American college of gastroenterology monograph on the management of irritable bowel syndrome

Abstract

This editorial reviews a recently published guideline on management of irritable bowel syndrome. The guideline illustrates problems arising from the quality of clinical trials used in systematic reviews and the potential impact of the inherent weaknesses of those trials on rating the strength of evidence and the resulting recommendations.

Keywords:

• American college of gastroenterology
• evidence
• GRADE
• guidelines
• irritable bowel syndrome
• medications

1. Introduction

The stated goal of the American college of gastroenterology (ACG) monograph [1] was to update the most recent systematic reviews commissioned by the ACG on irritable bowel syndrome since 2009 [2]. Such monographs are appreciated by different groups: clinicians use guidelines to select appropriate therapies for patients; editors, publishers and journal owners note the effects on the impact factor and advertising revenue; and third-party payers use guidelines to determine reimbursement for medications prescribed.

The methodology in developing the guidelines was the standards recommended in the Institute of Medicine’s “Clinical Practice Guidelines We Can Trust” (http://www.iom.edu/∼/media/Files/Report%20Files/2011/Clinical-Practice-Guidelines-We-Can-Trust/Clinical%20Practice%20Guidelines%202011%20Insert.pdf), including methodological experts and clinicians, the use of systematic reviews, explanation of the reasoning underlying the recommendation, ratings of the level of confidence in and strength of the evidence, articulation of the evidence in a standardized form, and consideration of external comments. Cochrane reviews (http://www.cochrane.org/sites/default/files/uploads/handbook/Whats%20new%20in%20Handbook%205_1_0.pdf) appraise quality of trials based on random sequence generation, allocation concealment, blinding of participants and study personnel, blinding of outcome assessment, incomplete outcome data, selective reporting and size. In general, the accompanying recommendations [1] regarding treatment of chronic idiopathic constipation (CIC) are non-controversial and will not be discussed here. This editorial focuses on the overall guidelines and recommendation of some of the treatments for irritable bowel syndrome (IBS).

The problems with guidelines on IBS (which are not limited to the specific guidelines discussed here) often stem from the quality of the trials that were used in the systematic reviews and the debatable ratings of the strength of the evidence, and the resulting recommendations. The best examples of these deficiencies pertain to the appraisal of antidepressants and alosetron.

2. Design of IBS trials used in systematic reviews

1. Trial duration: studies selected for the systematic review required that subjects be followed for at least 1 week. Although the monograph does not identify trials with only 1-week follow-up, one may question whether such a trial is really sufficient to make recommendations for clinical practice. There is also wide variation in the duration of studies for some drug classes (e.g., 4 – 12 weeks for antidepressants [3], in contrast to at least 12 weeks for 5-HT₃ receptor antagonists [4]).

2. Study endpoints: for eligibility, the IBS trials needed to include global assessment of improvement in IBS symptoms or improvement in abdominal pain. The mean number of stools per week during therapy was applied in the review of treatment trials for CIC, but not in IBS trials. Thus, the review did not appraise the effects of treatment on diarrhea and related symptoms. These diarrhea-related symptoms impact quality of life in IBS-diarrhea [5], and a survey of multi-national IBS experts considered altered bowel habit (17%) and bloating (54%) in addition to abdominal pain (29%) as most bothersome symptoms of IBS [6]. These issues raise questions as to the generalizability and utility of the IBS guidelines beyond the symptom of abdominal pain.

3. Heterogeneity of trial quality and size: many conclusions of the current monograph are consistent with or draw from current or prior meta-analyses, and their recommendations are non-controversial and strong. For example, the appraisal of 5-HT₃ antagonists for IBS-D [7] or selective 5-HT₄ receptor agonists in CIC [8] included several large RCTs involving thousands of patients, uniform inclusion criteria and validated primary and secondary endpoints.

   • However, there are trial quality factors for which the meta-analysis simply cannot control, even though heterogeneity (I² expressed as percent) may ‘qualify’ the strength of conclusions. These quality factors differ significantly within a drug class, perhaps best exemplified in the antidepressant systematic review [3], and include: differences in experimental design (e.g., type of baseline, duration of study), ethnic or racial differences associated with extremely low placebo response rates in Iranian patients in contrast to Western patients [9], and variations in the endpoints that assess ‘global improvement’. We drew attention to this potential weakness in the application of meta-analyses and cautioned that you cannot make a silk purse from a sow’s ear or from the little ears of different strains of sows [9].

4. The generation gap: it is challenging to base recommendations by assessing trials from more than a decade ago, which had size, design, endpoints, and documentation (e.g., allocation concealment) that differed markedly from recent trials that benefited from expertise of trial methodologists, changes in trial design, validated patient response outcomes [10], robust endpoints, and guidance from regulatory agencies including FDA [11] and EMA [12].

In summary, the design of IBS trials used in systematic reviews compromises the ratings of the confidence in the strength of the evidence, and the resulting recommendations emanating from the appraisal of pharmacological treatments of IBS.

