A study of nearly 860,000 U.S. adults over eight years finds that recreational cannabis legalization primarily brings new users into the market, particularly older adults, women, and college-educated individuals, rather than increasing how often existing users consume cannabis. These findings carry significant implications for clinical screening and patient counseling in states that have recently legalized.

As recreational cannabis legalization spreads across the United States, clinicians need to know not just whether more people are using cannabis but precisely which patients are newly initiating use. Older adults face heightened risk from cannabis due to polypharmacy, altered metabolism, and increased fall susceptibility. Women managing chronic conditions may have unique pharmacokinetic considerations. If legalization is quietly drawing these traditionally lower-use populations into cannabis consumption for the first time, the clinical imperative shifts from monitoring known users to proactively identifying and counseling newly initiated ones, especially those least likely to volunteer the information.

Recreational cannabis legalization has expanded rapidly across the United States, but how this policy change reshapes actual cannabis use patterns among adults has remained an open and clinically urgent question. This study leveraged the staggered timing of state legalization laws as a natural experiment, applying a difference-in-differences design to eight years of Behavioral Risk Factor Surveillance System (BRFSS) data from 38 states. Zero-inflated negative binomial and probit regression models separately estimated the probability of any cannabis use and the frequency of use among those already using, with state and year fixed effects to control for time-invariant confounders and secular trends.

The headline finding is a 0.94 percentage point absolute increase in any cannabis use associated with legalization (a 9.8% relative increase from the 2016 baseline of 9.6%), with 44% lower odds of reporting zero use. Critically, there was no significant increase in use frequency among existing users. Subgroup analyses revealed that the largest relative increases occurred among adults aged 60 and older (24 to 57% relative increases), women (18.7%), college graduates, and White adults (18.9%). Paradoxically, adults without a high school degree showed a significant decrease in use. These findings are based entirely on self-reported data from telephone surveys with response rates of 45 to 54%, and the authors note that differential social desirability bias in legalizing versus non-legalizing states cannot be ruled out. They call for longitudinal studies tracking individual-level trajectories and linking use changes to clinical outcomes.

New Users, Not More Use: What Eight Years of Data Tell Us About Cannabis Legalization

When states legalize recreational cannabis, conventional wisdom assumes it mostly gives existing users permission to be more open about what they were already doing. New data covering nearly 860,000 Americans over eight years suggest the reality is more surprising and more clinically consequential. This study gets something genuinely right that most prior analyses glossed over: the critical distinction between whether more people are using cannabis and whether people who already use cannabis are using it more often. Think of it this way. The question is not whether coffee drinkers are drinking more coffee after a new cafe opens on the corner. The question is whether people who never drank coffee before are now stopping in for a cup. That distinction matters enormously in clinical practice, because new initiates in their sixties and seventies carry a fundamentally different risk profile than a thirty-year-old who has used cannabis intermittently for a decade. The finding that legalization’s behavioral footprint lands disproportionately on older adults, women, and college-educated White Americans, groups that historically used less, reframes where clinicians should direct their attention.

That said, the study’s most important limitation deserves honest airing. Self-reported cannabis use in a telephone survey is sensitive to whether the respondent feels it is legal to use. Imagine asking people in a newly opened restaurant whether they enjoy dining out more frequently than asking people who still have to eat at home: the venue itself changes how honestly they answer, independent of whether they are actually eating more. In states that have legalized, respondents may simply be more candid, producing a statistical “increase” that partly reflects honesty rather than behavior change. The authors acknowledge this but do not weight it heavily enough in their conclusions. The parallel trends assumption, tested through event study analysis, holds up reasonably in the pre-legalization period, yet pre-trend parallelism does not guarantee post-treatment parallelism. Unmeasured shifts in cannabis culture, advertising, or enforcement intensity in legalizing states could violate this assumption in ways the data cannot detect. The paradoxical decrease in use among adults without a high school degree is particularly intriguing and underdeveloped. It might reflect differential market access, pricing barriers, or shifts in social norms, but the paper leaves this finding largely unexplored.

What I would tell a patient who is curious about cannabis for the first time after their state legalized is simple: please talk with me first, not to discourage you, but because your medications, your metabolism, and your health conditions create considerations worth a five-minute conversation. What I would tell a colleague is that we need to normalize asking about cannabis use, especially among older, female, and educated patients who are the least likely to volunteer the information and the most likely to be newly initiating. What I would tell a policymaker is that public health frameworks designed around the traditional cannabis user demographic need updating, because legalization is reaching new populations with distinct vulnerabilities. The broader lesson is one that applies well beyond cannabis: in policy research, who changes their behavior matters as much as whether behavior changes at all. A modest population-level prevalence increase concentrated in high-risk demographic subgroups can have outsized clinical and public health consequences that aggregate statistics conceal.

