Children with end-stage heart failure have limited options for durable mechanical circulatory support. The widely used Berlin Heart EXCOR pediatric ventricular assist device (VAD Berlin, Germany) is currently the only FDA-approved option for children. It has revolutionized pediatric heart failure management, successfully supporting several thousand children worldwide to heart transplant. Unfortunately, children receiving VAD support are exposed to greater risk of adverse events than adults. Approximately 28% of patients supported with the EXCOR will experience an adverse neurologic event (among which an estimated 12.5% are severe), and nearly half suffer major bleeding while on support.1,2 The focal point of these challenges is antithrombotic management.
The Edmonton Anticoagulation and Platelet Inhibition Protocol (“Edmonton Protocol”) was developed to guide antithrombotic therapies in the Berlin EXCOR Investigational Device Exemption (IDE) Trial.1 Experts in the field used the early European clinical experience and limited available literature to develop this protocol, and it remains the only tool of its kind. Standard agents used in the protocol include heparin, enoxaparin, warfarin, aspirin, and dipyridamole, and there is specific guidance as to when and how to initiate and adjust these therapies. Doses of antiplatelet agents are adjusted according to a sliding scale nomogram, using results of Thromboelastography (TEG), and its platelet function companion assay known as Platelet Mapping (PM, Haemonetics Corporation, Braintree, MA). During the IDE trial with the protocol in use, the adverse event rate was high, and when neurologic events occurred, antithrombotic labs were most often within the specified goal ranges. In the trial, only 9% of neurologic events appeared probably/definitely related to antithrombotic therapy.3
The high neurologic event rate has led to a multitude of center-specific modifications to the protocol in attempts to improve patient outcomes. Each pediatric VAD center in North America currently has a unique approach to antithrombotic therapy (ATT), and communication among all centers poses a challenge. Therein lie opportunities to improve outcomes for this patient population, but there is no consensus as to best AT practice. An important first step is to identify areas of high practice variability, as an indication of diagnostic and therapeutic uncertainty.4 By identifying these high-variability domains, targets for in-depth research can be determined.
The primary study hypotheses are as follows: 1) there is marked practice variation in antithrombotic management among those caring for children on the Berlin EXCOR; and 2) certain management strategies are shared among high-volume VAD centers.
The primary aim of this study is to explore the degree of practice variation in antithrombotic care for patients on the Berlin EXCOR and to simultaneously identify domains of highest practice variation. The secondary aim is to compare management strategies at high- versus low-volume pediatric VAD centers. The broader purpose of this study is to guide future research and possibly standardization of care in ATT for pediatric VAD recipients.
Materials and Methods
This descriptive, cross-sectional survey study assessed practice variation in antithrombotic management at pediatric VAD programs in North America. Because no applicable survey instruments existed, a novel questionnaire was developed based on current literature and expert opinion. The questionnaire was designed by Berlin Heart Inc. and their expert advisors (including those with expertise in hemostasis, patient care after VAD implantation, and bioengineering). The survey was reviewed by clinicians in pediatric hematology and was organized into the following domains: antithrombotic decision-making, application of the Edmonton Protocol, medication selection, medication monitoring, and specific challenges. Each respondent’s self-reported subjective impression of their adherence to the protocol was assessed, in addition to their subjective impression of the efficacy of the protocol. In addition to providing responses to closed-ended questions, respondents were asked to elaborate on their answers in free text form.
Pediatric VAD centers implanting one or more Berlin EXCOR devices between January 2012 and January 2016 were eligible for inclusion. Centers were excluded from the analysis if they failed to complete the survey. A physician at each program (either a surgeon or cardiologist identified as a leader of the VAD program) was contacted via email by Berlin Heart Inc. on March 21, 2017, and invited to complete the questionnaire. The email message provided background information as well as a link to the online survey instrument. If the survey was not completed within 2 weeks, a reminder message was sent, followed by another reminder 2 weeks later. Only one response was received from each institution, which was designated with a center identification number. Data from each responding center were deidentified before analysis.
For the purposes of this study, a high-volume VAD center was defined as one that implanted 14 or more EXCOR devices during the 4-year study period (an average of 3.5 implants per year). This represents the upper quartile of responding centers.
Based on administrative review at the University of Utah/Primary Children’s Hospital Institutional Review Board, this study was classified as exempt from full review based on the determination that it does not qualify as human subjects research.
Response distributions (numbers and percentages) were calculated for key variables of interest. To assess differences in certain practices between high- and low-volume centers, we conducted cross-tabulations of center volume and practice variables. To determine whether differences between high- and low-volume centers were statistically significant, exact Wilcoxon rank sum tests were performed for ordinal variables. Fisher’s exact tests were performed for categorical variables. Exact versions of these methods were employed to account for small cell counts, due to the small sample size. A p value of 0.05 was considered statistically significant. Statistical analyses were conducted using Stata version 13.1.
