Background Irritable bowel syndrome with diarrhoea is characterised by frequent, loose or watery stools with associated urgency, resulting in marked reduction of quality of life. Ondansetron, a 5-hydroxytryptamine-3 receptor antagonist, has been shown to benefit patients with irritable bowel syndrome with diarrhoea. Objective To evaluate the effect of ondansetron in irritable bowel syndrome with diarrhoea. Design Phase III, parallel-group, randomised, double-blind, multicentre, placebo-controlled trial in 400 patients, with embedded mechanistic studies. Setting Hospital, primary care and community. Participants Eighty participants meeting Rome IV criteria for irritable bowel syndrome with diarrhoea. Intervention Ondansetron 4 mg (dose titrated up to two tablets three times a day) or matched placebo for 12 weeks. Main outcome measures Clinical – Primary patient-reported end point was % ‘Food and Drug Administration-defined responders’ over 12 weeks. Secondary end points were worst abdominal pain intensity, worst urgency, stool consistency, stool frequency, anxiety, depression and dyspepsia at 12 and 16 weeks. Main outcome measures Mechanistic – Whole gut transit time, faecal water, protease (FP), bile acids and assessment of rectal sensitivity using a barostat. Results Clinical – The study closed early due to slow recruitment. Between 1 January 2018 and 11 May 2020, 80 patients were recruited and randomised (20% of target), 37 to ondansetron, 43 to placebo. Discontinuations (4 ondansetron; 2 placebo) meant 75 completed the 12-week trial treatment. There were four protocol violations. In the intention-to-treat analysis, 15 (40.5%) on ondansetron were primary end-point responders (95% CI 24.7% to 56.4%), and 12 (27.9%) on placebo (95% CI 14.5% to 41.3%), p = 0.19, adjusted OR 1.93 (0.73, 5.11). Pain intensity reduction occurred in 17 (46.0%) on ondansetron (95% CI 29.9% to 62.0%) and 16 (37.2%) on placebo (95% CI 22.8% to 51.7%), p = 0.32. Improvement in stool consistency occurred in 25 (67.6%) on ondansetron (95% CI 52.5% to 82.7%) and 22 (51.2%) on placebo (95% CI 36.2% to 66.1%), p = 0.07. Use of rescue medication, loperamide, was lower on ondansetron [7 (18.9%) vs. 17 (39.5%)]. Average stool consistency in the final month of treatment reduced significantly more on ondansetron, adjusted mean difference –0.5 [standard error (SE) 0.25, 95% CI (–1.0 to –0.02), p = 0.042]. Ondansetron improved dyspepsia score (SFLDQ), adjusted mean difference –3.2 points [SE 1.43, 95% CI (–6.1 to –0.4), p = 0.028]. There were no serious adverse events. Mechanistic – mean (SD). Ondansetron increased whole gut transit time between baseline and week 12 by 3.8 (9.1) hours on ondansetron, significantly more than on placebo –2.2 (10.3), p = 0.01. Mean volume to reach urgency threshold using the barostat increased on ondansetron by 84 (61) ml and 38 (48) ml on placebo, n = 8, p = 0.26. Ondansetron did not significantly alter protease, faecal water or bile acids. Changes in referral pathways substantially reduced referrals, impairing recruitment, which meant the study was underpowered. Conclusion Our results are consistent with previous studies and confirmed ondansetron improves stool consistency and urgency but showed minor effect on pain. We plan to undertake a simplified version of this trial overcoming the changed referral pathways by recruiting in primary care, using software linked to primary care records to identify and randomise patients with irritable bowel syndrome with diarrhoea to ondansetron or placebo and remotely follow their progress; thus minimising barriers to recruitment. Trial registration This trial is registered as ISRCTN17508514. Funding This project was funded by the National Institute for Health and Care Research (NIHR) Efficacy and Mechanism Evaluation programme and will be published in full in Efficacy and Mechanism Evaluation; Vol. 10, No. 9. See the NIHR Journals Library website for further project information. Plain language summary Background Irritable bowel syndrome with diarrhoea is characterised by frequent, loose, or watery bowel movements with marked reduction of quality of life. A previous small study suggested ondansetron benefits patients with irritable bowel syndrome with diarrhoea. Methods A clinical trial aiming to recruit 400 patients meeting established criteria for irritable bowel syndrome with diarrhoea from 18 centres throughout the UK. Patients received either ondansetron or placebo for 12 weeks but neither the investigator and nor patient could tell which they were receiving. They recorded their worst abdominal pain, stool frequency and consistency daily. The main end point was the proportion of patients meeting a standard recommended by the U.S. Food and Drug Administration (FDA). Being called a “FDA responder” meant they showed reductions to both pain and days with loose bowel movements. Other less important end points included pain intensity, stool consistency and frequency. We also measured the time for content to pass through the gut (whole gut transit time). Results The study closed early due to slow recruitment with 80 patients randomised. There were 40.5% of responders in the ondansetron group and 27.9% in the placebo group; however, due to low numbers these differences could be due to chance. Ondansetron produced a significant improvement in average stool consistency in the final month of treatment. Ondansetron slowed whole gut transit time which increased from baseline to 12 weeks by a mean of 3.8 hours, while it fell 2.2 hours on placebo, a difference unlikely to be due to chance. Conclusion These results are consistent with previous studies showing that ondansetron improves stool consistency and slows transit. However, because the numbers recruited were smaller than planned, the apparent improvement in “FDA responder” rate could have been due to chance. A further larger trial is needed to confirm the benefit of ondansetron which should be done in primary care where most patients are to be found. Scientific summary Background Irritable bowel syndrome (IBS), which affects around 10% of the population, accounts for 1.8 million consultations/year in primary care in England and Wales (0.6 million patients). Symptoms of IBS with diarrhoea (IBS-D) include frequent, loose or watery stools with associated urgency, which can severely limit socialising, travelling and eating out, with resulting marked reduction in quality of life and loss of work productivity. Around one-third of all IBS patients meet Rome criteria for IBS-D. When patients are asked to rank symptoms in order of importance, the erratic bowel habit is rated first, followed by abdominal pain and, for those with diarrhoea, urgency. Ondansetron, a 5-hydroxytryptamine-3 (5HT3) receptor antagonist, has an excellent safety record for over 20 years as an antiemetic, but is only exceptionally used in the treatment of IBS-D. It has, however, been shown to slow colonic transit and in a small randomised, placebo-controlled, crossover pilot study, benefited patients with IBS-D. While the current trial was ongoing, a separate trial in the USA using a fixed dose bimodal release formulation (3 mg ondansetron +9 mg delayed release formulation) also reported improvement in stool consistency but not pain. Objectives Our primary aim was to determine the efficacy of generic ondansetron compared to placebo in controlling the symptoms of IBS-D using the US FDA-recommended combined end point in which a responder is defined as a patient who met the response criteria for both pain and bowel habit for 6 out of 12 weeks of the trial. Secondary end points included its effect on the characteristic abnormalities of stool consistency, frequency and urgency as well as abdominal pain, satisfactory relief of IBS symptoms, mood and use of rescue medication and to determine the effect of 12 weeks ondansetron over the 1 month after discontinuation, as well as safety. The study also included mechanistic studies to examine the correlation of rectal sensitivity and compliance, faecal bile acids (FBAs) and proteases and postprandial sigmoid motility with the baseline symptoms of our IBS-D patients. We also attempted to determine whether ondansetron significantly altered these biomarkers compared to placebo. Methods Treatment of irritable bowel syndrome using titrated ondansetron trial (TRITON) was a multisite, parallel-group, randomised, double-blinded, placebo-controlled trial, with embedded mechanistic studies within selected sites. Our aim was to determine the superiority of ondansetron compared with placebo. We aimed to randomise 400 patients with IBS-D on a 1 : 1 basis to receive either ondansetron or placebo. Both treatments were administered for 12 weeks in oral doses ranging from 4 mg every third day to 24 mg daily. Dose titration was undertaken in the first 2 weeks of the study to avoid constipation, which at a standard dose occurs in