LAUSR

Circulation. 2020;142:e82–e84. DOI: 10.1161/CIRCULATIONAHA.120.048260 e82 Amit P. Amin, MD, MSc, MBA Sunil V. Rao, MD Richard G. Bach, MD Jeptha P. Curtis, MD Nihar Desai, MD, MPH Christian McNeely, MD Firas Al-Badarin, MD, MSc John A. House, MS Hemant Kulkarni, MD Frederick A. Masoudi, MD, MSc John A. Spertus, MD, MPH In Response: We appreciate the opportunity to respond to several letters regarding our article.1 Van Belle state “(our) study highlights a signal that should not be ignored.” O’Neill and Chieffo raise concerns about potential misclassification with big data and billing codes, stating “randomized clinical trials are warranted.” We concur and reemphasize our support for randomized clinical trials (RCTs) of mechanical circulatory support (MCS) devices.1 We believe that observational data are complementary to RCTs for postmarketing studies of effectiveness, safety, and cost-effectiveness; they support regulatory decision making for drugs and devices; they quantify the translation of RCTs into practice, and their value should not be overlooked. In this regard, our study1 stands independently of the results of RCTs. O’Neill and Chieffo point out limitations of determining shock severity due to lack of hemometabolic variables. Whereas information on lactate, or SCAI shock score could have enriched our inferences, our data reflect how MCS is used in contemporary practice. Given the lower incidence of cardiac arrest, mechanical ventilation, or cardiogenic shock observed in the Impella group and the trend of more frequent Impella use among patients who were not critically ill, it seems unlikely that hemodynamic/metabolic variables would be worse in Impella patients. Furthermore, the falsification end point analysis—a validated method of examining unmeasured confounding—showed that important unmeasured confounding was unlikely. The consistency of higher adverse events within highversus low-use hospitals and Impella era versus the pre-Impella era, add further strength to our findings. Last, our findings are congruent with analyses from NCDR2 and BMC23 registries. Chieffo et al raise an important point for patients with acute myocardial infarctions, “the timing of MCS and type of support might play an important role.” While those factors may theoretically affect outcomes, no RCTs demonstrate that MCS timing affects outcomes. The best evidence comes from Dhruva,2 where the association of Impella with higher rates of bleeding and mortality were consistent, regardless of device timing. We conducted additional analysis restricted to ST-elevation myocardial infarction (62% patients), the risks of death, bleeding, and acute kidney injury were uniformly higher with Impella versus intra-aortic balloon pump, irrespective of the presence or absence of cardiogenic shock (odds ratio [OR] ranging from 1.5–2.7, P<0.001 for all outcomes). To our knowledge, there is no RCT showing that earlier MCS use, in the setting of acute myocardial infarction shock improves outcomes. The ongoing Door-to-Unload trial (ClinicalTrials.gov. Unique identifier: NCT03947619) may shed further light. Helgestad et al question the exclusion of 828 patients who received both intra-aortic balloon pump and Impella. While “this subgroup might be particularly high-risk,” these patients could not reliably be assigned to either treatment © 2020 American Heart Association, Inc. RESPONSE TO LETTER TO THE EDITOR