OUP user menu

Contemporary management of and outcomes from cardiac device related infections

Ronan Margey, Hugh McCann, Gavin Blake, Edward Keelan, Joseph Galvin, Maureen Lynch, Niall Mahon, Declan Sugrue, James O'Neill
DOI: http://dx.doi.org/10.1093/europace/eup362 64-70 First published online: 12 November 2009


Aims To describe the incidence and management of cardiac device infection. Infection is a serious, potentially fatal complication of device implantation. The numbers of device implants and infections are rising. Optimal care of device infection is not well defined.

Methods and results We retrospectively identified cases of device infection at our institution between 2000 and 2007 by multiple source record review, and active surveillance. Device infection was related to demographics, clinical, and procedural characteristics. Descriptive analysis was performed. From 2000 to 2007, a total of 2029 permanent pacemakers and 1076 biventricular/implantable cardioverter–defibrillators (ICDs) or ICDs were implanted. Thirty-nine cases of confirmed device infections were identified—27 pacemaker and 12 bivent/ICD or ICD infections, giving an infection rate of 1.25%. Median time from implant or revision to presentation was 150 days (range 2915 days, IQR25% 35–IQR75% 731). Ninety percent of patients presented with generator-site infections. The most common organism was methicillin-sensitive Staphylococcus aureus (30.8%), followed by coagulase negative Staphylococcus (20.5%). Complete device extraction occurred in 82%. Of these, none had relapse, and mortality was 7.4% (n = 2/27). With partial removal or conservative therapy (n = 13), relapse occurred in 67% (n = 8/12), with mortality of 8.4% (n = 1/12). Median duration of antibiotics was 42 days (range 47 days, IQR25% 28–IQR75% 42 days). Re-implantation of a new device occurred in 54%, at a median of 28 days (range 73 days, IQR25% 8.5–IQR75% 35 days). Methicillin-Resistant Staphylococcus Aureus infection predicted mortality (P < 0.004, RR 37, 95% CI 5.3–250). Median follow-up was 36 months.

Conclusion Cardiac device infection is a rare complication, with significant morbidity and mortality. Complete hardware removal with appropriate duration of antimicrobial therapy results in the best outcomes for patients.

  • Pacemaker infection
  • Implantable cardioverter–defibrillator infection
  • Cardiac device infection
  • Cardiac-device endocarditis
  • Endocarditis


The use of permanent pacemakers (PPM) and implantable cardioverter–defibrillators (ICDs) continues to grow worldwide1,2 with over 3 million pacemakers and 250 000 ICDs in use.3 The rate of device implantations is increasing with aging of the general population and expanding indications.1,2 Similar to other prosthetic materials, infection complicates a small proportion of patients with these devices.

Along with the increase in device implantation, the incidence of device infection has also been increasing, but at a faster rate.4 In the USA, data from Medicare recipients from 1990 to 1999 showed an increase in the number of device infections from 0.94 per 1000 recipients to 2.11 per 1000 recipients, an increase of 124%.5 Other published reviews suggest the estimated rate of infection of endocardial leads is between 1 and 2%, but varies from 0.13 to 12.6%.6,7 The exact reason for this increase remains unexplained, but it is hypothesized to be due to increasing co-morbidities in device recipients, improved surveillance and detection of cardiac device infection (CDI), and improving survival of patients with devices.8

The clinical presentation of device infection ranges from superficial wound infection to frank device-related endocarditis, occurring in 10%.8 The condition is associated with substantial morbidity and mortality, which ranges in the published studies from 8% in those with complete device removal, up to 46% in those in whom the device is not explanted.9

It also results in significant financial costs, including prolonged hospitalization, prolonged antimicrobial therapy, management of systemic complications of sepsis, and the costs involved in device extraction and potential re-implantation. According to one US study, the mean hospital cost for treating a single PPM or ICD infection is $24 459 or $57 213, respectively.10

Data to guide treatment of patients with this condition are limited. However, the consensus from the published literature recommends prompt and complete device removal, combined with antimicrobial therapy for an appropriate duration.1113

To establish the incidence, management, complications, and outcome of CDI, we retrospectively reviewed all cases of confirmed CDI at our institution from 2000 until 2007.


We identified all patients who underwent pacemaker or ICD implantation at our institution between 2000 and 2007. Cases of CDI were identified by a retrospective search of the computerized hospital in-patient discharge database. The search terms ‘CDI’, ‘pacemaker infection’, ‘ICD infection’, ‘pacemaker–endocarditis’, and ‘ICD endocarditis’, were searched for. This generated a list of cases of potential CDIs. This list was crosschecked with procedural logbooks to ensure complete case identification. The hospital discharge database identified cases correctly in 93% of cases. In the past 3 years, active surveillance following device implantation has been carried out, and we used this as a secondary source to confirm case identification. Medical record review was then performed on all potential cases. Patients who fulfilled criteria for CDI were included in this analysis. A second database of all device implantations was created to allow comparison of those who developed CDI with those who did not. Patient confidentiality was protected at all times, and local ethical practice guidelines were adhered to throughout. The local ethics committee waived the requirement of informed consent.

