Erin Brockovich, the real-life character who brings fictional lawsuits, thinks my criticism of her trolling for clients in Avandia lawsuits is “shameful”. She should know from shameless.
I was very amused by Brockovich’s remark “It is no coincidence that thousands on Avandia now have heart attacks.” Really? Thousands of people who saw Erin Brockovich in the theaters have had heart attacks, and many others have had strokes. Some even contracted cancer! Coincidence, or has Ms. Brockovich put movie royalties ahead of safety?
8 Comments
Speaking of “straw dogs,” as she calls them… she sounds like she thinks you’re suggesting that trial lawyers’ internet advertisements be prohibited or something.
This rant of hers sounds like religious banter – Climate change alarmist use the same technique: wax epistemology, toss in some victim advocacy, say something about fat-cat corporations, and you have everything you need for a good daily outrage.
One of the problems with the legal system is the exploitation and/or misunderstanding of statistics. Brokovich’s statement is nonsense. Ted recognizes this, but doesn’t quite put his finger on what is wrong.
Welcome to the world of Bayes’ theorem and conditional probabilities. See http://en.wikipedia.org/wiki/Bayes'_theorem (example 2 therein).
See also the prosecutor’s fallacy http://en.wikipedia.org/wiki/Prosecutor%27s_fallacy
Many legal (and medical) errors are made due to a misunderstanding of conditional probabilities, and I see it all the time in real life and on the legal blogs and in medical testing. For example, on a recent blog criticizing the jury’s deliberation in 12 angry men, one commentator wrongly stated that if each piece of evidence proved guilt by 70% then with 4 pieces of such evidence he is guilty to a certainty of 99.2% =(1-.3**4).
In order to prove or disprove a relation between taking Avandia and having a heart attack, one would need the missing information to plug into the Bayes equation, namely the incidence of heart attacks in the general population. That also goes for proving (or disproving) a relation between seeing Erin Brokovich and having a heart attack.
Rarely in medicine are diagnostic signs pathognomonic. Rather, a diagnosis is made by a probabilistic analysis where Bayes’ theorem should come into play. Almost all tests have a finite probability of a false positive and a false negative.
I have a statistics question for VMS. Would data from a placebo group supply the missing data for the Bayes equation? In other words, would a relative risk derived from a double blind study with a placebo group sufficiently form a basis for proving a relation under the Bayes theorem?
Disclaimer: I don’t know what kind of studies have been done re: Avandia and do not know whether any such studies have been conducted.
No placebo trial could be done with patients presenting with indications for avandia. It is clear that those patients need and benefit from treatment, and so, giving them a placebo (and not treating someone in ways with known benefits) would never be approved.
VMS may have meant this, but its not the general population’s rate of heart attack we need. We aren’t handing out avandia to the general population. It would be the heart attack rate in people where avandia is indicated… btw, significantly elevated.
Just had a patient with serious COPD in for a followup for hospitalization for COPD exacerbation – patient was given a great drug for quitting smoking two weeks ago, and told me she had stopped taking it and did not plan to restart supposedly secondary to seeing an ad by a local lawyer that scared her about side effects… wonder how those side effects stack up to continued smoking in a patient with significantly advanced COPD. This is a recurring theme. These ads have population health consequences… and they are not positive. Frustrating!
When will we get a class action for the negative effects of this BS publicity about side effects?
Does not a lawyer’s advertisement recommending that people stop taking certain medications constitute the practice of medicine? How many lawyers are also doctor’s of medicine? And currently licensed to practice medicine?
Seems to be skating on thin ice.
Commentor: “Would data from a placebo group supply the missing data for the Bayes equation?”
Theoretically, the placebo group would give the background rate but only if the placebo group is representative. One of the pieces of “damning” evidence used against Vioxx was a surprising reduction of incidents in the placebo group about midway through the study while the incidence rate among Vioxx users continued unabated and unchanged. The only reasonable conclusion is that there was something wrong with the placebo group. Guess how it was presented though.
DA: “No placebo trial could be done with patients presenting with indications for avandia.”
True but presumably some study was done. I’m not so cynical to assume EB’s claim is entirely baseless. You still need to know the background rate for the general population though — because it’s possible that other treatments may have similar (but different) effect. Any attempts to derive it from the experiment groups would likely hopelessly skew it.
Why don’t you post that it is impossible to post a comment on Brockovich’s blog. You have to login to post a comment, but there is no where to signup for an account! Could it be a case of “free speech for me but not for thee”?
Also, she claims that drug litigation is not about lawyer profits, but about “empowering” the little guy, if that is the case, then why do they take a fee at all?