In 2022, of the 39 jurisdictions allowing medical cannabis use, 34 (87.2%) reported the number of patients (Table 1) and 18 (48.7%) reported the number of patient-reported qualifying conditions (Figure 1). However, some jurisdictions reported data only in certain years, did not report complete data (such as Illinois, which reported cumulative patient totals but not cumulative patient-reported qualifying conditions), or did not consistently publish reports. For example, New York published biennial reports in 2016, 2018, and 2022 but not 2020. Data sources are shown in Supplement Table 1.

Total patient number increased by 33.3% during the study period, from 3 099 096 in 2020 (in 30 reporting jurisdictions) to 4 132 098 in 2022 (in 34 reporting jurisdictions), with a 23.0% increase in the population prevalence of patients licensed for medical cannabis in this same period (175.0 per 10 000 in 2020 to 215.2 per 10 000 in 2022) (Table 1). In medical-only jurisdictions, the number of patients generally increased or stayed the same during this period; overall, it increased from 206.9 to 300.6 patients per 10 000 from 2020 to 2022 (Table 1). By contrast, only 2 adult-use jurisdictions had increasing enrollment (Massachusetts and Maine), whereas 13 (Arizona, Colorado, Illinois, Michigan, Montana, Nevada, New Jersey, New Mexico, New York, Oregon, Rhode Island, Vermont, and Washington) had decreasing enrollment by the end of the study period after the opening of adult-use dispensaries. Adult-use jurisdictions had 146.5 patients per 10 000 persons in the population in 2020 compared with 124.8 in 2022 (Table 1). This decreasing enrollment was most dramatic in Arizona, which had 295 295 patients in 2020 (411.3 per 10 000 persons in the population) and only 129 836 in 2022 (176.4 per 10 000) after the adult-use law was implemented in 2021. Overall, the ratio of the prevalence of patients in medical-only to adult-use jurisdictions increased from 1.4 in 2020 (206.9 vs. 146.5 patients per 10 000 persons in the population) to 2.4 in 2022 (300.6 vs. 124.8 patients per 10 000).

Chronic pain remained the most common patient-reported qualifying condition during the study period but decreased from 1 119 678 instances (65.7% of all patient-reported qualifying conditions) in 2020 to 934 603 (48.4%) in 2022 (Figure 1). The second and third most common patient-reported qualifying conditions in 2022 were anxiety (14.2%) and PTSD (13.0%), respectively; reports of anxiety increased 53-fold from 2020 to 2022 (5067 to 274 556). Overall, the percentage of reported conditions for which cannabis has substantial or conclusive evidence of therapeutic value according to the 2017 NASEM report decreased from 70.4% in 2020 to 53.8% in 2022 (Figure 2). The number of conditions in the “other” category (such as Sjögren syndrome, cerebral palsy, muscular dystrophy, and psychiatric conditions) increased considerably, with substantial contributions from vague categories, such as “all other conditions.”

In the United States, 29 of 39 jurisdictions with active programs (74.3%) reported data on authorizing clinicians in 2022 compared with 24 of 35 jurisdictions with active programs (68.6%) in 2020 (Table 2). There were 29 500 reported clinicians as of 2022 compared with 29 552 in 2020. When we limited states to those that released authorizing clinician data each year (2020 to 2022) and included states whose programs became active during the time frame, the number of authorizing clinicians increased from 24 252 in 2020 to 26 208 in 2022. Overall, in 2022 there were more authorizing clinicians per patient in nonmedical adult-use jurisdictions (12.9 clinicians per 1000 patients) than in medical-only jurisdictions (6.1 clinicians per 1000 patients). There was considerable heterogeneity in the number of authorizing clinicians per patient, ranging from 0.8 clinicians per 1000 patients in Oklahoma to 109 clinicians per 1000 patients in Mississippi (Table 3). In 2022, 53.5% of authorizing clinicians had an MD or DO degree (most of whom were described generally as “physician” in registry reports), 34.4% had an NP degree, and 11.0% were physician assistants. In the 4 jurisdictions that provided data on medical specialties, the most common in 2022 were family or internal medicine (63.4%), physical medicine and rehabilitation (9.1%), and anesthesia or pain (7.9%).

