Medical Cannabis Use Across Ages 19–65: U.S. Young and Middle Adults, 2018–2023

Abstract

Introduction.

This study provides national data on cannabis use type (medical vs. recreational-only) across ages 19–65 and associations with overall cannabis use prevalence and frequency, including developmental and historical trends and sociodemographic and policy associations.

Methods.

Data collected in 2018–2023 from individuals (N=33,647) ages 19–65 participating in the Monitoring the Future Panel study were analyzed in 2024–2025. Developmental and historical trends and regression analyses examined past 12-month cannabis use type: no use, recreational-only, or any medical cannabis used from their own written medical recommendation/prescription (with or without recreational use).

Results.

Medical use was reported by 2.6% [2.3%, 2.8%] of all respondents and 9.8% [9.0%, 10.6%] of those reporting past 12-month use. Among all respondents, medical use prevalence did not evidence significant developmental trends; among those reporting past 12-month use, there was an age-graded increase in medical use (p<.001) corresponding to an age-graded decrease in recreational-only use (p<.001). Medical use prevalence increased across time in states with medical use only policy (p=0.002) but not in other states, and was associated with being male (p<.001). Past 30-day cannabis prevalence and frequency were higher among medical than recreational-only use groups across ages (p<.001).

Conclusions.

Among U.S. young and middle adults, the proportion reporting medical use was consistent across age; observed age-graded increases in medical use among those reporting past 12-month cannabis use were due to decreasing recreational-only use. Medical use was associated with higher past 30-day frequency across age, indicating it acts as a consistent risk factor for daily or near-daily use.

Introduction

Cannabis use among young and middle adults (ages 19–30 and 35–65) has been rising.^1,2,3,4,5 Significant increases from 2018 to 2023 were observed for past 12-month use (38.7% to 42.4% for young adults; 19.0% to 29.3% for middle adults) and daily or near-daily use (8.6% to 10.4% for young adults; 4.3% to 7.5% for middle adults).³ High-frequency use is associated with cannabis use disorder among those using cannabis overall^6,7 and those reporting only medical use (use with a personal medical recommendation from a medical professional).⁸ Results from non-representative studies^9,10 indicate medical use (which can occur with or without recreational use^11,12) is associated with higher use frequency than recreational-only use. To the degree medical use is a particularly strong risk factor for high-frequency use, awareness of potential age and other differences in medical use likelihood may be important from a public health perspective, given rapidly increasing U.S. medical cannabis program enrollment.^13,14

Research on age-related differences in medical cannabis prevalence has provided mixed results. National studies have indicated lifetime medical use is most likely during the 30s and 40s.^8,15 Some national U.S. studies have found past 12-month medical use is most likely at older ages (65+);¹⁶ others find no age differences.^17,18 Data collected in 2015 from registered medical cannabis participants in eight states showed medical use age differences varied by state.¹⁹ Limited research has examined age-related differences in medical use frequency; one small study found similar medical use frequency across ages 18–30, 31–50, and 51–74.²⁰ Because cannabis use mental and physical health effects change with age,²¹ it is important to understand age-related differences in medical use prevalence and frequency.

Research examining medical cannabis prevalence and frequency should account for known sociodemographic and policy correlates. Medical use may be more likely for males than females,^8,15,19 those not married,⁸ and those not employed;^8,17 racial/ethnic associations have been mixed.^8,17 U.S. state policies legalizing medical and/or adult recreational use differ and are multi-dimensional,^22,23,24,25 but lifetime medical cannabis prevalence in 2018 was higher in states where both recreational and medical use were legalized (vs. medical only).¹⁵

The current paper used national data collected in 2018–2023 from U.S. adults ages 19–65 to examine three research questions (RQs): (1) What percentages of adults reported no cannabis use, recreational-only use, and medical use (with or without recreational use) in the past 12 months, and were there developmental or historical trends in these types of use? (2) To what extent did likelihood of cannabis use types vary by sociodemographic characteristics and state policy context? (3) Did people reporting medical (vs. recreational-only) use differ in past 30-day cannabis prevalence or frequency, and did associations vary by age?

