Efficacy and safety of non-pharmacological interventions for irritable bowel syndrome in adults

Yun-Kai Dai, Yun-Bo Wu, Ru-Liu Li, Wei-Jing Chen, Chun-Zhi Tang, Li-Ming Lu, Ling Hu

Abstract

Key Words: Nonpharmacological interventions; Irritable bowel syndrome; Network metaanalysis; Randomized controlled trials; Adults; Clinical practice

INTRODUCTION

Irritable bowel syndrome (IBS) is one of the most common chronic functional gastrointestinal disorders, which is characterized by abdominal pain, irregular defecation or changes in stool property[1,2]. Currently, about 15% of the general population around the world are suffering from this condition[3]. Because of its symptoms IBS affects patients’ work and daily lives and could lead to an increase in healthcare cost[4,5]. According to the latest Rome criteria (Rome IV)[6], IBS is classified into diarrhea predominant, constipation predominant, mixed and unclassified.

However, the pathogenesis of IBS remains unclear. Some factors such as unhealthy lifestyles and diets, psychological factors, visceral allergies, gastrointestinal motility dysfunction and intestinal microbiota alteration have been taken into consideration[7]. Therefore, routine pharmacotherapies (RPs) such as antipsychotics, antispasmodics, promotility agents, laxatives and antidiarrheics are recommended for the management of IBS. Although these interventions can relieve symptoms like abdominal pain, their effects are inadequate and may produce some unwelcome reactions including ischemic colitis and cardiovascular events[8]. Due to the chronicity and recurrence of IBS, many patients are intolerability to pharmacological interventions for a long time and then put their eyes on nonpharmacological interventions (NPI).

As an add-on treatment or alternative option, NPI for IBS include dietary and physical interventions, biofeedback therapy (BFT), cognitive behavioral therapy (CBT), probiotics, acupuncture and moxibustion therapy. Although previous meta-analyses of these therapies showed good efficacy in improving global IBS symptoms[9-14], these studies have concentrated on individual aspects of NPI and are not comprehensive. Therefore, the reliability of the evidence might fluctuate by various assessment outcomes, thereby leading to between-study heterogeneity and mitigating their efficacies in guiding clinical practice.

Network meta-analysis (NMA) is a powerful statistical technique that combines direct and indirect evidence to analyze multiple treatments from different studies and estimate the relative effects of all included treatments in the network simultaneously[15]. Moreover, NMA has the advantage of assisting medical decisionmaking through providing useful and evidence-based data[16]. Based on these, we used NMA to evaluate the comparative effects and rankings of all known NPIs on IBS.

MATERIALS AND METHODS

This study was conducted according to the Cochrane criteria, the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) statement[17]and relevant meta-analysis guidance[18].

Data sources and search strategy

Five electronic databases including OVID EMBASE, MEDLINE, Cochrane Library, PubMed and the Chinese database of CNKI were searched from their inception to January 12, 2020 without language limitation for randomized controlled trials (RCTs). Search strategies were performed with a combination of the following terms: Irritable bowel syndrome, randomized controlled trial, nonpharmacological interventions, biofeedback, cognitive behavioral therapy, probiotics, dietary, acupuncture and moxibustion. Detailed information for each database is displayed in Supporting Information S1. Some unpublished articles were searched in ClinicalTrials.gov and relevant data were obtained through contacting the investigators or authors. In case of duplicates, the most updated one was selected.

Inclusion and exclusion criteria

Relevant titles and abstracts were blindly evaluated and details of selected studies were independently analyzed by two researchers (Dai YK, Wu YB). Based on the PICOS (participants, interventions, comparisons, outcomes and study design) criteria, the following items were included in this NMA: IBS participants whose ages are 18 years or over should meet one of the Rome criteria versions (Rome II, III or IV)[19-21]; NPI should include at least one of the following treatments: Diet, biofeedback, CBT, probiotics, acupuncture or moxibustion; Outcomes should be at least one of these items such as overall clinical efficacy, IBS-SSS (symptom severity scale), SAS (selfrating anxiety scale) and SDS (self-rating depression scale). Moreover, treatment courses should be 4 wk or over. Studies with a Jadad score above 1 was selected for further analysis.

