Addressing Shoulder Weakness in Individuals With Rotator Cuff-Related Shoulder Pain: A Systematic Review With Meta-analysis.

Every result is a change within a treatment group from before to after the programme. There is no comparison between groups, so improvement could reflect time passing, natural recovery, or simply being measured twice, and the review cannot separate the treatment's effect from those. Thirteen of the randomised trials (43%) were at high risk of bias, mainly because the people measuring strength knew which group each patient was in, and because it was not reported how patients were assigned. When those studies were removed, several effects weakened. The studies measured strength in different positions, with different tools, different numbers of attempts and different ways of reporting results, so combining them produced substantial disagreement between studies in several analyses. Only 28 of the 33 included studies could be pooled, and the non-randomised studies included were all rated poor quality. Several headline results rest on very little data. The NMES findings come from 1 or 2 arms, and the large eccentric external rotation result comes from 5 arms with very high disagreement between them. The authors themselves say these should be interpreted cautiously. Thirteen studies did not state whether people with rotator cuff tears were included, so the number of participants with a full-thickness tear is likely underestimated, even though the authors tried to exclude this group. No included study reported the race or ethnicity of participants. No study measured muscle activation or muscle structure, so the review cannot say why strength changed.

Do not read the large effects for NMES or eccentric training as proof that these are the best exercises to choose. The NMES estimates came from only 1 or 2 study arms, the eccentric external rotation result came from 5 arms that disagreed strongly with each other, and the review pooled before-and-after changes within each group rather than comparing interventions against one another. The authors state plainly that their results cannot support one intervention being superior to another.