Our protocol stated that we would include non‐randomised studies, due to feasibility issues often raised when researching the environment. Following review by our contact editor, the decision was made to remove 24 non‐randomised CCTs on music. Initially, these studies were included and summarised separately to the 85 included RCTs on music. In general, the CCTs reported similar findings to the RCTs however the quality of the studies was lower. The removal of these studies did not alter the conclusions of the review. Since RCTs are generally feasible in the study of music interventions, and there are so many RCTs exploring the use of music in hospital, it was felt that the CCTs were not providing any further valuable information to the review, which has sought to summarise the best available evidence.

We initially planned to assess heterogeneity through using the I² statistic with a cut‐off point of 50% (i.e. > 50% would be considered heterogeneous). We have since revised this approach (to that specified in the data analysis section of the review) in alignment with revisions to the Cochrane Handbook for Systematic Reviews of Interventions (version 5; Higgins 2008), in which the use of cut‐off points are no longer recommended. Further to the release of the new Handbook (Higgins 2008), we have incorporated a 'Risk of bias' assessment utilising the new 'Risk of bias' tool.

In our exploration of heterogeneity for studies on music, we conducted subgroup analyses based on whether patients were offered a choice of music or given set pieces. These analyses were post‐hoc as they were not planned for in the protocol (although in our protocol we did specify that we would set out to find interventions which provided patient controls over the environment‐ such as access to lighting and ventilation controls). We felt that it would be useful to explore these variations in the way the music interventions were administered since patient choice and providing patients with a sense of control is often advocated as beneficial.