Health knowledge made personal
Join this community!
› Share page:
Search posts:

Statins Don't Cause Diabetes

Posted Feb 17 2010 6:06am
As of this morning, the only major source reporting the results of a recent study from Lancet is Business Week. However, the headlines that will likely pop up in the next few hours will likely be similar to theirs: "Cholesterol-Cutting Drugs Raise Diabetes Risk by 9% in Study." In my last post I discussed how the media likes bad news stories, so even though this is not yet all over the headlines, I will take a preemptive strike for potentially concerned patients.
The Lancet study is a large major meta-analysis of major statin trials, looking to see if there is a risk associated with statins and the development of diabetes.

The authors likely got the idea to do a meta-analysis from the JUPITER trial which showed that Crestor reduced heart attacks in patients who wouldn't have generally gotten a statin, except they had an elevated CRP (see Crestor: Get Ready to Ask Your Doctor for the CRP Test. ) One thing seen in this study was a potential increased risk of the development of diabetes. I mentioned this in my blog post when the data was first released that physician report of diabetes was increased in the Crestor group (270 reports of diabetes, vs. 216 in the placebo group; P=0.01). I stated that though one might be concerned about this, these were physician reports without confirmation. When looking at the study lab values in Jupiter, there were no significant differences with respect to the fasting blood glucose level, and the A1c level was pretty much the same (5.9% and 5.8%). I also mentioned that there was no biologically plausible reason to suspect Crestor, or any statin, as a cause of diabetes. In addition, both groups of patients in the JUPITER study had a 41% prevalence of metabolic syndrome, which is a pre-diabetic state. Therefore it should not have been surprising to find an increase in diabetes in both groups.

The meta-analysis looked at Jupiter and several other studies combined and found a 9% increase in the development of diabetes in patients taking statins.
So if this is true, how can you say that statins don't cause diabetes??
1. The 9% value is a relative risk, not an absolute risk. If 1/100 patients taking placebo have a side effect, and 2/100 patients taking a statin have a side effect, the relative risk increase is a doubling or 100%, but the real or absolute risk is only 1%. Drug companies and the media (and unfortunately some authors, as was the case in this study) like to talk about relative risk instead of absolute risk, because it makes there drug/data/news story look better. In this study, the actual increase in diabetes was 0.38% or about one third of 1%. Even though the result is statistically significant (true by scientific standards), the magnitude of the increase is so small that there is a high possibility that this is not true.
2. The study is a meta-analysis, which by design can not prove causation. The only way to really find this out is to do a large, randomized controlled study to specifically look at whether or not statins cause diabetes. This will never happen because 1) given the low absolute risk, the number of patients needed in this study would be tens to hundreds of thousands, which means the study would be ridiculously expensive and 2) with Lipitor going generic, no drug company is going to fund such a study. I spoke of the dangers of meta-analysis many times, specifically in talking about Nissen's meta-analysis of Avandia (multiple posts on this). Even in Nissen's own meta-analysis showing a "40% increase in heart attacks" in patients taking Avandia (relative risk), there was really NO DIFFERENCE in the absolute risk in his study! Of course that didn't stop the press, and millions of patients stopped taking Avandia. Unlike the randomized, controlled trial for statins and diabetes we will never see, there was a randomized controlled trial on cardiac safety and Avandia called the RECORD trial which was presented at the ADA this past summer. It showed NO RISK of heart attacks with Avandia. Don't be surprised if you never hear about this, since the press does not seem to be too fond of reporting good news.
3. As above, there is no biologic reason to believe that a statin medication would cause diabetes.

In other words, what you are seeing in the headlines is statistical garbage, that really doesn't mean anything. An incredibly low absolute risk found in a meta-analysis without a biologic reason to support a connection between statins and diabetes should not be cause for concern.
Post a comment
Write a comment: