4.1. Limitations
There are several limitations to our analysis. First, the nature of this intervention makes a double-blinding design impossible. Secondly, the sample size in each study is relatively small. Thirdly, the quality of some studies is relatively low. Some earlier studies did not detail the methods to generate the random numbers and/or declare the concealment of allocation, which got quality scores 2 or 3. Fourthly, substantial heterogeneity was presented. Although the major source of heterogeneity was detected through subgroup analysis, uncontrolled or unmeasured factors potentially produce bias. Fifthly, the longest follow- up duration was no more than 4 weeks which left it unclear that whether the efficacy of music can maintain or even be better after a longer follow-up. Sixthly, although SMD was used to pool the results, the difference between various subjective methods used in studies could still induce a significant heterogeneity and bias. Finally, various objec- tive indices were presented by polysomnography, but we only assessed the sleep efficiency which may miss some other useful information.
5. Conclusion
Music appears to be effective in treating acute and chronic sleep disorders. It is low cost and safe, and could be used to improve sleep quality in various populations with different ages and culture backgrounds, in hospital or in community. Our study also gives an indication that music shows a cumulative dose effect for chronic sleep disorders. A follow-up duration more than three weeks is necessary for assessing the efficacy of music, which have an implication for the design of trial evaluating the efficacy of music therapy for chronic sleep disorders.
Conflict of interest: None declared. Funding: None.
Ethical approval: None.