Introduction and aims Recruiting patients to palliative care randomised controlled trials (RCTs) is challenging. This paper describes and analyses recruitment trajectories for patients with malignant and non-malignant disease to a palliative care RCT, outlining lessons learned.

Methods Analysis of descriptive recruitment statistics (patient identification and response rates) to a Phase II pilot pragmatic single-blind fast track RCT, and subsequent Phase III RCT, of a breathlessness intervention service for advanced disease. Phase II piloted COPD patients only, whereas Phase III RCT incorporated two sub-protocols: for patients with malignant or non-malignant disease. Documentary analysis of: recruitment activity log, Trial Management and Advisory Group minutes and fieldnotes.

Results Recruitment targets for patients with non-malignant disease were achieved. The pathway to recruitment was through referral to the service therefore referral rates impacted on recruitment alongside response rates. Recruitment of cancer patients was considerably slower despite concerted efforts to increase referrals by raising the service profile. Referrals only improved for the latter when a researcher attended clinics, supporting clinical staff in patient identification. Predictably, response rates remained lower than for non-malignant disease patients.

Conclusion Recruitment was partly referral-driven, therefore gate-keeping did not explain the differences. Clinical inter-professional relationships consolidated in Phases 0-II drove early referrals of non-malignant disease patients. Local palliative care services were available for cancer patients. Consideration of the natural history and context of a service is therefore important when predicting recruitment. Pilot trials are informative, but should include qualitative elements and all disease groups. Placing researchers in relevant clinical settings is helpful.