Table 2.

Regression coefficients, 95% CI, and P values of the linear mixed-effect models estimating root-mean-square of successive differences between normal heartbeats (RMSSD), high frequency (HF), self-reported craving, and self-reported anxiety during alcohol exposure. Time was treated as a categorical variable. The corresponding outcome measures during the negative baseline were modeled as covariates. Each participant was modeled with a random intercept to account for the individual differences in baseline heart rate variability metrics, responses to alcohol cues, and responses to the intervention. The outcome measures during the positive baseline were used as reference levels. The P values indicated that the corresponding term significantly affected the outcome measures.

Variables		RMSSD	HF	Self-reported craving	Self-reported anxiety
Random intercept
	β coefficient	4.84	5.21	2.72	7.28
	95% CI	−2.86 to 12.53	−13.18 to 23.60	1.45 to 3.99	3.47 to 11.08
	P value	.21	.57	<.001a	<.001a
Negative baseline (covariate)
	β coefficient	0.76	0.73	0.30	0.43
	95% CI	0.64 to 0.91	0.55 to 0.91	0.08 to 0.53	0.06 to 0.80
	P value	<.001a	<.001a	.009b	.02c
Interaction between intervention and time (Block 1)
	β coefficient	−10.95	−26.65	−2.12	0.46
	95% CI	−19.48 to −2.41	−49.80 to −3.49	−4.16 to −0.08	−1.75 to 2.67
	P value	.01c	.03c	.04c	.68
Interaction between intervention and time (Block 2)
	β coefficient	−8.48	−21.59	−2.42	−0.46
	95% CI	−17.02 to 0.06	−44.74 to 1.57	−4.46 to −0.38	−2.67 to 1.75
	P value	.052c	.07	.02c	.68

^aSignificance level P<.001.

^bSignificance level P<.01.

^cSignificance level P<.052.