Eliciting and incorporating patients' opinions about missing data in randomised controlled trials

点击翻译按钮获取中文摘要

Missing data occurs when data is unavailable to be analysed and is a common challenge within clinical trials that can have serious consequences for the validity of results. The analysis of trials with missing data usually assumes the missing data are "missing at random", i.e. given an individual's past observed data, their probability of dropout does not depend on their present (or future) unobserved outcome.In many settings this assumption is implausible. For this reason, it is crucial to develop methods to assess the robustness of conclusions to departures from the missing at random assumption. Since we cannot base assumptions on data, an attractive approach is to incorporate experts' opinions about reasons for and distributions of the missing data in the trial's sensitivity analysis. In the past, experts have been defined as clinicians and methods to elicit their views have been developed. Patients have been overlooked in this process, even though they are likely to have important opinions to share regarding patient missing data. Currently, there is no method available to elicit patient's views in this important aspect of trial analysis.The current studentship aims to develop and test a practical, accessible approach that allows patient's opinions about missing data in a clinical trial to be meaningfully and accurately elicited and incorporated into a trial's sensitivity analyses.The project involves:1. Review of the literature on current expert elicitation methods available2. Based on the findings from 1, co-design with a patient panel a tool to elicit their views on missing data. This would include a series of workshops led by the student to incorporate the panel's views on what the new tool should look like, prioritise the key aspects to ensure feasibility and refine it.3. Evaluate the tool developed in 2, by implementing it with a group of patients, using a real-world trial as an example, and based on criteria outlined by Johnson et al. for Bayesian elicitation tools (reference: https://bit.ly/30j3iHV) including validity, reliability, responsiveness, feasibility4. Incorporate the opinions elicited in the application of the tool in (3) in a trial's sensitivity analysis, using pre-established methods, to assess the robustness of the findings5. Produce recommendations for the elicitation of patient's views regarding missing data in clinical trials