In patients with end-stage heart failure, left ventricular assist device (LVAD) implantation is the preferred therapy as a bridge to transplantation.1 Because of donor heart shortage and rising numbers of LVAD implants, this therapy also becomes an alternative for heart transplantation.2,3 However, LVAD implantation induces a hypercoagulability state that can result in potentially life-threatening cerebrovascular adverse events such as thromboembolic (TE) episodes.3 Therefore, anticoagulation is inevitable in patients with LVAD implants.4,5 Treatment is usually performed by a combination of oral anticoagulants with antiplatelet therapy. Unfortunately, this therapeutic strategy also increases the risk for hemorrhagic complications.6,7 Moreover, LVAD implant itself increases the risk of hemorrhagic events by leading to acquired von Willebrand disease.8 Hence, it is challenging to find a balance between the risk of TE and hemorrhagic complications in patients with LVAD implants.9,10 The complications may occur independent of the patient’s Interagency Registry for Mechanical Assisted Circulatory Support (INTERMACS) level and are major limiting factors for the duration of LVAD support.3
Almost 50% of heart failure patients have clinical signs of atrial fibrillation (AF).11 Atrial fibrillation results in blood stasis,12 and increases stroke risk13 and mortality.14 Atrial fibrillation must, thus, be regarded an additional risk factor of poor clinical outcome in LVAD-supported patients. Nevertheless, available data are controversial: one study reported more TE events in patients with preoperative AF compared with patients without preoperative AF,15 whereas two other studies,16,17 among them a large retrospective analysis of INTERMACS data,16 reported no increase in TE event rates in patients with AF. However, the INTERMACS data analysis reported a higher incidence of bleeding complications in patients with AF than in other patients.16
Despite LVAD support, right ventricular filling, and consequently cardiac output, may still be compromised in AF patients.17 Because right heart failure (RHF) is a well-known risk factor of mortality,18 this complication may influence survival in LVAD-supported patients with AF. Even so, available data are again controversial: one earlier study reported a strong trend for higher midterm mortality in LVAD-supported patients with AF,17 whereas two other studies reported no differences in midterm mortality between patients with and without AF.15,16
It was, therefore, the aim of the current study to compare by the use of a propensity score (PS)-adjusted statistical approach in consecutive LVAD patients with and without permanent AF the risk of thrombotic and hemorrhagic complications as well as overall survival during a 2 year follow-up period.
Materials and Methods
For the present investigation, all patients who received as first device Heartware (HVAD, HeartWare International, Inc, Framingham, MA) or HeartMate II (Thoratec Corp, Pleasanton, CA) implants at the Heart and Diabetes Center NRW, Germany, were eligible. In total, 432 patients who attended the mechanical circulatory support program of our clinic from January 2009 until May 2014 fulfilled this criterion and were considered for data analysis. However, we intended to include in our data analysis only those patients with definitive sinus rhythm (SR) or AF. Therefore, inclusion criterion was that a preoperative electrocardiogram had to be performed at our clinic. Consequently, we excluded 29 patients with missing electrocardiographic data. Additionally, 73 patients with paroxysmal AF (according to the medical report of the referring doctor) and 8 patients with cardiac thrombus formation and left atrial appendage excision during LVAD implantation were excluded, leaving 322 LVAD patients who were finally included in our data analysis (Figure 1). Written informed consent for scientific use of clinical data was obtained by all patients or their relatives (in case of unconscious patients). The investigation conforms to the principles outlined in the Declaration of Helsinki. Because of the retrospective study design, the need for an Ethics committee votum was waived.
All LVAD implants were performed via full median sternotomy using cardiopulmonary bypass. Before LVAD implantation, the cardiac rhythm of the patient was analyzed and compared with the first and after electrocardiographic records during preoperative hospital stay. In addition, transesophageal echocardiographic examination was performed to exclude cardiac thrombus formation and persistent foramen ovale or atrial septal defects.
