Table 2.

Estimates (SE) of residual variance (σ²), of monozygous and dizygous covariances (σ_monozygous² and σ_dizygous², respectively), of monozygous and dizygous correlations (ρ_monozygous and ρ_dizygous, respectively), and of heritability (H) under the AE model on log dense area and log dense area, separately

Estimate	Age adjusted for mean	Mean adjusted for age and covariates*	Random effect a function of age^†
Log dense area
σ²	0.63 (0.023)	0.60 (0.022)	0.43 + 0.015 (age 40)
σ_monozygous²	0.43 (0.025)	0.39 (0.024)	0.31 + 0.007 (age 40)
σ_dizygous²	0.16 (0.030)	0.13 (0.029)	0.09 + 0.004 (age 40)
ρ_monozygous	0.67 (0.021)	0.66 (0.022)	0.73 − 0.005 (age 40)
ρ_dizygous	0.25 (0.046)	0.21 (0.047)	0.21 + 0.001 (age 40)
H	0.66 (0.022)	0.65 (0.024)	0.71 (0.039)
−2 Log-likelihood	668.1	585.8	531.8
Log nondense area
σ²	0.45 (0.017)	0.24 (0.0089)	0.28 − 0.004 (age 40)
σ_monozygous²	0.34 (0.018)	0.16 (0.0096)	0.19 − 0.002 (age 40)
σ_dizygous²	0.15 (0.021)	0.05 (0.012)	0.06 − 0.001 (age 40)
ρ_monozygous	0.76 (0.017)	0.67 (0.022)	0.67 − 0.0002 (age 40)
ρ_dizygous	0.34 (0.042)	0.21 (0.044)	0.23 − 0.002 (age 40)
H	0.75 (0.017)	0.66 (0.023)	0.67 (0.036)
−2 Log-likelihood	−162.7	−1,165.3	−1,178.8

↵* Adjusted covariates for mean log dense area were age, height, weight, age at menarche, number of live births, years of oral contraception use, years of hormone replacement therapy use, menopausal status, and population (Australian vs North American). Adjusted covariates for mean log nondense area were age, height, weight, age at menarche, number of live births, age at first birth, pack years of smoking, and population (Australian vs North American).
↵† These analyses allowed the random effects variables to vary as a linear function of age (centered on 40 y).