Protocol for an adaptive platform trial of intended service user-derived interventions to equitably reduce non-attendance in eye screening programmes in Botswana, India, Kenya & Nepal

Background Only 30-50% of people referred to clinics during community-based eye screening are able to access care in Botswana, India, Kenya, and Nepal. The access rate is even lower for certain population groups. This platform trial aims to test multiple, iterative, low-risk public health interventions and simple service modifications with a series of individual randomised controlled trials (RCT) conducted in each country, with the aim of increasing the proportion of people attending.

Methods and Analysis We will set up a platform trial in each country to govern the running of a series of pragmatic, adaptive, embedded, parallel, multi-arm, superiority RCTs to test a series of service modifications suggested by intended service users. The aim is to identify serial marginal gains that cumulatively result in large improvements to equity and access. The primary outcome will be the probability of accessing treatment among the population group with the worst access at baseline. We will calculate Bayesian posterior probabilities of clinic attendance in each arm every 72 hours. Each RCT will continually recruit participants until the following default stopping rules have been met: >95% probability that one arm is best; >95% probability that the difference between the best arm and the arms remaining in the trial is <1%; or 10,000 people have been recruited. Lower thresholds may be used for RCTs testing interventions with very low risks and costs. The specific design of cluster RCTs will be determined by our research team once the intervention is known, but the population and outcome will be the same across all trials.

Discussion This APT will be used to identify effective service modifications, driving continuous improvements in access.

Ethics and Dissemination This trial has been approved by the research ethics committee at the London School of Hygiene and Tropical Medicine. Approvals for individual interventions will be sought from UK and local ethics committees. Results will be shared via local workshops, social media, and peer-reviewed publications.

Trial Registration: ISRCTN 53970958. Registered on 21 September 2023

Strengths and Limitations

• Randomised control trials are resource intensive and often require lengthy set up periods. The adaptive platform design allows for the evaluation of multiple interventions with a single outcome, governed by a predefined set of criteria

• The study defaults are designed to test multiple low-risk, incremental service modifications in series, and quickly identify those that are just as good as, or superior to the status quo.

• Our high default tolerance for type I error means that we will often incorrectly identify arms as superior when really there is no difference. This is acceptable when arms confer similar costs and negligible risks.

• Our default very low type II error rate means that we will very rarely mistakenly identify an inferior arm as being superior.

• Our trial is embedded within screening programmes and uses automated randomisation, allocation, data collection, and statistical testing to minimise resource requirements.