L26/O-014 Inositol supplementation in polycystic ovary syndrome (PCOS): a systematic review of randomised controlled trial focusing on data integrity

Abstract

Abstract Study question Are randomised controlled trials (RCTs) evaluating inositol for polycystic ovary syndrome (PCOS) methodologically trustworthy, and does excluding untrustworthy studies alter reported treatment effects? Summary answer Three of 28 RCTs met predefined trustworthiness criteria, which was insufficient to allow quantitative comparison of treatment effects between trustworthy and non-trustworthy studies. What is known already Inositol supplementation has been widely studied for the treatment of PCOS, with reported benefits on metabolic, hormonal, and reproductive outcomes. However, the quality of supporting evidence remains variable. While some meta-analyses suggest positive effects, others cite low-certainty findings. Nevertheless, inositol is frequently recommended and widely adopted by patients, often influenced by commercial and social media promotion. A 2024 systematic review formally assessed the trustworthiness of trials in this field. A 2024 systematic review was among the first to assess trial trustworthiness. Since then, tools such as the TRACT checklist have been developed to enable more rigorous integrity evaluation. Study design, size, duration We conducted a systematic review and integrity assessment of 30 RCTs evaluating inositol for PCOS, building on a prior review and updating the search to December 2025. Trial registration: PROSPERO CRD420251129602. Participants/materials, setting, methods We included RCTs reporting on metabolic, hormonal, or fertility outcomes in women with PCOS. Each trial was assessed using the Trustworthiness in RAndomised Controlled Trials (TRACT) checklist. We attempted to reproduce univariable statistical analyses and evaluated baseline randomisation using Monte Carlo simulations. Based on these criteria, studies were classified as meeting or not meeting trustworthiness standards. Treatment effects were then estimated separately for trustworthy and non-trustworthy trials to examine the impact of methodological quality. Main results and the role of chance From 30 RCTs, one had been retracted and one was labelled with an expression of concern. The remaining 28 RCTs were conducted predominantly in Europe and South Asia with 13 studies (46%) originating from Italy and 6 (21%) from India. The median sample size was 60 (range 20 to 526). Twenty-four trials (86%) began recruiting after 2010. From these 24, eight (33%) were prospectively registered, three (13%) were retrospectively registered, while 13 (52%) lacked any trial registration. After applying the TRACT checklist, 16 (57%) RCTs were identified as low risk, four as moderate risk (14%), and eight (26%) as high risk for trustworthiness issues. Overall, we could not reproduce 45% of reported p-values. In 12 papers (43%), more than 50% of p-values were non-reproducible. Simulation-derived baseline p-values deviated from the expected uniform distribution. Notably, five trials (18%) exhibited >40% of baseline p-values that were statistically significant (<0.05). After integrating the TRACT assessment, reproducibility of statistical analyses and evaluation of baseline characteristics, three studies were classified as trustworthy, while the remaining 25 were classified as not meeting trustworthiness criteria. The number of studies meeting trustworthiness criteria was insufficient to allow a quantitative comparison of treatment effects between trustworthy and non-trustworthy studies. Limitations, reasons for caution This study applied novel methods to evaluate trustworthiness of included RCTs. Inconsistency in reported statistical results may reflect limited statistical knowledge or error in processing raw data. Deviations from the expected uniform [0,1] distribution should only serve as an approximation of the null distribution under properly conducted randomised treatment allocation Wider implications of the findings Our findings highlight limitations regarding trial registration, statistical reproducibility and integrity of randomisation in included studies. These issues undermine confidence in reported treatment effects and underscore the need for higher-quality and transparently reported trials to support clinical decision-making in this field. Trial registration number Yes