Surveys were distributed to 37 VAD programs and 17 (45.9%) completed the survey. Participating centers are listed in Table 1. Four centers (23.5%) were defined as high volume (implanting 14 or more devices over the 4-year period), whereas the majority (n = 13) of programs were low volume (Figure 1). Among the 20 nonresponding centers, 15.0% (n = 3) were high-volume centers. The median number of devices implanted at all responding centers was 9, or approximately 2.25 devices per year. The median number of implants at high- and low-volume centers was 18 and 9, respectively.
At 52.9% of centers (9/17), one or two clinicians made all antithrombotic decisions for patients on VAD support (Table 2). The four high-volume centers either limited the decision-making to two individuals (n = 2) or conversely involved a group of four or more (n = 2). This distribution was not significantly different than the approaches of the low-volume centers as discussed below. One of these scenarios was described as a multidisciplinary team consisting of a surgeon, intensivist, pharmacist, and a heart failure cardiologist. No center incorporated hematologists into the primary decision-making, but 76.5% (13/17) consulted them for particularly challenging cases. International colleagues, and Berlin Heart representatives were also involved with difficult cases.
Application of the Edmonton Protocol
All 17 centers (100%) followed the Edmonton Protocol to some extent. Although 47.1% (8/17) followed the Edmonton Protocol either “extremely” or “very closely,” only 5.9% (1/17) perceived it to be very effective (Figure 2). Most of the high-volume centers (3/4, 75.0%) “slightly” followed the protocol and perceived it to be “somewhat effective.” In the entire cohort, there appeared to be no relationship between degree of adherence to the protocol and its perceived efficacy (p = 0.153). Ten centers (66.7%) deviated in at least two aspects of the protocol (Table 3). One center commented that they deviate in “every” aspect of it.
Deviations from the protocol often involved medication selection (Table 3), with 8 of 15 (53.3%) of those reporting deviations, modifying antiplatelet agents (4/15), anticoagulants (3/15), or both (1/15). One center specified that they used higher than recommended doses of aspirin and used heparin longer than in the protocol and returning to heparin at times of infection or inflammation.
Nearly 2 of 3 of centers used clopidogrel (Plavix) for their patients (11/17; 64.7%). There was no apparent relationship between perceived risk nor perceived seriousness, of stroke and clopidogrel use (results not shown). Four centers (23.5%) used direct thrombin inhibitors for their patients, and the agent of choice, when reported, was bivalirudin. The reasons for the use of alternative anticoagulants were not specified and all of those using direct thrombin inhibitors were low-volume VAD centers. Many centers (12/17; 70.6%) incorporated steroids into their regimes. Steroids were used “sometimes,” “frequently,” or “always” at 9 of 17 centers (52.9%). Three other centers commented on a suspected inflammatory response in the absence of infection as an indication for steroid use.
The majority of centers (14/15; 93.3%) deviated from the Edmonton Protocol in either timing of lab monitoring or type of tools used (Table 3). Seven reported modifying the timing of testing, and seven reported modifications in type of antithrombotic testing. Four centers modified both aspects of the protocol.
Most centers (11/17; 64.7%) used TEG/PM as the primary tool to monitor antiplatelet therapies. One of these centers used TEG/PM in combination with multiplate aggregometry. Two centers not using the TEG/PM used unspecified alternative means of platelet aggregometry testing. One center did not use any form of platelet monitoring. Another reportedly used urine thromboxane testing in conjunction with TEG/PM. Despite the predominant use of TEG/PM, it was never perceived to be “always accurate,” most reported that it is “sometimes accurate” (8/17; 47.1%) or “almost always accurate” (6/17; 35.3%).
Confidence in antithrombotic testing was a concern at most sites. Many (11/15; 73.3%) lacked confidence in lab results and the majority (8/17; 53.3%) expressed specific concerns about antiplatelet monitoring in particular (Table 4). Anticoagulant testing represented another area of concern, with 33.3% (5/15) of respondents lacking confidence in these tests. Logistics and timing of tests were of concern for 26.7% (4/15) of centers.
All but one center (16/17; 94.1%) used anti-Xa levels to monitor heparin effectiveness, though 6 of 17 (35.3%) used some combination of anti-Xa, TEG, and activated partial thromboplastin time (aPTT). The four centers using activated clotting times (ACTs) also used either an aPTT or an anti-Xa in conjunction.
An overview of the challenges in antithrombotic management is illustrated in Table 4. The greatest patient management concern was a lack of confidence in lab results, including TEG and platelet activity testing. There was also some reported difficulty with both under- and overanticoagulation at VAD centers (20% and 13% of respondents, respectively). Variability in clinician management was of concern at 2 sites, though interestingly these centers reported that one and three individuals managed therapies. One center specified postoperative hemostasis as a specific challenge.
