Starting Medical Cannabis After Age 50: What a Driving-Simulator Study Can Actually Establish
A small study of adults starting medical cannabis did not demonstrate a different change in simulated driving performance from matched controls over one month. It also did not establish that driving after cannabis is safe. The distinction matters when a study is offered to support an opinion about a particular driver’s impairment: a treatment-initiation study and an acute intoxication experiment answer different questions.
Ennis and colleagues followed adults aged 50 and older with chronic nonmalignant pain in a prospective, nonrandomized feasibility study. Their original research, published in JMIR Formative Research in 2026, is available at doi:10.2196/79735.
What was actually compared?
The investigators enrolled 23 participants in the medical marijuana group and 21 controls at baseline. Assessments occurred before treatment initiation and approximately one month afterward. Nineteen participants in each group attended the follow-up assessment, but attendance did not mean everyone completed every simulator outcome.
This was human observational research. Treatment was not randomly assigned, and the investigators did not administer a standardized THC dose immediately before each drive. Participants used different products and routes under real-world conditions. Consequently, the study evaluates the feasibility of following treatment initiation and explores associated performance changes; it does not isolate the causal effect of a particular dose at a specified time after consumption.
Why the apparently faster reaction time needs context
At follow-up, one divided-attention scenario produced a statistically significant between-group difference: the medical marijuana participants with available data responded faster than controls. For that scenario, there were 14 participants with data in the treatment group and seven in the control group; reported mean reaction times were 2.57 and 5.79 seconds, respectively, with P=.02.
That comparison alone does not establish a treatment benefit. A difference between groups at follow-up is distinct from evidence that the groups changed differently from baseline. In the sex-adjusted longitudinal models, none of the six driving outcomes had a statistically significant group-by-time interaction. For the same divided-attention scenario, the interaction estimate was −2.47 seconds, with a 95% confidence interval from −6.76 to 1.82 and P=.25. The interval includes no difference and effects in either direction.
The appropriate description preserves both findings: one exploratory follow-up comparison was significant, while the models did not demonstrate different changes over time. Selecting only the favorable comparison would omit the central longitudinal result.
Missing simulator data are a substantive limitation
Simulator sickness was a major reason for incomplete driving data. The authors found that participants with missing follow-up driving data were more likely to be female. Adjusting for sex addresses one measured difference, but does not prove that the remaining participants represent everyone who enrolled.
Small outcome-specific samples, numerous comparisons, and missing measurements make both positive and null findings less definitive. Failure to detect a difference is not an equivalence test, and this feasibility study was not a certification that each participant could drive safely.
Urine screening did not measure driving-time blood THC
At follow-up, 17 of the 19 medical marijuana participants screened positive for marijuana in urine. These results helped verify exposure. They were not quantitative whole-blood THC concentrations and did not establish the interval between use and the simulator assessment. A positive urine screen therefore cannot fill the study’s gaps concerning acute dose, timing, or contemporaneous impairment.
For related interpretation, see our discussion of the separate questions of cannabis exposure, impairment, and crash risk.
How to evaluate an expert’s reliance on this paper
Ask whether the opinion concerns starting a treatment regimen, an acute episode after consumption, or responsibility for an actual collision. Then ask which measured outcome supports that opinion, how missing observations were handled, and whether the expert is relying on the follow-up comparison or the longitudinal interaction.
The study supplies preliminary evidence about a specific older chronic-pain population in a simulator. It supplies neither a universal waiting period after cannabis nor an individualized determination of impairment. Applying it to a case requires the person’s product, route, consumption and driving times, relevant clinical history, and independently documented behavior.
Primary research
Ennis N, Hou Y, Kloss K, Rogers J, Classen S. Medical Marijuana Initiation and Simulated Driving Performance Among Mid-to-Late-Life Adults With Chronic Pain: Prospective Observational Feasibility Cohort Study With Matched Controls. JMIR Form Res. 2026;10:e79735. https://doi.org/10.2196/79735.
