Table 3.

Patient ratings of physician communication^*

Preferred role	Actual role	N (%)	Adjusted % providing top rating	OR (95% CI)	P
All Patients*		4825 (100)	55.8%
Patient-controlled		1739 (36)	55.3%	0.93 (0.81–1.06)	0.27
Shared		2799 (58)	57.0%	1.0	Ref
Physician-controlled		287 (6)	47.5%	0.67 (0.51–0.87)	0.002
	Patient-controlled	1914 (40)	57.4%	0.98 (0.86–1.11)	0.75
	Shared	2264 (47)	57.9%	1.0	Ref
	Physician-controlled	647 (13)	43.7%	0.55 (0.45–0.66)	<0.001
Effect of Actual Role Stratified by Preferred Role
Patient-controlled
	Patient-controlled	1067 (22)	57.0%	1.03 (0.83–1.29)	0.78
	Shared	494 (10)	56.2%	1.0	Ref
	Physician-controlled	178 (4)	44.8%	0.62 (0.43–0.89)	0.009
Shared control
	Patient-controlled	785 (16)	58.5%	0.95 (0.80–1.14)	0.60
	Shared	1674 (35)	59.6%	1.0	Ref
	Physician-controlled	340 (7)	43.3%	0.50 (0.39–0.64)	<0.001
Physician-controlled
	Patient-controlled	62 (1)	50.9%	1.22 (0.58–2.57)	0.59
	Shared	96 (2)	46.4%	1.0	Ref
	Physician-controlled	129 (3)	35.6%	0.61 (0.33–1.13)	0.12

^*P for interaction of preferred and actual role=0.76, so only main effects model is presented in the top box; stratified results are also presented. These multivariable logistic regression models also adjusted for all variables in Table 1.