Please find here the details of a proposed PhD project in medical statistics.
"Understanding how complex interventions do or do not work"
Supervisors
Dr Toby Prevost, General Practice and Primary Care Research Unit, University of Cambridge
www.medschl.cam.ac.uk/gppcru
Prof Simon Thompson, Director, MRC Biostatistics Unit, Institute of Public Health, Cambridge
www.mrc-bsu.cam.ac.uk
Background
Lifestyle behaviours are important in preventing, delaying and limiting the progression of chronic diseases. Changing such behaviours is challenging to accomplish and typically difficult to evaluate. Randomised trials often test complex interventions that involve several potentially impactful "components" (such as: a "theory" that specifies a linked "chain" of correlated variables that is assumed to apply; "persons" delivering the intervention; and the organisation of the content of the intervention to be delivered in "parts" and over "time"). These trials also involve the measurement of "intermediate variables" that may be causally connected with the behaviour.
This project will be based around the ProActive trial of a complex intervention to encourage greater physical activity in those at higher risk of developing Type 2 Diabetes (published in the Lancet in 2008). A second multiple-behaviour dataset, the Addition plus trial, will be also be available.
Purpose
The purpose of the project is to apply and develop statistical methods to understand how behavioural interventions do or do not work, and to inform the design of future complex intervention trials in this important area.
With multiple components, and a chain repeated over time in three trial arms, there are several areas in which any limitation in intervention effectiveness could be chosen to be investigated via research questions such as: Was the assumed underlying theory observed to apply well? To what extent can the assessment of causal relationships be inhibited by measurement error? Could the "parts" of the intervention, and their delivery, inform the potential effectiveness? How could we improve the design of future trials?
Possible lines of research
These could be tackled by means of the following:
· Developing effective methods of analysis for intermediate variables to understand any causal relations to behaviour change.
· How to make best use of the repeated timepoints and multiple arm data to identify when a causal chain is operating.
· Exploring the potential of individual level within-arm data to make inferences about behaviour change.
· The impact of measurement error on the assessment of any causal relations.
· The potential of trading bias with variability in combining and correcting self-reported and objective measures of behaviour.
· Using simulation methods to assess the potential effect of fuller delivery of components of the intervention, with parameters informed by an available 10% data-rich sub-study.
· Investigating how targeting of "parts" of the intervention could increase the effect on behaviour.
· Design features - optimising the timing and frequency of collecting intermediate variables; when to capture behaviour change, and when to deliver the intervention.
· Evaluating the potential improvement in power from making use of intermediate variables when testing for interaction between subgroups.
· Extension of the techniques developed above for continuous variables to binary and ordinal variables.
Further details
The project is for three years full-time, beginning 1st October 2008, and includes a stipend of £12,940 pa with discretionary increase based on age and experience, associated support costs and all relevant University and College fees.
Candidates should have a good first degree in mathematics with a substantial component of statistics or an MSc in Statistics.
The General Practice and Primary Care Research Unit (GP&PCRU) is part of the University Department of Public Health and Primary care, and is housed with the MRC Biostatistics Unit in the Institute of Public Health.
The GP&PCRU is one of only five research groupings in general practice and primary care in the UK awarded a 5* rating in the last national research assessment exercise and is a partner in the National Institute for Health Research School (NIHR) for Primary Care Research from which this studentship is funded. The GP&PCRU comprises seven research groups, including the Behavioural Science group headed by Prof Stephen Sutton, the Statistics and Modelling Group headed by Senior Statistician Dr Toby Prevost and the newly established Qualitative Research Group headed by Senior Lecturer Dr Simon Cohn. Owing to the growing interdisciplinary nature of the research conducted at the GP&PCRU, we are seeking the best overall candidate for one of three proposed projects relating to these core research strengths.
Application and selection process
Please submit a full CV, including the names and contact details of two academic referees.
Please also include a covering letter indicating your relevant experience and interest in the statistical and substantive areas.
Applications, preferably by email, should be send jointly to Helen Ludford ([log in to unmask]), Stephanie Vo ([log in to unmask]) and Toby Prevost ([log in to unmask]), to whom informal enquiries are welcome (email or phone 01223 330593).
The closing date for applications is Wednesday 23rd July 2008.
Short-listed candidates will be invited for interview during the week of 11th August 2008.
Dr. A.T.Prevost
Senior Statistician
General Practice and Primary Care Research Unit
Department of Public Health and Primary Care
University of Cambridge
Institute of Public Health
Forvie Site, Robinson Way
Cambridge CB2 0SR
Tel: +44 (0) 1223 330593
Fax: +44 (0) 1223 762515
|