Likelihood ratios for modern screening mammography. Risk of breast cancer based on age and mammographic interpretation

Abstract

Objective: To determine the sensitivity, specificity, and likelihood ratios (LRs) for modern screening mammography by decade of age and mammographic interpretation.

Design: Cross-sectional.

Setting: Nine counties in northern California.

Participants: A total of 26,057 women aged 30 years and older who underwent a total of 41,747 first and subsequent screening mammographic examinations at the Mobile Mammography Screening Program of the University of California, San Francisco, from April 1985 to September 1991.

Measurements: Breast cancer risk profile, 2 standard mammographic views per breast, and follow-up of abnormal and normal mammograms of contacting women's physicians and by linkage to the regional Surveillance, Epidemiology, and End Results tumor registry. False-negative examinations were normal examinations that occurred within 13 months of a diagnosis of invasive breast cancer or ductal carcinoma 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 years, and 91.2% for ages 70 years and older (P = .04). Specificity was similar for all ages, ranging from 92.6% to 95.2%. Of all abnormal first screening examinations, 92.9% were reported as "additional evaluation needed." The LRs for that category ranged from 5.2 to 8.8 and did not vary with age. Based on the risk of breast cancer before mammography, which increases with age, the risk of breast cancer after mammography associated with these LRs were 0.01 for ages 30 to 39 years, 0.02 for 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 mammography reported as "suspicious for malignancy" ranged from 88 to 144 and did not vary across age groups. These LRs were associated with a risk of breast cancer about 10 times greater than when mammography was reported as "additional evaluation needed."

Conclusion: Most abnormal first screening mammography are interpreted as "additional evaluation 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. The LRs for screening mammography interpreted as "suspicious for malignancy" are high (about 124) and are associated with a substantial increase in the risk of breast cancer irrespective of age, but these interpretations comprise only a small proportion of abnormal mammography.