Population estimates of survival in women with screen-detected and symptomatic breast cancer taking account of lead time and length bias
Background Evidence of the impact of breast screening is limited by biases inherent in non-randomised studies and often by lack of complete population data. We address this by estimating the effect of screen detection on cause-specific fatality in breast cancer, corrected for all potential biases, using population cancer registry data. Methods Subjects (N = 26,766) comprised all breast cancers notified to the West Midlands Cancer Intelligence Unit and diagnosed in women aged 50–74, from 1988 to 2004. These included 10,100 screen-detected and 15,862 symptomatic breast cancers (6,009 women with interval cancers and 9,853 who had not attended screening). Our endpoint was survival to death from breast cancer. We estimated the relative risk (RR) of 10-year cause-specific fatality (screen-detected compared to symptomatic cancers) correcting for lead time bias and performing sensitivity analyses for length bias. To exclude self-selection bias, survival analyses were also performed with interval cancers as the comparator symptomatic women. Findings Uncorrected RR associated with screen-detection was 0.34 (95% CI 0.31–0.37). Correcting for lead time, RR was 0.49 (95% CI 0.45–0.53); length bias analyses gave a range of RR corrected for both phenomena of 0.49–0.59, with a median of 0.51. Self-selection bias-corrected estimates yielded a median RR of 0.68. Interpretation After adjusting for various potential biases, women with screen-detected breast cancer have a substantial survival advantage over those with symptomatic breast cancer.
KeywordsPopulation screening Mammography Lead-time bias Length bias Self-selection bias
Prue Allgood was supported by a grant from the Princess Grace Hospital, London. Iris Nagtegaal was supported by the Dutch Cancer Society. The screening histories project at the West Midlands QA Reference Centre is supported by a grant from the Breast Cancer Research Trust. The authors are grateful to NHS Trusts, private hospitals and NHS breast screening services in the West Midlands for providing cancer registration and breast screening data and to Rosie Day at the WMCIU for extracting the breast cancer data used in this study from the WMCIU’s cancer registration database.
- 3.Bundred NJ (2007) Bs22 prognostic index in screen-detected breast cancer. ANZ J Surg 77:A5–A6Google Scholar
- 9.Collett D (1994) Modelling Survival Data in Medical Research. Chapman and Hall, New YorkGoogle Scholar
- 10.Breslow NE, Day NE (1987) Statistical Methods in Cancer Research II. Design and Analysis of Cohort Studies. IARC Scientific Publications, LyonGoogle Scholar
- 11.Duffy SW, Nagtegaal N, Wallis M, Cafferty F, Houssami N, Warwick J et al (2008) Correcting for lead time and length bias in estimating the effect of screen-detection on cancer survival. Am J Epidemiol (in press)Google Scholar
- 13.Duffy SW, Cuzick J, Tabar L, Vitak B, Chen THH, Yen MF et al (2002) Correcting for non-compliance bias in case-control studies to evaluate cancer screening programs. Appd Stat 51:235–243Google Scholar