However, publications would be excluded once the following items appeared: Meeting abstracts; incomplete or imprecise data; ambiguous treatment courses; unavailable full texts; cross-sectional studies or reviews.

Data abstraction and quality evaluation

Two investigators (Dai YK, Wu YB) independently performed data extraction and methodological quality assessment. The following data should be extracted from each included trial: Study ID (first author and publication year), general characteristics of patients (gender, age and sample size), diagnostic criteria, details of interventions, treatment courses, primary and secondary outcomes and adverse events. Some absent information was obtained by contacting corresponding authors. The risk of bias of each study was assessed using the Cochrane Collaboration Recommendations assessment tool[22]. Six domains with the evaluation of risk bias were as follows: Random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcomes assessment, incomplete outcome data and selective reporting. Each domain of the included publications was judged as low, unclear or high risk. As for the evaluation of evidence quality, the Grading of Recommendations Assessment, Development and Evaluation (GRADE) was used with the online guideline development tool (https://gdt.gradepro.org/app/). Quality of evidence in this NMA was assessed as high, moderate, low and very low quality[23].

Statistical analysis

Compared with results of standard and pairwise analyses, NMA results can afford more precise estimates and rank interventions to inform clinical decisions[24,25]. Therefore, in order to compare the efficacy and safety of each NPI across RCTs, a NMA was conducted using Stata version 13.0 software. For each treatment, we produced a pooled relative risk for dichotomous outcomes or standardized mean difference (SMD) for continuous variable data with their corresponding 95% confidence intervals (CI) to summarize the effect of each comparison tested using a random-effect model as a conservative estimate. Evidence of direct and indirect multiple-intervention comparisons were examined through producing a network plot where node sizes corresponded to the number of study participants while connection sizes referred to the number of studies for each intervention. According to the Bayesian framework and the Markov chain Monte Carlo method, we evaluated and processed research dataa prioriusing WinBUGS version 1.4.3 (MRC Biostatistics Unit, Cambridge, United Kingdom). Three Markov chains and noninformative uniform and normal priori distributions were used to fit the model[26,27]. Then, 10 thinning intervals each Markov chain and 50000 iterations were equipped so as to obtain their posterior distributions. Of all the simulation iterations, the first 20000 were applied to annealing for the elimination of impacts of the initial value while the last 30000 were used for sampling. Heterogeneity analysis was quantified using the inconsistency index statistic (I2)[28]. TheI2value above 50% was regarded as heterogeneity throughout the study. Accordingly, we conducted sensitivity analysis to verify the robustness of results and test the source of heterogeneity in each RCT. Surface under the cumulative ranking curve (SUCRA) probability value was used to rank the examined interventions[29].

RESULTS

Study selection

All of the 1592 articles were identified from five data libraries based on the wellestablished retrieval. Ultimately, 40 RCTs[30-69]including 4196 participants were selected in the NMA according to the inclusion and exclusion criteria. The study selection process is shown in Figure 1. The baseline characteristics of the included studies are summarized in Table 1.

Risk of bias evaluation

The quality of each included RCT was evaluated using the Cochrane Risk of Bias Assessment Tool[70]including these factors:

(1) Selection bias:Thirty trials grouped patients according to detailed randomized algorithms while the remaining ten only described “randomization.” Therefore, the thirty trials were assessed as “l(fā)ow risk” while the other ten were viewed as “unclear risk.” As for the allocation concealment, four trials were evaluated as “l(fā)ow risk” within detailed information while the remaining 36 trials were viewed as “unclear risk” because of insufficient information.

(2) Performance bias and detection bias:Twelve trials provided information on blinding and were blinded to the outcome assessors. Therefore, both performance bias and detection bias were assessed as “l(fā)ow risk.” However, the remaining 28 trials failed to provide adequate information on blinding. Therefore, both of the two biases were viewed as “unclear risk.”

