COSMOS: Interrater and Intrarater Reliability Study of a Novel Outcome Measure.
Journal Articles
Overview
Research
Identity
Additional Document Info
View All
Overview
abstract
The vast majority of patients with minor stroke achieve what are considered good or excellent outcomes on the modified Rankin Scale (0-1/0-2), yet many are dissatisfied with their outcomes. There is a need for a functional outcome measure tailored for minor stroke that better reflects the spectrum of clinical outcomes within this population. We developed the Canadian Outcome Scale for Minor Stroke (COSMOS) and performed an interrater and intrarater reliability study. COSMOS is a 7-point scale ranging from 0 (no symptoms) to 6 (loss of independence for an instrumental or basic activity of daily living or worse), which accounts for performance limitations and losses of a person's hobbies or passions and of their employment, educational, service, or caregiving pursuits, besides just activities of daily living. One hundred test case vignettes were developed. Stroke physicians, fellows, and research nurses/staff were invited to review training materials and provide the COSMOS grade for 20 cases representing all COSMOS grades (0-6). After a minimum 2 weeks' wash-out period, participants were asked to grade the same 20 cases again. Interrater and intrarater agreement were assessed using Cohen κ, weighted κ, percentage agreement, and intraclass correlation coefficient. Among 33 participants (18 attending physicians, 9 stroke fellows, and 6 research staff/nurses; median 12.5 years of experience), COSMOS had substantial interrater reliability (80.5% agreement [95% CI, 75.7%-85.3%]; Cohen κ, 0.77 [95% CI, 0.72-0.84]) and almost-perfect intrarater reliability overall (87.1% agreement [95% CI, 84.4%-89.7%]; Cohen κ, 0.85 [95% CI, 0.82-0.88]); weighted κ showed almost perfect agreement for both interrater (0.88 [95% CI, 0.85-0.92]) and intrarater reliability (0.92 [95% CI, 0.90-0.94]). The overall chance-adjusted simultaneous intrarater/interrater agreement using intraclass correlation coefficient was 0.95 (95% CI, 0.94-0.97). Results were similar with substantial to almost-perfect agreement when considering key subgroups based on position (attendings, fellows, research nurses/staff) and years of experience. In conclusion, the newly proposed COSMOS scale demonstrated substantial interrater and intrarater reliability. The scale merits further study in cohort studies and clinical trials of minor stroke.
