医学
冲程(发动机)
危险系数
心室辅助装置
单变量分析
比例危险模型
内科学
心脏病学
外科
目的地治疗
置信区间
审查(临床试验)
心力衰竭
多元分析
病理
工程类
机械工程
作者
Yuting Chiang,Daniel Cox,Jacob N. Schroder,Mani A. Daneshmand,Laura Blue,Chetan B. Patel,Adam D. DeVore,Muath Bishawi,Carmelo A. Milano
摘要
Stroke remains a significant complication of left ventricular assist device (LVAD) therapy. We performed a single-center retrospective study evaluating patients undergoing first-time HeartWare HVAD (Medtronic Inc) or HeartMate 3 (Abbott Inc) implantation from September 2009-February 2018.Exclusion criteria were age <18 and preoperative ECMO. The primary endpoint was stroke-free survival. Stroke was defined as new neurological deficits persisting >24 hours with corresponding radiographic evidence. Risk factors evaluated included demographics, medical comorbidities, heart failure etiology, LVAD indication, INTERMACS profile, and device type. Univariate predictors (P < .15) and variables clinically suspected to raise stroke risk were entered in a multivariate hazard regression model, specified using backward selection of covariates and accounting for competing risks of transplant/LVAD exchange.A total of 163 HVAD and 84 HM3 patients were analyzed. Median follow up (until death, censoring for transplant/LVAD removal, or end of follow up) was 1.2 years in HVAD patients and 1.4 years in HM3 patients. Stroke occurred in 24 HVAD patients (15 ischemic, 9 hemorrhagic) and 6 HM3 patients (4 ischemic, 2 hemorrhagic). One-year stroke-free survival was 76.8% for HVAD and 84.3% for HM3. Thirty-day mortality following stroke was 41.7% for HVAD and 66.7% for HM3; 54.2% of HVAD strokes were disabling compared to 83.3% of HM3 strokes. Age, LVAD indication, and device type were associated (P < .15) with stroke on univariate analysis. On multivariate analysis, the HVAD was associated with significantly higher stroke risk (hazard ratio, 2.57; 95% confidence interval, 1.02-6.44; P = .045).Different LVAD models appear to be associated with significantly different stroke risks.
科研通智能强力驱动
Strongly Powered by AbleSci AI