Secondary Logo

Journal Logo

Adult Circulatory Support

Preoperative Predictors of Mortality in Short-Term Continuous-Flow Ventricular Assist Devices

Bozso, Sabin J.*; Buchholz, Holger*; Pidborochynski, Tara; Freed, Darren H.*,†; MacArthur, Roderick G. G.*; Conway, Jennifer

Author Information
doi: 10.1097/MAT.0000000000000902
  • Free


Acute heart failure represents a significant therapeutic challenge, with mortality rates reported as high as 70%.1 This is despite significant advances in both medical and interventional therapies.2 This acute decompensation in heart function commonly occurs secondary to decompensated heart failure in the setting of a preexisting cardiomyopathy, following a massive myocardial infarction or postcardiotomy. The definitive therapy for both advanced and refractory heart failure is heart transplantation; however, a shortage of donor organs precludes the broad application of this therapy.

Ventricular assist devices (VADs) were originally developed as a bridge to recovery for patients developing postcardiotomy cardiogenic shock.3,4 The utility of these devices has since been expanded to include long-term applications such as bridge to transplantation and even destination therapy.5 To meet these indications, both temporary and durable device options exist. To achieve the best short- and long-term outcomes, VAD therapy should be initiated expeditiously and in the appropriate patient population.6 Patients with acute cardiogenic shock often have multiorgan failure in addition to an uncertain neurologic status that can complicate the decision to proceed with initiation of mechanical circulatory support. In these situations, implantation of a durable VAD has been shown to lead to poor patient outcomes and are not considered cost-effective.7 A temporary VAD can be utilized in these situations as a bridge to decision, recovery, or allow more time for assessment for candidacy for a long-term VAD.

Currently, there are a limited number of short-term VAD options in the adult population. The Impella (ABIOMED, Inc., Danvers, MA) and Tandem Heart (Tandem Life, Pittsburgh, PA) are short-term pumps approved for use as rescue therapy.5 The Thoratec CentriMag (Abbott Inc., Chicago, IL) is a nondurable, centrifugal, continuous-flow pump with a magnetically levitated rotor with no bearing or seals. The CentriMag is designed to support the left ventricle (LVAD), right ventricle (RVAD) or both ventricles (BiVAD) and has the added advantage of potentially less thrombus formation compared with the other short-term devices.8

The challenge in implementing temporary VAD support lies with optimizing patient selection. Numerous studies have been performed investigating the role of preoperative and postoperative clinical and biochemical variables, resulting in the development of several indices used clinically.9–12 However, these studies focused exclusively on durable VADs, and the few studies that have investigated temporary VAD support have either neglected preoperative variables or have not focused exclusively on short-term VADs.13,14 Therefore, we sought to describe our single-center experience with the CentriMag as a short-term, continuous-flow (STCF) VAD and elucidate preoperative risk factors for mortality, in an attempt to provide further clarity when selecting patients for mechanical circulatory support with STCF-VADs.


The local Research Ethics Board approved the study protocol, and the requirement for individual patient consent was waived. Study data were collected and managed using REDCap, Nashville, TN15 electronic data capture tools hosted and supported by the Women and Children’s Health Research Institute at the University of Alberta.

Patient Population

This was a retrospective study of all adult patients (>18 years) supported with the CentriMag cannulated in a VAD configuration from June 1, 2009, to December 31, 2015, at the Mazankowski Alberta Heart Institute, University of Alberta, Edmonton, Alberta, Canada. All patients were cannulated using bypass cannulas in an LVAD, an RVAD, or a BiVAD configuration as depending on the clinical situation, and only cannulation with the first device was analyzed. All cannulations were performed centrally and either apically or transmitral via the right superior pulmonary vein. Device choice was made based on the etiology of acute heart failure and the need for support. Patient demographic and clinical characteristics were collected.

All patients included in our study population met criteria for Interagency Registry of Mechanically Assisted Circulatory Support (INTERMACS) Profile 1 at the time of STCF-VAD implantation. However, 46% of our cohort had mechanical circulatory support (MCS) initiated postcardiotomy; thus, if they failed to wean from cardiopulmonary bypass (CPB) these patients were assigned an INTERMACS Profile 1. We recognize that this scale was designed for heart failure patients but felt as these patients were dependent on mechanical circulatory support and unable to wean from CPB that an INTERMACS Profile 1 was the most appropriate designation at the time of implant.16


The primary outcome of this study was decannulation from the STCF-VAD because of transplantation, ventricular recovery, or conversion to a long-term VAD, defined as a successful bridge, and death or death within 30 days of weaning off the device, defined as an unsuccessful bridge. Complete follow-up data were available for all patients.

