Lead supervisor: Dr Hayley Jones, Senior Lecturer in Medical Statistics, Population Health Sciences, University of Bristol
Co-supervisors: Dr Penny Whiting (Bristol), Professor Chris Hyde (Exeter), Dr Huiqin Yang (Exeter)
With an increasing number of tests available for many diseases, statistical methods are needed to determine the ‘best’ test or combination of tests. Data on test accuracy are often available from multiple studies and statistically pooled using meta-analysis. Working with case studies including tests for coeliac disease, this project will evaluate and develop advanced meta-analysis models and provide guidance to systematic reviewers on what data are required.
New diagnostic tests are being developed at an ever-increasing rate. Healthcare providers and guideline developers such as NICE are faced with decisions as to which test or, likely, combination of tests to use. Although a number of factors play a part in these decisions, test accuracy (measured by the sensitivity and specificity), is critical.
Systematic reviews and meta-analyses, which synthesise data from all relevant studies, are increasingly used to summarise test accuracy. However, methodological development has almost entirely focused on evaluation of each test in isolation. These methods are insufficient to answer important questions about comparative test accuracy and the ‘joint’ accuracy of tests used in combination. A small advanced literature is developing around these issues, but the models proposed require further exploration and validation before they can be widely recommended. Further, some of these methods require information that appears to be rarely available from primary studies – and most have focused on quantifying the accuracy of only two tests. In practice there will usually be more than two competing tests, with different studies reporting data for different tests. Methods for ‘network meta-analysis’ (NMA) of intervention effectiveness are now well developed, but NMA of test accuracy is in its early infancy.
1. To investigate and evaluate recently suggested statistical methods for meta-analysis of the comparative or joint accuracy of two tests.
2. To assess the potential contribution of indirect evidence to such a meta-analysis.
3. To investigate the availability of the required data in study publications. Drawing on these findings, the student will provide guidance to systematic reviewers on which data to extract.
4. To develop Bayesian evidence synthesis / NMA models, extending comparative and joint meta-analysis models to the case of three or more tests, including networks of test comparisons, e.g. A vs B, B vs C, and A vs C.
Methods will be applied to case study data sets addressing important clinical questions, including data from an ongoing systematic review of the accuracy of multiple tests for coeliac disease. We are currently extracting data from over 100 studies, each evaluating the accuracy of one or more tests.
The student will be based at the department of Population Health Sciences, Bristol Medical School, a major centre for collaborative and multi-disciplinary research, with staff from a wide range of academic disciplines. He/she will receive excellent training and support from a supervisory team with expertise across multiple areas of diagnostic test evaluation, including statistical (particularly Bayesian) methods for meta-analysis, systematic review methodology, and health economic evaluation.
The student will be a member of the multi-parameter evidence synthesis (MPES) research group at the University of Bristol, which has pioneered much of the methodological development in network meta-analysis of intervention effectiveness and is the home of the NICE Guidelines Technical Support Unit. Through co-supervision by academics at the University of Exeter, the student will also be closely connected with PenTAG and the Exeter Test Group, groups which are particularly experienced in health technology assessment, test evaluation and input into policy making.
UK applicants for a studentship must have obtained, or be about to obtain, a First or Upper
Second Class UK Honours degree. EU students will need to have gained the equivalent
qualifications outside the UK in an area appropriate to the skills requirements of the project.
Applicants with a Lower Second Class degree will be considered if they also have a Master’s
degree or have significant relevant non-academic experience.
As this project requires strong quantitative skills, applications are particularly welcome from individuals with an MSc in Medical Statistics, Statistics or similar. Some experience with or knowledge of Bayesian statistical analysis is desirable.
Please make an online application for this project here
Applications close at 17:00 on 25th November 2019.
Contact: Hayley Jones [email protected]