Sunday, May 4, 2008

"Brilliant Sleaze" Award Goes to Pfizer's Heart Risk Calculator

In a previous blog entry,, I mentioned the whistleblower suit filed on behalf of Dr. Jesse Polansky against Pfizer, relating to alleged off-label marketing of Lipitor. In that earlier post I alluded to risk calculators as one part of the Pfizer Lipitor marketing strategy. When I first wrote the post, based on my reading of the complaint in the suit, I admit to not having understood very well just what some of the allegations were. Now that I have had some further correspondence with Dr. Polansky and others who know more about these things, I can see that I failed to report what is actually an extremely clever (and if true quite slimy) example of marketing winning out over science.

This gets pretty intricate so see if you can follow me here.

The current cholesterol treatment guidelines (written by a group of experts, most of whom have serious conflicts of interest with manufacturers of statins) depend a good deal on the physician’s ability to assign to each patient a risk level for future heart attacks or other cardiovascular events. Here are four risk group—high, moderately high, moderate, and low—each with its own cutoff level of LDL. That is, the higher the risk group you are in, the lower the level of LDL you should have in your bloodstream, before a physician recommends to you that you should take a statin. Bottom line—drugmakers will sell more statins (like Pfizer’s Lipitor) if somehow more patients can be nudged over from a lower risk classification into a higher one.

Now, for reasons I have covered in previous posts, this entire theory of when statins are good for you is probably deeply flawed, and as a whole shows how industry marketing has come to dominate science to an embarrassing extent. But for present purposes ignore all that. Imagine that the guidelines actually represent good science. The only remaining issue that concerns us is—did we accurately classify the individual patient as being high or moderate or low risk? Suppose that at certain risk level, the guidelines say you should take a statin if your LDL is over 130; and at a different risk level, the guidelines say you should take one if your LDL is over 100. We will not ask any more if there is evidence that patients with those numbers get any real benefit from statins. All we will ask is whether the patients were correctly placed in the given risk category.

It appears that the NIH-based National Cholesterol Education Project (NCEP) came up in the past with two versions of a heart-attack risk calculator, based on the Framingham data. One is a computerized calculator that lives on an NIH website. This calculator relies on continuous variables and so provides an accurate assessment of heart attack risk.

The other NCEP calculator is a paper-and-pencil calculator, designed for docs who are less computer savvy. The paper calculator classifies patients by a yes/no algorithm rather than by continuous variables, and so without boring you (or me) with all the math, it turns out to misclassify a number of patients as high-risk who are actually moderate-risk. It appears that the paper calculator does not misclassify any patient downward by mistake, only upward in terms of risk level. So in effect the paper-and-pencil NCEP calculator does exactly what a company like Pfizer would like it to do, if the goal is to increase Lipitor sales.

Now, Pfizer could have at this point simply thanked its lucky stars and kept quiet about all this, but that would not be an intelligent marketing plan by today’s standards. So what Pfizer (according to Dr. Polansky) proceeded to do was a very complicated trick. It placed a computerized risk calculator on its Lipitor website. It labeled the calculator as being the same as the NCEP calculator, thus disarming any possible suspicion that the average doc might harbor about using a drug-company-supplied device. But the Pfizer calculator is not the same as the NCEP computerized calculator (the accurate one). Rather, the Pfizer on-line risk calculator exactly emulates the NCEP paper-and-pencil calculator, the one that regularly and routinely misclassifies patients as high-risk and therefore needing statins at lower LDL levels. (By contrast, AstraZenica and Merck have Framingham risk calculators on their websites, and those match the more accurate NCEP computerized calculator—suggesting how much extra work the Pfizer programmers had to go through to get their computerized calculator to act like a paper-and-pencil one.)

Dr. Polansky has done some preliminary number-crunching (he tells me he is working on more precise figures) that shows that if the less accurate calculator were universally used, the net result would be extra millions of patients who would be placed on statins, who really ought not to be (even according to the flawed “science” that says they need them at all).

It appears that Pfizer, not content with its coup so far, has worked hard to get its version of the computerized calculator distributed to docs in the form of PDA programs, and distributed by other on-line information sources, but on these details I remain fuzzy.

Dr. Polansky has lodged complaints and queries with people at both NCEP and Pfizer about these matters. Basically, no one at either place seems to think there is any problem. NCEP didn’t initially seem bothered that its two calculators gave different answers and that the average doc would remain clueless that this is so. (After being badgered, NCEP did admit that maybe some sort of warning notice should be added.) NCEP also seems unconcerned that a commercial firm would deliberately misuse NCEP-created materials for its own marketing advantage, while leaving NCEP accountable for the results.

Now, I am not sure whether this ploy gets a brilliance award or a sleaze award, so I will grant it (if all the facts turn out to be as alleged) a “brilliant sleaze award.”


Bernard Carroll said...

Brilliant sleaze, indeed! Never underestimate the willingness of corporations to put stealth marketing before science.

Anonymous said...

Dr. Xuemei Huang, UNC chapel Hill has proposed a clinical study of 1600 participants to determine if an association exists btn lipophilic statin use and parkinson's disease, based upon earlier findings of positive correlation btn low LDL and parkinson's disease. Dr. Duane Graveline MD, developed and maintains a web site for statin sufferers,, to which thousands of individuals have logged on to record the adverse effects suffered from statins. There exists much scientific data connecting lipophilic statin use to the development of neurodegenerative diseases--alzheimer's, ALS and Parkinson's.
Please access Dr. Graveline's site to appreciate the myraid functions of the metabolic products of the mevalonate pathway that is blocked by HMG CoA reductase inhibitors.

Anonymous said...
This comment has been removed by a blog administrator.
Anonymous said...

I do not know how much of this is true but it sure seems eminently plausible and Pfizer ought to be embarrassed about it. Within Pfizer, there are people like Dr. Michael Berelowitz, who are preaching the mantra of "Medical Professionalism". After reading blog posts like this, it is clear that either the medial groups that people like Dr. Berelowitz lead within Pfizer are toothless tigers or Dr. Berelowitz and his leadership are just paying lip service to "Medical Professionalism". For Pfizer's sake, I hope many of the things you mention here are not true.