Table 3.

Effect estimates for automated measures of mammographic density on case–control status, n = 3,830 (278 cases and 3,552 controls)

MIP, VPD, and PD one at a time in the modelMIP and VPD included together in the modelMIP and APD included together in the model
OutcomeEstimate (95% CI)PEstimate (95% CI)PEstimate (95% CI)P
(A)
VPD (raw)1.776 (1.295–2.438)3.7 × 10−41.470 (1.047–2.063)0.0259
APD (processed)1.340 (1.176–1.527)1.1 × 10−51.265 (1.107–1.448)6.1 × 10−4
MIP (processed)0.767 (0.674–0.871)4.6 × 10−50.811 (0.707–0.930)0.00280.816 (0.714–0.932)0.0027
(B)
VPD (raw)2.217 (1.476–3.336)1.3 × 10−41.959 (1.290–2.977)0.0016
APD (processed)1.474 (1.259–1.730)1.7 × 10−61.428 (1.219–1.675)1.1 × 10−5
MIP (processed)0.759 (0.660–0.871)9.4 × 10−50.791 (0.686–0.911)0.00120.784 (0.682–0.902)6.6 × 10−4

NOTE: Estimates (point estimates and 95% CIs) are presented as ORs. A, with full adjustment (age, BMI, menopausal status, HRT use, parity, age at first birth, and machine type) and B, full adjustment plus additionally adjusting for acquisition parameters. OR estimates, CI estimates, and P values (Wald tests) are based on estimated coefficients for VPD, APD, and MIP in logistic regression models with case–control status as outcome.