The modified Rankin scale (mRS) is a standard measure of outcome used in stroke research, but interrater variability exists and unlike imaging markers in which central adjudication is often used, central adjudication of the mRS has not been validated. This can be problematic for studies in which local adjudication of the mRS may lead to observer bias. Video based central adjudication can help with the problem of observer bias and a previous study showed good agreement between central and local evaluation of mRS. In order to better address this, a secondary analysis of the REVASCAT study was completed to look at the feasibility of central mRS adjudication and to compare phone versus video-based central adjudication against the outcome standard of 24-hour infarct volume on neuroimaging.
Patients were assigned a 90-day mRS both locally using a structured interview in a face-to-face visit by an mRS-certified neurologist not involved in the patient’s management as well as centrally. The central adjudication of the mRS was done initially by an external mRS-certified nurse by telephone call using a structured interview, but due to discordance of local and central scores, was changed to a video adjudication method by a single external, mRS-certified neurologist. Significantly, the authors found:
- Central adjudication, including video adjudication, was feasible; patient information remained secure, video protocol was brief, and the laptop used for video-taping was portable.
- The inter-rater agreement between local and central assessors was higher when the central assessment was done by video, with 62.5% concordance when central assessment was done by phone compared to 86.8% concordance when central assessment was done by video. In a subset of 55 patients who received both modalities of central adjudication, the agreement was 67.3%, with a wider range of assigned mRS by the phone assessor compared to the video assessor in comparison to the local adjudicator.
The majority of local face-to-face interviews were rated by a central video evaluator as reliable (71.7%), but 10.4% were rated as poor and 17.9% as acceptable. The higher the quality of face-to-face interview, the better the agreement between video based and local assessment.
What does this mean for clinical studies? This does support the feasibility of central adjudication of the mRS using video-taped recordings, which can have a significant impact on the way stroke research is done. Neuroimaging is easily assessed centrally by blinded investigators, but imaging markers do not necessarily correlate with quality of life, which is a much more clinically significant endpoint for both patients and physicians.
