September 24, 2026
Host
Welcome to our discussion about one of the most heated controversies in modern medicine: whether thimerosal, a mercury-based preservative in vaccines, caused autism. We're going to unpack the original hypothesis, trace the epidemiological studies that tested it, and consider why the evidence ultimately exonerated vaccines. Along the way, we'll talk about laboratory research, environmental mercury, public fear, alternative therapies, and what happens when science and advocacy collide. There's a lot to cover, so let's get into it. To begin, can you set the stage for why this theory gained so much traction? I'm ready to have my assumptions challenged.
Guest
Thimerosal contains ethylmercury and had been used in vaccines since the 1930s to prevent bacterial contamination. As vaccine schedules expanded in the 1990s, babies received more shots, so their total ethylmercury exposure increased, at least on paper. Around the same time, autism diagnoses rose sharply. Some parents and researchers began connecting those dots. They argued that mercury, even in small amounts, could be neurotoxic, and that children with autism showed similar signs to mercury poisoning. The theory seemed compelling because it offered a clear toxic cause for a devastating and still poorly understood condition. That context matters enormously.
Host
Walter Orenstein had been the first to propose a study evaluating the risk of thimerosal, but preliminary results were horribly flawed, as discussed by conferees at Simpsonwood. The study hung under a cloud, and advocates wedded to the thimerosal-causes-autism hypothesis roundly dismissed its final conclusions. Still, Orenstein's suggested study wasn't the only one. Eight more followed, each seeking to clarify whether thimerosal exposure through vaccines predicted neurological harm. So let's walk through them, because the pattern is striking. Could you start with what happened in Scandinavia?
Guest
In August 2003, Paul Stehr-Green published a paper in the American Journal of Preventive Medicine. He studied children with autism in Sweden and Denmark from the mid-1980s through the late 1990s. He found that the risk of autism increased after thimerosal had been removed from vaccines. By the late 1990s, when health officials had completely eliminated thimerosal, the number of children with autism was higher than it had ever been. That's exactly the opposite of what you'd expect if thimerosal caused autism. He concluded that existing data did not support the hypothesis.
Host
That's a counterintuitive finding, but some might say Sweden and Denmark are small countries with different vaccine schedules. I want to dig into whether that pattern was mirrored elsewhere. Before we move on, could you explain what a removal-related increase might actually mean? Is it evidence against causation, or could it just reflect better diagnosis during that same period? I think that's an important nuance, because critics often argue that rising autism rates after removal don't prove anything about the vaccine itself, since diagnostic practices changed.
Guest
That's a fair challenge. But what's compelling is that the same pattern appeared in multiple large datasets. In September 2003, Kreesten Madsen, an epidemiologist from the University of Aarhus in Denmark, published a study in Pediatrics. He examined medical records of 1,000 children diagnosed with autism between 1971 and 2000. Like Stehr-Green, Madsen found that between 1992 and 2000, after thimerosal had been removed from Danish vaccines, the incidence of autism skyrocketed. He concluded the data did not support a correlation between thimerosal-containing vaccines and the incidence of autism.
Host
Then there was the Hviid study, also from Denmark, published in JAMA in October 2003. Let me get this straight: Anders Hviid looked at the records of Danish children between 1990 and 1996, a period during which thimerosal had been removed from vaccines. He also found that the number of children with autism increased after elimination. His conclusion, again, was that results do not support a causal relationship between childhood vaccination with thimerosal-containing vaccines and development of autistic spectrum disorder. This seems like Denmark provided a natural experiment. What about the United Kingdom?
Guest
In September 2004, Jon Heron from the University of Bristol published a study in Pediatrics. He examined records of 14,000 children who had received different amounts of thimerosal in vaccines between 1991 and 1992. He wanted to see if the amount of thimerosal babies received correlated with neurological problems. He did find a relationship, but not the expected one: the more thimerosal children received, the less likely they were to be hyperactive or to have difficulties with hearing, movement, or speech. That is exactly opposite to what parents concerned about thimerosal would have predicted. Heron found no convincing evidence that early exposure had any deleterious effect.
Host
And the same month, Nick Andrews, an epidemiologist from the Communicable Disease Surveillance Center in London, published a study also in Pediatrics. He examined records of 100,000 children who had received different amounts of thimerosal. Like Heron, Andrews found that the more thimerosal children received, the less likely they were to develop neurological problems such as attention deficit disorder. The amount of mercury in vaccines didn't presage autism. He concluded there was no evidence that thimerosal exposure via vaccines caused neurodevelopmental disorders. These independent findings really start to reinforce each other, don't they?
Guest
They do. In 2004, after reviewing more than 200 epidemiological and biological studies, a committee of the Institute of Medicine issued a statement. The committee concluded that the body of epidemiological evidence favors rejection of a causal relationship between thimerosal-containing vaccines and autism. It further found that potential biological mechanisms for vaccine-induced autism generated to date are theoretical only. The institute recommended that autism research dollars be spent on more fruitful leads. For many mainstream researchers, that was the decisive moment, but the controversy continued, especially among advocacy groups and some parents.
Host
Canada provided another layer. Eric Fombonne, an epidemiologist at McGill University in Montreal, surveyed 28,000 children born between 1987 and 1998. During that period, babies could have received vaccines containing anywhere from zero to more than 200 micrograms of thimerosal. Fombonne found that the group with the highest risk of autism had received no mercury. He concluded findings ruled out an association between autism and high levels of thimerosal exposure. He also attributed the rising numbers to broadened diagnostic concepts, increased awareness, and improved access to services.
Guest
Then in September 2007, Bill Thompson at the CDC published what many consider the most comprehensive and definitive study to date. He carefully determined the exact amount of mercury that 1,000 children had received in vaccines and performed more than forty separate neurological and psychological tests. The study, published in the New England Journal of Medicine, took four years to complete. Its results were consistent with the other six: vaccines containing mercury hadn't caused harm. That's a remarkable convergence across countries, datasets, and study designs.
Host
Finally, in January 2008, Robert Schechter and Judy Grether from California's Department of Public Health looked at autism rates from 1995 to 2007. California had removed thimerosal from vaccines six years before the end point, so if thimerosal were a strong driver, rates should have declined. They found the opposite: rates continued to increase. In an accompanying editorial titled "Thimerosal Disappears but Autism Remains," Eric Fombonne wrote that parents of autistic children should be reassured that autism in their child did not occur through immunizations.
Guest
Those wedded to the notion that thimerosal caused autism were furious. They declared the epidemiological studies meaningless and were sickened that the Institute of Medicine had been persuaded by them. What really mattered, they said, were the biological studies, like those of Richard Deth and Mady Hornig using cells and mice. On NBC's Meet the Press, David Kirby argued that you need to look at biology, toxicology, and the cellular level. He claimed the IOM gave a preponderance of emphasis to epidemiological evidence while giving short shrift to biological evidence. That framing turned out to be deeply misleading.
Host
I want to pause here and ask a fundamental question, because it cuts to the heart of the debate. Epidemiological studies are sometimes dismissed as merely statistical, as if they can't prove causation in the way laboratory experiments can. But what makes them such a powerful tool for detecting rare vaccine side effects? Could you ground that in a concrete example, because I think many people don't appreciate just how sensitive these studies are, even for problems affecting one in ten thousand or one in a hundred thousand children.
Guest
Take the 1998 rotavirus vaccine. The FDA licensed it, and the CDC recommended it for all infants after trials in 10,000 babies. But after it had been given to 1 million babies, epidemiological studies found it was a rare cause of intussusception, a life-threatening intestinal blockage. The problem wasn't trivial; babies can die from bacteria entering the bloodstream or massive bleeding. The risk was about one in 10,000. Of the 1 million children who received it, 100 suffered intussusception and one died. Within months, the CDC withdrew its recommendation.
Host
That's a powerful example. But the sensitivity goes deeper. Natural measles infection can cause thrombocytopenia, a dangerous decrease in platelets, and measles vaccine can also cause it, albeit rarely, in about one of every 25,000 vaccinated children. Epidemiologists detected that too. Even more striking, in 1976 the swine flu vaccine was linked to Guillain-Barré Syndrome at a rate of about one in 100,000. Problems as rare as one in 10,000, one in 25,000, and one in 100,000 were readily detected.
