Medicinal cannabis use among young adults during California’s transition from legalized medical use to adult-use: a longitudinal analysis

Figures & data

Table 1. Sample characteristics of key variables of interest from Wave 1 to Wave 6.

	Wave 1 2014–2015 n = 366	Wave 2 2015–2016 n = 339	Wave 3 2016–2017 n = 322	Wave 4 2017–2018 n = 302	Wave 5 2019–2020 n = 260	Wave 6 2020–2021 n = 251
			mean (SD) or % (n)
Demographic characteristics
Female sex	33.9% (124)	33.6% (114)	33.9% (109)	35.1% (106)	38.8% (101)	37.1% (93)
Hispanic ethnicity	45.1% (165)	44.8% (152)	44.7% (144)	43.7% (132)	43.8% (114)	46.2% (116)
Current age	21.23 (2.48)	22.21 (2.47)	23.18 (2.46)	24.21 (2.46)	26.16 (2.45)	27.19 (2.47)
Age of cannabis use initiation	15.24 (2.33)	15.25 (2.30)	15.28 (2.30)	15.27 (2.31)	15.27 (2.24)	15.23 (2.27)
Medicinal cannabis use
Medical cannabis patient	57.4% (210)	53.7% (182)	49.1% (158)	38.1% (115)	11.5% (30)	6.0% (15)
Self-reported medicinal use	24.3% (89)	27.5% (89)	22.3% (65)	25.3% (65)	28.8% (65)	23.8% (47)
Health symptoms
BSI-Global Severity Index	51.98 (10.41)	50.57 (10.89)	50.93 (11.01)	51.19 (12.43)	54.58 (12.29)	52.63 (13.29)
BSI-Anxiety	50.32 (9.85)	49.17 (10.11)	50.11 (10.11)	50.13 (11.17)	53.04 (11.57)	52.32 (11.74)
BSI-Depression	53.01 (10.14)	52.55 (10.06)	52.18 (10.10)	53.22 (10.80)	55.78 (11.53)	55.44 (11.33)
BSI-Somatization	51.54 (9.32)	50.31 (9.16)	51.21 (9.80)	51.56 (10.76)	54.06 (11.10)	52.43 (11.23)
Current non-minor pain	35.3% (128)	26.7% (88)	22.2% (71)	26.1% (76)	27.6% (70)	24.4% (60)
Cannabis use motives
Sleep motives	3.38 (1.28)	3.25 (1.22)	3.11 (1.25)	2.98 (1.20)	2.98 (1.21)	2.83 (1.22)
Pain motives	3.03 (1.34)	2.75 (1.25)	2.67 (1.26)	2.48 (1.13)	2.67 (1.19)	2.43 (1.23)
Coping motives	2.26 (1.14)	1.98 (0.92)	1.96 (0.91)	2.05 (0.92)	2.21 (0.99)	2.13 (1.06)
Cannabis use
Daily/near daily use in the past 90 days	71.6% (262)	70.5% (239)	65.8% (212)	55.9% (162)	58.9% (149)	58.0% (141)
Use of concentrates in the past 90 days	58.7% (215)	54.2% (176)	53.9% (158)	40.1% (111)	36.5% (88)	28.2% (64)
Problematic cannabis use (SDS)	2.48 (2.71)	2.81 (2.70)	2.81 (2.87)	2.92 (2.97)	3.13 (3.35)	2.96 (3.16)

1) SD = standard deviation; BSI = Brief Symptom Inventory; SDS = Severity of Dependence Scale.

2) Four-digit year notations represent dates of data collection within each wave.

3) Cannabis use motives subscales rate the frequency of motives on a scale from 1 (“almost never/never”) to 5 (“almost always/always”).

4) SDS (Severity of Dependence Scale) denotes the sum score ranging 0–15, with a higher score indicating higher level of dependence.

Table 2. Model fit indices for longitudinal latent class analysis.

						VLMR	LMR	BLRT	Average Class Posterior Probabilities
Class	AIC	BIC	sBIC	Class Distribution (%)	Entropy	p-value
1	4023	4070	4031	−	−	−	−	−	−
2	3754	3851	3772	27/73	.744	.001	.001	<.0001	0.87, 0.95
3	3641	3789	3669	20/40/39	.687	.004	.004	<.0001	0.84, 0.87, 0.90
4	3611	3810	3648	31/35/15/18	.687	.12	.12	<.0001	0.77–0.89
5	3601	3851	3648	17/6/13/33/30	.721	.51	.51	.05	0.77–0.93

1) AIC = Akaike Information Criterion; BIC = Bayesian Information Criterion; sBIC = Sample size-adjusted Bayesian Information Criterion; VLMR = Vuong-Lo-Mendell-Rubin test; LMR = Lo-Mendell-Rubin test; BLRT = bootstrapped likelihood ratio test.

2) Bolded is the selected model.

Figure 1. Longitudinal patterns of medicinal cannabis use from the selected 3-class model: Recreational Users (39.3% of the sample; low self-reported medicinal use and low-to-decreasing patient status), Recreational Patients (40.4%; low self-reported medicinal use and high-to-decreasing patient status), and Medicinal Patients (20.3%; high self-reported medicinal use and high-to-decreasing patient status). Y-axis displays the estimated probabilities of patient status and self-reported medicinal cannabis use in each class. Descriptively, findings show that patient status decreased but self-reported medicinal cannabis use remained steady in each class. W1-W6 = Waves of data collection. Four-digit year notations represent dates of data collection within each wave.

