Statins: Diabetes Risk Outweighed by CV Benefit

November 16, 2012

LOS ANGELES— "Although it does appear that statins may increase the [chance] of diabetes in high-risk patients, this should not stop us from using these drugs, as this small risk is outweighed by the substantial reduction in cardiovascular events." That was the conclusion of Dr Barton Duell (Oregon Health & Science University, Portland), who was the discussant of five new studies on this issue reported at last week's American Heart Association 2012 Scientific Sessions.

Presenter of one of the studies, Dr David Waters (San Francisco General Hospital, CA), agreed with this view. He pointed out that the risk of diabetes vs cardiovascular benefits of statins in patients with prediabetes had been referred to as a "double-edged sword." But noting that cardiovascular events were much more serious than an increased risk of diabetes, Waters commented: "One edge of the sword is an awful lot sharper than the other."

One edge of the sword is an awful lot sharper than the other.

Summarizing the new results presented at the AHA session, Duell noted that all five studies showed increased incidence of diabetes in statin users; the absolute increase in risk is low and appears to be proportional to the pretreatment risk of diabetes and the statin dose; and diuretics, beta blockers, and antidepressants may synergistically increase the diabetes risk seen with statins.

As background, Duell reviewed two published meta-analyses:

  • A review of 13 trials with a total of 91 140 participants showed a 9% increased risk for incident diabetes with statins vs placebo [1]. "This translates into one case of diabetes per 255 subjects over four years (0.4%), so an increased risk of 1% per decade. This is a very low risk," Duell said.

  • A meta-analysis of five trials of high-dose vs low-dose statins, with a total of 32 753 participants, which showed a 12% increased risk for incident diabetes over a 2-5 year follow-up [2]. "This works out to two cases of diabetes per 1000 patient-years (0.2% per year, 2% over a decade). This is still very low in comparison with cardiovascular benefits."

Duell added the mechanism behind the diabetes risk is unclear. "This is likely to be multifactorial and perhaps varies from one population to another." He said it is important that people not be discouraged from using statins because of this risk. "These really are excellent drugs, but new strategies are needed to help reduce the risk of diabetes in statin users, such as increased use of bile-acid sequestrants, reduced use of niacin, and better glucose monitoring," he added.

Risk vs Benefit in TNT and IDEAL

In his presentation, Waters reported an analysis of diabetes risk vs cardiovascular event reduction in two large studies of high-dose vs low-/moderate-dose statins: TNT and IDEAL.

Combining these two trials gave a total of 15 056 patients (after those with diabetes were excluded), who were divided into two groups: those at low risk of new-onset diabetes (8825 patients with zero to one risk factor), and those at high risk (6231 patients with two to four risk factors).

Results showed no difference in the development of new-onset diabetes with high-dose vs low-/moderate-dose statin in the low-risk population, but a 24% increase with the high-dose statin in high-risk patients. However, cardiovascular events were significantly reduced with the high-dose statin in both risk groups.

New-Onset Diabetes and Cardiovascular Events: Patients at Low Risk of Diabetes

Outcome Atorvastatin 80 mg (%) Atorvastatin 10 mg/simvastatin 20–40 mg (%) HR (95% CI) p
New-onset diabetes 3.2 3.3 0.97 (0.77–1.22) 0.773
CV events 8.5 9.8 0.87 (0.76–0.99) 0.042

New-Onset Diabetes and Cardiovascular Events: Patients at High Risk of Diabetes

Outcome Atorvastatin 80 mg (%) Atorvastatin 10 mg/simvastatin 20–40 mg (%) HR (95% CI) p
New-onset diabetes 14.3 11.9 1.24 (1.08–1.42) 0.003
CV events 10.1 12.0 0.82 (0.71–0.96) 0.011

Asked about "net clinical benefit," Waters said this was difficult to work out, as "new-onset diabetes really does not compare with cardiovascular events, which were defined as cardiovascular death, MI, or stroke."

He added: "New-onset diabetes may count as one stroke against, but dropping dead from a cardiovascular event could be 100 strokes against. If you had to choose between one of these events, you are going to choose the diabetes."

Similar Results in Taiwanese Study

Another study looking at risk vs benefits was presented by Dr Kang-Ling Wang (Taipei Veterans General Hospital, Taiwan).

Using data from Taiwan's national health insurance program, which covers 99.5% of the population, the study looked at new diabetes cases according to statin use in patients with abnormal glucose tolerance not receiving diabetes medications. Propensity scoring was used to select 3069 matched pairs of patients, one of whom was taking a statin and the other not.

Results showed a higher incidence of new-onset diabetes in statin users, but this was outweighed by a reduced all-cause mortality and morbidity (all hospitalizations and emergency-room visits).

New Onset Diabetes vs Mortality/Morbidity Benefits With Statins

End point Statin users (%) Nonusers (%)
New-onset diabetes 8.5 7.1
All-cause mortality 1.4 2.4
All hospitalizations and ER visits 53 60

Diabetes Risk Also in Asian Population

Another study documented "a clear association" with statin therapy and new-onset diabetes in Asian patients, which the author, Dr Sang-Ho Park (Cheonan Hospital, Republic of Korea), said had not been definitely shown before.

Park and colleagues used electronic medical records to analyze the occurrence of new-onset diabetes in 3099 statin patients and 9300 controls. Results showed after a median follow-up of two years, the incidence of new-onset diabetes was 2.75% in the whole population, but was higher in the statin group (4.8%) than the controls (2.1%). The increase remained significant after adjustment for many covariates and propensity score in Cox proportional hazard models.

Risk of New-Onset Diabetes in Statin Users vs Nonusers

Analysis type HR (95% CI) p
Unadjusted analysis 1.49 (1.21–1.85) <0.001
Adjusted for covariates and propensity score 1.34 (1.01–1.78) 0.041

Other results from multivariate analysis suggested that age and beta-blocker and diuretics use might also be potential independent predictors for new-onset diabetes.

NAVIGATOR: Statins and Diuretics Implicated

Both statins and diuretics were found to be associated with increased risk of new-onset diabetes, and beta blockers showed a borderline effect, in an analysis of the NAVIGATOR trial. The trial had compared nateglinide (Starlix, Novartis) and valsartan (Diovan, Novartis) in the prevention of diabetes and cardiovascular outcomes in 9518 treatment-naive patients with impaired glucose tolerance and at least one other cardiovascular risk factor. Fasting plasma glucose was measured regularly and oral glucose-tolerance tests were performed annually.

In the current analysis, presented by Dr Lan Shen (Duke University, Durham, NC), the risk of developing new-onset diabetes in association with starting beta blockers, diuretics, or statins was calculated in a Cox proportional hazard model adjusting for confounders that changed over time.

Results were presented as a single hazard ratio that represents the difference in rate of progression to new-onset diabetes between patients receiving a particular medication class and those not.

Hazard Ratio for Developing Diabetes With Diuretics, Statins, and Beta Blockers

Type of medication HR (95% CI)
Diuretics 1.35 (1.12–1.61)
Statins 1.30 (1.09–1.55)
Beta blockers 1.19 (0.97–1.46)

In comparison, no increased risk of diabetes was found with calcium blockers.

WHI: Risk Escalates With Combinations of Certain Drugs

Finally, Dr Rhonda Cooper-DeHoff (University of Florida Colleges of Pharmacy and Medicine, Gainesville) presented an analysis of the Women’s Health Initiative(WHI) suggesting strong trends for increased new-onset diabetes risk with antidepressants, thiazide diuretics, and beta blockers, as well as statins. The risk was shown to increase substantially with each class of these drugs used in combination.

The current analysis involved data on 139 554 postmenopausal women without diabetes at baseline who were followed for 11 years and asked annually about new diabetes diagnosis or new use of diabetes medications.

Risk of New-Onset Diabetes With Increasing Number of Drug Classes Used

Drug classes being used (n) Incident diabetes (%) HR (95% CI)
0 (n=108 096) 7.5 1
1 (n=26 2550 11 1.13 (1.07–1.20)
2 (n=4768) 14 1.31 (1.15–1.48)
3–4 (n=435) 16 1.80 (1.18–2.75)

In terms of individual drug-class risks, statins showed the highest risk, with a hazard ratio of about 1.5. Antidepressants, thiazides, and beta blockers were associated with a smaller risk, with a hazard ratio of around 1.2.

Duell reports being on a speaker's bureau for Merck; consulting for or serving on an advisory board for Pfizer, Genzyme, Merck, Aegerion, and Amarin; and receiving research grants from Genzyme, Pfizer, Bristol-Meyers Squibb, and Cerenis. Waters reports honoraria from Pfizer, Genentech, and Roche and consulting for or serving on an advisory board for Anthera, Pfizer, Sanofi, Aegerion, Cerenis, Genentech, Roche, and Servier. Wang, Shen, and Cooper-DeHoff report no conflicts of interest.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.