A percentage is not a dose

Modern cannabis really is stronger.

A major 2025 systematic review in Annals of Internal Medicine examined 99 studies involving more than 220,000 participants and found predominantly unfavorable associations between high-concentration THC products and psychosis or schizophrenia and cannabis use disorder. That should concern anyone interested in responsible cannabis policy.

It is also worth noticing what the review could not tell us. More than 95% of the studies were judged to have moderate or high risk of bias. A separate 2025 systematic review in the American Journal of Psychiatry reached a similar general conclusion about concern surrounding higher-potency products, while rating the overall certainty of the evidence very low.

That is not evidence that potency is harmless. Quite the opposite. It is a reason to take potency seriously without pretending we already understand exactly how risk changes at every concentration.

A bottle of vodka is more concentrated than a bottle of beer. That matters. But nobody would estimate a person’s alcohol exposure from the percentage on the bottle without asking how much they drank.

Cannabis somehow invites exactly that mistake.

A 70% THC vape contains a more concentrated product than 20% flower. Actual exposure still depends on how much is consumed, how it is delivered, how often it is used, and a variety of behavioral and pharmacokinetic factors. This problem is significant enough that researchers have proposed standardized THC units, analogous in concept to alcohol units, specifically because product percentage does not provide a common measure of dose across flower, concentrates, vapes, and edibles. One influential proposal defines a standard unit as 5 mg THC.

Those numbers happen to look similar. They answer completely different questions.

This is why I have trouble with the proposal that cannabis flower should be returned to 5% THC as a sensible “starting point.” Regulating concentration may prove useful. I can find no high-quality evidence establishing 5% THC flower as a biological safety threshold for adolescents or adults. The recent systematic reviews do not identify one.

A 5% cap can be proposed as a precautionary policy.

It should not be mistaken for a scientifically discovered line between safe and unsafe cannabis.

Daily use is a warning sign, not a diagnosis

Frequency matters too, perhaps even more clearly.

A systematic review and meta-analysis of prospective studies found a strong dose-response relationship between cannabis-use frequency and subsequent cannabis use disorder. Risk increased progressively from infrequent use through monthly, weekly, and daily use. Daily use deserves clinical attention.

But there is an important difference between saying daily use increases the risk of addiction and saying daily use demonstrates addiction.

Cannabis use disorder is a disorder of behavior and consequences, not simply a tally of days on which cannabis was consumed.

Those questions tell me much more than a calendar.

Tolerance and withdrawal complicate this further. Both can occur with cannabis. Both matter clinically. Neither, by itself, proves that a person has lost control of the behavior.

Medicine already understands this distinction in other settings. Physiological adaptation to a repeatedly administered drug and addiction to that drug are not automatically the same event.

Cannabis medicine is beginning to confront the same diagnostic problem directly. A 2026 study examined 2,054 Australian adults prescribed medical cannabis. Among the 986 people who reported using cannabis only as prescribed, applying all 11 conventional DSM-5 cannabis-use-disorder criteria classified 20% as having CUD. When tolerance and withdrawal were excluded under the prescription-specific framework, the estimate fell to 9%. Moderate-to-severe CUD fell from 5% to 2%.

This matters when we read that people who use cannabis daily may experience withdrawal and therefore return to cannabis.

Perhaps they do. But returning does not tell us why they returned.

Their original symptom may have returned. Withdrawal may be unpleasant. They may enjoy cannabis. They may judge, wisely or unwisely, that the benefits remain worth the costs. Or they may have developed a cannabis use disorder and find themselves unable to govern the behavior despite wanting to.

Those stories can look remarkably similar if the only thing we measure is repeated use.

Clinically, they are not similar at all.

The teenager may have been telling the truth

This brings me back to the patient described in the original essay.

He first tried cannabis at eight. His mother died. His father was incarcerated. By twelve, he was using cannabis daily. By sixteen, he was struggling with depression, anger, school failure, suspension, and suicidal thoughts.

Asked why he used cannabis, he said:

The essay responds that his life told a different story.

I am not sure those stories contradict each other.

A behavior can solve one problem while quietly creating several others. Clinically, understanding the problem it solves is often essential to changing the behavior.

This is especially relevant because adolescent cannabis use and psychiatric vulnerability do not appear to travel in only one direction.

A longitudinal analysis from the large Adolescent Brain Cognitive Development study followed nearly 12,000 young people. Adolescents who eventually initiated cannabis already had more psychosis-spectrum symptoms beforehand, and those symptoms and the distress associated with them increased in the period leading up to cannabis initiation. The investigators found evidence consistent with shared vulnerability and self-medication, alongside mixed evidence for additional increases after cannabis initiation.

That does not absolve cannabis.

