25gmds072 10.3205/25gmds072 urn:nbn:de:0183-25gmds0728 Meeting Abstract Ringlstetter Ringlstetter Rieke R

Institut für Biometrie, Medizinische Hochschule Hannover, Hannover, Germany

author Müller-Vahl Müller-Vahl Kirsten K

Klinik für Psychiatrie, Sozialpsychiatrie und Psychotherapie, Medizinische Hochschule Hannover, Hannover, Germany

author Koch Koch Armin A

Institut für Biometrie, Medizinische Hochschule Hannover, Hannover, Germany

author Großhennig Großhennig Anika A

Institut für Biometrie, Medizinische Hochschule Hannover, Hannover, Germany

author German Medical Science GMS Publishing House

Düsseldorf

610 matching individual patient data frequentist borrowing re-analysis 20251103 engl This is an Open Access article distributed under the terms of the Creative Commons Attribution 4.0 License. Dieser Artikel ist ein Open-Access-Artikel und steht unter den Lizenzbedingungen der Creative Commons Attribution 4.0 License (Namensnennung). M0631 072 Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie 70. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e. V. (GMDS) V: Medical Biometry 1: Methoden für die Studienplanung Jena 20250907 20250911 Abstr. 57 TextIntroduction: The allocation in randomized controlled trials is usually an equal 1:1 ratio between intervention and control groups. Some trials use an unbalanced 2:1 ratio, allocating more individuals to the intervention group to improve inclusion rates or to obtain more safety data . Starting in 2019, the randomized controlled trial (RCT) CANNA-TICS compared the cannabis extract nabiximols versus placebo (2:1 allocation ratio), with 64 patients in the treatment group and 33 in the placebo group. The primary endpoint was assessed based on the change in the Yale Global Tic Severity Scale-Total Tic Score (YGTSS-TTS) between baseline and end of treatment, using a dichotomous responder criterion. In the sample size calculation, the placebo group was assumed to have almost no effect; however, a small responder effect was observed, contradicting this assumption, as the CANNA-TICS trial failed to demonstrate the efficacy of Nabiximols. This re-analysis explores a strategy for borrowing controls to address unequal allocation ratios. To achieve this, we used controls from another RCT with a very similar patient population for our matching approach.Methods: First, we described and compared the trial populations between the allocation groups at baseline using the IPD. For patient matching, we selected a matching ratio of one match partner per available control from the CANNA-TICS trial, using the SAS macro developed by Mortensen et al. . The baseline YGTSS-TTS (± 1 point) and gender were used as key matching variables to minimize bias in the populations. After matching, the modified CANNA-TICS population was re-analyzed using Mantel-Haenszel risk difference (RD) estimation, consistent with the primary analysis, and generalized linear model (GLM) regression to calculate the odds ratio (OR) accounting for matching.Results: Matching was successful for 26 patients, increasing the CANNA-TICS control group size from 33 to 59. Re-analyzing the modified data using the original trial analysis strategy (-0.15 (RD) [-0.27; -0.03], p = 0.01) and the GLM regression (3.86 (OR) [1.19; 12.59], p = 0.02) led to statistically significant results. Notably, the treatment effects were consistent with the initial analysis results (-0.13 (RD) [-0.28; 0.01], p = 0.07).Discussion and conclusion: The sample size calculation in the CANNA-TICS trial accounted for the unbalanced allocation ratio by increasing the overall power. However, a null effect in the placebo group was mistakenly assumed. Borrowing controls from a similar trial increased power and led to significant results without biasing the treatment effect. Two key learnings can be highlighted: First, in a trial where psychological factors may influence the primary endpoint measure, assuming a null effect may not be appropriate. Second, comparable trial data can support further analyses – for example, if unbalanced ratios are necessary but increased power is needed. This approach has the potential to improve the design of future trials, providing a frequentist alternative to Bayesian methods.The authors declare that they have no competing interests.The authors declare that an ethics committee vote is not required. Nay J Haslam A Prasad V Justification for unequal allocation ratios in clinical trials Nay J, Haslam A, Prasad V. Justification for unequal allocation ratios in clinical trials: A scoping review. Contemp Clin Trials. 2024;139. DOI: 10.1016/j.cct.2024.107484 https://doi.org/10.1016/j.cct.2024.107484 Müller-Vahl KR Pisarenko A Szejko N Haas M Fremer C Jakubovski E CANNA-TICS: Efficacy and safety of oral treatment with nabiximols in adults with chronic tic disorders - Results of a prospective, multicenter, randomized, double-blind, placebo controlled, phase IIIb superiority study 2023 Psychiatry Res 115135 Müller-Vahl KR, Pisarenko A, Szejko N, Haas M, Fremer C, Jakubovski E, et al. CANNA-TICS: Efficacy and safety of oral treatment with nabiximols in adults with chronic tic disorders - Results of a prospective, multicenter, randomized, double-blind, placebo controlled, phase IIIb superiority study. Psychiatry Res. 2023;323:115135. DOI: 10.1016/j.psychres.2023.115135 https://doi.org/10.1016/j.psychres.2023.115135 Haas M Jakubovski E Kunert K Fremer C Buddensiek N Häckl S ONLINE-TICS: Internet-Delivered Behavioral Treatment for Patients with Chronic Tic Disorders 2022 J Clin Med 250 Haas M, Jakubovski E, Kunert K, Fremer C, Buddensiek N, Häckl S, et al. ONLINE-TICS: Internet-Delivered Behavioral Treatment for Patients with Chronic Tic Disorders. J Clin Med. 2022;11:250. DOI: 10.3390/jcm11010250 https://doi.org/10.3390/jcm11010250 Mortensen LQ Andresen K Burcharth J Pommergaard HC Rosenberg J Matching Cases and Controls Using SAS® Software 2019 Front Big Data Mortensen LQ, Andresen K, Burcharth J, Pommergaard HC, Rosenberg J. Matching Cases and Controls Using SAS® Software. Front Big Data. 2019;2. DOI: 10.3389/fdata.2019.00004 https://doi.org/10.3389/fdata.2019.00004 0 0 0 0