(3) Attrition bias:Twenty-three trials were evaluated as “unclear risk” for their incomplete data while the remaining seventeen trials were estimated as “l(fā)ow risk” because they reported withdrawal or dropout.

(4) Reporting bias:Because the complete implementation scheme could be acquired, the bias of all the trials was assessed as “l(fā)ow risk.”

(5) Other bias:Considering the lack of information in this item, all included RCTs were estimated as “unclear risk.” The detailed quality evaluation of the included studies is shown in Figure 2.

Network evidence

There were ten regimens in this study as follows: RPs, placebo, probiotics, probiotics + RPs, BFT, BFT + probiotics, CBT, acupuncture, moxibustion and acupuncture + moxibustion. The network graphs of these regimens with different outcomes are displayed in Figure 3.

Primary outcome

Overall clinical efficacy:There were 30 RCTs reporting overall clinical efficacy. As displayed in Table 2, RPs, probiotics, probiotics + RPs, acupuncture, BFT and acupuncture + moxibustion had better overall clinical efficacy than placebo; Probiotics + RPs, acupuncture and BFT had better overall clinical efficacy than RPs and probiotics. The differences among the above mentioned treatments were statistically significant. As shown in Figure 4, the SUCRA plot indicated that acupuncture ranked first, followed by BFT and probiotics + RPs. Meanwhile, heterogeneity analysis (Figure 5A) showed good homogeneity (I2= 0.0%,P= 0.997), and sensitivity analysis (Figure 5B) indicated strong stability in the ranking of all treatments for overall clinical efficacy. Furthermore, the symmetry funnel plot of this endpoint was observed in Figure 6.

Secondary outcomes

IBS-SSS:The improvement of IBS-SSS was reported in seven RCTs with five interventions (RPs, placebo, probiotics, CBT and acupuncture). Compared withplacebo (Table 3), CBT (SMD = 2.39, 95%CI: 1.71, 3.07), RPs (SMD = 2.15, 95%CI: 1.39, 2.90) and probiotics (SMD = 0.30, 95%CI: 0.07, 0.52) had significantly statistical differences. CBT (SMD = 2.09, 95%CI: 1.46, 2.73) and RPs (SMD = 1.85, 95%CI: 1.13, 2.57) were superior to probiotics. CBT (SMD = 0.24, 95%CI: -0.09, 0.57) was better than RPs. According to the SUCRA plot (Figure 7), CBT was the optimal intervention, RPs was the second and acupuncture was the third.

Table 1 Characteristics of the studies included in the network analysis

Cheng et al[46], 2017 China IBS-D (Rome III)19/22 18/21 E: 36.27 ± 2.78 C: 41.69 ± 12.63 N/A 8 CBT RPs d, f, o N/A N/A Kang et al[47], 2016 China IBS-D (Rome III)17/23 16/24 E: 44.5 ± 6.4 C: 42.5 ± 7.2 N/A 4 Probiotic + RPs RPs a, i, j N/A N/A Robin et al[48], 2016 France IBS (Rome III)31/161 31/156 E: 45.3 ± 15.7 C: 45.4 ± 14.1 N/A 12 Probiotics Placebo a, b, e, m N/A E: 10 C: 0 Zhang et al[49], 2016 China IBS (Rome III)12/18 14/16 E: 40.7 ± 11.4 C: 36.3 ± 14.1 E: 3.58 ± 2.04 C: 3.88 ± 2.36 4 Probiotics RPs a N/A E: 0 C: 2 Han et al[50], 2016 Korea IBS (Rome III)13/10 11/12 E: 45.7 ± 9.55 C: 42.5 ± 10.07 N/A 4 Probiotics Placebo a, k, l, p N/A N/A Jia et al[51], 2016 China IBS (Rome III)16/14 22/10 E: 40.08 ± 13.23 C: 41.31 ± 11.82 N/A 8 CBT RPs f, o N/A N/A Choi et al[52], 2015 South Korea IBS (Rome III)a: 20/34 b: 35/25 C: 35/23 d: 25/31 26/31 E: a: 44.8 ± 13.4 b: 48.9 ± 14.2 C: 46.2 ± 13.8 d: 45.9 ± 12.8 C: 48.5 ± 13.2 N/A 6 Probiotics + RPs Placebo a, b, m N/A E: 4/8/8/8 C: 6 Jia et al[53], 2015 China IBS (Rome III)N/A N/A E: 44.74 ± 11.98 C: 40.85 ± 13.87 N/A 8 CBT RPs d, o N/A N/A Shi et al[54], 2015 China IBS-D (Rome III)28/32 25/35 E: 40.2 ± 10.8 C: 38.5 ± 9.1 E: 8.6 ± 3.8 C: 7.3 ± 2.1 4 AP RPs a N/A N/A Li[55], 2015 China IBS-D (Rome III)N/A N/A E: 46 C: 46 E: 4.2 C: 4.2 4 AP RPs + Probiotics a, e, g N/A N/A Ye et al[56], 2015 China IBS (Rome III)N/A N/A 43.59 ± 12.17 2.42 ± 1.27 4 BFT + Probiotics Probiotics o, r, v N/A N/A Zheng[57], 2014 China IBS-D (Rome III)49/40 49/36 40/42 52/34 E: 38.75 ± 18.32 42.66 ± 16.75 42.51 ± 16.78 C: 42.29 ± 18.30 E: 72.91 ± 76.70 78.83 ± 99.19 77.51 ± 84.56 C: 87.67 ± 90.28 d 4 AP RPs b, k, l, o, q, s N/A E: 3 C: 0 Zhu et al[58], 2014 China IBS-D (Rome III)9/6 7/6 E: 47.470 ± 0.896 C: 40.920 ± 10.136 E: 3.0 C: 3.5 4 MB Placebo d, t, u N/A N/A Kong[59], 2014 China IBS-D (Rome III)14/16 9/21 E: 40 ± 9 C: 38 ± 11 E: 5.87 ± 6.52 C: 6.21 ± 6.33 4 AP+MB RPs a, d, e N/A N/A He et al[60], 2014 China IBS-D (Rome III)N/A N/A 37.3 ± 10.4 3.7 ± 2.1 4 BFT + RPs RPs a, g, i, n, v N/A N/A Cheryl et al[61], 2014 South Africa IBS (Rome III)2/52 0/27 E: 48.15 ± 13.48 C: 47.27 ± 12.15 E: 9.58 ± 10.32 C: 10.05 ± 9.36 6 Probiotics Placebo b, d N/A E: 1 C: 0 Lesley Britain IBS (Rome III)15/73 15/76 E: 44.66 ± 11.98 N/A 4 Probiotics Placebo a, d, e, f, m N/A N/A

AP: Acupuncture; BFT: Biofeedback therapy; C: Control group; CBT: Cognitive behavior therapy; E: Experiment group; F: Female; IBS: Irritable bowel syndrome; IBS-C: Constipation-predominant irritable bowel syndrome; IBS-D: Diarrhea-predominant irritable bowel syndrome; M: Male; MB: Moxibustion; N/A: Not applicable; RPs: Routine pharmacotherapies (including antispasmodic, laxative, antidiarrheic, antidepressant, glutathione); TCM: Traditional Chinese medicine. a: Overall clinical efficacy; b: Adverse effect rate; c: Recurrent rate; d: IBS-QOL (Quality of life); e: Clinical symptoms scores (abdominal pain/discomfort, flatulence, diarrhea, stool frequency, stool consistency); f: IBS-SSS (IBS symptom severity scale); g: The expression of immunohistochemistry (5-HT, TNF-α, IL-8, IL-10, ); h: TCM symptom scores; i: HAMA & HAMD (The Hamilton Anxiety & Depression Rating Scale); j: Change in intestinal flora (Escherichia coli, Lactobacillus, Bifidobacterium, Enterococcus faecalis); k: Bristol Stool Form Scale; l: Frequency of clinical symptoms (abdominal pain, diarrhea, constipation); m: SGA (subject’s global assessment); n: BSS (Bowel Symptoms Scale); o: SAS and SDS (self-rating anxiety scale and self-rating depression scale); p: VAS-IBS (Visual Analogue Scale); q: SF-36 (The Medical Outcomes Study 36-item Short-form Healthy Survey); r: Total and specific scores of GSRS (Gastrointestinal Symptom Rating Scale); s: The weekly average number of days with normal defecations; t: fMRI Examination; u: The Birmingham IBS Symptom Scale; v: Rectal distention threshold comparison; w: Visceral Pain threshold.

