A Randomized Control Trial of a Virtually Delivered Program for Increasing Upper Limb Activity After Stroke.

Only 66 people were analysed, and the trial was powered for a large effect. The main difference was 368 counts a day with a confidence interval starting at 6, so the result is barely distinguishable from no effect. Everyone could already move the affected arm, with at least 30 degrees of shoulder movement and enough finger and thumb movement to grip, so this says nothing about patients with more severe impairment. The comparison was against a waiting list, and those participants were still in contact with study staff, so the effect of that attention cannot be separated from the treatment. The programme is three things at once (home exercise, device feedback and therapist sessions), so this trial cannot show which component produced the change. The two-month follow-up and the analysis of everyone treated had no control group; the authors state those results do not account for natural recovery or usual care. Exercise logs were available for only 22 of the 33 treated participants (67%), partly because therapists sometimes forgot to record them, so adherence rests on two-thirds of the group. Participants covered a wide range of impairment and activity levels and the study was too small to show who benefited. Several outcomes were tested with no adjustment for multiple comparisons. Some participants may have kept exercising after the programme finished, which the authors say could explain part of the difference at four weeks.

One author, Dr Menon, holds a stake in commercialising the sensor used in the trial and may gain financially through BioInteractive Technologies, the company that makes the wrist-worn TENZR device. The other authors declare no competing interests. Funding came from the Canadian Institutes of Health Research, the Canada Research Chairs Program, a CIHR Fellowship and the Canada Brain Research Fund; the authors state the funders had no role in study design, data collection, analysis, interpretation or writing.

Do not read the 45% increase in daily arm activity reported for everyone who received the programme as proof that it works. That analysis pooled all participants, including those who had been on the waiting list, and had no control group. The authors say it does not account for natural recovery or usual care. The controlled result is the 368-count difference at four weeks, which only just reached statistical significance. Do not credit the sensor alone either, since the exercises and the therapist sessions were delivered at the same time.