Objectives: The objectives were to determine the most appropriate design for a definitive study, the most appropriate primary outcome measure and recruitment and retention rates, and the non-inferiority margin. We also undertook a realist-informed process evaluation and rehearsed the health economic data collection tool and analysis.
Design: A pilot randomised controlled trial over a 15-month period, with a realist-informed process evaluation. In parallel, we rehearsed the health economic evaluation and explored patients’ preferences to inform a preference elicitation exercise for a definitive study.
Setting: The setting was NHS dental practices in North West England.
Participants: A total of 217 low-risk patients in eight high-street dental practices participated.
Interventions: The current practice of using dentists to provide NHS dental check-ups (treatment as usual; the control arm) was compared with using dental therapists to provide NHS dental check-ups (the intervention arm).
Main outcome measure: The main outcome measure was difference in the proportion of sites with bleeding on probing among low-risk patients. We also recorded the number of ‘cross-over’ referrals between dentists and dental therapists.
Results: No differences were found in the health status of patients over the 15 months of the pilot trial, suggesting that non-inferiority is the most appropriate design. However, bleeding on probing suffered from ‘floor effects’ among low-risk patients, and recruitment rates were moderately low (39.7%), which suggests that an experimental design might not be the most appropriate. The theory areas that emerged from the realist-informed process evaluation were contractual, regulatory, institutional logistics, patients’ experience and logistics. The economic evaluation was rehearsed and estimates of cost-effectiveness made; potential attributes and levels that can form the basis of preference elicitation work in a definitive study were determined.
Limitations: The pilot was conducted over a 15-month period only, and bleeding on probing appeared to have floor effects. The number of participating dental practices was a limitation and the recruitment rate was moderate.
Conclusions: Non-inferiority, floor effects and moderate recruitment rates suggest that a randomised controlled trial might not be the best evaluative design for a definitive study in this population. The process evaluation identified multiple barriers to the use of dental therapists in ‘high-street’ practices and added real value.
Future work: Quasi-experimental designs may offer more promise for a definitive study alongside further realist evaluation.
Trial registration: Current Controlled Trials ISRCTN70032696.