Current Infectious Disease Reports

, Volume 14, Issue 4, pp 359–366

Ventricular Assist Device Infections

Authors

    • Infectious Diseases and MicrobiologyAlfred Health, and Monash University
  • Donald Esmore
    • Department of Cardiothoracic SurgeryAlfred Hospital and Monash University
Cardiovascular Infections (DP Devine, Section Editor)

DOI: 10.1007/s11908-012-0261-9

Cite this article as:
Spelman, D. & Esmore, D. Curr Infect Dis Rep (2012) 14: 359. doi:10.1007/s11908-012-0261-9

Abstract

Ventricular Assist Devices(VAD) are the commonest form of cardiac mechanical support, used as bridge to transplantation but also as destination therapy in non-transplant-eligible patients in whom transplantation is considered unsuitable based on age criteria. Infections are common and can significantly impact on patient outcome. Strategies to prevent and treat infections have not been assessed by randomised controlled trials, a difficult task due to the multiplicity of devices and their ongoing evolution. This review summarises the recent literature on infections in VAD-supported patients, including the recently proposed definitions, microbiology, diagnosis, treatment and preventive strategies.

Keywords

Ventricular assist deviceBacteraemiaDriveline infectionPocket infection

Introduction

It has been estimated that over 2 million people worldwide have endstage heart failure [1]. For many patients cardiac transplantation is life saving; however there remains a severe shortage of donor hearts. Estimates of the number of patients in USA with endstage heart disease who may benefit heart transplantation range between 15,000 and 60,000 [2, 3]. However the recent number of heart transplants in USA annually is between 2000 and 2500[4•]. Unfortunately for every 5 patients transplanted approximately one patient dies on the waiting list. In 2008 there were 441 deaths on the waiting list and 2163 patients transplanted [5]. An alternative therapy for such endstage heart failure evolved in the mid-80s when devices such as the Thoratec-PVAD, Novacor and Heartmate devices were successfully deployed as a “Bridge to Transplant.” As this therapy evolved and matured 60 % to 70 % of “bridged” patients were successfully transplanted.

In 2001, the efficacy of VAD therapy was demonstrated in the randomised controlled REMATCH study where endstage heart failure patients implanted with the Heartmate-XVE device demonstrated a 48 % reduction in the risk of death when managed with VAD compared with those managed by optimal medical therapy [6]. The United Network of Organ Sharing registry data have estimated that 9000 cardiac transplant candidates have undergone VAD support since 1999 [7]. Estimates of the number of devices implanted in the USA each year is approximately 1700, compared with 430 per year in Europe [8].

One of the major limiting factors for successful VAD deployment is the frequency and impact of infection in this patient group. The REMATCH study documented a 28 % probability of infection within 3 months of device implantation with sepsis being the most common contributor to death of patients in the VAD arm [6].

Ventricular Assist Devices

Clinically available Ventricular Assist Devices (VAD) include both external paracorporeal and implantable devices. External devices have two transcutaneous cannulae, the inflow connecting to the left ventricular apex and the outflow to the ascending aorta. Implantable devices have a similar configuration but being battery powered have a transcutaneous driveline to a drive module and external battery pack. Implantable devices are usually sited in the epigastrium in the pre-peritoneal space but more recently directly attached to the apex of the failing heart [9, 10].

The earlier pulsatile VAD have been largely replaced by smaller continuous flow axial or rotary devices. Pulsatile devices have been generally larger in size with more rigid drive lines, and are volume-displacement i.e. “rocking,” factors that may make infection more likely [11]. Continuous flow devices are reportedly associated with lower incidence of infection, findings that may be explained by their smaller, simpler structure with less surface area [12•, 13•],a minimum of moving parts [14], limited device movement per se and a smaller and more flexible driveline[15].

Ongoing developments include the use of even smaller pumps which can be placed in the pericardial space avoiding the need for abdominal surgery and more recently an “intraventicular” assist device implanted through a small thoracotomy [16, 17].

Definitions and Infections in VAD Supported Patients

Infection in a patient with a VAD include those infections that are device related as well as infections not specifically related to the device. Previously there have been no consensus standard definitions for VAD associated infections. Multiple definitions have therefore been used thereby confounding attempts for meaningful comparison between reports. A significant recent advance has been the publication in 2011 of proposed standard definitions of infections in patients with VAD by the International Society for Heart and Lung Transplantation (ISHLT) [18••].

