Dr. Mary Talley Bowden on the Joe Rogan Experience #2335, fact-checked

What the evidence shows: HHS did launch the COVID-19 Community Corps on April 1, 2021, and Houston Methodist was indeed the first US hospital system to require COVID-19 vaccination for its roughly 26,000 employees, announcing the policy on March 31 to April 1, 2021, so the timing and the first-in-the-nation points are roughly accurate. The Community Corps itself, however, was documented as a grassroots network of trusted community voices meant to encourage vaccination, not a fund to censor people, and CDC records tie the surrounding effort to a $3 billion vaccine-access award, not an $11.5 billion figure. A peer-reviewed benefit-cost analysis put the total cost of the related We Can Do This public education campaign plus associated vaccination costs at $8.3 billion for April 2021 to March 2022, and no authoritative source supports an $11.5 billion figure or the characterization of the program as a censorship slush fund. Overall the claim is mixed: the launch date and Houston Methodist first-mover facts hold up, but the dollar figure and the propaganda-and-censorship framing are unsupported.

What the evidence shows: Published toxicology data put ivermectin's acute oral LD50 in rodents in roughly the range Bowden cites, with reported values that vary widely by species and sex: pharmacology reviews list about 25 mg/kg in mice and about 50 mg/kg in adult rats, and WHO and other datasets span from roughly 11.6 mg/kg to 40 mg/kg or higher in mice, so her 10 to 80 mg/kg framing is broadly consistent with the literature. Her stated human COVID dose of 0.4 mg/kg is also accurate as a description of what protocols used: reviews describe 0.2 to 0.4 mg/kg as the usual antiparasitic dose considered extremely safe in humans. The numbers she gives, and her underlying point that lethal animal doses far exceed the 0.4 mg/kg human dose, are correct. Context worth noting: this favorable safety margin does not mean ivermectin works for COVID, as large randomized trials found no consistent clinical benefit. On the specific figures cited, the claim is largely accurate.

What the evidence shows: Bowden offers an uncontrolled personal case series with no comparison group, so a reported 100 percent out-of-hospital rate among self-selected early-treatment patients cannot establish that any treatment worked. The largest randomized, placebo-controlled trials of the ivermectin-based approach she is known for found no benefit: the TOGETHER trial (New England Journal of Medicine, 2022) concluded that early ivermectin did not result in a lower incidence of hospital admission due to progression of COVID-19. The NIH-funded ACTIV-6 trial, including a higher-dose arm (600 micrograms per kilogram for 6 days), likewise found ivermectin did not improve time to recovery and did not reduce hospitalizations or deaths, stating the findings do not support its use in outpatients with COVID-19. Anecdotal outcome counts from a single practice, absent a control group, are not evidence that the treatment prevented hospitalizations. The claim is unsupported by controlled evidence.

What the evidence shows: The 4.4 percent figure is Dr. Joseph Varon's own self-reported mortality for his Houston hospital in 2020, not an independently audited or peer-reviewed result, and it is presented here as evidence that his ivermectin-based protocol saved lives. Published US data do show much higher averages: a study of 104,590 patients across 21 US health systems found hospital mortality near 18.6 percent early in the pandemic, falling to 7.3 percent by mid-2021 (PMC9518859), and a meta-analysis put pooled ICU mortality around 35.5 percent (PMC8013495), so a roughly 20 percent hospital average is plausible for 2020. The comparison is still misleading because a single hospital's self-reported death rate cannot be set against national averages without adjusting for patient age, illness severity at admission, and pandemic timing, all of which strongly affect outcomes. The implied cause, ivermectin, is not supported: large randomized trials such as the NIH ACTIV-6 higher-dose study found ivermectin did not reduce recovery time, hospitalization, or death (PMC9941969). Comparing an unadjusted, self-reported single-center rate to national averages to credit an unproven treatment makes this claim misleading.

