Statistical methods for the evaluation of costs and effects of new surgical procedures

Project

Evaluation of new interventions requires balancing of the resulting costs against the benefits due to improved survival and quality of life. Data arising from randomised clinical trials provides evidence for the joint analysis of costs and benefits. When the intervention to be evaluated is a surgical procedure, a number of complications may arise. This PhD will consider some practical and methodological issues in this context. These include the following.
•Delays between patient recruitment and surgery, due to surgical waiting lists, complicate the estimation of treatment costs and benefits if the comparator can be administered without delay. The impact of different choices around the timing of the economic comparison will be studied, comparing traditional intention to treat analysis (from the date of randomisation) against methods that aim to estimate unbiased treatment effects from the date of the surgery (for those who receive survive to surgery) or from the date that surgical assessment begins.
•Due to the complicated nature of surgery there may be variation in outcomes due to developing expertise, patient characteristics, surgical expertise and hospital factors. Methods for dealing with this potentially complicated correlation structure in trial-based cost-effectiveness analysis will be investigated using multi-level and non-linear regression models.
•In contrast to the consistent formulation and dosing used in drug trials, the exact surgery received by individual patients will vary. For example, a recent trial in heart surgery reported that only a partial operation could be completed for some patients, whilst a smaller number did not have the planned surgery at all. Moreover, subsequent treatments were dependent on the outcome of the initial operation. In order to investigate how these different components of the intervention contribute to the overall treatment effect it may be useful to consider modern methods for non-randomised studies. In particular mediation methods will be investigated.
•Based on results of (ii) and (iii) the student will investigate methods for assessing cost-effectiveness of the surgical options according to patient characteristics, sometimes referred to as stratified medicine.