Jump to: Page Content, Site Navigation, Site Search,
You are seeing this message because your web browser does not support basic web standards. Find out more about why this message is appearing and what you can do to make your experience on this site better.
Published 1 September 2009, doi:10.1136/bmj.b3512
Cite this as: BMJ 2009;339:b3512
| The first 150 words of the full text of this article appear below. |
Despite Collins and Altmans re-analysis of data from the THIN database to validate the QRISK equation for predicting cardiovascular disease,1 adoption of QRISK in primary care is premature because key issues about the handling of missing data and the use of social deprivation indices remain unresolved.
Collins and Altman again highlight that complete data were available for just over a quarter of subjects. We appreciate that imputation methods were applied, but we question use of age-sex means of QRESEARCH data for lipid concentrations and blood pressures. This implies that QRESEARCH data were missing completely at random within age-sex strata—an assumption acknowledged as incorrect when the developers of QRISK revised their equation2. It also implies that observed QRESEARCH data reflect age-sex norms in the population—an assumption questioned by the developers comparison of their data with the health survey for England.3 We call for additional validation using data from bespoke cohort
Richard Morris, professor of medical statistics and epidemiology1, Irene Petersen, MRC training fellow1, Louise Marston, research statistician1, Kate Walters, clinical senior lecturer1, James Carpenter, reader in medical and social statistics2, Irwin Nazareth, professor of primary care1
1 Department of Primary Care and Population Health, University College London, Royal Free Campus, London NW3 2PF, 2 Medical Statistics Unit, London School of Hygiene and Tropical Medicine, London WC1E 7HT
r.morris@pcps.ucl.ac.uk