The ISHLT proposal divides infections into three groups. Firstly, VAD-specific infections are those involving any part of the device hardware including the pump, cannula, anastomoses, pocket, drive line, or anatomical spaces where these components are sited. The second group of infections are VAD-related infections and include blood stream infections (BSI), mediastinitis and surgical site infections. The third group of infections are those that are not related to the presence of VAD, such as urinary tract infections and pneumonia [18••]. Despite these proposed definitions, remaining challenges include the diagnosis of VAD-related BSI in the presence of a central venous catheter [18••]. Confirmation of VAD related-BSI may require support from the technique of differential time to positivity or require molecular typing methods which are not widely available [18••].

The ISHLT authors emphasise the need for validation of the proposed definitions which may require future modifications. The ultimate goal of such definitions is to allow meaningful comparisons between reports and facilitate the identification of infection risk factors to then direct interventions to optimise infection management and prevention.

A further obstacle to meaningful comparison between studies is the use of infection prevalence or attack rate rather than the more informative incidence rate or infections per unit of support time. Prevalence does not take into account the duration of VAD support. Some authors estimate that the prevalence of infection has remained stable while the duration of support has significantly increased indicating that the incidence rate has decreased [19, 20].

Attempted comparisons may also be confounded by variations in device type, implantation surgical technique and differing investigation modalities used for infection diagnosis. Common limitations remain single institution reports, use of historical data and the absence of a control arm.

Infections in Patients with VAD

Two studies published in 2001 and 2003 summarised a total of 21 studies with prevalence of infection ranging from 4.5 % to 72 % [20, 21]. More recent studies from 2007 to 2010 have reported a prevalence between 52 % and 76 % [13•, 22, 23]. Genovese, reporting infection only in the first 60 days following implantation, found that 42 % of 195 VAD recipients had infection in that time period [24]. In a study of a paediatric population 63 % of 25 patients experienced an infection with 44 % having a septic episode [4•].

Few recent reports include infection rates per 100 patient months [25•], or patient years [26•, 27•] or 1000 support days [28]. Slaughter in 2009 demonstrated a significant decrease in device related infections (from 0.90 to 0.48 events per patient year), non device related infections (from 1.33 to 0.76 events per patient year) and sepsis (from 1.11 to 0.39 events per patient year) with continuous flow devices compared with pulsatile devices [27•]. These authors found that the year of implantation was predictive of outcome with more recent implantation associated with improved freedom from bacteraemia, driveline and pump pocket infection. They suggested that decreased infections with continuous flow devices were likely related to increased physician experience. The Interagency Registry for Mechanical Circulatory Support (INTERMACS) collects data from more than 90 institutions and reports adverse events including infection in VAD recipients. In the second INTERMACS 2010 annual report on 1092 VAD implanted between June 2006 and March 2009 there was a significantly lower rate of infection adverse events with continuous flow compared with pulsatile flow devices (11.8 v 28.29/100 patient months) [25•].

Blood Stream Infections

In 2001 Gordon summarised 7 previous studies of blood stream infections (BSI) in patients with VADS in which the prevalence ranged from 16 % to 49 % [21]. Three more recent studies published in 2009 and 2010 documented BSI attack rates of 17.9 %, 43.4 % and 76.8 % [23, 24, 29]. Also the INTERMACS registry [12•] reported that 32 % of infective episodes in 593 patients from 88 institutions were related to positive blood cultures. Apart from these reports of prevalence, actual rates per 1000 support days have also been reported. Gordon reported an overall incidence rate of 5 (range 1.8–8.4)/1000 support days in 7 studies mentioned earlier [21]. A subsequent publication by Simon reported a VAD-related BSI rate of 3.1 per 1000 VAD support days [30]. More recently Schaffer has reported BSI incidence rates of 0.42 events per year for continuous flow devices and 0.98 events per year for pulsatile flow devices [26•].

Drive Line Infections

Four recent reports from 2009 to 2010 documented a drive line infection (DLI) prevalence of between 17.4 % and 21 % [12•, 13•, 24, 31]. Morshuis [32] reported an event rate of 0.34 per person years, with 6 infections in 5 of 33 patients and a prevalence of 15 % [32]. Zierer, who studied DLI which developed at least 30 days after implantation, demonstrated the importance of duration of followup [3]. These authors reported that 23 % of patients at a median time of 158 days after implantation developed DLI, but importantly the DLI risk progressively increased with the duration of support, reaching 94 % at 1 year.

