Model may guide statin treatment in diabetes patients

May 17 2016

By Eleanor McDermid

Researchers have developed a model to predict the benefits of statin treatment in individual patients with Type 2 diabetes, but experts question whether individualised treatment is always the best choice.

The team used data from three major statin trials to develop a predictive model based on eight clinical variables: age, gender, smoking status, blood pressure, history of cardiovascular disease, statin treatment and levels of non-high-density-lipoprotein cholesterol and glucose.

The model, which can predict a patient's absolute risk of cardiovascular events and the risk reduction provided by statin treatment, is based on outcomes of around 9500 patients with Type 2 diabetes from the lipid-lowering arms of the ASCOT and ALLHAT trials, and from the CARDS study.

Using the model, the median estimated 10-year risk of major cardiovascular events among patients not given a statin ranged from 15% to 21%, depending on the study. The median absolute risk reduction provided by 10 years of statin treatment was 3.2%.

Frank Visseren (University Medical Centre Utrecht, the Netherlands) and study co-authors found that the prediction model had moderate ability to discriminate between patients who did and did not have cardiovascular events. It tended to underestimate the risk of lower-risk patients and overestimate that of higher-risk patients.

In clinical practice, the usefulness of the model would vary according to the physician's willingness to prescribe statins, the team reports in Circulation: Cardiovascular Quality and Outcomes.

For example, if a physician routinely prescribed statins to all diabetic patients, then the model would be of no use, but if a physician reserved statin treatment for patients with a 10% risk of cardiovascular events over 10 years according to the model, then 13% of patients would be spared treatment.

The researchers note that a prediction model is no better than blanket treatment among high-risk patients, because they will benefit equally from either strategy. "Therefore, the additional value of a prediction-based strategy lies in selective treatment of patients in the lower risk groups", they say.

But in a linked editorial, Suzanne Arnold and Mikhail Kosiborod, from Saint Luke's Mid America Heart Institute and University of Missouri-Kansas City, USA, question whether personalised medicine is advisable in this patient group.

The entire benefit of the strategy would come from avoiding side effects and costs, they say, but this scenario involves a drug that is available in generic form and has a low risk of adverse effects. They also note the "disconnect" between guideline recommendations and clinical practice and ask if introducing more complexity, in the form of personalised treatment, would increase or reduce appropriate treatment of patients.

"When the harm in treating a small proportion of additional patients is minimal, it is hard to justify adding further complexity to the current treatment paradigm", the editorialists conclude.

Licensed from medwireNews with permission from Springer Healthcare Ltd. ©Springer Healthcare Ltd. All rights reserved. Neither of these parties endorse or recommend any commercial products, services, or equipment.