Table 2.

Regression results for costs of consequences (log scale, with TEL as reference)

Independent variables*	Co-efficient (β)	Excess risk, Exp(β)-1	t-statistic	Statistical significance
STD	1.57	3.79	2.59	< 0.01
RP	0.78	1.19	1.26	n.s.
Log baseline costs of consequences	0.40	0.50	4.76	<0.0001
Log baseline costs of consequences x STD	−0.23	−0.21	−2.18	<0.05
Log baseline costs of consequences x RP	−0.13	−0.12	−1.18	n.s.

Notation: STD denotes standard continuing care program; RP denotes relapse prevention program, TEL denotes telephone-based continuing care program (reference). n.s. denotes not statistically significant, log denotes natural logarithm. Regressions were run with SAS procedure PROC MIXED.

^*Earlier models tested the following independent variables, but dropped them from the final model because they proved to be not statistically significant: Treatment condition x Log baseline cost of consequences, Log income, Log baseline income, Treatment condition x Log baseline income, Baseline investment, Treatment condition x Log baseline investment, Program, Treatment condition, and Sequence