Objective. Childhood diarrhea accounts for substantial morbidity and mortality worldwide. Multiple studies in children have shown that Lactobacillus, administered orally, may have antidiarrheal properties. We conducted a meta-analysis of randomized, controlled studies to assess whether treatment with Lactobacillus improves clinical outcomes in children with acute infectious diarrhea.
Methods. Studies were sought in bibliographic databases of traditional biomedical as well as complementary and alternative medicine literature published from 1966 to 2000. Search terms were “competitive inhibition,” “diarrhea,” “gastroenteritis,” “Lactobacillus,” “probiotic,” “rotavirus,” and “yog(h)urt.” We included studies that were adequately randomized, blinded, controlled trials in which the treatment group received Lactobacillus and the control group received an adequate placebo and that reported clinical outcome measures of diarrhea intensity. These inclusion criteria were applied by blind review and consensus. The original search yielded 26 studies, 9 of which met the criteria. Multiple observers independently extracted study characteristics and clinical outcomes. Data sufficient to perform meta-analysis of the effect of Lactobacillus on diarrhea duration and diarrhea frequency on day 2 were contained in 7 and 3 of the included studies, respectively.
Results. Summary point estimates indicate a reduction in diarrhea duration of 0.7 days (95% confidence interval: 0.3–1.2 days) and a reduction in diarrhea frequency of 1.6 stools on day 2 of treatment (95% confidence interval: 0.7–2.6 fewer stools) in the participants who received Lactobacillus compared with those who received placebo. Details of treatment protocols varied among the studies. A preplanned subanalysis suggests a dose-effect relationship.
Conclusion. The results of this meta-analysis suggest that Lactobacillus is safe and effective as a treatment for children with acute infectious diarrhea.
Diarrhea is common among children and contributes substantially to pediatric morbidity and mortality worldwide. In the United States, an estimated 21 million to 37 million episodes of diarrhea occur among 16.5 million children younger than 5 years of age annually.1 Three million physician visits per year are related to diarrhea,1 as are 163 000 hospitalizations, or 13% of all hospitalizations for children in this age group.2 Given the ubiquity of acute infectious diarrhea (ID) and its associated burdens on children, families, and the health care system, all parties desire a therapy that is safe, relatively inexpensive, and effective in ameliorating the course of illness.
For at least a century, researchers have hypothesized that live bacterial cultures, such as those found in yogurt, may help treat and prevent diarrhea.3 The bacterial genus Lactobacillus, which is found in normal human intestinal and perineal flora, has been studied frequently in children with regard to its antidiarrheal properties since the 1960s.4,5 Studies published in the world literature have concluded that Lactobacillus is indeed safe and effective in treating and preventing ID; antibiotic-associated diarrhea; and diarrhea in children who are unusually susceptible as a result of poor nutrition, impaired immune status, or frequent exposure to pathogens. Despite these reports, health professionals in the United States do not routinely recommend Lactobacillus,6 perhaps believing that its effectiveness has not yet been proved.7,8 We therefore conducted a meta-analysis of existing randomized, controlled studies to test the hypothesis that treatment with Lactobacillus improves clinical outcomes in children with ID.
We sought trials that involved human subjects in the traditional biomedical literature as well as the complementary and alternative medicine literature. To this end, the following databases were searched: from 1966 to 2000, Medline, PubMed, EMBase, CCTR (Cochrane Controlled Trials Register), DARE (Database of Abstracts of Reviews of Effectiveness), CINAHL (Cumulative Index to Nursing and Allied Health); from 1985 to 2000, AMED (Allied and Alternative Medicine), MANTIS (Manual, Alternative and Natural Therapy), Complementary and Alternative Medicine Citation Index, and AltHealthWatch. The search terms for the disease/therapy pairing used to interrogate the databases were “diarrhea,” “gastroenteritis,” or “rotavirus,” in combination with “competitive inhibition,” “Lactobacillus,” “probiotic,” or “yog(h)urt.” Search terms were modified slightly to correspond to the subject headings and tree structures of some databases. We also sought trials in reference sections of other clinical trials and review articles. Key investigators in the field were contacted and asked to provide other known clinical trials.
