Abstract
Background:
As greater numbers of states in the United States and countries in the world continue to legalize cannabis for medical use, it has become increasingly important to assess patterns of cannabis use in individuals using cannabis for medical symptoms over time. A public health concern is that, like recreational cannabis, some individuals using cannabis for medical reasons may develop detrimental patterns of use, leading to the development of a cannabis use disorder (CUD).
Methods:
In a 9-month longitudinal cohort study following a 12-week randomized, waitlist-controlled trial in 149 adults who used cannabis to alleviate insomnia, pain, depressed mood, or anxiety (RCT: NCT03224468), we assessed whether patterns of cannabis use for the 9 months following the RCT were associated with the development of CUD.
Results:
We identified five unique trajectories of use; 31 participants (21%) had low stable or no use, 50 (34%) had medium stable use, 19 (13%) had high stable use, 26 (17%) showed de-escalating and 23 (15%) showed escalating use over 9 months following the RCT. Of 149 participants enrolled, 19 (13%) met diagnostic criteria for CUD at 12 months. Only the escalating cannabis use pattern predicted significantly higher rates of CUD compared to the low or no use category (OR = 4.29, 95% CI = 1.21 to 10.87, p = 0.02).
Conclusions:
These data indicate that most individuals using cannabis for medical symptoms have a stable pattern of use over the first year. Escalation of use may be a detrimental pattern that warrants further concern.
Keywords: cannabis, cannabis use disorder, patterns, addiction
1. Introduction
As of September 2021, approximately a dozen countries, including Canada, the United Kingdom, Australia, and thirty-six states and the District of Columbia in the United States, have legalized cannabis for medical use. It has become increasingly important to assess patterns of cannabis use over time in individuals using cannabis for medical symptoms. A significant public health concern is that, like recreational cannabis, some individuals using cannabis for medical reasons may develop escalating use patterns leading to cannabis use disorder (CUD) (Hasin 2018). Those who develop CUD are characterized by the compulsive seeking, loss of control and reinstatement to use cannabis despite significant problems associated with its consumption. Approximately 90% of those seeking treatment for CUD report difficulty achieving and maintaining abstinence (Budney and Hughes 2006), and adults seeking treatment for CUD average more than 10 years of daily or almost daily use and around six attempts to reduce or stop consumption (Budney and Hughes 2006, Copeland, Rooke et al. 2013). These individuals develop physical dependence on the drug, reporting tolerance to many of the effects of delta-9-tetrahydrocannabinol (THC)(Gonzalez, Cebeira et al. 2005, Lichtman and Martin 2005). Those with CUD also develop withdrawal symptoms including anxiety, insomnia, appetite disturbance and depression (Budney and Hughes 2006).
In contrast to medicines that undergo Food and Drug Administration (FDA) review, most non-FDA approved cannabis products available for medical use lack information about dosage, safety, efficacy, and side effects. How much cannabis is used, when, and how often, is left up to the discretion of patients themselves, and little research regarding patterns of use of cannabis has been conducted (Freeman and Lorenzetti 2020, Lorenzetti, Hindocha et al. 2022). Little is known about tolerability, amounts, duration of use, or whether patients reach a stable and regular dose of cannabis versus escalate and exhibit a loss of control over their cannabis use. It is important to determine whether medical cannabis use is more like a stable medication (e.g., blood pressure medication, where a specific dose is taken each day), or whether it is uncontrolled (e.g., like alcohol or medicines with high misuse liabilities, which may be prone to problematic/binge use). To better understand patterns of use, we examined cannabis use over a period of nine months among participants who participated in a longitudinal study following a 3-month clinical trial of medical cannabis in Massachusetts (NCT03224468). Here we analyzed participants’ self-report of medical cannabis use over 9 months of use (in months 3-12, after the initial 3-month randomized clinical trial) and assessed whether these patterns predicted CUD.
