Steve Kirsch's Survey Reveals Possible Autism Signature
Steve Kirsch put one item to a Facebook group of parents of autistic children: if your child became suddenly autistic within 90 days of a vaccination, how many normal days did you see between the shot and the change? Seventy-nine percent said zero or one. He treats that as inconsistent with "vaccines do not cause autism." An outside reading of the same poll called it something thinner and more awkward: not proof, a temporal pattern that has not been properly tested, and a question radioactive enough to end a career.
Both can be true. The poll cannot carry the conclusion. The question can still be real. Start with what was counted. The sentence begins with "if." It asks only about children already described as regressing within 90 days of a shot, and then asks how soon. It does not ask what share of autistic children regress, what share of regressions follow a vaccine, or what share of vaccinated children regress at all. There is no denominator of children who were vaccinated and stayed as they were. There is no count of regressions in the 90 days before a scheduled visit. Parents were recruited from a group willing to host the survey. "Suddenly autistic" is their category, not a clinic's. Memory supplies the date. A vaccination is the most documented day in a toddler's month, and a parent looking for the last normal day will land on it.
That is enough to stop the headline. Seventy-nine percent of a selected group, answering a conditional question about a memorable day, is not 79 percent of autism. Clustering at day zero is what recall bias produces even when the shot is irrelevant. It is also what you would see if the shot mattered. A biased instrument cannot tell those apart.
The biological calendar makes the trap worse. Regressive autism, loss of language or social behaviour after a period that looked typical, clusters in the second year of life. So do several routine vaccines, including MMR. Two things that both happen around 15 months will look linked in a parents' poll. Large registry studies, Madsen in Denmark, Hviid's later Danish work, Taylor's review, have not found a higher autism risk after MMR, including in children who already had a sibling with autism. Those studies have their own limits. They use diagnoses, not a daily behaviour log, and a short regression can be coded years later. They are still better evidence than a Facebook poll, and they constrain the size of any effect. A toxin that produced autism in a large fraction of vaccinated children would have shown up. A rare cascade in a susceptible minority is harder to exclude, and harder to prove.
Which is the actual research question. Not "do vaccines cause autism?" as a slogan, and not "are parents imagining it?" as a rebuttal. The answerable version is narrower. Among children with documented regression, is onset bunched in the days after a recorded vaccination beyond what age, visit schedules and recall predict? The design is dull and decisive: prospective cohorts, or existing records with vaccination dates and onset dates taken from notes written at the time, a pre-specified window, a control window before the visit, and a comparison with other salient days, illnesses, fevers, antibiotics. Objective exposure, objective onset. No poll.
That study is oddly missing. Not because the method is obscure. Because the result is career-ending in both directions. A null result, published by someone already suspect, will be called a cover-up. A positive result, published by someone inside a university or a public-health agency, will be called misconduct before the methods are read. Agencies have spent twenty years saying the question is closed. Reopening it with a design that could move the answer is not a neutral act. Parents who watched a child change in a week hear the closure as a refusal to look at their calendar. Researchers who have seen what a viral claim does to measles coverage hear the question as a trap.
The cost of not asking it is the poll. A conditional question in a sympathetic group produces a number that is revealing, and the absence of the proper study is taken as evidence that the number was suppressed. If a vaccine/autism link was proven, this would be a big threat to Big Pharma, so don't expect any study anytime soon.
https://kirschsubstack.com/p/parent-poll-shows-nearly-80-of-kids