SAS and SDS:In this NMA, seven RCTs with five treatments (RPs, probiotics, BFT, CBT and acupuncture) reported improvement of SAS and SDS. As show in Table 4, CBT (SMD = 3.44, 95%CI: 1.49, 5.39), acupuncture (SMD = 3.39, 95%CI: 1.19, 5.58) and RPs (SMD = 3.13, 95%CI: 1.28, 4.97) had better significant improvement of SAS than probiotics. CBT (SMD = 0.31, 95%CI: -0.31, 0.94) was superior to RPs. As for the improvement of SDS, Table 4 showed that CBT (SMD = 2.97, 95%CI: 1.70, 4.23), BFT (SMD = 2.81, 95%CI: 1.86, 3.77), acupuncture (SMD = 2.36, 95%CI: 1.01, 3.72) and RPs (SMD = 2.27, 95%CI: 1.06, 3.49) were better than probiotics. CBT (SMD = 0.15, 95%CI: -0.68, 0.99) was superior to BFT. Acupuncture (SMD = 0.09, 95%CI: -0.51, 0.69) was better than RPs. Meanwhile, the SUCRA plot suggested that CBT was the mostfavorable treatment in the improvement of SAS and SDS (Figure 8).

Table 2 Risk ratios with 95% confidence interval of overall clinical efficacy

Table 3 Standardized mean difference with 95% confidence interval of irritable bowel syndrome symptom severity scale

Table 4 Standardized mean difference with 95% confidence interval of self-rating anxiety scale and self-rating depression scale

Adverse effects

A total of sixteen RCTs with six interventions (RPs, placebo, probiotics, probiotics + RPs, acupuncture and moxibustion) reported adverse effects. There were no significant statistical differences among these treatments (Table 5). According to the SUCRA plot (Figure 9), acupuncture was the most favorable intervention, probiotics was the second and moxibustion was the third.

Table 5 Risk ratios with 95% confidence interval of adverse effects

Figure 1 Flow diagram. IBS: Irritable bowel syndrome; RCTs: Randomized controlled trials.

Quality estimates based on the GRADE system

For the primary endpoint, the quality of estimates was “l(fā)ow” (Figure 10). Considering the details of GRADE criteria, the result was possibly derived from quality ratings of direct and indirect comparisons within RCTs, thereby leading to imprecision and unclear risk of bias.

DISCUSSION

NMA is used to analyze trials with multiple interventions and provides rankings for them[71]. Although RPs for IBS can benefit patients, inevitable adverse effects have to be admitted. Accordingly, NPI for IBS have been developed. In this study, to compare the different NPIs, a NMA of multiple NPI comparisons was conducted. Results showed the comprehensive analysis of data for retrievable IBS interventions at present. Based on the SUCRA values, acupuncture was most likely to improve overall clinical efficacy and least likely to result in adverse effects. CBT was most likely to lower the scores of IBS-SSS and SAS and SDS. In summary, when NPIs are used as an alternative therapy in treating IBS, acupuncture and CBT had better efficacy in relieving IBS symptoms.

Figure 2 Risk of bias graph.

Figure 3 Network evidence of four endpoints. A: Overall clinical efficacy; B: Irritable bowel syndrome symptom severity scale; C: Self-rating anxiety scale and self-rating depression scale; D: Adverse effects.

