Small intestinal bacterial overgrowth (SIBO) is a disease defined by an overgrowth of bacteria in the small intestine.¹ The diagnosis of SIBO is based on a positive lactulose or glucose breath test or a small bowel aspirate culture containing > 10³ CFU of bacteria.^1-3 SIBO has been associated with a wide array of symptoms including diarrhea, constipation, bloating, and abdominal pain and appears to be prevalent in the disorder of the gut-brain interaction such as irritable bowel syndrome (IBS).⁴ Specifically, in IBS, SIBO may play a mechanistic role in the development of maldigestion, malabsorption, low-grade inflammation, and altered motility.^5,6

Three large multicenter randomized controlled trials have shown that non-absorbable broad-spectrum antibiotic, rifaximin reduces abdominal pain and diarrhea in patients with diarrhea predominant IBS.^7,8 Rifaximin appears to have a transient effect at reducing the relative abundance of certain taxa such as Enterobacteriaceae.⁹ Similarly in SIBO, symptoms such as bloating appears to be associated with an increase in taxa such as Enterobacteriaceae, Escherichia-Shigella, and a Clostridium.¹⁰ To this date, no large randomized controlled trials have evaluated the efficacy of antibiotics in reducing the symptoms of SIBO.

In the last decade, 2 meta-analyses of small trials have shown that antibiotics can eradicate SIBO in most cases, based on repeat testing after treatment.^11,12 However, no study has evaluated the pooled efficacy of antibiotics in treating symptoms in patients with SIBO. Hence, we performed a systematic review and meta-analysis on the symptomatic response to antibiotics in patients with SIBO. In addition, given the large overlap between the 2 conditions, we compared the efficacy of antibiotics in treating the symptoms of IBS in patients with and without SIBO.

Search Strategy and Study Selection

This meta-analysis followed the protocols described by Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA).¹³ We searched databases MEDLINE (inception to March 3, 2021), EMBASE (inception to March 18, 2021), Web of Science (inception to March 3, 2021), and Cochrane (inception to March 3, 2021).

Search terms for MEDLINE, Web of Science, and Cochrane are summarized in the Appendix. The inclusion and exclusion criteria are described in Table 1. Both randomized control trials (RCTs) and prospective cohort studies were included. RCTs that reported outcomes for patients with SIBO as a subgroup analysis were also included. For cross-over studies, the first treatment before the cross-over was included. For studies that did not report dichotomous outcomes, study authors were contacted to retrieve data. Only prospective studies were included. Abstracts of the studies found using the search terms were extracted and reviewed (W.T. and A.R.). All potentially relevant studies that met the inclusion and exclusion criteria had their full manuscripts reviewed and the data was extracted. The 2 authors reviewed the data independently. Any disagreements were resolved through discussions.

Table 1 . Eligibility Criteria for Systematic Review and Meta-analysis

Inclusion criteria
RCT or prospective cohort studies
SIBO diagnosed based on glucose or lactulose breath test, or using small bowel aspirate
Dichotomous outcome reporting on patients’ symptoms
Compared efficacy of antibiotics vs placebo/no treatment in patients with SIBOOR
Compared efficacy of antibiotics in treating IBS patients with SIBO vs no SIBO

Exclusion criteria
Studies that compared efficacy of antibiotics to other antibiotics or probiotics
Retrospective studies, case series, and case studies

RCT, randomized controlled trial; SIBO, small intestinal bacterial overgrowth; IBS, irritable bowel syndrome.

Foreign language papers were translated into English. Meeting abstracts from Digestive Disease Science Week (May 2010 to March 2021), and the American College of Gastroenterology (September 2003 to March 2021), and United European Gastroenterology (October 2013 to March 2021) annual meetings were reviewed.

Outcomes

There were 2 primary outcomes of interest for this study. One was to assess the efficacy of antibiotics in relieving symptoms in patients with SIBO. The second was to compare the symptomatic response in IBS patients with and without SIBO. If the outcome of interest was not presented, or was not presented as a dichotomous result, study authors were contacted to obtain such results.

Data Abstraction

Extracted data included, the study year, study design, method of diagnosing SIBO, the dose and duration of antibiotics, blinding, patient disease, IBS subtypes, country the study took place in, length of follow-up, patient sample size and source, age, sex, racial distribution, and outcome(s).

