Table 4.

Adjusted change in annual ICS days-of-supply for lower and higher-cost ICS users in 2003: restricted vs. unrestricted coverage

	Change in ICS Days-of-Supply from 2003 to 2004		Change in ICS Days-of-Supply from 2003 to 2004: Restricted vs. Unrestricted Coverage (Difference-in-difference)
	Restricted Coverage	Unrestricted Coverage	Restricted vs. Unrestricted	(95% CI)	p-value
All Subjects (n = 1,802)	-30.27	-14.81	-15.47	(-24.98, -5.95)	0.001
Lower-cost ICS Users in 2003 (n = 1,140)	-26.38	-13.55	-12.84	(-24.34, -1.33)	0.029
Continued using lower-cost ICS in 2004 (n = 960)	-13.08	0.63	-13.71	(-25.52, -1.90)	0.023
Switched to higher-cost ICS in 2004 (n = 25)	-54.24	29.51	-83.75	(-171.58, 4.07)	0.060
Higher-cost ICS Users in 2003 (n = 662)	-37.40	-15.60	-21.80	(-38.46, -5.13)	0.010
Continued using higher-cost ICS in 2004 (n = 354)	-39.81	-1.19	-38.62	(-56.96, -20.28)	< .001
Switched to lower-cost ICS in 2004 (n = 212)	0.71	-18.04	18.75	(-27.46, 64.96)	0.425

Notes: Each row represents a separate OLS regression model. Among lower-cost ICS users in 2003, 155 subjects had no ICS drug use in 2004, among higher-cost ICS users in 2003, 96 subjects had no ICS drug use in 2004. We calculated mean changes in ICS days-of-supply for each coverage group by using the lincom command in Stata 8.2 and the mean levels of the covariables.