Pump Pocket Infection

The prevalence of pump pocket infection (PPI) in two recent reports is 7 % and 15 % [12•, 31]. However some authors combine PPI with mediastinitis and have reported an aggregate prevalence of 6.1 % and 15.1 % [24, 29]. Significantly higher PPI prevalence has also been reported: in the implantable-Lionheart European experience 35 % of 23 patients developed PPI [22]. Monskowski in 2006 reported a rate of 42 % with 25 pocket infections in 60 implant recipients [33]. In this report PPI was defined as “purulent fluid within the VAD pocket or a microbe cultured from the pump pocket at explant.” It is unclear whether a positive culture in the absence of clinical or operative evidence of infection was included as a PPI or considered a contaminant and excluded. Mouskowski reported that many (11 of 25) patients who had pocket infections also had driveline infections [33].

Endocarditis

There have been a variety of definitions used for VAD associated endocarditis. Simon used culture of the internal surfaces of explanted devices as well as histopathological evidence for the diagnosis of VAD associated endocarditis [30]. High rates of positive bacterial cultures (35 % to 48 %) of explanted devices have been reported [30, 34]. It is uncertain whether these isolates indicate infection or contamination. There is a paucity of data available on more modern devices.

The recent ISLHT definitions outline Transthoracic Echocardiography (TTE) abnormalities of VAD-related endocarditis with a vegetation “adjacent to or in the outflow cannula, or in an area of turbulent flow such as regurgitant jets on implanted material” [18••]. Recent reports suggest VAD-associated endocarditis is rare [35].

Pathogenesis and Host Risk Factors

Infection pathogenesis may involve one or a combination of patient, surgery, device or bacterial factors. These have previously been outlined and are summarised in Table 1. Direct bacterial inoculation of any part of the device intra- or post-operatively or from metastatic BSI can occur. Local infection may remain uncomplicated or spread further to other parts of the device or give rise to BSI.
Table 1

Risk factors for infections in patients with VADS

Risk factor

First Author

Older agea

Hermann [34]

Longer Ventilation

Haemodialysis

Malani [56]

Hypo-albuminaemia

Poston [57]

Length of ICU

Reoperation

Preoperative renal diseasea

Holman [51]

Poor nutrition

Gordon [1]

Immuno-suppression

Indwelling catheter

Prolonged hospital stay

Diabetesa

Simon [30]

Cardiogenic shock

Holman [11]

Biventricular failure

Cardiopulmonary bypass time

Saito [29]

Pulsatile device

Total parenteral nutritionb

Aslam [37]

aTopkara [13] found that these factors were not predictive of infection

bAslam [37] predictive of fungal infection

Microbiology of Infection in Patients with VAD

Recent data from INTERMAC outlined that most infections following pulsatile VAD implantation are bacterial (87 %) followed by fungi (9 %) with the remainder consisting of viral(1 %), protozoal (0.3 %) or unknown(2 %) [36].

VAD related infections are commonly caused by gram-positive cocci, frequently Staphylococcus aureus and Coagulase negative staphylococci (CNS), most often arising from the patients’ own commensal or acquired flora. Hospitalised patients are at risk of acquiring antimicrobial resistant hospital flora including methicillin resistant Staphylococcus aureus (MRSA), Vancomycin resistant enterococci (VRE) and resistant gram-negative bacteria such as Acinetobacter and pseudomonas [28].

Different spectrums of organisms have been identified at different sites. Schaffer in 2011 reported that the common organisms causing BSI were coagulase negative staphylococci (45 % of cases), Candida species (13 %) and S. aureus (11 %); for DLI S. aureus (44 %), pseudomonas (28 %) and Serratia (9 %); for PPI coagulase negative staphylococci (24 %), enterococcus (24 %) and S. aureus (19 %) [26•]. Similarly Topkara reported that staphylococci and pseudomonas were the common pathogens found with both driveline and pump pocket infections [13•]. Two recent reports confirmed polymicrobial DLI and one study reported multiple organisms from PPI [13•, 33].

