Studies in people and in animal models suggest that atorvastatin, rosuvastatin (Crestor), and simvastatin are associated with elevated risk for diabetes, whereas pravastatin is associated with lower risk. In this population-based cohort study, investigators in Canada used administrative databases to assess incident diabetes between 1997 and 2010 in 471,000 older patients (age, [≥]66) who were newly treated with statins (median age at treatment onset, 73). After adjustment for multiple confounders, with pravastatin as the reference drug, risk for new diabetes was 10%, 18%, and 22% higher in participants who received simvastatin, rosuvastatin, and atorvastatin, respectively. Corresponding numbers needed to harm were 363 (simvastatin), 210 (rosuvastatin), and 172 (atorvastatin). Fluvastatin and lovastatin were not associated with excess risk. Compared with low statin doses (pravastatin, fluvastatin, and lovastatin at all doses; atorvastatin, <20 mg; rosuvastatin, <10 mg; simvastatin, <80 mg), moderate and high statin doses were associated with higher diabetes risk.

Comment: In this study, compared with pravastatin, moderate- and high-potency statins (simvastatin, rosuvastatin, and atorvastatin) were associated with higher risk for incident diabetes. Using low doses of these drugs or low-potency statins (fluvastatin and lovastatin) might mitigate this risk. Notably, these results are biologically plausible: Simvastatin diminishes insulin secretion, whereas pravastatin improves insulin sensitivity and inhibits gluconeogenesis.

An editorialist recommends: "When total cardiovascular risk favours statin treatment, a low dose, low potency agent should be used to begin with." However, some readers will disagree and will argue that, for some high-risk patients, benefits of high-potency statins outweigh the small absolute risk for incident diabetes.