When Florida legalized smokable medical cannabis in 2019, the amount of THC dispensed per certified patient jumped by roughly 42% and kept climbing. However, this study measures what pharmacies recorded as sold, not what patients actually consumed, and it lacks the health outcome data needed to determine whether the increase was harmful, beneficial, or neutral.

Florida operates the largest medical cannabis program in the United States by patient enrollment, and decisions made about product form availability here send ripple effects across other state programs. When a single legislative change can produce a rapid, population-wide shift in THC dispensing patterns, clinicians certifying patients and policymakers designing program guardrails need to understand the magnitude and the limits of that evidence. This study offers one of the first quantitative estimates of what happens when smokable forms are added to a mature medical cannabis market, and the questions it raises about monitoring, dosing, and safety infrastructure are immediately relevant to every state considering similar expansions.

Medical cannabis programs across the United States have expanded rapidly, yet the population-level pharmacological consequences of specific policy decisions within these programs remain poorly characterized. Florida’s passage of Senate Bill 182 in March 2019, which legalized smokable cannabis flower for certified patients, provided a natural experiment to study exactly this kind of question. The researchers used an interrupted time series design applied to aggregate weekly dispensing reports from the Florida Department of Health, covering approximately 98 weeks of data. The underlying rationale is straightforward: smokable cannabis delivers THC with substantially higher bioavailability and faster onset than oral preparations, and adding it to a program’s product menu could meaningfully alter the total THC reaching patients at the population level.

Under the primary assumption that smokable flower contained 20% THC (a figure borrowed from other state programs, not measured in Florida products), the study found a statistically significant immediate level increase of 138.45 mg per patient per week (95% CI: 102.69 to 174.20), corresponding to a 42% jump. A sustained upward trend of 5.62 mg per patient per week (95% CI: 4.35 to 6.89) continued across the 35-week observation period. Under a conservative 10% THC assumption, the level increase was a more modest 35.10 mg (95% CI: 5.31 to 64.88), approximately an 11% change. Critically, the study cannot determine whether this increased dispensing represents actual consumption, substitution from illicit sources, or redistribution. No health outcome data were assessed. The authors acknowledge that a comparison state was not available and that longer follow-up with individual patient records would be needed to draw clinical safety conclusions.

Dispensed Is Not Consumed: Unpacking a Major Study on Smokable Cannabis and THC

When Florida added smokable cannabis to its already massive medical program in 2019, it did not just expand a product menu. It triggered an immediate, measurable, and sustained shift in the entire program’s THC dispensing profile that no one was actively tracking in real time. A new study tries to quantify that shift, and the numbers are striking: a 42% jump in dispensed THC per certified patient under the researchers’ primary assumption, with continued weekly increases across the entire observation window. The study genuinely contributes something important to the literature. It applies an appropriate quasi-experimental design to complete census-level administrative data from the largest medical cannabis program in the country, and it demonstrates that form-specific policy changes are not pharmacologically neutral. Adding smokable flower to a program is not like adding another brand of capsule to a formulary. It introduces a route of administration with fundamentally different bioavailability, and this study catches the population-level consequences of that difference in real time. That is a contribution worth taking seriously, and the authors deserve credit for testing the robustness of their findings under two different THC concentration assumptions and for their unusually thorough acknowledgment of limitations.

But the gap between what was dispensed and what any of it means for patient health turns out to be the most important story. The central methodological problem is this: the THC concentration in smokable flower, the variable that drives the headline finding, was not measured from Florida products. It was estimated from data gathered in nine other states. Imagine trying to estimate how much alcohol a city consumed based on liquor store sales receipts, but having to guess the alcohol content of every bottle because the labels were never recorded. Your total would be directionally useful, but the specific number would depend entirely on your guess. Here, the difference between the 20% and 10% assumptions produces level increases that vary nearly fourfold, from 138 mg to 35 mg. That is not a trivial sensitivity range. And even the dispensed figure itself is a step removed from what matters clinically. A prescription for 30 pills tells you what was dispensed; it does not tell you whether the patient took them, shared them, or left them in a drawer. The same logic applies here, and the comparison to oral dosing guidelines, while arithmetically interesting, layers an additional extrapolation (oral-to-inhaled bioavailability) onto an already estimated quantity.

What I would tell a patient is this: the data suggest your program is dispensing substantially more THC than oral guidelines recommend, but we do not know how much you personally are using, whether it is helping, or whether it is causing harm. The dosing conversation is genuinely important and currently underspecified for inhaled routes. What I would tell a colleague is to treat the precise milligram figures with appropriate skepticism, but to recognize the broader signal: we lack validated inhaled dosing guidelines, and that gap matters in clinical practice. And what I would tell a policymaker is that the decision to add smokable forms immediately and significantly shifted population-level dispensing in a direction that existing safety benchmarks cannot comfortably accommodate, not because the evidence says it is harmful, but because we simply do not have the outcome data to know. The policy creates an obligation for prospective safety surveillance that does not currently appear to exist. Dispensing data are a useful early-warning system for policy-driven changes in population exposure, but they are a starting point for safety evaluation, not an endpoint. The gap between “dispensed” and “clinically significant exposure” must be explicitly bridged before regulatory or clinical conclusions can follow.

This study sits at the intersection of policy epidemiology and clinical pharmacology, an early-stage signal detection effort rather than a clinical outcomes trial. For clinicians working in medical cannabis, it provides quantitative context for a phenomenon most practitioners have observed intuitively: when patients gain access to smokable forms, their total THC exposure profile changes substantially. It does not, however, provide the individual-level or health-linked data needed to translate that observation into specific clinical guidance. The research arc for this question requires, at minimum, difference-in-differences studies with comparison states and linkage to emergency department, cardiovascular, and mental health outcome registries.