For each patient, preoperative and perioperative variables were collected. These variables were selected based on the ease of collection in a real-world patient population as well as emphasis in prior VAD risk scores.11,12,14 These variables included demographic data such as age at implant, sex, body mass index, preoperative dialysis, preimplant diagnosis, and VAD strategy. Perioperative variables such as extracorporeal membrane oxygenation (ECMO) and concomitant cardiac surgery were also collected. Additionally, a variety of standard preoperative biochemical variables were collected within 72 hours before implantation and analyzed. Preimplant renal insufficiency was defined as a creatinine >176 μmol/L (equivalent to 2 mg/dL) as defined previously.16 A bilirubin cutoff of 34.2 μmol/L was chosen, and high creatinine defined as a serum creatinine >220 μmol/L, based on previously described VAD risk scores.12 Pre-VAD renal failure was defined as the presence of preimplant renal insufficiency or dialysis within 72 hours of VAD implantation. Platelet counts were collected and divided by a factor of 10 to allow for smaller numbers and ease of interpretation. The left ventricular internal diastolic dimension (LVIDd) was measured on preimplantation echocardiography, and patients were excluded if they underwent echocardiography >30 days before implantation.

Ten types of VAD-related complications were explored in this cohort as outlined by the INTERMACS Adverse Event Definitions.16 These included major bleeding, neurologic dysfunction, major infection, ischemic organ damage, hepatic dysfunction, acute renal failure, respiratory failure, venous thromboembolism event, myocardial infarction, and device malfunction.

Statistical Analysis

Descriptive statistical methods were used. Continuous variables are described as median with interquartile range (IQR) because of non-normality of the data, and absolute numbers are presented with proportions for categorical variables. All preoperative clinical and biochemical variables were assessed. The duration of support was calculated as the number of days between implant and removal of the device, regardless of the reason for decannulation. Univariate logistic regression model with each variable was performed. Variables for the multiple logistic regression were selected if the p value <0.2. A multiple logistic regression model was fitted based on the variables selected from the univariate analysis. Variables in the multiple logistic regression model were considered significant and kept in the model if the p value was <0.05. Analysis was performed as two groups: all VADs and LVAD/BiVAD, excluding isolated RVAD. Variables eliminated at the multivariable stage were tested for confounding by assessing β coefficient change with a standard set at 15%. Because high creatinine is a component of preimplantation renal failure, only preimplantation renal failure was included in the model (Fisher exact test for association p < 0.001). The model for all VADs was significant with a likelihood ratio of 13.44 (p = 0.0002). The model for LVAD/BiVAD was also significant with a likelihood ratio of 10.4 (p = 0.0013).

Patient Management

Device selection in our institution depends on the acuity of the clinical situation and potential for recovery. For patients who acutely deteriorate, insertion of an intra-aortic balloon pump is first-line therapy, followed by either institution of ECMO or short-term VAD if clinical deterioration occurs. Conversion to a short-term VAD usually occurs in the context of a bridge to recovery or bridge to decision. If a patient was to be converted to a long-term device or be bridged to transplantation with a short-term device, a comprehensive transplant assessment occurred. Postoperative STCF-VAD management included starting heparin when the chest tube drainage was decreased to <120 ml/12 h and there were no other concerns for bleeding. The heparin was titrated to maintain a target partial thromboplastin time (PTT) of 55–65.


Demographics and Clinical Outcomes

The demographic and baseline preoperative characteristics of the study population are outlined in Table 1. From June 2009 to December 2015, 61 patients (41 male [77%]) with a median age of 54.6 years (IQR, 45.6–62.6 years) were supported with an STCF-VAD. The most common reason for implantation was cardiomyopathy in 27 (44%), including 12 chronic ischemic cardiomyopathy, 9 dilated cardiomyopathy, and 6 valvular cardiomyopathy. Acute coronary syndrome accounted for 18 (30%) and postheart transplantation for 7 (12%) implants. At the time of implantation, 18 recipients (30%) were supported by ECMO with a mean duration of 3.1 days before conversion to an STCF-VAD. Thirty-one (51%) of the implanted devices were LVADs, 13 (21%) were RVADs, and 17 (28%) were BiVADs.

