Table 3.

Relationship between aspirin use and dichotomized breast density as the outcome, including interactions between aspirin use and age and race.

Effect Modification by Age
	OR (95% CI)* for non-dense vs. dense
Aspirin use at age 40-49 (n=7784)	0.48 (0.37 – 0.63)
Aspirin use at age 50-59 (n=7944)	0.77 (0.66 – 0.90)
Aspirin use at age 60-69 (n=6658)	0.87 (0.76 – 0.98)
Aspirin use at age ≥70 (n=3614)	0.93 (0.79 – 1.09)
	Pinteraction = 0.0002**
Effect Modification by Race
	OR (95% CI)* for non-dense vs. dense
Aspirin use for race = white (n=13519)	0.89 (0.80 – 0.98)
Aspirin use for race = black (n=10662)	0.70 (0.62 – 0.79)
Aspirin use for race = other (n=1819)	0.78 (0.57 – 1.08)
	Pinteraction = 0.011***

^*Logistic regression results for a models with dichotomized density (dense=BI-RADS 3,4 and non-dense=BI-RADS 1,2), adjusted for age, race, and BMI.

^**Chi-square likelihood ratio test (df. = 3) comparing the logistic regression model with and without interaction between aspirin use and age.

^***Chi-square likelihood ratio test (df. 2) comparing the logistic regression model with and without interaction between aspirin use and race.