Twelve studies were included: eight RCTs; one controlled trial; one crossover design; and two studies with no control group (one of which was a single-case study). The number of participants ranged from one to 42. Nine studies were classified as of acceptable methodological quality and three were considered to be of low quality. Follow-up was not reported beyond the end of the treatment.
Three of five studies that targeted function reported a statistically significant improvement over time or compared to control in pain, muscle strength or on a shoulder rating scale; two studies did not report such improvements.
Two studies that targeted activity level reported a statistically significant increase in hand function in the motor training group compared to control, but not for upper extremity function or pinch grip.
The five studies that used a combined programme all reported some improvement in outcomes. These included manual muscle testing, functional skills and daily activities in the motor training group, which was statistically significant in two of the studies.