one-quarter of patients. This was achieved by frequent consultation with the research nurse, starting with 1 × 4 mg tablet per day and increasing in increments every 2 days to a maximum of 2 tablets thrice daily. If constipation developed, the treatment was stopped to allow the return of bowel movements and then restarted at a lower dose, typically one every alternate day or one every second day. Rescue medication of loperamide was discouraged but allowed exceptionally for uncontrolled diarrhoea and was documented in the daily diary. The primary outcome of response for both reduction in pain intensity and improvement in stool consistency was assessed over the 12 weeks post randomisation. Secondary and safety outcomes were measured up to 16 weeks post randomisation. Symptoms that were recorded daily included (1) stool consistency and abdominal pain (measured by both paper diary and daily text message); (2) stool frequency, urgency of defaecation, use of rescue medication (defined as the use of loperamide) over 12 weeks of treatment and the answer to the question in the diary ‘Overall, have you had satisfactory relief from your IBS symptoms in the past week?’. Irritable bowel syndrome symptom severity [measured by the IBS Severity Scoring System (IBS-SSS)], dyspepsia [using the Short Form Leeds Dyspepsia Questionnaire (SFLDQ)], quality of life and mood [using the IBS Quality of Life (IBS-QOL) and Hospital Anxiety and Depression Scale (HADS) questionnaires], and somatic symptoms [using the Patient Health Questionnaire 12 Somatic Symptoms (PHQ-12) questionnaire] were assessed by patient-reported questionnaires at the baseline and 12 weeks post randomisation. The trial also assessed possible underlying mechanisms of any effect of ondansetron on changes in the primary and secondary end points. Whole gut transit was measured at baseline and 12 weeks using radio-opaque markers and an abdominal X-ray. High-resolution manometry was performed at baseline and after 8–11 weeks of treatment at two centres to assess whether ondansetron decreased the number of high-amplitude propagating contractions (HAPCs) or increased the percentage time occupied by cyclical retrograde propagated contractions. Barostat assessment was performed at baseline and after 8–11 weeks of treatment at two centres in order to assess if ondansetron increases rectal compliance or decreases sensitivity (manifested as increased pressure thresholds for pain and urgency). Stool samples were assessed for faecal water % (FW), faecal protease (FP) and FBAs. Clinical results The study closed early due to slow recruitment with just 80 patients randomised; 37 to ondansetron and 43 to placebo. Four patients discontinued ondansetron and one placebo during 12-week randomised treatment. Four were excluded from the per-protocol population due to major protocol violations. In the intention to treat (ITT) analysis, 15 patients (40.5%) on ondansetron achieved the primary end point response [95% confidence interval (CI) 24.7% to 56.4%], compared to 12 (27.9%) patients on placebo (95% CI 14.5% to 41.3%), p = 0.19, adjusted OR 1.93 (0.73, 5.11). Response for pain intensity reduction was achieved by 17 (46.0%) on ondansetron (95% CI 29.9% to 62.0%) and 16 (37.2%) on placebo (95% CI 22.8% to 51.7%), p = 0.32, adjusted OR 1.61 (0.63 to 4.12). Response for stool consistency improvement was reported by 25 (67.6%) on ondansetron (95% CI 52.5% to 82.7%) and 22 (51.2%) on placebo (95% CI 36.2% to 66.1%), p = 0.07, adjusted OR 2.45 (0.92, 6.52). Overall use of the rescue medication, loperamide, was 39.5% (n = 17) on placebo compared with 18.9% (n = 7) on ondansetron. However, by week 12, loperamide use fell to 13.5% on ondansetron versus 25.6% on placebo. Average stool consistency in the final month of treatment fell significantly more on ondansetron than placebo, adjusted mean difference –0.5 [standard error (SE) 0.25, 95% CI (–1.0 to –0.02), p = 0.042]. Ondansetron improved the dyspepsia score (SFLDQ) significantly more than placebo; the largest reduction being in symptoms of indigestion and nausea. The adjusted mean difference in the total score compared to placebo was –3.2 points [SE 1.43, 95% CI (–6.1, to –0.4), p = 0.028]. Ondansetron was well tolerated with most adverse reactions being mild or moderate and not significantly greater than on placebo. The commonest was constipation, reported in 32% on ondansetron and 23% on placebo, of which 75% and 80%, respectively, were rated as mild. Just two patients withdrew citing constipation as the cause. Mechanistic results Results are expressed as mean (SD). Comparing baseline and week 12 showed ondansetron increased average whole gut transit 3.78 (9.1) hours on ondansetron significantly more than placebo –2.2 (10.3), p = 0.01. Mean volume to reach urgency threshold using the barostat increased on ondansetron by 84 (61) ml and 38 (48) ml on placebo, n = 8; the difference was not significant, p = 0.26. Too few underwent manometry to allow meaningful assessment of the effect of ondansetron but anecdotally one patient who had a dramatic clinical improvement showed a loss of HAPCs and an increase in retrograde contractions, but this could have been due to chance. Ondansetron appeared not to significantly alter FP, though overall the increase in whole gut transit time from baseline to week 12 was correlated with a decrease in FP. There were no significant changes in FBAs and no evidence that ondansetron altered these though we did confirm that we had effectively excluded those with bile acid diarrhoea (BAD). The ratio of secondary to primary bile acids, a measure of bacterial metabolism of bile acids, increased substantially on ondansetron from 9.7 (7.08) to 21.4 (32.9) and less so on placebo from 22.84 (58.23) to 28.61 (31.42). However, owing to small numbers and wide variability these differences were not significant. Limitations Two previous studies in Nottingham had recruited 120 and 136 IBS-D patients within 2 years so we did not anticipate problems with recruitment. However, changes in referral pathways from primary to secondary care substantially reduced referrals to our coinvestigators who were all in secondary care, thus impairing recruitment. The power calculations required 400 to achieve 90% power to detect a 15% difference in primary end point, so the study is substantially underpowered. Use of loperamide did somewhat complicate interpretation since those on placebo used more rescue medication reducing the size of the effect on transit and stool consistency. Conclusion Despite being underpowered for our primary end point, our results are consistent with previous studies and confirmed ondansetron improves stool consistency but showed little effect on pain. Ondansetron significantly slowed whole gut transit time. Ondansetron reduced sensitivity to rectal distension more than placebo without altering compliance, but numbers were too small to achieve statistical significance. This could plausibly contribute to the reduction in urgency and stool frequency but needs repeating with larger numbers to be sure it was not due to chance. We found no evidence that rectal sensitivity was related to either faecal protease or bile acids. The manometry studies were underpowered but anecdotally ondansetron appeared to alter rectosigmoid motor patterns in a way that could reduce inflow of stool to the rectum. Future work We plan to do a simplified version of this trial, using an efficient and remote process, to overcome the changed referral pathways by recruiting in primary care. We will search for patients who have had a diagnosis of chronic diarrhoea and the recommended screening including a normal full blood count, a negative tissue transglutaminase (excluding coeliac disease) and a normal faecal calprotectin using software linked to primary care records. This will allow rapid screening of large numbers of patients to identify and approach patients with IBS-D who meet criteria to take part in a randomised trial of ondansetron or placebo, thus minimising barriers to recruitment. We would remove the pain threshold, which would increase the number of eligible patients and facilitate recruitment. Not allowing loperamide as rescue medication would simplify interpretation and dropouts would be treated as treatment failures. Further streamlining by removing all additional tests that were included in the current trial, as well as efficient trial processes, including e-consent, remote blood and stool samples (if required), and online questionnaires would also optimise recruitment. Study registration This trial is registered as ISRCTN17508514. Funding This project was funded by the National Institute for Health and Care Research (NIHR) Efficacy and Mechanism Evaluation programme and will be published in full in Efficacy and Mechanism Evaluation; Vol. 10, No. 9. See the NIHR Journals Library website for further project information.