Table 3.

Association between Number of Surfaces Restored with Composite and Change from Baseline in BPA from Regression Models^*

	Type of Composite Surface Restored
	All Surfaces				Posterior Occlusal Surfaces
	N	eβ	95% CI	P-value	N	eβ	95% CI	P-value
Change from Baseline to Visit:
Next day after composite placement†	89	1.37	1.13, 1.67	0.002	89	1.47	1.18, 1.83	<0.001
14 days after composite placement	81	0.93	0.78, 1.16	0.64	81	0.94	0.75, 1.18	0.60
6 months after composite placement	77	0.93	0.81, 1.08	0.35	77	0.88	0.74, 1.04	0.13

^*The regression models analyzed the outcome of log-transformed BPA corrected for specific gravity. Values presented as e^β are the back-transformed coefficients. Models adjusted for baseline BPA concentration and household income as fixed covariates, and level of canned food consumption in the day prior to urine sample collection and season of urine sample collection as time-varying covariates. The available sample sizes for BPA concentration measured after treatment visits 2 and 3 were insufficient for modeling analyses.

^†Results in table included all participants. Excluding the participant with 8 surfaces restored at visit 1 and with high next-day BPA concentration of 81.8 ng/mL, results for next day were: all surfaces e^β = 1.08 (95% CI 0.81, 1.47), P=0.57; posterior-occlusal surfaces e^β =1.19 (95% CI 0.86, 1.64), P=0.29.