[Three of the authors respond:]
Complementary roles of observational studies and randomized controlled trials (RCTs) have been recognized.1 Results from RCTs comparing statins head-to-head for long-term cardiovascular prevention are, in fact, very few. The Pravastatin or Atorvastatin Evaluation and Infection Therapy (PROVE-IT) trial2 and the recent Incremental Decrease in Endpoints through Aggressive Lipid Lowering (IDEAL) trial3 compared 2 different statin regimens with a focus on comparing intensive versus moderate cholesterol-lowering therapy, rather than the different statins per se. The latter, comparing high-dose atorvastatin versus usual-dose simvastatin, failed to achieve significance in the primary endpoint of major coronary events.3 Compared to subjects enrolled in clinical trials, the present observational study4 evaluated the effectiveness of statins in all patients ≥ 65 year of age with a diverse risk profile. Thus, our study provided evidence in real-world practice.
Our study was not an equivalence trial but a study of the effectiveness of the different statins prescribed to the population at large. Posterior power calculation is theoretically less meaningful in the observational study setting, where patients from the 3 most populated provinces in Canada were included.4 The confidence intervals around the point estimate of 1.0 we observed were quite narrow and suggested a class effect of statins.
Missing patient cholesterol information represents a limitation in the study, especially for the study of statin dose effect. However, there is no obvious reason to believe that cholesterol levels were significantly different across 5 statin groups. In our study population, the median dose used across statins was comparable, and in our analyses we adjusted for dose equivalence. There is a possibility that switching to atorvastatin caused „contamination.” However, as reported in our sensitivity analysis, we found that patients who switched to atorvastatin or to other statins had similar risk profiles (as measured by similar rates of cardiac outcomes and medication use). Additionally, similar results were obtained when we kept patients who switched in the analysis (intention to treat) or removed them. Thus we believe that the clinical risk was similar across groups.
We agree that higher doses of statins should be used to attain the lowest low-density lipoprotein target possible, regardless of which statin is prescribed.5
Footnotes
Competing interests: None declared.