3. GRADE quality appraisal and the impact on recommendations

The very nature of the grading of the quality of evidence may be subjective, particularly for ‘moderate (or low)’ quality. The recommendation for evidence of such quality is often “further research is likely (or very likely) to have an important impact on our confidence in the estimate of effect and may change the estimate”. In the case of alosetron for IBS-diarrhea (adjudicated as moderate quality of evidence), it is unclear what additional information could conceivably change the quality of evidence, given the total of 13 trials containing 8173 patients, the largest number of patients ever appraised with a drug for IBS.

Despite the adjudication of the review group, some of the guidance is equivocal and impacted by the selective approach to appraise only global endpoints and abdominal pain. For example, the quality of evidence on antidepressants is adjudicated as high, even though several meta-analyses of antidepressant efficacy are based on small single-center or tertiary-referral trials (raising questions of generalizability), heterogeneous drug classes, markedly different endpoints, and overall patient numbers in 18 trials of 1100 patients (416 on tricyclic antidepressants [TCA], 176 on selective serotonin reuptake inhibitor [SSRIs] and 508 on placebo, or an average of 61 patients per trial). In fact, the cited [3] heterogeneity index (I²) was 35% for TCA and 49% for SSRI, suggesting at least moderate heterogeneity that would almost certainly be regarded as potentially large heterogeneity if the 95% CI of I² was calculated, as recommended by Ioannides et al. [13].

With the (questionably) high quality of evidence, antidepressants as a class are given a weak recommendation as ‘effective in symptom relief in IBS’.

The quality of evidence for alosetron is (questionably) moderate, and alosetron, as an effective treatment in females with IBS, is given a weak recommendation. In a systematic review and meta-analysis published in 2009 [4], there were 8 trials with 3214 alosetron, 1773 placebo and an average of 623 patients per trial, 10 times higher than the antidepressant trials! The global efficacy endpoints (adequate relief or pain), study design and duration, patient subgroup and gender were uniform in all trials. The relative risk of global IBS symptoms or abdominal pain [4] persisting with alosetron compared with control was 0.79 (95% CI 0.69, 0.90). This analysis was confirmed by the meta-analysis of 10 trials of alosetron (6232 patients), which showed relative risk of response of 1.23 (1.15 – 1.32) for relief of abdominal pain and discomfort, and 1.55 (1.40 – 1.72) for the 4 trials (1732 patients) that assessed the global improvement of IBS symptoms [7].

It is difficult to concur with the authors’ appraisal of the quality of evidence and recommendation regarding alosetron, given the uniformity of trial design and efficacy demonstrated in several published meta-analyses. The consensus GRADE methodology synthesizes several factors that may influence the overall recommendation: quality of evidence, applicability to all patient groups, benefits versus risks, patient values and cost. It appears that the panel’s summary appraisal is influenced by assessment that alosetron “can be expensive and not freely available”, or the risk of complications. However, post-marketing and clinical experience show that constipation is eminently manageable with dose titration and therefore an irrelevant risk in clinical practice and the ∼ 0.1% risk of ischemic colitis of the non-gangrenous, superficial mucosal variety are unassociated with significant morbidity or mortality [14]. Global efficacy and safety have been confirmed in trials conducted after introduction of the risk management strategy, confirming relief of diarrhea and urgency [15] and improved quality of life [16], which are not appraised in the current guideline document. Thus, one may reasonably conclude that the evidence on alosetron was high quality and the recommendation to use alosetron for the appropriate clinical indication is strong.

In summary, grading of the quality of evidence appears to be vulnerable to subjective interpretation given the diverse factors such as quality of evidence, applicability to all patient groups, benefits versus risks, patient values and cost that need to be appraised. Hence, the methodological process intended to remove bias is potentially subjective too.

4. Expert opinion

The guideline documents provide an appraisal of the current state of treatment of IBS. Many of the conclusions are indeed evidence-based with unequivocal evidence of efficacy (e.g., linaclotide and lubiprostone for CIC or IBS-C); other conclusions are more contentious because the evidence for efficacy is based on weak studies and the recommendations are therefore weak, either because the medication is not efficacious or the studies were of poor quality to appraise efficacy. The utility of recommendations in guidelines is therefore variable, and one could argue that a narrative review by content expert(s) based on the plethora of published systematic reviews of drugs for IBS would provide just as useful guidance to clinical gastroenterologists managing patients with all IBS subgroups. The old adage “do not confuse lack of efficacy from the failure of a trial to demonstrate efficacy” needs to be remembered [17], particularly with medications tested > 10 years ago. The field of IBS would make significant advances if there were more high-quality clinical trials based on robust evidence of mechanisms that are perturbed in IBS, selection of patients in this multifactorial disease based on biomarkers such as colonic transit and bile acid malabsorption [18,19] and, perhaps, by incorporation of modeling in addition to available evidence in the development of clinical practice guidelines [20].

Declaration of interest

The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.