Cardiac device infection was defined according to guidelines from the Mayo Clinic Group12 and only cases meeting these criteria were included.

Clinical evidence of device infection included local signs of infection or inflammation at the generator site, wound dehiscence, erosion, or purulent discharge. Presence of a PPM or ICD-related endocarditis was confirmed when valvular or lead vegetations were detected by echocardiography, or if the Duke criteria for infective endocarditis were met. A CDI was microbiologically confirmed based on positive cultures from the generator site, lead(s), and/or blood (in the presence of local infective signs at the generator, no other cause of bacteraemia, and resolution of the infection with device removal). Relapse was defined as recurrence of the device infection with the same organism, confirmed on a similar antibiogram.

Baseline demographic data, co-morbidities, device information, time from last implantation or revision, antibiotic prophylaxis use, symptoms and signs, laboratory data, microbiological data, antibiotic management, and device extraction information were recorded on all patients. Outcomes and relapse were also recorded.

Statistical analysis

Categorical measures were summarized by using counts and percentages; continuous variables were summarized by using either means with standard deviations or medians with an inter-quartile range, depending on data skew. Univariate comparison between continuous variables was performed with the Mann–Whitney U test, and for categorical data, comparison was performed with the Fisher's exact test. All univariate predictors were then considered candidate predictors for a logistic regression model, using the enter method. To avoid multicolliniarity, number of leads implanted was excluded as this correlates highly with type of device implanted. All analyses were performed using SPSS, version 12.1, software.


Between 1 January 2000 and 31 December 2007, a total of 2029 PPM (1295 single chamber, 569 dual chamber), and 1076 bivent/ICD or ICD systems were implanted at our centre. During this same time frame, we identified 39 cases that met our criteria for a CDI. This represents 1.25% of total number of devices implanted.

Baseline demographics and clinical characteristics

Median age of patients with a CDI was 69 (range 26–89, IQR25% 60–IQR75% 78). Sixty-nine percent were male. Devices included 27 PPM and 12 ICD (bivent/ICD 6 and ICD 6). Most patients had single-chamber devices (56%) (Table 1). Transvenous placement was the most common lead insertion technique used. The pulse-generator was located in the left prepectoral area in 100%. Median fluoroscopy time was 3.38 min (IQR25% 2.30–IQR75% 9.26, range 1–32.5). Indications were according to contemporaneous ACC/AHA/ESC guidelines for device implantation. Ten percent of patients had temporary pacing in place at the time of original device implantation or revision, and three (7.7%) had a fever documented within the previous 24 h before device revision or implantation. A CDI occurred in 54% following initial de novo device implantation, and in 46% following revision or generator replacement. In the majority of cases, local instillation with intravenous neomycin solution to the pocket site was performed, until this became commercially unavailable from 2006. All patients received antibiotic prophylaxis with a first generation cephalosporin (unless penicillin allergic, in that situation, they were treated with erythromycin) at the time of skin incision and pocket creation and for 48 h following implantation. Local installation of neomycin antibiotic solution into the pocket before closure was performed until late 2006, when this product was no longer commercially available. Since 2006, an active surveillance programme for the detection of Methicillin-Resistant Staphylococcus Aureus (MRSA) has been undertaken. The rate of diabetes, renal impairment, and corticosteroid use were 28, 43, and 15%, respectively. All implantations were performed by five operators, each performing >150 device implants per annum, not all of whom were present from the study outset.

View this table:
Table 1

Baseline demographics and clinical characteristics

Clinical presentation with cardiac device infection

The median duration from device implant or revision to presentation with confirmed CDI was 150 days (range 5–2920 days, IQR25% 35–IQR75% 731 days) (Table 2). Fever, chills and malaise were the most common presenting symptoms. Evidence of generator site inflammation was present in 90%. Frank erosion or purulent discharge could be identified in 66%. A third of cases met diagnostic criteria for cardiac device related endocarditis. The majority had non-specific laboratory abnormalities, including elevation in leukocyte count, anaemia, elevated erythrocyte sedimentation rate, or C-reactive protein level.

