Nelson Christopher P, Lambert Paul C, Squire Iain B, Jones David R
Centre for Biostatistics and Genetic Epidemiology, Department of Health Sciences, University of Leicester, 2nd Floor, Adrian Building, University Road, Leicester LE1 7RH, UK.
Eur Heart J. 2008 Apr;29(7):941-7. doi: 10.1093/eurheartj/ehn079. Epub 2008 Mar 6.
To illustrate the application of relative survival to observational studies in coronary heart disease (CHD) and potential advantages compared with all-cause survival methods. Survival after myocardial infarction (MI) is generally assessed using all-cause or cause-specific methods. Neither method is able to assess the impact of the disease or condition of interest in comparison with expected survival in a similar population. Relative survival, the ratio of the observed and the expected survival rates, is applied routinely in cancer studies and may improve on current methods for assessment of survival in CHD.
Using a cohort of subjects after a first recorded acute MI, we discuss the application of relative survival in CHD and illustrate a number of the key issues. We compare the findings from relative survival with those obtained using Cox proportional and non-proportional hazards models in standard all-cause survival. Estimated survival rates are higher using relative survival models compared with all-cause methods.
Estimates obtained from all-cause mortality fail to disentangle mortality associated with the condition of interest from that due to all other causes. Relative survival gives an estimate of survival due to the disease of interest without the need for cause of death information.