The standard anticoagulation regimen for patients with LVAD implants at our clinic is phenprocoumon with aspirin. The phenprocoumon dose was checked by daily monitoring of the international normalized ratio (INR). For all patients with LVAD implants, the INR target range was 2.3–2.8, irrespectively whether or not they had AF. Outpatients performed INR self-monitoring and management as previously described.19 Moreover, results were sent to our clinic daily via a short electronic message. This patient feedback enables the clinic to contact the patient when INR levels were out of the target range and to advise them to perform dosage adjustment, if necessary. The daily aspirin dose of LVAD recipients is 100 mg. Pharmacological rhythm control during LVAD support was performed according to the American Heart Association guidelines for the management of patients with permanent AF using β-adrenergic receptor blockers, calcium channel blockers, digoxin, and amiodarone, if necessary.20
For data analysis, patients were divided into two groups. Group 1 consisted of patients with SR (designated SR group; n = 205), group 2 consisted of patients with permanent AF (designated AF group; n = 117). Baseline characteristics and data on the clinical status of the patients were available through a prospectively maintained database of our mechanical circulatory support program. This database includes demographic parameters, routine preoperative laboratory indices, preoperative clinical parameters, and major event categories such as RHF, stroke, gastrointestinal bleeding, pump thrombosis, and mortality. Moreover, we calculated from several preoperative parameters the CHA2DS2-VASc score13 and the HAS-BLED score.21 The former score is used to estimate stroke risk in AF patients (≥2 points indicate the need for oral anticoagulation), whereas the latter score is used to estimate bleeding risk (≥3 points indicate an increased bleeding risk).
Primary end point was 2 year overall survival. We followed the patients from the day of LVAD implantation for up to 2 years, irrespectively whether they were transplanted, weaned, or remained on device. Data completeness was 100%. Secondary end points were perioperative RHF, pump thrombosis, gastrointestinal bleeding, and stroke. RHF was considered in case of low mean arterial pressure and central venous pressure values > 18 mm Hg with cardiac index values < 2.0 L/min/m2 in the absence of elevated pulmonary artery pressure/pulmonary capillary wedge pressure values > 18 mm Hg, requiring extracorporeal circulatory membrane oxygenation (ECMO) implantation or requiring inhaled nitric oxide or inotropic therapy for a duration of more than 1 week after LVAD implantation. A stroke was considered present when a functional relevant motoric, sensory, or cognitive neurological deficit persisted for at least 24 h. The diagnosis was verified by multislice computed tomography and confirmed by a neurologist within 24 h. Gastrointestinal bleedings were detected by endoscopy and the necessity of blood transfusions. Pump thrombosis was assumed by low pump flow with an increased power consumption, hematuria, elevated lactate dehydrogenase (LDH) (reference range: 100–600 unit/L) and serum-free hemoglobin (reference range: 0–40 mg/dl) concentrations and verified by computer-gated tomography, transthoracic echocardiography, and right heart catheterization.22
Categorical variables are summarized as percentages. Continuous variables following a normal distribution are presented as means and SDs, and those not normally distributed (as assessed by the Kolmogorov Smirnov test) are displayed as median and interquartile range. We used Fisher’s exact test, the unpaired t-test and the Mann–Whitney U test, respectively, when appropriate to assess group-specific differences in categorical variables and normally and non-normally distributed continuous variables. We generated Kaplan–Meier estimator to assess the probability of overall survival and freedom from secondary end points during follow-up as a function of time after LVAD implantation. Secondary end points were only assessed in patients on device, whereas patients were censored whether they died on device, or were weaned, or transplanted. We also performed PS-adjusted Cox regression analysis and logistic regression analysis to evaluate the association of the study groups with the primary and secondary end points. Data are expressed as hazard risk (HR, Cox regression analysis) or odds ratio (OR, logistic regression analysis) with 95% confidence interval (CI). The PS-adjusted analyses were performed to control for selection bias because of the result of non-random group assignment.23 The PS derivation model was constructed using multivariable logistic regression, with study group as the binominal dependent variable and those preoperative variables listed in Table 1. The model’s reliability and predictive ability were measured with the Hosmer–Lemeshow test and the c-index, respectively. Study group and the PS covariate were then included in Cox regression/logistic regression models predicting postoperative morbidity or morality. In addition to the PS-adjusted analyses, multivariable logistic regression models including all preoperative variables listed in Table 1 were also fit for each of the outcomes, including mortality. In addition, we generated competing outcome curves for both study groups to report the probability of weaning, transplantation, ongoing device support, or death during follow up. The p values < 0.05 were considered statistically significant. We applied the statistical software package SPSS, version 21 (IBM Corp, Armonk, NY) to perform the analyses.
Baseline characteristics of the study cohort are shown in Table 1. The AF group was significantly older and had a lower BMI than the SR group. Moreover, compared with the SR group the prevalence of hypertension was higher and the percentage of patients with ECMO implants was lower in the SR group. However, diagnosis, concomitant diagnoses such as diabetes mellitus, peripheral arterial occlusive disease, previous myocardial infarction, and other characteristics like INTERMACS level, redo sternotomy, hemofiltration, resuscitation, and intra-aortic balloon pump (IABP) implantation did not differ significantly between groups. The CHA2DS2-VASc score was significantly higher in the AF group than in the SR group, whereas the HAS-BLED score was comparable between groups. Of the SR and AF group, 81 and 55 patients, respectively, received HMII implants, and 124 and 62, respectively, received HVAD implants. The type of device did not differ significantly between the two groups (p = 0.199). Kidney function was similar in the SR and AF group (creatinine levels: 1.54 ± 0.95 vs. 1.59 ± 0.76 mg/dl, respectively; p = 0.696) and so was liver function (bilirubin levels: 1.71 ± 2.11 vs. 1.73 ± 1.43 mg/dl, respectively; p = 0.559). Moreover, pulmonary vascular resistance was similar between groups (223 ± 135 vs. 244 ± 121 dyn s/cm5, respectively; p = 0.473), whereas central venous pressure was significantly higher in the AF group than in the SR group (12.7 ± 6.4 vs. 10.5 ± 5.9 mm Hg; p = 0.006). The PS model that we used to adjust clinical outcomes for non-random group assignment was reliable (Hosmer–Lemeshow test p = 0.356), as well as moderately discriminate (c-statistic = 0.72, 95% CI: 0.66–0.78).
Primary End Point
During follow-up, 58 patients were transplanted and 7 patients were weaned. Although significantly more patients were on destination therapy in the AF group than in the SR group (35.9% vs. 19.0%; p = 0.001), the Kaplan–Meier estimates of the probability of heart transplantation were comparable between groups (21.5% vs. 28.9%; p = 0.239). The probability of weaning was 4.4% in the SR group and 2.2% in the AF group (p = 0.291). Competing outcomes of the two groups are illustrated in (see Figure S1, Supplemental Digital Content, http://links.lww.com/ASAIO/A134). In total, 128 patients died, of whom 71 patients belonged to the SR group and 57 patients belonged to the AF group. Out of these 128 patients, 118 died during LVAD support, 8 after transplantation, and 2 after weaning. In the AF and SR group, causes of death were RHF in 21.4% (n = 25) and 10.2% (n = 21), respectively (p = 0.008), multiorgan failure with sepsis in 14.5% (n = 17) and 11.2% (23), respectively (p = 0.386), stroke in 10.3% (n = 12) and 6.8% (n = 14), respectively (p = 0.293), pump thrombosis in 0.9% (n = 1) and 2.4% (n = 5), respectively (p = 0.423), rejection after heart transplantation in 2.6% (n = 3) and 2.4 (n = 5), respectively, (p > 0.999), and cardiac arrhythmia after weaning in 1.7% (n = 2) and 0% (n = 0), respectively (p = 0.131). In the AF group, RHF and sepsis were the cause of death in 10 and 1 patients, respectively, during the first postoperative month and in 15 and 16 patients, respectively, thereafter. The corresponding data for the SR group were 6 and 1 patients, respectively, during the first postoperative month and 15 and 22 patients, respectively, thereafter. Two year survival was 65.4% in the SR group and 51.3% in the AF group. Events per 100 patient-years were 23 and 39, respectively. The PS-adjusted HR of 2 year mortality was higher in the AF group than in the SR group (Figure 2) and was for the AF group (reference: SR group) = 1.48 (95% CI: 1.02–2.15; p = 0.038). The relationship between study group and mortality obtained from the multivariable logistic model was similar to that of the PS-adjusted analysis (HR 1.53; 95% CI: 1.05–2.21; p = 0.026). Moreover, results did not change substantially by inclusion of the type of device into the statistical model (HR 1.53; 95% CI: 1.05–2.22; p = 0.026).
Secondary End Points
In the SR and AF group, the incidence of perioperative RHF was 42.4% and 49.6%, respectively, and did not differ significantly between study groups (Table 2). The 2 year incidence of stroke, pump thrombosis, and gastrointestinal bleeding varied between 13.6% and 28.6%, depending on study group and complication. Regarding strokes, the incidence of TE events was in the SR and AF group 15.7% and 11.3%, respectively (p = 0.593). The corresponding values for hemorrhagic events were 12.9% and 11.3%, respectively (p = 0.965). The linearized incidence of TE strokes was in the SR and AF group 12.5 and 10.6 per 100 patients-years. The corresponding values for hemorrhagic strokes were 9.6 and 9.8 per 100 patient-years. The PS-adjusted and multivariable-adjusted HRs of the secondary end points were comparable between groups (Table 2).