View this table:
Table 2

Clinical symptoms and signs and microbiologic results

Microbiological findings

Out of 39 CDI cases, a causative organism could be identified in 62%. The most frequent causative organism was Methicillin Sensitive Staphyloccoccus aureus (30.8%), followed by Coagulase negative Staphylococcus (20.5%), and Streptococcus species (7.7%). The rate of MRSA infection was 5.1%. Blood cultures were performed in 84% of the CDI group, and were positive in 54% of these cases. Cultures of generator site tissue and lead tips were performed in those undergoing device extractions (82%) and were positive in 38 and 18%, respectively. In cases where all three swabs were positive, the same causative organism was identified in each case. Of those patients in whom blood cultures were negative, all had already received antibiotic therapy by the time cultures were drawn.

Echocardiographic findings

Most (87%) patients underwent transthoracic echocardiography during their admission. Thirty-six percent also underwent transoesophageal echocardiography. In those in whom echocardiography was performed, vegetations were identified on the lead in 18%, and involving the heart valves in 5%. The tricuspid valve was the only valve involved.

Device extraction, replacement, and antibiotic management

Eighty-two percent underwent device explantation at the time of their presentation with a CDI. Five (12.8%) had relapsing CDI and underwent device removal after failure of conservative therapy.

The entire system was removed in 84% of those undergoing extraction, with only partial removal occurring in 16%. The vast majority of patients underwent percutaneous lead extraction using manual traction. Sixteen percent had lead extraction surgically, three for transvenous leads, two for epicardial leads.

No complications occurred in those undergoing percutaneous lead extraction. In two patients, the transvenous lead was cut in the subclavian vein, and the remaining intracardiac portion removed percutaneously using a gooseneck snare placed from the femoral vein. In those undergoing surgical device extraction, one patient died from relapsing sepsis, two other patients developed acute renal failure, which resolved with conservative management.

A replacement cardiac device was implanted in 54% (17 PPM, 4 bivent/ICD or ICDs). Median time from device extraction to implantation of a new device was 28 days (range 1–74 days, IQR25% 8.5–IQR75% 35 days). A temporary wire was required in 12.8% (five cases). The majority of cases had transvenous lead placement (90.5%), with 9.5% requiring epicardial leads. A new pulse generator was placed in the contralateral prepectoral area in all cases.

All patients received antimicrobial therapy prior to and following device extraction. Intravenous antibiotic therapy was used for a median of 14 days, with subsequent oral antibiotics being used for a median of 14 days. The median total duration of antibiotics was 42 days (range 9–56 days, IQR25% 28–IQR75% 42 days). The majority of patients were treated with combination antimicrobials (95%). The most commonly used antibiotics were the b-lactams, followed by vancomycin. Adverse effects of antibiotics included diarrhoea in 7.7%, and neutropenia with associated drug fever in 12.8%.


The median length of stay in hospital was 21 days (3–74 days). Three (7.7%) of the patients died during index hospitalization. 12.8% experienced relapse, with 80% of these patients subsequently undergoing total device extraction. Clinical presentation, management, and hospital course of these patients is summarized in Table 3.

View this table:
Table 3

Clinical management and outcomes of cardiac device infection

In our study, complete device extraction was undertaken in 82%. Of these patients, none had a subsequent relapse, and their mortality was 7.4% (2/27). Of those managed with partial removal or conservative therapy, relapse occurred in 67% (n = 8/12) of these cases, with mortality occurring in 8.4% (n = 1/12).

92.4% remain infection free at their last outpatient visit with a median follow-up of 36 months (range 3–95 months).

Three patients died during hospitalization. An 89-year-old female expired 64 days after presentation, 57 days after complete device extraction due to unresolving multi-organ failure from sepsis. A 68-year-old male expired 28 days after presentation, 14 days after partial device extraction due to unresolving multi-organ failure from sepsis. Finally, a 26-year-old male expired 47 days after presentation, 35 days after complete device extraction from multi-organ failure from unresolving sepsis.

Univariate comparison of cardiac device infection survivors vs. cardiac device infection deaths

A comparison between those patients who died due to CDI despite therapy vs. those who survived was performed to try and identify any predictors of an adverse outcome. Due to the small numbers of deaths, only univariate comparisons were possible. The presence of pre-existing renal impairment at the time of original device implantation was the sole baseline co-morbidity that showed a trend towards but did not reach statistical significance (P = 0.07, relative risk 0.83, 95% CI 0.66–1.02). No procedural related factor such as device revision, temporary pacing wire use, number of leads implanted, or fluoroscopy time as a marker of procedure duration, differed between the two groups. The presence of documented MRSA infection (all cases had positive blood cultures and local cultures for MRSA) was significantly different between both groups, and associated with adverse outcomes (P = 0.004, relative risk 37, 95% CI 5.3–250). Finally, markers of severe systemic septicaemia, such as requirement for inotrope support, acute renal, and hepatic dysfunction, were also significantly different between both groups (P = 0.07, relative risk 9.2, 95% CI 0.95–90.9; 0.05, relative risk 0.85, 95% CI 0.63–1.02; 0.009, relative risk 0.66, 95% CI 0.42–1.05, respectively). This more pronounced septic state is also reflected in the longer length of hospitalization in the patients who eventually died, which showed a trend towards significance (P = 0.069).