We first limited the entire set of trials to original studies that involved Lactobacillus treatment of ID in children, in which clinical outcomes were reported. Studies included in the meta-analysis were adequately randomized, blinded, controlled trials in which the treatment group received Lactobacillus (any species or strain) and the control group received a suitable placebo. A killed Lactobacillus species was considered unsuitable as a placebo, because there is evidence that this agent itself may have antidiarrheal properties.9–11 Randomization was considered adequate when a study was described as randomized, even if the precise randomization method was not reported. When Lactobacillus and control were indistinguishable and when the individuals who recorded clinical data were blinded as to which treatment was given to which subjects, blinding was considered adequate. Because no consensus exists regarding a consistent definition of diarrhea, the manner in which diarrhea was defined was not a criterion for inclusion or exclusion. Studies in which patients had received recent antibiotics were excluded. No language restrictions were used in the literature search or selection process.
The studies that originated from the database search were examined by 2 of the authors (C.F. and M.M.G.). These authors independently read only the methods section of the studies and were blinded to all information about the author, site, journal, year, or title of each study. Two authors (C.W.V.N. and D.A.C.) resolved any disagreements regarding the inclusion of studies, which arose infrequently and were attributable to reader misunderstanding rather than to true differences of opinion. Study results were not weighted on the basis of assessments of quality.12
The primary outcome measures were the characteristics of the clinical course of diarrhea. Diarrhea was defined variously in the studies as an increase in duration or frequency of diarrhea, increase in volume of stool, or decrease in consistency of stool, as noted by caregivers or investigators. Diarrhea was considered to be infectious by clinical diagnosis, with or without confirmation by laboratory testing. Subjects with bloody as well as nonbloody diarrhea were described in the included studies.
Because 1 or more measures of diarrhea intensity were used in the studies, we presumed a priori that the best measure of diarrhea intensity has components of duration (days of diarrhea), frequency (number of stools per day), and amount (volume of diarrheal stool). We abstracted data from each study using outcomes that best approximated these 3 components (duration, frequency, and amount).
A secondary outcome was whether subjects received additional medical intervention, such as intravenous fluid administration, or additional contact with a health care provider in an ambulatory or inpatient setting. We also noted any reports of adverse reactions as a result of treatment.
The full articles of all studies that met the inclusion criteria were translated into English, if necessary, and reviewed. Data were extracted independently by 2 authors (C.F. and M.M.G.) with subsequent verification by a third author (C.W.V.N.). Each study was examined for sample size, study site, patient demographics, strain of Lactobacillus, definitions of diarrhea, infectious pathogens, adverse effects, and the outcome measures described above. Disagreements, which were infrequent and were entirely attributable to misreading by 1 of the authors, were resolved by clarifying discussion, rereading, and consensus.
The studies were analyzed separately for each measure of diarrhea intensity. Measurements of diarrhea duration were converted to days, maintaining the number of significant digits in the original units of time. Diarrhea frequency was reported as the number of loose stools per day. We calculated an absolute difference between the Lactobacillus and control groups for each of the outcomes in each study. Given that the distribution of duration and frequency of diarrhea may be right-skewed (as a result of potential long-duration or high-frequency cases), inferences based on an assumed normal distribution are not ensured. Although nonparametric analysis would be preferable, we did not have access to original data and therefore could analyze only the means reported in the studies. Nevertheless, summary estimates of the effects of Lactobacillus across the studies are likely to be more accurate and stable than in any of the individual studies. Summary measures were determined by a random effects model because of significant heterogeneity of point estimates among the studies. In the meta-analysis, outcomes across the included studies were examined for evidence of publication bias using the Begg and Mazumdar13 adjusted rank correlation test and the Egger et al14 regression asymmetry test. The dose-response relationship was tested using least-squares linear regression of the log of the dose used in each study and the mean change in diarrhea duration measured in days.