These findings provide the most up-to-date estimates of the number of patients licensed to receive medical cannabis in the United States, as well as the most complete data set on clinicians who authorized medical cannabis. From 2020 to 2022, the total number of reported patients increased 33.3%. Combined with data from our previously published report (9), this analysis indicates a 610% increase in patient number in the United States from 2016 to 2022. The increased number of patients since 2020 was likely driven by the passage of new medical cannabis laws and increasing enrollment in existing programs—largely medical-only programs (9). Indeed, most jurisdictions with active laws allowing nonmedical adult use reported decreased enrollment. This decrease may result from patients opting out because they no longer need legal cover for nonmedical adult use, because of licensing fees or the inconvenience of certification visits, or because they are using nonmedical adult-use products medically (12). Rapid growth in patient numbers also likely reflects increasing acceptance of cannabis: A 2022 national Pew poll reported that 88% of Americans agreed that cannabis should be legal for medical purposes (13). Further, the study period encompassed the COVID-19 pandemic, during which many medical cannabis markets were designated “essential” industries, thereby increasing access through curbside pickup, delivery services, and telemedicine visits for licensure (14).

The proportion of patient-reported qualifying conditions with a substantial or conclusive evidence basis decreased from 70.4% (2020) to 53.8% (2022), partly driven by increasing licensure for anxiety and PTSD in 2022 (limited evidence; 14.2% and 13.0% of conditions, respectively). In addition, incomplete reporting of medical conditions for licensure (that is, “other conditions” in Figure 2) resulted in increased numbers of patients indicating conditions that lacked a substantial or conclusive evidence basis because vague categories, such as “all other medical conditions,” cannot be evaluated for efficacy. The COVID-19 pandemic may have affected the number of people who obtained cannabis licensure for anxiety and PTSD because pandemic-related stress often resulted in worsened mental health conditions (15). Some studies have suggested associations between the pandemic and increased cannabis use, especially among people with mental health conditions (16, 17). The increase in licensure for anxiety (5067 in 2020 to 274 556 in 2022) is likely also due to jurisdictions with large patient populations and newer laws, such as Pennsylvania, allowing licensure for anxiety. Chronic pain remained the most common patient-reported qualifying condition, constituting 65.7% of total conditions in 2020 and 48.4% in 2022. This finding may be due to the high population prevalence of chronic pain (18) and the fact that it is commonly comorbid with other patient-reported qualifying conditions, such as cancer or multiple sclerosis (19). Further, inadequate relief from conventional pain medications (20, 21), including opioids (22, 23), has led some patients to seek alternative treatment options (24). Indeed, a growing body of observational literature shows that some people substitute cannabis for pain medications, largely because of fewer reported adverse effects and better symptom management (24–26). These findings have contributed to legislative updates allowing cannabis in place of opioids or other pain medications (27).

The authorizing clinician data in the current study provide a complementary angle to view national medical cannabis trends. The range in number of clinicians per jurisdiction may reflect differential requirements for authorizing cannabis. For instance, an analysis of the 34 jurisdictions (33 states and Washington, DC) with medical cannabis laws in 2019 showed that only 9 states and Washington, DC, required clinicians to register with a state or jurisdiction program to certify patients and only 9 jurisdictions required the clinician to complete a course or training to certify patients (28). The clinical support received by medical patients from their authorizing clinician may also differ because authorization volumes vary widely, exemplified by data from Colorado’s registry in 2021, where 67.8% of authorizing clinicians (221 of 326) recommended cannabis for 20 or fewer patients and 7.7% (25 of 326) authorized it for 1000 or more patients (29). The top 3 clinicians in Colorado wrote 6538, 6340, and 6170 recommendations for medical cannabis in 2021 alone. Last, we note that on average, medical-only jurisdictions have substantially fewer authorizing clinicians per 1000 patients: 6.1 in 2022 compared with 12.9 in adult-use jurisdictions. This may be because medical-only jurisdictions have more licensed patients on average. It may also be a function of program longevity: Adult-use jurisdictions typically have longer-standing programs, and clinicians may become comfortable authorizing cannabis due to the more accepting cultural context.

