Patterns of Cannabis Use among Canadian Youth over Time; Examining Changes in Mode and Frequency Using Latent Transition Analysis

Figures & data

Table 1. Descriptive statistics of the COMPASS study sample, 2017/2018 to 2018/2019.

N = 17,161
		2017/2018 % (N)^a
		Females	Males	Total
Grade	9	36.8 (3372)	37.3 (2987)	37.1 (6359)
	10	35.5 (3247)	34.8 (2790)	35.2 (6037)
	11	27.7 (2534)	27.9 (2231)	27.8 (4765)
Ethnicity	Nonracialized	71.7 (6560)	70.9 (5675)	71.3 (12,235)
	Racialized	28.3 (2593)	29.1 (2333)	28.7 (4926)
Spending Money	Zero	19.0 (1741)	22.3 (1787)	20.6 (3528)
	$1–$40	41.3 (3781)	40.0 (3200)	40.7 (6981)
	$41–$100	12.1 (1109)	10.0 (804)	11.1 (1913)
	>$100	12.2 (1121)	14.4 (1151)	13.2 (2272)
	Don’t Know	15.3 (1401)	13.3 (1066)	14.4 (2467)
Current smoking (last 30 Days)	No	95.6 (8752)	94.6 (7573)	95.1 (16,325)
	Yes	4.4 (401)	5.4 (435)	4.9 (836)
Current e-cigarette Use (last 30 Days)	No	86.8 (7944)	82.1 (6574)	84.6 (14,518)
	Yes	13.2 (1209)	17.9 (1434)	15.4 (2643)
Current binge drinking (≥ once a month)	No	89.6 (8201)	89.5 (7171)	89.6 (15,372)
	Yes	10.4 (952)	10.5 (837)	10.4 (1789)
					2018/2019 % (N)
		Females	Males	Total	Females	Males	Total
Marijuana/Cannabis Use (last 12 Months)b	No use	84.6 (7745)	84.0 (6727)	84.3 (14,472)	71.2 (6514)	69.8 (5588)	70.5 (12,102)
	Less than once a month	7.4 (681)	6.1 (492)	6.8 (1173)	14.3 (1310)	11.4 (909)	12.9 (2219)
	1–3 times per month	4.4 (399)	4.6 (365)	4.5 (764)	7.6 (692)	7.3 (583)	7.4 (1275)
	Weekly or more often	3.2 (295)	4.8 (387)	4.0 (682)	6.4 (585)	10.2 (814)	8.2 (1399)
Smoked cannabis (last 12 Months)	No	85.7 (7848)	85.5 (6844)	85.6 (14,692)	74.2 (6793)	73.4 (5877)	73.8 (12,670)
	Yes	14.3 (1305)	14.5 (1164)	14.4 (2469)	25.8 (2360)	26.6 (2131)	26.2 (4491)
Vaped cannabis (last 12 Months)	No	97.4 (8917)	96.7 (7745)	97.1 (16,662)	92.2 (8437)	89.7 (7185)	91.0 (15,622)
	Yes	2.6 (236)	3.3 (263)	2.9 (499)	7.8 (716)	10.3(823)	9.0 (1539)
Ate/Drank cannabis (last 12 Months)	No	95.1 (8706)	95.2 (7623)	95.2 (16,329)	87.4 (8004)	87.2 (6981)	87.3 (14,985)
	Yes	4.9 (447)	4.8 (385)	4.8 (832)	12.6 (1149)	12.8 (1027)	12.7 (2176)

a 336 total students were removed from original sample due to missing covariate data.

b Cells do not sum to total N due to missing data (70 students were missing cannabis use frequency data in 2017/2018, and 166 were missing for 2018/2019).

Table 2. Model fit statistics for LCA models (COMPASS Study 2017/2018–2018/2019).

