Even when relatively complex and nuanced clinical outcomes are captured during the conduct of a clinical trial, common analysis strategies often fail to capture important differences in outcomes that are of importance to patients and their families. This may be especially true in the evaluation of treatments for acute ischemic stroke (AIS); the outcome is commonly measured using the 7-level, well-accepted modified Rankin Scale (mRS) while the most common method of analysis is based on a dichotomization of the endpoint. This approach discards substantial, clinically important differences in outcomes. Further, non-homogeneous treatment effects (e.g., benefit in one end of the clinical spectrum and harm in the other may be obscured or difficult to quantify in a valid manner using this approach).
An analysis method that more accurately captures the value of outcomes to patients and their families is to apply a utility weight to each level of the mRS, with the utility representing the relative value of the outcome from a patient-centered perspective. With this approach, the goal of treatment is to maximize the expected utility-weighted mRS (uw-mRS). While precise utility weights for each specific patient are generally unavailable, population-based utility weights have been derived from both patient and caregiver perspectives. These utility weights are very similar, allowing the creation of a valid uw-mRS for use in the analysis of clinical trial (Chaisinanunkul et al. Stroke 2015;46:2238-43). In contrast, the common dichotomization of the mRS is equivalent to a weighting of (1, 1, 1, 0, 0, 0, 0), a weighting that makes little clinical sense.
The Diffusion-weighted Imaging or Computerized Tomography Perfusion Assessment with Clinical Mismatch in the Triage of Wake up and Late Presenting Strokes Undergoing Neurointervention with Trevo (DAWN) clinical trial (Jovin TG et al. International Journal of Stoke 2017 DOI 10.1177/1747493017710341) is a pivotal Bayesian, adaptive, enrichment design trial with the goal of determining the effect of a clot retrieval strategy on the outcomes of patients with AIS with the primary analysis based on the uw-mRS. Eligible patients, outside the time window for treatment with tPA undergo imaging to identify those with a mismatch between their clinical findings and core infarct size, and are then randomized to standard medical care versus clot retrieval. The primary outcome is the uw-mRS at 90 days. The Bayesian, adaptive enrichment trial design included: (1) frequent interim analyses with stopping rules for futility or success based on the predictive probability of success; and (2) prespecified rules for enrichment of the population based on core infarct size when that would substantially increase the probability of trial success. The trial was stopped, based on the prespecified stopping rule for predicted success at the first interim analysis at which that option was available. Full patient follow up confirmed treatment benefit, both based on the analysis of the uw-mRS and on the dichotomized analysis of the mRS.