Mapping algorithms from non-preference to preference-based outcome measures: do they really work in practice?
Project Reference: NDPH/MT16/015
The use of ‘mapping’ algorithms to predict utilities to enable the estimation of quality-adjusted life years (QALYs) from non-preference based instruments has become a popular approach in health economic evaluation. Around one quarter of the QALY estimations informing recent National Institute of Health and Care Excellence (NICE) appraisals in England and Wales involved the implementation of a mapping statistical model. Moreover, the use of mapping techniques has received increasing attention from researchers, who advise caution when using mapping algorithms in practice. These concerns include the poor quality of reporting of many such studies, the inaccuracy of utility predictions for poor health states, and failure to capture uncertainty around means and across individuals. These issues are likely to introduce important biases when estimating treatment effects between competing alternatives, which may ultimately impact patient safety. To date, a thorough investigation of the potential implications of using mapping algorithms to estimate preference-based estimates of treatment effects using clinical trial data has not been conducted. This DPhil proposal will explore this issue using data from a number of existing trials, and will provide guidance on the appropriateness of using this method to inform health outcomes when conducting economic evaluations.
Research Experience, Research Methods and Training
This studentship will provide experience and training in literature review methods, statistical methods to conduct mapping algorithms, and the use of parametric and non-parametric methods to propagate uncertainty of mapping models in practice. The student will have access to real trial data to test the study hypothesis and will received training in econometrics and advanced programming techniques in statistical software such as Stata.
This project would suit a candidate with a strong quantitative (e.g. mathematics, statistics or economics) background and/or a proven track record of previous experience in health outcomes research and an interest in the development of methodology to conduct economic evaluations.