National Institute for Health Research Grant
Alan Kimber, Reader in Statistics, Robin Mitra, Lecturer in Statistics and Dave Collett, Visiting Professor in S3RI and Associate Director of Statistics and Clinical Audit at NHS Blood and Transplant, have won an award from the National Institute for Health Research to develop improved methods of analysing survival data on patients receiving organ transplants.
NHS Blood and Transplant collects data on all UK transplant recipients and follows them up until death. It thus has a large and rich data set of survival times which it uses, for example, in assessing transplant benefit to patients. As transplantation is often successful, there can be high proportions of censored observations, that is, transplant patients who are still alive at the end of the appropriate study period, which considerably reduces the effective sample size for modelling patient survival behaviour. So, even though patient numbers are fairly high, making the best use of data is very important.
As is common with routinely collected patient data, values may be missing from the data set on variables, known as covariates, that are likely to influence survival times. These missing values can cause difficulties in the data analysis. For example, the smoking history of a lung donor is an important covariate for the prognosis of the recipient, but information on smoking history is not available for all donors. Similarly, the severity of abdominal swelling in patients waiting for a liver transplant is known to affect their prognosis, but this information is sometimes not available.
This award will allow development of methods for missing covariate techniques to improve analyses of transplant survival data in a range of problems arising from organ donation and transplantation. Appropriate techniques will be implemented, in collaboration with NHS Blood and Transplant, and activities will be undertaken to embed the recommended methods in the work of medical statisticians and the transplant community. The award will also support an early career statistician to work on the project.