Quality Assessment

To assess the bias in randomized controlled studies comparing the effectiveness of antibiotics compared to no antibiotics in patients with SIBO, the Cochrane Handbook for Risk Assessment¹⁴ was used. Specifically, each study was evaluated on its random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, and select reporting. Studies that were assessed to have 2 or more categories with high risk of bias were deemed poor quality. To assess the bias in studies assessing the effect of antibiotics on patients with or without SIBO in IBS and the non-randomized studies evaluating the efficacy of antibiotics on SIBO, the revised Newcastle-Ottawa Scale¹⁵ was used. Specifically, the studies were scored on the representativeness of the exposed cohort, selection of the non-exposed cohort, ascertainment of exposure, demonstration that the outcome of interest was not present at the start of the study, comparability of cohort on the basis of the design or analysis, assessment of outcome, if follow-up was long enough for the outcome to occur, and the adequacy of follow-up were graded. If the total score was 8 or 9/9 then they were considered high quality, if 7 then medium quality, and if 6 or less low quality.

Statistical Methods

A random effect model was used for analysis. For the first outcome of interest, the pooled relative risk (RR) with 95% confidence intervals (CIs) were calculated for the number of patients that improved on antibiotics compared to those not on antibiotics. Similarly, for the second objective the RR for those improved on antibiotics was calculated for those with SIBO vs no SIBO. We utilized a 3-level hierarchical model to account for the inclusion of both randomized and non-randomized studies for the first meta-analysis. For the second meta-analysis a 2-level hierarchical model was employed as the analysis involved only the prospective cohort portions of each study. Where possible, the intention-to-treat results were used in the analysis. For the purpose of statistical analysis, studies that had 0 as the outcome for both groups had their values changed to 0.5 for both groups. The heterogeneity of the studies was calculated using Q test and I² for the 3-level and 2-level hierarchical models respectively and the publication bias was assessed using Egger’s test. Furthermore, data were stratified by continent and duration of treatment and length of follow-up and the RR with 95% CI for the risk of improvement were calculated for the 2 primary outcomes. The meta-analysis is implemented by R. The escalc function was used to calculate effect size or outcome measures, and the rma function in metafor package was used to perform the random effects meta-analysis.

The search identified 694 citations, of which 647 were excluded based on the title or abstract, and 46 were evaluated further in detail. Of these, 38 were excluded for reasons listed in Figure 1, and 8 were included in the final analysis. There were 4 studies that evaluated the symptomatic response to antibiotics compared to no antibiotics in patients with SIBO,^16-19 2 studies evaluated the symptomatic response in patients with or without SIBO in IBS,^20,21 and 2 studies evaluated both.^22,23 One study¹⁸ did not report on a dichotomous outcome and the authors were available to provide the relevant data.

Figure 1. Schematic for study inclusion. SIBO, small intestinal bacterial overgrowth.

Efficacy of Antibiotics in Treating Symptoms in Small Intestinal Bacterial Overgrowth

In total, 6 studies evaluated the efficacy of antibiotics in relieving symptoms in patients with SIBO. Of the 196 patients included in the analysis, 101 received antibiotics, and 95 received placebo or no antibiotics. The reported female distribution ranged from 19-62%. Out of the 6, 2 compared rifaximin to placebo,^16,19 1 compared rifaximin with bran to placebo with bran,¹⁸ 1 compared norfloxacin to placebo,²³ 1 compared neomycin to placebo,²² and another compared rifaximin to no antibiotics.¹⁷ Patients in the included studies had concomitant diseases in addition to SIBO including diverticulosis, IBS, cystic fibrosis, and Crohn’s disease (Table 2). Two studies used glucose breath test, 3 studies used lactulose breath test, and 1 used duodenal aspirate to diagnose SIBO. Various ways were used to assess symptomatic response including Crohn’s Disease Activity Index (CDAI), improvement in gastrointestinal symptoms, and improvement in Bristol stool scale. The treatment and follow-up duration were 7-14 days and 0-30 days, respectively. Overall, the RR (95% CI) of improvement was 2.46 (1.33-4.55), P = 0.004 with antibiotics (Fig. 2). The pooled response rate was 49.5% vs 13.7% for those treated with antibiotics vs no antibiotics. The number needed to treat (NNT) for antibiotics to relieve symptoms in SIBO was 2.8, favoring the use of antibiotics. There was no significant heterogeneity between the studies with Q = 6.37, P = 0.270. There was no publication bias based on the Egger’s test (t = 1.1289, df = 4, P = 0.132) although the small sample size may limit interpretation.

