Eight RCTs (n=343) met the inclusion criteria. Seven RCTs met more than five of the PEDro criteria.
Immediate effects (three RCTs): Aromatherapy plus acupuncture showed a significantly better effect than acupuncture alone (OR -1.94, 95% CI -2.79 to -1.09; one RCT). Slow-stroke back massage showed better immediate effect than no massage (OR -2.12, 95% CI -2.6 to -1.63; one RCT). Intramuscular electric stimulation was significantly more effective compared to Hemisling (OR -1.35, 95% CI -2.03 to -0.68; one RCT).
Short-term effects (five RCTs): No significant differences were demonstrated between BoNT/A and placebo (three RCTs); heterogeneity between the studies was high (I2=83%). There was no significant difference between BoNT/A and triamcinolone acetonide.
Long-term effects (five RCTs): Intramuscular electric stimulation showed significantly better long-term benefits compared with Hemisling (OR 0.79, 95% CI -1.37 to 0.20; one RCT); BoNT/A showed no significant benefits when compared to placebo (three studies) or triamcinolone acetonide (one RCT) or when triamcinolone acetonide was compared to placebo (one RCT).
Overall, there was no significant difference in the incidence of adverse events between interventions and control (OR 1.45, 95% CI 0.29 to 7.38; three RCTs). No relationship was identified between subluxation (one RCT) or spasticity (two RCTs) and hemiplegic shoulder pain. Passive shoulder external rotation increases with decreasing hemiplegic shoulder pain.
Results of sensitivity analyses were reported.