Table 5.

Linear Mixed-Effects Longitudinal Model^a: Endometrial Thickness^b Response Across Time Points (N=208, compliant model)

Outcome Variable	Independent Variable	Parameter Estimate	Standard Error	pValuec
Endometrial Thicknessb	Intercept	−0.401	0.262	0.13
	Site (ISU=0)	0.528	0.049	≤0.0001
	Treatment (80 mg)	0.032	0.056	0.57
	Treatment (120 mg)	0.043	0.055	0.43
	Time (mo)	−0.008	0.001	≤0.0001
	Estrogen Exposured (y)	0.022	0.007	0.0013
	Plasma 17-β Estradiol (pg/mL)	0.006	0.002	0.0086
	Alcohol Intakee (g/week)	−0.001	0.0004	0.023
	Serum Thyroid-Stimulating Hormone (µIU/L)	−0.020	0.011	0.066

^aOverall model vs model with obligatory design variables only: likelihood ratio = 12.149; df=4; p≤0.0001. Likelihood ratio based on maximum likelihood estimation; estimates and significances based on restricted maximum likelihood estimation.

^bNatural logarithm of endometrial thickness as variance stabilizing transformation.

^cP values are based on asymptotic normality of t-distribution; variables left in model were significant at p≤0.10 level (acceptable Type I error rate in modeling endometrial thickness). Additional variables (i.e., body weight, lactation duration, etc.) did not remain in model. The interaction between treatment and time (log likelihood ratio=3.996; df=2) was not significant (p=0.14).

^dCalculated at baseline, estrogen exposure in years = age at menopause – age at menarche; estrogen exposure was not related to plasma 17-β estradiol (Pearson correlation coefficient = −0.05).

^eRepresents usual weekly alcohol intake at baseline, which was assessed at baseline from a dietary questionnaire.