Were our estimates of overdiagnosis with mammography screening in the United States "based on faulty science"?

Abstract

The author responds to criticisms that his and his coauthor’s estimates of overdiagnoses with mammography screening are “based on faulty science” by reviewing the statistical analyses and comparing global data with those of Connecticut.

Figure 1.

Annual incidence of early- and late-stage breast cancer in women younger than 40 years of age, 1976–2010, Surveillance, Epidemiology, and End Results 9 (lower panel reproduced from Figure 1 of Bleyer and Welch [2] with 2009 and 2010 data added, in brackets) and in women 35 to 39 years of age at diagnosis (upper panel). The straight red lines are linear regressions of the late-stage incidence for 1980–2010. Lower panel from New England Journal of Medicine, A Bleyer, HG Welch, “Effect of three decades of screening mammography on breast-cancer incidence”, 2012;367:1998–2005. Copyright © 2012 Massachusetts Medical Society. Reprinted with permission.

Figure 2.

Annual incidence of common cancers in women 40 years of age and older in Connecticut, 1973–2010. The solid blue lines are linear regressions (with lung cancer an exception) of incidence rates during 1935–1973. The dotted blue lines are projections to 2010 of the trend prior to 1973. The solid brown lines are linear regressions of the Surveillance, Epidemiology, and End Results (SEER) 1973–2010 era.

Figure 3.

Annual incidence of breast cancer in 35- to 39- and 40- to 44-year-old women, 1973–2010, Surveillance, Epidemiology, and End Results 9. Dotted curves indicate invasive cancer and solid curves indicate all cancer, including in situ.

Figure 4.

Steady state incidence since 1990 in invasive breast cancer not due to hormone-replacement therapy (HRT) in women 40 years of age and older in Connecticut and in Surveillance, Epidemiology, and End Results (SEER) 9. The bold curves designate all nine SEER registries; the lighter curves represent Connecticut. Horizontal lines indicate mean value after impact on incidence stabilized and “HRT years” are excluded.

Figure 5.

Annual incidence of invasive breast cancer in women of all ages, 1975–2010, Surveillance, Epidemiology, and End Results (SEER) 9. Adapted from age-adjusted SEER incidence of breast cancer 1975–2010 (figure in [1]), with an additional year (2010) of data from SEER added. The circled area designates the change in incidence, or lack thereof, of invasive breast cancer incidence during 2003–2010.

Figure 6.

Breast cancer incidence in SEER 9 registries since 1976. Upper panel: Increase in annual incidence of in situ and localized breast cancer from prescreening mammography (1976–1985) to postscreening steady state (1995–2010), age 40 and older. Lower panel: Average annual percentage change (AAPC) in breast cancer incidence 1976–2010 among women 35 to 39 years of age.

Figure 7.

Incidence of breast cancer in white females from 1950 to 1985 according to five areas and Surveillance, Epidemiology, and End Results (SEER) 9 data. All black elements are from the original figure in Kessler et al. [17].

Figure 8.

Annual breast cancer incidence in women in Connecticut of age 40 and older since 1935 [8] and by extent of disease at diagnosis and of age younger than 40 years of age after 1973 when Connecticut became a Surveillance, Epidemiology, and End Results (SEER) site. Abbreviation: DCIS, ductal carcinoma in situ.

Figure 9.

Cumulative number of women diagnosed with early- and late-stage breast cancer since 1979, age 40 and older, Surveillance, Epidemiology, and End Results 9.