The current study indicates similar complication rates of stroke, pump thrombosis, and gastrointestinal hemorrhage, but significantly higher 2 year mortality in patients with permanent AF than in patients with SR.
To the best of our knowledge, this is the first study to show increased midterm mortality in AF patients with LVAD implants. Data are in general agreement with a smaller study by Enriquez et al.,17 reporting a strong trend for higher midterm mortality in patients with persistent AF (HR = 2.65 [95% CI: 0.96–7.35]), but not in patients with paroxysmal AF (HR = 0.77 [95% CI: 0.25–2.40]). Compared with the AF group of our study, the SR group was younger, had lower central venous pressure values, and the prevalence of hypertension was lower. The SR group had, however, also more preoperative interventions than the AF group. In the era of donor heart shortage, old age (>65 years) and LVAD implants are relative contraindications for heart transplantation. At our institution, younger patients listed ‘high urgent’ for a donor heart first receive IABP and ECMO implants if they are hemodynamically unstable, except patients with INTERMACS levels 1. Therefore, adjustment is necessary to prevent bias caused by preoperative confounding. Thus, it is important that even after PS or multivariable adjustment results remained significant. The differences may at least in part be explained by the relatively high frequency of RHF as cause of death in AF patients. This assumption is in general agreement with a meta-analysis that reported in patients with left ventricular dysfunction an odds ratio of 2.98 (95% CI: 2.02–4.39) between right ventricular dysfunction and overall mortality.18 Moreover, our results are in line with increased mortality in patients with new-onset of AF after LVAD implantation.24 However, the present results contrast the aforementioned large retrospective INTERMACS data analysis.16 Of note, our results differ from those of the INTERMACS data analysis in that RHF was much more often the primary cause of death, especially in the AF group (21.4% in the current study vs. 4.1% in the INTERMACS data analysis). It is also noteworthy that completeness of follow-up data differed markedly between the two analyses, being at the end of the follow-up period 100% in the present investigation and obviously <10% in the INTERMACS database analysis. This may also be an explanation for the difference in RHF among the two studies and we would speculate at this stage that as a consequence of our complete and thorough follow-up, RHF was more often detected in these patients and found to be the cause of death. Another explanation is the significantly higher percentage of patients on destination therapy in the AF group than in the SR group, which may have influenced patient allocation. The percentage of patients who were transplanted or weaned was non-significantly higher in the SR group than in the AF group. This may at least in part have influenced overall survival. As the higher midterm mortality in the AF patients of our study is most likely disease-related, our results do not contradict LVAD implantation.
The recommended INR target range for patients supported with centrifugal LVAD implants is 2.0–3.025 and our anticoagulation regimen was in line with these recommendations. Generally, AF seems to increase the risk of TE events in LVAD-supported patients.15 Therefore, it has been assumed that a higher INR target range is necessary in LVAD-supported patients with AF than in LVAD-supported patients with SR.16 Indeed, similar incidence of TE events has been reported in LVAD-supported SR patients with INR target range of 1.5–2.0 and in AF patients with INR target range of 2.0–2.5.17 In our study, the incidence of TE events was similar in SR and AF patients, although the INR target range was identical for both groups. It may well be that an INR target range of 2.3–2.8 is sufficient to minimize TE events, independently whether or not AF is present. However, it is also noteworthy that the patients of our study performed INR self-monitoring. It has been demonstrated that INR self-monitoring results in less INR values outside the target range than INR measurement by the patient’s health-care provider. This strategy is, therefore, able to reduce TE events.26
It was not surprising that gastrointestinal bleeding incidence was comparable between the patients with AF and SR, because the INR target range was identical for both groups. In the retrospective INTERMACS data analysis,16 the reported higher bleeding incidence in AF patients than in non-AF patients has been explained by a higher INR target range to reduce TE events. Therefore, there is a tendency in some centers to reduce the INR target range below 2.0 for patients supported with axial flow LVAD devices like HeartMate II.27 However, this trend also showed a higher risk of pump thrombosis.28 It is also noteworthy that in conventional groups requiring oral anticoagulation with vitamin K antagonists, the incidence of major bleeding events is substantially lower at INR values of 2.0–3.07,26 than in studies in LVAD-supported patients.16,17 In our AF group, for instance, the average HAS-BLED score of 1.46 indicates a much lower bleeding risk (approximately 4% within 2 years) than the observed incidence of gastrointestinal bleedings (27.5% within 2 years). In addition, in AF patients with an INR target range of 2.0–2.5 the bleeding risk was similar to non-AF patients with an INR target range of 1.5–2.0.17 Collectively, results indicate that a low INR target range may not reduce bleeding complications substantially. These data are in line with the hypothesis that instead of oral anticoagulation intensity acquired von Willebrand disease is the major cause of bleeding complications in LVAD-supported patients.29 Altogether, a target INR range of 2.3–2.8 may be acceptable for minimizing TE events without increasing the risk of bleeding complications.