Figure 2. Forrest plot of studies comparing the efficacy of antibiotics in small intestinal bacterial overgrowth (SIBO). The pooled response rate was 49.5% vs 13.7% for those treated with antibiotics vs no antibiotics, respectively. Relative risk (95% CI) of improvement was 2.46 (1.33-4.55), P = 0.004. Heterogeneity was Q = 6.37, P = 0.270.

Table 2 . Study Characteristics of Studies Evaluating Symptomatic Response for Antibiotics in Patients With Small Intestinal Bacterial Overgrowth

Study	Study type	Country	Sample size (% female)	Disease	Method to diagnose SIBO	Criteria for symptomatic improvement	Antibiotic used	Duration of therapy (day)	Duration of follow-up	Rate of improvement on antibiotics	Rate of improvement on placebo
Biancone et al,16 2000	Non-randomized	Italy	14 (50%)	Crohn’s disease	Glucose breath test	Change in CDAI	Rifaximin 400 mg BID	7	7	1/7 (0.0%)	1/7 (0.0%)
Pimentel et al,22 2003	RCT	US	93 (62%)	IBS (Rome I)	Lactulose breath test	≥ 50% reduction in composite score of abdominal pain, diarrhea, and constipation	Neomycin 500 mg BID	10	7	21/46 (45.7%)	7/47 (14.9%)
D’Incà et al,18 2007	RCT	Italy	22	Diverticular disease	Lactulose breath test	Global symptomatic improvement	Rifaximin 600 mg BID	14	End of treatment	8/12 (66.7%)	0/10 (0.0%)
Ghoshal et al,23 2016	RCT	India	34 (19%)	IBS (Rome III)	Duodenal aspirate 103 CFU	No longer meeting Rome III criteria for IBS	Norfloxacin 400 mg BID	10	30	12/19 (63.2%)	0/15 (0.0%)
Ghoshal et al,19 2018	RCT	India	13 (54%)	Rome III IBS-C or FC	Lactulose breath test (methane)	BSS ≥ 3	Rifaximin 400 mg BID	14	7	5/6 (83.3%)	4/7 (57.1%)
Furnari et al,17 2019	RCT	Italy	23 (48%)	Cystic fibrosis	Glucose breath test	≥ 50% reduction in composite GI score	Rifaximin 10 mg/kg TID (up to 400 mg TID)	14	21	4/11 (36.4%)	2/9 (22.2%)

SIBO, small intestinal bacterial overgrowth; CDAI, Crohn’s Disease Activity Index; BID, twice a day; RCT, randomized controlled trial; US, United States; IBS, irritable bowel syndrome; CFU, colony forming unit; IBS-C, constipation predominant IBS; FC, functional constipation; BSS, Bristol stool scale; GI, gastrointestinal; TID, three times a day.

Data were stratified by continent, duration of treatment, and length of follow-up (Supplementary Table 1). There was no difference in RR when studies were stratified as coming from Asia,^19,23 Europe,^16-18 or the United States²² (P = 0.985), with RR of 3.80 (0.32-44.74), 3.30 (0.49-21.92), and 3.07 (1.44-6.51), respectively. There was also no difference in RR when stratified by treatment duration, with treatments ≤ 10 days^16,19,22 having an RR of 4.48 (1.08-18.57), vs treatments > 10 days^17,18,23 having an RR of 1.69 (0.89-3.19), P = 0.220. Duration of follow-up had no significant difference either, with ≤ 7 days^16,18,19,22 having an RR of 2.40 (1.14-5.05) vs > 7 days^17,23 having an RR of 4.29 (0.40-46.53), P = 0.647.

