A, Left ventricular assist device (LVAD) recipients with ESRD. B, Left ventricular assist device (LVAD) recipients without ESRD. HMO indicates health maintenance organization; ICD-9, International Classification of Diseases, Ninth Revision (ICD-9); USRD, United States Renal Data System.
ESRD indicates end-stage renal disease; LVAD, left ventricular assist device.
eTable. Adjusted association of end-stage renal disease (ESRD) with time to death among left ventricular assist device (LVAD) recipients
Customize your JAMA Network experience by selecting one or more topics from the list below.
Bansal N, Hailpern SM, Katz R, et al. Outcomes Associated With Left Ventricular Assist Devices Among Recipients With and Without End-stage Renal Disease. JAMA Intern Med. 2018;178(2):204–209. doi:10.1001/jamainternmed.2017.4831
Do outcomes after placement of a left ventricular assist device (LVAD) for the treatment of advanced heart failure in patients with end-stage renal disease (ESRD) differ compared with those of patients without ESRD?
In this 11-year study of a nationally representative patient population of LVAD recipients, 81.9% of patients with ESRD prior to LVAD implantation died during follow-up (vs 36.4% among patients without ESRD) and the median time to death was 16 days after LVAD placement for patients with ESRD (vs 2125 days for those without ESRD).
Left ventricular assist device recipients with preexisting ESRD have an extremely poor prognosis.
Left ventricular assist devices (LVADs) are widely used both as a bridge to heart transplant and as destination therapy in advanced heart failure. Although heart failure is common in patients with end-stage renal disease (ESRD), little is known about outcomes after LVAD implantation in this population.
To determine the utilization of and outcomes associated with LVADs in nationally representative cohorts of patients with and without ESRD.
Design, Setting and Participants
We described LVAD utilization and outcomes among Medicare beneficiaries after ESRD onset (defined as having received maintenance dialysis or a kidney transplant) from 2003 to 2013 based on Medicare claims linked to data from the United States Renal Data System (USRDS), a national registry for ESRD. We compared Medicare beneficiaries with ESRD to a 5% sample of Medicare beneficiaries without ESRD.
ESRD (vs no ESRD) among patients who underwent LVAD placement.
Main Outcomes and Measures
The primary outcome was survival after LVAD placement.
Among the patients with ESRD, the mean age was 58.4 (12.1) years and 62.0% (96) were male. Among those without ESRD, the mean age was 62.2 (12.6) years and 75.1% (196) were male. From 2003 to 2013, 155 Medicare beneficiaries with ESRD (median and interquartile range [IQR] days from ESRD onset to LVAD placement were 1655 days [453-3050 days]) and 261 beneficiaries without ESRD in the Medicare 5% sample received an LVAD. During a median follow-up of 762 days (IQR, 92-3850 days), 127 patients (81.9%) with and 95 (36.4%) without ESRD died. more than half of patients with ESRD (80 [51.6%]) compared with 11 (4%) of those without ESRD died during the index hospitalization. The median time to death was 16 days (IQR 2-447 days) for patients with ESRD compared with 2125 days (IQR, 565-3850 days) for those without ESRD. With adjustment for demographics, comorbidity and time period, patients with ESRD had a markedly increased adjusted risk of death (hazard ratio, 36.3; 95% CI, 15.6-84.5), especially in the first 60 days after LVAD placement.
Conclusions and Relevance
Patients with ESRD at the time of LVAD placement had an extremely poor prognosis, with most surviving for less than 3 weeks. This information may be crucial in supporting shared decision-making around treatments for advanced heart failure for patients with ESRD.
Left ventricular assist devices (LVADs) are used as both a bridge to heart transplant and destination therapy for patients with advanced heart failure,1 and have been shown to prolong survival and improve quality of life.1-3 The Randomized Evaluation of Mechanical Assistance for the Treatment of Congestive Heart Failure (REMATCH) trial demonstrated that LVAD placement as destination therapy improved survival compared with medical therapy alone.1 In 2008—fueled by advances in LVAD technology, scarcity of donor hearts,1 and a growing number of patients who are ineligible for heart transplantation due to comorbidity4—the US Food and Drug Administration approved the use of LVADs as destination therapy for advanced heart failure, paving the way for expansion in the use of these devices.5-7
The prevalence of heart failure is approximately 40% among patients with end-stage renal disease (ESRD), and 37% die from heart failure.8 Few studies have evaluated outcomes among LVAD recipients with kidney disease, and these have largely focused on patients with earlier stages of kidney disease not receiving dialysis and those receiving acute dialysis for acute kidney injury, rather than those with ESRD.9,10 To address this knowledge gap, we designed a study to compare national trends in utilization and outcomes after LVAD placement among patients with vs without ESRD, defined as having received maintenance dialysis or a kidney transplant, at the time of LVAD placement.
We used the United States Renal Data System (USRDS), a national registry for ESRD, linked to Medicare claims to define a cohort of beneficiaries with ESRD who received their first LVAD between 2003 and 2013. We defined ESRD as having received maintenance dialysis or a kidney transplant for treatment of chronic kidney disease. In the USRDS, a person is identified as having ESRD when a physician certifies the disease on the Center for Medicare and Medicaid Services’ (CMS) Medical Evidence form, or when there is other evidence that the patient has received maintenance dialysis or a kidney transplant. At the time of our study, patients receiving maintenance dialysis for the treatment of acute kidney injury did not appear in the USRDS registry. For comparison, we used the 5% Medicare sample linked to the USRDS registry to identify a nationally representative cohort of Medicare beneficiaries who did not have ESRD at the time of LVAD placement over the same time period. Institutional review board approval was obtained from the University of Washington.
Our primary exposure was LVAD placement based on an International Classification of Diseases, Ninth Revision (ICD-9), code (37.66, 37.62, and 37.65) search of Medicare claims to ascertain the date of each patient’s first LVAD placement.5 There were 467 patients with ESRD in the USRDS registry that had at least 1 LVAD placed between 2003 and 2013 (Figure 1A). Of these, we excluded patients who recovered kidney function prior to LVAD placement, did not have Medicare Parts A and B, or were covered by a health maintenance organization (HMO) during the year before LVAD placement. Because the focus of our study was on patients with ESRD at the time of LVAD placement, we excluded those for whom LVAD placement had preceded ESRD onset, yielding an analytic cohort of 155 LVAD recipients with ESRD at the time of LVAD placement.
Among members of the 5% Medicare sample, we identified 504 patients who received an LVAD between 2003 and 2013 (Figure 1B). We excluded patients with ESRD at the time of LVAD placement and those who did not have Medicare Parts A and B or had HMO coverage during the year before LVAD placement, yielding an analytic cohort of 261 LVAD recipients without ESRD at the time of LVAD placement.
The primary outcome was death after LVAD placement. Date of death was ascertained from the USRDS Standard Analysis Files for patients with ESRD and from the Medicare Denominator File for those without ESRD. Additional outcomes included death during the index hospitalization and receipt of a heart transplant during follow-up (ICD-9 procedure codes 37.5, 37.51, and 33.6). All patients were followed until death or the end of follow-up (December 31, 2013). Outcomes at discrete time periods after LVAD placement are reported for the subset of patients who died on or before December 31, 2012, to allow for at least 1 year of follow-up for all patients.
We ascertained demographic characteristics including age, sex, and race (white, black, other) at the time of LVAD placement. We used ICD-9 diagnostic codes to identify the following comorbid conditions based on inpatient and outpatient Part A and B claims: history of cancer, cirrhosis, coronary artery disease, heart failure, dementia, diabetes, dyslipidemia, chronic obstructive pulmonary disease, hypertension, peripheral arterial disease, and stroke. We estimated burden of comorbidity using the Quan score, a comprehensive index of comorbidity based on ICD-9 codes.11 Time period of LVAD placement was categorized as follows to account for advances in LVAD technology over time: 2003 to 2006, 2007 to 2009, and 2010 to 2013. For patients with ESRD, we included information on modality of renal replacement therapy (hemodialysis, peritoneal dialysis or kidney transplant) at the time of LVAD placement. Medicare claims were used to examine health care utilization during the index hospitalization when the LVAD was placed, including receipt of cardiopulmonary resuscitation, enteral and/or parenteral nutrition, intubation, tracheostomy, placement of an implantable cardioverter defibrillator, placement of a cardiac pacemaker, days in the intensive care unit (ICU) or coronary care unit, length of stay, and Medicare Parts A and B costs.
We examined temporal trends in LVAD placement among patients with and without ESRD using denominator populations of ESRD and non-ESRD Medicare beneficiaries during each time period from the USRDS registry and 5% Medicare sample, respectively. We described patterns of care and outcomes during the index hospitalization using mean (SD) or median (interquartile range [IQR]) values for continuous variables and percentages for categorical variables.
We generated Kaplan-Meier curves to compare survival probability among patients with and without ESRD. We examined the percentage of surviving patients who had received a heart transplant at discrete time points after LVAD placement (at 1, 3, 6, 9, and 12 months).
We used Cox proportional hazard models to examine the association of ESRD (vs no ESRD) with time to death after LVAD placement, adjusting for demographics, comorbid conditions during the year before cohort entry, tertile of Quan score, and time period of LVAD placement. Follow-up time was defined as time from of the date of LVAD placement to death or the end of follow-up (December 31, 2013), whichever came first. We included an interaction term for time (>60 days; P < .001) in final Cox models to accommodate a violation in the proportional hazards assumption identified by examining Kaplan-Meier curves and Schoenfeld residuals.
We also tested for an interaction with time period (2003-2008 and 2009-2013) to determine whether the magnitude of the association of ESRD with time to death after LVAD placement changed over the study period.
From 2002-2013, 155 Medicare beneficiaries with ESRD and 261 beneficiaries without ESRD in the Medicare 5% sample received an LVAD. Most patients (123 [79.3%]) with ESRD were receiving maintenance hemodialysis, 12 (7.7%) were receiving peritoneal dialysis and 15 (9.7%) had a functioning kidney transplant at the time of LVAD placement.
Among the patients with ESRD, the mean age was 58.4 (12.1) years, and 96 (62.0%) were male. Among the non-ESRD patients, the mean age was 62.2 (12.6) years, and 196 (75.1%) were male. Compared with patients without ESRD, those with ESRD were more likely to be black (58 [37.4%] vs 53 [20.3]%) and have a greater burden of comorbidity (Quan score, median [IQR]: 9 [7-10] vs 6 [4-8]) (Table 1).
Incidence rates of LVAD placement per 10 000 person-years for the years 2003 to 2006, 2009 to 2009, and 2010 to 2013 were, respectively, 0.32, 0.47, and 0.45 for patients with ESRD and 0.06, 0.13, and 0.25 for patients without ESRD.
Receipt of intensive procedures and time spent in the ICU during the index hospitalization were similar for LVAD recipients with and without ESRD (Table 2). Over half of patients with ESRD (80 [51.6%]) died during the index hospitalization compared with only 11 (4.2%) of those without ESRD. Among patients who survived the index hospitalization, the median length of stay was slightly longer for those with ESRD (37 days [29-45 days] vs 32 days [21-48 days]). Among patients who survived to hospital discharge, those with ESRD were less likely to be discharged home (38 of 76 [50%] vs 172 [68.5%]) and had a shorter time to readmission (37 days [13-118 days] vs 81 days [26-196 days]) (Table 2).
The median follow-up time was 762 days (IQR, 92-3850 days). Overall, 127 LVAD recipients with ESRD (81.9%) had died by the end of follow-up compared with 95 (36.4%) of those without ESRD. The median time to death for patients with ESRD was 16 days (IQR, 2-447 days) compared with 2125 days (IQR, 565-3850 days) for those without ESRD. Kaplan-Meier curves showed significantly shorter survival among those with vs without ESRD, with the greatest difference occurring in the first 60 days after LVAD placement (Figure 2), during which time 92 patients with ESRD (72.4%) and 6 patients without ESRD (6.3%) died.
In adjusted analyses that included an interaction term for time from LVAD placement (P < .001), ESRD was a dominant risk factor for death after LVAD placement (eTable in the Supplement). The interaction for time period (2003-2008 vs 2008-2013) did not reach statistical significance (P = .06).
Among patients with ESRD, 9 (5.8%) went on to receive a heart transplant compared with 56 of those without ESRD (25%). For patients with at least 1 year of follow-up data available, only 4 cohort patients with ESRD (2.9%) were alive with a heart transplant 1 year after LVAD placement, compared with 29 patients without ESRD (13.0%) (Figure 3).
Among nationally representative cohorts of Medicare beneficiaries who received an LVAD over an 11-year period, mortality rates among patients with ESRD—patients who had received maintenance dialysis or a kidney transplant prior to LVAD placement—were exceptionally high compared with those without ESRD. Less than half of patients with ESRD at the time of LVAD placement survived to hospital discharge, and their median survival was only 16 days, whereas 96% of patients without ESRD survived to discharge and their median survival was almost 6 years. At 1 year after LVAD placement, 75.2% of patients with ESRD had died and only 2.9% were alive with a heart transplant. By comparison, 20.1% of patients without ESRD had died and 13.0% were alive with a heart transplant.
Survival rates among LVAD recipients without ESRD in our study were comparable with those of previous reports.2,3,5,7 The REMATCH trial reported a 1-year survival of 52% among LVAD recipients.1 Among Medicare free-for-service beneficiaries hospitalized for LVAD placement, 30-day mortality was 9% and 1-year mortality was 31% in 2011, representing a substantial decrease since 2004.7
Several prior studies9,10,12 have examined the relationship between kidney function prior to LVAD placement and subsequent outcomes; however, they largely focused on patients with less advanced chronic kidney disease and those receiving acute dialysis for treatment of acute kidney injury.10,12,13 Our study focused on patients with ESRD at the time of LVAD placement (median duration of ESRD was 4 years). In an analysis of 280 participants from the REMATCH trial,14 those with an estimated glomerular filtration rate (eGFR) of less than 34 mL/min/1.73 m2 not receiving dialysis had 2-fold higher odds of in-hospital mortality after LVAD placement compared with those with higher levels of eGFR.14 Among 227 LVAD recipients at a single center, survival was more limited among the 55 patients who were treated with acute dialysis for acute kidney injury at the time of LVAD placement.15 To our knowledge, only 1 study16 may have included patients with ESRD at the time of LVAD placement. This study included 85 patients who were receiving dialysis at the time of LVAD placement but did not specify whether these patients were receiving acute or maintenance dialysis. Patients with advanced kidney disease (including those receiving dialysis) had a 3-month mortality rate of 22%, which was almost 3 times that for other cohort members.16 However, the number of patients receiving dialysis in this study was small, the study did not report outcomes specifically for the subset receiving dialysis, and the study did not distinguish between patients receiving acute vs maintenance dialysis.
Unlike other groups undergoing LVAD placement, patients with ESRD represent a unique population already receiving 1 form of chronic life support.17 Patients receiving maintenance dialysis are more likely than patients with other chronic illness to receive intensive medical care at the end of life.17 The median survival of 16 days after LVAD placement and high in-hospital mortality for patients with ESRD in this study suggests that these procedures are either being offered at a time when patients are actively dying or that they actually precipitate this process. A small subset of patients with ESRD did survive for longer periods of time and did go on to receive a heart transplant, suggesting that some patients with ESRD may derive benefit from LVADs.
Recent studies demonstrate that end-of-life discussions for patients with LVADs and their family members, caregivers, and physicians are often challenging, particularly decisions about whether and when to deactivate these devices toward the end of life.18-28 Our results may help to inform treatment decisions in patients with ESRD and heart failure who are contemplating LVAD placement. These patients should ideally be informed about the very poor prognosis of most patients with ESRD who receive an LVAD. In the context of each patients’ goals and values, this information should then factor into shared decision-making around LVAD placement.21
A strength of our analysis is that we describe outcomes after LVAD placement among nationally representative cohorts. Unlike prior studies of outcomes among LVAD recipients with kidney failure, which we suspect have included mainly patients receiving acute dialysis, we studied Medicare beneficiaries with ESRD. However, these analyses have the following limitations. We relied on diagnostic and procedure codes in administrative data to define and characterize the cohort, and lacked detailed information on the clinical context in which LVADs were placed (including information on the Interagency Registry for Mechanically Assisted Circulatory Support [INTERMACS] profile). We did not have information on the indication for LVAD placement (bridge to transplant or destination therapy), nor did we have information on the type of LVAD that was placed. However, this approach has been used in other studies based on administrative data when manual medical review is not possible.5 We were unable to conduct separate analyses among subgroups within the ESRD population owing to the very small number of patients in the cohort.
Left ventricular assist device recipients with ESRD at the time of placement have a very poor prognosis, with less than half surviving to hospital discharge and most surviving less than 3 weeks. This information may be crucial in supporting shared decision-making around treatments for advanced heart failure for patients with ESRD.
Corresponding Author: Nisha Bansal, MD, MAS, Kidney Research Institute, University of Washington, 908 Jefferson St, Third Floor, Seattle, WA 98104 (firstname.lastname@example.org).
Accepted for Publication: July 24, 2017.
Published Online: December 18, 2017. doi:10.1001/jamainternmed.2017.4831
Author Contributions: Dr Hailpern had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Bansal, Hall, O'Hare.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Bansal, Hailpern.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Bansal, Hailpern, O'Hare.
Obtained funding: Hall, Kurella Tamura, O'Hare.
Administrative, technical, or material support: Kurella Tamura, Kreuter, O'Hare.
Study supervision: Katz, Hall, O'Hare.
Conflict of Interest Disclosures: None reported.
Funding/Support: This work was supported by the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) (1U01DK1021501; principal investigators: O’Hare and Kurella Tamura).
Role of the Funder/Sponsor: The NIDDK had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The data reported herein have been supplied by the USRDS. The interpretation and reporting of these data are the responsibility of the authors and in no way should be seen as an official policy or interpretation of the US government. None of the authors received compensation for participation in this study.