From a pharmacological perspective, the rapid onset and high peak plasma concentrations associated with inhaled THC create a distinctly different risk-benefit profile than oral or sublingual routes. Clinicians certifying patients should be aware that program-level dispensing volumes may far exceed what any existing dosing framework envisions for oral preparations, and that no validated dosing ceiling exists for inhaled cannabis in clinical use. Drug interaction considerations remain underexplored, particularly for patients on CNS depressants, anticoagulants metabolized via CYP3A4, or medications with narrow therapeutic windows. The most actionable recommendation from this evidence is to initiate an explicit dosing and consumption tracking conversation with every patient who accesses smokable products, documenting actual use patterns rather than relying on dispensing records as a proxy.

This is a quasi-experimental interrupted time series analysis of aggregate administrative dispensing data, published as a short report. In the evidence hierarchy, ITS designs occupy a middle tier: stronger than simple before-and-after comparisons, but weaker than controlled quasi-experiments or randomized trials. The single most important inference constraint is the absence of a concurrent comparison group, which means that secular trends in Florida’s rapidly growing program, such as changing patient demographics or market maturation, cannot be separated from the specific policy effect with certainty.

This study extends the work of Boehnke and colleagues (2022), who documented rapidly rising registration across U.S. medical cannabis programs from 2016 to 2020, by adding a dispensing-volume dimension to the enrollment trends. It confirms what program-level data from other states have suggested informally: expanding the range of available product forms, particularly to include inhalable options, is associated with meaningful shifts in aggregate utilization. However, unlike studies that have linked dispensary access to health outcomes at the individual level, this analysis stops at the dispensing measure and does not connect to clinical endpoints. Lorenzo and colleagues (2023) examined characteristics of Florida medical cannabis patients and found gender-specific patterns of use and health-related problems, providing contextual evidence that the patient population accessing these products is heterogeneous in ways that aggregate dispensing data cannot capture. The current study’s findings are broadly consistent with the direction of prior literature, but the absence of a control state and the reliance on estimated THC concentrations limit direct comparability with policy evaluations in other jurisdictions.

The most consequential analytic choice was the assumption of 20% THC concentration for smokable flower. The sensitivity analysis at 10% demonstrated that while the directional finding held, the magnitude of the level change dropped nearly fourfold (from 138 mg to 35 mg), and the confidence interval for the 10% scenario barely excluded zero (lower bound 5.31 mg). Had the authors been able to use Florida-specific product testing data, the true estimate might fall anywhere within or even outside this range. Additionally, a difference-in-differences approach using a comparable state (for example, one that expanded its medical program over the same period but did not add smokable forms) would have provided a far more credible counterfactual. Without it, the observed increase could partially reflect natural program growth, changing patient mix, or provider behavior shifts unrelated to smokable legalization.

The most likely overinterpretation is that this study demonstrates smokable cannabis is unsafe or that Florida patients are consuming dangerously high levels of THC. The 492 mg per week figure is an estimated average of dispensed product per certified patient, not a confirmed consumption measure. It includes patients who may not have purchased any cannabis at all during a given week but remained in the denominator as certified registrants, likely underestimating per-active-user amounts while simultaneously saying nothing about what was actually ingested. The 42% figure, while attention-grabbing, applies only under the 20% THC assumption; the 10% scenario yields roughly 11%. Treating either number as definitive misrepresents the underlying measurement uncertainty. Similarly, characterizing this as a causal study overstates what a single-group ITS without a control comparison can establish. It identifies a temporal association, not a proven policy effect.

This study documents a rapid and statistically significant increase in dispensed THC per certified patient following Florida’s legalization of smokable medical cannabis. It does not establish a causal link to the policy change with certainty, does not measure actual consumption, and provides no health outcome data. Its contribution lies in demonstrating that form-specific policy decisions produce measurable, population-level shifts in dispensing, underscoring the need for prospective pharmacovigilance infrastructure within large medical cannabis programs.

Does this study prove that smokable cannabis is dangerous?

No. This study measured changes in how much THC was dispensed by pharmacies, not what patients actually consumed or whether anyone experienced harm. No health outcomes were assessed. The increase in dispensed THC raises questions that future research with clinical data will need to answer, but this study alone does not support claims of danger or safety.

Are Florida patients really using 492 mg of THC every week?

That figure is an average of total dispensed THC divided by all certified patients, including those who may not have made a purchase in a given week. It also depends on an assumed 20% THC concentration in flower that was not directly measured in Florida products. Actual individual consumption could be significantly higher for active users or significantly lower for many registrants.

Should I be concerned about my medical cannabis dosing based on this study?

This study is a useful reminder that the dosing conversation for inhaled cannabis remains underspecified in clinical practice. If you use smokable medical cannabis, it is reasonable to discuss your actual consumption patterns, frequency, and any side effects with your certifying physician. Population-level dispensing averages do not translate directly to your individual situation.

Does this mean Florida’s decision to legalize smokable cannabis was a mistake?

The study does not make that judgment, and the data do not support it. What it suggests is that adding smokable forms to a medical program is a pharmacologically significant policy decision that changes dispensing patterns in measurable ways, and that programs of this scale would benefit from prospective monitoring systems that can track utilization alongside health outcomes.

Have thoughts on this? Share it: