Statin use before diabetes diagnosis and risk of microvascular disease: a nationwide nested matched study

Abstract

Background: The role of statins in the development of microvascular disease in patients with diabetes is unknown. We tested the hypothesis that statin use increases the risk of diabetic retinopathy, diabetic neuropathy, diabetic nephropathy, and gangrene of the foot in individuals with diabetes.

Methods: We identified all patients living in Denmark who were aged 40 years or older and were diagnosed with incident diabetes between Jan 1, 1996, and Dec 31, 2009. We obtained patients' data from the Danish Patient Registry and information on drug use from the Danish Registry of Medicinal Product Statistics. We randomly selected 15,679 individuals from the database who had used statins regularly until their diagnosis of diabetes (statin users) and matched them in a 1:3 ratio with 47,037 individuals who had never used statins before diagnosis (non-statin users). Our primary outcome was to compare the cumulative incidence of diabetic retinopathy, diabetic neuropathy, diabetic nephropathy, or gangrene of the foot in statin users versus non-statin users. We analysed data with Cox regression models, adjusted for covariates including sex, age at diabetes diagnosis, and method of diabetes diagnosis. To address potential biases between statin users and non-statin users, we made adjustments to our analysis with a propensity score and with other factors. Median follow-up was 2·7 years (range 0-13).

Findings: During 215,725 person-years of follow-up, 2866 patients developed diabetic retinopathy, 1406 developed diabetic neuropathy, 1248 developed diabetic nephropathy, and 2392 developed gangrene of the foot. Compared with non-statin users, statin users had a lower cumulative incidence of diabetic retinopathy (hazard ratio 0·60, 95% CI 0·54-0·66; p<0·0001), diabetic neuropathy (0·66, 0·57-0·75; p<0·0001), and gangrene of the foot (0·88, 0·80-0·97; p=0·010), but not diabetic nephropathy (0·97, 0·85-1·10; p=0·62). These results were similar after adjusting for the competing risk of death, after matching for a propensity score, after adjusting for visits to a family doctor, and by stratification on covariates. The corresponding multivariable adjusted hazard ratio for risk of diabetes in the total population was 1·17 (95% CI 1·14-1·21; p<0·0001).

Interpretation: Use of statins before diagnosis of incident diabetes was not associated with an increased risk of microvascular disease. Whether statins are protective against some forms of microvascular disease-a possibility raised by these data-will need to be addressed in other studies similar to ours, in mendelian randomisation studies, and preferably in randomised controlled trials.

Funding: Herlev Hospital, Copenhagen University Hospital.