II. Methods
This study was approved by the Partners Human Research Committee. All participants provided written informed consent. Study procedures took place between June 2017 and July 2020 and have been previously described in detail (Gilman, Schuster et al. 2022). Briefly, adults, aged 18-65, with desire to use cannabis for depressed mood, pain or insomnia were recruited through advertising and assessed at baseline, at 2 weeks, and then monthly for 12 months after initiating cannabis. Daily cannabis use and CUD at screening or baseline, assessed using the CUD Checklist of the Diagnostic and Statistical Manual of Mental Disorders 5 (DSM-5), was exclusionary. Data for frequency of cannabis use were collected monthly, via phone at months 4, 5, 7, 8, 9, 10, and 11, and via in-person visits at months 3, 6, and 12. Participants were asked to report cannabis use using a 4-point ordinal rating scale for number of days spent using cannabis per month, with ratings: (1) Less than monthly, (2) Less than once a week, (3) 1-4 days a week, and (4) 5 or more days a week. CUD was assessed by doctoral-level or trained staff at 12 months after the beginning of the study, using the DSM-5, with number of symptoms ranging from 0-11. A CUD diagnosis was defined as endorsement of 2 or more symptoms, with 2-3 symptoms considered ‘mild,’ 4-5 symptoms considered ‘moderate,’ and 6 or more symptoms considered ‘severe.’ These categories were collapsed into a binary variable of CUD or no CUD for our prediction models, due to low cell counts of moderate and severe CUD. Finally, participants were asked about their primary method of use (vaped, smoked, or oral use).
We identified cannabis use frequency patterns over the 9-month period by fitting monthly ratings of cannabis use in the follow up period using a 2-knot piecewise multilevel ordinal regression model with subject-varying intercepts and slopes. We generated smoothed estimates of change in cannabis use, computing a person’s average level per month on the latent variable underlying the ordinal rankings, and categorized participants based on the average of their smoothed estimates. This resulted in five distinct characterizations of cannabis use patterns: (low or no use (<1 day per week), medium stable use (1-4 days per week), high stable use (5-7 days week), escalating use (a transition from lower [<1 day per week or 1-4 days per week] to higher [1-4 days/week or 5-7 days per week] use), or de-escalating use (a transition from higher [5-7 days per week or 1-4 days per week] to lower [1-4 days/week or <1 day per week] use). These categories were used as categorical predictors of CUD. The key effects of interest were a set of 4 dummy-coded variables contrasting the low or no use group against the other four groups. We also tested whether method of use was a categorical predictor of trajectory group or of CUD. Missingness was handled naturally via the multilevel modeling approach, with the assumption of missing at random given the model covariates.
III. Results
A total of 149 (of 186 enrolled) participants had complete 12-month data (Table 1). 100 (68%) of these participants had stable use patterns: 31 (21%) had low stable or no use, 50 (34%) had medium stable, and 19 (13%) had high stable use, while 26 (17%) participants showed de-escalating and 23 (15%) showed escalating use (Figure 1A).
Table 1:
Participant characteristics
| Measure | % (n) or Mean (SD) |
|---|---|
| Sample size | 149 |
| Age; M (SD) | 37.5 (14.6) |
| Sex; % (n) | |
| Female | 68.5% (102) |
| Male | 31.5% (47) |
| Race; % (n) | |
| Asian | 6% (9) |
| Black or African American | 5.4% (8) |
| Multi-racial | 2.7% (4) |
| Unknown | 2.7% (4) |
| White | 83.2% (124) |
| Hispanic or Latino; % (n) | 5.4% (8) |
| Education level; % (n) | |
| High school | 3.4% (5) |
| Part college | 20.8% (31) |
| College 2 years | 2% (3) |
| College 4 years | 32.9% (49) |
| Part grad school | 40.3% (60) |
| Unknown | 0.7% (1) |
| Education years; M (SD) | 16.4 (2.3) |
| Cannabis Use Frequency at Baseline | |
| < monthly; % (n) | 57% (85) |
| < weekly; % (n) | 17.4% (26) |
| 1-4 days; % (n) | 23.5% (35) |
| 5-7 days; % (n) | 2% (3) |
| Primary reason for cannabis; % (n) | |
| Anxiety/depressed mood | 40.9% (61) |
| Insomnia | 23.5% (35) |
| Pain | 35.6% (53) |
Figure 1.
A. Latent estimates of participants’ trajectories of cannabis use from 3 to 12 months (obtained from piecewise ordinal regression model) grouped into five categories of use. Orange lines are data from participants who developed CUD, blue lines are data from participants who did not. Black lines and points are within-category averages. B. Percentage of participants diagnosed with CUD at the 12-month follow-up sessions by category of cannabis use trajectory. Error bars represent 95% credible intervals.
