Background: Outcome in acute stroke trials is often based on short-term mRS, but there are few data from prospective population-based studies on how this measure translates into long-term outcomes. We evaluated the relationship between 1-month mRS and 5-year disability, death, quality of life, and healthcare cost in a population-based cohort study.
Methods: In 3-month survivors of ischaemic stroke in the Oxford Vascular Study (2002-2014), we used logistic regression to determine predictors of 5-year death/disability, including mRS score at 1 month, age, and sex. The analyses were repeated for different subgroups, including thrombectomy-eligible, atrial fibrillation-related, and lacunar strokes. Hospital resource use and institutionalization data up to 31 August 2015 were collected, and mean censor-adjusted costs were reported with 95% CIs from 1000 bootstrap estimates. A general gamma linear model was used with 1-month mRS controlling for age, sex, and comorbidities. 5-year quality-adjusted life expectancies (QALE) generated from survival data and EQ-5D-derived utility scores were stratified by 1-month mRS.
Results: Among 1,425 survivors, mRS score was a strong independent predictor of 5-year death/disability, with a step-change from mRS 2 to 3: adjusted odds ratio for mRS 3-5 vs 0-2: 35.57, 95%CI 17.40-72.71, p<0.0001. This step-change was also seen for 5-year mortality: adjusted hazard ratio for mRS 3-5 vs 0-2: 1.84, 95%CI 1.59-2.14, p<0.0001. Trends were consistent across the stroke subgroups. mRS score was the only independent predictor of 5-year healthcare costs (p<0.0001) aside from age, again with a step-change from mRS 2 to 3: £8,817.67 (95%CI 6,207.98-10,688.86) vs 29,692.73 (95%CI 21,117.18-31,367.53). 5-year QALE dropped incrementally with rising mRS: mRS=0 - 3.47 (95%CI 3.25-3.66); 1 - 2.91 (2.77-3.04); 2 - 2.64(2.46-2.81); 3 - 1.72 (1.52-1.91); 4 - 1.23 (1.03-1.45); 5 - 0.31 (0.02-0.76).
Conclusions: Our results reaffirm the practice in acute stroke trials of using short-term mRS as the primary outcome measure. The step change between mRS 2 and 3 for both death/disability and healthcare costs supports the traditional dichotomous outcome, but the incremental drop in QALE with each mRS grade shows that ordinal analysis is also valid.