The prevalence of fungal VAD infections in recent reports varies. Bagdasarin reported 7 cases of VAD associated fungemia in 292 VAD recipients between 1996 and 2009 at a rate of 0.1 infections/1,000 days of device support [28]. In contrast Aslam in 2010 reported a retrospective chart review of 300 VAD recipients [37]. One hundred and eight (36 %) patients developed infection and of these 21 % were fungal in aetiology, mostly candida species. Of the 23 fungal infections there were 15 BSI, 6 DLI, 5 PPI and 5 VAD related endocarditis. Eight of the 23 (34.8 %) were polymicrobial infection [37].

Clinical Evaluation and Diagnosis

A comprehensive clinical history and physical examination is essential [1, 18••]. It is critical to obtain suitable specimens e.g. blood cultures, swabs and aspirates before initiation of antibiotic therapy. These aspects are outlined in a recent review [18••]. The clinical presentations of VAD related infections have previously been described [18••, 20].

If a PPI is considered possible, then ultrasound imaging and guided aspiration should be considered [1]. Most pockets contain fluid and its presence does not necessarily indicate infection. However the absence of, or a minimal amount of, fluid would indicate that another site of infection is likely [20]. Ultrasound may also demonstrate fluid around the driveline or cannula. Of the other imaging modalities, MRI is contraindicated. A labelled white cell scan may define infection of the device including the driveline tract anatomy or infection at another site [1, 3, 20].

Two recent studies have reported experience with CT and leucocyte SPECT/CT. It has been considered that the value of CT may be limited by the presence of artifact [1]. However Carr in 2010 reported the experience of CT in 23 of 42 VAD recipients and concluded that this modality may help visualise areas not seen by other imaging [38]. There is a recent preliminary report of the use of leucocyte SPECT/CT for detection of suspected VAD infections suggesting that this modality may be valuable in defining the location and extent of infection [39].

Timing of VAD Related Infections

The highest incidence of infection occurs in the immediate post operative period i.e. in the first few weeks or months [13•, 20, 29]. Topkara found that the majority of non VAD related infections e.g. pneumonia and urinary tract infection occurred before discharge whereas most VAD associated infections including BSI, DLI and PPI occurred following discharge [13•]. There is one report that only 15 % of driveline infections occurred in the first 30 days with the majority occurring thereafter [3]. In a study of BSI the time interval between the device implant and BSI was 23 days with 28 days for fungi, 24 days for gram positive cocci and 19.5 days for gram negative bacilli [21]. There was no statistically significant difference between these time intervals. Similarly Aslam in 2010 also found that there was no statistically significant difference in the time interval between VAD implantation and the occurrence fungal compared with bacterial infection [37].

Treatment

Initial management is guided by the suspected site of the infection, the severity of the clinical illness, and the likely organism with consideration of the institutions microbial resistance pattern. Patients should be reviewed by cardiothoracic, cardiology and infectious diseases specialists. Treatment often requires aggressive antibiotic therapy with surgical drainage of an abscess or driveline wound revision. Empirical antibiotic choice should include cover against resistant gram-positive organisms (eg vancomycin) and resistant Gram negative organisms (eg aminoglycoside or a broad spectrum beta-lactam agent). Isolation of the causative organism will then direct subsequent antibiotic therapy. Biofilm active agents such as rifampicin and fluoroquinolones should be considered. For therapy of antibiotic resistant gram positive infection following VAD insertion, there is a recent report of successful therapy with the newly available daptomycin [40]. The optimal treatment of fungal infection is uncertain with the efficacy of the newer agents such as lipid formulations of amphotericin and echinocandins yet to be studied.

The optimal duration of antibiotics is unknown. Some authors suggest that VAD related infection should have antibiotic management similar to a patient with prosthetic valve endocarditis with a course of at least 6 weeks. Simon reported fewer relapses, especially in patients with S. aureus VAD-related BSI if treated with continuous antibiotic therapy before, during and after transplantation compared with those patients treated with short term antibiotics [30]. Prolonged courses of vancomycin to treat MRSA bacteraemia has resulted in the development of heterogeneous intermediate S aureus (VISA) infection [41].

For PPI some authors consider optimal therapy may require explantation and/or transplantation [3, 42]. However, this is not always a realistic or viable option. Drainage of an abscess, either percutaneously or by surgical incision may be indicated [43]. Some infections have been successfully treated by surrounding the device pocket in omentum or a muscle flap or with the use of antibiotic impregnated polymethylmethacrylate [21, 43]. Omentoplexy has also been used to treat mediastinitis following VAD implantation [44].

