Measuring and Analysis of Outcome in Stroke Trials

Hong, Keun Sik

Abstract

Stroke trials are broadly categorized into acute stroke trials and secondary stroke prevention trials. In acute stroke trials, National Institute of Health Stroke Scale is the most widely employed neurological scale measuring stroke severity. Modified Rankin scale and Glasgow Outcome Scale are global functional scales, while Barthel Index is a measure of activity of daily living. To analyze therapeutic efficacy, measured outcomes are usually dichotomized into "good" or "bad" according to arbitrary criteria set based on expert consensus. While the dichotomized analysis allows physicians to easily interpret clinical trials' findings, it has weaknesses of cut-off point dilemma, and a great chance of missing clinically meaningful but modest therapeutic efficacy. Shift analysis, which incorporates therapeutic effects over the entire range of clinical outcomes, is recognized for greater sensitivity in detecting treatment efficacy, but it is not easy to clinically interpret trial findings with shift analysis. In the field of stroke clinical research, recent studies have introduced a metric of disability-adjusted life years (DALY) lost which has been widely employed in population level data analysis to estimate global and regional burden of diseases. DALY lost metric indicating healthy life years lost due to disability and premature death has several advantages: a continuous scale measured with a more powerful statistical method; a common metric of life years lost that enables to compare stroke with other health conditions; an easier-to-understand metric to laymen. In secondary stroke prevention trials, most common primary outcome measures are recurrent stroke and composite of major vascular events. Composite endpoint analysis can index the overall therapeutic impact over polyvascular beds as well as cerebrovascular bed, and capture more vascular events to reduce sample size requirements. However, it has a widely-known limitation of equal weighting of all endpoint components, irrespective of their different impact on patients' lives. A recent study introduced a new weighting strategy of DALY lost analysis which indicates, compared to a nonfatal myocardial infarction, a 1.48-fold greater DALY lost with a nonfatal stroke, and a 2.25-fold greater loss of DALY with vascular death.