Our study has several strengths, but also some limitations. Strengths are the PS- and multivariable-adjustment of clinical outcome variables, the relative large sample size of frequently used devices, the identical oral anticoagulation and anti-aggregation therapy protocol for all patients, and the 100% completeness of 2 year follow-up data. Limitations are that 1) data on re-hospitalization rates and quality of life were not available, 2) arterial pressures were not recorded, and 3) arrhythmic disorders under LVAD support were not assessed. Therefore, we cannot definitively rule out that results on midterm mortality were attenuated because some patients in the SR group developed paroxysmal/permanent AF.
In conclusion, our data indicate higher midterm mortality but similar thrombotic and bleeding events in LVAD-supported patients with permanent AF than in patients with SR. However, results do not contradict LVAD implantation in patients with AF.
1. Stehlik J, Hosenpud JD, Edwards LB, Hertz MI, Mehra MR. International Society for Heart and Lung Transplantation: ISHLT International Registry for Heart and Lung Transplantation–into the fourth decade, from strength to strength. J Heart Lung Transplant 2013.32: 941950.
2. Kirklin JK, Naftel DC, Pagani FD, et al. Long-term mechanical circulatory support (destination therapy): On track to compete with heart transplantation? J Thorac Cardiovasc Surg 2012.144: 584603; Discussion 597.
3. Kirklin JK, Naftel DC, Pagani FD, et al. Seventh INTERMACS annual report: 15,000 patients and counting. J Heart Lung Transplant 2015.34: 14951504.
4. Lloyd-Jones D, Adams RJ, Brown TM, et al. Executive summary: Heart disease and stroke statistics–2010 update: A report from the American Heart Association. Circulation 2010.121: 948954.
5. Camm AJ, Lip GY, De Caterina R, et al.; ESC Committee for Practice Guidelines (CPG): 2012 focused update of the ESC Guidelines for the management of atrial fibrillation: An update of the 2010 ESC Guidelines for the management of atrial fibrillation. Developed with the special contribution of the European Heart Rhythm Association. Eur Heart J 2012.33: 27192747.
6. Lamberts M, Lip GY, Ruwald MH, et al. Antithrombotic treatment in patients with heart failure and associated atrial fibrillation and vascular disease: A nationwide cohort study. J Am Coll Cardiol 2014.63: 26892698.
7. Koertke H, Zittermann A, Wagner O, et al.; Christ of Huth: Telemedicine-guided, very low-dose international normalized ratio self-control in patients with mechanical heart valve implants. Eur Heart J 2015.36: 12971305.
8. Amer S, Shah P, Hassan S. Gastrointestinal bleeding
with continuous-flow left ventricular assist devices. Clin J Gastroenterol 2015.8: 6367.
9. Menon AK, Götzenich A, Sassmannshausen H, Haushofer M, Autschbach R, Spillner JW. Low stroke rate and few thrombo-embolic events after HeartMate II implantation under mild anticoagulation. Eur J Cardiothorac Surg 2012.42: 319323; Discussion 323.
10. Litzler PY, Smail H, Barbay V, et al. Is anti-platelet therapy needed in continuous flow left ventricular assist device patients? A single-centre experience. Eur J Cardiothorac Surg 2014.45: 5559; Discussion 59.
11. Eapen ZJ, Greiner MA, Fonarow GC, et al. Associations between atrial fibrillation and early outcomes of patients with heart failure and reduced or preserved ejection fraction. Am Heart J 2014.167: 369375.e2.