	Number of classes	df	Log-likelihood	AIC	BIC	Entropy
Females (2017/2018)	1	25	–11,924	23,861	23,904	1.00
	2	18	–7110	14,245	14,338	1.00
	3	11	–6997	14,033	14,175	0.97
	4	4	–6997	14,047	14,154	0.95
	5	Negative
Males (2017/2018)	1	24	–10,829	21,670	21,712	1.00
	2	18	–6557	13,141	13,231	1.00
	3	11	–6397	12,834	12,974	0.97
	4	4	–6397	12,848	13,037	0.59
	5	Negative
Females (2018/2019)	1	25	–19,300	38,613	38,656	1.00
	2	18	–12,225	24,476	24,568	1.00
	3	11	–12,052	24,145	24,287	0.92
	4	4	–11,947	23,948	24,140	0.91
	5	negative
Males (2018/2019)	1	25	–17,622	35,257	35,299	1.00
	2	18	–11,071	22,168	22,259	1.00
	3	11	–10,790	21,620	21,760	0.97
	4	4	–10,683	21,421	21,609	0.96
	5	negative

Lowest AIC and BIC values are highlighted in bold.

Table 3. LTA model results among females in the COMPASS 2018/2019 sample.

		Regular multimode	Occasional smoking	Occasional multimode	Nonuse
Class prevalence:	2017–2018	3.4%	8.7%	3.0%	84.9%
	2018–2019	7.4%	10.9%	10.0%	71.7%
Item response probabilities:
Use frequency	No use	0.00	0.00	0.00	1.00
	Less than once a month	0.01	0.65	0.70	0.00
	1–3 times per month	0.34	0.25	0.27	0.00
	Weekly or more often	0.66	0.11	0.03	0.00
Smoked	Yes	0.99	1.00	0.75	0.00
Vaped	Yes	0.44	0.06	0.35	0.00
Ate/Drank	Yes	0.67	0.07	0.67	0.00
Transition probabilities:
		2018–2019
2017–2018	Regular multimode	0.94	0.03	0.00	0.02
	Occasional smoking	0.24	0.43	0.21	0.12
	Occasional multimode	0.21	0.07	0.63	0.09
	Nonuse	0.02	0.08	0.08	0.83

Table 4. LTA model results among males in the COMPASS 2018/2019 sample.

		Regular multimode	Smoking	Occasional multimode	Nonuse
Class prevalence:	2017–2018	3.4%	9.6%	2.6%	84.4%
	2018–2019	9.4%	11.8%	7.8%	71.0%
Item response probabilities:
Use frequency	No use	0.00	0.00	0.00	1.00
	Less than once a month	0.02	0.49	0.67	0.00
	1–3 times per month	0.25	0.27	0.27	0.00
	Weekly or more often	0.73	0.24	0.06	0.00
Smoked	Yes	0.99	1.00	0.68	0.00
Vaped	Yes	0.68	0.03	0.40	0.00
Ate/Drank	Yes	0.79	0.08	0.57	0.00
Transition probabilities:
		2018–2019
2017–2018	Regular multimode	0.94	0.02	0.01	0.04
	Smoking	0.35	0.45	0.11	0.09
	Occasional multimode	0.31	0.13	0.49	0.08
	Nonuse	0.03	0.08	0.06	0.83

Figure 1. Visual depiction of transition probabilities for different modes of cannabis use among females in the COMPASS Study (2017/2018–2018/2019). In this Sankey diagram, the size of the colored bars corresponds to the magnitude of the transition probabilities in Table 3, by which thicker bars represent greater probability.

Figure 2. Visual depiction of transition probabilities for different modes of cannabis use among males in the COMPASS Study (2017/2018–2018/2019). In this Sankey diagram, the size of the colored bars corresponds to the magnitude of the transition probabilities in Table 4, by which thicker bars represent greater probability.

Table 5. Multinominal logistic regression model predictors of baseline class prevalence among females in the 2017/2018 COMPASS sample.

		Regular multimode vs. nonuse	Occasional smoking vs. nonuse	Occasional multimode vs. nonuse
		OR (95% CI)
Grade	9 (ref)
	10	2.49 (1.60–3.89)	2.56 (1.99–3.29)	2.00 (1.24–3.22)
	11	2.82 (1.76–4.51)	3.57 (2.75–4.52)	3.22 (2.00–5.16)
Weekly spending money	Zero (ref)
	$1–$40	1.73 (1.02–2.92)	1.20 (0.92–1.58)	0.88 (0.51–1.51)
	$41–$100	1.75 (0.94–3.25)	1.48 (1.06–2.05)	1.41 (0.76–2.60)
	More than $100	1.50 (0.79–2.83)	1.54 (1.11–2.13)	1.75 (0.98–3.13)
	Don’t Know	0.89 (0.43–1.85)	0.84 (0.58–1.20)	1.28 (0.70–2.32)
Current smoking	No (ref)
	Yes	28.11 (18.08–43.68)	9.64 (6.60–14.07)	4.17 (2.09–8.35)
Current E-cigarette use	No (ref)
	Yes	7.09 (4.90–10.27)	4.35 (3.41–5.55)	6.13 (3.91–9.63)
Current binge drinking	No (ref)
	Yes	13.16 (9.05–19.13)	4.72 (3.62–6.15)	5.66 (3.55–9.05)

Table 6. Multinominal logistic regression model predictors of baseline class prevalence among males in the 2017/2018 COMPASS sample.

		Regular multimode vs. nonuse	Smoking vs. nonuse	Occasional multimode vs. nonuse
		OR (95% CI)
Grade	9 (ref)
	10	2.30 (1.39–3.80)	1.36 (1.08–1.71)	2.14 (1.22–3.75)
	11	2.88 (1.72–4.83)	1.89 (1.48–2.41)	4.43 (2.57–7.64)
Weekly spending money	Zero (ref)
	$1–$40	2.27 (1.08–4.78)	1.27 (0.98–1.64)	2.47 (1.20–5.10)
	$41–$100	1.96 (0.84–4.59)	1.22 (0.84–1.77)	3.16 (1.40–7.13)
	More than $100	3.94 (1.89–8.21)	1.37 (0.99–1.88)	1.89 (0.81–4.41)
	Don’t Know	2.00 (0.86–4.67)	1.07 (0.76–1.50)	2.70 (1.21–6.03)
Current smoking	No (ref)
	Yes	13.67 (8.91–20.95)	6.48 (4.57–9.18)	1.64 (0.69–3.88)
Current E-cigarette use	No (ref)
	Yes	16.28 (10.39–25.50)	6.11 (4.91–7.61)	10.24 (6.38–16.42)
Current binge drinking	No (ref)
	Yes	5.74 (3.93–8.40)	4.61 (3.60–5.92)	2.74 (1.61–4.65)

Table A1. Attrition analysis comparing baseline cannabis use between participants who were linked vs. not linked.

		2017/2018
		Linked		Not linked		Chi-Sqa
		n	%	n	%	p-value	Cramer’s V
Total		18,824	100.0%	15,125	100.0%	<.0001	0.1904
Cannabis use frequency (past 12 months)	No use	15,328	81.4%	10,204	67.5%
	Less than once a month	1350	7.2%	1413	9.3%
	1–3 times per month	851	4.5%	1073	7.1%
	Weekly or more often	780	4.1%	1986	13.1%
	Missing	515	2.7%	449	3.0%
Smoked it (Last 12 Months)	No	15,644	83.1%	10,533	69.6%	<.0001	0.1593
	Yes	3180	16.9%	4592	30.4%
Vaped it (Last 12 Months)	No	18,206	96.7%	13,947	92.2%	<.0001	0.1
	Yes	618	3.3%	1178	7.8%
Ate/drank it (Last 12 Months)	No	17,841	94.8%	13,436	88.8%	<.0001	0.1097
	Yes	983	5.2%	1689	11.2%

a Chi-square tests exclude missing values.

Table A2. Missing data analysis comparing included vs. excluded participants due to missing covariates (COMPASS 2017/2018–2018/2019).

