Table 2.

A multilevel regression model associating continuous-valued treatment response with writing activity during different worry diary task assignments (n=1686). Number of observations (GAD-7 measurements)=13,205.

Effect			Estimatea		95% CI		P value
Fixed effects
	Intercept	−0.087		−0.260 to 0.087		.33
	Session number	−0.147		−0.158 to −0.135		<.001
	Treatment moderation effect, entries 1 (change/session)b	−0.001		−0.002 to 0.001		.23
	Treatment moderation effect, entries 2 (change/session)	0.002		0.000 to 0.003		.03
	Treatment moderation effect, entries 3 (change/session)	0.005		0.001 to 0.010		.02
	Baseline effect, entries 1	0.014		0.004 to 0.023		.006
	Baseline effect, entries 2	−0.003		−0.017 to 0.010		.61
	Baseline effect, entries 3	−0.031		−0.066 to 0.004		.08
	Age	−0.009		−0.013 to −0.005		<.001
	Sex	0.086		−0.029 to 0.201		.14
Random effects
	Residual variance	0.64		—c		—
	Between-patients intercept standard deviation	0.90		—		—
	Between-patients slope standard deviation	0.10		—		—
	Intercept-slope correlation	−0.15		—		—

^aEstimate: regression coefficient.

^bThe interaction between session number and each entry variable was interpreted as the treatment moderation effect.

^cNot applicable.