What the evidence shows: Bowden was one of three doctors (Apter v. HHS/FDA, alongside Robert Apter and Paul Marik) who sued the FDA in 2022 over its August 2021 'You are not a horse. You are not a cow' ivermectin posts. After a district court dismissed the case on sovereign immunity, the Fifth Circuit revived it in 2023, holding that the FDA 'is not a physician' and can inform but not endorse, denounce, or advise. In March 2024 the FDA settled by agreeing to delete the horse tweet and two similar posts from X, LinkedIn, and Facebook and to retire a related consumer advisory. So the core factual claim (she sued over the horse tweet and the FDA removed it) is accurate, and her own aside that 'it hasn't really changed anything' is fair, but the settlement was not a court finding: the FDA did not admit any wrongdoing, disputed that it exceeded its authority, and did not change its position that clinical data do not show ivermectin is effective against COVID-19. Overall assessment: largely accurate, with the caveat that the 'we won' framing rests on a settlement in which the FDA conceded no legal violation.

What the evidence shows: The three states Bowden names as successful match the legislative record: Idaho enacted Senate Bill 1211, signed April 14, 2025, allowing ivermectin to be sold without a prescription, and Arkansas enacted SB189 (Act 396) on March 25, 2025, authorizing ivermectin for human use to be sold without a prescription. Tennessee was the earlier adopter, having passed its law in 2022, so it predates the most recent legislative session rather than being a fresh success. Contemporary reporting counted roughly 15 states introducing such bills during the 2025 wave rather than 17, and the exact split of four still pending versus ten failed cannot be independently confirmed and appears to reflect a fluid, point-in-time tally. Note that these laws remove a prescription requirement but do not change the underlying medical evidence: the FDA and NIH have found ivermectin is not effective for and is not approved to prevent or treat COVID-19. The count of introduced bills and the pass or fail breakdown are approximate, but the core claim (three states enacting OTC ivermectin laws, named as Tennessee, Idaho, and Arkansas) is largely accurate.

What the evidence shows: The pharmacology of ivermectin supports the metabolism point. The FDA-approved STROMECTOL (ivermectin) label states the drug is metabolized in the liver and that ivermectin and its metabolites are excreted almost exclusively in the feces over about 12 days, with less than 1 percent of the dose appearing in the urine. A study of human liver microsomes concluded that cytochrome P4503A4 (CYP3A4), a hepatic enzyme, is the predominant isoform responsible for ivermectin metabolism. So the core statement that ivermectin is cleared by the liver rather than the kidney is correct, which is why renal impairment or a kidney transplant does not by itself change ivermectin's elimination. The broader clinical inference of no harm is more than the metabolism data alone establish, since transplant patients take CYP3A4-affecting immunosuppressants and drug interactions still warrant caution, but the specific pharmacological claim stated on air is largely accurate.

What the evidence shows: Bowden's figures track a 2017 JAMA cohort study (Downing et al.) of 222 novel therapeutics approved by the FDA between 2001 and 2010. That study found 71 of them (32.0 percent) had a postmarket safety event, defined as a market withdrawal, a new boxed warning, or an FDA safety communication. The median time from approval to the first such event was 4.2 years (interquartile range 2.5 to 6.0 years), and the study estimated that roughly 31 percent of therapeutics had experienced an event by about 10 years post-approval. Her roughly one-third figure, the four-year median, and the roughly 10-year framing all align with the published results, so the claim is roughly accurate.

What the evidence shows: Merck's arthritis painkiller Vioxx (rofecoxib) was on the market from May 1999 until its withdrawal in September 2004, and peer-reviewed sources report it brought in more than 2.5 billion dollars annually, roughly 10 percent of Merck's worldwide sales, at its peak. Over its roughly five years on sale that implies cumulative revenue in the range of about 10 to 12 billion dollars, so Rogan's 12 billion figure is at the high end but in the right ballpark. In November 2007 Merck agreed to pay a fixed 4.85 billion dollars to settle 26,600 lawsuits representing 47,000 plaintiffs over heart attack and stroke claims. That settlement was indeed a fraction of the drug's lifetime sales, supporting the core of the claim. Roughly accurate.

What the evidence shows: US life expectancy fell sharply in 2020 and 2021 during the COVID-19 pandemic, but it has been rising since, not falling. Federal mortality data reported by NPR show life expectancy climbed to about 77.5 years in 2022, the first increase in two years. It rose again in 2023 and reached a record high of roughly 79 years in 2024, the highest ever recorded in the United States, driven by falling deaths from COVID-19, drug overdoses, heart disease, and cancer, according to STAT reporting on CDC data. Framing the trend as life expectancy having 'gone down' describes the pandemic dip while ignoring the multi-year recovery that was already in the published record by the time of this conversation, making the statement misleading.

