Thursday, March 11, 2004

Diseases That You Die Of Trump Those That You Don't 

I have a patient with a 5 year risk for a cardiovascular event of 9%. His blood pressure is well controlled on a thiazide diuretic and beta-blocker. He came in for my opinion about changing his anti-hypertensive regimen.

Seems a dermatologist told him the beta-blocker may be causing his "skin lupus" to get worse. It is under control unless he is exposed to a lot of sunshine. Since he is an avid traveller and sun worshipper he wanted to change.

He lives out of town and maintains more than one primary care doctor. They stopped the beta-blocker and changed him to Norvasc. His BP went up. The Norvasc was changed to Verapamil. Again, BP went up.

We decided to go back to the original BP regimen and put up with some skin issues.

Wednesday, March 10, 2004

New Cholesterol Goals? The PROVE IT trial 

Recent information about more aggressive LDL cholesterol goals caught my attention for several reasons:
1. I don't see much drug vs drug literature.
2. Lipitor has less outcomes data than Zocor and Pravachol. I've anticipated a class effect but have been reluctant to initiate therapy with this drug. If I inherit a patient on Lipitor I usually leave them alone.
3. I've not been completely convinced that the LDL < 100 threshold was ever as important as the NCEP guideline made it out to be. Until recently, I think they were guility of setting goals too agressive for the available research.

As usual, the hype surrounding the research isn't as impressive if I bother to read the actual NEJM article (I have full text access but have only provided the abstract here).

I'm OK with validity (double blind, small drop out, intent to treat analysis). I do have some reservations because the study protocol describes monthly 10 day courses of the antibiotic Tequin or placebo without including the results or evidence the subjects were matched for this.
Of note, the study was set up to determine equivalence by the maker (Bristol-Myers Squibb) of the drug (pravastatin) that compares less favorably (atorvastatin).
The endpoints are "combined":
1. Primary endpoint = death from any cause, MI, unstable angina that results in an admission, or revascularization by either PTCA or CABG and CVA.
2. Secondary endpoint = death from CAD, MI, or revascularization.

What I'm struggling with here are the various sweeping conclusions. Let me explain by putting the article through some analysis:
1. The authors say "The benefit of high-dose atorvastatin as compared with standard-dose pravastatin emerged as early as 30 days and was consistent over time (Fig. 3)." I don't agree. Figure 3 hazard ratios cross unity for 30 and 90 days.
2. They say: "The benefit of high-dose atorvastatin was consistent across the prespecified subgroups." That isn't the case regarding age, diabetes, smoking, prior statin therapy, or MI with ST elevation (Fig. 5).
3. The discussion section states: "The reduction in the rate of death from any cause was of borderline significance (28 percent, P=0.07) suggesting that more aggressive lipid lowering is important not only to reduce the risk of recurrent ischemia, but possibly also to decrease the risk of fatal events." Sorry, but no matter how appealing the theory and trend, they shouldn't say that. I find it disturbing when authors use language like "trend" or "borderline". There simply isn't an all cause death benefit.

How about the results?
1. The primary endpoint relative risk reduction (RRR) is actually an absolute risk reduction (ARR) from 26.3 to 22.4 percent = 3.9% so number needed to treat with atorvastatin 80 mg instead of pravastatin 40 mg for 2 years to prevent a combined endpoint is 26. The Yahoo link above points out the difference in cost for the two drugs is Lipitor ($1,400) versus Pravachol ($900). So, 1400 - 900 = $500 x 2 years x 26 patients = $26,000 to prevent one combined outcome.
2. The secondary endpoint has NNT = 38.
3. The NNT to prevent a revascularization = 40.
4. The NNT for unstable angina requiring hospitalization = 77.
5. There is no all cause death benefit.
6. There is no death benefit from CHD.
7. There is no combined MI benefit.
8. Regarding safety, there is a statistical difference in the disease oriented outcome (DOE) of abnormal lab work (ALT> 3 times normal, ARR from 3.3% to 1.1% or NNH = 45) but it did not translate into a patient oriented outcome (POEM) of either discontinuation for symtpoms or actual rhabdomyolysis.

Bottom Line:
1. This is now good evidence that lower LDL goals for high risk patients result in better outcomes. The NCEP guideline was premature in this regard.
2. There is no mortality benefit.
3. The benefit isn't for at least 180 days.
4. I don't think one drug is superior.
5. Pravachol now has an 80 mg indication.
6. I would like more information on safety.
7. Combined endpoints were required.

Tuesday, March 02, 2004

Believing is Seeing 

I read with interest a recent DB's Medical Rants post about the adoption of evidence-based interventions.

The Februrary 17, 2004 Circulation has two articles of interest:
1. Impact of Combination Evidence-Based Medical Therapy on Mortality in Patients With Acute Coronary Syndromes
2. We Must Use the Knowledge That We Have to Treat Patients With Acute Coronary Syndromes

It seems to me it isn't enough to prove an intervention is effective.
I have to balance competing interests (time, cost, convenience...).
Number needed to treat (NNT) levels the playing field.
Cost benefit analysis helps to decide among equally effective interventions.

Let's take the first article: Impact of Combination....
1. I'm not surprised the individual treatments were all independently show to work. It just seems unusual that it took so long to measure the benefit of the real world effect of combining them.
2. It is also important to remember this is secondary prevention. It is easier to show a benefit in this group than the same intervention for primary prevention. Not many of my colleagues know this.
3. The discussion section does a nice job of summarizing the original work for the interventions that I now take for granted.
4. Results are "odds ratios" instead of risk reduction. I can live with that. I just wish they would go ahead to do the absolute risk reduction (ARR) and NNT.

The second article is: Focused Perspective, We Must Use...
1. They were nice enough to give some NNT's. Examples: GUSTO = aspirin would save 9 lives per 1000 treated, beta blockers would save 11 lives per 100, and ACE inhibitors would save 23 lives per 1000. They even remind us 13 lives per 1000 would be saved by not using calcium-channel blockers.
2. Unfortunately, they are guilty of one of my pet peeves. When referring to the first article after noting the benefits of combination therapies they say: There was also a strong trend (P=0.08) toward increased survival in patients who underwent revascularization procedures...". The 95% CI = 0.05-1.24 which means the trend is irrelevant. It is either stastistically valid or it isn't. The authors lose credibility for me when they are soft on this issue.

This page is powered by Blogger. Isn't yours?