At month 12, 19 (13%) participants met diagnostic criteria for CUD, defined as 2 or more out of 11 possible symptoms of CUD; 15 (10.1%) met criteria for mild CUD, 2 (1.3%) met criteria for moderate CUD, and 2 (1.3%) met criteria for severe CUD. Of those with any CUD, use patterns were as follows: 1 (3%) participant had low stable or no use pattern, 8 (16%) had medium stable pattern, 2 (11%) had high stable use, 1 (4%) had de-escalating use, and 7 (30%) had escalating use. Only the escalating cannabis use pattern predicted significantly higher rates of CUD compared to the low or no use category (OR = 4.29, 95% CI = 1.21 to 10.87, p = 0.02) (Figure 1B).
The primary method of cannabis use reported at all time points was vaped cannabis, followed by smoked cannabis, and then oral cannabis (Table 2). Method of use was not associated with likelihood of developing CUD (all ps > 0.3). Method of use was also not associated with membership in any use category.
Table 2:
Measures of cannabis use patterns by time point
| Measure | Month 3 | Month 6 | Month 12 |
|---|---|---|---|
| Sample size | 163 | 163 | 149 |
| Frequency of use; % ((n) | |||
| < weekly | 31.3% (51) | 30.7% (50) | 36.9% (55) |
| 1-4 days | 46.6% (76) | 42.3% (69) | 39.6% (59) |
| 5-7 days | 22.1% (36) | 27.0% (44) | 23.5% (35) |
| Times used per day; M (SD) | 0.98 (0.68) | 1.10 (0.79) | 1.01 (0.69) |
| Mode of use; % (n) | |||
| Oral | 23.9% (39) | 27.0% (44) | 20.8% (31) |
| Smoked | 19.0% (31) | 18.4% (30) | 26.2% (39) |
| Vaped | 30.1% (49) | 33.7% (55) | 34.9% (52) |
IV. Discussion
Among adults using cannabis for medical reasons, cannabis use patterns over a year suggest that most individuals using medical cannabis either maintain (68%) or decrease (17%) their cannabis use frequency pattern over time, though a significant minority (15%) escalate use.
Those who escalated their use were at significant risk for developing CUD compared to the low or no use group. This is not surprising as emerging evidence shows that cannabis can be highly addictive to some individuals; US national data reports that 3 out of 10 cannabis users develop CUD, 23% of whom are symptomatically severe (≥ 6 CUD criteria) (Hasin, Saha et al. 2015). In the current sample of individuals using cannabis for medical symptoms, 13% met criteria for CUD, a lower rate than what is nationally reported for recreational users. These current data suggest that clinicians should be specifically attentive to risk for CUD in those using for medical purposes who escalate their frequency of cannabis use over time. Method of use was not related to development of CUD.
This study has limitations. Though participants reported frequency of use each month, we did not collect measures of potency or dose, and previous analyses of this data indicated that some people who reported cannabis use did not show urinary cannabis metabolites (Gilman, Schmitt et al. 2021). This could indicate that some participants used sham products that did not contain any or very little THC. Participants in this study were using cannabis specifically for depressed mood, pain, or insomnia, so the results may not generalize to those who use cannabis for other medical conditions. Because CUD is a rare outcome, sample size of those who develop CUD is small. Larger studies should be conducted to replicate this finding. While all participants reported medical motives for cannabis use, recreational use was not queried, and therefore participants may also have been using for recreational purposes. These participants were not non-users at baseline, as ∼25% of participants were using cannabis at least weekly at the baseline visit. Finally, the 13% of those who develop CUD include people with only 2 symptoms of CUD; the clinical significance of mild CUD is unclear. Some of the CUD criteria may be associated only with increasing frequency (i.e. tolerance) and not necessarily with problematic or hazardous use. Future studies can examine clinical outcomes of mild CUD, as well as other potentially interesting predictors of CUD including demographic characteristics, more fine-grained patterns of cannabis use (e.g., time of day), and motives for use.
In summary, these observational data indicate that most individuals using cannabis for medical symptoms have a stable pattern of use over the first year. Escalation of use may be a detrimental pattern that may warrant further concern.
Highlights.
Individuals are increasingly using cannabis for medical symptoms.
We assessed patterns of cannabis use for one year after initiating medical cannabis use.
