Emerging evidence about the impact of different statins on the risk of diabetes

Emerging evidence about the impact of different statins on the risk of diabetes

Recent evidence has shown that statins can increase the incidence of new-onset diabetes, leading to an “Read the FDA warning for consumers” official warning by the FDA.  This has huge implications for practice, as these drugs are widely used to reduce the risk of cardiovascular disease in patients with and without diabetes.  Selecting the right treatment and balancing the proven benefits of statins whilst minimising the potential for harm presents a difficult dilemma for patients and clinicians.  A recent meta-analysis directly addressed this dilemma.

Clinical question

In patients without diabetes who are taking a statin, what is the effect of different drug-dose combinations on the incidence of new-onset diabetes?


The reviewers conducted a comprehensive literature search for RCTs published between 1994 and October 2012.  They excluded trials of less than 1 year duration with fewer than 1,000 participants.  They conducted blind, independent assessment of the quality of the studies.

17 RCTs comprising 113,394 patients were included in their analysis.  The analyses they carried out were:

  1. high-dose statins versus placebo (9 studies, n = 64,137)
  2. moderate-dose statins versus placebo (11 studies, n = 63,558)
  3. high-dose versus moderate dose statins (3 studies).

Extensive sensitivity analyses and statistical modelling were planned to investigate the robustness of the findings and identify the optimal treatment and dose.


The reviewers estimated the odds ratio for each drug-dose combination compared with placebo.

In the high dose studies, the event rate of new onset diabetes was 7.28% in the treatment group compared with 7.09% in the control group.  Meta-analysis found that rosuvastatin at a 20mg dose had the highest odds ratio compared with placebo (1.25;  95% CI 0.75-2.01) whilst lovastatin 40mg  had the lowest (0.97; 95% CI 0.58-1.61).

In the moderate-dose studies, the event rate was 8.18% in the treatment group compared with 7.95% in the control patients.  Meta-analysis again found that rosuvastatin 10mg had the highest odds ratio (1.10;  95% CI 0.78-1.58) whilst pravastatin had the lowest (0.99; 95% CI 0.68-1.41)

The reviewers’ networked meta-analysis also enabled them to directly compare the performance of each drug at each dose.


The reviewers concluded that:

  1. there was a gradient for the risk for new-onset diabetes across different types and doses of statins,
  2. pravastatin therapy was numerically associated with the lowest risk,
  3. high-dose pravastatin therapy provided the most robust safety profile compared with the other high-dose statins.

These findings need to be interpreted with some caution.  As shown above, the reviewers reported broad confidence intervals around the estimates of increased risk.   These broad confidence intervals are as we would expect for such a small absolute difference in risk.  However, they claim that the findings were robust across various sensitivity analyses, suggesting that they are likely to be correct.  Nevertheless, the findings need to be confirmed in direct comparisons between drugs.

The reviewers investigated the effect of body mass index and cholesterol reduction and concluded that these factors did not influence the risk of developing diabetes.


  • “High dose” was:  atorvastatin 80 mg, lovastatin 20 to 40 mg, pravastatin 40mg, rosuvastatin 20 mg and simvastatin 40 mg.
  • “Moderate dose” was:  atorvastatin 10 mg, pravastatin 10-20 mg, rosuvastatin 10 mg.
  • It’s not clear whether these definitions were made in advance of conducting the search.
  • There were important differences between the studies in how the measured new-onset diabetes.
  • No detailed information is provided about the quality of the studies.
  • No detailed information is provided about the patients in the studies, other than their treatment status and outcome.
  • The event rate for new onset diabetes in the trials was around 7-8%.
  • The risk of harm needs to be balanced against cardiovascular benefit.


Navarese EP, Buffon A, Andreotti F, Kozinski M, Welton N, Fabiszak T, Caputo S, Grzesk G, Kubica A, Swiatkiewicz I, Sukiennik A, Kelm M, De Servi S, Kubica J. Meta-Analysis of Impact of Different Types and Doses of Statins on New-Onset Diabetes Mellitus. Am J Cardiol. 2013 Apr 15;111(8):1123-30. doi: 10.1016/j.amjcard.2012.12.037. Epub 2013 Jan 24

Douglas Badenoch
I am an information scientist with an interest in making knowledge from systematic research more accessible to people who need it. This means you. I've been attempting this in the area of Evidence-Based Health Care since 1995. So far the results have been mixed. For some reason we expected busy clinicians to search databases and appraise papers instead of seeing patients. We also expected publishers to make the research freely available to the people who paid for it.. Ha! Hence The National Elf service.

Follow me here:

Leave a Comment

Your email address will not be published. Required fields are marked *