What the evidence shows: The 700-day figure traces to a Yale-led study of post-vaccination syndrome (PVS), released in February 2025 as a medRxiv preprint, which detected the S1 subunit of spike protein circulating in the blood of some PVS participants more than 700 days after their last dose (42 participants and 22 controls). That is a measurement of spike protein detected in a symptomatic subset, not evidence that the body continues to produce spike for 700 days: co-senior author Akiko Iwasaki explicitly said the vaccine mRNA is unlikely to be the source at such a late point, that researchers do not know what is driving the late-phase spike, and that spike was not causally linked to symptoms because other PVS participants had no measurable spike. The authors stressed the work was early-stage, unreplicated, and not yet peer reviewed. Bowden's framing that spike continues to be produced up to 700 days later overstates and mischaracterizes a preliminary detection finding in a small, symptomatic cohort, so the claim is misleading.

What the evidence shows: Federal data on the Countermeasures Injury Compensation Program (CICP) support the core of this claim. A review of the program (drawing on the December 2024 GAO report on HRSA's CICP) found that of the roughly one quarter of claims adjudicated as of mid 2024, only 92 (about 2.6 percent) were deemed eligible for compensation, meaning well over 97 percent of decided claims were denied. Of those 92 eligible claims, only 52 were COVID vaccine related and just about $400,000 had actually been paid out for COVID countermeasure injuries, consistent with only a few dozen people compensated at the time this June 2025 episode aired. Two important caveats: the 98 percent figure applies to claims that have been decided, not all claims filed (a large share remain pending), and the high denial rate reflects the program's strict causation and filing standards rather than proof that the underlying injuries were fabricated. On the numbers cited, the statement is largely accurate.

What the evidence shows: The relevant program is the federal Countermeasures Injury Compensation Program (CICP), which handles COVID-19 vaccine injury claims. Congressional Research Service analysis confirms that CICP pays only reasonable medical expenses, lost employment income, and a death benefit, and does not cover pain and suffering, which keeps individual awards small. That report also documents an overwhelming denial rate: as of the April 2025 update, HRSA had decided 4,111 of 13,764 COVID-19 countermeasure claims and denied 4,044 of them (98.4%), finding only 67 eligible for compensation (1.6%). Consistent with those limits, published tallies of the small number of paid COVID-19 claims show most awards clustered in the low four figures (roughly $1,000 to $5,500 for common injuries such as myocarditis), so Bowden's figure of an average award around $4,000 for the few claimants who are paid is roughly accurate, though a handful of severe-injury outliers were far larger.

What the evidence shows: The best available data point the opposite way. A peer-reviewed 20-year study (2002 to 2022) of more than 2 million NCAA athletes, published in Circulation, found that sudden cardiac death incidence decreased over the period, falling roughly 29 percent every 5 years (an overall decline of about 70 percent), while non-cardiovascular deaths stayed stable. The same study recorded no deaths attributed to COVID-19-related myocarditis, and the single post-2020 myocarditis death was giant cell myocarditis, a pattern not linked to COVID-19. The viral '29 to 290, tenfold' figure does not appear in this or other registry-based epidemiology and typically comes from comparing incompatible media-tracked tallies rather than measured incidence rates. Status: False.

What the evidence shows: Health Canada's Fifth Annual Report on Medical Assistance in Dying reports about 15,300 MAID provisions in Canada in 2023, a 15.8 percent increase over the 13,241 provisions in 2022. That confirms the figure of more than 15,000 people receiving MAID in 2023. Status: accurate.