Methods

Study Sample

Data were obtained from the Monitoring the Future (MTF) Panel study.³ MTF surveys new nationally-representative samples of U.S. 12^th grade students each year.²⁶ From each annual 12^th grade sample, approximately 2,450 individuals are selected for Panel study follow-up which is randomized to begin one or two years after high school (modal ages [hereafter referred to simply as age] 19 or 20).³ Six biennial young adult surveys are collected (ages 19/20, 21/22, 23/24, 25/26, 27/28, 29/30); middle adult surveys are collected every five years (ages 35, 40, 45, 50, 55, 60, and 65). Through 2021, elements of consent were included in the introductory letter for all Panel surveys; from 2022 onward, informed consent has been obtained at the beginning of each survey.³ A University of Michigan institutional review board approved the study.

Medical cannabis use measures were available for both young and middle adult surveys starting in 2018. Medical cannabis use was asked on 1/6th of young adult surveys¹ in 2018, and half of young adult surveys from 2021. All middle adult surveys included medical use. The current study utilized data collected in 2018–2023 (analyzed in 2024/2025) from respondents ages 19–65 with the opportunity to answer medical cannabis measures (12^th grade cohorts from 1976–2022; see Appendix Table 1). Of the 92,261 respondents purposively selected for panel participation from the noted cohorts, 36,016 (39.0%) responded to one or more panel surveys (see Statistical Analysis below for attrition adjustments), providing a total of 44,148 observations (see Appendix). Of these, 1,119 (2.5% of 43,906) were excluded because they did not reside in a U.S. state or had missing data on state of residence; 3,103 (4.8% of 43,906) had missing or conflicting data on measures used to code cannabis use type, leaving 40,926 observations from 33,647 respondents (36.5% of the original 92,216). By age group, 27,671 middle adults provided 31,955 observations (mean=1.15; range=1–2): 8,971 young adults provided 6,026 observations (mean=1.49; range=1–3). Of all observations, 51.6% were female, 47.2% male, and 1.2% other or prefer not to answer.² Racial/ethnic distribution was 12.8% Hispanic, 3.1% non-Hispanic Asian, 11.4% non-Hispanic Black, 68.0% non-Hispanic White, and 4.6% another non-Hispanic race/ethnicity or multiracial.

Measures

Respondents were asked the number of occasions they used cannabis³ in the last 12 months and the last 30 days; response options included 0, 1–2, 3–5, 6–9, 10–19, 20–39, and 40 or more. If respondents reported any 12-month use, they were asked if they had obtained any cannabis used in the past 12 months from their own medical marijuana written recommendation/prescription (young adults); or how much of the cannabis they used came from their own written recommendation/prescription (middle adults) (see Appendix Table 2 for specific wording). Medical use was defined as using any cannabis from one’s own medical recommendation/prescription. Cannabis use type was coded as no past 12-month use, recreational-only use, or medical use with or without recreational use (hereafter referred to as medical use). Past 30-day prevalence indicated any past 30-day day use versus none. Past 30-day frequency was coded in integer values of 0, 1, 4, 7, 14, 29, and 40 occasions in the past 30 days.

At age 18, respondents self-reported sex (female, male, or [starting in 2021] other or prefer not to answer) and racial/ethnic identity (coded as Hispanic; non-Hispanic Asian; non-Hispanic Black; non-Hispanic White; or another non-Hispanic race or ethnicity or multiracial [multiracial possible from 2005 onward]). At each panel survey, respondents reported marital status (married vs. other); parental status (have children vs. not); income assistance (if any total household income came from unemployment compensation or other government assistance); and employment status (one full-time job vs. other). Respondents reported the state where they lived during the last 12 months; states were then coded for U.S. Census divisions. For multivariable models, age was coded as 19–30, 35–45, or 50–65; year of data collection was coded using dummy indicators (2018, 2019, 2020 [referent], 2021, 2022 2023).

State cannabis policy categories were coded using state medical cannabis²⁷ and recreational laws²⁸ effective on January 1 of each year from 2018–2023: no legal use, medical use only, or recreational+medical use (all states with adult recreational use also allowed medical use).

Statistical Analysis

Descriptive analyses used survey procedures in SAS v9.4 (SAS Institute, Inc., Cary, NC) with domain statements to obtain adjusted estimates with robust standard errors. Developmental and historical trend analyses (RQ1) used Joinpoint v5.3.0.0 (Statistical Research and Applications Branch, National Cancer Institute, Rockville, MD) specifying the weighted BIC (Bayesian Information Criterion) model selection method.^29,30 For these analyses, data were aggregated to either age level (developmental trends) or year level (historical trends). Regression analyses for RQ2 and 3 used Mplus v7.4 (Muthén & Muthén, Los Angeles, CA) specifying maximum likelihood parameter estimates, type=complex with subpop statements. For RQ2 (covariate associations with cannabis use type), two multinomial regression models regressed the 3-level cannabis use type measure on sociodemographics, state policy, division, age (categorical), and year. Model 1 used “no past 12-month use” as the outcome referent to compare recreational-only versus no use, and medical versus no use. Model 2 used “recreational-only use” as the referent to compare medical versus recreational-only use (no use vs. recreational-only use also was compared, but this was already done in Model 1). For RQ3 (cannabis use type and past 30-day use prevalence and frequency associations), logistic and negative binomial regression models were fit including RQ2 covariates. Additional models examined age differences with age group × medical use interactions.

All analyses were weighted using MTF panel analysis weights to adjust for sampling and nonresponse.³¹ In regression models, missing covariate data were addressed by using full information maximum likelihood (FIML). Benjamini-Hochberg tests with a false discovery rate of 5% adjusted for multiple testing.³² Comparison of raw p-values with their Benjamini-Hochberg critical values (BHCV) indicated p=0.028 was the largest value that remained smaller than its BHCV. Discussion of results will be limited to those with p-values of 0.028 or lower.

Results

Out of the total 40,926 observations, 73.7% of observations reported no past 12-month cannabis use, 23.7% recreational-only use, and 2.6% medical use. Among those reporting 12-month use, 9.8% reported medical use. Table 1 provides descriptive statistics.

Table 1.

Descriptive Statistics

Variable	% (95% CI)
Past 12-month cannabis use type (n=40,926)
No 12-month use	73.7 (73.1, 74.4)
Recreational-only use	23.7 (23.1, 24.3)
Medical usea	2.6 (2.3, 2.8)
Past 30-day cannabis use (n=40,926)
Prevalence	17.4 (16.8, 18.0)
Frequency [range = 0 to 40]	3.0 (2.8, 3.1)
Sex (n=40,883)
Female	51.6 (50.8, 52.3)
Male	47.2 (46.5, 48.0)
Other	1.2 (1.0, 1.4)
Race and ethnicity (n=40,506)
Hispanic	12.8 (12.2, 13.4)
NHb Asian	3.1 (2.8, 3.3)
NH Black	11.4 (10.9, 12.0)
NH White	68.0 (67.2, 68.8)
Another NH race/ethnicity, multiracial	4.6 (4.3, 5.0)
Married (n=40,482)
No	47.8 (47.1, 48.6)
Yes	52.2 (51.4, 52.9)
Parental status (n=40,735)
Do not have children	42.3 (41.5, 43.0)
Have children	57.7 (57.0, 58.5)
Employment status (all respondents) (n=40,367)
Not working one full time job	35.5 (34.9, 36.2)
Working one full time job	64.5 (63.8, 65.1)
Income assistancec (n=37,335)
No	87.3 (86.8, 87.8)
Yes	12.7 (12.2, 13.2)
State cannabis policy categorization (n=40,926)
No legal use	32.1 (31.4, 32.8)
Legal medical use only	38.0 (37.4, 38.7)
Legal recreational+medical use	29.9 (29.2, 30.5)
Age (n=40,926)
19/20	4.0 (3.7, 4.3)
21/22	5.6 (5.2, 5.9)
23/24	5.8 (5.5, 6.1)
25/26	5.9 (5.6, 6.2)
27/28	6.3 (5.9, 6.6)
29/30	6.6 (6.3, 7.0)
35	9.5 (9.2, 9.9)
40	9.9 (9.5, 10.2)
45	10.8 (10.4, 11.1)
50	11.0 (10.7, 11.4)
55	11.5 (11.2, 11.9)
60	11.4 (11.0, 11.7)
65	1.7 (1.5, 1.8)
Division (n=40,926)
New England	5.2 (4.9, 5.5)
Middle Atlantic	12.3 (11.8, 12.8)
East North Central	16.6 (16.1, 17.1)
West North Central	7.0 (6.7, 7.4)
South Atlantic	20.1 (19.5, 20.7)
East South Central	6.3 (6.0, 6.7)
West South Central	11.6 (11.1, 12.1)
Mountain	7.3 (7.0, 7.7)
Pacific	13.6 (13.1, 14.1)

^aMedical use indicates any medical cannabis use with or without recreational use.

^bNH = Non-Hispanic.

^cIncome from unemployment/governmental assistance.

