The Value of Large Multi-center Observational Datasets in Explaining Sources of Variation in Patient Outcomes



This program is for the research community.


In many areas of health care our understanding of the sources of variation in patient outcomes is incomplete because of a lack of high quality randomized studies. Precise estimates of the amount of variation explained by patient characteristics, provider identities and complex interventions is unlikely to ever be available. The next best source of evidence comes from large multi-center observational studies. The presenter has considerable experience of carrying out such research in the UK, for the most part to investigate the quality of surgical care provided within the NHS. When carrying out such research the overriding principle has been to consider audit projects as epidemiological studies. This implies that epidemiological methods should be used to generate high quality quantitative evidence on the processes and outcomes of care as well as on their determinants. As a consequence, the methodological requirements of audit projects do not differ from those of any epidemiological study: (1) representative sample of patients, (2) unbiased measures of processes, outcomes and their determinants, and (3) appropriate methods of analysis. The single most important factor to produce high quality evidence on the quality of care is that methods (e.g. third-party validation of data quality and case ascertainment) must be put in place to ensure that the collected data is unbiased, complete and accurate. Once this is ensured the analytic methods are complex but relatively straightforward.

In this presentation Professor John Browne will use the UK National Mastectomy and Breast Reconstruction Audit as an illustrative case. The focus will be on exploring the amount of variation in post-reconstructive surgical outcomes that can be explained by patient characteristics (e.g. age, obesity), treatment plan (e.g. free versus pedicled flap, use of post-operative radiotherapy) and surgical provider (e.g. low versus high volume centers). The main methodological challenges will be explored and a case will be made for commencing similar projects in the United States.

Date & Time(s)


Memorial Sloan Kettering Cancer Center
1275 York Avenue, Room M-107
New York, NY 10065


Center for Health Policy and Outcomes


John Browne, PhD
Associate Professor of Epidemiology and Public Health
Department of Epidemiology and Public Health
University College Cork
Honorary Senior Lecturer
Health Services Research Unit, London School of Hygiene and Tropical Medicine