DLI are often managed with wound care, local irrigation, immobilisation of the driveline exit site and intravenous antibiotics targeting the causative organism [1, 3, 29, 44]. However, relapse is common and this may then warrant long term suppressive antibiotics. Local treatment with hydrogen peroxide, chlorhexidine [1, 3, 20, 29] or sterile honey local application have also been used. Excisional debridement of the exit site proximally until the tissue is adherent to the driveline may be required [3] There are also recent reports of vacuum assisted systems in the successful management of driveline infections [13•, 45].

Treatment Outcome

The impact of infection has been variably reported in terms of mortality, duration of support, successful bridge to transplantation and use of hospital resources. Generally most infections can be cured or successfully contained until transplantation [20]. Infections generally result in frequent patient nursing visits, prolonged intravenous antibiotics, hospital admissions, attendant surgical interventions, transfusions with blood or blood products and overall compromises the patients quality of life [13•, 46].

The most common cause of death in the REMATCH study was sepsis contributing to 41 % of the deaths [6]. Since that study mortality due to VAD associated infections has markedly decreased, therapy now having moved from larger problematic volume deplacement devices to smaller continuous flow pumps. In a 2009 report infection was the primary cause of death in only 7.7 % of patients including 2.4 % of patients within the first month following implantation and 11.3 % of patients >1 month following implantation [47].

There are conflicting reports on the impact of VAD infection on survival. Although Gordon [21] had reported that VAD associated BSI was associated with death on the device, other earlier studies reported that long term survival was similar in patients with or without VAD related infection [30, 46, 48]. However a number of recent reports contradict these findings. Topkara found that VAD sepsis was associated with increased mortality at 2 years compared with those without sepsis (61.9 % v 18 %) [13•]. Holman reported from the INTERMAC registry data that infection did result in a significant adverse impact on survival with those patients with infection during the first month after implant having significantly worse survival than patients who had their first infection in the second or third month after implant [12•].

Significant increased mortality was found in patients with fungal infections following VAD implantation [37]. These authors reported an all cause mortality of 91 % and a VAD-related mortality of 65 % in this population. Consistent with these findings, Gordon had previously reported that when BSI were stratified by pathogen, fungemia had the highest hazard ratio for death on device (ie 10.9) [21].

Patients with VAD related infections have longer support time than patients without infections and in some reports a longer waiting time to transplantation [30, 33, 49]. This was true of DLI, PPI and endocarditis [33]. Also, patients with sepsis are less likely to be successfully “bridged” to cardiac transplantation, however in this report transplant survival rates at 1 year are unaffected [50•]. Pre transplant VAD-related infection can recur following cardiac transplantation often with the same causative organism, but now in the immunosuppressed post transplant environment [33].

The development of DLI or PPI had no impact on survival in report by Topkara [13•]. In a further report, the presence of DLI did not influence the frequency of proceeding to subsequent cardiac transplantation [3]. However Monkowski reported shorter median survival in those with PPI (350 days) compared with those without infection (>2400 days) [33]. VAD related endocarditis was associated with worse survival [33].

Prevention

Multiple strategies to prevent infections have been used and/or recommended. Unfortunately the paucity of clinical studies of such strategies makes assessment of their efficacy difficult.

Pre-Operative Measures

The avoidance of prolonged hospitalisations prior to VAD implantation diminishes the risk of colonisation with antibiotic resistant organisms and hospital acquired infection [13•, 51]. Pre operative nasal mupirocin has been used in an attempt to eliminate nasal staphylococcal carriage and therefore subsequent staphylococci infection [1, 20]. In a 2011 survey, Walker and his colleagues reported that 61.9 % (13 of 25) of centres that responded used a decolonisation strategy, either nasal mupirocin or chlorhexidine washes or both [52].

Antibiotic Prophylaxis

Systemic antibiotics are routinely given before and up to 48 h to 72 h post operatively [1, 30]. Some authors suggest continuing the antibiotic prophylaxis until removal of tubes but there is no evidence to support this longer use which would increase the risk of antibiotic resistant super infection [53].

There is no standard antibiotic prophylaxis choice. Vancomycin, levofloxacin (or trovafloxacin) and rifampicin were used in the original REMATCH study [6]. Different antibiotic regimes have subsequently been published [1, 3, 13•, 26•, 30, 53].

