Table 4.

Results of multivariable logistic regression models (total and stratified); Outcomes: alcohol, tobacco, cannabis consumption, and episodes of drunkenness decreased 5 months post-intervention.

	Stratified by estimation of peer use at baseline (T0) and follow-up (T1)
	Total analysis group		Under-/accurately at T0		Overestimated at T0		Under-/accurately at T0		Overestimated at T0
	(n = 1,255)		Under-/accurately at T1		Under-/accurately at T1		Overestimated at T0		Overestimated at T1
	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	p	OR (95% CI)	P
Alcohol consumption
Intervention (IG vs. CG)	1.91 (1.42-2.56)	<0.001	*	*	6.28 (2.00-19.8)	0.002	1.82 (0.48-6.90)	0.371	1.86 (1.32-2.63)	<0.001
Tobacco consumption
Intervention (IG vs. CG)	0.68 (0.38-1.22)	0.194	0.11 (0.00-2.10)	0.139	0.42 (0.03-6.00)	0.520	0.39 (0.06-2.47)	0.315	0.71 (0.31-1.60)	0.406
Cannabis consumption
Intervention (IG vs. CG)	1.33 (0.67-2.65)	0.414	0.83 (0.26-2.65)	0.753	11.7 (1.24-110)	0.032	*	*	1.74 (0.44-6.95)	0.430
Episodes of drunkenness
Intervention (IG vs. CG)	1.32 (0.98-1.80)	0.072	1.34 (0.00-253.0)	0.753	2.45 (0.63-9.48)	0.191	1.84 (0.41-8.26)	0.419	1.27 (0.89-1.82)	0.187

^*Sample size in this subgroup was too small to derive reliable estimates (algorithm did not converge).

CG, Control group; CI, Confidence Interval; IG, Intervention group; OR, Odds ratio; p-value (derived by means of a multivariable logistic regression model, additionally adjusted for alcohol, tobacco, or cannabis consumption or episodes of drunkenness at baseline; University treated as random effect). Bold values indicate significant effects.