Digital Mirror Therapy and Action Observation Therapy for Chronic Stroke: A Pilot Randomized Controlled Trial.

Nineteen patients split across three arms gives roughly 6 to 7 per group, far too few to detect anything but a very large difference between approaches. Only the assessor was blinded; participants and therapists knew which group they were in, which can influence effort and delivery. The control group received active conventional occupational therapy, so the comparison is between two active treatments, not against no treatment. Three patients dropped out because of confirmed COVID-19 cases or a hospital shutdown, and missing data were handled by carrying the last observation forward. The total therapy dose was 900 minutes, which the authors note is lower than the 1,800 minutes in some prior mirror therapy studies and the 600 to 1,620 minutes in prior action observation trials. The number of repetitions per task category was not reported, so the true dose delivered to each patient is unclear. Baseline characteristics varied across stroke type, lesion side, and lesion location, adding variability in an already small sample.

Do not read the within-group improvements in the digital therapy groups as proof they outperform conventional occupational therapy. The between-group comparison found no significant differences on any outcome, and with only 6 to 7 patients per group the study was not powered to detect a real difference. Equally, do not read the null between-group result as proof the digital methods are useless. All three groups exceeded the minimal clinically important difference for arm movement and function, so the null reflects a lack of statistical power as much as a lack of effect.