Modeled developmental trends for cannabis use type prevalence (years combined) are shown in Figure 1 (see Appendix Table 3 for age-specific prevalence estimates). Among all respondents, recreational-only use prevalence was approximately 40% across ages 19 to 28 (slope −0.001, p=0.896), decreased across ages 28 to 40 (to 19.8%; slope −0.072, p=0.020), and then continued to decrease (at a slower rate) across ages 40 to 65 (to 12.6%; slope −0.014, p=0.004). There were no significant developmental trends in medical use among all respondents. Among those reporting past 12-month use, medical use prevalence increased linearly across age from 5.2% to 16.8% across ages 19 to 65 (slope 0.011, p<.001).

Figure 1. Developmental trends from ages 19–65 in cannabis use type among U.S. adults, 2018–2023.

Notes: Unweighted n = 40,926 all respondents; 10,219 respondents reporting past 12-month use. Medical use indicates any medical cannabis use with or without recreational use. Parentheses show standard errors for all slope estimates.

Modeled historical trends from 2018–2023 for cannabis use type estimates (ages combined) stratified by state policy type are shown in Appendix Table 4. There was not a significant historical trend in recreational-only use prevalence in any of the three state policy groups. Medical use prevalence increased across historical time from 1.1% in 2018 to 5.4% in 2023 in states with medical use only (0.008, p=0.002), but there were no significant historical trends in other states (Appendix Figure 1).

Table 2 provides multivariable associations between covariates and cannabis use type. The odds of medical versus no use were higher for males and respondents identifying as other sex or prefer not to answer (vs. females); non-Hispanic White (vs. Hispanic or non-Hispanic Asian) respondents; those not married (vs. married); those receiving income assistance (vs. not); those living in states with either legal medical use only or legal recreational+medical use (vs. no legal use); and those ages 35–45 (vs. 50–65). The odds of medical versus recreational-only use were higher for males (vs. females); those living in states with either legal medical use only or legal recreational+medical use (vs. no legal use); and those ages 50–65 (vs. 19–30). The odds of reporting recreational-only versus no use were higher for: males and respondents identifying as other sex or prefer not to answer (vs. females); non-Hispanic White (vs. Hispanic, non-Hispanic Asian, and non-Hispanic Black) respondents; respondents identifying as another non-Hispanic race/ethnicity or multiracial (vs. non-Hispanic White); those not married (vs. married) or without children (vs. with children); those receiving income assistance (vs. not); those in states with legal recreational+medical use (vs. no legal use); and those ages 19–30 and 35–45 (vs. 50–65).

Table 2.

Associations between sociodemographics, state policy, and past 12-month cannabis use type among U.S. adults

				MODEL 1		MODEL 2
Variable	No use	Recreational-only use	Medical usea	Recreational-only use vs. no use	Medical use vs. no use	Medical use vs. recreational-only use
	% (95% CI)	% (95% CI)	% (95% CI)	AOR (95% CI) p	AOR (95% CI) p	AOR (95% CI) p
Sex
Male	72.3 (71.3, 73.3)	24.8 (23.8, 25.7)	2.9 (2.6, 3.3)	1.18 (1.09, 1.27) <.001	1.52 (1.27, 1.83) <.001	1.29 (1.07, 1.56) 0.008
Other	48.2 (41.7, 54.8)	46.6 (40.0, 53.2)	5.2 (2.2, 8.2)	1.76 (1.32, 2.35) <.001	2.62 (1.37, 4.99) 0.003	1.49 (0.79, 2.82) 0.222
Female	75.7 (74.9, 76.5)	22.2 (21.3, 23.0)	2.2 (1.9, 2.4)	(ref)	(ref)	(ref)
Race
Hispanic	68.0 (65.6, 70.3)	29.6 (27.4, 31.9)	2.4 (1.8, 3.0)	0.81 (0.72, 0.93) 0.002	0.60 (0.43, 0.82) 0.001	0.73 (0.53, 1.01) 0.055
NH Asian	78.7 (75.3, 82.2)	19.9 (16.6, 23.2)	1.4 (0.6, 2.2)	0.54 (0.43, 0.67) <.001	0.28 (0.15, 0.50) <.001	0.51 (0.29, 0.93) 0.027
NH Black	76.7 (74.4, 79.0)	21.0 (18.9, 23.1)	2.3 (1.3, 3.3)	0.81 (0.70, 0.94) 0.005	0.88 (0.56, 1.39) 0.588	1.09 (0.69, 1.74) 0.714
NH Other/multiracialb	65.4 (61.5, 69.2)	31.1 (27.3, 35.0)	3.5 (2.3, 4.7)	1.24 (1.03, 1.50) 0.022	1.10 (0.75, 1.61) 0.622	0.89 (0.60, 1.31) 0.548
NH White	74.8 (74.1, 75.5)	22.6 (21.9, 23.2)	2.6 (2.4, 2.8)	(ref)	(ref)	(ref)
Married
No	63.7 (62.6, 64.8)	33.2 (32.1, 34.3)	3.1 (2.8, 3.5)	1.68 (1.54, 1.83) <.001	1.88 (1.53, 2.32) <.001	1.12 (0.90, 1.39) 0.300
Yes	82.9 (82.3, 83.6)	15.0 (14.4, 15.6)	2.0 (1.8, 2.3)	(ref)	(ref)	(ref)
Parental status
Do not have children	62.6 (61.4, 63.8)	34.5 (33.4, 35.7)	2.9 (2.5, 3.2)	1.28 (1.16, 1.40) <.001	1.01 (0.77, 1.32) 0.963	0.79 (0.60, 1.04) 0.094
Have children	81.9 (81.2, 82.5)	15.8 (15.2, 16.4)	2.3 (2.1, 2.6)	(ref)	(ref)	(ref)
Employment status
Working full-time job	75.4 (74.6, 76.2)	22.2 (21.5, 23.0)	2.4 (2.1, 2.6)	0.99 (0.93, 1.07) 0.876	0.84 (0.69, 1.03) 0.086	0.85 (0.69, 1.04) 0.107
Not doing so	70.3 (69.2, 71.4)	26.8 (25.7, 27.8)	2.9 (2.6, 3.3)	(ref)	(ref)	(ref)
Income assistance
Yes	64.1 (62.0, 66.1)	31.8 (29.8, 33.8)	4.2 (3.2, 5.1)	1.32 (1.18, 1.46) <.001	1.77 (1.35, 2.32) <.001	1.35 (1.02, 1.78) 0.035
No	74.5 (73.8, 75.2)	23.2 (22.5, 23.9)	2.3 (2.1, 2.5)	(ref)	(ref)	(ref)
State marijuana policyc
Recreational+medical	66.4 (65.1, 67.6)	29.5 (28.4, 30.7)	4.1 (3.6, 4.6)	1.42 (1.25, 1.62) <.001	6.58 (4.17, 10.38) <.001	4.63 (2.90, 7.38) <.001
Medical use only	75.2 (74.3, 76.2)	21.7 (20.7, 22.6)	3.1 (2.7, 3.4)	1.04 (0.94, 1.15) 0.428	6.57 (4.40, 9.80) <.001	6.31 (4.19, 9.48) <.001
No legal use	78.9 (77.8, 80.0)	20.6 (19.5, 21.7)	0.5 (0.3, 0.7)	(ref)	(ref)	(ref)
Age
19–30	58.6 (57.0, 60.1)	38.9 (37.4, 40.4)	2.5 (2.1, 3.0)	2.96 (2.67, 3.28) <.001	1.30 (0.93, 1.80) 0.121	0.44 (0.31, 0.61) <.001
35–45	77.6 (76.7, 78.5)	19.3 (18.5, 20.1)	3.1 (2.8, 3.5)	1.66 (1.54, 1.79) <.001	1.85 (1.56, 2.20) <.001	1.11 (0.93, 1.33) 0.241
50–65	85.1 (84.5, 85.7)	12.8 (12.2, 13.4)	2.1 (1.9, 2.3)	(ref)	(ref)	(ref)
Year
2018	79.2 (77.9, 80.5)	18.7 (17.4, 19.9)	2.2 (1.7, 2.7)	0.85 (0.77, 0.95) 0.005	0.84 (0.64, 1.09) 0.190	0.98 (0.73, 1.30) 0.877
2019	78.4 (77.2, 79.7)	19.6 (18.3, 20.8)	2.0 (1.6, 2.4)	0.91 (0.81, 1.03) 0.125	0.72 (0.53, 0.98) 0.039	0.79 (0.57, 1.09) 0.155
2020	76.7 (75.3, 78.0)	20.6 (19.3, 21.9)	2.7 (2.1, 3.3)	(ref)	(ref)	(ref)
2021	70.8 (69.4, 72.2)	26.7 (25.4, 28.1)	2.5 (2.1, 2.9)	0.98 (0.88, 1.10) 0.728	0.83 (0.60, 1.15) 0.266	0.85 (0.61, 1.19) 0.340
2022	70.0 (68.5, 71.5)	27.2 (25.8, 28.7)	2.8 (2.3, 3.2)	0.99 (0.88, 1.10) 0.797	0.97 (0.72, 1.32) 0.848	0.99 (0.72, 1.35) 0.925
2023	70.8 (69.5, 72.2)	26.1 (24.8, 27.4)	3.0 (2.6, 3.5)	1.05 (0.94, 1.18) 0.366	1.14 (0.86, 1.53) 0.360	1.09 (0.80, 1.47) 0.589

