Table 1.
Baseline variable | No. of women (%) | Follow-up, 1000 women-years | No. of invasive breast cancers | Hazard ratio* (95% CI) |
---|---|---|---|---|
Total | 132,439 (100) | 874 | 2704 | - |
Age (years) | ||||
40–49 | 60,325 (45.6) | 418 | 977 | 1 [Reference] |
50–59 | 43,878 (33.1) | 317 | 1057 | 1.41 (1.29–1.53) |
60–73 | 28,236 (21.3) | 139 | 670 | 2.26 (2.04–2.50) |
Continuous | Median (IQR) = 50 (44–58) | LR-χ2(df = 1) = 309.1 (p < 0.001) | ||
BMI (kg/m2): < 55 years | ||||
Underweight (< 18.5) | 1087 (1.3) | 7 | 13 | 0.70 (0.41–1.22) |
Healthy (≥ 18.5 to 25) | 33,808 (38.8) | 246 | 623 | 1 [Reference] |
Overweight (≥ 25 to 30) | 26,428 (30.3) | 186 | 513 | 1.10 (0.98–1.23) |
Obese (≥ 30 to 35) | 13,126 (15.1) | 92 | 221 | 0.96 (0.82–1.11) |
Morbidly obese (≥ 35) | 12,707 (14.6) | 87 | 211 | 0.98 (0.84–1.15) |
Continuous | Median (IQR) = 26.6 (23.0–31.2) | LR-χ2(df = 1) = 0.02 (p = 0.89) | ||
BMI (kg/m2): ≥ 55 years | ||||
Underweight (< 18.5) | 570 (1.3) | 3 | 12 | 1.18 (0.66–2.10) |
Healthy (≥ 18.5 to 25) | 14,905 (32.9) | 85 | 308 | 1 [Reference] |
Overweight (≥ 25 to 30) | 16,440 (36.3) | 92 | 449 | 1.35 (1.17–1.56) |
Obese (≥ 30 to 35) | 7665 (16.9) | 44 | 194 | 1.22 (1.02–1.46) |
Morbidly obese (≥ 35) | 5703 (12.6) | 33 | 160 | 1.33 (1.10–1.60) |
Continuous | Median (IQR) = 27.4 (23.8–31.0) | LR-χ2(df = 1) = 6.9 (p = 0.01) | ||
BI-RADS density | ||||
Fatty | 10,387 (7.8) | 61 | 107 | 0.59 (0.48–0.72) |
Scattered | 46,206 (34.9) | 309 | 786 | 1 [Reference] |
Heterogeneous | 57,158 (43.2) | 376 | 1338 | 1.76 (1.61–1.93) |
Extremely dense | 18,688 (14.1) | 128 | 473 | 2.31 (2.04–2.63) |
Continuous | Median (IQR) = 3 (2–3) | LR-χ2(df = 1) = 91.8 (p < 0.001) |
n = 132,439
95% CIs from Wald tests
LR-χ2(df = 1) is the likelihood-ratio trend test for each variable alone
BI-RADS Breast Imaging Reporting and Data System, BMI body mass index, df degrees of freedom, IQR interquartile range, 95% CI 95% confidence interval
*Hazard ratios from proportional-hazards Cox models (model for BI-RADS density is adjusted for continuous age and continuous BMI, models for age and BMI are univariable)