Table 3. The association of longitudinal latent classes with covariates.

Covariate	Medicinal Patients vs. Recreational Users		Recreational Patients vs. Recreational Users		Medicinal Patients vs. Recreational Patients
	UOR [95% CI]	AOR [95%CI]	UOR [95%CI]	AOR [95%CI]	UOR [95%CI]	AOR [95%CI]
Baseline age	1.14* [1.00, 1.30]	1.16* [1.01, 1.32]	1.07 [0.93, 1.23]	1.07 [0.93, 1.23]	1.07 [0.94, 1.21]	1.08 [0.95, 1.23]
Female sex	0.57 [0.28, 1.14]	0.56 [0.27, 1.16]	0.44* [0.23, 0.86]	0.42* [0.21,0.82]	1.29 [0.60, 2.78]	1.34 [0.62, 2.89]
Hispanic ethnicity	1.05 [0.54, 2.05]	1.10 [0.55, 2.23]	0.79 [0.43, 1.48]	0.75 [0.39, 1.44]	1.33 [0.66, 2.65]	1.47 [0.73, 2.99]
Age of cannabis use initiation	1.00 [0.86, 1.17]	0.98 [0.83, 1.16]	0.98 [0.87, 1.11]	0.98 [0.85, 1.13]	1.02 [0.87, 1.19]	1.00 [0.85, 1.17]

*p < .05.

1) UOR = Unadjusted Odds Ratio; AOR = Adjusted Odds Ratio; CI = Confidence Interval.

2) In adjusted models, all covariates were entered simultaneously.

Table 4. Estimated trajectories of health symptoms, cannabis use motives, and cannabis use across longitudinal latent classes.

	Estimated growth factor means			Significant class pairwise differences in growth factor means	Overall comparison (Wald χ^2)
Longitudinal characteristics	Class 1: Recreational Users	Class 2: Recreational Patients	Class 3: Medicinal Patients
Health symptoms
BSI-Global Severity Index					1.72
Intercept	50.37***	51.13***	52.48***	–
Slope	0.38	0.16	0.25	–
BSI-Anxiety					3.11
Intercept	48.98***	49.01***	51.76***	–
Slope	0.67**	0.43	0.21	–
BSI-Depression					0.90
Intercept	52.05***	52.52***	51.83***	–
Slope	0.50*	0.63*	0.41	–
BSI-Somatization					5.72
Intercept	50.14***	50.15***	53.80***	3>1*, 3>2*
Slope	0.38	0.31	0.23	–
Current non-minor pain					133.26***
Intercept	−1.37***	−1.29***	0.69**	3>1***, 3>2***
Slope	−0.03	−0.30**	−0.05	–
Cannabis use motives
Sleep motives					34.13***
Intercept	3.02***	3.39***	3.95***	3>1***, 3>2**
Slope	−0.09**	−0.11***	−0.13**	–
Pain motives					46.50***
Intercept	2.63***	2.82***	3.71***	3>1***, 3>2***
Slope	−0.11***	−0.10**	−0.11**	–
Coping motives					2.95
Intercept	2.03***	2.01***	2.20***	–
Slope	0.01	0.04	−0.04	–
Cannabis use
Daily/near daily use in the past 90 days					234.14***
Intercept	−0.17	1.78***	2.30***	3>1***, 2>1***
Slope	−0.49***	0.06	−0.35**	2>1**, 2>3*
Use of concentrates in the past 90 days					117.74***
Intercept	−0.74**	1.74***	0.97*	3>1**, 2>1***
Slope	−0.29**	−0.39***	−0.30**	–
Problematic cannabis use (SDS)					6.24
Intercept	2.44***	2.65***	2.61***	–
Slope	−0.01	0.24**	0.04	–

*p < .05, **p < .01, ***p < .001.

BSI = Brief Symptom Inventory; SDS = Severity of Dependence Scale; Wald χ² = Wald Chi-Square test.

Figure 2. Trajectories of health symptoms, cannabis use motives, and cannabis use across three longitudinal latent classes: Recreational Users (39.3%), Recreational Patients (40.4%), and Medicinal Patients (20.3%). The left Y-axis displays estimated mean scores (solid lines) and the right Y-axis displays estimated probabilities (dashed lines). Trajectories with statistically significant upward or downward slopes are represented by lines ending in arrows (see Table 4 for slope p-values). Findings show that between waves 1–6, depression or/and anxiety symptoms increased in recreational groups and pain decreased in Recreational Patients (panels a, c, e). Pain motives, sleep motives, and concentrate use decreased in all groups, daily/near daily use decreased in Recreational Users and Medicinal Patients, and problematic cannabis use increased in Recreational Patients (panels b, d, f). BSI-GSI = Brief Symptom Inventory - Global Severity Index; BSI-ANX = BSI-Anxiety subscale; BSI-DEP = BSI-Depression subscale; BSI-SOMA = BSI-Somatization subscale; SDS = Severity of Dependence Scale; pain = current non-minor pain; W1-W6 = Waves of data collection.