Efficacy of Antibiotics in Treating Symptoms in Patients With Irritable Bowel Syndrome and Evidence of Small Intestinal Bacterial Overgrowth

Next, we evaluated whether antibiotic therapy led to a higher rate of symptom improvement in patients with IBS and evidence of SIBO compared to those with IBS and without evidence of SIBO. There were 4 prospective studies identified through the search which included a total of 266 IBS patients, of whom 172 had SIBO and 94 did not. The female distribution ranged from 19-67%. Two studies used Rifaximin,^20,21 while 1 used neomycin,²² and 1 used norfloxacin²³ (Table 3). Criterion for improvement in symptoms evaluated composite or global symptoms in 3 studies while the study by Ghoshal et al²³ defined a response as no longer meeting Rome III criteria for IBS. The RR (95% CI) for improvement after antibiotic therapy in IBS patients with SIBO compared to no SIBO was 2.07 (1.40-3.08), P = 0.0003 (Fig. 3). The pooled response rate was 51.2% in the SIBO group vs 23.4% in the no SIBO group, respectively. The NNT was 3.6. No heterogeneity was observed with I² = 0.01%, P = 0.390. Again, no publication bias was seen based on Egger’s test (t = 0.4197, df = 2, P = 0.716), although the interpretation may be limited due to the small sample size.

Figure 3. Forrest plot of studies comparing the efficacy of antibiotics in irritable bowel syndrome (IBS) patients with or without small intestinal bacterial overgrowth (SIBO). The pooled response rate was 51.2% vs 23.4% in the SIBO and no SIBO group, respectively. Relative risk (95% CI) for improvement was 2.07 (1.40-3.08), P = 0.0003. Heterogeneity was I² = 0.01%, P = 0.300.

Table 3 . Study Characteristics of Studies Evaluating Symptomatic Response for Antibiotics in Irritable Bowel Syndrome Patients With Small Intestinal Bacterial Overgrowth Compared to No Small Intestinal Bacterial Overgrowth

Study	Analyzed study type	Country	Sample size (% female)	IBS type	Method to diagnose SIBO	Criteria for symptomatic improvement	Antibiotic used	Duration of therapy (day)	Duration of follow-up (day)	Rate of improvement for patients with SIBO	Rate of improvement for patients without SIBO
Pimentel et al,22 2003	Cohort	US	55 (62%)	Rome I IBS-D: 48% IBS-C: 35% Other: 13%	Lactulose breath test	≥ 50% reduction in composite score of abdominal pain, diarrhea, and constipation	Neomycin 500 mg BID	10	7	21/46 (45.7%)	3/9 (33.3%)
Ghoshal et al,23 2016	Cohort	India	40 (19%)	Rome III IBS	Duodenal aspirate 103 CFU	No longer meeting Rome III criteria for IBS	Norfloxacin 400 mg BID	10	30	12/19 (63.2%)	3/21 (14.3%)
Rezaie et al,21 2019	Cohort	US	93 (67%)	Rome III IBS-D	Lactulose breath test	≥ 30% decrease weekly abdominal pain and ≥ 50% decrease in day/wk with BSS type 6 or 7 stool during ≥ 2 wk	Rifaximin 550 mg TID	14	28	37/62 (59.6%)	8/31 (43.9%)
Zhuang et al,20 2020	Cohort	China	78 (33%)	Rome IV IBS-D	Lactulose breath test	50% improvement in the global GI symptoms	Rifaximin 400 mg TID	14	14	18/45 (40.0%)	8/33 (24.2%)

IBS, irritable bowel syndrome; SIBO, small intestinal bacterial overgrowth; US, United States; IBS-D, diarrhea predominant IBS; IBS-C, constipation predominant IBS; BID, twice a day; CFU, colony forming unit; BSS, Bristol stool scale; TID, three times a day; GI, gastrointestinal.

Again, data were stratified by continent, duration of treatment, and length of follow-up (Supplementary Table 1). RR of studies in Asia^20,23 vs United States^21,22 was 2.38 (1.33-4.28) vs 1.79 (1.05-3.06); P = 0.483. RR for studies with duration of treatment of < 14 days was 2.18 (0.96-4.98)^22,23 and 1.99 (1.25-3.19) for ≥ 14 days^20,21; P = 0.851. RR for studies with duration of follow-up of ≤ 14 days^20,22 was 1.86 (1.06-3.25) and > 14 days^21,23 was 2.24 (1.29-3.90); P = 0.647.

