« CANOE and Diabetes as a Surrogate Endpoint | Main | Independent Events »

Jun 28, 2010


Feed You can follow this conversation by subscribing to the comment feed for this post.

What do you make of the statin articles in this week's Archives of Internal Medicine?

I read through them today and need to spend a little more time on them.

First impressions:

The attack on JUPITER seems over the top. I have a number of issues with JUPITER, but the suggestion that it was the only study of its kind to suggest benefits in primary prevention seems like a willful misreading of the literature. It makes me suspicious of all the authors' analyses of "flaws" in JUPITER.

The meta-analysis is surprising for how small the point estimate of a mortality benefit is. I've thought that the real number was closer to 15% in primary prevention but this analysis concluded about 8% (barely missing statistical significance, but that's not really relevant). I'm wondering how much ALLHAT (which had significant flaws) dragged down the number, but overall it may be the case that the relative mortality benefit in primary prevention really is smaller than in secondary prevention. (The absolute benefit is, of course, much smaller than in secondary prevention.) The paper doesn't present estimates of reduction in MI and stroke, which is too bad -- it really does seem to be the first meta-analysis to really sort out patients without prior CVD. Presumably they are planning to publish a follow-up paper with those numbers. I'm uncertain why this paper didn't make it into NEJM, JAMA, or Annals and thus if it has a problem that the reviewers picked up but that I'm not noticing so far. (Anyone care to weigh in on this?)

Any return to this post with a more informed analysis of Archives publications would be greatly appreciated. I too wondered why these were not published in one of the big three. I find this topic fascinating, and beyond the methodology and motivations of authors, the implications for population health, the dogma of "statins = good, put them in the water," and the public's awareness of this class of rx and the desire to take them make this kind of appraisal so important.

I found the analysis reasonable and did not walk away with any skepticism. However, I did sense an ax grinding away in these submissions--clearly they have opinions.


The failure of primary prevention trials to provide much in the way of a mortality benefit (versus the clear benefit in secondary prevention trials) means that trials - and we in clinical practice - are doing a very poor job of selecting the right types of patients who will benefit from statins.

We should be sorting out who is at higher risk in primary prevention - who in fact actually is a secondary prevention patient among the primary prevention "masses" by virtue of the fact that they have subclinical cardiovascular disease. In my case, that means carotid atherosclerosis imaging. I find many so-called primary prevention patients who actually have quite advanced disease. Since none of the primary prevention trials actually selected out for these types of patients, it is no wonder that they have failed to demonstrate a true mortality benefit. Therefore, we should not be selecting based on risk factors alone - that much seems clear from the meta-analysis, and from the individual trials themselves.

If you don't think this approach is evidence-based, you have to answer the counterfactual question, which is why the large subset of primary prevention patients with demonstrated atherosclerosis would NOT benefit proportionately from statin therapy (in relation to their baseline risk).

It would seem wise to keep the main points of your thoughtful post in mind when reviewing the maeltrom of editorial comments unleashed by the recent meta-analysis published in Lancet-Oncology on Angiotensin Receptor Blockade and risk of cancer. An unpopular result does not make a study "junk". The authors are careful to state that the clinical implications are unknown. (All the "he said-she said-I asked-they didn't reply" aside, the exclusion of the VALUE trial does appear to be an important limitation).

The comments to this entry are closed.