Paper
Abstract Background Hospital readmissions following left ventricular assist device (LVAD) remain a frequent comorbidity, associated with decreased quality of life and increased resources utilization. This study sought to determine causes, predictors, and impact on survival of hospitalizations during HeartMate 3 (HM3) support. Methods All patients implanted with HM3 between November 2014 to December 2019 at Columbia University Irving Medical Center were consecutively enrolled in the study. Demographics and clinical characteristics from the index admission and the first outpatient visit were collected and used to estimate 1‐year and 900‐day readmission‐free survival and overall survival. Multivariable analysis was performed for subsequent readmissions. Results Of 182 patients who received a HM3 LVAD, 167 (92%) were discharged after index admission and experienced 407 unplanned readmissions over the median follow up of 727 (interquartile range (IQR): 410.5, 1124.5) days. One‐year and 900‐day mean cumulative number of all‐cause unplanned readmissions was 0.43 (95%CI, 0.36, 0.51) and 1.13 (95%CI, 0.99, 1.29). The most frequent causes of rehospitalizations included major infections (29.3%), bleeding (13.2%), device‐related (12.5%), volume overload (7.1%), and other (28%). One‐year and 900‐day survival free from all‐cause readmission was 38% (95%CI, 31–46%) and 16.6% (95%CI, 10.3–24.4%). One‐year and 900‐day freedom from 2, 3, and ≥4 readmissions were 60.7%, 74%, 74.5% and 26.2%, 33.3%, 41.3%. One‐year and 900‐day survival were unaffected by the number of readmissions and remained >90%. Male sex, ischemic etiology, diabetes, lower serum creatinine, longer duration of index hospitalization, and a history of readmission between discharge and the first outpatient visit were associated with subsequent readmissions. Conclusions Unplanned hospital readmissions after HM3 are common, with infections and bleeding accounting for the majority of readmissions. Irrespective of the number of readmissions, one‐year survival remained unaffected.