Table 4.

Association between type of biopsy and subsequent mammography performance measures^*

Mammography performance measures	Documented FNA or cyst aspirations,† OR (95% CI)	Documented core biopsies,† OR (95% CI)	Documented excisional biopsies, surgical biopsy, or not otherwise specified,† OR (95% CI)	P‡
Sensitivity§	0.9 (0.55 to 1.47)	1.3 (0.87 to 2.16)	1.2 (0.56 to 3.07)	.5306
Specificity║	0.58 (0.55 to 0.61)	0.69 (0.66 to 0.72)	0.66 (0.61 to 0.71)	<.001
PPV1¶	1.05 (0.82 to 1.31)	1.30 (1.09 to 1.54)	1.17 (0.8 to 1.64)	<.05
PPV2#	0.95 (0.70 to 1.28)	0.70 (0.57 to 0.86)	0.99 (0.61 to 1.55)	<.01

^*Odds ratios (ORs) and 95% confidence intervals (CIs) for a correct interpretation when comparing screening mammograms with prior documented biopsies (fine needle aspiration [FNA] or cyst aspiration, core biopsies, excisional biopsies, surgical biopsy, or not otherwise specified) with screening mammograms without a biopsy history were computed from multivariable logistic regression models including age at mammogram (in 5-year intervals), months since last mammogram, breast density, availability of comparison film, mammography registry, and year of examination as covariates. The model for sensitivity used fewer categories for age at mammogram and months since last mammogram and did not include the availability of comparison film because of small numbers. The analysis was based on 1 703 328 screens among women without a biopsy history (7119 associated with a subsequent cancer), 13 339 screens among women with a prior documented FNA or cyst aspiration (99 associated with a subsequent cancer), 22 889 screens with prior documented core biopsies (167 associated with a subsequent cancer), and 7682 screens among women with documented prior excisional biopsies, surgical biopsy, or not otherwise specified (40 associated with a subsequent breast cancer). PPV1 = positive predictive value 1; PPV2 = positive predictive value 2.

^†Based on biopsies documented in the cancer registries or pathology databases before screening mammograms performed from January 2, 1996, through December 31, 2005.

^‡Two-sided P values for overall association of biopsy history with performance measure in the multivariate models were calculated using likelihood ratio tests.

^§Odds of a positive screen (Breast Imaging Reporting and Data System [BI-RADS] = 0, 4, 5, or 3 with recommendation for additional evaluation) among screening mammograms from women with a subsequent cancer diagnosis (sensitivity), comparing examinations from women with a documented biopsy (FNA or cyst aspirations, core biopsies, excisional biopsies) with screening mammograms from women without a biopsy history.

^║Odds of a negative screen (BI-RADS assessment = 1, 2, or 3 without immediate follow-up) among screening mammograms from women without a subsequent cancer diagnosis (specificity), comparing screening mammograms from women with a documented biopsy (FNA or cyst aspirations, core biopsies, excisional biopsies) with screening mammograms from women without a biopsy history.

^¶Odds of a subsequent cancer among positive screening mammograms (BI-RADS = 0, 4, 5, or 3 with recommendation for additional evaluation) (PPV1), comparing screening mammograms from women with a documented biopsy (FNA or cyst aspirations, core biopsies, excisional biopsies) with screening mammograms from women without a biopsy history.

^#Odds of a subsequent cancer among screening mammogram with a recommendation for biopsy (BI-RADS assessment = 4, 5) (PPV2), comparing screening mammograms from women with a documented biopsy (FNA or cyst aspirations, core biopsies, excisional biopsies) with screening mammograms from women without a biopsy history.