Clinical Insight | CED Clinic

A double-blind, double-dummy crossover RCT from Johns Hopkins, published in JAMA Network Open on May 1, 2026, finds that combining cannabis edibles with alcohol impairs driving more than either substance alone. At 25 mg THC with a 0.05% BAC, impairment exceeded what driving at the legal limit of 0.08% BAC would produce from alcohol alone – a finding with direct implications for every patient who uses cannabis and drinks, drives, or lives in a legal-use state.

Co-use of cannabis and alcohol is common, increasing, and under-studied. Most patients I see don’t think of cannabis and drinking as “stacking” substances the way they would think about mixing two medications. This study demonstrates clearly that they are additive in their impairment of driving — and that the interaction can exceed what the legal alcohol limit alone would produce. That’s a clinical conversation that belongs in every cannabis certification visit.

Zamarripa et al. conducted a within-participant, double-blind, double-dummy, 7-session crossover trial at Johns Hopkins University School of Medicine, enrolling 25 healthy adults who reported recent binge drinking and prior cannabis/alcohol co-use. Participants consumed brownies containing 0, 10, or 25 mg delta-9-THC, paired with placebo or alcohol drinks targeting 0%, 0.05%, or 0.08% breath alcohol concentration (BrAC). The primary outcome was a composite global drive score (GDS) from a standardized driving simulator, paired with standardized field sobriety testing and the DRUID cognitive-psychomotor battery.

All active drug conditions except the 10 mg THC alone condition produced measurable driving impairment relative to placebo. The combination of 0.05% BrAC with 25 mg THC yielded impairment comparable to 0.08% BrAC with 10 mg THC – meaning modest alcohol plus a higher edible dose matched the impairment seen at the legal limit. The 0.08% BrAC plus 25 mg THC combination was significantly worse than 0.08% BrAC alone. Field sobriety tests worsened at the 0.08% BrAC condition but missed impairment in several other conditions where the driving simulator detected meaningful decrements. THC pharmacokinetics were not significantly altered by alcohol co-administration.

Why the Edibles Conversation in the Exam Room Just Changed

One of the questions I get from patients more than almost any other is some version of: “I only had a couple of drinks and I took an edible earlier – am I okay to drive?” I’ve never had a clean, controlled, peer-reviewed answer to that question. This study is the closest we’ve come.

The design is worth understanding because it’s unusually rigorous. Double-blind, double-dummy means participants didn’t know whether their brownie contained THC or whether their drink contained alcohol. Seven sessions, each separated by a week, crossing every combination of two THC doses and two alcohol levels against respective placebos. Within-participant means every person served as their own control. For a study of a behavior as complex as impaired driving, this design is close to the gold standard.

The finding I keep coming back to is the 0.05% BrAC plus 25 mg THC combination. That’s below the legal limit for alcohol – in most U.S. states you would pass a breathalyzer. But when paired with a 25 mg edible, the composite driving impairment in this trial matched what the 0.08% BrAC plus 10 mg THC condition produced. In other words, being under the legal alcohol limit did not protect driving performance when a meaningful edible dose was on board. That matters for patients. A lot.

The pharmacokinetic piece is also notable. Alcohol did not change THC or metabolite blood concentrations – so the interaction isn’t about altered THC absorption. It’s a pharmacodynamic effect: two substances acting on partially overlapping neural pathways produce additive impairment even when neither is changing the other’s blood levels. The brain, in other words, doesn’t care what the breathalyzer says.

There’s also a sober warning buried in the standardized field sobriety test data. These tests – the walk-and-turn, one-leg stand, and related assessments – are what law enforcement uses roadside to determine impairment. In this study, sobriety tests flagged impairment at the 0.08% BrAC condition. They missed it in several co-use conditions where the driving simulator found meaningful decrements. That’s a detection gap. Current tools aren’t calibrated for cannabis-alcohol co-use, and this study demonstrates what that gap looks like in practice.

A few clinical caveats worth naming. The sample was 25 participants, all of whom reported prior co-use – not naive users. Driving was simulated, not on-road. The THC doses were controlled and known, which is very different from the uncontrolled potency variability in commercial edibles, where a “25 mg” product may deliver substantially more or less depending on batch and formulation. And the participants were young adults with a mean age of 25. I see patients ranging from 20 to 90, and I’d want to see this data extended to older adults, to patients with chronic conditions, and to populations using edibles therapeutically rather than recreationally.

None of those caveats change what I tell patients. If you use cannabis in any form – edibles especially, given their delayed and prolonged onset – and you drink at any level, the combination impairs your driving more than either substance alone. The impairment is not reliably detectable by current roadside tools. Blood THC levels don’t tell you whether you’re impaired, and blood alcohol at a legal level doesn’t tell you you’re safe if cannabis is on board. There’s no clean threshold the way there is for alcohol alone.

What I want from the research that follows this trial: dose-response modeling across a wider THC range, data on inhalation versus edibles (the delayed absorption from edibles is clinically distinct from smoked or vaped cannabis), and performance data in patient populations rather than healthy young adults. The public health and policy implications here are significant. We are in a moment of rapid legalization without a correspondingly mature detection and safety infrastructure. This study moves the scientific baseline. The clinical and legal systems now have to decide what to do with it.

Cannabis and driving research has been constrained for decades by the same Schedule I barriers that limited all cannabis science. The literature that exists is largely observational, with significant variability in how “impairment” is measured and how cannabis use is verified. Prior controlled trials have mostly focused on smoked or inhaled cannabis, where the pharmacokinetic profile, rapid onset, shorter duration, differs substantially from edibles. Edibles produce delayed peak THC concentrations (often 1 to 4 hours post-ingestion) and prolonged psychoactive effects, making the driving risk window harder to predict intuitively. This study is among the first RCTs to isolate the edible formulation specifically and combine it with a dose-controlled alcohol challenge in a within-participant design. A CED Clinic-relevant paper from 2025 using a mobile app-based driving behavior study found cannabis use associated with meaningful driving changes but lacked the controlled alcohol co-use component this Johns Hopkins trial adds.