12. Shively BK, Gelgand EA, Crawford MH. Regional left atrial stasis during atrial fibrillation and flutter: Determinants and relation to stroke. J Am Coll Cardiol 1996.27: 17221729.
13. Lip GY, Nielsen PB, Skjøth F, Lane DA, Rasmussen LH, Larsen TB. The value of the European society of cardiology guidelines for refining stroke risk stratification in patients with atrial fibrillation categorized as low risk using the anticoagulation and risk factors in atrial fibrillation stroke score: A nationwide cohort study. Chest 2014.146: 13371346.
14. Dries DL, Exner DV, Gersh BJ, Domanski MJ, Waclawiw MA, Stevenson LW. Atrial fibrillation is associated with an increased risk for mortality and heart failure progression in patients with asymptomatic and symptomatic left ventricular systolic dysfunction: A retrospective analysis of the SOLVD trials. Studies of left ventricular dysfunction. J Am Coll Cardiol 1998.32: 695703.
15. Stulak JM, Deo S, Schirger J, et al. Preoperative atrial fibrillation increases risk of thromboembolic events after left ventricular assist device implantation. Ann Thorac Surg 2013.96: 21612167.
16. Xia Y, Stern D, Friedmann P, Goldstein D. Preoperative atrial fibrillation may not increase thromboembolic events in left ventricular assist device recipients on midterm follow-up. J Heart Lung Transplant 2016.35: 906912.
17. Enriquez AD, Calenda B, Gandhi PU, Nair AP, Anyanwu AC, Pinney SP. Clinical impact of atrial fibrillation in patients with the HeartMate II left ventricular assist device. J Am Coll Cardiol 2014.64: 18831890.
18. Iglesias-Garriz I, Olalla-Gómez C, Garrote C, et al. Contribution of right ventricular dysfunction to heart failure mortality: A meta-analysis. Rev Cardiovasc Med 2012.13: e62e69.
19. Koertke H, Zittermann A, Minami K, et al. Low-dose international normalized ratio self-management: A promising tool to achieve low complication rates after mechanical heart valve replacement. Ann Thorac Surg 2005.79: 19091914.
20. January CT, Wann LS, Alpert JS, et al.; American College of Cardiology/American Heart Association Task Force on Practice Guidelines: 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: A report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines and the Heart Rhythm Society. J Am Coll Cardiol 2014.64: e1e76.
21. Pisters R, Lane DA, Nieuwlaat R, de Vos CB, Crijns HJ, Lip GY. A novel user-friendly score (HAS-BLED) to assess 1-year risk of major bleeding in patients with atrial fibrillation: The Euro Heart Survey. Chest 2010.138: 10931100.
22. Oezpeker C, Zittermann A, Ensminger S, et al. Systemic thrombolysis versus device exchange for pump thrombosis management: A single-center experience. ASAIO J 2016.62: 246251.
23. D’Agostino RB Jr. Propensity scores in cardiovascular research. Circulation 2007.115: 23402343.
24. Kormos RL, Teuteberg JJ, Pagani FD, et al.; HeartMate II Clinical Investigators: Right ventricular failure in patients with the HeartMate II continuous-flow left ventricular assist device: Incidence, risk factors, and effect on outcomes. J Thorac Cardiovasc Surg 2010.139: 13161324.
25. Feldman D, Pamboukian SV, Teuteberg JJ, et al.; International Society for Heart and Lung Transplantation: The 2013 International Society for Heart and Lung Transplantation Guidelines for mechanical circulatory support: Executive summary. J Heart Lung Transplant 2013.32: 157187.
26. Heneghan C, Ward A, Perera R, et al.; Self-Monitoring Trialist Collaboration: Self-monitoring of oral anticoagulation: Systematic review and meta-analysis of individual patient data. Lancet 2012.379: 322334.
27. Katz JN, Adamson RM, John R, et al.; TRACE study: Safety of reduced anti-thrombotic strategies in HeartMate II patients: A one-year analysis of the US-TRACE study. J Heart Lung Transplant 2015.34: 15421548.
28. Starling RC, Moazami N, Silvestry SC, et al. Unexpected abrupt increase in left ventricular assist device thrombosis. N Engl J Med 2014.370: 3340.
29. Suarez J, Patel CB, Felker GM, Becker R, Hernandez AF, Rogers JG. Mechanisms of bleeding and approach to patients with axial-flow left ventricular assist devices. Circ Heart Fail 2011.4: 779784.