Walker undertook a survey of centres performing VAD surgery [52]. Of the 23 centres that responded 42.9 % were using a 4 drug regimen, 23.8 % were using a 3 drug regimen and the remainder were using a 1 or 2 drug regimen. Most centres were using vancomycin and rifampicin and the majority of the 3 and 4 drug regimens included fluconazole. A gram-negative antibiotic, either a fluoroquinolone, cefepime, ceftazidime or piperacillin—tazobactam, was also commonly used. Extended antifungal prophylaxis has been used in the setting of a high incidence of fungal infection [1]. Nystatin has been given when there is a risk of oral or oesophageal fungal infection [20]. In Aslam’s recent report, the use of anti-fungal prophylaxis (mostly fluconazole) was not associated with decreased fungal infections [37]. Herrandez has reported on the in vitro effectiveness of dipping VAD into solutions containing amphotericin, but the clinical use is yet to be reported [54].

Prophylactic regimens may change according to local antimicrobial resistance patterns and whether the individual patient has or has had known infection or colonisation with specific bacteria. Antibiotic dosing protocols should ensure optimal dosing and administration timing of within 1 h prior to surgery, consideration of a second antibiotic dose during prolonged surgery and limiting duration to no longer than 48 h. Vancomycin use in the setting of renal dysfunction may require serum monitoring with dose modification.

Operative Measures

Detailed intra-operative infection prevention guidelines have previously been published [1, 20, 43]. Specifically a number of possible strategies have been suggested to protect the pump and driveline from bacterial contamination intra-operatively. These include wrapping both the device and the driveline in pads soaked in antibiotics such as vancomycin/gentamicin, soaking the driveline velour in an antibiotic solution prior to implantation and the use of antimicrobial impregnated driveline coatings to prevent infection and facilitate driveline and tissue integration [11, 20, 55]. N-acetylcysteine has also been recommended in an attempt to minimise surface biofilm production [54]. None of these strategies are in common usage.

Driveline management is considered of paramount importance. Tunnelling of the driveline to the contralateral side, to give a longer subcutaneous course, is recommended with the skin incision that is snug around the driveline [12•]. There should be immediate immobilisation of the driveline to avoid pulling and limit movement, thereby facilitating integration with tissue [3, 12•].

Post Operative Measures

Early extubation and removal of catheters and tubes is recommended [51]. Patient nutrition should be optimised to improve wound healing [20]. Because of the reported association between total parenteral nutrition and fungal infection in VAD recipients’, enteral feeding is preferred [37].

Continued driveline immobilisation to prevent movement back and forth at the exit site will facilitate integration or tissue in growth at the exit site [36]. An elastic abdominal binder and drain tube attachment device has been used for this purpose [3]. The driveline exit site should undergo daily antiseptic cleansing: hydrogen peroxide, chlorhexidine, silver impregnated dressings such as silver sulphadiazine (SSD) and povidone iodine solutions have been used for this purpose [1, 3, 13•, 20, 36]. Patients and their families need instruction on optimal exit site and driveline care and should be encouraged to report bleeding or discharge at the driveline site [12•]. If there was adequate driveline exit wound site healing after 30 days, Holman considered showering an option, but bathing and swimming was not allowed [51].

Chinn and his colleagues recommended antibiotic prophylaxis for VAD recipients undergoing procedures that may generate transient bacteraemia such as dental extractions and colonoscopy [43].

Conclusions

Although first used over 20 years ago, VAD therapy has evolved as a powerful and increasingly viable therapy in the management of endstage heart failure. Increasing success with the elderly non-transplant eligible destination therapy patient cohort remains an area of ongoing development, with clinical deployment increasing world-wide. “Bridge” numbers remain relatively static with ~80 % of patients being successfully transplanted even if their clinical course has been confounded by attendant infections. Infections however in the permanently—supported DT patient is a major threat to the supported patients’ quality of life, serious adverse event profile, resource consumption and attendant survival. Strategies to prevent and treat such infections have not been assessed in comparative studies. The recent formulation of consensus definitions will hopefully allow for objective comparisons between different institutions.

Disclosure

No potential conflicts of interest relevant to this article were reported.

Copyright information

© Springer Science+Business Media, LLC 2012