Table 3.
Effect of covariates on change in dense breast volume across menopause. (N=2568 subjects with complete BMI data)
| Annualized Change in DV (95% CI)* cm3 | interaction p-value | |
|---|---|---|
| Overall Change | −2.22 (−2.74, −1.71) | NA |
| Premenopausal DV | ||
| Below median (<=54 cm3) | −0.97 (−1.71, −0.22) | <0.0001 |
| Above median (>54 cm3) | −3.46 (−4.40, −2.73) | |
| Premenopausal BMI | ||
| Normal (<25 kg/m2) | −2.31 (−3.84, −0.77) | 0.56 |
| Overweight (25–29 kg/m2) | −2.02 (−2.65, −1.40) | |
| Obese (>=30kg/m2) | −2.72 (−3.84, −1.61) | |
| Postmenopausal Hormone Therapy | ||
| Not Current | −2.14 (−2.69, −1.60) | 0.18 |
| Current | −3.58 (−5.57, −1.59) | |
| Unknown | −4.29 (−7.43, −1.15) | |
| Postmenopausal Hormone Therapy (among those known) | ||
| Not Current | −2.13 (−2.66, −1.59) | 0.64 |
| Current Estrogen | −3.16 (−5.60, −0.72) | |
| Current Estrogen + Progesterone | −3.09 (−6.68, 0.50) | |
| Family History of Breast cancer (among those known) | ||
| No family history | −2.25 (−2.83, −1.68) | 0.87 |
| Family History | −2.14 (−3.33, −0.95) | |
| Parity (among those known) | ||
| Nulliparous | −2.27 (−3.17, −1.37) | 0.88 |
| Parous | −2.18 (−2.81, −1.55) | |
| Age at First Birth (among those known) | ||
| Nulliparous | −2.27 (−3.16, −1.37) | 0.66 |
| <30 Years | −2.55 (−3.57, −1.53) | |
| 30+ Years | −1.96 (−2.76, −1.16) | |
| Race (among those known) | ||
| Caucasian | −2.42 (−3.12, −1.73) | 0.30 |
| Asian | −1.68 (−2.56, −0.80) | |
| Other | −2.91 (−4.59, −1.23) | |
| Alcohol Use (among those known) | ||
| None | −2.06 (−2.79, −1.33) | 0.12 |
| <=1/day | −2.79 (−3.63, −1.95) | |
| >=2 /day | −1.01 (−2.59, 0.57) | |
| Previous Biopsies | ||
| None | −2.16 (−2.73, −1.58) | 0.36 |
| <=1/day | −1.80 (−3.16, −0.43) | |
| >=2 /day | −3.55 (−5.60, −1.51) | |
*
Coefficients estimated by linear mixed models adjusted for age, BMI, log dense volume, BMI change across menopause, time from premenopausal mammogram and the interaction between dense volume and time from premenopausal mammogram.
**
The effect of each variable on change over time was estimated by fitting an interaction between time (years) and the premenopausal covariate of interest; p-values reflect overall interaction for each covariate.