LGG reduced both the duration of diarrhea and the hospital stay duration, especially in specific patient subsets. A striking finding was the time to improvement in stool consistency, which more investigators have confirmed since 2010[8,34,40]. In the whole range of diarrhea cases, the management of stools with this probiotic strain showed a modest beneficial effect on the number of stools per day and the time to improvement in stool consistency. However, no reduction in stool frequency was observed on days 2 and 3. Compared with the placebo group, the risk of diarrhea lasting more than 3 and 4 d was reduced by LGG administration. In both groups, similar rates of vomiting and adverse effects were observed.

Evidence from RCTs confirmed the beneficial effect of LGG on rotavirus-induced diarrhea[42]. In addition to interference with viral replication, most recent studies have shown that LGG prevented injuries to the epithelium and ameliorated rotavirusinduced diarrhea by modulating immune cells, such as dendritic cells and inflammatory cytokines[43,44]. The marked statistical difference in the diarrhea duration with a higher dosage of probiotics reflected greater effectiveness, which confirmed the dose dependence of dendritic cell activation. Treatment efficacy was related to the dose of LGG[45]. As confirmed in the study of Szajewska et al[13]in 2013, the importance of a daily LGG dose is high, and a dosage of 1010CFU/day is needed for a positive effect. The statistical heterogeneity between studies can be explained by the timing of the LGG intervention, which was directly correlated with indicators such as the duration of diarrhea before study enrollment. Although the heterogeneity persisted in the subgroup with the shortest duration of diarrhea before study enrollment,probiotics should be applied early in the course of disease. Moreover, symptoms are usually mild at the early stage. Differences in prominent pathogens, sanitation conditions, and common comorbidities lead to dissimilarities between various study locations. Due to differences in the treatment effect among regions, the implications for clinical practice should be evaluated. The nutrient intake and dietary structure of humans have continuously changed in recent decades, which may have caused the reduced effectiveness of LGG reflected in the results of the trials conducted in the 2010s.

Figure 3 Lactobacillus GG vs control with regard to the presence of diarrhea. A: Diarrhea lasting ＞ 3 d; B: Diarrhea lasting ＞ 4 d. LGG: Lactobacillus rhamnosus GG; CI: Confidence interval.

Probiotics manipulate and restore the gut microbiota, thus benefitting the prevention of diarrhea. Various therapeutic interventions designed to alter the microbiota range from probiotic administration to fecal microbiota transplantation[46,47]. However, due to the limited number of included studies and the selflimiting nature of disease, strategies should also be discussed in detail. Vomiting was reported as an adverse event in numerous studies[48,49], and it is one of the most common symptoms associated with diarrhea[50,51]. Additionally, less frequent clinical symptoms were observed in the probiotic groups[4], although our meta-analysis showed no improvement in the risk or duration of vomiting.

Safety

The safety of probiotic supplementation is generally certain. Nevertheless,pathologies correlated with the use of probiotic products to treat gastrointestinal disorders have been identified, such as endocarditis, sepsis, and bacteremia[52-54].Unfortunately, the most prevalent strain implicated in the adverse effects was Lactobacillus rhamnosus. Conversely, most authors in our analysis did not report adverse effects or the adverse effects were not thought to be correlated with LGG treatment. In addition to the interventions, the primary illness contributed the most to the participant drop-out rate. A higher frequency of negative effects attributed to probiotics was found in catheterized (82.5%) and immunosuppressed (66%)participants[55]. Further safety evaluations of probiotics are necessary in the clinical setting, especially for susceptible individuals, such as those with immunodeficiency,immunosuppression, or malnourishment.

