Table 3 Examples of useful external pilot studies

FEMUR study [29] ⁺

Testing if it would be possible to randomise by primary care groups (in the 1990s) as a precursor to a trial to see if a whole systems approach could reduce falls in older people

Showed that it was unlikely to be possible to achieve an effect given the intervention proposed within primary care groups and the main trial did not proceed

UK BEAM [30]

Testing a cluster randomisation design with back pain patients recruited from general practices (the clusters) after randomisation

Changes needed to be made in the main trial design (moving from cluster randomisation to individual randomisation) prior to proceeding to the definitive trial because of a phenomenon in cluster randomised trials now well-known: identification and recruitment bias

COMQUOL [31]

Testing feasibility of randomising secure mental health wards (no large definitive trials in this area)

Confirmed feasibility of this approach and funding was sought for main trial, but was never secured

STarT MSK [32]

Testing recruitment rate and GP fidelity to an in intervention testing a tool to manage musculoskeletal pain

Neither recruitment rate nor GP fidelity were met, but retention was good and when the tool was used treatment was mostly in line with the tool. Based on GP feedback, the trial processes were modified prior to proceeding to the main trial