A positive association between anxiety disorders and cannabis use or cannabis use disorders in the general population- a meta-analysis of 31 studies

Table 5 Results of the moderator analyses (subgroup-analyses and univariate meta-regression)

	Subgroup of studies	Anxiety vs. cannabis use/no use			Anxiety vs. CUD/no CUD (or no use)
		Nstudies	OR(95% CI)	p _two-tailed	Nstudies	OR(95% CI)	p _two-tailed
Overall mean weighted OR		15	1.24 (1.06-1.45)	.006*	13	1.68 (1.23-2.31)	.001*
Subgroup analyses (Figure S3)
ORs adjusted for confounders	Yes	12	1.24 (1.04-1.47)	.014*	10	1.66 (1.17-2.37)	.005*
	No	3	1.28 (.78-2.08)	.326	3	1.86 (1.10-3.15)	.021*
	Yes vs. No		Q(1) = .04; p = .848			Q(1) = 8.96; p = .003*
Clinical diagnosis of anxiety (based on DSM/ICD)	Yes	9	1.29 (1.04-1.61)	.021*	11	1.87 (1.43-2.44)	<.001*
	No	6	1.17 (.93-1.48)	.186	2	1.14 (.78-1.66)	.509
	Yes vs. No		Q(1) = .30; p = .586			Q(1) = 36.91; p < .001*
Meta-regression		N studies	slope	*slope p* _two-tailed	N studies	slope	*slope p* _two-tailed
Predictor: Year of publication Outcome: weighted OR
		15	.02	.299	13	-.005	.901

Note: Confounders were: other substance use and/or other substance use disorders and/or past AD and/or other psychiatric illnesses and/or demographics. The subgroup analyses were conducted using the so-called mixed-effects model of meta-analysis [57]. According to this model the studies within each subgroup were combined using the random-effects model. However, since the number of subgroups was fixed (rather than randomly selected out of many subgroups), the overall mean weighted ORs in both groups were compared statistically using the between-groups Q-statistic based on the fixed-effect model of meta-analysis with df = 1 (number of subgroups-1). This approach to comparing ORs in independent subgroups of studies is equivalent to the independent samples t-test. Figure S3 is located in the Additional file 1.
Abbreviations: AD anxiety disorder, CUD cannabis use disorder (cannabis dependence and/or abuse/harmful use), df degrees of freedom, OR odds ratio.
*p < .05.

ISSN: 1471-244X