The New York Times this morning delivers a sobering post-mortem on the ALLHAT trial and its impact on medical practice. For those not familiar with this landmark $130 million study funded jointly by the National Institutes of Health and several drug firms, ALLHAT compared blood pressure lowering medications, including low-cost diuretics, to see which ones were more effective. While the results have been dubbed controversial -- often by researchers with ties to the drug industry -- the generally-agreed upon takeaway was that low-cost diuretics ought to be used as first-line therapy for lowering blood pressure.

But as today's story points out, medical practice didn't change much in the succeeding years. New drugs came along; the researchers, some with financial ties to industry, questioned each others' motives; and a government-funded push (so-called academic detailing) to educate rank-and-file physicians about the results paled besides the sophisticated and stepped-up marketing campaign waged by drug firms.

Comparative effective studies like ALLHAT, the article suggests, may not be the panacea for rising health care costs that some health care reformers are looking for. “There’s a lot of magical thinking that it will all be science and won’t be politics,” Sean Tunis, a former Center for Medicare and Medicaid Service official who now runs the Center for Medical Technology Policy, which pushes for such trials, told the Times.

This story raises a host of interesting questions, from which I drew several conclusions:

1. Payment mechanisms must reinforce comparative effectiveness information or its impact will be minimal. In England, where the National Institute for Clinical Excellence conducts comparative-effectiveness and cost-effectiveness studies, the single-payer National Health Services uses that information to determine what it will and won't pay for. Next year's debate over establishing a comparative effectiveness agency here must arrive at a national consensus that its studies will be used to make payment decisions by CMS and other insurers. That doesn't mean physicians and medical consumers can't use the less effective or less cost-effective options, whether in response to industry marketing or a heartfelt belief that it is what is in their own best interests. It just means that they will have to pay a higher price for access to those medical interventions. (Another option, of course, is for medical suppliers to lower the price of technologies deemed less effective or only as good as older technologies. Comparative effectiveness research also holds out hope of introducing technological price competition into the health care equation, something that heretofore has not existed. Most new technologies come into use at much higher price points that the older technologies they seek to supplant.)

2. The Food and Drug Administration has to be brought into the comparative effectiveness equation. Right now, when a new drug or device comes along, the new studies, usually conducted in isolation from existing technology, upsets the balance sheet constructed by older comparative effectiveness studies. In the future, initial clinical trials for new technologies seeking regulatory approval should include comparative arms so physicians and payers know immediately where the new technology stands in the pantheon of potential treatments. This will require a change in law from the FDA's current standard, which states only that a new technology must be safe and effective, which is defined as better than nothing (placebo).

3. Finally, the double blind controlled clinical trial, long the gold standard for medical research, may not be the only or even the best standard for comparative effectiveness research. Carolyn Clancy of the Agency for Healthcare Research and Quality suggests near the end of today's article that registries of patient outcomes (the story says "insurance records") can be mined to determine what works best in medicine. This has been done in several well-known cases to uncover safety problems (Vioxx by the FDA's David Graham, who used records from Kaiser Permanente; and Trasylol by Dennis Mangano of the Ischemia Research and Education Foundation -- see this GoozNews post). It's time to move beyond safety and think of this tool in terms of comparative effectiveness.

AHRQ, under Clancy's leadership, has sought to position itself as the potential home for of a beefed-up comparative effectiveness research program. Last year the agency issued a major report on how to construct useful patient outcome registries. Interest in this approach is building.

Alas, getting the research community to agree that the approach is statistically valid is another matter. Graham and Mangano went through a lot of grief before getting other scientists at FDA to accept the validity of epidemiological studies based on patient records.