Replicability and transparency in physiotherapy research: Time to wake up

Replicability and transparency in physiotherapy research: Time to wake up François Jabouille, Ata Farajzadeh, Matthieu P. Boisgontier School of Rehabilitation Sciences, Faculty of Health Sciences, University of Ottawa Partial support No support Power analysis justification Introduction Methods Conclusion Over the past decade, a replication crisis has been recognized in several scientific disciplines,1,2,3 raising concerns about the validity and reliability of published research and eroding scientific credibility. While this crisis has been widely discussed in other fields, its impact on physiotherapy research has received little attention. This study4aimed to provide an overview of the current state of physiotherapy research practices that could suggest low replicability. Extracted data •Research topic •Replication •Preregistration & registered report •Sample size justification •Hypothesis statement and support •Statistical reporting •P-values distributions •Data sharing •Preprints These results suggest shortcomings in the replicability and transparency of the physiotherapy literature. Addressing these issues requires the adoption of best practices, including registered reports, justifying sample sizes, publishing negative results, and sharing data. Delays in implementing these practices risk exacerbating bias in the literature and undermining the scientific credibility of physiotherapy. 1. Murphy J et a. (2025). Estimating the replicability of sports and exercise science research. Sports Medicine. 2. Twomey R et al. (2021). Nature of our literature: a registered report on the positive result rate and reporting practices in kinesiology. Communications in Kinesiology, 1(3). 3. Open Science Collaboration. (2015). Estimating the reproducibility of psychological science. Science, 349(6251), aac4716. 4. Jabouille F et al. (2025) Replicability and transparency in physiotherapy research: Time to wake up. MedRxiv Results 4 National Associations 4 Journals Excluded (n = 708) Review (n = 168) Perspective (n = 120) Commentary (n = 116) Letter to the editor (n = 71) Synopsis (n = 69) Editorial (n = 58) News (n = 32) Correction (n = 22) Protocol (n = 20) Case report (n = 18) Guidelines (n = 9) Historical essay (n = 2) Announcement (n = 2) Association business (n = 1) 1173 articles (2022-2024) 465 original articles 0% 25% 50% Original articles 37% 33% 63% 83% 81% Quantitative studies 28% 24% 12% 1% 0% 0% 75% 0% 25% 50% 75% Full support Sample size justification 53% 17% 30% Partial support No support 38% 59% 59% 50% 0% 25% Proportion of fragile p-values (Expected if 80% power) 30% 2022 2023 2024 35% 18% 47% 0 40 80 120 160 0.00 0.98 1.96 2.94 3.92 4.90 5.88 6.86 7.84 8.82 Z statistic Numberof p-values 0 250 500 750 Numberof p-values 40% Previous study Pilot study Benchmark 83% 26% Full support References Hypothesis support Association Business Announcement Historical Essay Guidelines Case Report Protocol Correction News Editorial Synopsis Letter To The Editor Commentary Perspective Review Original Number of articles 400 300 Quantitative Type of articles Qualitative Other 200 100 0 • 0 replication studies • 37% used a sample size justification Rigorous practicesPublication bias Open practices • 6 studies (1%) • 400 days to publish in a journal • 0 registered reports • 28% of preregistration • 12% shared their data • 83% of positive results • 39% of fragile p-value per article • Gap around the p = .05 threshold Preprints