Table 1
Table 1:
Baseline Demographic and Preoperative Characteristics

Patients were supported with an STCF-VAD for a median duration of 11 days (IQR, 4–29 days), with the longest duration being 83 days. Ten recipients (16%) were subsequently converted to a long-term VAD (HeartWare HVAD, Framingham, MA or HeartMate II LVAD, Pleasanton, CA). Fourteen patients (23%) were weaned for recovery, and 8 (13%) patients were transplanted. Overall, 32 (52%) patients had a successful bridge as previously defined by our group, and all were discharged from hospital. Twenty-nine (48%) patients had an unsuccessful bridge, with 26 (43%) deaths occurring with the STCF-VAD in place and 3 (5%) deaths within 30 days of decannulation. Among discharged patients, 81% were alive at a median follow-up of 42 months.

The average left ventricular ejection fraction among patients with an unsuccessful bridge and successful bridge was 20 ± 15% and 24 ± 17%, respectively. The LVIDd among patients with an unsuccessful bridge and successful bridge was 5.1 ± 1.3 and 5.6 ± 1.4 cm. For those patients not on ECMO or cardiopulmonary bypass (n = 13) at the time of implantation, 67% with available data were on intravenous vasoactive medications (n = 28/42).

Device Performance and Complications

There were no device-related mechanical failures. Five (8.2%) of the recipients required a short-term device exchange or removal of a section of the circuit for evolving thrombus. Adverse events encountered by our cohort of patients are summarized in Table 2. Acute renal failure and major bleeding occurred most commonly, complicating 21 (34%) and 20 (33%) runs, respectively. Twelve (20%) runs developed neurologic dysfunction, whereas 11 (18%) patients required prolonged ventilation or tracheostomy.

Table 2
Table 2:
Adverse Events

Significant Preoperative Predictors of Mortality

Univariate analysis revealed that preimplant renal insufficiency or dialysis, platelet count, elevated bilirubin, and high creatinine were significantly associated (p value <0.05) with an unsuccessful bridge (Table 3). There was no significant correlation between age, VAD strategy or diagnosis, and an unsuccessful bridge (p values >0.2). Five variables (male sex, postcardiotomy implant, preimplant renal insufficiency or dialysis, platelet count, and elevated bilirubin) entered multivariable regression analysis to determine independent predictors of an unsuccessful bridge (Table 4). Only preimplant renal insufficiency or dialysis remained significant (odds ratio [OR] = 7.53; 95% CI: 2.10–27.1; p = 0.002] in multivariable analysis of all VAD types. Multivariable regression analysis excluding RVAD demonstrated similar results with preimplant renal insufficiency or dialysis remaining a significant predictor of an unsuccessful bridge in the LVAD/BiVAD group (OR = 7.23; 95% CI: 2.01–25.9; p = 0.002). For both models included in this study the goodness of fit test was nonsignificant with a p value close to 1 indicating that the models fit the data well.

Table 3
Table 3:
Univariate Analysis
Table 4
Table 4:
Multivariable Analysis


Although previous studies have investigated preoperative risk factors for mortality in durable VADs and established the utility of clinical indices based on postimplantation variables, this analysis sought to add to the growing literature on short-term VAD support. These results demonstrate that an STCF-VAD can be successfully utilized to bridge patients with acute heart failure to ventricular recovery, transplantation, or conversion to durable VAD in just over half of cases. In addition, intermediate-term follow-up suggests survival in patients discharged from hospital after a successful bridge is excellent, with 81% surviving to a median of 42 months, and last, we found that renal insufficiency or dialysis before implantation was a significant predictor of an unsuccessful bridge.

Several studies have been performed investigating the role of preoperative and postoperative clinical and biochemical variables resulting in the development of several clinical indices used clinically. However, these studies exclusively investigated patients undergoing implantation of durable VADs. Kalogeropoulos et al.12 evaluated various right ventricular failure prediction models after durable VAD implantation and found that they performed only modestly. A large study including over 200 patients found 13 significant predictors of mortality postdurable VAD implantation.11 These included age >50 years, clinical data such as redo surgery, on ECMO or intra-aortic balloon pump (IABP) at the time of implant, and laboratory values including platelets <100 × 103/μL. The authors concluded that mortality can be predicted preoperatively, and a scoring system may assist physicians in guiding optimal patient selection for implantation.

Few studies have assessed for predictors of mortality in STCF-VADs, and the few that have focused primarily on postoperative variables. Worku et al.14 developed a risk stratification scoring system to predict survival after STCF-VAD implantation. They found that a bilirubin level >88.9 μmol/L, female sex, and diagnosis of postcardiotomy shock were significant predictors of mortality. The biochemical variables, however, were collected on postoperative day 3 rather than preimplantation. Furthermore, this group excluded patients supported with ECMO before conversion to STCF-VAD. A study examining outcomes after STCF-VAD implantation for refractory cardiogenic shock in 90 patients found ongoing cardiopulmonary resuscitation to be an independent risk factor for mortality.13 However, this study grouped veno-arterial-ECMO(VA-ECMO) and STCF-VAD together; thus, it did not provide an isolated analysis of STCF-VADs. The use of STCF-VADs in specific patient populations, such as postcardiotomy, has also been investigated.17 The largest study examining the use of STCF-VADs included over 11,000 patients and focused on incidence, outcomes, and cost analysis without analysis of adverse events.18

The results of this current study lend support to the use of STCF-VADs in patients with acute heart failure including a broad list of indications for STCF-VAD implantation, such as preimplant ECMO, to represent a real-world experience. Over 50% of STCF-VAD recipients survived to discharge from hospital and over 80% of those patients survived to intermediate-term follow-up. The device used in this analysis was reliable with no pump failures and rare development of clots in the pump. The most common complication in our series was acute renal failure in 34% of the runs. In addition, 33% experienced major bleeding and 20% developed neurologic dysfunction. These reported rates are less than those seen in prior reports of complications post-STCF-VAD implantation.17 Our results suggested that preimplant renal insufficiency or dialysis is a predictor of an unsuccessful bridge. This finding is likely related to the critical preimplantation state of patients being considered for STCF-VAD support. This result highlights the importance of early initiation of STCF-VAD support in acute heart failure before end-organ dysfunction occurs, likely precluding overall functional recovery.

As discussed, prior studies have either focused on preimplantation risk factors in durable VADs or on postimplantation risk factors in STCF-VADs. Therefore, our study addresses a gap in the literature regarding preimplantation risk factors in short-term devices. Some have argued that preoperative risk assessment in patients undergoing emergent surgery by using biochemical parameters is unreliable and that the accuracy of these indicators improves after patient stabilization.19 Nonetheless, our results suggest that preimplantation renal insufficiency or dialysis is an independent predictor of an unsuccessful bridge in patients requiring support with an STCF-VAD. This finding is intuitive and accepted by most clinicians; however, this report provides objective evidence to suggest that preimplantation renal failure is a variable that can, and should, be measured before STCF-VAD cannulation, given the significant predictive value it provides. Measuring variables after implantation may predict mortality; however, the decision regarding support has already been made. Identifying predictive preoperative parameters will help clinicians make the most informed decision regarding candidacy for STCF-VAD support and ability to counsel patients and their families about potential outcomes.

Although this is one of the larger series of adults supported with an STCF-VAD, there are several limitations inherent to the retrospective nature and small sample size. In addition, although this analysis concentrated on several potential risk factors, it was not meant to imply that these are the only risk factors that may affect survival in these patients. Our analysis was unable to capture cardiac index data and aggregate doses of vasoactive medications before STCF-VAD implementation. Although these data would better inform the degree of cardiogenic shock before MCS initiation, all patients included in our study population met criteria for INTERMACS Profile 1 at the time of STCF-VAD implementation. Finally, a selection bias may exist that could have an impact on the results and outcomes. However, all patients in our study were an INTERMACS Profile 1, indicating that they were all critically ill as would be expected in a cohort of patients where short-term support is utilized.

In conclusion, STCF-VADs can successfully bridge critically ill adult patients to recovery, a long-term device, or transplantation in just over half of cases, with an acceptable complication rate. Preoperative renal insufficiency or dialysis is strongly correlated with an unsuccessful bridge in our patient population, likely reflecting the severity of illness preimplantation. Further studies are required to determine whether these factors remain significant in a larger population.


This research has been facilitated by the Women and Children’s Health Research Institute through the generosity of the Stollery Children’s Hospital Foundation and supporters of the Lois Hole Hospital for Women and the Health Outcomes Improvement (HOI) Fund from Maternal, Newborn, Child and Youth Strategic Clinical Network


1. Goldberg RJ, Samad NA, Yarzebski J, Gurwitz J, Bigelow C, Gore JM. Temporal trends in cardiogenic shock complicating acute myocardial infarction. N Engl J Med 1999.340: 1162–1168.
2. Kannel WB. Incidence and epidemiology of heart failure. Heart Fail Rev 2000.5: 167–173.
3. DeBakey ME. Left ventricular bypass pump for cardiac assistance. Clinical experience. Am J Cardiol 1971.27: 3–11.
4. Spencer FC, Eiseman B, Trinkle JK, Rossi NP. Assisted circulation for cardiac failure following intracardiac surgery with cardiopulmonary bypass. J Thorac Cardiovasc Surg. 1965;49:56–73.
5. Peura JL, Colvin-Adams M, Francis GS, et al. American Heart Association Heart Failure and Transplantation Committee of the Council on Clinical Cardiology; Council on Cardiopulmonary, Critical Care, Perioperative and Resuscitation; Council on Cardiovascular Disease in the Young; Council on Cardiovascular Nursing; Council on Cardiovascular Radiology and Intervention, and Council on Cardiovascular Surgery and Anesthesia: Recommendations for the use of mechanical circulatory support: Device strategies and patient selection: A scientific statement from the American Heart Association. Circulation 2012.126: 2648–2667.
6. Reynolds HR, Hochman JS. Cardiogenic shock: Current concepts and improving outcomes. Circulation 2008.117: 686–697.
7. Hoefer D, Ruttmann E, Poelzl G, et al. Outcome evaluation of the bridge-to-bridge concept in patients with cardiogenic shock. Ann Thorac Surg 2006.82: 28–33.
8. Hoshi H, Shinshi T, Takatani S. Third-generation blood pumps with mechanical noncontact magnetic bearings. Artif Organs 2006.30: 324–338.
9. Oz MC, Goldstein DJ, Pepino P, et al. Screening scale predicts patients successfully receiving long-term implantable left ventricular assist devices. Circulation 1995.92(suppl 9): 173
10. Rao V, Oz MC, Flannery MA, Catanese KA, Argenziano M, Naka Y. Revised screening scale to predict survival after insertion of a left ventricular assist device. J Thorac Cardiovasc Surg 2003.125: 855–862.
11. Klotz S, Vahlhaus C, Riehl C, Reitz C, Sindermann JR, Scheld HH. Pre-operative prediction of post-VAD implant mortality using easily accessible clinical parameters. J Heart Lung Transplant 2010.29: 45–52.
12. Kalogeropoulos AP, Kelkar A, Weinberger JF, et al. Validation of clinical scores for right ventricular failure prediction after implantation of continuous-flow left ventricular assist devices. J Heart Lung Transplant 2015.34: 1595–1603.
13. Takayama H, Truby L, Koekort M, et al. Clinical outcome of mechanical circulatory support for refractory cardiogenic shock in the current era. J Heart Lung Transplant 2013.32: 106–111.
14. Worku B, Naka Y, Pak S, et al. Predictors of mortality after short-term ventricular assist device placement. Ann Thorac Surg 2011.92: 1613
15. Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)–a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform 2009.42: 377–381.
16. UAB School of Medicine: Interagency Registry for Mechanically Assisted Circulatory Support. Available at: Accessed July 9, 2017.
17. Mohite PN, Sabashnikov A, Patil NP, et al. Short-term ventricular assist device in post-cardiotomy cardiogenic shock: Factors influencing survival. J Artif Organs 2014.17: 228–235.
18. Stretch R, Sauer CM, Yuh DD, Bonde P. National trends in the utilization of short-term mechanical circulatory support: Incidence, outcomes, and cost analysis. J Am Coll Cardiol 2014.64: 1407–1415.
19. Rix TE, Bates T. Pre-operative risk scores for the prediction of outcome in elderly people who require emergency surgery. World J Emerg Surg 2007.2: 16

mechanical circulatory support; risk factors; outcomes

Copyright © 2019 by the American Society for Artificial Internal Organs