| Clinically relevant evaluation of prognostic models for CHD |
| IAN WHITE (cambridge University) |
|
| Prognostic models constructed from observational data are widely used to predict cardiovascular disease. This talk will consider how to quantify the value of adding a new risk factor to the model, and illustrate the ideas using data from 31 cohort studies. |
| |
| Propensity score models in observational studies |
| MILISA BLAGOJEVIC (Keele University) |
| The focus is on the estimation of the treatment effect in observational studies where baseline differences between treatment groups are inevitable. The method of propensity scores successfully balances the groups for observed covariates, thus removing some of the bias inherent in treatment allocation, and also avoids the issue of building a parsimonious model. |
| |
| Methodological issues in using routine data to a) model risk of emergency admissions and b) determine adverse drug reactions |
| PETER DONNAN (Dundee University) |
| Encashed prescriptions in the community are easily linked to hospitalisation and death in Tayside, Scotland using a single identifier for all NHS encounters. This facilitates research in a number of areas. This talk will focus on predicting risk of emergency admissions as well as determining adverse drug reactions in the community. |
| |
| Meeting Contact: GILLIAN LANCASTER (g.lancaster@lancaster.ac.uk) |
| Organising Group(s): Jonit meeting with The RSS Primary Health Care Study Group and Society of Acadmic Primary Care, Special Interest Epidemiology Group |