Guest
Right. If autism affects one in 150 children, and even if thimerosal accounted for only 1 percent of autism—one in 15,000 children—epidemiological studies would have found it. Instead, after examining records of hundreds of thousands of children across North America and Europe, investigators couldn't find any evidence of a relationship. It wasn't that their studies were poorly designed or part of a vast international conspiracy. They couldn't find a relationship because it wasn't there to be found. That distinction is crucial.
Host
But here's where the argument got complicated. David Kirby pleaded on Meet the Press to dismiss epidemiological studies in favor of laboratory studies. He was asking for science-in-reverse. Typically, scientists determine that something is a problem through epidemiological studies before determining why it's a problem in the laboratory. That's because animal and cell studies can be quite misleading. I want to explore why that hierarchy exists, because it's central to understanding why the thimerosal hypothesis collapsed.
Guest
In the 1950s, researchers found that hamsters injected with SV40, a monkey virus that contaminated early polio vaccine, developed enormous tumors. Yet many epidemiological studies over fifty years have shown SV40 doesn't cause cancer in people. Similarly, adenovirus caused cancer in hamsters but not in humans. In the 1970s, laboratory cells exposed to large amounts of formaldehyde became cancerous, while morticians who work with formaldehyde aren't at greater risk. These are classic examples of animal and cell studies failing to predict human outcomes.
Host
And laboratory studies can be falsely reassuring too. The Cutter incident in the 1950s is a grim illustration. After Salk's polio vaccine was licensed, Cutter Laboratories produced a batch that wasn't completely inactivated. More than 100,000 children were inadvertently injected with live, dangerous polio virus. Seventy thousand got mild polio, 200 were permanently paralyzed, and ten were killed. Cutter had tested the vaccine extensively in cells, mice, and monkeys, but those studies missed the problem. Edwin Lennette, a leading virologist, said the only way to determine whether something is a problem in people is to test it in people.
Guest
That warning remains true today. In November 2007, Merck suspended an HIV vaccine trial. The vaccine had been remarkably effective in mice and monkeys, but it failed miserably when tested in people. Peggy Johnston, director of the AIDS Vaccine Program at NIH, echoed Lennette: "Mice lie, monkeys sometimes lie, and humans never lie." Cigarette smoking offers another example. In 1939, Alton Ochsner proposed that smoking caused lung cancer. Lab animal studies were inconclusive, but epidemiological studies in the 1950s clearly showed that smokers were at greater risk.
Host
Bradford Hill, the lead investigator on the British smoking study, had the perfect response to those who dismissed epidemiological evidence as merely statistical. He said, "In this particular problem, what experiment can one make? We may subject mice or other laboratory animals to such an atmosphere of tobacco smoke that they can neither sleep nor slumber. And lung cancer may or may not develop. What then? We may have strengthened the evidence, but we must, I believe, invariably return to man for the final proof."
Guest
Beyond the epidemiological findings, understanding mercury in the environment also undermined the thimerosal theory. Mercury is released by burning coal, rock erosion, and volcanoes. It settles into lakes, rivers, and oceans, where bacteria convert it to methylmercury. Methylmercury is everywhere—in fish, water, infant formula, and breast milk. A typical breast-fed child ingests almost 400 micrograms of methylmercury during the first six months of life, more than twice the mercury ever contained in all vaccines combined. And methylmercury is excreted more slowly than ethylmercury.
Host
This connects to a basic principle from the chemist Paracelsus, who declared in the sixteenth century, "The dose makes the poison." The quantity of a substance matters enormously. The earth contains heavy metals like lead, cadmium, arsenic, beryllium, and chromium; large amounts can be toxic, but very small amounts are found in most people. It also works the other way: large amounts of water can cause seizures by depleting minerals, but that doesn't mean water is neurotoxic. The dose, not the mere presence, determines harm.
Guest
The clinical picture didn't fit either. Lyn Redwood and Sallie Bernard argued that symptoms of mercury poisoning were indistinguishable from autism. But Karin Nelson, a neurologist from NIH, and Margaret Bauman, a neurologist from Harvard Medical School, showed the disorders are quite different. Children with mercury poisoning have a narrowing of their visual fields, can become severely psychotic, and have smaller than normal heads. Children with autism don't have visual problems, aren't psychotic, and have larger than normal heads. They concluded the typical clinical signs are not similar.