Most respondents expressed concern about the likelihood of both bleeding and thrombotic risk at their respective centers, with 56.3% (9/16) reporting this combination to be the greatest perceived risk for VAD patients. Risk of ischemic stroke outweighed the perceived risk for hemorrhagic stroke risk (reported as a likely event by 56.3% versus 12.5% of centers, respectively).
Observations Among High- Versus Low-Volume VAD Centers
Between high- and low-volume centers, there was no statistically significant difference in the number of clinicians involved in antithrombotic decision-making (p = 0.665); use of clopidogrel (p = 0.669); nor degree of adherence to the Edmonton Protocol (p = 0.335) as illustrated in Table 5.
For heparin monitoring, there were statistically more low-volume centers using the aPTT in comparison with high-volume centers (9/17 vs. 0/4; p = 0.029). All 4 of these high-volume centers reported using anti-Xa levels for heparin monitoring instead of the aPTT. Although 76.9% (10/13) of low-volume centers rely on TEG/PM as the primary platelet monitoring tool, 75% (3/4) of the high-volume programs have abandoned the use (p = 0.099).
In this study, we demonstrate marked practice variation among antithrombotic providers for children on the Berlin Heart EXCOR VAD. Although there are frequent deviations, in particular regarding antiplatelet therapies, the Edmonton Protocol remains a cornerstone of antithrombotic management as illustrated by its use at 100% of VAD centers in this study. High-volume VAD centers use the anti-Xa assay for heparin monitoring and none currently use TEG/PM as the primary platelet monitoring tool. Beyond these, there were no other substantial management differences between high- and low-volume VAD centers. Concern was shared among all centers about the high risk of adverse events, predominantly ischemic stroke, which mirrors the findings in the Berlin IDE and compassionate use study cohorts in which 28–30% of patients suffered neurologic dysfunction, with a predominance of ischemic stroke.1,2
Pronounced practice variation in this cohort is not surprising, as this has been reported widely in non-VAD pediatric and adult anticoagulation literature.5–8 This is, however, the only study to date examining antithrombotic practice variation for children on ventricular assist devices, including stratification by high- and low-volume centers. Substantial practice variation has been reported in anticoagulation management and transfusion practices for pediatric extracorporeal membrane oxygenation (ECMO).7 Interestingly, only 8% of this ECMO cohort was on an alternative anticoagulant, and less than a quarter of VAD centers in our study had used alternative anticoagulants. Notably all of the centers using direct thrombin inhibitors at the time of this survey were low-volume VAD centers. We hypothesize that the experience of these groups (either positive or negative) had not been widely disseminated at the time of this study, as these therapies had neither been widely adopted or rejected. Based on recent personal communications, the proportion of patients on alternative agents such as bivalirudin is increasing at pediatric VAD centers, though this is not yet reflected in the literature, so these low-volume centers may represent early adopters of this technology. This once again highlights the need for improved collaboration and broad sharing of VAD experience among the community. Nonheparin alternatives, as well as the use of steroid, represent promising realms for future study based on preliminary single-center findings.9–11
A study of bridging practices for anticoagulated children with mechanical heart valves has also shown practice variability.8 These authors also noted that providers rely heavily on their own clinical experience to make practice decisions. In this study, 56.3% of respondents cited their own experience as their most commonly used “resource,” even in comparison to clinical practice guidelines and expert consultation.8 These findings suggest that much can be gained by sharing collective experiences within the field of pediatric VAD care.
This study clearly identifies several key areas of practice variation that almost certainly correlate with important knowledge gaps. First, centers take a highly variable approach in selecting antiplatelet agents. This was a prominent source of deviation from the Edmonton Protocol. The widespread use of clopidogrel, despite its minimal role in the original Edmonton Protocol, also suggests that there may be benefit in this approach, as supported by the recently published Stanford Modified Antithrombotic Guidelines.11 Second, there is no consensus as to best practice for monitoring antiplatelet therapies. Despite skepticism about the accuracy of TEG/PM in this survey, it is used pervasively (at 64.7% of centers). There does not appear to be a widely accepted alternative for platelet function testing based on this study. Technical considerations such as large required blood volumes and long turn-around times are among the challenges in platelet function testing for children. Finally, when compared with smaller centers, there are subtle differences in management at the high-volume VAD centers surveyed. The only statistically significant difference between high- and low-volume VAD centers was that high-volume centers more consistently use the anti-Xa assay to monitor heparin effectiveness. It is noteworthy that for platelet monitoring no high-volume centers rely exclusively on TEG/PM, although most low-volume centers continue to rely on this modality. Based on the assumption that high-volume VAD centers have the greatest cumulative experience, thus may be ahead of the learning curve in adjusting their practice to clinical outcomes, we expected to find more pronounced differences in management when comparing high- and low-volume centers. The lack of dissimilarity may be due to small sample size but may also reflect a background reliance on the Edmonton Protocol to guide the overall care of these patients.