Notes: Unweighted n = 40,926 (missing covariate data addressed using full information maximum likelihood). AOR = adjusted odds ratio. CI = confidence interval. NH = Non-Hispanic. Model 1 used “no use” as the outcome referent category in the multinomial regression model; Model 2 used “recreational-only use” as the outcome referent category. Although not shown here, both multinomial regression models also included U.S. Census division. Boldface indicates statistical significance (p≤0.028).

^aMedical use indicates any medical cannabis use with or without recreational use.

^bNH Other/multiracial = another NH race/ethnicity or multiracial.

^cState policy categories: recreational+medical = policy allowing legal adult recreational and medical use; medical use only = policy allowing legal medical use only; no legal use = policy allows no legal recreational or medical use.

Table 3 presents associations between cannabis use type, covariates, and past 30-day cannabis use at the respondent level among those reporting past 12-month use (covariate associations not discussed). The odds of any past 30-day use were higher among respondents reporting medical use versus recreational-only use. Use frequency was more than twice as high for the medical use group (21.2 occasions) versus recreational-only use group (10.2 occasions). As frequent or daily or near-daily cannabis use often is defined as use on 20+ occasions in the past 30 days,³ respondents reporting medical use on average engaged in daily or near-daily use. Models including age group × medical use interaction terms (not tabled) showed no significant interactions. Observed higher prevalence and frequency among those reporting medical use did not vary by age.

Table 3.

Associations with past 30-day cannabis use among U.S. adults reporting past 12-month cannabis use

