Table 4. Multivariable models for regression of HT prescribing frequency on provider characteristics: model 1 for primary care providers and model 2 for OB/GYNs ^a.

Characteristics	Beta = Adjusted Difference in HT Prescribing Frequency	95% Confidence Interval	P value
Model 1: Primary Care Provider Only
Northeast (vs. Northwest)	-7.19	-9.18 to -5.20	<0.001
Years at health care organization	0.15	0.04 to 0.25	0.01
Average age of patient at visit (years)	-0.04	-0.30 to 0.23	0.81
Feel well prepared to counsel	-1.47	-3.85 to 0.91	0.18
Expert knowledge of trials	1.14	0.56 to 1.73	<0.001
Risk of EP appropriately described (vs. exaggerated)	-0.20	-0.74 to 0.34	0.48
WHI results relevant to women in my practice	0.26	-0.49 to 1.01	0.50
Agree women want to come off HT	-0.37	-1.00 to 0.26	0.25
Agree a scientific case has been made for colon cancer prevention	0.42	-0.48 to 1.33	0.33
Model 2: OB/GYNs Only
Northeast (vs. Northwest)	-11.87	-18.97 to -4.77	0.001
Years at health care organization	0.60	0.18 to 1.02	0.01
Average age of patient at visit (years)	-1.69	-3.16 to -0.22	0.02
Feel well prepared to counsel	10.15	3.03 to 17.27	0.01
Expert knowledge of trials	-0.38	-3.82 to 3.07	0.78
Risk of EP appropriately described (vs. exaggerated)	-1.96	-3.85 to -0.07	0.04
WHI results relevant to women in my practice	-1.21	-3.52 to 1.11	0.33
Agree women want to come off HT	0.81	-1.68 to 3.31	0.49
Agree a scientific case has been made for colon cancer prevention	-0.94	-4.34 to 2.47	0.56

^aP-values from linear regression model with provider prescribing frequency as the outcome and all variables listed in models as independent variables. All variables with p<0.1in the bivariate analyses were included in the multivariate models. Number of providers included in the regression analyses: 173 for model 1 (PCP) and 42 for model 2 (OB/GYN). Regression coefficients reflect mean difference in number of HT prescription days/visit for unit change in provider characteristic after controlling for other variables.