Fatal Neurologic Dysfunction During Continuous-Flow Left Ventricular Assist Device Support

• Published in: The Annals of thoracic surgery Vol. 107; no. 4; pp. 1132 - 1138
• Main Authors: Kawabori, Masashi, Conyer, Ryan T., Kurihara, Chitaru, Sugiura, Tadahisa, Civitello, Andrew B., Frazier, O.H., Morgan, Jeffrey A.
• Format: Journal Article
• Language: English
• Published: Netherlands Elsevier Inc 01.04.2019
• Subjects: Aged; Cause of Death; Databases, Factual; Female; Heart Failure - diagnosis; Heart Failure - mortality; Heart Failure - surgery; Heart-Assist Devices - adverse effects; Humans; Logistic Models; Male; Middle Aged; Multivariate Analysis; Nervous System Diseases - etiology; Nervous System Diseases - mortality; Nervous System Diseases - physiopathology; Prognosis; Proportional Hazards Models; Retrospective Studies; Risk Assessment; Severity of Illness Index; Survival Analysis; Treatment Outcome
• ISSN: 0003-4975; 1552-6259
• DOI: 10.1016/j.athoracsur.2018.10.012

During continuous-flow left ventricular assist device (CF-LVAD) support, neurologic dysfunction (ND) is a common complication and can be fatal. Few reports provide detailed data on neurologic mortality in such patients. Therefore, we examined ND-related mortality during CF-LVAD support. Between November 2003 and March 2016, 526 patients underwent implantation of a CF-LVAD (403 HeartMate II [Thoratec, Pleasanton, CA] and 123 HVAD [HeartWare International, Framingham, MA]) at our center. We categorized ND as hemorrhagic or ischemic and recorded resulting deaths. Records were reviewed to determine preoperative demographics, perioperative variables, prevalence and causes of postimplantation ND, duration of support until ND, time from ND to death, laboratory data and medications at the time of ND, post-ND treatment procedures, and causes of hemorrhagic ND. We also performed Cox multivariable logistic regression analysis to identify predictors of ND-related mortality by calculating odds ratios and confidence intervals. Neurologic dysfunction occurred in 141 patients (26.8%), 48 (9.1%) of whom subsequently died. Median duration of left ventricular assist device support before ND was 230 days (range, 3 to 2,422), and median time from ND to death was 3.5 days (range, 0 to 55). Parenchymal hemorrhage was the most frequent cause of early conversion (76.7%). In the Cox multivariable regression analysis, predictors of fatal ND were age, ischemic cause of heart failure, history of stroke, and longer intraoperative aortic cross-clamp time. Our study elucidates the characteristics and risk factors of patients who died of ND during CF-LVAD support. Further studies are required to find ways to decrease the incidence of fatal ND during CF-LVAD support.