		2017/2018						2018/2019
		Included		Excluded		Chi-Sqa		Total		Male		Chi-Sqa
		n	%	n	%	p-value	Cramer’s V	n	%	n	%	p-value	Cramer’s V
Total		17,161	100.0%	344	100.0%	<.0001	0.0426	17,161	100.0%	344	100.0%	<.0001	0.0535
Cannabis use frequency (past 12 months)	No use	14,472	84.3%	183	53.2%			12,102	70.5%	211	61.3%
	Less than once a month	1173	6.8%	14	4.1%			2219	12.9%	30	8.7%
	1–3 times per month	764	4.5%	15	4.4%			1275	7.4%	24	7.0%
	Weekly or more often	682	4.0%	26	7.6%			1399	8.2%	62	18.0%
	Missing	70	0.4%	106	30.8%			166	1.0%	17	4.9%
Smoked it (Last 12 Months)	No	14,692	85.6%	289	84.0%	.4026	0.0063	12,670	73.8%	237	68.9%	.0395	0.0156
	Yes	2469	14.4%	55	16.0%			4491	26.2%	107	31.1%
Vaped it (Last 12 Months)	No	16,662	97.1%	329	95.6%	.1141	0.0119	15,622	91.0%	303	88.1%	.0586	0.0143
	Yes	499	2.9%	15	4.4%			1539	9.0%	41	11.9%
Ate/drank it (Last 12 Months)	No	16,329	95.2%	326	94.8%	.7426	0.0025	14,985	87.3%	293	85.2%	.0237	0.0089
	Yes	832	4.8%	18	5.2%			2176	12.7%	51	14.8%
Grade	9	6359	37.1%	160	46.5%	.0015	0.0272
	10	6037	35.2%	105	30.5%
	11	4765	27.8%	79	23.0%
Ethnicity	Nonracialized	10,842	63.2%	186	54.1%	.0017	0.0237
	Racialized	6319	36.8%	153	44.5%
	Missing	0	0.0%	5	1.5%
Weekly spending money	Zero	3528	20.6%	47	13.7%	.6635	0.0117
	$1–$40	6981	40.7%	89	25.9%
	$41–$100	1913	11.1%	21	6.1%
	More than $100	2272	13.2%	33	9.6%
	Don’t Know	2467	14.4%	25	7.3%
	Missing	0	0.0%	129	37.5%
Current smoker	No	16,325	95.1%	224	65.1%	<.0001	0.0303
	Yes	836	4.9%	26	7.6%
	Missing	0	0.0%	94	27.3%
Current E-cigarette user	No	14,518	84.6%	139	40.4%	.1716	0.0104
	Yes	2643	15.4%	33	9.6%
	Missing	0	0.0%	172	50.0%
Current binge drinker	No	15,372	89.6%	280	81.4%	.1517	–0.0108
	Yes	1789	10.4%	24	7.0%
	Missing	0	0.0%	40	11.6%

a Chi-square tests exclude missing values.

Table A3. Invariance testing for differences by sex.

2017/2018
Model	Parameters	Log-lik	BIC	Delta G^2 (LR chi-square)	df	p-value	Entropy
1. Unconstrained	43	–25,156	50,731	191	20	<.0001	0.99
2. IRPs constrained	25	–25,192	50,628	263	38	<.0001	0.98
3. CPs constrained	22	–25,199	50,612	277	41	<.0001	0.99
Likelihood ration test for equal IRPs:
	Delta G^2	Delta df	p-value
Model 1. vs. 2.	73	18	.0000
Likelihood ratio test for equal CPs:
	Delta G^2	Delta df	p-value
Model 2. vs. 3.	14	3	.0034
2018/2019
Model	Parameters	Log-lik	BIC	G^2 (LR chi-square)	df	p-value	Entropy
1. Unconstrained	43	–34,487	69,394	363	20	<.0001	0.96
2. IRPs constrained	25	–34,531	69,306	450	38	<.0001	0.94
3. CPs constrained	22	–34,561	69,337	485	41	<.0001	0.97
Likelihood ratio test for equal IRPs:
	Delta G^2	Delta df	p-value
Model 1. vs. 2.	88	18	.0000
Likelihood ratio test for equal CPs:
	Delta G^2	Delta df	p-value
Model 2. vs. 3.	35	3	.0000