Abstract 14662: Treatment Combinations for Elderly Patients and Those With Comorbidities After an Acute Myocardial Infarction

IntroductionFollowing an acute myocardial infarction (AMI), the elderly and patients with comorbidities often do not receive guideline-recommended treatment. Because such patients were generally excluded from randomized, controlled trials (RCTs), the consequences of undertreatment are unknown. Thus, we used an artificial neural network (ANN) to simultaneously examine 128 different treatment combinations among patients after an AMI.HypothesisPost-AMI guideline-recommended treatments are effective for elderly patients and patients with comorbidities not considered in RCTs.MethodsUsing a sample of 124,031 Medicare beneficiaries with a previous AMI, we used 3 categories of predictors (comorbidities, procedures and demographics) and 7 medications (ACE inhibitors/ARBs, β-blockers, calcium channel blockers, clopidogrel, diuretics, nitrates and statins) in an ANN to predict one-year mortality. Separate models were optimized on distinct training and validation sets, and optimized treatment regimens were calculated for query cases using the training model. The recommended regimen and patient features were then used in a validation model to estimate the probability of survival under the recommended treatment. We used 10-fold cross-validation and estimated ANNs for the complete sample, those older than 80 years, patients with diabetes (DM), heart failure (HF), chronic kidney disease (CKD) and chronic obstructive pulmonary disease (COPD) separately.ResultsApproximately 15% of the full sample had not filled a prescription for any of the medications we considered. Among patients with DM and CKD, 8% and 14% of patients had not filled a prescription. Using our ANN, we found that, among the 128 different possible treatment combinations we examined, a combination of ACE inhibitors/ARBs, β-blockers and statins or a combination of β- blockers and statins were most commonly recommended, on the basis of reducing one year mortality. However, only 2% of patients in each group were actually taking these recommended medications.ConclusionsOur ANN demonstrated that both elderly patients and patients with comorbidities benefit from guideline-recommended treatments.