By Stan Deresinski, MD, FACP, FIDSA

Clinical Professor of Medicine, Stanford University

SYNOPSIS: Almost two-fifths of patients with left ventricular assist devices develop associated infections. The optimal appropriate management remains to be determined.

SOURCE: Blanco-Guzman MO, Wang X, Vader JM, et al. Epidemiology of left ventricular assist device infections: Findings from a large nonregistry cohort. Clin Infect Dis 2021;72:190-197.

Blanco-Guzman and colleagues retrospectively reviewed cases of left ventricular assist device (VAD) infections seen at a single center from 2009 to 2015. Of the 455 who had had continuous flow left VAD placement during that time, 174 patients (38.6%) developed infection an incidence of 36.9 per 100 patient-years of VAD support. Infections occurred 5-2,266 days (median 150 days) after placement. The infections were considered VAD-specific and VAD-related in 146 (83.9%) and 28 (16.1%), respectively. VAD-specific infections involved the device hardware or the body surfaces containing the hardware and included those involving the pump or cannula, driveline, or pocket. VAD-related infections are those including endocarditis, mediastinitis, and bloodstream infections (BSI) that can occur in the absence of the device, but that are seen more commonly in patients with VADs.

Two-thirds of VAD-specific infections involved the driveline, while one-third of the VAD-related infections were BSI. Local infections (driveline or pocket infections, or mediastinitis without BSI) accounted for 108 (62.1%) of the cases and endovascular infections (pump/cannula infection, endocarditis, BSI) accounted for 66 (37.9%) of the cases. Of importance, 29.2% of patients considered to have endovascular infection did not meet systemic inflammatory response syndrome (SIRS) criteria on their initial evaluation.

Computed tomography (CT) was performed on 142 patients (81.6%) and was normal in 37.7%. Driveline stranding was the most common finding, seen in 70 patients (49.3%), while a drainable fluid collection was identified in 15 patients (10.6%). Infection was caused by Gram-positive bacteria in 62.6%. Approximately one-fifth of endovascular infections were caused by methicillin-susceptible Staphylococcus aureus or Enterococcus. Gram negatives accounted for 15% of infections and with similar frequencies in local and endovascular infections. Candida (all non-albicans) was associated with six infections, five of which were endovascular.

The median duration of antibiotic therapy for both local and endovascular infections was six weeks. Almost three-fourths of the 155 patients who survived to completion of their initial antibiotic therapy were started on antimicrobial suppression (84.3% with endovascular and 66.3% with local infection). Approximately one-fifth underwent a surgical procedure to control infection most commonly driveline revision and debridement of the mediastinum or pocket. The median survival time from initiation of infection was 28 months, but it was only 14 months for those with endovascular infection compared to 35 months with local infection.


This study confirms the known remarkably high incidence of infection in left VAD patients, which occurred in 38.6% at a rate of 36.9 per 100 patient-years in this cohort. The majority were caused by Gram-positive organisms. Candida infection occurred in only six patients, five of whom had an endovascular source.

An important observation was that almost one-third with infections involving the pump, cannula, or bloodstream with or without endocarditis did not meet SIRS criteria at the time of initial evaluation, suggesting that the diagnosis often may be delayed. Almost one-third lacked fever and leukocytosis. Measurement of procalcitonin levels is not helpful during the first weeks after VAD placement since it becomes elevated, often markedly, in the absence of infection in the first days after the procedure before resolving toward normal by 14-30 days.1

CT scans were helpful in localizing some local infections and assisted in allowing drainage of fluid collections, but were less sensitive than desired. I would guess that positron emission tomography/CT would be a better diagnostic and localizing procedure, albeit at greater cost.

The choice of antibiotic therapy is dictated by the microbiological results, but the duration of therapy is poorly defined in many of the infections seen in association with left VADs. Even more poorly defined are the role and duration of suppressive therapy. Clearly, we need clinical trials to provide guidance in the optimal prevention and management of these infections, as well as effective means to detect infections at onset and to define their location.


  1. Kettner J, Holek M, Franekova J, et al. Procalcitonin dynamics after long-term ventricular assist device implantation. Heart Lung Circ 2017;26:599-603.