What the evidence shows: Bowden founded the advocacy group Americans for Health Freedom in late 2023, which recruits politicians to pledge that the mRNA COVID shots should be pulled off the market, so the roughly 252 figure is consistent with the group's own steadily growing tally (public reports cite about 250 elected officials in early 2026 and 300 by mid-2026). That count is self-reported by the organization and is not independently audited. The medical premise behind the pledge, that the shots should be removed, runs counter to regulators and peer-reviewed evidence: the FDA states the benefits of the mRNA COVID-19 vaccines outweigh their risks and that it stands behind their safety and effectiveness, and a peer-reviewed review concludes the benefits clearly outweigh the slim risk of myocarditis and pericarditis. FactCheck.org has also rebutted the related claim that the FDA is required to pull the vaccines over residual DNA. The tally is roughly accurate as the organization's own count, but the implied conclusion that these vaccines warrant removal is contrary to the scientific and regulatory consensus.

What the evidence shows: Peer-reviewed research directly contradicts this claim. A systematic review and meta-analysis of test-negative studies found seasonal influenza vaccination reduced the risk of hospitalization with laboratory-confirmed influenza among adults by 41 percent (95 percent CI 34 to 48), with protection of 51 percent among adults aged 18 to 64 and 37 percent among those 65 and older. A separate population study of Ontario residents aged 65 and older estimated vaccination reduced influenza-associated pneumonia and influenza hospitalizations by 19 percent (95 percent CI 4 to 31) and reduced death within 30 days of such hospitalization by 25 percent (95 percent CI 13 to 37). Multiple independent studies thus show measurable reductions in both influenza hospitalization and death, so the statement that the flu shot has never been shown to prevent hospitalization or death is false.

What the evidence shows: A 2025 Cleveland Clinic preprint tracking 53,402 health system employees during the 2024 to 2025 season did find vaccinated workers had a higher rate of influenza (hazard ratio 1.27, a calculated vaccine effectiveness of negative 26.9 percent), so the direction of Bowden's claim has a real basis, though the figure was about 27 percent, not 24 percent, and it applied to laboratory-confirmed influenza, not general sickness. The study was a non peer-reviewed preprint limited to relatively healthy healthcare workers in Ohio, and per PolitiFact it did not represent the general population and, critically, did not measure the vaccine's main purpose: reducing severe illness, hospitalization and death. Experts told PolitiFact the analysis was subject to testing bias (vaccinated staff were more likely to be tested) and other confounders, and the study's own coauthors rejected using it to claim the vaccine should be withdrawn. Presenting this single-season, one-population preprint as evidence that flu shots make people 24 percent more likely to get sick omits these limitations, so the claim is mixed.

What the evidence shows: National estimates from NHANES do not support a 74 percent deficiency rate. An analysis of NHANES 2001 to 2018 (71,685 participants) found combined moderate and severe deficiency (serum 25(OH)D below 50 nmol/L) at about 24.6 percent, with insufficiency at 40.9 percent and sufficiency at 34.5 percent. Using the National Academy of Medicine thresholds, NHANES 2011 to 2014 found only 5.0 percent at risk of deficiency and 18.3 percent at risk of inadequacy, meaning roughly 73 percent of the population was sufficient. A higher cutoff of 20 ng/mL (50 nmol/L) yields the largest common figure, about 41.6 percent deficient overall, and the 82.1 percent rate reported at that threshold applies specifically to Black Americans, not the whole country. The 74 percent figure appears to conflate deficiency with the broader insufficient plus deficient categories or is simply overstated: it is Exaggerated.

What the evidence shows: Published surveys of the mRNA clinical development landscape do not support figures of 500 mRNA shots or 33 self-amplifying candidates. A 2025 peer-reviewed review of therapeutic and self-amplifying RNA reports that only about six saRNA vaccine candidates have entered human clinical trials (COVAC1, ARCT-154, RBI-4000, LNP-nCOV saRNA-02, ARCT-021 and GEMCOVAC-OM), and that just one saRNA product, Arcturus's ARCT-154, has been approved anywhere. The pivotal ARCT-154 program, published in Nature Communications, describes that single saRNA COVID-19 vaccine rather than any large slate of self-amplifying shots. Broader tallies of all mRNA candidates (vaccines plus therapeutics) run in the low hundreds across cancer and infectious disease, most in early phases, well below a total of 500 self-amplifying-capable products, and the specific 500 and 33 figures trace to advocacy blog posts rather than a registry or regulatory source. The claim is exaggerated: the documented saRNA clinical pipeline is roughly six candidates, not 33.