It is striking that at half of all responding centers, only one or two providers manage all VAD AT decisions. This may be due to a limited number of providers with this expertise at most VAD programs but may also reflect some benefit in minimizing unwanted practice variation by reducing the number of care providers.
This study further highlights that the vast majority of centers are low volume, implanting fewer than four EXCOR devices per year. An important implication is that collaborative learning will be essential to improve patient outcomes in this setting. Another important finding is that best practice in selection and monitoring of antiplatelet therapies is unclear. Prospective study will be of vital importance moving forward in this regard.
The lack of granular detail accompanying survey responses is one of the key limitations of this study. Although this study was not designed to collect detailed practice-specific data and clinical outcomes, it does serve the purpose of identifying key areas of practice variation and important knowledge gaps, which can be addressed in a larger scale fashion. Of note also, half of all eligible centers participated in the survey so despite a diverse selection of high- and low-volume centers, these results may not be representative of the entire VAD community. Nevertheless, given that the proportion of high-volume centers in the survey nonresponders was similar to the proportion of high-volume centers in the study (15% vs. 23.5%), the study cohort may be characteristic of the heart failure community at large.
An additional limitation is the inability to control for patient factors that influence AT practice variation. The pediatric heart failure population is heterogeneous, and specific patient characteristics do necessitate some degree of individualized care. With a small, diverse population, it would be unrealistic to control for patient-specific factors, but there is value in a programmatic overview of general antithrombotic practice as illustrated by this study. By focusing on key domains in this survey, rather than detailed single-center practice, is it possible to provide an overview of current practice to identify areas for future standardization of care. Interestingly, in other antithrombotic literature, substantial practice variation exists even when controlling for patient factors in large studies.5
In summary, this study illustrates marked practice variation in antithrombotic management among those caring for patients on the Berlin EXCOR. Targeted study may enable standardization of practice in the realms of antiplatelet medications and monitoring, which represent considerable knowledge gaps. Collaborative learning may allow rapid adaptation of clinical practice based on outcomes in this small, diverse cohort of children.
1. Fraser CD Jr, Jaquiss RD, Rosenthal DN, et al.; Berlin Heart Study Investigators: Prospective trial of a pediatric ventricular assist device
. N Engl J Med 2012.367: 532–541.
2. Almond CS, Morales DL, Blackstone EH, et al. Berlin Heart EXCOR pediatric ventricular assist device
for bridge to heart transplantation in US children. Circulation 2013.127: 1702–1711.
3. Steiner ME, Bomgaars LR, Massicotte MP; Berlin Heart EXCOR Pediatric VAD IDE Study Investigators: Antithrombotic Therapy in a Prospective Trial of a pediatric ventricular assist device
. ASAIO J 2016.62: 719–727.
4. Wennberg JE, Barnes BA, Zubkoff M. Professional uncertainty and the problem of supplier-induced demand. Soc Sci Med 1982.16: 811–824.
5. Wasfy JH, Kennedy KF, Chen JS, Ferris TG, Maddox TM, Yeh RW. Practice variation
in triple therapy for patients with both atrial fibrillation and coronary artery disease: Insights from the ACC’s National Cardiovascular Data Registry. JACC Clin Electrophysiol 2016.2: 36–43.
6. Garcia DA, Ageno W, Libby EN, Bibb J, Douketis J, Crowther MA. Perioperative anticoagulation for patients with mechanical heart valves: A survey of current practice. J Thromb Thrombolysis 2004.18: 199–203.
7. Bembea MM, Annich G, Rycus P, Oldenburg G, Berkowitz I, Pronovost P. Variability in anticoagulation management of patients on Extracorporeal Membrane Oxygenation: An international survey. Pediatr Crit Care Med 2013.14:77–84.
8. Nguyen N, Sharathkumar A. Current perioperative anticoagulation practices in children with prosthetic mechanical heart valves. Congenit Heart Dis 2015.10: E210–E215.
9. Rutledge JM, Chakravarti S, Massicotte MP, Buchholz H, Ross DB, Joashi U. Antithrombotic strategies in children receiving long-term Berlin Heart EXCOR ventricular assist device
therapy. J Heart Lung Transplant 2013.32: 569–573.
10. Byrnes JW, Prodhan P, Williams BA, et al. Incremental reduction in the incidence of stroke in children supported with the Berlin EXCOR ventricular assist device
. Ann Thorac Surg 2013.96: 1727–1733.
11. Rosenthal DN, Lancaster CA, McElhinney DB, et al. Impact of a modified anti-thrombotic guideline on stroke in children supported with a pediatric ventricular assist device
. J Heart Lung Transplant 2017.36: 1250–1257.