LIKELIHOOD RATIOS FOR MODERN SCREENING MAMMOGRAPHY - RISK OF BREAST-CANCER BASED ON AGE AND MAMMOGRAPHIC INTERPRETATION

Objective.-To determine the sensitivity, specificity, and likelihood r atios (LRs) for modern screening mammography by decade of age and mamm ographic interpretation. Design.-Cross-sectional. Setting.-Nine counti es in northern California. Participants.-A total of 26 057 women aged 30 years and older who underwent a total of 41 747 first and subsequen t screening mammographic examinations at the Mobile Mammography Screen ing Program of the University of California, San Francisco, from April 1985 to September 1991. Measurements.-Breast cancer risk profile, 2 s tandard mammographic views per breast, and follow-up of abnormal and n ormal mammograms by contacting women's physicians and by linkage to th e regional Surveillance, Epidemiology, and End Results tumor registry. False-negative examinations were normal examinations that occurred wi thin 13 months of a diagnosis of invasive breast cancer or ductal carc inoma in situ. Results.-The sensitivity of first screening mammography increased with age: 77.3% for ages 30 to 39 years, 86.7% for ages 40 to 49 years, 93.6% for ages 50 to 59 years, 94.1% for ages 60 to 69 ye ars, and 91.?% for ages 70 years and older (P=.04). Specificity was si milar for all ages, ranging from 92.6% to 95.2%. Of all abnormal first screening examinations, 92.9% were reported as ''additional evaluatio n needed.'' The LRs for that category ranged from 5.2 to 8.8 and did n ot vary with age. Based on the risk of breast cancer before mammograph y, which increases with age, the risk of breast cancer after mammograp hy associated with these LRs were 0.01 for ages 30 to 39 years, 0.02 f or ages 40 to 49 years, 0.05 for ages 50 to 59 years, 0.07 for ages 60 to 69 years, and 0.07 for ages 70 years and older. The LRs for mammog raphy reported as ''suspicious for malignancy'' ranged from 88 to 144 and did not vary across age groups. These LRs were associated with a r isk of breast cancer about 10 times greater than when mammography was reported as ''additional evaluation needed.'' Conclusions.-Most abnorm al first screening mammography are interpreted as ''additional evaluat ion needed'' and are associated with LRs of about 7. Given this low LR , the risk of breast cancer after mammography is primarily influenced by a woman's age-specific risk of breast cancer before mammography. Th e LRs for screening mammography interpreted as ''suspicious for malign ancy'' are high (about 124) and are associated with a substantial incr ease in the risk of breast cancer irrespective of age, but these inter pretations comprise only a small proportion of abnormal mammagraphy.