The clinical action point here is straightforward even if the underlying pharmacology is not: counsel patients that edibles and alcohol do not cancel each other out and do not operate on independent impairment tracks. Patients should not drive after combining these substances, particularly within the 4-hour window following edible ingestion when THC blood levels may still be rising. For patients in Massachusetts, where both alcohol and adult-use cannabis are legally available, the interaction risk is a routine part of the responsible use conversation. Document that counseling.

This is a within-participant randomized crossover trial, the strongest design available for within-person drug effect comparisons. Double-blinding and a double-dummy control make demand bias unlikely. The primary limitation is sample size: 25 participants is adequate to detect large effects but insufficient to characterize subgroup variation by sex, age, or cannabis use history. Driving simulation, while validated and widely used in this literature, is not fully equivalent to real-road performance.

Prior observational work has consistently linked cannabis use to increased crash risk, with odds ratios in the range of 1.2 to 1.7 depending on study methodology. The controlled human performance literature is smaller and more mixed, partly because prior studies used smoked cannabis with faster-onset, shorter-duration kinetics that don’t map cleanly to edible use patterns. The Vandrey group at Hopkins has contributed several earlier controlled studies on cannabis and cognition; this study extends that work specifically to the oral route and the alcohol interaction.

The field sobriety test detection gap confirmed here is consistent with prior work suggesting that THC blood levels and standard sobriety tests are poor predictors of driving impairment compared to performance-based measures. This creates a meaningful legal and clinical disconnect: a driver could pass current roadside detection while being meaningfully impaired on simulator-validated measures.

The global drive score is a composite, and individual driving metrics within it might show different effect sizes. Separating lane-keeping from braking latency from speed variability could produce a more granular picture of how exactly cannabis and alcohol interact rather than just confirming that they do. A longer post-dosing window, particularly for the edible conditions, could also reveal whether impairment peaks and resolves differently than the session timing captured.

Because all 25 participants had prior co-use experience, the results may underestimate impairment in naive or less frequent users. The authors appropriately note this as a limitation. Whether the same effects scale proportionally in heavier cannabis users, who may show some tolerance, is unknown from this dataset.

The most likely misread is that 10 mg THC “didn’t impair driving.” That’s not what the data show. The 10 mg THC alone condition did not reach statistical significance on the composite GDS relative to placebo, but it did produce measurable effects on some individual measures, and the study was not powered to detect small effects. Absence of a statistically significant composite finding at that dose is not the same as confirmed safety.

A second misread: because alcohol and THC didn’t alter each other’s pharmacokinetics, some commentators may conclude the interaction is purely additive and therefore predictable. The additive nature of the behavioral effect is the finding,  it does not mean the interaction is benign or easily managed. Two predictable impairments adding together to exceed a legal threshold is the clinical problem, not a reassurance.

This well-designed Johns Hopkins RCT establishes that cannabis edibles and alcohol produce additive driving impairment exceeding the effect of either substance alone, that the interaction can surpass impairment seen at the legal U.S. BAC limit, and that standard field sobriety tests miss impairment in several co-use conditions. For patients who use cannabis and drink, the message is clear: do not drive after combining them. For clinicians, this paper belongs in your cannabis counseling protocol.

I only had a couple of drinks after my edible. Am I okay to drive?

This study suggests not. Two substances that both impair driving combine to produce greater impairment than either alone, and even alcohol below the legal limit (0.05% BrAC) paired with a 25 mg edible produced meaningful driving decrements in this trial. The safest approach is not to drive after using any combination of cannabis and alcohol, particularly within the hours following edible use when THC levels may still be rising.

Does this apply to vaping or smoking, or just edibles?

This trial studied edibles specifically. Smoked and inhaled cannabis have faster onset and shorter duration than edibles, so the impairment timing is different. The interaction with alcohol is likely present for all routes of cannabis use, but the delayed and prolonged effect of edibles makes the risk window longer and harder to predict intuitively. This is one reason edibles carry particular caution in the driving context.

Can I pass a breathalyzer even if I’m impaired by edibles and alcohol?

In some co-use conditions in this study, driving simulation detected meaningful impairment while field sobriety tests did not flag it. A breathalyzer measures alcohol, not cannabis or their combined behavioral effect. Passing a breathalyzer at 0.04% BrAC tells you nothing about your driving performance when cannabis is also on board.

I take edibles medically at night. Should I avoid alcohol entirely?

That’s a conversation worth having with your CED Clinic physician based on your dose, timing, and specific medical situation. As a general principle, the combination produces more impairment than cannabis alone, and driving should not be undertaken in the hours following any edible use. If you drink socially the same evening you’ve taken a therapeutic edible, plan not to drive.

Is 25 mg of THC a large edible dose?

It is a moderate-to-high dose for a non-daily consumer, but it is commonly available in recreational markets and is not unusual for medical patients who have built tolerance. Importantly, commercial edibles vary widely in actual THC delivery relative to their label claims. A product labeled 25 mg may deliver more or less depending on formulation and batch — another reason that the impairment window from edibles is difficult to predict without clinical guidance.

1. Zamarripa CA, Lin S, Klausner M, et al. Impact of Cannabis Edibles Combined With Alcohol on Driving, Field Sobriety Performance, and Subjective Effects: A Within-Participant Crossover Trial. JAMA Network Open. 2026;9(5):e269842. doi:10.1001/jamanetworkopen.2026.9842 PMID: 42065887

Have thoughts on this? Share it: