Table 4.
Body mass index (BMI) stratified, multivariable polytomous logistic regression for the association of race with mammographic density (referent outcome is almost entirely fatty) among 37 839 women*
| Variable | No. participants† | Extremely dense, No. (%) |
Heterogeneously dense, No. (%) |
Scattered fibroglandular tissue, No. (%) |
Almost entirely fatty, No. (%) (Referent) |
|||
|---|---|---|---|---|---|---|---|---|
| 1597 (4.2) |
10 630 (29.1) |
20 686 (54.7) |
4926 (13.0) |
|||||
| %‡ | OR (95% CI) | %‡ | OR (95% CI) | %‡ | OR (95% CI) | %‡ | ||
| BMI categories underweight and normal weight, BMI < 25.0 kg/m2§ | ||||||||
| Race | ||||||||
| Non-Hispanic white | 8650 | 11.4 | 1.00 (Referent) | 47.8 | 1.00 (Referent) | 38.7 | 1.00 (Referent) | 2.1 |
| African American | 2249 | 11.8 | 0.82 (0.59 to 1.16) | 40.3 | 0.61 (0.45 to 0.84) | 44.3 | 0.75 (0.55 to 1.02) | 3.6 |
| Among postmenopausal | ||||||||
| Non-Hispanic white | 6031 | 7.8 | 1.00 (Referent) | 43.8 | 1.00 (Referent) | 45.8 | 1.00 (Referent) | 2.7 |
| African American‖ | 1612 | 8.8 | 0.95 (0.65 to 1.40) | 36.3 | 0.62 (0.44 to 0.88) | 50.4 | 0.74 (0.53 to 1.04) | 4.4 |
| BMI category overweight, BMI 25.0–30.0 kg/m2§ | ||||||||
| Race | ||||||||
| Non-Hispanic white | 6867 | 2.1 | 1.00 (Referent) | 31.5 | 1.00 (Referent) | 59.1 | 1.00 (Referent) | 7.4 |
| African American | 4076 | 2.8 | 1.54 (1.13 to 2.10) | 29.9 | 1.01 (0.85 to 1.20) | 59.3 | 1.02 (0.87 to 1.20) | 8.0 |
| Among postmenopausal | ||||||||
| Non-Hispanic white | 5124 | 1.2 | (Referent) | 27.1 | (Referent) | 62.9 | (Referent) | 8.8 |
| African American‖ | 2985 | 1.8 | 1.83 (1.19 to 2.83) | 24.8 | 1.03 (0.85 to 1.24) | 63.8 | 1.06 (0.89 to 1.26) | 9.7 |
| BMI category obese, BMI > 30.0 kg/m2§ | ||||||||
| Race | ||||||||
| Non-Hispanic white | 7649 | 0.5 | 1.00 (Referent) | 14.4 | 1.00 (Referent) | 62.9 | 1.00 (Referent) | 22.2 |
| African American | 8347 | 0.6 | 1.21 (0.76 to 1.93) | 13.3 | 0.89 (0.79 to 0.99) | 60.6 | 0.87 (0.80 to 0.94) | 25.5 |
| Among postmenopausal | ||||||||
| Non-Hispanic white | 5514 | 0.4 | 1.00 (Referent) | 12.0 | 1.00 (Referent) | 63.2 | 1.00 (Referent) | 24.4 |
| African–American‖ | 5714 | 0.5 | 1.36 (0.73 to 2.54) | 11.4 | 0.98 (0.85 to 1.13) | 61.0 | 0.93 (0.85 to 1.02) | 27.0 |
Fully adjusted for all covariates, model includes parity and age at first birth, age group, BMI at age 18 years, age at menarche, family history of breast cancer, alcohol use, history of oral contraceptive use, and menopausal status. The results from the polytomous logistic regression can be interpreted as the log odds of either extremely dense, heterogeneously dense, or scattered fibroglandular tissue compared with the log odds of almost entirely fatty (the referent mammographic breast density category). Results using fully conditional specification (FCS) multiple imputation. BMI = body mass index; CI = confidence interval.
Approximated strata population derived from multiple imputations.
Presented as proportion within variable strata with breast density category. No. (%) represents the total number of participants within strata and proportion of total population.
P value for multiplicative interaction between race and BMI categories <.001.
Among postmenopausal women only (underweight and normal weight, n = 7643; overweight, n = 8108; obese, n = 11 229). Models among postmenopausal women only further adjusted ever hormone replacement therapy and age at menopause.