With the exception of the potential factors mentioned earlier, genetic findings in IBS pathogenesis should also be taken into consideration. Gazouliet al[72]confirmed that single nucleotide polymorphisms in genes of serotonergic signaling pathway are associated with at least a subgroup of IBS. For instance, patients who carry an S allele or S/S genotype have differences in the central processing of visceral pain, which could result in a high susceptibility to negative emotional memory and contribute to enhanced visceral pain perception[73,74]. As is well-known, visceral hypersensitivity has been deemed as an important neurological evidence underlying the pathogenesis of abdominal pain in IBS, and visceral pain is associated with a dysregulation of the brain-gut axis[75,76]. Some clinical investigations have confirmed the efficacy of acupuncture in the regulation of the abnormal brain activities and improving visceral hypersensitivity in IBS sufferers[77,78]. Moreover, numerous animal studies have also suggested that acupuncture could significantly reduce the peripheral blood flow of rats with 5-hydroxytryptamine positive reactant content and improve visceral hypersensitivity[79-81].

Figure 4 Surface under the cumulative ranking curve plot of overall clinical efficacy.

As a typical psychosomatic disease, IBS sufferers have more or less cognitive biases and negative coping styles[82,83]. A few studies have shown that CBT could improve these negative emotions and mental tension by means of relaxation training, respiratory training and hypnotherapy, which made them identify uncontrollable stressors[84-86]. Not only that, CBT could also correct their negative coping styles to relieve psychosomatic damage caused by IBS symptoms, thereby improving the overall well-being and quality of life of these patients[87]. Based on this evidence, our findings may supplement the recommendations of existing guidelines and identify specific NPI with better effects.

Consistency is viewed as a one-way comparative relationship between direct and indirect evidence in an NMA[88]. It would be lack of transitivity if there was an inconsistency in a statistical analysis. In this paper, although heterogeneity analysis indicated good homogeneity and sensitivity analysis suggested strong stability in overall clinical efficacy, clinical heterogeneity such as the improvement of IBS-SSS, SAS and SDS, which were evaluated by an excessive personal opinion from professional practitioners or participants should be noticed. Meanwhile, comprehensive evaluation of outcome measurements on different IBS types should also be seriously considered.

There were several limitations in this study. First, although RCTs are insusceptible to many biases, some certain defects in them including design, conduct, analysis and reporting may lead to bias. In this NMA, the methodological quality of all RCTs was moderate and quality estimates based on the GRADE system showed “Low,” which may originate from some overlooked details on randomization and blinding, especially for CBT, BFT, acupuncture and moxibustion that were hard to blind. Second, strict inclusion and exclusion criteria were used in this study, but the number of each NPI in all included trials had relatively large differences (acupuncture /moxibustion: 13 trials, CBT: 4 trials, BFT: 5 trials and probiotics: 18 trials), which was likely to influence the strength of the evidence. Third, although all included RCTs were assessed based on the Cochrane Risk of Bias Assessment Tool, any assessment of bias is subjective. We have to admit that no quantitative index could assess only artificial risk of bias so far. Finally, 32 (80%) of the included RCTs were conducted in China, which may reduce the universality of our results.

Figure 5 Heterogeneity and sensitivity analysis. A: Heterogeneity analysis; B: Sensitivity analysis. CI: Confidence interval; OR: Odds ratio.

CONCLUSION

In conclusion, evidence from this NMA showed that acupuncture could be beneficial for patients with IBS because of improved overall clinical efficacy and less adverse effects. CBT had preferable effects in lowering the scores of IBS-SSS, SAS and SDS. However, more RCTs should be performed to confirm the impact of NPIs on other IBS symptoms, and additional high-quality clinical research should be conducted to offer more powerful evidence in the future.

Figure 6 Funnel plot of overall clinical efficacy. BFT: Biofeedback therapy; RPs: Routine pharmacotherapies.

Figure 7 Surface under the cumulative ranking curve plot of irritable bowel syndrome symptom severity scale.

Figure 8 Surface under the cumulative ranking curve plot of self-rating anxiety scale and self-rating depression scale. A: Self-rating anxiety scale; B: Self-rating depression scale.

Figure 9 Surface under the cumulative ranking curve plot of adverse effects.

Figure 10 Grading of Recommendations Assessment, Development and Evaluation quality grading assessment.

ARTICLE HIGHLIGHTS