Nineteen RCTs were included in this review (total n=1,513). Overall study quality was low (16 studies scored 1 on the Jadad scale). None of the trials reported allocation concealment. Outcome assessors were blinded in only four studies.
Quantitative meta-analysis based on nine trials (n=704) found a small to moderate statistically significant reduction in levels of pain/anxiety in music therapy patients over controls (SMD -0.35, 95% CI -0.55 to -0.14). Analyses of pain (five trials) and anxiety (five trials) separately found significant reductions in favour of music therapy, although significant statistical heterogeneity was found for the anxiety analysis (I2=52.4%).
Subgroup analyses indicated that the effect was significant for passive but not active music therapy and only where music therapy was compared with standard care (not placebo or an active control). Subgroup analysis by type of outcome measure (self-report or observed) indicated that only observed measures showed a significant benefit of music therapy. Music therapy was more effective when combined with other modalities rather than given in isolation.
There was no significant statistical heterogeneity in any of the main analyses and no evidence of publication bias. Further sensitivity analyses related to type of music and therapist effects were reported.