Application prospects

Preventing or correcting dehydration through treatment with zinc or 0.9% saline solution is the main approach used for diarrhea management[56]. However, during diarrhea episodes, infectious symptoms are not fully alleviated and the gut microbiota is not restored by rehydration measures[57]. Probiotics were investigated as therapeutic agents for diarrhea. The mechanisms by which probiotics alleviate diarrhea are described below. Host defenses are reinforced by enhanced antimicrobial peptide secretion. Probiotics prevent disruption of gut barrier integrity and stimulate the expression of junctional adhesion and tight junction molecules[58-61]. They produce short-chain fatty acids and induce the production of IgA to resist infections[62-64]. In epithelial cells and mucin, probiotics compete for binding sites to arrest pathogen colonization[65]. Probiotics can specifically and nonspecifically interfere with the viral cycle, thus impeding the progression of rotavirus-induced diarrhea[66-68]. The prevalence of diarrhea is seasonal, and almost all cases of rotavirus-induced diarrheaoccur from January to May in Russia[38]. By contrast, in regions where rotavirus is not prevalent, bacterial diarrhea commonly occurs from June to October[38]. Influenza seasons, dietary habits, and antibiotic use must be considered when evaluating heterogeneity in further studies. The efficacy of probiotic treatment was altered based on host and environmental factors[12]. Overall, our study supported the previous systematic reviews which concluded that LGG is an effective treatment for children with acute diarrhea.

Figure 4 Lactobacillus GG vs control with regard to stool number and consistency. A: The average stool number per day (high dose and low dose); B: Stool frequency on day 2; C: Stool frequency on day 3; D: The mean time to improvement in stool consistency. LGG: Lactobacillus rhamnosus GG; CI: Confidence interval;SD: Standard deviation.

Figure 5 Lactobacillus GG vs control. A: The duration of hospital stay (hours); B: The hospital stay duration of rotavirus-positive children (hours). LGG:Lactobacillus rhamnosus GG; CI: Confidence interval; SD: Standard deviation.

Conclusions and limitations

Although most studies have suggested that LGG is efficacious, limited identification of pathogens, small sample sizes, varying therapeutic strategies, and methodological limitations such as articles without a strictly blinded design, including a lack of a standard clinical parameter format, weakened the conclusions and precluded further analyses across studies[69]. For example, Czerwionka-Szaflarska et al[28]did not specifically define the treatment applied, although a significantly reduced duration of diarrhea was detected. Salazar-Lindo et al[41]partially depicted the duration of diarrhea in children with or without LGG treatment. Although factors varied in the trials, according to the same criterion for both groups, no evidence suggests that a poor study design leads to overestimation of probiotic efficacy[4]. Appropriate subgroups, such as those stratified by etiology and nutritional status, are indispensable. In 2016, approximately 8.4% of children (480000) presenting with diarrhea ultimately died due to the condition worldwide (https://data.unicef.org/topic/child-health/diarrhoeal-disease/). Assessments of the availability of vaccines, the applicability of probiotics, and the effectiveness of current treatments under severe conditions and cost-effect analyses must be performed to optimize therapeutic strategies for acute diarrhea management in children.

In summary, the following conclusions were cautiously established: LGG reduces the duration of diarrhea, particularly in patients with rotavirus-positive diarrhea receiving a dosage no less than 1010CFU per day and in patients treated at the early stage. In addition, studies conducted in Asia and Europe showed greater treatment efficacy. The therapeutic effect of LGG supplementation on the stool number per day and hospital stay duration associated with rotavirus-induced diarrhea is high.

Figure 6 Lactobacillus GG vs control with regard to vomiting. A: The number of participants with vomiting [number (%)]; B: The duration of vomiting (hours).LGG: Lactobacillus rhamnosus GG; CI: Confidence interval; SD: Standard deviation.

ARTICLE HIGHLIGHTS

who received a course of LGG had better outcomes, including a markedly reduced duration of diarrhea, especially those with rotavirus-positive diarrhea, those who received no less than 1010CFU per day, and those treated at the early stage. Furthermore, studies conducted in Asia and Europe reported greater treatment efficacy. The therapeutic effect of LGG supplementation on the stool number per day and hospital stay duration associated with rotavirus-induced diarrhea was high.

Research perspectives

Our study found better outcomes among children with acute diarrhea who were treated by LGG supplementation. Limited identification of pathogens, small sample sizes, and a lack of a standard clinical parameter format precluded further analyses across studies, thus weakening the evidence required to guide clinical practice. Investigations are required to assess the costeffectiveness of treating diarrhea with probiotics.