Cancer Causes & Control

, Volume 23, Issue 10, pp 1625–1634

Estimating prevalence of distant metastatic breast cancer: a means of filling a data gap

  • Mark S. Clements
  • David M. Roder
  • Xue Qin Yu
  • Sam Egger
  • Dianne L. O’Connell
Original paper

DOI: 10.1007/s10552-012-0040-9

Cite this article as:
Clements, M.S., Roder, D.M., Yu, X.Q. et al. Cancer Causes Control (2012) 23: 1625. doi:10.1007/s10552-012-0040-9

Abstract

Purpose

To develop and validate a method for estimating numbers of people with distant cancer metastases, for evidence-based service planning.

Methods

Estimates were made employing an illness-death model with distant metastatic cancer as the illness state- and site-specific mortality as an outcome, using MIAMOD software. To demonstrate the method, we estimated numbers of females alive in Australia following detection of distant metastatic breast cancer during 1980–2004, using data on patient survival from an Australian population-based cancer registry. We validated these estimates by comparing them with direct prevalence counts.

Results

Relative survival at 10 years following detection of distant metastases was low (5–20 %), with better survival experienced by: (1) females where distant metastatic disease was detected at initial diagnosis rather than subsequently (e.g., at recurrence); (2) those diagnosed in more recent calendar years; and (3) younger age groups. For Australian females aged less than 85 years, the modeled cumulative risk of detection of distant metastatic breast cancer (either at initial diagnosis or subsequently) declined over time, but numbers of cases with this history rose from 71 per 100,000 in 1980 to 84 per 100,000 in 2004. The model indicated that there were approximately 3–4 prevalent distant metastatic breast cancer cases for every breast cancer death. Comparison of estimates with direct prevalence counts showed a reasonable level of agreement.

Conclusions

The method is straightforward to apply and we recommend its use for breast and other cancers when registry data are insufficient for direct prevalence counts. This will provide estimates of numbers of people who would need ongoing medical surveillance and care following detection of distant metastases.

Keywords

PrevalenceEpidemiologyMetastatic cancerBreast cancerStatistical models

Abbreviations

NSW

New South Wales

MIAMOD

Mortality and incidence analysis model

HR

Hazard ratio

CI

Confidence interval

Copyright information

© Springer Science+Business Media B.V. 2012

Authors and Affiliations

  • Mark S. Clements
    • 1
    • 2
  • David M. Roder
    • 3
  • Xue Qin Yu
    • 4
  • Sam Egger
    • 4
  • Dianne L. O’Connell
    • 4
    • 5
    • 6
    • 7
  1. 1.Department of Medical Epidemiology and BiostatisticsKarolinska InstitutetStockholmSweden
  2. 2.National Centre for Epidemiology and Population HealthAustralian National UniversityCanberraAustralia
  3. 3.National Breast and Ovarian Cancer CentreSydneyAustralia
  4. 4.Cancer Council NSWSydneyAustralia
  5. 5.University of SydneySydneyAustralia
  6. 6.University of New South WalesSydneyAustralia
  7. 7.University of NewcastleNewcastleAustralia