A British Medical Journal editorial confirms that scientific misconduct by then-Dr. Andrew Wakefield was even worse than previously assumed. The resulting media-fueled panic led parents to refuse vaccination in large numbers, and childhood scourges such as measles soared as a result, with disability and even death resulting. Wakefield was being financed by lawyers hoping to sue the vaccine industry. [Respectful Insolence, CNN, AP, Adler]
16 Comments
Think of all the kids who DIED from preventable diseases because this man’s claims convinced people not to have there kids vaccinated. He should be prosecuted.
What jumps out at me, in addition to the fraud and conflict of interest, is that the original study was based on 12 patients! Is there something about medical studies where the need for statistically significant sample sizes is not required?
An investigation published by the British medical journal BMJ concludes the study’s author, Dr. Andrew Wakefield, misrepresented or altered the medical histories of all 12 of the patients whose cases formed the basis of the 1998 study — and that there was “no doubt” Wakefield was responsible.
http://www.cnn.com/2011/HEALTH/01/05/autism.vaccines/index.html?hpt=T1&iref=BN1
It is interesting to note that Lancet seems to have a history of publishing fraudulent studies such as the ones that claimed that the US was responsible for 654,965 excess deaths in Iraq.
http://en.wikipedia.org/wiki/Lancet_surveys_of_Iraq_War_casualties
Richard Nieporent (comment 1) raises an important point. How did such modest work have such a dramatic effect? The interplay between lawyers, doctors, media, and activists is worthy of a book length study. Representative Dan Burton gave credence to Wakefield even after heath officials begged him not to do so. Public flogging is called for for him!
Similar problems are associated with the priest scandal of a couple of years ago. Father Shanley’s conviction was based upon the discredited “repressed memory” as attested to by a known charlatan.
“is that the original study was based on 12 patients! Is there something about medical studies where the need for statistically significant sample sizes is not required?”
Richard:
Pilot studies, especially those involving a new idea, tend to be small. It is too expensive and time consuming to review new ideas with large studies. Usually, after several small pilot studies, if it looks like something interesting has been found, either large studies and/or “meta” studies are done. (A meta-analysis is taking data from published studies and statistically combining it to see if the analysis shows significant results. The published studies do not necessarily have to have directly involved the question reviewed by the meta-analysis — but, whether that is valid or proper as to any particular published study used in the meta-analysis can be a very dicey question.).
What Wakefield was contending would (if true) have been ground-breaking (which may explain publication in Lancet). However, a lot of questions started popping up, since no one could replicate any of the major claims.
Then the information about Wakefield’s conflicts of interests started coming out. In addition to the payments from UK and US mass tort lawyers, Wakefield also held a patent on a measles vaccine — which if the MMR vaccine was declared unsafe, would have been its readily available replacement (just get a Pharma Company interested in signing a royalty agreement and gearing up production).
In the meantime, there were a lot of media-hysteria articles that “vaccines cause autism leading scientists claim.” How many of those were plants by attorneys with friendly reporters, vs. the media-lemming effect, is unknown.
When Brian Deer began investigating and reporting, Wakefield’s story began publicly unraveling — leading, eventually, to Lancet retracting the article, and the various government proceedings in the US and UK.
Pilot studies, especially those involving a new idea, tend to be small. It is too expensive and time consuming to review new ideas with large studies.
That does not address my objection that with a sample size of twelve the results are not statistically significant. With such a small sample size you can by chance get any result. You can just as easily get a positive result or a null result independent of what the actual effects
are. In other words the result are meaningless. This is not the way to do scientific research.
Can any of the lawyers that made payments to this guy be held to account in any way for this? Can any settlements paid out by defendants in these suits be recovered?
What do the words “peer review” really mean anymore? Anything and nothing at all?
Jonathan Bailey:
No.
As an aside, I wonder if this story is going to be covered on the Oprah Winfrey program, probably the most prominent show that pushed this idea. She was quick enough to denounce James Frey for his falsified memoir. Wonder if she’ll have the guts to tell her millions of followers that she and Jenny McCarthy promoted a charlatan whose fraud has killed children.
This was a reporting of an adverse side effect suspiciously linked to a common medical intervention, not a randomized controlled trial. Thats how many unexpected adverse effects are found after a drug makes it through clinical trials. 12 deaths or serious injuries would be more than enough to set off alarms for just about any drug/intervention. So, ‘statistical significance’ wouldn’t apply in the same way it would in other study designs.
Without the knowledge of fraud, secondary gain, and bad science that we have now, that paper was a sensational and compelling ‘alarm’ forcing everyone to take a careful look at a potentially devestating side effect.
As someone who occasionally reviews articles for medical journals, it is not fair in my experience to say that the peer review process means ‘nothing’. I do, however, agree with a statement in one of the articles about this paper that scientific journals spend a lot of time being skeptical about methods and interpretation of data, but essentially no time worrying about fraud.
Richard: “statistical significance” is not a relevant issue here for several reasons.
