A reference hospital, specialized in cardiology, has standardized clopidogrel, ticagrelor and recently the incorporation of prasugrel was approved with an estimate of fifteen patients with stent implantation monthly. Stent thrombosis is a rare but serious complication and it is recommended that patients be treated with antiplatelet therapy. Considering the existing therapeutic options and the low adherence to treatment, a cost study was requested to discuss a disinvestment proposal.

Perspective of the hospital; Population: patients with myocardial infarction and stent implantation; Intervention: prasugrel; Comparators: clopidogrel and ticagrelor; Outcomes of interest: stent thrombosis and major bleeding; and types of study: meta-analyses and randomized clinical trials. Literature review was performed in the Medline database, via Pubmed and performed meta-analysis. Cost estimates: data collection made in the institution's database.

Thirteen articles were selected; there are no differences in the outcomes for prasugrel and ticagrelor; discontinuation of treatment is the most important risk factor for stent thrombosis and major bleeding is an important predictor of nonadherence to treatment. Quantitative analysis: three clinical trials. All showed a superiority of prasugrel and ticagrelor over clopidogrel for outcome of stent thrombosis (Odds Ratio, OR .60 Confidence Interval, CI 95 percent [.40; .90]), which was committed by the heterogeneity of studies (I2 = 64 percent); and favorable outcome for clopidogrel compared to ticagrelor and prasugrel for the outcome of major bleeding (OR 1.28 CI 95 percent [1.10; 1.50]). Annual cost with treatment of the fifteen patients: clopidogrel = USD5,765.86; prasugrel = USD41,047.48; and ticagrelor = USD44,081.30.

The evidences found support the opening of a disinvestment process and suggest to the managers of the institution a reflection on: strategy to optimize adherence to treatment; and especially in the cost of opportunity with new technologies about 10 times more expensive.