Monday, November 18, 2013

New Cholesterol Guidelines—The Devil in the Details

I had been holding off on commenting on the cholesterol treatment guidelines recently issued by the American Heart Association and the American College of Cardiology:
http://circ.ahajournals.org/content/early/2013/11/11/01.cir.0000437738.63853.7a
--because I know that many of the friends that I have previously quoted here, the thoughtful naysayers in the put-everybody-on-statins campaign, were sharpening their statistical knives to go after the new document, and I hoped to be able to link to a truly scholarly dissection of the issue.

I am perhaps jumping the gun a bit egged on by today’s All Things Considered on NPR, which had a lead story on the critics of the risk calculator that’s part of the guidelines, a debate that apparently blew up at the Heart Association’s big annual meeting. So with apologies for my own inability to drill down with any statistical acumen, I will offer a preliminary read.

If you just looked at the guidelines quickly, you’d say that the guideline writers had had a sudden attack of brilliance. Some of the basic points that this blog has been carping on for several years now are reflected in the document:

  • Treatment with statins should be based on the patient’s heart-disease-stroke risk level and not on a blood level of cholesterol.
  • There’s no evidence that shooting for any specific target level of cholesterol is better than simply placing the patient on a reasonable dose of a statin and keeping them there.
Not only that, but the guideline panel seems actually to have taken on board some of the criticisms of conflicts of interest. The chair of the panel reports no conflicts, and while a few of the panelists have seriously long lists of industry funding, many do not. And there’s quite detailed disclosure. (I admit that I did not check to see if any of the people reporting no COI actually could be found to be on the industry payroll.)

So—what’s not to like? Well, a number of things. What the panel giveth, they then taketh away, usually in the fine print of the document.

The panel realizes that there’s no good evidence to recommend a particular target level of cholesterol. But they are impressed (as we’ll see) with the studies that show that lower LDL cholesterol is associated with lower risk. They then urge the physician to prescribe a “moderate” or a “high” intensity dose of their favorite statin, with the idea that the end result will be a certain percentage drop in the LDL level—with the additional proviso that if you don’t achieve that level of drop, you will need to up the dose. So the idea of treating to a target level went out the front door and snuck back in the rear window.

Where did this panel get the idea that you need to reduce LDL by a certain percentage? Well, it turns out that they were quite enamored of one particular study, the CTT meta-analysis in Lancet in 2012. I tried to explain at some length:
http://brodyhooked.blogspot.com/2012/05/statins-in-water-supply-continued-why.html
--why that study was deeply flawed because it answered the wrong question. Moreover, that was not an independent clinical trial; it was a re-analysis of data from many previous trials, and when analyzed by people who are not on the statin bandwagon, all those previous trials yielded much less optimistic results in favor of prescribing statins. Yet if you look at the evidence table that the current panel relies on, “CTT” jumps off the page at every turn.

Next we come to the point that NPR addressed today—the risk calculator. If you no longer treat with statins based on level of blood LDL, but based on the patient’s level of risk, then it becomes critical that you have a reliable measure of that risk. The critics are shouting today that the calculator that the panel favors way overestimates heart-disease risk. Before I get into that discussion I will wait to see one figure that I think is critical—how many additional people will be prescribed statins in the US based on this risk calculator? There are no doubt some people that could benefit from statins who are not now getting them. But there are a whole bunch more IMHO who are currently on statins because their blood tests looked bad and yet they have no diagnosed heart disease and are at reasonably low risk. So if the calculator leads to a net increase in statin prescribing, we know it has to be a move in the wrong direction.

But wait, as those TV ads say, there’s more. Two things are almost completely absent from these guidelines, that ought to be front and center in any practical approach to evidence-based prescribing of statins today. First is the concept of NNT. I didn’t see anywhere where the guidelines stated an actual NNT. The assumption is that if a study shows that statins offer any benefit for any particular group, then you should prescribe the statin—never mind if you have to give a statin to 400 people to prevent 1 cardiovascular event. (If you had an NNT, then you could talk further about shared decision making, giving the patient the facts and letting them decide if they wanted a statin—another concept curiously lacking from these guidelines.) Finally, what’s massively missing is any significant discussion of the downside of statins. According to these guidelines, a vanishingly small number of patients have any serious side effects. The possibility, for example, that as many as half of all patients on statins have some muscle aches or weakness, or the emerging worry that dementia may be more common in statin users, gets no mention whatsoever.

Bottom line—it looks worrisomely like what I concluded with CTT—even if these guidelines were not written by people in the pay of the drug industry, they could just as well have been. And somehow, while important new evidence against the routine use of statins, and suggesting that we really don’t know much about the true mechanism by which these drugs work in the cases where they do, crept into the guidelines for the first time, the bottom line is largely unaffected by such enlightened thoughts. We’re back to putting statins in the water supply.

NOTE ADDED 11/20: I can now cite a comment by some real experts, Drs. John Abramson and Rita Redberg, in the New York Times:
http://www.nytimes.com/2013/11/14/opinion/dont-give-more-patients-statins.html?_r=0
These two estimate that the new guidelines would increase the number of people considered candidates for statin therapy by 70 percent, with 18 percent of these at risk for the adverse effects of statins, with no evidence of fewer deaths from heart disease or stroke as a result. (They mention what most statin advocates ignore, that even the most successful clinical trials tend to show that statins reduce the risk of heart attack or stroke, but still have no impact on overall mortality.)

No comments: