Table 2.

Likelihood of Patients treated within the Same Physician Group or by the Same Physician to Receive BMD Testing or Prescription Osteoporosis Medications

	Two Level Model (Physician Group and Patients)		Three Level Model (Physician Group, Physicians, and Patients)
	Crude OR (95% CI)	Adjusted* OR (95% CI)	Crude OR (95% CI)	Adjusted* OR (95% CI)
Glucocorticoid users, receipt of DXA
Physician Group ICC	1.61(1.30 – 1.98)	1.55 (1.16 – 2.10) 0.12	1.26 (1.09 – 1.44)	1.12 (0.98 – 1.28) 0.04
Individual Physician ICC	–	–	1.87 (1.39 – 2.52)	1.87 (1.24 – 2.84) 0.11
Glucocorticoid users, receipt of OP Rx
Physician Group ICC	1.41(1.20 – 1.67)	1.36 (1.18 – 1.57) 0.10	1.22 (0.88 - 1.69)	1.14 (0.87 – 1.49) 0.02
Individual Physician ICC	–	–	1.56 (1.34 – 1.82)	1.53 (1.33 – 1.77) 0.13
Nursing home residents, Receipt of OP Rx
Physician Group ICC	1.18 (1.02 – 1.36)	1.12 (1.00 – 1.25) 0.04	1.08 (0.91 – 1.30)	1.04 (0.95 – 1.14) 0.01
Individual Physician ICC	-	-	1.32 (1.04 – 1.67)	1.22 (0.98 – 1.51) 0.05

ICC = Intra-class correlation coefficient

These pair-wise odds-ratios (ORs) are referent to patients treated by different physicians and were computed using Alternating Logistic Regression (ALR).

^*results for long term glucocorticoid users were adjusted for the number of physicians in the group; physician covariates including specialty, age, gender, and years since obtaining medical degree; and patient covariates including sex, age, prior fracture, number of provider visits, number of comorbid conditions, and prednisone use (cumulative dose and new vs. prevalent use).

Results for the nursing home residents were adjusted for facility-level covariates including the number of beds in the facility (i.e. size), for-profit status, rural (vs. urban), and proportion of Medicare patients; and patient covariates including gender, age, race, whether ambulatory, and history of falling, cognitive impairment, GERD, dysphagia, esophogitis or peptic ulcer, breast cancer, thromboembolic disease, tobacco use, and alcohol abuse.