By Randolph J. Nudo
Prediction of who will recover after stroke has been a perennial focus for both researchers and clinicians in the field of neurorehabilitation. The prospects of applying a population-based model to predict outcome in individual patients might ultimately allow more focused approaches to stroke rehabilitation and foster a better distribution of precious health care resources. Aside from anatomical biomarkers, such as the integrity of the corticospinal tract, recent attention has focused on the proportional recovery rule, formally proposed in this journal more than 10 years ago by Prabhakaran et al,1 who described a surprisingly linear relationship between Fugl-Meyer Assessment upper extremity scores obtained within 3 days after stroke and those obtained at 3 months poststroke, illustrating the general principle of spontaneous recovery with a level of predictability not previously appreciated. This relationship appears to hold for most individuals (so-called “fitters” or “recoverers”), but a subset of individuals (so-called “non-fitters” or “non-recoverers”) fall off the linear regression line. First applied to upper limb motor impairment, the proportional recovery rule has been examined in a variety of motor and nonmotor impairments, and results have generally been in agreement with the initial linear relationship. Recent controversy surrounding the proportional recovery rule has been based on statistical factors such as mathematical coupling and nonlinearity of outcome scales, questioning not only the accuracy but also the underlying validity of this predictive population-based model. Two articles in the current issue of Neurorehabilitation and Neural Repair highlight some of the emerging views and suggestions for future research regarding this model. The first article by Senesh and Reinkensmeyer examines the reasons why “non-fitters” do not recover according to the proportional recovery algorithm. They argue that the local slope of the linear regression reflects the difficulty of test item scores related to arm and hand movement at follow-up, consistent with the view that non-fitters lack sufficient corticospinal tract. They suggest that at least some non-fitters may have a heightened response to intensive movement training and should be targeted early after stroke for such rehabilitative training. In the second article by Kundert et al, the statistical validity of the proportional recovery rule is examined in the context of recent criticisms regarding its underlying assumptions. Despite 2 recent articles critical of statistical relationships of baseline impairment scores to follow-up scores, especially when used for patient-level predictions, Kundert et al contend that the systematic non-artifactual relationship between initial impairment and motor recovery provides a valid statistical and biologically meaningful model, and that future studies of proportional recovery should use more sophisticated analysis techniques and rigorous methods to assess validity, including comparisons to alternative models.
|1.||Prabhakaran, S, Zarahn, E, Riley, C, et al. Inter-individual variability in the capacity for motor recovery after ischemic stroke. Neurorehabil Neural Repair. 2008;22:64–71. doi:10.1177/1545968307305302
Google Scholar | SAGE Journals | ISI