Primarily, “statistical significance” is relevant for, well, statistical studies. That is, studies which attempt to measure an effect while controlling for other effects. It’s not relevant for studies which select the patients based on the effect to be studied. Obviously writing the case histories of 12 children with autism has no “statistical significance”, since there were not 12 random children — but it’s absurd to expect that every new idea be ignored until there are enough cases observed to do a randomized study.
Thus, a doctor who sees (what may be) a new disease might sometimes write a paper describing a single patient. The point is to document the new phenomenon so that others who see something similar will be able to make the connection. Once enough anecdotes accumulate it may be time for a larger-scale study to see if there really is something there.
Consider a hypothetical doctor reporting about 12 people in the UK who didn’t eat vegetables and later got scurvy. The doctor is proposing that something in vegetables helps prevent scurvy. Is this “statistically significant”? Of course not.
Assume for the sake of argument that scurvy and eating of vegetables have nothing to do with each other, but that neither is terribly rare. In a country as large as the UK it would be easy to find 12 people satisfying both conditions. Does that mean that we should ignore this report?
The truth is that the scientific method is more than about statistics. It’s also about mechanisms, and it’s mostly about degree of confidence in various assertions. So initially we assume there is no connection between vegetables and scurvy. After this paper including 12 case studies, our belief in this possible connection goes up a little — but only a little. How much depends on things such as the details of the case study, the past reputation of the reporter, etc. Depending on the details of the case studies, perhaps the author has a proposed mechanism for the connection. If that proposal makes sense our confidence goes up a little more — but we still are very skeptical. However, if enough such reports come in our level of belief (while low) may be high enough to justify spending money to do a controlled study — take a group of people and randomly assign them to eat or not each vegetables. Now is the time to insist on statistical significance.
The real problem with the Wakefield story is not that papers reporting 12 case studies are bad. It’s that that laypeople can be manipulated into putting their children at risk based on a case study involving 12 people.
Even if it was an honest paper, Wakefield’s report should have been a minor component in our understanding of autism and vaccines. Instead it became a driver of public policy, in that many parents chose to not immunize their children based on such a paper. Why did this happen? Because the press could sell newspapers trumpeting this study, and because the people who read the newspapers lack basic scientific literacy.
Richard, you make it sound like it’s the newspaper’s fault.
Bob
The real problem with the Wakefield story is not that papers reporting 12 case studies are bad. It’s that that laypeople can be manipulated into putting their children at risk based on a case study involving 12 people.
I agree with you completely. My point was not that researchers are not supposed to report observations, but that it is then necessary to devise a study to confirm or disprove the observations. The first part of the scientific method is to come up with a hypothesis. That should have been the role of Wakefield’s paper. Next there should have been one or more studies carried out to validate or disprove his hypothesis. Instead, the publication was used as proof that vaccination caused autism.
Why did this happen? Because the press could sell newspapers trumpeting this study, and because the people who read the newspapers lack basic scientific literacy.
Yes the press was once again irresponsible. However, so was Lancet for allowing Wakefield’s results to be interpreted that way by the press. Lancet knows that almost all of the papers it publishes that contain a hypothesis as to the cause of some disease turn out not to be valid. They should have conveyed that information to the press when they reported Wakefield’s hypothesis as if it was proof that vaccinations caused autism.
In regards peer review.
Peer review is a modern substitute for real science. Peer review means NOTHING.
The only thing that matters is being able to reproduce the results, period. Johnny Researcher says that by doing this, that and another thing, a certain set of results are obtained. Peer review has someone look at that and thumbs up or thumbs down it. So what. That some one has looked it over in no way indicates that the reported results are reproducible. If one truly understands the physical world, then they’ll be able to predict that input X, yields result Y. See F=mA, laws of thermodynamics, stress strain relationships, certain dose of antibiotics killing certain bugs, etc.
Now, if someone goes out on their own, and tries to replicate what Johnny Researcher did, and does the exact same this, that and another thing and gets the exact same results, and THEN goes on to start manipulating this, than and another thing in controlled ways to see how the results change, THAT is science and actually has credibility that the reported effect may in fact be a response to the inputs.
Peer review is mere opinion. Independent replication is fact.
Opinion is the realm of politicians, lawyers and other bull spit artists. Fact is the realm of real science, engineering and physical reality.
Do we know if any kids actually got sick because their parents chose not to vaccinate? I’d say they might be in a position to take action.
L’s comment above corrects Richard Nieporent’s misuse of “statistical significance”. But Mr. Nieporant was absolutely correct when stating that the paper just didn’t have the power to generate the reaction that ensued.
Keep in mind that vaccinations have been in use for decades and have been subject to study over that time. A paper involving 12 patients should be reviewed and checked against the mountains of data extant. That is common sense as well as good Science. It is not surprising that the paper did not hold up to scrutiny. Actual fraud is not surprising either.
I saw another article that said some the 12 children were preselected for the ‘study’ because they had known, pre-existing conditions. Others were members of a ‘parents against vaccination’ group that were willing to exaggerate symptoms so that the study has the results they wanted.
In short, it doesnt seem that there was any real investigation done at all. He had already written the conclusion before the first subject was tested.