Table 3.

Linear mixed models analysis of the effect of academic detailing on physician recommendations for (1) mammography at follow-up and (2) clinical breast examination (CBE) at follow-up (N = 710 for both)*

	Effect of Academic Detailing on Physician Recommendation of Mammography at Follow-up						Effect of Academic Detailing on Physician Recommendation of CBE at Follow-up
	Beta^†	SE	t	OR^‡	95% CI^§	P	Beta^†	SE	t	OR^‡	95% CI^§	P
Years of practice^‖							−0.003	0.006	−0.48			.63
Medical school	0.19	0.23	0.86			.39	−0.40	0.30	−1.35			.18
Percentage of patients insured by Medicaid or Medicare	−0.03	0.006	−4.47			<.0001	−0.04	0.007	−5.26			<.0001
Barriers to screening^‖							−0.48	0.26	−1.87			.06
Knowledge of risk factors for breast cancer	1.32	0.23	5.79			<.0001	1.62	0.24	6.83			<.0001
Age (40 to 49, >50)				0.97	0.58, 1.62	.91				1.09	0.57, 2.05	.80
Baseline mammography proportion				5.17	3.15, 8.46	<.0001	n/a
Baseline CBE proportion	n/a									1.37	0.73, 2.56	.33
Intervention^¶^**				1.85	1.25, 2.74	.002				2.13	1.31, 3.46	.002

* Using medical audit data, after completion of academic detailing intervention; analyzed via GLIMMIX in SAS, to account for clustering of patients.
† Beta is the standardized regression coefficient, ie, a standardized measure of the change in outcome attributable to one predictor with the remaining predictors held constant.68
‡ The reference is the control group.
§ 95%confidence intervals.
‖ The contribution of this factor was null so it was deleted from the model.
¶ Intervention model with mammography screening as outcome adjusted for baseline rates of recommendations for mammography screening, whether attended US medical school, follow-up knowledge of breast cancer risk factors, follow-up barriers to breast cancer screening, percentage of patients enrolled in Medicare or Medicaid, percentage of patients enrolled in other insurance.
** Intervention model with CBE as outcome adjusted for baseline CBE rates, number of years of practice, whether attended US medical school, follow-up knowledge of breast cancer risk factors, follow-up barriers to breast cancer screening, percentage of patients enrolled in Medicare or Medicaid, percentage of patients enrolled in other insurance.