Premium
Predicting the lung cancer burden: accounting for selection of the patients with respect to general population mortality
Author(s) -
Heinävaara Sirpa,
Hakulinen Timo
Publication year - 2006
Publication title -
statistics in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.996
H-Index - 183
eISSN - 1097-0258
pISSN - 0277-6715
DOI - 10.1002/sim.2443
Subject(s) - lung cancer , selection (genetic algorithm) , medicine , population , excess mortality , intensive care medicine , statistics , oncology , demography , environmental health , computer science , mathematics , artificial intelligence , sociology
Incidence, survival, prevalence and mortality are the elements of the cancer burden. The cancer burden is thus contributed by the same individuals from the diagnosis of cancer (incidence) until possible death from cancer (mortality). It would therefore be natural that predictions of future cancer burden, those needed by health administration, for example, could be based on individual data. This paper presents a new model for estimating future cancer burden, a model where individuals are followed from birth to death with or without a diagnosis of cancer. The model can formally be expressed as a summation of log‐likelihoods of getting cancer and of surviving the cancer and other causes of death until death or censoring. The new model is illustrated with data of Finnish males with or without a diagnosis of lung cancer in calendar period 1987–1997. Incidence is modelled with a general age‐cohort model with a drift and survival from cancer with a parametric mixture model. In a model for survival from other causes of death, selection of patients with respect to general population mortality is accounted for. Future cancer burden is illustrated with short‐term predictions of prevalence and mortality. Copyright © 2005 John Wiley & Sons, Ltd.