After adjusting for changes in demographics and comorbidities over time (red squares), we found that long-term mortality after myocardial infarction (MI) decreased significantly from 1995 to 2004, with a 3% reduction in the risk of death each year. Next, we adjusted for the use of statins, beta-blockers, angiotension-converting enzyme inhibitors/angiotensin-II receptor blockers, and nonaspirin antiplatelet drugs after discharge, which eliminated the association between time trend and mortality (blue circles). This means that the change in mortality may be primarily attributable to increased use of recommended drugs. In contrast, adjusting for the use of coronary interventions, but not for the study drugs (green triangles), diminished the time trend slightly but did not eliminate the effect completely Hazard ratios for calendar years were estimated from the multivariate Cox proportional hazards regression model that includes indicators for each calendar year, age, gender, race, comorbidities (previous MI, non-MI coronary artery diseases, heart failure, cerebrovascular diseases, peripheral vascular diseases, atrial fibrillation, aortic aneurysms, diabetes, hypertension, chronic pulmonary diseases, peptic ulcer diseases, liver diseases, chronic kidney diseases, dialysis, malignancy, rheumatoid arthritis, osteoarthritis, human immunodeficiency virus infection, dementia, depression, other mental disorders, obesity, and alcohol abuse) and health service use measures (number of visits, number of generic medications prescribed, and any hospitalization). The p value for the trend was estimated from the same model replacing the indicators for each calendar year by 1 continuous variable for calendar year. CV = cardiovascular.