Association between diabetes (D) and statins (S) is well demonstrated. However, results are conflicting according to the different S with for example association with atorvastatin or rosuvastatin but not with others in the Takker's meta-analysis. Little is known about the risk in real world. We investigated this safety signal in a pharmacovigilance database.

Using Vigibase^®, the WHO Global Individual Case Safety Report (ICSR) database with over 14 million reports, we performed a disproportionality analysis including ICSRs until March 2017 only with age (≥18 years) and gender known. Cases were ICSRs with D (defined as reports registered under the HLT MedDRA terms “D mellitus” or “hyperglycemic conditions”) and non-cases all other ICSRs. Drug exposure to S (atorvastatin, cerivastatin, fluvastatin, lovastatin, pitavastatin, pravastatin, rosuvastatin, simvastatin) was identified using the ATC code C10AA, i.e. HMG CoA reductase inhibitors defined as “suspected” or “concomitant”. Strength of the link was quantified by crude Reporting Odds-Ratio (ROR) with 95% CI. P threshold was 0.05.

Among the 9,372,588 ICSRs, 13,071 involved S plus D. Most of the patients were women (69.3%) with age mainly between 45 and 64 years (57.0%). The 3 most frequently involved S were atorvastatin (62.4%) followed by simvastatin (20.4%) and rosuvastatin (11.7%). A signal was found between D and S in general [ROR=4.29 (4.21–4.38)] and all other S in particular. The higher ROR value was found with atorvastatin [7.27 (7.10–7.45)] followed by rosuvastatin [3.02 (2.86–3.17)] and the lowest with cerivastatin [1.38 (1.07–1.78)]. Sensibility analyses investigating ROR with one S after exclusion of cases registered with the other S show similar results.

The present study found a clear signal of S for all S showing that the diabetogenic risk of S is a class property. This is the first study investigating the risk of D according to the different S in the context of real world.