Subanalyses were planned a priori to discern modification of reductions in diarrhea intensity by subject age group, use of adjunctive therapy such as oral rehydration solution (ORS), strain of Lactobacillus, dose and duration of Lactobacillus treatment, location of subject population, whether the patients were ambulatory or inpatients, and infectious agent. All analyses were performed using Stata 7.0 (Stata Corp, College Station, TX).25
The initial search for studies involving Lactobacillus treatment of ID in children yielded 26 journal articles (as of August 1, 2000). A search of the Medical Editors’ Trial Amnesty did not detect any relevant unpublished studies. Eleven studies10,16–25 met inclusion criteria on the basis of examination of the methods sections. Fifteen studies were excluded because they were not controlled,5,26–28 were not randomized,5,26–29 were not blinded,5,26–28,30 had no placebo group,5,26–28,31–36 had an inadequate placebo group (killed Lactobacillus),25 or had exclusion11 or reallocation37 of subjects after randomization. Two other studies were subsequently excluded from the meta-analysis because of methodological issues discovered on examination of the entire article. One of these articles reported data from a subanalysis only,24 and the other reported aggregated data (subjects had either ID or antibiotic-associated diarrhea).25
Characteristics of the included studies are described in Table 1. Eight of the 9 included studies involved inpatients exclusively,10,17–23 and 1 study16 was a multicenter trial with a minority of outpatient subjects. Subjects in all 9 studies received at least ORS in addition to Lactobacillus or control as a part of the experimental protocol. Some subjects in 4 of the studies also received intravenous fluids.19,21–23 Diarrhea duration and frequency of diarrheal stools on day 2 of treatment were the only clinical outcomes reported in multiple studies. We were therefore not able to examine stool amount or frequency of diarrhea stools at other times as outcome measures.
One study16 that examined children who had ID and were receiving ORS did not follow clinical outcomes of subjects who became ill enough during the course of the study to require intravenous fluids. With this study protocol, a small and equal proportion of subjects were removed from the Lactobacillus and control groups. Omission of these data made an intention-to-treat analysis impossible. However, recalculation of the meta-analysis without this study did not result in a significantly different point estimate for reduction of diarrhea duration (0.8 days; 95% confidence interval [CI]: 0.2–1.3 days).
Reduction of Diarrhea
The summary point estimate from the meta-analysis indicates a significant reduction in diarrhea duration of 0.7 days (95% CI: 0.3–1.2 days) in subjects who were given Lactobacillus compared with control subjects (Fig 1). Only 7 studies reported variance in the measurement of diarrhea duration (Table 2).10,16–21 We could not calculate variance because individual subject data were not available, so only these 7 studies could be included in the meta-analysis to estimate the effect of Lactobacillus on duration. Similarly, only 3 studies reported variance in the measurement of diarrhea frequency on day 2 of treatment (Table 2).17,18,22 The summary point estimate for frequency for these 3 studies was 1.6 fewer stools in subjects who were given Lactobacillus than in control subjects (95% CI: 0.7–2.6 fewer stools; Fig 2). No publication bias was detected.
Preplanned subanalyses were performed only when the study characteristic of interest was reported in 3 or more studies. When the studies that were performed in developed countries were analyzed,16–18,20,21 the summary point estimate showed a decrease of 0.8 days of diarrhea (95% CI: 0.1–1.5 days) in subjects who were given Lactobacillus compared with control subjects. Studies that used only live Lactobacillus preparations16–21 also demonstrated a reduction in diarrhea duration of 0.8 days (95% CI: 0.3–1.3 days), whereas studies that included subjects with ID of all causes (not just rotavirus)10,16,17,19,21 demonstrated reduction of diarrhea duration of 0.5 days (95% CI: 0.1–1.0 days). We could not infer any effects of various Lactobacillus strains because of heterogeneity of study results.
In all included studies, adverse reactions that were consistent with signs and symptoms usually associated with ID occurred equally in patients who received Lactobacillus and those who received placebo, except in 2 studies that reported decreased vomiting in the Lactobacillus group.17,22 One study22 reported adverse reactions outside the usual clinical spectrum of ID: myoclonic jerks were noted in 1 patient in the Lactobacillus group and 1 patient in the control group. The analysis of subjects who required additional interventions was not possible, as insufficient data were reported.
A dose-response relationship appears across the 8 included studies that reported diarrhea duration. A significant positive linear association exists between the log of the Lactobacillus dose and the reduction in diarrhea duration in days (P < .01; Fig 3).
The results of this meta-analysis suggest that Lactobacillus is safe and effective as a treatment for ID in children, reducing diarrhea duration by approximately two thirds of a day and reducing the frequency of diarrhea on the second day of treatment by 1 to 2 stools. Furthermore, our preplanned subanalyses suggest that all children with ID, rather than just a subset, may benefit from Lactobacillus. Specific indications and limited use of Lactobacillus in children with ID have been proposed in the Lactobacillus literature.9,16,19,22,24 We did not find, however, that the effect of Lactobacillus on diarrhea duration was modified by country of study or live versus killed Lactobacillus preparation. We also found that Lactobacillus therapy benefited not only cases with documented rotavirus diarrhea but also cases of ID caused by a variety of pathogens, as would be found in ambulatory clinical settings.
Exactly how Lactobacillus exerts its probiotic effect is unclear. Some have postulated that Lactobacillus enhances the immune response,20 elaborates antimicrobial substances,38–40 and occupies intestinal mucosal sites, inhibiting the attachment and growth of pathogenic organisms by achieving competitive exclusion and microbial balance.41 The dose-effect relationship noted in this meta-analysis suggests that Lactobacillus is most effective above a threshold dose (10 billion colony-forming units during the first 48 hours) that reduces diarrhea duration by more than half of 1 day. Although this relationship could support any of the postulated mechanisms, it has been shown that a similar dose of 1010 to 1011 colony-forming units of the species Lactobacillus GG results in colonization of the intestine and inhibition of attachment by pathogens.42 Higher doses of Lactobacillus may lead to a shorter duration of diarrhea. Careful thought is warranted, however, in applying the concept of dose-response relationship to a probiotic agent that, in the case of live Lactobacillus, can replicate. The proposed mechanisms may also explain observations that credited Lactobacillus with the ability to prevent antibiotic-associated diarrhea43,44 and traveler’s diarrhea.45,46 Children who are susceptible to the development of diarrhea as a result of poor nutrition, impaired immune status, or frequent exposure to infectious agents have also been shown to benefit from Lactobacillus administration.34,47–49
Several issues should be kept in mind when evaluating these findings. First, the methods of the included studies differed as to how diarrhea was defined, how diarrhea intensity was measured, which strain of Lactobacillus was used, and how Lactobacillus was administered. Despite these differences, the studies did suggest a consistent conclusion, showing significant reductions in duration and frequency of diarrhea in children who were given Lactobacillus, even when we accounted for possible heterogeneous treatment effects across the studies. Second, the subjects in the studies were almost exclusively inpatients. Lactobacillus-associated reductions in diarrhea intensity might be less pronounced in children who are not sick enough to require hospitalization. However, most subjects in the included studies required only oral rehydration, and some studies excluded children with severe dehydration before enrollment.10,18 The conclusions from this meta-analysis may therefore be plausibly generalized to ambulatory populations with diarrhea. Moreover, diarrhea durations reported in these studies are consistent with the usual course of childhood ID,50 regardless of the severity of illness or location of patient care. In outpatients, the use of Lactobacillus may in fact serve to prevent hospitalization and other adverse outcomes. Third, although none of the included studies was performed in the United States, most originated in developed countries, with incidences and causes of ID similar to those in the United States.50 Extension of the conclusions of this meta-analysis to children in the United States is therefore reasonable. Fourth, most subjects were children younger than 3 years. Younger children are more susceptible to clinical consequences of ID and thus may have the most to gain from Lactobacillus administration. Fifth, we had determined measures of diarrhea intensity a priori that we believed would best relate to the clinical burden of diarrhea. For example, a large amount or frequency of diarrheal stool, regardless of duration of illness, might have a close association with the likelihood of dehydration and subsequent medical intervention or with socioeconomic burden such as parental days of work lost or number of diapers purchased. We were constrained, however, by the outcomes most commonly reported in the studies, namely diarrhea duration and frequency. Finally, publication bias remains a concern, although there was no statistical evidence of its presence. Specifically, we note that this meta-analysis includes studies funded by pharmaceutical and food companies, raising the possibility that sources of funding with vested interests may have biased toward submission and publication of only those studies that found therapy to be beneficial.
These limitations suggest additional research. For example, a large randomized and controlled trial, funded by a nonvested party, could test whether high-dose Lactobacillus is an effective treatment for ID in an ambulatory pediatric population, as no such study has been published. Use of consistent measures of diarrhea intensity would help prevent the challenges of interpretation presented by the variety of measures used in the studies included in this meta-analysis. An outpatient pediatric population and consistent diarrhea intensity measures could also be applied to studies in children with other gastrointestinal disorders and studies examining other treatments.
Should Lactobacillus be used to treat children with ID? Our results indicate that Lactobacillus seems safe and reasonably effective in reducing diarrhea duration and frequency. A crude family-centered cost analysis would also favor Lactobacillus use in children with signs and symptoms of ID. A 48-hour course of a Lactobacillus product is commercially available for approximately $1031 and on average could save approximately 17 hours of caring for a sick child with diarrhea, and 1 to 2 diapers. This framework may help a physician counsel a specific patient’s family when the patient receives a diagnosis of ID. We conclude that Lactobacillus can be recommended in the treatment of children with ID.
THIS NEW CHALLENGE TO PEER REVIEW COULD/WILL BE A DISASTER!
Peer review certainly has its critics, and it takes time, but it’s certainly better than letting public relations departments and the public judge science on instant advertising in the press and on TV.Biotech Firms Bypass Journals to Make News
“It used to be that a scientific breakthrough was taken seriously only if it first appeared in a peer-reviewed journal. But in the race to grab the spotlight, some companies are rushing to release information via esoteric publications that have less-stringent criteria or in news releases.
The upshot: investors and the public may be led to believe certain claims that could later prove to be exaggerated. ‘It undermines public trust in science if key results are released without peer review,’ says Philip Campbell, editor-in-chief of Nature, a 133-year-old British research journal that has published the likes of Charles Darwin.
PPL—which shot to fame in 1997 after helping clone Dolly the sheep— insists it didn’t know about Immerge’s imminent publication. Although it accepts that peer review of experiments ‘is the gold standard’ in scientific publishing, the company says it is often forced to override the convention. ‘We’re a public company and we decided to make a limited press release … as soon as we felt that we had something [stock] price-sensitive,’ says Alan Colman, PPL’s research director. In the high-stakes world of stem cells and cloning, he adds, ‘people don’t have time to hang around and wait for a peer review’…
… Advanced Cell Technology, Inc said it had created a human embryo clone. It reported the details in an obscure 2-year-old Internet-based publication called e-biomed: The Journal of Regenerative Medicine. The paper, which was peer-reviewed, was widely hailed as a landmark and hit the front page of newspapers around the world, including The Wall Street Journal.
… But there is now a chorus of detractors who point to serious potential flaws in the experiment. Advanced Cell’s cloned embryo, these scientists say, had divided into just 6 cells after 5 days, and then died.“
Naik G. Wall Street Journal. January 28, 2002
Noted by JFL, MD
C.F. is supported by grant K08 HS00002 from the Agency for Health Care Research and Quality. D.A.C. is a Robert Wood Johnson Generalist Faculty Scholar.
We thank Susan L. Klawansky, MLS, AHIP, for assistance in database searching; Alyson Littman, MPH, for translation and editing support; Jeff Brosco, MD, PhD, for translation support; and Phillip I. Tarr, MD, for expertise regarding the disease entity and editing support.