Most individuals using cannabis had a stable pattern of use over the first year.
Those with escalating cannabis use patterns were at higher risk of CUD.
Escalation of use may be a detrimental pattern in medical cannabis users.
Funding/Support:
This work was funded by 5R01DA042043; PI: JMG.
Role of Funder/Sponsor:
The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Conflict of Interest Disclosures:
AEE has served as a consultant to Charles River Analytics (NIDA SBIR grant) and Karuna Pharmaceuticals (Chair Data Monitoring Board). Other investigators report no potential conflicts.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Previous Presentations: This work had not been presented.
Data Sharing Statement:
All data, code, and materials used in the analyses can be provided by Jodi Gilman and Massachusetts General Hospital pending scientific review and a completed data use agreement/material transfer agreement. Requests for all materials should be submitted to Jodi Gilman at jgilman1@mgh.harvard.edu.
References
- Association, A. P. (2013). “Diagnostic and Statistical Manual of Mental Disorders: DSM-5. 5th ed.” [DOI] [PubMed] [Google Scholar]
- Budney AJ and Hughes JR (2006). “The cannabis withdrawal syndrome.” Curr Opin Psychiatry 19(3): 233–238. [DOI] [PubMed] [Google Scholar]
- Copeland J, Rooke S and Swift W (2013). “Changes in cannabis use among young people: impact on mental health.” Curr Opin Psychiatry 26(4): 325–329. [DOI] [PubMed] [Google Scholar]
- Freeman TP and Lorenzetti V (2020). “‘Standard THC units’: a proposal to standardize dose across all cannabis products and methods of administration.” Addiction 115(7): 1207–1216. [DOI] [PubMed] [Google Scholar]
- Gilman JM, Schmitt WA, Wheeler G, Schuster RM, Klawitter J, Sempio C and Evins AE (2021). “Variation in Cannabinoid Metabolites Present in the Urine of Adults Using Medical Cannabis Products in Massachusetts.” JAMA Netw Open 4(4): e215490. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gilman JM, Schuster RM, Potter KW, Schmitt W, Wheeler G, Pachas GN, Hickey S, Cooke ME, Dechert A, Plummer R, Tervo-Clemmens B, Schoenfeld DA and Evins AE (2022). “Effect of Medical Marijuana Card Ownership on Pain, Insomnia, and Affective Disorder Symptoms in Adults: A Randomized Clinical Trial.” JAMA Netw Open 5(3): e222106. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gonzalez S, Cebeira M and Fernandez-Ruiz J (2005). “Cannabinoid tolerance and dependence: a review of studies in laboratory animals.” Pharmacol Biochem Behav 81(2): 300–318. [DOI] [PubMed] [Google Scholar]
- Hasin DS (2018). “US Epidemiology of Cannabis Use and Associated Problems.” Neuropsychopharmacology 43(1): 195–212. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hasin DS, Saha TD, Kerridge BT, Goldstein RB, Chou SP, Zhang H, Jung J, Pickering RP, Ruan WJ, Smith SM, Huang B and Grant BF (2015). “Prevalence of Marijuana Use Disorders in the United States Between 2001-2002 and 2012-2013.” JAMA Psychiatry 72(12): 1235–1242. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lichtman AH and Martin BR (2005). “Cannabinoid tolerance and dependence.” Handb Exp Pharmacol(168): 691–717. [DOI] [PubMed] [Google Scholar]
- Lorenzetti V, Hindocha C, Petrilli K, Griffiths P, Brown J, Castillo-Carniglia A, Caulkins JP, Englund A, ElSohly MA, Gage SH, Groshkova T, Gual A, Hammond D, Lawn W, Lopez-Pelayo H, Manthey J, Mokrysz C, Liccardo Pacula R, van Laar M, Vandrey R, Wadsworth E, Winstock A, Hall W, Curran HV and Freeman TP (2022). “The iCannToolkit: a tool to embrace measurement of medicinal and non-medicinal cannabis use across licit, illicit and cross-cultural settings.” Addiction 117(6): 1523–1525. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
All data, code, and materials used in the analyses can be provided by Jodi Gilman and Massachusetts General Hospital pending scientific review and a completed data use agreement/material transfer agreement. Requests for all materials should be submitted to Jodi Gilman at jgilman1@mgh.harvard.edu.