Effect of home exercise programs in patients with non-tears rotator cuff-related shoulder pain: a systematic review and meta-analysis.

Only 12 trials with 475 patients in total were included, because the inclusion criteria were strict, so the pooled estimates rest on small numbers. The comparison was pinned to 12 weeks. Seven of the 12 trials had no follow-up after the home exercise period, and the authors say the lack of long-term data is a central limitation. The control groups were not consistent. Some patients got supervised exercise alone, others got exercise plus manual therapy, massage, taping, acupuncture or electrotherapy, and that mix could itself explain some of the spread in results. The home programmes were not standardised either: different exercises, different doses and only five of the 12 trials progressed the dose over time. No trial could blind patients, and none could blind the physiotherapists, so expectations could have shaped the self-reported pain and disability scores. Six trials were rated at high risk of performance bias for exactly this reason. Adherence data were too sparse and too inconsistent to analyse, which the authors say puts the validity of the estimated treatment effects in doubt. The trials were mostly small, and the authors themselves say the results should be interpreted with caution because of the limited strength of the available evidence.

The authors state that they have no conflict of interest with any financial organisation regarding the material discussed. The work was partially supported by the Chair of Economic Model Transformation of the Department of Finance, Economy, and Public Administration of the Valencian Government. The review was registered on PROSPERO in advance and had ethics committee authorisation from Miguel Hernández University of Elche.

Do not read this as proof that supervision is unnecessary or that home and supervised exercise are equivalent. The review found no significant difference, which is not the same as establishing equivalence, and most comparisons rested on three or four small trials. One in four included trials was at high risk of bias, patients could not be blinded in any of them, and heterogeneity was substantial for pain and function.