Service development or research: make sure you know which game you are playing.

Background Benign headache is a common neurological symptom in both primary care (5% of consultations) and secondary care (20-30% of neurology referrals). Question Does a simple patient information sheet improve headache severity, frequency and impact as measured by the HIT-6 score? Methods A randomised controlled trial (RCT) in an NHS setting. Results The trial terminated early due to incomplete recruitment over a two year period. The information sheet had strong face validity with patients and patient groups but no meaningful conclusions can be drawn because of recruitment problems. One hundred and sixty eight patients were randomised from a projected sample size of 220 and only 62 fully completed the trial. Conclusions We analysed the reasons for trial failure, and they fall into a number of distinct groups: (1) major service configuration/re-orientation occurred concurrently as the trial was run with 'Choose and Book' and the '18 week targets' being introduced; (2) our aim was a classical evidence-based superiority trial, whereas the PCT aim was demand management; (3) there was a funding and resource shortfall. Our experience generates discussion about appropriate level of evaluation required for service development.