Wednesday, May 23, 2012

Statins in the Water Supply, Continued--Why the New Meta-Analysis Is Unreliable

In the immediately previous post, I said that I expected more commentary to appear on the CTT meta-analysis of statins for low-risk patients recently published in Lancet. In this post I will 1) report on one of those commentaries and 2) fill in what I failed to say in the previous post, which is why this whole issue is relevant here.

First, commentary, by Dr. David Newman of Mt. Sinai in New York, and author of the excellent book, Hippocrates' Shadow: http://smartem.org/content/data-drugs-and-deception-true-story

Dr. Newman asks why the CTT meta-analysis, which said that low-risk patients benefit from statin therapy, is so much at odds with previous reviews, conducted by reputable groups, whoch all agreed they wouldn't. He decided that it's because the older reviews answer one question and the CTT meta-analysis, another.

The old reviews, in his (and my) opinion, answer the real question: if patients are put on statins, will they benefit in terms of lower rates of heart/vessel disease and/or lower overall mortality? The answer is that right now there's no proof of that for lower-risk individuals without known heart or vessel disease.

The reason that the CTT folks got as different answer is because they asked a different question: if you go on statins, and if the result of going on  statins is that your cholesterol drops by a certain amount, will you have less heart/vessel disease and lower overall mortality? They found that the answer is yes, and apparently Dr. Newman thinks they crunched their numbers reliably and so this is probably a true result.

Now, suppose I'm a family doc and Mr. Jones, a 50-year-old man with apparent low risk for cardiac disease, comes for his annual visit, and I have to decide whether to put him on a statin. The older reviews address my decision, showing that if all I know is what I know now, then I cannot predict reliably that Mr. Jones would benefit.

The CTT analysis is irrelevant. Maybe if I put Mr. Jones on a statin drug, then he'll drop his LDL (bad) cholesterol by a certain number of points, or maybe he won't. There could be many reasons why he won't--maybe his body works differently, or maybe he forgets to take his medications, or who knows what. Since I cannot guess how he'd respond to the statin, the new analysis really provides me with no useful information. The old studies provide useful information, suggesting that I might have a talk with Mr. Jones about his options, but certainly not specifically recommend a statin drug.

Put another way, the CTT analysis shows how one particular subpopulation of people placed on statins respond, and that for whatever reason--because of the statin, or for some other reason--they seem to do well. But the overall population (low-risk patients without existing disease) is not addressed.

Dr. Newman notes that given this highly selective subpopulation that they looked at, the CTT folks have no grounds to recommend what they did, which is that guidelines should be revised massively to increase the number of patients for whom statins are recommended. The CTT folks (I'd add) were also disingenuous to the extent of their not openly addressing how and why their study came to such different conclusions from the earlier meta-analyses, coyly suggesting it was because they are so much smarter and the old analyses were plain wrong.

OK, so now why am I once again ragging about cholesterol and statins, on a blog that's supposed to be about the ethics of the medicine/drug industry interface? I cannot find clear evidence that the CTT group is in the pay of the drug industry; but they are basically acting as if they were. They are illegitimately recommending that a vast number of new patients should be prescribed statins. They are doing so based on an incorrect analysis, in a way that would probably not be obvious to the average reader, but as Dr. Newman shows is quite evident to the expert who understands data analysis. The Lancet is aiding and abetting this distortion of the data, and by appending an enthusiastic commentary, is even tooting the horn louder in favor of prescribing more drugs. All in all, one more sorry illustration of how the medical literature all too often serves drug industry revenues rather than patient health.

No comments: