Open AccessBiasing the Interpretation of Mammography Screening Data by Age Grouping: Nothing Changes Abruptly at Age 50
Author(s) -
Kopans Daniel B.,
Moore Richard H.,
McCarthy Kathleen A.,
Hall Deborah A.,
Hulka Carol,
Whitman Gary J.,
Slanetz Priscilla J.,
Halpern Elkan F.
Publication year - 1998
Publication title -
the breast journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.533
H-Index - 72
eISSN - 1524-4741
pISSN - 1075-122X
DOI - 10.1046/j.1524-4741.1998.430139.x
Subject(s) - medicine , mammography , predictive value , breast cancer , biopsy , age groups , incidence (geometry) , age adjustment , cancer , demography , cancer detection , breast cancer screening , epidemiology , physics , sociology , optics
This study demonstrates how data can be, and have been inappropriately grouped. Analysts have suggested that many of the parameters of mammographic breast cancer screening (recall rates, recommendations for biopsy, the positive predictive value of a biopsy, and cancer detection rates) change abruptly at age 50. We evaluated these parameters in the Massachusetts General Hospital Screening program by specific ages to determine whether there was a change at this age. Of the total of 72,229 studies, 19,988 (29%) of the women were ages 40–49; 20,116 (29%) were ages 50–59; 19,268(25%) were ages 60–69; and 12,857 (17%) were ages 70–79. Among these, 4,509 (6%) women were recalled for additional evaluation; 832 (1%) biopsies were recommended; and 219 (0.3%) cancers were diagnosed with an overall positive predictive value for a mammographically initiated biopsy of 26%. There were no abrupt changes in any of the parameters at the age of 50. The recall rate declined slightly from approximately 7.3% at age 40 to 5.2% at age 79, while the rate of biopsies recommended was virtually constant with age varying from 1% to 2%. The positive predictive value and the yiield of cancers increased steadily with age rising from 15% at age 40 to 43% by the age of 79. The cancer detection rate for combined prevalence and incidence among the screened women rose from approximately 1.9 per 1,000 women at age 40 to 5.0 per 1,000 by age 79. The grouping of data can lead to misinterpretation of the results. Screening guidelines should not be predicated on the false assumption that these variables change abruptly at age 50.