Variable	Past 30-day prevalence		Past 30-day frequency
	% (95% CI)	AOR (95% CI) p	Mean (95% CI)	aIRR (95% CI) p
Past 12-month cannabis use type
Medical usea	89.2 (86.8, 91.5)	4.74 (3.67, 6.13) <.001	21.2 (20.0, 22.5)	2.11 (1.96, 2.27) <.001
Recreational-only use	63.8 (62.4, 65.2)	(ref)	10.2 (9.8, 10.7)	(ref)
Sex
Male	68.0 (66.1, 69.9)	1.23 (1.09, 1.39) 0.001	12.3 (11.7, 13.0)	1.28 (1.18, 1.38) <.001
Other	74.3 (65.8, 82.8)	1.45 (0.93, 2.26) 0.104	13.8 (10.9, 16.7)	1.27 (1.01, 1.60) 0.040
Female	64.0 (62.3, 65.8)	(ref)	10.0 (9.4, 10.6)	(ref)
Race
Hispanic	64.1 (60.2, 68.0)	0.87 (0.72, 1.06) 0.177	10.0 (8.8, 11.2)	0.83 (0.72, 0.95) 0.007
NH Asian	62.4 (54.5, 70.4)	0.83 (0.58, 1.19) 0.308	6.6 (4.8, 8.4)	0.58 (0.44, 0.76) <.001
NH Black	71.8 (67.2, 76.3)	1.32 (1.04, 1.68) 0.023	13.0 (11.3, 14.8)	1.13 (0.98, 1.31) 0.101
NH Other/multiracialb	69.6 (62.8, 76.5)	1.16 (0.83, 1.62) 0.380	12.5 (10.3, 14.8)	1.03 (0.86, 1.24) 0.761
NH White	65.7 (64.3, 67.1)	(ref)	11.3 (10.8, 11.8)	(ref)
Married
No	67.8 (66.1, 69.5)	1.18 (1.04, 1.35) 0.014	11.8 (11.3, 12.4)	1.14 (1.04, 1.25) 0.007
Yes	63.2 (61.4, 65.0)	(ref)	10.1 (9.5, 10.7)	(ref)
Parental status
Do not have children	67.0 (65.2, 68.8)	0.99 (0.86, 1.15) 0.930	11.4 (10.8, 12.0)	0.95 (0.87, 1.05) 0.340
Have children	65.1 (63.4, 66.8)	(ref)	11.1 (10.5, 11.7)	(ref)
Employment status
Working full time job	65.0 (63.4, 66.7)	0.91 (0.81, 1.03) 0.124	10.7 (10.2, 11.2)	0.90 (0.83, 0.97) 0.004
Not doing so	68.3 (66.4, 70.3)	(ref)	12.1 (11.4, 12.8)	(ref)
Income assistance
Yes	72.4 (69.2, 75.5)	1.28 (1.08, 1.52) 0.006	14.3 (13.1, 15.5)	1.27 (1.16, 1.40) <.001
No	65.1 (63.7, 66.6)	(ref)	10.6 (10.2, 11.1)	(ref)
State marijuana policyc
Recreational+medical	67.9 (65.9, 69.9)	0.89 (0.72, 1.11) 0.315	11.7 (11.0, 12.4)	0.97 (0.84, 1.13) 0.713
Medical use only	66.7 (64.7, 68.7)	0.95 (0.80, 1.12) 0.550	11.6 (10.9, 12.3)	0.97 (0.86, 1.09) 0.620
No legal use	63.3 (60.6, 66.0)	(ref)	10.2 (9.3, 11.1)	(ref)
Age
19–30	67.3 (65.3, 69.3)	1.01 (0.86, 1.20) 0.879	11.7 (11.0, 12.4)	1.13 (1.01, 1.26) 0.037
35–45	63.6 (61.6, 65.7)	0.86 (0.76, 0.98) 0.028	10.6 (9.9, 11.2)	0.97 (0.89, 1.06) 0.462
50–65	67.1 (65.1, 69.1)	(ref)	11.0 (10.4, 11.7)	(ref)
Year
2018	65.4 (62.0, 68.8)	1.17 (0.95, 1.43) 0.146	11.4 (10.3, 12.5)	1.06 (0.94, 1.20) 0.332
2019	65.8 (62.5, 69.0)	1.20 (0.98, 1.47) 0.075	10.4 (9.5, 11.4)	0.98 (0.86, 1.11) 0.714
2020	62.1 (58.9, 65.3)	(ref)	10.9 (9.9, 11.9)	(ref)
2021	68.5 (65.8, 71.1)	1.30 (1.07, 1.57) 0.008	12.0 (11.0, 12.9)	1.03 (0.91, 1.17) 0.617
2022	66.5 (63.6, 69.3)	1.22 (1.00, 1.48) 0.045	11.2 (10.3, 12.1)	1.02 (0.90, 1.14) 0.794
2023	67.2 (64.6, 69.8)	1.26 (1.04, 1.52) 0.017	11.3 (10.5, 12.1)	1.03 (0.91, 1.17) 0.604

Notes: Unweighted n = 10,219 (missing covariate data addressed using full information maximum likelihood). aIRR = adjusted incidence rate ratio. AOR = adjusted odds ratio. CI = confidence interval. NH = Non-Hispanic. Although not shown here, both regression models also included U.S. Census division. Boldface indicates statistical significance (p≤0.028).

^aMedical use indicates any medical cannabis use with or without recreational use.

^bNH Other/multiracial = another NH race/ethnicity or multiracial.

^cState policy categories: recreational+medical = policy allowing legal adult recreational and medical use; medical use only = policy allowing legal medical use only; no legal use = policy allows no legal recreational or medical use.

Discussion

In this national U.S. young and middle adult sample, past 12-month medical cannabis use was reported by almost 1 in 10 of those reporting past 12-month use compared with approximately 3% of all respondents. Developmental trends in recreational-only prevalence decreased after young adulthood; there was no developmental trend change in medical prevalence among respondents overall. In states with medical use only policy, historical trends in medical prevalence increased from 2018–2023. Sociodemographic characteristics and state policy were associated with cannabis use type; medical use was associated with higher past 30-day prevalence and frequency. Results highlight the need for increased awareness of connections between medical use and high-frequency use across adulthood, and call for research examining whether consequences differ between daily or near-daily medical versus recreational use.

