Ordinal vs dichotomous analyses of modified Rankin Scale, 5-year outcome, and cost of stroke

Abstract

ObjectiveTo compare how 3 common representations (ordinal vs dichotomized as 0–1/2–6 or 0–2/3–6) of the modified Rankin Scale (mRS)—a commonly used trial outcome measure—relate to long-term outcomes, and quantify trial ineligibility rates based on premorbid mRS.MethodsIn consecutive patients with ischemic stroke in a population-based, prospective, cohort study (Oxford Vascular Study; 2002–2014), we related 3-month mRS to 1-year and 5-year disability and death (logistic regressions), and health/social care costs (generalized linear model), adjusted for age/sex, and compared goodness-of-fit values (C statistic, mean absolute error). We also calculated the proportion of patients in whom premorbid mRS score >1 or >2 would result in exclusion from trials using dichotomous analysis.ResultsAmong 1,607 patients, the ordinal mRS was more strongly related to 5-year mortality than both the 0–1/2–6 and 0–2/3–6 dichotomies (all p < 0.0001). Results were similar for 5-year disability, and 5-year care costs were also best captured by the ordinal model (change in mean absolute error vs age/sex: −$3,059 for ordinal, −$2,805 for 0–2/3–6, −$1,647 for 0–1/2–6). Two hundred forty-four (17.1%) 3-month survivors had premorbid mRS score >2 and 434 (30.5%) had mRS score >1; both proportions increased with female sex, socioeconomic deprivation, and age (all p < 0.0001).ConclusionThe ordinal form of the 3-month mRS relates better to long-term outcomes and costs in survivors of ischemic stroke than either dichotomy. This finding favors using ordinal approaches in trials analyzing the mRS. Exclusion of patients with higher premorbid disability by use of dichotomous primary outcomes will also result in unrepresentative samples.

Ticagrelor Added to Aspirin in Acute Ischemic Stroke or Transient Ischemic Attack in Prevention of Disabling Stroke

“…1 Disability 3 months after stroke onset is an independent predictor of future death. 1 Disability after a transient ischemic attack (TIA) or a minor ischemic stroke may be owing to the index stroke, comorbidities, or a subsequent stroke. 2,3 Hence, tackling disability by reducing recurrent ischemic stroke is one major objective of any short-term treatment.…”

Which version of the modified Rankin Scale should we use for stroke trials?

“…The slope at any point on this curve represents the positive likelihood ratio. In this paper's 6 5-year mortality model, the AUC is explained as the likelihood that a randomly sampled stroke patient who died has a higher mRS score than a randomly sampled stroke case who survived. In predicting 5-year mortality, the base model with just age and sex has an AUC of 0.80.…”

Disabling stroke in persons already with a disability

“…[35][36][37] Given the simple dichotomized mRS endpoints that are routinely used to quantify favorable outcomes in stroke trials (which tend to define favorable outcome as an mRS score of <2 for thrombolysis or 3 for thrombectomy), enrolling patients with preexisting disabilities introduces practical difficulties in adjusting for different levels of premorbid disability when quantifying the utility of an intervention, especially when the good outcome is defined as functional independence (mRS score ≤2). 36,38 The road ahead…”

“…4 Apart from satisfying the normative demands of nondiscrimination, enrolling patients with disabilities in trials can further help to ensure more representative and generalizable samples, especially as increasing longevity drives a higher proportion of patients with stroke with increased premorbid baseline mRS score over time. 38 Outcome measures that capture accumulated functional dependence (such as change in mRS score from before stroke to after stroke, the modified Rankin shift, 34 or the weighted mRS 33 ) rather than the mere binary state transfer from nondependent to dependent could become standard outcomes for future studies. These outcome measures may allow more widespread inclusion of persons with disability without compromising data construct validity.…”

“…These outcome measures may allow more widespread inclusion of persons with disability without compromising data construct validity. 38 With respect to thrombolysis, data from stroke registries suggest that those with significant disability account for <10% of treated cases, despite accounting for a much larger percentage of patients who present with acute stroke. 39,40 This disparity demands careful evaluation.…”