Estimation and projections of cancer prevalence from cancer registry data

Stat Med. 2002 Nov 30;21(22):3511-26. doi: 10.1002/sim.1304.


A method, PIAMOD (Prevalence, Incidence, Analysis MODel), which allows the estimation and projection of cancer prevalence patterns by using cancer registry incidence and survival data is presented. As a first step the method involves the fit of incidence data by an age, period and cohort model to derive incidence projections. Prevalence is then estimated from modelled incidence and survival estimates. Cancer mortality is derived as a third step from modelled incidence, prevalence and survival. An application to female breast cancer is given for the Connecticut State by using data from the Connecticut Tumor Registry (CTR), 1973-1993. The age, period and cohort model fitted incidence quite well and allowed us to derive long-term projections up to 2030. Patients' survival was also projected to future years according to a scenario approach based on two extreme hypotheses: steady, that is, no more improvements after 1993 (conservative), and continuously improving at the same rate as during the observation period. Age-standardized estimated incidence shows a changing trend around the year 2005, when it starts decreasing. Age-standardized prevalence is expected to increase and change trend at a later date. Breast cancer mortality is projected as decreasing, as the combined result of no further increase in incidence and improving cancer patients' survival. An easy-to-use PIAMOD software package, on which work is in progress, will be made available to individual cancer registries and/or health planning institutions or authorities once it is developed. The use of the PIAMOD method for cancer registries will allow them to provide results of paramount importance for the whole community involved in the assessment of future disease burden scenarios in an evolving society.

MeSH terms

  • Adult
  • Aged
  • Aged, 80 and over
  • Breast Neoplasms / epidemiology*
  • Breast Neoplasms / mortality
  • Connecticut / epidemiology
  • Epidemiologic Methods*
  • Female
  • Humans
  • Incidence
  • Middle Aged
  • Models, Biological*
  • Prevalence
  • SEER Program
  • Survival Analysis