Richard Riley is a Reader in Biostatistics at the University of Birmingham, with a joint post in the School of Mathematics and the School of Health and Population Sciences. He is also a Statistics Editor for the BMJ, an Associate Editor of Statistics in Medicine, and a co-convenor of the Cochrane Prognosis Methods Group.
In terms of research, Richard is an expert on statistical methods for meta-analysis (the combination of results across multiple studies) and prognosis research (the study of future outcomes in those with existing disease). In meta-analysis, he specialises in methods for dealing with multiple correlated outcomes, and for synthesising individual participant data (IPD). In prognosis, Richard co-leads the PROGRESS initiative (PROGnosis RESearch Strategy), that seeks to improve the standards of prognosis research. This includes statistical methods to identify prognostic factors, to develop and validate prognostic models (risk prediction models), and to identify predictors of treatment response for stratified medicine. He combines his meta-analysis and prognosis interests through the development and validation of prognostic models using IPD from multiple studies, and through meta-analysis of prognostic factor studies. He has received numerous healthcare related grants, from funders including the MRC and NIHR, and has published over 60 applied and methodological research articles. He currently supervises 6 PhD students and co-supervises 3 others.
In terms of teaching, Richard delivers a year-long Medical Statistics module to undergraduate and postgraduate maths students. This covers the basic principles of statistical methods for medical and epidemiological research. Many students who take the module go on to have a career in medical statistics. Richard also occasionally lectures on the Masters in Public Health.
Richard delivers three training courses:
(1) AN INTRODUCTION TO STATISTICAL METHODS FOR META-ANALYSIS
This course covers the basic principles for undertaking a meta-analysis of results extracted from the literature. It focuses mainly on the synthesis of results from randomised trials of interventions, but also includes discussion of prognostic factor studies and diagnostic test studies. The course covers the process of data extraction, and the concept of fixed-effect and random-effects meta-analysis models. Continuous, binary and time-to-event outcomes are covered, and the core meta-analysis methods are covered, such as inverse-variance, Dersimonian and Laird (random effects), and exact likelihood methods for binary data. It also explains how to interpret the summary results from a meta-analysis, and display them appropriately using forest plots, confidence intervals, and prediction intervals. Methods for examining potential publication bias are covered, including contour-enhanced funnel plots and tests for asymmetry. Meta-regression is also detailed, for examining treatment-covariate interaction, and the threat of ecological bias and potential advantages of individual participant data are discussed. For more information or if you would like this course run for your team, contact: firstname.lastname@example.org
(2) STATISTICAL METHODS FOR INDIVIDUAL PARTICIPANT DATA (IPD) META-ANALYSIS
The course covers the fundamental statistical methods and principles for meta-analysis when IPD (Individual Participant/Patient Data) are available from multiple related studies. The course considers continuous, binary and time-to-event outcomes. It focuses mainly on the synthesis of randomised trials of interventions, and how to estimate treatment effects and treatment-covariate interactions (effect modifiers). However, it also covers the issues for IPD meta-analysis of observational studies examining prognostic factors and risk prediction (prognostic) models. Methods for examining bias in IPD meta-analyses is also described, alongside approaches for combining IPD and non-IPD studies. Key messages are illustrated with real examples, and participants get the chance to conduct a variety of IPD analyses within STATA, to practice the key methods and to reinforce the learning points. The course assumes an understanding of regression methods (such as linear, logistic, and Cox) and the basic principles of traditional meta-analysis methods that synthesise only published results. For more information or if you would like this course run for your team, contact: email@example.com
(3) PROGNOSIS RESEARCH: CONCEPTS, METHODS AND CLINICAL APPLICATION
Prognosis research provides information crucial to understanding, explaining and predicting future clinical outcomes in people with existing disease or health conditions. This 3-day course is designed to introduce the key components of prognosis research to health professionals and researchers, including: (i) a framework of different prognosis research questions (overall prognosis, prognostic factors, prognostic models, and stratified medicine); (ii) key principles of study design and methods; (iii) interpretation of statistical results about prognosis; and (iv) the limitations of current prognosis research, and how the field can be improved. The course is based on 4 articles published in BMJ/PLOS Medicine (February 2013) and course combines seminars and lectures by international experts in the field (Prof. Harry Hemingway, Prof. Douglas Altman, Dr. Richard Riley, Prof. Danielle van der Windt, Dr. Kate Dunn, Dr. Pablo Perel) with group work and case studies. For more information and registration: http://www.progress-partnership.org/training.html
Richard Riley qualified with a BSc in Mathematics from the University of Nottingham (1998), and went on to complete an MSc with distinction (1999) and a PhD (2005) in Medical Statistics from the University of Leicester (2005). He worked as a Research Associate (1999 to 2001) in the Centre for Biostatistics at the University of Leicester, and as a Lecturer in the Centre for Medical Statistics and Health Evaluation at the University of Liverpool (2006-2008), where he also completed a Research Fellowship in Evidence Synthesis awarded by the Department of Health. He is now a Reader in Biostatistics at the University of Birmingham, with a joint post in the School of Health and Population Sciences, and the School of Mathematics.
Richard specialises in the application and development of statistical methods for evidence synthesis and meta-analysis. His main methodological research interests include:
Statistical models for multivariate meta-analysis of multiple outcomes
Statistical methods for undertaking an individual participant data (IPD) meta-analysis
Approaches to combining IPD with aggregate data in meta-analysis
Investigating and dealing with publication and availability bias in IPD meta-analysis
Particular clinical applications of interest (both in primary studies and systematic reviews) include:
Identifying and evaluating diagnostic tests
Identifying and evaluating prognostic factors and biomarkers
Developing, validating and assessing the impact of prognostic models and risk prediction models
Facilitating stratified medicine, in particular by identifying patient-level factors that interact with treatment effect (‘predictive markers’, ‘treatment-covariate interactions’)
A particular research passion is to improve the quality, design, conduct, analysis and reporting of prognosis research studies.
Richard has been a Statistics Editor of the BMJ since 2009, an Associate Editor of Statistics in Medicine since 2011, and a co-convenor of the Cochrane Prognosis Methods Group since 2007. He enjoys teaching medical statistics to mathematics students, and leads training courses on basic and advanced statistical issues, to either statistical or non-statistical professions. He also speaks regularly at national and international conferences.