Estimating cancer prevalence using mixture models for cancer survival
Article first published online: 10 APR 2002
Copyright © 2002 John Wiley & Sons, Ltd.
Statistics in Medicine
Volume 21, Issue 9, pages 1257–1270, 15 May 2002
How to Cite
Phillips, N., Coldman, A. and McBride, M. L. (2002), Estimating cancer prevalence using mixture models for cancer survival. Statist. Med., 21: 1257–1270. doi: 10.1002/sim.1101
- Issue published online: 10 APR 2002
- Article first published online: 10 APR 2002
- Manuscript Accepted: MAY 2001
- Manuscript Received: AUG 2000
Knowledge of cancer prevalence is useful for estimating the ongoing level of resources utilized in the treatment of disease and is of some public health interest. Cancer prevalence is estimated first as the proportion of persons previously diagnosed (PD) with cancer that are still alive; and second as the proportion of individuals in the population who were previously diagnosed with cancer and who have not been cured (NC). The proportion of cases that are cured is estimated by assuming that the cured and uncured cases have distinct survival patterns. The hazard for cured cases is assumed to be a multiple of the hazard from causes other than cancer in the general population. The hazard for uncured cases is assumed to have two independent components: one corresponding to the disease-specific hazard, and the other a multiple of the population hazard from ‘other causes’. Future prevalence estimates are obtained by projecting the survival of current prevalent cases as well as the survival of future incident cases. Copyright © 2002 John Wiley & Sons, Ltd.