In the full population, the prevalence of recreational-only use decreased across age, while the prevalence of medical use was stable. As a result of this shift in cannabis use away from recreational-only and toward medical use, the proportion of all people using cannabis who reported recreational-only use decreased across age, and the proportion reporting medical use increased across age. This observation does not imply causality; additional factors likely underly decreasing recreational-only use and stable medical use across age. The current findings of stable developmental trends in medical use prevalence among all respondents contrast with earlier national studies finding medical use was higher at ages 18–49 (vs. 50+) during 2013–2015^8, and at ages 26–45 (vs. 16–25 or 46+) in 2018.¹⁵ Given rapid historical increases in U.S. medical cannabis program enrollment,^13,14 age distributions of medical cannabis use may be shifting with historical time. That is, although developmental trends in overall medical prevalence were stable, the individual likelihood of reporting medical use versus non-use or recreational-only use did vary by age.

State policy of medical only use or recreational+medical use was associated with increased odds of medical use among this adult population. These findings are in line with prior research showing medical only policies are associated with increased use among adults but not adolescents.³³ With increasing state legalization of adult recreational cannabis use, more adults are able to obtain cannabis legally without a physician’s medical recommendation and choose whether or not to use for medical reasons. This raises the question of whether identifying use with medical recommendations remains relevant for public health purposes, or if research should focus on medical use motivations instead. If research is focused on prevalence estimates for medical cannabis use, relying solely on use with medical recommendation misses the fact that cannabis purchases and use without a medical recommendation may be medically motivated (e.g., pain, anxiety). Recreational policy is estimated to decrease medical cannabis sales^14,34 and is associated with some degree of individual transition out of exclusively medical use.¹¹ However, recreational policy also is associated with movement from non-use to medical use^9,35 likely through destigmatization and shifts in perceived risks. The current study found medical use increased from 2018–2023 in states with medical only policy; no significant trends were observed in recreational+medical use states. Benefits of medical cannabis program participation—even in states with legal recreational use—include obtaining physician guidance regarding administration routes, dosage, and possible interactions with other medications^21,36 (an issue that grows in importance with age). Other benefits vary with state policy but may include access to higher THC potency products, increases on limits for growing or possession quantity, and state sales or excise tax exclusions.^37,38,39,40 Research focusing on clinical outcomes among those using cannabis for medical purposes with and without medical recommendations is needed.

The current study found male sex was associated with higher odds of medical use. Male sex historically has been associated with higher cannabis prevalence and frequency,^41,42 but sex differences are narrowing.³ Men are more likely to report specialist and primary care physician support for medical cannabis use than women.⁴³ Reasons for medical cannabis use vary by sex: women are more likely to report use for anxiety, depression, and pain while men are more likely to report use for psychosis, alcohol or other drug use, seizures, and tumors.⁴⁴ Further research on sex differences in medical cannabis use is warranted.

Findings regarding higher past 30-day cannabis prevalence and frequency associated with medical use were striking; mean frequency in the medical use group was equivalent to daily or near-daily use. High-frequency use has been associated with increased use disorder⁶ and other acute risks⁴⁵ which may be higher in later adulthood due to metabolism changes, lower cognitive reserves, and increased likelihood of drug interactions and comorbidities.⁴⁶ Medical cannabis dosing recommendations for some conditions (e.g., chronic pain) involve daily administration,⁴⁷ raising questions regarding how to evaluate high-frequency use risk among those using medical cannabis. Additional research is needed examining how high-frequency use risks may differ between medical and non-medically motivated use in regards to when, how, and with what potency cannabis is used. Regardless of these differences, however, using cannabis from a medical recommendation was associated with increased high-frequency use across age, possibly increasing risk for associated negative outcomes.

Limitations

This study was based on nationally-representative U.S. 12^th-grade student samples; individuals not attending school in 12^th grade were not included. School dropout is associated with higher cannabis use;^48,49 the sample may underrepresent adults with adolescent cannabis use experience. Only general state-level cannabis policy coding and a single-item measure of medical cannabis use were used. Future research could explore the effects of (a) policy implementation and enforcement, (b) more detailed medical cannabis use measures including specific sources, use methods, frequency, and quantity consumed, and (c) confounders such as psychiatric and medical disorders.

Conclusions

Past 12-month medical cannabis use was reported by 2.6% of U.S. young and middle adults; about one in ten who used cannabis did so with a medical recommendation. The proportion of those using cannabis reporting medical use became larger with age, and was more prevalent where state policy allowed medical or recreational use. Medical cannabis use was a risk factor for high-frequency use.