Another enormous longitudinal cohort, involving more than 463,000 adolescents, found that past-year cannabis use was associated with later diagnoses of psychotic, bipolar, depressive, and anxiety disorders, with particularly strong associations for psychotic and bipolar disorders. Those findings provide plenty of reason to discourage adolescent use. They still do not establish potency as the single causal mechanism connecting cannabis with those outcomes.

Several things can be true at once.

A vulnerable teenager may discover that cannabis temporarily reduces distress. Frequent cannabis use may subsequently make portions of his psychiatric or functional picture worse. Underlying vulnerability may increase both his likelihood of using cannabis and his likelihood of later psychiatric illness.

Once we recognize that, the clinical question changes.

A lower THC percentage does not answer those questions.

Detectable THC is not the same as ongoing impairment

There is another appealingly simple biological story in the argument about modern cannabis.

THC is lipophilic and can remain detectable in frequent users long after acute intoxication has passed. From there, the claim becomes that adolescents using potent products carry higher THC levels in their bodies, creating prolonged exposure that suggests cognitive problems “will only worsen.”

The first part is pharmacokinetics.

The last part outruns it.

A systematic review found that frequent cannabis users can retain measurable blood THC for days after stopping. Some participants remained above commonly used per se driving thresholds after prolonged abstinence. The investigators’ practical conclusion was that residual blood THC does not necessarily establish recent cannabis use.

That distinction does not make the adolescent cognitive literature reassuring.

A 2024 JAMA Pediatrics systematic review and meta-analysis involving more than 438,000 young people found cannabis use associated with lower grades, more absenteeism and dropout, and lower rates of high-school completion and postsecondary attainment. Earlier and more frequent use tended to be associated with worse outcomes.

But cognition itself is more nuanced. A JAMA Psychiatry meta-analysis of 69 studies found a small overall association between frequent or heavy cannabis use and poorer cognitive performance in adolescents and young adults. Among studies requiring more than 72 hours of abstinence, the estimated difference became much smaller and was no longer statistically significant. The authors cautioned that previous research may have overstated the magnitude and persistence of cognitive deficits.

That study does not prove there are no lasting developmental effects. It was not designed to settle that much larger question.

It does show why tissue persistence should not be casually converted into evidence of continuous cognitive impairment.

The cognitive concerns are serious enough without asking residual THC to prove something it has not proved.

What the alcohol analogy actually teaches us

The original essay uses alcohol as a model for cannabis regulation.

I think that is a useful comparison.

We just may be learning different lessons from it.

Alcohol comes in beer, wine, spirits, and products of substantially higher concentration. We do not require whiskey to contain the alcohol concentration of beer.

Instead, concentration becomes one component of a larger regulatory system.

We regulate age of purchase, serving quantities, packaging, labeling, retail practices, taxation, intoxicated driving, and commercial behavior. Adults retain the ability to choose among products carrying very different concentrations and risk profiles.

Cannabis policy could become considerably more sophisticated along the same lines.

Concentrates may deserve rules that flower does not. A 100 mg edible package is a different exposure problem from a 5 mg serving. A product engineered to permit repeated high-dose inhalation deserves consideration different from something far more self-limiting.

A percentage still matters.

It simply does not have to carry the entire regulatory system on its back.

I am not arguing that a fourteen-year-old should be expected to thoughtfully self-regulate access to high-potency cannabis. Developmental vulnerability is precisely why strong youth protections, meaningful age enforcement, prevention efforts, and restrictions on products and marketing that appeal to children are justified.

The autonomy question enters when an argument about protecting teenagers becomes an argument that adults should also be limited to 5% THC.

Adults make risk-bearing choices constantly. They drink alcohol. Ride motorcycles. Ski. Participate in contact sports. Decline medications. Accept the risks of medications they prefer. Eat things their doctors wish they would eat less often.

Public health certainly regulates some of these choices. Medicine certainly advises against some of them.

But we do not ordinarily presume that the existence of risk means an adult has lost the capacity to choose.

That distinction becomes particularly important when addiction enters the argument.

If the possibility that some people will lose control becomes enough to treat everyone as though they cannot exercise control, we have quietly changed subjects.

We are no longer talking only about addiction.

We are talking about how much agency adults are permitted to retain when their choices involve risk.

That is not an argument for a laissez-faire cannabis market. I do not think the existing market is particularly well designed around health.

It is an argument for putting regulation where the risk actually changes.

Protect children aggressively. Make high-dose products harder to use accidentally. Tell consumers what they are actually taking. Treat concentrates differently when the evidence justifies it. Warn people honestly about psychosis, cannabis use disorder, impairment, pregnancy, driving, and early frequent use. Build treatment systems for the adolescents and adults who do lose control.

And keep studying potency.