Application of Modern Methods to Predict Risk and Estimate Treatment Effectiveness in Cardiogenic Shock

PROJECT SUMMARY/ABSTRACT Acute myocardial infarction complicated by cardiogenic shock (AMI-CS) is a major driver of global morbidity and mortality. Due to the challenges of conducting randomized trials for these acutely ill patients, the evidence for mechanical circulatory support devices (MCS) commonly used to treat AMI-CS is limited, resulting in a lack of consensus on optimal treatment strategies and consequent wide variation in care practices. Thus, improving our understanding of whether and how recently published randomized trials of MCS should influence contemporary clinical practice, as well as applying rigorous observational approaches to understand the causal effects of MCS devices in purposefully designed shock registries would fill critical gaps in our current understanding of AMI-CS management. The overarching goal of this grant is to apply state-of-the-science methods to a combination of clinical trial and observational cohorts to identify the optimal use of MCS devices in AMI-CS. In Aim 1, we will append individual patient data from recent landmark clinical trials of MCS devices to several contemporary cardiogenic shock registries including the Cardiogenic Shock Working Group (CSWG) Registry, the American Heart Association’s Cardiogenic Shock Registry, and the American College of Cardiology’s CathPCI Registry. To these appended datasets, we will apply novel transportability methods to first evaluate the applicability of the trials’ results to contemporary U.S. practice, and then extend inferences from the trial to new clinically relevant target populations. In Aim 2, we will evaluate the safety and effectiveness of MCS device treatment for AMI-CS in the CSWG and AHA Shock Registries through application of the target trial framework. We will take advantage of the collection of detailed time-stamped hemodynamic parameters, laboratory values, and device treatments in these unique registries, and apply grace-period cloning approaches that can account for dynamic changes in patient status to examine the effects of alternative treatment approaches in AMI-CS. In Aim 3, we will apply traditional and machine-learning approaches to assess heterogeneity of treatment effects of MCS use among patients with AMI-CS and develop tools to create individualized estimates of treatment benefit and harm. “Effect score” models will be developed and validated within the CSWG and AHA Shock, and applied to randomized trial data to identify heterogeneous treatment responses. This research has the potential to fill critical gaps in our understanding of the benefits and risks associated with mechanical circulatory support devices for the treatment of AMI-CS, leveraging the application of innovative methodological approaches to multiple powerful new datasets. The insights gained from this knowledge may lead to improvements in the care and outcomes of individuals with this highly morbid cardiovascular condition for which evidence remains scant and randomized clinical trials are unlikely to be conducted soon.