- ↵Parashar UD, Holman RC, Clarke MJ, Bresee JS, Glass RI. Hospitalizations associated with rotavirus diarrhea in the United States, 1993 through 1995: surveillance based on the new ICD-9-CM rotavirus-specific diagnostic code. J Infect Dis.1998;177 :13– 17
- ↵Metchnikoff E. Sur la flore du corps humain. Manchester Lit Philos Soc.1901;45 :1– 38
- ↵Pene P, Linhard J, Bernou JC. [The colibacillus-lactobacillus combination in the treatment of diarrhea in adults, children and infants]. Semaine Des Hopitaux.1966;42 :241– 244
- ↵Practice parameter: the management of acute gastroenteritis in young children. American Academy of Pediatrics, Provisional Committee on Quality Improvement, Subcommittee on Acute Gastroenteritis. Pediatrics.1996;97 :424– 435
- ↵Kaila M, Isolauri E, Saxelin M, Arvilommi H, Vesikari T. Viable versus inactivated Lactobacillus strain GG in acute rotavirus diarrhoea. Arch Dis Child.1995;72 :51– 53
- ↵Simakachorn N, Pichaipat V, Rithipornpaisarn P, Kongkaew C, Tongpradit P, Varavithya W. Clinical evaluation of the addition of lyophilized, heat-killed Lactobacillus acidophilus LB to oral rehydration therapy in the treatment of acute diarrhea in children. J Pediatr Gastroenterol Nutr.2000;30 :68– 72
- ↵Boulloche J, Mouterde O, Mallet E. Management of acute diarrhea in infants and toddlers. Controlled study of the antidiarrheal efficacy of killed Lactobacillus acidophilus (LB strain) versus a placebo and a reference agent (loperamide). Ann Pediatr.1994;41 :457– 463
- ↵Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ.1997;315 :629– 634
- Stata Corporation. Stata Statistical Software, Release 7.0.Co llege Station, TX: Stata Corporation;2000
- ↵Pant AR, Graham SM, Allen SJ, et al. Lactobacillus GG and acute diarrhoea in young children in the tropics. J Trop Pediatr.1996;42 :162– 165
- ↵Girola M, Ventura P. [Efficacy of probiotic preparation with living, freeze-dried lactic acid bacteria and yeast on child diarrhoea]. Arch Med Int.1995;61– 71
- Fourrier A, Lequien P. [The treatment of infantile gastroenteritis by the use solely of a combination of colibacillus and lactobacillus. Apropos of 56 cases]. Ann Pediatr.1968;15 :491– 495
- ↵Contreras G, Corti C, Cassani E. [Lactobacillus acidophilus in childhood diarrhea]. Compend Invest Clin Latinoam.1983;3 :114– 116
- Bin LX. [Controlled clinical trial in infants and children comparing Lacteol Fort sachets with two antidiarrhoeal reference drugs]. Ann Pediatr.1995;42 :396– 401
- ↵Taborska J, Pazdiora P. [Smecta and Lactobacillus adicophilus ND in the treatment of acute diarrhoea in children]. Cesk Pediatr.1997;52 :29– 33
- ↵Isolauri E, Juntunen M, Rautanen T, Sillanaukee P, Koivula T. A human Lactobacillus strain (Lactobacillus casei sp strain GG) promotes recovery from acute diarrhea in children. Pediatrics.1991;88 :90– 97
- ↵Coconnier MH, Lievin V, Bernet-Camard MF, Hudault S, Servin AL. Antibacterial effect of the adhering human Lactobacillus acidophilus strain LB. Antimicrob Agents Chemother.1997;41 :1046– 1052
- Axelson LT, Chung TC, Dobrogosz WJ, Lindgren SE. Production of a broad spectrum antimicrobial substance by Lactobacillus reuteri. Microb Ecol Health Dis.1989;8 :131– 136
- ↵Selva M, Jacobus NV, Deneke C, Gorbach SL. Antimicrobial substance from a human Lactobacillus strain. Antimicrob Agents Chemother.1987;31 :1231– 1233
- ↵Bernet MF, Brassart D, Neeser JR, Servin AL. Lactobacillus acidophilus LA 1 binds to cultured human intestinal cell lines and inhibits cell attachment and cell invasion by enterovirulent bacteria. Gut.1994;35 :483– 489
- ↵Saxelin M, Elo S, Salminen S, Vapaatalo M. Dose response colonization of faeces after oral administration of Lactobacillus casei strain GG. Microb Ecol Health Dis.1991;4 :1– 8
- ↵Arvola T, Laiho K, Torkkeli S, et al. Prophylactic Lactobacillus GG reduces antibiotic-associated diarrhea in children with respiratory infections: a randomized study. Pediatrics.1999;104(5) :e64 . Available at: http://www.pediatrics.org/cgi/content/full/104/5/e64
- ↵Oksanen PJ, Salminen S, Saxelin M, et al. Prevention of travellers’ diarrhoea by Lactobacillus GG. Ann Med.1990;22 :53– 56
- ↵Ruiz-Palacios G, Guerrero ML, Hilty M, et al. Feeding of a probiotic for the prevention of community-acquired diarrhea in young Mexican children. Pediatr Res.1996;39 :1089
- Copyright © 2002 by the American Academy of Pediatrics