Pot Debate Alex Berenson DR Michael Hart on the Joe Rogan Experience #1246, fact-checked

What the evidence shows: NIMH's website did lower its headline US schizophrenia prevalence figure around 2017, but the characterization of a secretive, unexplained 1.1%-to-0.3% swap is misleading. Since at least 1993 NIMH reported a one-year prevalence of 1.5% among adults (1.2% among children) to Congress, based on the 1980s Epidemiologic Catchment Area (ECA) study; NIMH's current published statistics page instead cites a range of 0.25%-0.64%, drawn from the 2001-2003 National Comorbidity Survey Replication and Medical Expenditure Panel Survey data. NIMH's own page discloses that these newer household-survey and medical-claims sources exclude institutionalized, incarcerated, and homeless populations, and states that these excluded groups may have higher prevalence, meaning the lower figure reflects a narrower methodology and case definition rather than a real sevenfold drop in schizophrenia's occurrence. The revision appeared as an update to NIMH's own public statistics page rather than a concealed alteration, though NIMH did not pair it with a dedicated public explanation of the methodology change. Overall: a real and substantial downward revision occurred and was under-explained, but it was a public methodology change documented on NIMH's own site, not a secret alteration of a stable epidemiological fact.

What the evidence shows: CDC surveillance data confirm a real rise in suicide among adolescent females: the rate for females aged 15-19 rose from 2.4 to 5.1 per 100,000 between 2007 and 2015, more than doubling (about a 113% increase), so the "50%" figure understates the documented change even as it correctly signals an upward trend; the absolute rate, however, remained in the low single digits per 100,000 throughout the period. Current NIMH data (2023) show suicide rates rise sharply with age and are far higher among men than women at every age bracket: females aged 15-24 had a rate of about 5.5 per 100,000, versus 29.2 per 100,000 for men aged 45-64 and 40.7 per 100,000 for men 75 and older, the single highest rate of any group; white men (28.0 per 100,000) have higher rates than Black, Hispanic, or Asian men, though American Indian/Alaska Native men have the highest rate by race. This supports the general direction of Berenson's second claim, middle-aged and older men do have far higher absolute suicide rates than young women, but "middle-aged white men" is an oversimplification since the highest-rate subgroup is men 75 and older, not middle-aged men specifically, and multiple demographic groups have elevated rates. No source establishes a causal link between social media use and the increase in young-female suicide; researchers describe the relationship as correlational and unresolved.

What the evidence shows: The figure traces to a 2018 study of 240 early-phase psychosis patients treated in Switzerland (Lausanne), which found that among the 82 patients with a cannabis use disorder, 38 (46%, close to but not exactly the "50%" cited) displayed violent behavior over a 36-month follow-up, versus 15% (24/158) of patients without a cannabis use disorder. The study had no matched control group of healthy non-users, and its authors explicitly stated that "the small sample size excludes the generalization of these results" and that the finding is "mainly correlational," noting they could not rule out that inherently violent patients used cannabis to self-medicate rather than cannabis driving the violence. The cohort's mean age was about 24 and 67.5% were male, consistent with Berenson's "20-something men" framing, but that demographic reflects the study's convenience sample, not a population-level rate. Broader meta-analyses since have found a statistical association between cannabis use and violence (odds ratios around 2, and up to roughly 5.8 for heavier cannabis misuse), but researchers describe the evidence as confounded by factors like other substance use, impulsivity, and possible reverse causation, and stop short of establishing that cannabis causes violence. A public letter signed by roughly 100 scholars and clinicians accused Berenson, whose book "Tell Your Children" is built around this argument, of inferring causation from correlation and cherry-picking data. The underlying statistic is approximately accurate to its source study but is presented without the small-sample, no-control-group, and correlational caveats the original researchers themselves attached to it.

What the evidence shows: Critics did not dispute Berenson's underlying murder counts for Alaska, Colorado, Oregon, and Washington between 2013 and 2017, but found the causal inference unsupported. PolitiFact reported that Berenson's four-state comparison cherry-picks dates, since Colorado and Washington legalized in 2012 (not 2014, as Berenson has stated elsewhere), and it coincides with a national homicide uptick in 2015; it also relies on raw counts rather than population-adjusted rates. University of Oregon economist Benjamin Hansen built synthetic-control models comparing Colorado and Washington to statistically similar non-legal states and found actual homicide rates were at or below what pre-legalization (2000-2012) trends predicted, implying no demonstrated legalization effect. A 2023 systematic review of comparative longitudinal studies likewise found no significant difference in violent crime between Colorado/Washington and matched comparison states from 1999-2016, while noting that rigorous studies isolating crime effects (as opposed to possession-arrest data) remain scarce and results across the broader literature are inconsistent. Overall, the specific 40% figure is not a fabrication, but using it to demonstrate a legalization-crime causal link is misleading given the absence of population/rate adjustment and the contrary findings of controlled analyses.

What the evidence shows: The claim asserts that NSAIDs decrease gonadotropins in a way that raises testosterone by roughly 25%. The best-known controlled human trial on this topic, a randomized study of 31 healthy men published in PNAS, found the opposite direction of effect: 1,200 mg/day of ibuprofen for six weeks increased luteinizing hormone (LH, a gonadotropin) by 23% at 14 days and 33% at 44 days, while free testosterone showed no significant change, and the testosterone-to-LH ratio fell about 18-23%. The researchers described this pattern as "compensated hypogonadism," where the pituitary raises gonadotropin output to maintain near-normal testosterone despite reduced testicular responsiveness, not a state where gonadotropins fall and testosterone rises. A separate NHANES-based population study of NSAID users found no significant association between regular NSAID use and testosterone once confounders were controlled. No study reviewed reports a ~25% testosterone increase from NSAID-induced gonadotropin suppression. The specific figure and mechanism described do not match the peer-reviewed literature, which instead points to gonadotropins rising, not falling, with NSAID use.

What the evidence shows: The cited figures match a single retrospective study (Nguyen et al., The American Surgeon, 2014) of 446 trauma patients at one Level I trauma center, in which 82 THC-positive TBI patients had 2.4% in-hospital mortality versus 11.5% among THC-negative patients, with THC positivity reported as independently associated with survival after statistical adjustment. Subsequent, larger studies have not consistently replicated this finding: a 2020 retrospective cohort of 854 severe TBI patients across six Ohio trauma centers found that positive THC screening did not predict discharge mortality once age, Glasgow Coma Scale score, injury severity, and other confounders were controlled for, with the authors attributing the earlier result to methodological differences in how the 2014 study handled its statistical model. A 2024 study of 3,729 TBI patients found cannabis-positive patients had a numerically lower mortality rate (3.9% versus 4.8%), but the difference was not statistically significant (p=0.3), and the authors explicitly said they could not replicate the original study's significant survival advantage. A 2020 systematic review concluded that evidence on marijuana exposure and TBI outcomes remains inconclusive, noting only one study had found the mortality connection at all, and calling for larger, better-designed research. The specific 2.4%/11.5% statistic is real but comes from one small, single-center retrospective study whose effect has not been consistently reproduced in later, better-controlled work, and it does not establish a proven protective mechanism of THC on brain injury outcomes.