Blinding of outcome assessors and its association with outcome in a randomized open-label stroke trial. Issue 5 (June 2023)

• Record Type: Journal Article
• Authors: van der Ende, Nadinda AM; Roozenbeek, Bob; Broderick, Joseph P; Khatri, Pooja; Lingsma, Hester F; Dippel, Diederik WJ
• Is Part Of: International journal of stroke. Volume 18:Issue 5(2023)
• Page Start: 562
• Page End: 568
• Publication Date: 2023-06
• Subjects: Clinical trial -- ischemic stroke -- acute stroke therapy -- intervention -- methodology -- neurology; 616.8005
• DOI: 10.1177/17474930221131706
• Languages: English
• ISSNs: 1747-4930

Abstract

Background: It is challenging for outcome assessors to remain blinded during outcome assessment in trials with prospective randomized open blinded endpoint (PROBE) design. If assessors are able to guess the correct treatment allocation more often than expected based on chance, the assessors may have been not properly blinded. Aims: We aimed to assess blinding of outcome assessors in a stroke trial with PROBE design and its association with outcome. Methods: We used data of the Interventional Management of Stroke (IMS) III trial. The modified Rankin Scale (mRS) at 90 days was assessed by local assessors who were unaware of treatment allocation. To assess success of blinding, each assessor was asked to guess the patient's treatment allocation. We assessed whether the percentage of correct guesses was higher than chance (i.e. 50%). The association between correctly guessed treatment allocation and the mRS at 90 days was analyzed with ordinal logistic regression stratified by treatment allocation. We tested for interaction of correctly versus incorrectly guessed treatment allocation with actual treatment allocation on the mRS. Patients with missing data on guessed treatment allocation and patients who died prior to 90-day assessment were excluded. Results: In total, 459 patients were included in this study. The assessors guessed the correct treatment allocation significantly more often than expected (267/459, 58.2%, one-sided p = 0.0003). Correctly guessed treatment allocations were associated with better mRS scores in the intervention group (common odds ratio (cOR): 2.28, 95% confidence interval (CI): 1.50–3.48) and with worse mRS scores in the control group (cOR: 0.47, 95% CI: 0.27–0.83) (pinteraction < 0.001). Conclusions: Assessors may not always be truly blinded for treatment allocation in clinical trials, and their guesses may be associated with outcome. Although causality between the assessors' guess and patient outcome cannot be determined, future trials with subjective outcome should make efforts to ensure blinding and should report their blinding method and the success of blinding like the IMS III trial. Clinical Trial Registration: URL:https://clinicaltrials.gov . Unique identifier: NCT00359424.