Our results highlight the need for better surveillance methods to adequately understand outcomes of medical cannabis use and thoughtful strategies and public health efforts to reduce harms from increased cannabis availability. The Food and Drug Administration has not approved any cannabinoid-based products for the most common patient-reported qualifying conditions. Therefore, real-world data collection on health outcomes among patients licensed for medical cannabis (for example, in Florida [30], Minnesota [31], and the United Kingdom [32]) could aid in the development of cannabis therapeutics, such as by increasing understanding of which formulations or cannabinoid content are useful or harmful in specific populations. To limit negative health effects associated with cannabis use, the following could be considered: implementing appropriate safety and delta-9-tetrahydrocannabinol concentration testing for cannabis products, clarifying the uncertain boundaries between medical and nonmedical markets, and increasing education and training for clinicians and cannabis dispensary employees (33). In addition, despite increased use of cannabis for medical and nonmedical purposes, enforcement of cannabis prohibition continues to disparately affect groups that have been marginalized for racial or socioeconomic reasons (5, 34).

This study has some limitations. Contents of registry reports are inconsistent and may vary by year (for example, Illinois did not report the number of authorizing physicians after 2020). The limited jurisdictions included may not generalize to other jurisdictions in the United States, especially for reports of patient-reported qualifying conditions. We also drew data from varied sources, such as Freedom of Information Act requests and program websites, which may affect data quality. California’s lack of reliable data and omission from the current analysis biases our results because it has the oldest medical cannabis program and the largest population of any jurisdiction. Because we used aggregated reports rather than individual-level data, we have no information on the primary patient-reported qualifying condition for each patient, outcomes of use, or why some patients chose not to renew their licenses after nonmedical adult-use laws were implemented. We classified the specialty of authorizing clinicians on the basis of their most advanced clinical training, which may lose some granularity in aggregate. Last, although the NASEM report is the most comprehensive data source on cannabis efficacy for health conditions, these categories are imperfect for the following 3 reasons. First, they are broad. Second, the studies included in the 2017 NASEM report used products that do not generalize to those in the legal cannabis marketplace because products in the existing market generally do not meet standards for pharmaceutical-grade manufacturing practices, they have different formulations (for example, no dispensaries sell synthetic delta-9-tetrahydrocannabinol [that is, dronabinol]), and the delta-9-tetrahydrocannabinol and cannabidiol concentrations in cannabis flower in existing clinical trials are far lower than what is found in the marketplace. And third, as research continues, cannabinoids are being shown to be effective for more conditions, such as cannabidiol for Lennox–Gastaut syndrome and Dravet syndrome (35, 36). This categorization scheme is especially imperfect for chronic pain given the heterogeneity of clinical presentations and the variability in treatment response depending on underlying mechanism (37).

In conclusion, among jurisdictions with available data, medical cannabis licensure has increased since 2020, from 3 099 096 patients (175.0 per 10 000 persons in the population) to 4 132 098 patients (215.2 per 10 000) in 2022. However, licensure decreased in 13 of 15 adult-use jurisdictions. The number of authorizing clinicians varied widely by jurisdiction, with approximately 7.7 authorizing clinicians per 1000 licensed patients overall. The percentage of patient-reported qualifying conditions without a substantial evidence basis increased from 29.6% (2020) to 46.2% (2022). Although chronic pain remains the most common patient-reported qualifying condition, this finding should be interpreted with caution given the broad nature of this category and the heterogeneity of chronic pain (38). Improved reporting efforts from states on patient characteristics (such as cannabis use patterns and length of use) as well as clinician credentials and training may help enhance understanding of this rapidly shifting medical cannabis use landscape. Much work is needed to enhance research and develop clinical guidance for appropriate medical cannabis use.