Type of study: Statistical re-analysis and “simulation” of data from published studies (though not a meta-analysis)

The finding: Several recent studies suggesting that long-term outcome (of a variety of deficits) in stroke is robustly quantitatively predicted by initial symptom severity, may be significantly flawed.

Extensive statistical argumentation (ambitious readers can follow links to two appendices that explain the propositional, theorem, and simulation-based backing of the authors’ methods) seems to reveal that in situations where there is much greater variability in baseline (post-stroke) than outcome conditions, correlations will be inflated and extreme. Under these conditions, the idea that baseline impairment accounts for up to 90% of the variance and determines proportionate recovery does not hold.

Strength and weaknesses: While pointing out that “Stroke researchers [and clinicians] are used to thinking of recovery as a complex, multi-factorial process,” the authors first provide a succinct yet well-referenced review of the nature and backing of the “proportional recovery rule.” The idea that initial deficit severity is related to long-term outcome is not new, but the finding that most patients (and even rats) can be predicted to recover 70% of lost function, with around 90% of the variance explained by initial deficit severity, is relatively new (or at least newly, and seemingly robustly, supported). The statistical analysis employed here is dense, but the main text does not go beyond correlations and standard deviations. Still, they are related to each other in ways that make these 7 pages pretty slow going. Rigor and readability were not correlated with one another in a statistically or comprehensibility significant way by the authors. There is just enough discussion of the nature of stroke victims as patients and subjects to show that the authors understand what they are studying as more than a mere abstraction. The discussion section summarizes things nicely and reveals the paper to be an attempt at being constructive, rather than a mere takedown.

Relevance: The review and critique of the proportional recovery rule are quite important to those of us working in neurological settings. The proportional recovery rule has been applied to paresis, aphasia, and “may even apply generally across cognitive domains.” The evidence supporting this idea has been published in several prestigious clinical journals in the past three years particularly. Familiarity with this concept and with its possible shortcomings will inform case contemplation as well as conversations with neurologist consultees (who may be unfamiliar with, wedded to, or skeptical of it), patients, and families.