Host
Despite this mounting evidence, the public fight escalated. In June 2005, Generation Rescue hired Fenton Communications to design a full-page ad in the New York Times titled "MERCURY POISONING AND AUTISM: IT ISN'T A COINCIDENCE." It featured quotes from politicians like John Kerry and Dan Burton. A year later, another ad in USA Today asked, "IF YOU CAUSED A 6,000% INCREASE IN AUTISM WOULDN'T YOU TRY TO COVER IT UP, TOO?" That set a tone of accusation that deeply affected real people.
Guest
Some scientists faced direct threats. Sarah Parker, an assistant professor of pediatrics at the University of Colorado, wrote an article criticizing the Geiers' studies and debated David Kirby on radio. She received threatening emails and phone calls. Callers said, "You need to retract this paper immediately, or else," and "Don't you have children of your own?" Fearing for her eighteen-month-old daughter, she called police, obtained a restraining order, and quit answering her phone for about a year. She never published another paper on vaccine safety.
Host
The Institute of Medicine also had to implement extraordinary security. Marie McCormick, a professor at Harvard's School of Public Health, headed the committee that found no relationship between vaccines and autism. She said the meeting venue had to be changed so the committee could leave without going through the audience. They stayed in the same hotel, came on a single bus, and entered through a garage under the National Academy of Sciences building. One committee member resigned over what he described as a low probability but high impact threat.
Guest
Journalists were targeted as well. Kim Strassel, a Wall Street Journal editorial board member, wrote about the politics of autism. A woman called to confirm she had written the editorial, then Strassel's name and contact details were published on the Internet and circulated to parent groups. She received threatening emails, including one suggesting that if she had a child, she better hope she took good care of him. Strassel said it was the most awful threat someone can make. She could look after herself, but you leave your kids and they're not under constant supervision.
Host
The CDC also received a series of threatening emails. One stated, "Forgiveness is between you and God. It is my job to arrange a meeting." Another lamented, "I'd like to know how you people sleep straight in bed at night knowing all the lies you tell and the lives you know full well you destroy with the poisons you push." The CDC contacted the FBI, instructed staff on safety issues, hired more security guards, and showed employees how to respond if pies were thrown in their faces.
Guest
Meanwhile, some parents were offered a vast array of alternative therapies, all claimed to work but based on ill-founded or disproved theories. Defeat Autism Now, or DAN, promoted a "three R's" approach: remove, reinoculate, and repair the gut. They claimed autism was caused by toxic substances entering through a leaky gut. But studies had failed to prove autistic children have leaky guts, and brain-damaging toxins were never identified. Despite that, DAN doctors ordered expensive, unproven tests and treatments, including chelation, hyperbaric oxygen, and strict diets.
Host
The secretin story is particularly instructive. Some parents were convinced secretin injections treated autism, but a North Carolina study showed that children injected with secretin improved according to parents, and so did children injected with salt water. That didn't mean both worked; it meant parents had a strong desire to see medicines make their children better. They had emotional and financial investments in the idea. Secretin eventually fell out of favor, but chelation, hyperbaric oxygen, and other therapies likely would meet the same fate if carefully tested.
Guest
The personal cost is real. Sharon Humiston, a doctor from Rochester, New York, told Congress that the worst day was not when her son was diagnosed with autism, but when specialists told her his progress was minimal after a year and a half of intense behavioral therapy. Her family tried gluten-free diets, megavitamins, anti-yeast medications, and cranio-sacral massage. Each therapy was supposed to get at the cause; each was expensive, and each was a failure. That desperation makes people vulnerable to false hope.
Host
Jim Laidler, a physician, wrote about his own seduction by alternative therapies. He considered himself a very scientific person. After his son's diagnosis, desperation got the better of him. His family started with vitamins and minerals, then moved to gluten- and casein-free diets, secretin, and chelation. Some seemed to work for a while, which spurred them to try the next thing. The breaking point came during a Disneyland vacation when his son, on a strict gluten-free diet, snatched a waffle and ate it. Nothing happened. Laidler realized he had been chasing random fluctuations.
