Article type
Year
Abstract
Background: Randomised controlled trials (RCTs) are based on the assumption that the effects of interventions are constant across all participants. This may not always be true in complex interventions which are known to depend on provider’s expertise and experience and may also be vulnerable to context variables (such as whether therapy is delivered within a trial). Music therapy, with music experiences and a therapeutic relationship as the elusive key components, is a complex intervention with an emerging evidence base.
Objectives: To investigate time trends in effect over the recruitment period in music therapy RCTs, which might indicate an impact of delivering therapy within a trial.
Methods: We included recent (2006-present), methodologically rigorous RCTs on music therapy versus standard care for people with mental disorders (schizophrenia or depression) that used a parallel single-blind design to assess changes in a continuous scale of mental health (symptoms of schizophrenia or depression). Music therapy had to be delivered by experienced therapists. Individual patient data were examined graphically and statistically.
Results: Three trials were included. In all trials, the comparative effects of the interventions (music therapy versus standard care) were not constant, but shifted over time. One trial suggested a? reverse effect in the first ten participants than in later participants. The other trials suggested a smaller effect in those first participants than in later ones. Whereas the trials overall suggested superiority of music therapy, this was not the case for the first participants recruited.
Conclusions: For some complex interventions, it cannot be assumed that effects remain constant over the recruitment period of a trial. Delivering therapy within a trial may be an unsettling experience for experienced therapists, especially at the beginning of a trial. If replicated, our findings will have implications for the analysis and reporting of RCTs on complex interventions.
Objectives: To investigate time trends in effect over the recruitment period in music therapy RCTs, which might indicate an impact of delivering therapy within a trial.
Methods: We included recent (2006-present), methodologically rigorous RCTs on music therapy versus standard care for people with mental disorders (schizophrenia or depression) that used a parallel single-blind design to assess changes in a continuous scale of mental health (symptoms of schizophrenia or depression). Music therapy had to be delivered by experienced therapists. Individual patient data were examined graphically and statistically.
Results: Three trials were included. In all trials, the comparative effects of the interventions (music therapy versus standard care) were not constant, but shifted over time. One trial suggested a? reverse effect in the first ten participants than in later participants. The other trials suggested a smaller effect in those first participants than in later ones. Whereas the trials overall suggested superiority of music therapy, this was not the case for the first participants recruited.
Conclusions: For some complex interventions, it cannot be assumed that effects remain constant over the recruitment period of a trial. Delivering therapy within a trial may be an unsettling experience for experienced therapists, especially at the beginning of a trial. If replicated, our findings will have implications for the analysis and reporting of RCTs on complex interventions.