Abstract 138: Assessing Disability Outcome in Acute Stroke Clinical Trials: Increased Study Power with Linear (ALDS) Compared with Ordinal (mRS) Scales
Background: The most common primary endpoint in acute stroke trials is the modified Rankin Scale (mRS) assessment of global disability. However, the mRS lacks granularity (assigns patients among only 7 broad levels) and is ordinal (scale levels unknown distances apart), constraining study power. Instruments that evaluate disability along an interval, continuous scale theoretically should increase study power. The Academic Medical Center linear disability scale (ALDS) is one such scale, assigning patients to scores from 0-100.
Methods: Consecutive patients undergoing 3 month final outcome visits in the NIH FAST-MAG phase 3 trial were assessed with the mRS and ALDS. Scale scores were compared using splines and linear regression. Model trial power was computed using 4000 random simulations.
Results: Among 55 patients, mean age was 71.2 (SD±14.2), 67% were male, and entry NIHSS was 10.7 (SD±9.5). At 90d, the median mRS was 3 (IQR 1 - 4, range 0-5, lower better) and median ALDS score 78.8 (IQR 3.3-100, range 0-100, higher better). The mRS and ALDS logit score were highly correlated, r = 0.94. ALDS logit score also correlated strongly with 90d Barthel Index(r=0.92) and 90d NIHSS (r=0.87), as well as modestly with entry NIHSS (r = 0.38). The ALDS showed greater discriminant validity than the mRS, with a mean of only 1.9 (SD±3.2) patients sharing each ALDS logit value versus a mean of 8.0 (SD±3.6) patients sharing each mRS score value, p < 0.001. Over a broad range of treatment effect magnitudes, trial power was 25-30% higher when the ALDS rather than the mRS was used as the primary trial endpoint (Figure).
Conclusion: Among patients enrolled in an acute neuroprotective stroke trial, the ALDS, a linear disability scale derived using item response theory, showed strong convergent validity and superior discrimination characteristics compared with the modified Rankin Scale, and considerably increased trial power to detect clinically meaningful treatment benefits.
Author Disclosures: N.N. Chaisinanunkul: None. S. Starkman: None. J.A. Gornbein: None. S.A. Hamilton: None. F. Chatfield: None. R.A. Conwit: None. J.L. Saver: None.
- © 2014 by American Heart Association, Inc.