Of note, there were 2 studies which measured the symptomatic response rates for patients who had their SIBO eradicated ie, had their breath test normalize after antibiotic treatment.^21,22 The pooled response rate was 20/25 (80.0%) for those with negative post-treatment breath test compared to 11/40 (27.5%) for those who had a baseline negative test, RR = 2.85 (1.66-4.89) (P = 0.001).

Quality Assessment

Of the 5 RCTs included in evaluating the efficacy of antibiotics in SIBO, 1 study was deemed to be of poor quality (Table 4).¹⁷ Two studies were deemed to be at high risk for selective reporting, but otherwise had low or unclear risk for the other categories.^18,23 The last 2 studies were deemed to not be at high risk of bias, although there were multiple biases for which an assessment was impossible. The study by Biancone et al¹⁶ was graded using the Newcastle-Ottawa Scale as it was a non-randomized study and was graded as low-quality. For the 4 studies which assessed the efficacy of antibiotics in IBS patients with or without SIBO, 3 studies were graded as poor quality,^20,22,23 and 1 study was considered medium quality (Supplementary Table 2).²¹

Table 4 . Cochrane Risk of Bias Tools for the Studies Evaluating the Efficacy of Antibiotics in Small Intestinal Bacterial Overgrowth

Study	Random sequence	Allocation concealment	Blinding of participants and personnel	Blinding of outcome assessment	Incomplete outcome	Selective reporting
Pimentel et al,22 2003	Unclear risk	Low risk	Low risk	Low risk	Unclear risk	Unclear risk
D’Incà et al,18 2007	Unclear risk	Unclear risk	Unclear risk	Unclear risk	Low risk	High risk
Ghoshal et al,23 2016	Low risk	Low risk	Low risk	Low risk	Low risk	Unclear risk
Ghoshal et al,19 2018	Low risk	Low risk	Low risk	Low risk	Low risk	High risk
Furnari et al,17 2019	High risk	High risk	High risk	High risk	High risk	Unclear risk

To our knowledge, this is the first systematic review to evaluate the efficacy of antibiotics in relieving symptoms in patients with SIBO, and the first study to evaluate whether IBS patients with evidence of SIBO are more likely to improve with antibiotics than IBS patients without evidence of SIBO. Our analysis revealed that antibiotics may be effective at relieving symptoms in SIBO and that IBS patients with SIBO appear more likely to respond to antibiotics than IBS patients without SIBO. The latter findings support a precision medicine approach where testing for SIBO in IBS patients may help identify those more likely to respond to antibiotics.

Previous studies have evaluated the efficacy of antibiotics in eradicating SIBO (ie, achieving a negative breath test after antibiotics).^11,12 Gatta et al¹² identified a total of 32 studies using rifaximin and found an eradication rate of 70.8%, whereas Shah et al¹¹ found an eradication rate of 51.1% evaluating all antibiotics in 10 studies. In this study, we analyzed symptomatic improvement which is arguably the most clinically relevant treatment outcome.

There is increasing evidence of an overlap between SIBO and IBS. The recent meta-analysis by Shah et al⁴ found that SIBO is highly prevalent in IBS, approximately 31.0% (95% CI, 29.4-32.6), with an OR = 4.9 (95% CI, 2.8-8.6) compared to healthy controls. In addition, animal models have shown a high prevalence of SIBO in post-infection IBS models following Campylobacter jejuni infection.⁵ Lastly, gut microbiome dysbiosis caused by chronic stressors appears to be normalized by rifaximin and is correlated with improvements in visceral hypersensitivity in rats.²⁴ Consistent with these points, our results suggest that IBS patients with SIBO are more likely to respond to rifaximin than IBS patients without SIBO.

