Three Bayesian trials with adaptive sample size and response-adaptive randomisation
Presented by Dr Tom Snelling
Gastroenteritis remains a major health problem for remote Aboriginal children. There is pressing need to identify better treatment and preventive strategies in this finite population of high disease burden children. NICE-GUT and ORVAC are NHMRC-funded trials of the treatment and prevention of gastroenteritis in Aboriginal children respectively. We use Bayesian inference and adaptive sample size in these two-arm trials with fixed ratio of assignment to each arm. This is expected to mitigate the risk of an inconclusive result, minimise needless enrolment if the proposed strategy is futile, and allow early implementation if the treatment benefit is greater than the minimum clinically important difference.
Poor immunisation timeliness is a major contributor to the population susceptibility to infectious diseases. Text message reminders are likely to be cheap and effective for improving timeliness of routine childhood immunisation, but the optimal type and timing of messages (in relation to the vaccine due date) is unknown. AuTOMATIC will compare 12 different text message type/timing combinations against each other and standard care. To maximise the efficiency of finding the optimal message type/timing, we are using both adaptive sample size and response-adaptive randomisation. The latter is expected to allow for focussed enrolment to the best performing arms.