This study sits within a growing body of quasi-experimental policy research that has gradually shifted the field’s understanding from broad aggregate estimates toward subgroup-specific effects. Earlier work by Cerda and colleagues (2020) found modest increases in overall use and cannabis use disorder through 2016, but lacked the temporal depth and demographic granularity offered here. By extending through 2023 and testing explicit interactions across age, sex, education, and race, Hawkins and colleagues advance the evidence base from “does legalization increase use?” to the more clinically actionable question of “in whom does it increase use?”

For clinicians, the pharmacological implications of these demographic shifts deserve careful attention. Older adults newly initiating cannabis face heightened sensitivity to THC due to reduced hepatic metabolism, increased body fat distribution altering drug distribution, and a higher prevalence of concomitant medications with potential cannabis interactions, including anticoagulants, benzodiazepines, and antihypertensives. Women may metabolize THC differently due to hormonal variation and body composition. Clinicians practicing in states that have recently legalized should consider integrating a routine, nonjudgmental cannabis screening question into annual wellness visits, with particular attention to patients aged 55 and older and women managing chronic conditions.

This is an original quantitative study using a quasi-experimental difference-in-differences design applied to repeated cross-sectional survey data. It occupies a middle tier of the causal evidence hierarchy: stronger than standard observational cross-sectional studies because it exploits variation in policy timing as a natural experiment, but weaker than a randomized trial. Its validity rests on the parallel trends assumption, the premise that cannabis use trajectories in legalizing and non-legalizing states would have remained similar absent the policy change, which is supported by pre-trend analysis but can never be definitively proven.

This study extends and refines earlier findings from Cerda and colleagues (2020), who documented modest overall use increases through 2016 using a similar difference-in-differences approach but with less temporal reach and coarser demographic stratification. By covering 2016 through 2023, a period during which many additional states legalized, the present analysis captures a more mature policy landscape. The finding that effects concentrate in older and female adults is consistent with emerging national survey trends showing cannabis use growing fastest among adults over 50, but provides stronger quasi-causal framing than descriptive trend analyses alone. Importantly, the null finding on use frequency among existing users adds nuance to prior work suggesting possible increases in cannabis use disorder, because it implies that new initiation, not escalation, is the primary mechanism for any population-level increase in problematic use.

The most consequential analytic choice is the reliance on self-reported cannabis use as the outcome variable. If differential reporting bias accounts for even a fraction of the observed legalization effect, meaning people in legalizing states are simply more willing to admit use rather than actually using more, then the true behavioral effect could be substantially smaller. An alternative analysis incorporating objective dispensary sales data or biomarker validation as a cross-check could meaningfully alter the magnitude of estimated effects. Additionally, the study tested multiple subgroup interactions without formal correction for multiple comparisons; a Bonferroni or false discovery rate adjustment could render some of the smaller subgroup effects, such as specific age-band estimates, non-significant, though the overall pattern of demographic heterogeneity would likely persist.

The most likely overinterpretation is conflating the extensive-margin finding (more people trying cannabis) with an intensive-margin finding (existing users consuming more). These two patterns carry very different implications for cannabis use disorder risk and clinical intervention strategy. The study explicitly finds no significant increase in use frequency among existing users. A related misreading is assuming the study proves that legalization caused these use increases. While the difference-in-differences design provides stronger causal inference than standard observational work, it falls short of definitive proof because the parallel trends assumption cannot be empirically verified for the post-treatment period. Finally, readers should resist the assumption that younger adults and racial minorities are most affected by legalization; this study shows the opposite, with older, White, female, and college-educated adults driving the observed increases.

This study contributes the most current and demographically detailed quasi-experimental evidence on how recreational cannabis legalization reshapes adult use in the United States. It establishes that legalization primarily brings new users into the market, not that it intensifies use among existing consumers. It does not establish definitive causation, cannot rule out reporting bias, and does not measure health outcomes. For clinicians, the practical takeaway is clear: proactive, nonjudgmental cannabis screening should now extend beyond traditional user demographics, with particular attention to older adults, women, and patients in recently legalizing states.

Does cannabis legalization make everyone use cannabis more often?

No. This study found that legalization increases the number of people who try or begin using cannabis, but it does not increase how often people who already use cannabis continue to use it. The effect is about new users entering the market, not existing users consuming more.

Who is most likely to start using cannabis after legalization?

The data suggest that adults aged 60 and older, women, college graduates, and White adults showed the largest relative increases in cannabis use after legalization. These are groups that historically used cannabis at lower rates. Younger adults and some racial and ethnic minorities did not show significant increases.

Should I talk to my doctor before trying cannabis for the first time?

Absolutely. If you are an older adult, take medications for chronic conditions, or have never used cannabis before, a conversation with your physician is important. Cannabis can interact with common medications, affect balance and cognition, and behave differently in older bodies. A brief discussion can help you make a safer, better-informed decision.

Does this study prove that legalization is dangerous?

No. The study measures changes in use patterns, not health outcomes. It did not examine whether new cannabis users experienced adverse health events. Whether the observed increases in use among older adults and women translate into clinical harms is an important question that requires further research with linked health outcome data.

Have thoughts on this? Share it: