Table 3.

Pooled sample^* to test whether the joint effect of age at exposure and DDT differ according to age at breast cancer (BC) diagnosis

Age at first exposure, y	Study 1	Study 2	P difference ‡
	Premenopausal BC	Early postmenopausal BC
	diagnosis ages <50 y	diagnosis ages 50–54 y
	ORlog2DDT† (95% CI)	ORlog2DDT† (95% CI)	By age at diagnosis	By age at exposure and diagnosis
<3	3.70 (1.22 to 11.26)	0.92 (0.52 to 1.63)	.03§	.03#
3–13	5.16 (1.92 to 13.82)	1.88 (1.37 to 2.59)	.59‖	.049**
14+	0.98 (0.51 to 1.88)	2.26 (1.22 to 4.20)	.06¶	Reference

^*Pooled sample includes Study1: cases diagnosed by age 50 years (n = 129) and year-of-birth matched controls (n = 129) and Study 2: cases diagnosed from 50 to 54 years (n = 153) and year-of-birth matched controls (n = 432). BC, breast cancer; CI, confidence interval; OR, odds ratio estimated by conditional logistic regression.

^†Model includes: p, p'-DDT(log₂-transformed as a continuous variables), o, p'-DDT(log₂-transformed as a continuous variable), year of blood draw (continuous) and parity (continuous), 3 two-way product terms: log₂(p, p'-DDT) X age at exposure (dichotomized as <3 years vs. 14+ years), log₂(p, p'-DDT) X age at exposure (dichotomized as 3-13 years vs. 14+ years), and log₂(p, p'-DDT) X age at diagnosis study (in <50 years study vs. in 50-54 years study), and 2 three-way product terms: log₂(p, p'-DDT) X age at exposure (dichotomized as <3 years vs. 14+ years) X age at diagnosis study (in <50 years study vs. in 50-54 years study) and log₂(p, p'-DDT) X age at exposure (dichotomized as 3-13 years vs. 14+ years) X age at diagnosis study (in <50 years study vs. in 50-54 years study). Odds Ratios (ORs) presented here are estimated from contrasts calculated using linear combinations of relevant terms from this model. The DDT OR represents a one-unit change in log₂(p, p'-DDT), corresponding to an estimated effect for a twofold increase in p, p'-DDT, a range encompassed within the interquartile range of the study sample (see Table 1).

^‡ P values were based on probability greater than chi-square tested in conditional logistic regression models in SAS 9.3. All tests were two-sided.

^§ P value resulting from test of product term: log₂(p, p’-DDT) × age at first exposure (dichotomized as <3 years vs 14+ years).

^‖ P value resulting from test of product term: log₂(p, p’-DDT) × age at first exposure (dichotomized as 3–13 years vs 14+ years).

^¶ P value resulting from test of product term: log₂(p, p’-DDT) × age at diagnosis study (in <50 years study vs in 50–54 years study).

^# P value resulting from test of product term: log₂(p, p’-DDT) × age at first exposure (dichotomized as <3 years vs 14+ years) × age at diagnosis study (in <50 years study vs in 50–54 years study).

^** P value resulting from test of product term: log₂(p, p’-DDT) × age at first exposure (dichotomized as 3–13 years vs 14+ years) × age at diagnosis study (in <50 years study vs in 50–54 years study).