Replication and extension of a meta-analysis of antidepressants for irritable bowel syndrome: a comparison of odds ratios and risk ratios using artificial intelligence-powered tools
Lefteris Teperikidisa,b,c, Christos Mademlisd, Georgios Hatzinakose, Nikolaos Lazaridisf
Synthesa, Inc, NY, USA; School of Medicine, Aristotle University of Thessaloniki, Thessaloniki, Greece; Hippokration General Hospital, Aristotle University of Thessaloniki, Thessaloniki, Greece; Medical School, Hippokration General Hospital, Aristotle University of Thessaloniki, Thessaloniki, Greece; St. Luke’s Hospital, Thessaloniki, Greece; The Royal Free Hospital, University College London Institute for Liver and Digestive Health, London, United Kingdom
aSynthesa, Inc., 19 West 24th St., New York, NY 10010, USA (Lefteris Teperikidis); bClinical Research Unit, Special Unit for Biomedical Research and Education (SUBRE), School of Medicine, Aristotle University of Thessaloniki, Thessaloniki, Greece (Lefteris Teperikidis); cThird Department of Cardiology, Hippokration General Hospital, Aristotle University of Thessaloniki, Thessaloniki, Greece (Lefteris Teperikidis); dSecond Propedeutic Department of Internal Medicine, Medical School, Hippokration General Hospital, Aristotle University of Thessaloniki, Thessaloniki, Greece (Christos Mademlis); eDepartment of Gastroenterology, St. Luke’s Hospital, Thessaloniki, Greece (Georgios Hatzinakos); fRoyal Free Unit for Endoscopy, The Royal Free Hospital, University College London Institute for Liver and Digestive Health, London, United Kingdom (Nikolaos Lazaridis)
We read with great interest the recent article by Temido et al, evaluating the efficacy of antidepressants in irritable bowel syndrome (IBS) through a systematic review and meta-analysis of randomized, double-blind, placebo-controlled trials [1]. Their study represents a significant contribution to the IBS literature by applying high methodological standards and demonstrating clinically meaningful benefits across various symptom domains.
To evaluate the reproducibility and extend the generalizability of these findings, we used a novel large language model (LLM)-based tool we developed for title and abstract screening. We replicated the original study’s selection process using a broad search strategy (PubMed and Scopus, total of 43,487 citations; 28,645 after deduplication) on May 27, 2025. Our tool successfully identified all 20 studies reported by Temido et al, plus 6 additional randomized controlled trials reporting binary outcomes suitable for inclusion in the meta-analysis [2-7]. We also identified 2 relevant studies that, like 4 in the original work, lacked extractable binary/dichotomous outcome data [8,9]. Our second LLM tool—designed to auto-generate R code for meta-analysis—was used to replicate the original meta-analytic computations and extend them.
Using the original dataset of 16 trials (n=1,428), we replicated the meta-analysis in R using the {meta} package. The model used was: effect measure: odds ratio (OR); model: Mantel-Haenszel (MH); between-study variance estimator: restricted maximum likelihood (REML); and confidence interval method: Hartung-Knapp. These align closely with the methodology reported by Temido et al, who also used a random-effects model, REML, and conducted intention-to-treat analyses via Stata v16.
The resulting pooled effect size using our script was slightly higher than that of Temido et al (OR 3.18 vs. 3.02), with a broader confidence interval (95%CI 2.13-4.73 vs. 2.16-4.2). This numerical difference was probably due to software-specific implementation differences, including continuity corrections and default tau2 estimators. Despite these minor discrepancies, both analyses confirmed the significant benefit of antidepressants in improving IBS symptoms.
We also conducted a parallel analysis using risk ratio (RR) as the effect measure—an approach often considered more clinically intuitive for interpreting data from randomized controlled trials. We then repeated both OR and RR meta-analyses after incorporating 6 newly identified studies, expanding the dataset to 22 trials (n=1946). Across all 4 analyses, the findings consistently supported the clinical efficacy of antidepressants (Fig. 1).
Figure 1 Composite figure showing 4 forest plots—odds ratio (OR) and risk ratio (RR) meta-analyses for both the original (16-study) and updated (22-study) datasets
CI, confidence interval
We commend the authors for their rigorous study and suggest that future publications consider including both OR and RR metrics to broaden interpretability across audiences. We also highlight the value of integrating artificial intelligence-based review pipelines to complement traditional evidence synthesis.
References
1. Temido MJ, Cristiano M, Gouveia C, Mesquita B, Figueiredo P, Portela F. Antidepressants in irritable bowel syndrome:a systematic review and meta-analysis of randomized controlled trials. Ann Gastroenterol 2025;38:284-293.
2. Myren J, Groth H, Larssen SE, Larsen S. The effect of trimipramine in patients with the irritable bowel syndrome. A double-blind study. Scand J Gastroenterol 1982;17:871-875.
3. Tabas G, Beaves M, Wang J, Friday P, Mardini H, Arnold G. Paroxetine to treat irritable bowel syndrome not responding to high-fiber diet:a double-blind, placebo-controlled trial. Am J Gastroenterol 2004;99:914-920.
4. Wright-Hughes A, Ford AC, Alderson SL, et al. Low-dose titrated amitriptyline as second-line treatment for adults with irritable bowel syndrome in primary care:the ATLANTIS RCT. Health Technol Assess 2024;28:1-161.
5. Drossman DA, Toner BB, Whitehead WE, et al. Cognitive-behavioral therapy versus education and desipramine versus placebo for moderate to severe functional bowel disorders. Gastroenterology 2003;125:19-31.
6. Tack J, Broekaert D, Fischler B, Van Oudenhove L, Gevers AM, Janssens J. A controlled crossover study of the selective serotonin reuptake inhibitor citalopram in irritable bowel syndrome. Gut 2006;55:1095-1103.
7. Talley NJ, Kellow JE, Boyce P, Tennant C, Huskic S, Jones M. Antidepressant therapy (imipramine and citalopram) for irritable bowel syndrome:a double-blind, randomized, placebo-controlled trial. Dig Dis Sci 2008;53:108-115.
8. Myren J, Løvland B, Larssen SE, Larsen S. A double-blind study of the effect of trimipramine in patients with the irritable bowel syndrome. Scand J Gastroenterol 1984;19:835-843.
9. Sharbafchi MR, Afshar Zanjani H, Saneian Z, Feizi A, Daghaghzadeh H, Adibi P. Effects of duloxetine on gastrointestinal symptoms, depression, anxiety, stress, and quality of life in patients with the moderate-to-severe irritable bowel syndrome. Adv Biomed Res 2023;12:249.
Notes
Conflict of Interest: Lefteris Teperikidis is co-founder of Synthesa, Inc., the company that develops the tools used in this validation study. Lefteris Teperikidis has consulted for SCRIPPS Research, Callibr BV, Parexel, Bruker GmbH, IVDeology, Pharmassist, Accuscript, Remedica and PARI GmbH, outside the present work. The other authors have no conflict of interest to declare