Predicting Individual Treatment Response to rTMS for Motor Recovery After Stroke: A Review and the CanStim Perspective.

The search used only one database, PubMed, and only articles in English, French, or German. Many included studies were small or uncontrolled: 14/26 had 30 or fewer participants and 16/26 had no sham control. Only 4 of the 11 sham-controlled trials were randomized and double-blind. Study designs, rTMS protocols, session numbers, outcome measures, and assessment timing varied widely. Most studies examined chronic stroke and upper limb function, so acute, subacute, and lower limb recovery are less covered. The two largest retrospective studies came from the same research group and accounted for nearly 75% of the total patient population. The lesion-location finding did not adjust for lesion size, so smaller lesions in subcortical patients could explain part of the effect. Only four studies measured restriction in activities and participation, although that is often more relevant to daily functioning than impairment scores. No predictor was validated for clinical use, and the multivariate model that explained 82% of variance has not been validated for long-term recovery.

Do not use these factors to decide who receives rTMS. This review identified candidate predictors from heterogeneous studies, and it did not validate a test or show that rTMS works.