Why Should We Swallow What These Studies Say?
By Jerome P. Kassirer
Sunday, August 1, 2004; Page B03
In a society rife with conflicts of interest, disclosure of such conflicts is usually a good tonic. In finance, we can read the fine print and decide whether to invest or seek other advice. We can hedge our bets. But in medicine, where decisions on treatment can have lasting effects, mere disclosure isn't enough. Patients need advice they can act on without having to calibrate how likely it is to be biased. Physicians and scientists with financial ties to the pharmaceutical industry should not just have to disclose conflicts -- they shouldn't be permitted to issue guidelines at all.
But they are permitted, and they do so routinely. The most prominent recent example of this is how the federal government came up with and then defended new recommendations on cholesterol levels for individuals with a high risk of heart disease. It's an enlightening -- and depressing -- story.
On July 13, the National Cholesterol Education Program (NCEP), part of the National Institutes of Health, unveiled tougher guidelines for cholesterol levels -- guidelines so stringent that millions of Americans at risk of heart disease would have to take costly statin drugs to meet the new lower limits. What the NCEP didn't unveil was that most panel members who helped write the recommendations had financial ties to the pharmaceutical companies that stood to gain enormously from increased use of statins.
Critics immediately complained about the hidden financial ties, and demanded disclosure. Within days, the highly respected sponsors of the cholesterol guidelines -- the NIH, the American Heart Association (AHA) and the American College of Cardiology (ACC) -- posted the disclosures on the NCEP's Web site. The extent of the connections was stunning: Of the nine members of the panel that wrote the guidelines, six had each received research grants, speaking honoraria or consulting fees from at least three and in some cases all five of the manufacturers of statins; only one had no financial links at all.
If all the members with conflicts had recused themselves, in fact, only two would have been left.
That didn't look too good, so a day or so later, another note appeared on the site, attempting to make the guidelines seem more credible. It explained that the panel's draft proposals had been "subjected to multiple layers of scientific review," first by the NCEP's coordinating committee, "consisting of 35 representatives of leading medical, public health, voluntary, community, and citizen organizations and Federal agencies," and then by the scientific and steering committees of the heart association and the college of cardiology. "Altogether approximately 90 reviewers scrutinized the draft," the note said. So the message to the public: No need to worry about pro-industry bias.
Meanwhile, the heart association, whose journal Circulation had published the guidelines, attempted to dispel any hint of bias or collusion with industry. It sent an e-mail message to its board of directors, its national strategic team, its communications advisory team and more than 30 prominent physicians who have worked closely with the organization. It reminded them of the association's conflict of interest policy, namely that "if in any situation, a panelist has a current relationship that could unduly influence guidelines or statements, that individual recuses himself from that aspect of the work of the writing panel. This process ensures that the guidelines are not inappropriately influenced in any way." (Apparently, none of the panelists had felt their drug-company connections required recusal.) The association added that "in the future, the AHA plans to publish online disclosures for all articles and guidelines published in any of our journals, whether they are developed by the AHA or not."
Disclosure and recusal, the association is essentially saying, can clean up a messy problem. But the fact is, patients deciding whether to take these drugs, and physicians deciding whether to prescribe them, still don't know whether the NCEP panel members consciously or subconsciously colored their analysis in favor of statin makers Merck, Novartis, AstraZeneca, Pfizer and Bristol-Myers Squibb.
Cholesterol guidelines have broad impact. They help doctors decide how aggressively to prescribe drugs. When the guidelines promote greater use of statins, they also raise the cost of care. And statins are not as benign as vitamins. In some cases, they cause liver and muscle injury; in rare instances, they have led to kidney failure.
So why in the world did three major organizations choose such a conflicted panel to write the guidelines? Quite likely one reason was that the panelists were experts in the field. Most of them had helped to write the preceding round of cholesterol guidelines three years earlier.
Is it imaginable that using conflicted experts is the best way of getting an unadulterated assessment of clinical data? I don't think so. The best collective decisions arise from diverse and independent views. When companies with identical interests are underwriting virtually all the researchers, decision makers can become susceptible to "group think." The military has a name for this sort of trap -- "incestuous amplification."
Does the review of the guidelines by 90 other leading experts guard against potential bias? Not really. Many of the other reviewers had conflicts of interest, too. The heart association conceded the point in its e-mail message, saying, "The best scientists are asked to speak and provide advice, not only by the NIH and the AHA, but by industry. Because of this, many of the top scientists working on any issue will likely have received some funding over time from pharmaceutical companies." Even if the 90 other scientists had no conflicts, how diligently did they examine the data or challenge the expert panelists? Was there dissent among those reviewers? If so, did anyone listen? We do not know.
There certainly is no unanimity of opinion regarding the lower cholesterol targets. A University of British Columbia group (with no connection to drug manufacturers) has drawn different conclusions from the same studies. We are left to contemplate whether contradictory evidence was minimized by people looking through green-colored glasses.
In describing the use of disclosure in business circles, an economist recently said, "It has become a truism on Wall Street that conflicts of interest are unavoidable. In fact, most of them only seem so, because avoiding them makes it harder to get rich. That's why full disclosure is so popular: It requires no substantive change." Disclosure covers up the question of bias with a patina of honesty. It tells you little or nothing about whether the disclosed conflicts of interest actually tainted a report. Because the issue of bias challenges the very integrity of researchers, people in the field shy away from talking about it.
© 2004 The Washington Post Company