Notably, not all patients with SIBO improve with antibiotics. This is likely due to the overlapping multifactorial nature of SIBO. Specifically, for IBS, some patients appear to have evidence of altered immune activation/low-grade inflammation,^25,26 increased intestinal permeability,²⁷ visceral hypersensitivity,^28,29 and autoimmunity,³⁰ to name but a few of the proposed mechanisms. It is also important that some patients with SIBO may receive inappropriate antibiotics or may be underdosed. In fact, Gatta et al¹² found that drug dose and co-therapy was independently associated with a higher eradication rate in their meta-regression. Similarly in our study, the lowest response rate was seen in the study with the lowest dose of antibiotics (rifaximin 400 mg twice a day for 7 days).¹⁶ It is also important to note that CDAI assesses hematocrit, extra-intestinal manifestations of inflammatory bowel disease, and abdominal mass, which are not expected to change with antibiotics. Therefore, it is not surprising that no change was observed. In addition, antibiotic efficacy may also be influenced by the presence of mucosal biofilms that protect the bacteria from antibiotics.³¹ Lastly, gut microbiome dysbiosis (which does not fulfill the criteria for SIBO) may contribute to symptoms in IBS, and thus patients with negative breath tests or small bowel aspirate cultures may still respond to antibiotics.⁶ A recent study has shown that a substantial portion of duodenal aspirates may be contaminated, hence, false positive results may also affect the overall rate of response in the included studies.³²

For the analysis evaluating the rate of symptomatic improvement in IBS patients with or without SIBO, 4 studies were included. The RR was 2.07 (1.40-3.08), with a higher efficacy in patients with SIBO. This analysis had less heterogeneity (I² = 19.8%) when compared to the first analysis. This is likely because only studies with IBS were included, with 3 of the 4 studies evaluating global symptoms. These results suggest that the response to antibiotics in IBS patients, in part, is mediated by the presence of SIBO.

There are several limitations to this study. There was significant heterogeneity in the first analysis and the sample sizes for all included studies were relatively small (n = 13-93), such that some of the studies may have been underpowered. At least 1 study in each analysis was not blinded. In addition, given the small number of available studies, RCTs, unrandomized prospective studies, and subgroup analyses were all analyzed together in this meta-analysis. This likely reflects the high heterogeneity seen in the first meta-analysis where this difference is more apparent. However, in the absence of a well-designed, large, multicenter clinical trial, our analysis provides value in summarizing the current evidence.

In conclusion, antibiotics appear to provide symptomatic relief in patients with SIBO. In patients with IBS, testing positive for SIBO predicts a higher likelihood of response to antibiotics. A large, multicenter double-blind randomized trial to validate these findings is needed.

Note: To access the supplementary tables mentioned in this article, visit the online version of Journal of Neurogastroenterology and Motility at http://www.jnmjournal.org/, and at https://doi.org/10.5056/jnm22187.

None.

Will Takakura and Jiajing Wang reports no conflict of interest. Mark Pimentel is a consultant for Bausch Health, Ferring Pharmaceuticals Inc, Salvo, Ardelyx, 9 meters, and Vivante Health Inc. Mark Pimentel has received grant support from Bausch Health and Synthetic Biologics. Ali Rezaie is a consultant/speaker for and has received grant support from Bausch Health. In addition, Cedars-Sinai Medical Center has licensing agreements Hobbs Medical, Aytu Biosciences and Gemelli Biotech. Ali Rezaie and Mark Pimentel have equity in Gemelli Biotech, and Mark Pimentel has equity in 9 meters and Synthetic Biologics. William D Chey reported being a Board member of the American College of Gastroenterology, International Foundation of Functional GI Disorders, and the Rome Foundation; compensation as a consultant from AbbVie, Alnylam, Atmo, Bayer, BioAmerica, Ironwood Pharmaceuticals, QOL Medical, Nestle, Phathom Pharmaceuticals, RedHill Biopharma, Salix/Valeant, Takeda, and Vibrant; grant/research support from BioAmerica, Commonwealth Diagnostics International, QOL Medical, Salix; stock/stock options in Isothrive, Kiwi Bioscience, GI on Demand, and Modify Health; and patents relating to methods and kits for identifying food sensitivities and intolerances, digital manometry, and a rectal expulsion device.

Will Takakura, Ali Rezaie, William D Chey, and Mark Pimentel: planning and/or conducting the study, collecting and/or interpreting data, and/or drafting the manuscript; and Jiajing Wang: planning and/or conducting the study, statistical analysis, collecting and/or interpreting data.