Background: Patient preferences for treatment can pose problems for the conduct of randomised controlled trials: patients with a preference may refuse participation and thereby potentially compromise external validity. Moreover, randomising patients with a preference may affect treatment efficacy and threaten internal validity. Aims: This study compared baseline characteristics and short-term psychological outcomes of patients who selected their treatment and those who agreed to random allocation. Methods: Men participating in the prostate testing for cancer and treatment (ProtecT) study and who were randomised to active monitoring (n = 138) were compared with those who had refused randomisation and selected this management (n = 180). Socio-demographic data were collected at baseline, and anxiety and depression data were collected at baseline and six month follow-up. Socio-demographic characteristics were compared across these two groups in univariable analyses, and then linear regression was used to compare levels of anxiety and depression at follow-up with adjustments for confounders. Results: Participants who selected active monitoring were more affluent (based on occupation details) and had less anxiety at baseline than those who were randomised. There were no differences with respect to age and marital status. Levels of anxiety and depression at six months follow-up were similar across the two groups of men. Conclusions: This study found some differences at baseline between the socio-demographic and psychological status of those randomised and self-selecting treatment, but no psychological differences at short-term follow-up. Further empirical evidence is required to assess whether preferences impact upon the process and outcome of randomised controlled trials.
Contemporary Clinical Trials (2006) 27(5) 413-419 [doi:10.1016/j.cct.2006.04.008]