Sensitivity analyses indicate results were robust in studies at risk of selection and detection biases in comparison to studies at low risk of bias. There was high‐quality evidence that stretch did not have clinically important effects on joint mobility in people with or without neurological conditions if performed for less than seven months. Nine studies involving 635 participants reported 41 adverse events including numbness, pain, Raynauds’ phenomenon, venous thrombosis, need for manipulation under anaesthesia, wound infections, haematoma, flexion deficits and swelling but it was not possible to statistically analyse these data. The short‐term effect of stretch on activity limitations (SMD 0.1 95% CI ‐0.2 to 0.3 5 studies with 356 participants) and participation restrictions were uncertain (SMD ‐0.2 95% CI ‐0.6 to 0.1 2 studies with 192 participants). In people with non‐neurological conditions, there was high‐quality evidence that stretch did not have clinically important short‐term effects on pain (SMD ‐0.2, 95% CI ‐0.4 to 0.1 7 studies with 422 participants) and moderate‐quality evidence that stretch did not have clinically important short‐term effects on quality of life (SMD 0.3, 95% CI ‐0.1 to 0.7 2 studies with 97 participants).
Five studies involving 145 participants reported eight adverse events including skin breakdown, bruising, blisters and pain but it was not possible to statistically analyse these data. No trials examined the short‐term effects of stretch on quality of life or participation restrictions in people with neurological conditions. In people with neurological conditions, it was uncertain whether stretch had clinically important short‐term effects on pain (SMD 0.2 95% CI ‐0.1 to 0.5 5 studies with 174 participants) or activity limitations (SMD 0.2 95% CI ‐0.1 to 0.5 8 studies with 247 participants). There was high‐quality evidence that stretch did not have clinically important short‐term effects on joint mobility in people with neurological conditions (MD 2° 95% CI 0° to 3° 26 studies with 699 participants) or non‐neurological conditions (SMD 0.2, 95% CI 0 to 0.3, 19 studies with 925 participants). The effect of these biases were unlikely to be important, given that there was little benefit with treatment.
However, most studies were at low risk of detection bias for objective outcomes including range of motion, and the majority of studies were free from attrition and selective reporting biases. Just over half the studies (51%) were at low risk of selection bias all studies were at risk of detection bias for self reported outcomes such as pain and at risk of performance bias due to difficulty of blinding the intervention. No study performed stretch for more than seven months. Forty‐nine studies with 2135 participants met the inclusion criteria.