Guest
Then came the final epidemiological verdict. In January 2008, six years after thimerosal had been removed from vaccines given to young infants in the United States, California health officials announced the most recent autism rates. Because children with autism are typically diagnosed between three and five years of age, six years should have been enough to see a decrease. The rates hadn't decreased; they were increasing dramatically. Robert Davis, then head of the Immunization Safety Office at the CDC, said, "If you remove cars from the highways, you'll see a marked decrease in auto-related deaths. If thimerosal was a strong driver of autism rates and you remove it from vaccines, researchers would have seen some sort of decline—and they didn't."
Host
David Kirby faced this evidence and refused to admit defeat. On November 11, 2006, before an audience of parents in San Diego, he offered another explanation. He claimed children were getting autism because China was burning more coal, causing mercury-filled plumes to envelop the West Coast. He showed slides labeled "Shanghai Plume" and "Mongolian Plume." He also cited California wildfires and an increase in cremations releasing mercury from dental fillings. Arthur Allen, his debate opponent, called it patently ridiculous.
Guest
Some parents were angered by the portrayal of autistic children as vaccine-damaged. Kathleen Seidel, a librarian from Peterborough, New Hampshire, and creator of neurodiversity.com, became a passionate defender. She wrote to David Kirby, urging him to consider how it feels for an autistic person to hear incessant descriptions of autism as an unmitigated tragedy caused by poisoning. She said autistic citizens will always have the diagnosis, and they are affected by how the label is bandied about. She exposed untold motives and unsavory relationships among those trumpeting the mercury theory.
Host
I want to push back a little, because some might argue that laboratory studies can still offer valuable biological plausibility even if they don't prove causation in humans. After all, without a mechanism, an epidemiological association can feel unsatisfying. How do we reconcile the need for both types of evidence without falling into the trap of science-in-reverse? I think this is a real tension, and I wonder if you agree that lab work still has a role, just a different one.
Guest
That's a fair point. Laboratory research is essential for understanding mechanisms once a real risk is identified. But the hierarchy matters because, as Edwin Lennette said, the only way to determine whether something is a problem in people is to test it in people. In the thimerosal case, the lab findings were not consistent with a human risk, and the epidemiological studies were negative. Good science moves from population-level signal to laboratory investigation, not the other way around. When advocates reversed that order, they built a theory on laboratory anomalies that didn't reflect human health outcomes.
Host
So the takeaway for parents and the public might be that epidemiological surveillance—post-licensure monitoring—isn't some cold, statistical exercise. It's the same tool that caught rare vaccine side effects and saved lives, from rotavirus to swine flu. The same tool exonerated thimerosal. And the failure of autism rates to decline after removal is as close to a natural experiment as we'll get. That should be reassuring, even if it means accepting that autism remains a complex, poorly understood condition.
Guest
Communication is the hard part. When scientists say "no evidence of harm," it can sound like "we don't care." When parents hear "the data are clear," they may feel their lived experience is being dismissed. The history shows that trust erodes when people feel threatened or unheard. But trust also can't be built on false certainty or unproven cures. The challenge is to hold compassion for families while insisting on rigorous evidence. That balance remains difficult, but it's the only honest path forward.
Host
Maybe we can close on a note about science as a process. It isn't perfect; it was wrong about some things, and it changed course when evidence demanded. But the thimerosal question wasn't a case of scientists hiding data. It was a case of multiple independent studies converging on the same null result. That doesn't make autism less real or less urgent. It just means we need to look elsewhere for causes and continue to support affected individuals.
Guest
Absolutely. The anxiety came from a reasonable place: autism rates were rising, and parents wanted answers. But the research path that followed was rigorous and conclusive. Thimerosal was removed as a precaution, and autism rates still rose. That is powerful evidence. If anything, the episode shows that public health systems can respond to concerns while also generating the data to test those concerns. The remaining challenge is to direct resources toward more promising leads in autism research and away from disproven theories.
Host
That's a perfect place to end. We've traced the story from a flawed preliminary study to eight epidemiological investigations, all exonerating thimerosal. We've seen how lab studies can mislead, how environmental mercury reframes the dose question, and how alternative therapies often trade on false hope. Most of all, we've seen why the hierarchy of evidence matters. Thank you for this rich conversation. To our listeners, if you found this useful, keep asking critical questions and follow the data where it leads. Goodbye.