Aspirin and statins for prevention of coronary artery disease

Practicing cardiologists often joke about putting “statins” in the drinking water to stem the epidemic of atherosclerotic cardiovascular disease, but curiously, I don’t believe that refers, even in jest, to a cocktail combination of drugs. On the other hand, the problems surrounding polypharmacy do come up frequently in management discussions, particularly in the treatment of elderly patients. Here, the discussion revolves around drugs needed to treat contemporaneous comorbidities, such as arthritis, hypertension, and diabetes. Cardiologists also discuss incremental drug therapies to control angina, stepwise pharmacologic management of heart failure, and combination therapy for the management of acute coronary syndromes.

In his interesting article, Pignone breaks new ground in his careful analysis of the cost-utility of aspirin and statins, alone and in combination, for the primary prevention of coronary artery disease (CAD) events in men. Coronary artery disease events are defined as angina, stroke, myocardial infarction, and death from CAD. Coronary artery disease defined as only hard end points would be expected to yield somewhat different results. The authors report a potential conflict of interest, with consultancies, expert testimony, and grants received from Bayer and Pfizer, manufacturers of products analyzed in this article.

This article led me to review the American Heart Association guidelines for the primary prevention of cardiovascular disease.¹ These guidelines set the goal of low-dose aspirin use for patients at “higher risk” for CAD, especially those with a 10-year risk > 10%. Cholesterol lowering is also recommended if the CAD risk is high and the low-density lipoprotein cholesterol level exceeds a threshold. The guidelines do not explicitly discuss the incremental cost utility of these interventions. This combined cost-utility analysis is the major conceptual advance contained in the article by Pignone and colleagues.

The authors employ a Markov model, which, according to Wikipedia has become a ubiquitous tool in bioinformatics since its introduction in the 1960s. Whether their model accurately reflects what would be discovered in a prospective controlled trial is, of course, at issue. Table 1 in Pignone and colleagues’ original article lists the many base-case estimates constructed,² reminding the reader of the potentially incorrect assumptions inherent in the best theoretical analyses, including variables entered and those omitted. At least for those variables entered, sensitivity analyses were performed to show that the estimates of cost utility were robust enough to withstand major variability in base-case assumptions.

The authors acknowledge that although aspirin was both more effective and less costly than no therapy in the base case, the quality adjusted life years and money saved were both modest compared with no therapy. Benefits increase with the risk of CAD. They also note that errors in estimating the risk of hemorrhagic stroke due to aspirin, particularly at low baseline levels of CAD risk, could invalidate their model. Additional empirical research is needed, particularly in patients at low risk. The authors also acknowledge that they did not model a discontinuation rate for statins related to liver enzyme abnormalities. Such discontinuation would decrease the cost effectiveness of statins. In my experience, discontinuation resulting from subjective complaints of myalgia, particularly when statins are used for primary prevention, is a much larger issue.

In sum, this analytic model suggests that aspirin alone is efficacious and cost effective in men with an average level of risk for CAD. The addition of a statin is more cost effective as risk increases. Just imagine how powerful linked data from Medicare Parts A, B, and D could be in furthering such analyses! Right now, in part related to privacy concerns, drug and clinical data exist in their own isolated silos.