Comparison of overall device population with those who developed cardiac device infection

From retrospective catheterization logbook review, we created an overall database on every individual receiving a device implant during the study period. From the data recorded, sex, age at implantation, device type, number of leads implanted, and fluoroscopy time were recorded on all cases, including those who ultimately developed infection (Table 4). Unfortunately, no baseline characteristics or co-morbid conditions were reliably recorded in the procedural logs to allow inclusion of these factors in our modelling.

View this table:
Table 4

Comparison of overall device population with those who developed cardiac device infection

In univariate comparison, screening time was significantly different between those who eventually developed a CDI compared with the remaining population (7.58 vs. 5.14 min; P = 0.02). Age showed a trend towards significance, and sex was not significantly different between both groups (P = 0.06 and P = 0.1, respectively). However, implantation of a biventricular device or two or more leads was significantly higher in those who developed a CDI compared with those who did not, P = 0.0001 (Table 4).

All univariate predictors were then considered candidate predictors for a logistic regression model, using the enter method. In contrast, screening time, although significant in the univariate model, was not a significant predictor in the regression model (Table 5). However, age at implantation, and more significantly, implantation of a biventricular device predicted device infection (P = 0.03, OR 0.96, 95% CI 0.94–0.98 for age, and P = 0.001, OR 7.57, 95% CI 2.4–23.7 for implantation of a biventricular device vs. non-biventricular device).

View this table:
Table 5

Logistic regression model of Population without device infection vs. cardiac device infection population


Cardiac device infection remains a serious complication, despite improvements in device design and implantation technique. With expanding indications for device use and the advancing age of the recipient population, rates of CDI have increased.14 Previously published literature estimates the rate of endocardial lead infection at 1–2%.6 In our study, the overall rate of device infection was 1.25% (39/3105), with a pacemaker infection rate of 1.33% (27/2029) and a biventricular pacemaker or ICD infection rate of 1.1% (12/1076).

In 61% of cases, a causative organism could be identified. Fifty-four percent of all patients had positive blood cultures at the time of presentation. The most common organism identified was Staphylococcal species, and empiric therapy should always include anti-staphylococcal therapy while waiting for a definitive microbiological diagnosis in a suspected case. The overall MRSA rate was 5.1% of all cases. The patients who developed MRSA infections were more likely to develop sepsis-related complications and the presence of MRSA infection independently predicted death. Prior studies have recommended that empiric MRSA therapy be initiated when CDI is suspected11,14 and the adverse outcome of patients with MRSA infection in our study suggests support for this strategy. On the basis of our study findings, we have adopted a strict screening protocol to identify MRSA colonization pre-device implantation, and if found, in addition to attempts to eradicate the organism in advance of implantation, all colonized patients receive glycopeptide antibiotic prophylaxis. Our decolonization regime consists of nasal mupirocin for 5 days for nasal carriage and chlorhexidine baths for 5 days if MRSA is isolated from the groin.

Cardiac device infection results in serious morbidity for patients, with prolonged hospital stays, repeated procedures for device extraction and re-implantation, and costs the health providers significant expenditure.10 In our study, patients who died as a consequence of CDI were more likely to the pre-existing renal impairment and culture the MRSA organism. This suggests that patients with either of these factors should be treated more aggressively from the outset. It is worth noting that this group also experience significantly more complications as a result of the septic state and have longer lengths of hospitalizations as a result.

Until recently risk factors for device infection have been identified from a number of case–control studies in the past.15,16 These studies have identified number of procedures, number of operators, use of temporary pacing wires, presence of central venous catheter, impaired LV function, and renal impairment as independent risk factors. An inverse relationship exists between operator volume and the rate of device infection, with those implanting the most devices having the lowest infection rates.17 However, valuable prospective data regarding the risk factors for developing device infection have been identified in the large prospective PEOPLE study, a multi-centre follow-up study of all device recipients in 44 centres in France.18 This study identified fever within 24 h of device implantation, use of a temporary pacing wire, and the need for early re-intervention after device implant as the strongest predictors of future device infection. There was an inverse correlation between development of infections and antibiotic prophylaxis.18 A previous study from the Mayo Clinic group established a number of potential risk factors to predict potential device infection,19 identifying long-term steroid use (OR 13.9, P = 0.03) and the implantation of greater than 2 leads (OR 5.41, P = 0.01), as independent predictors of pacemaker infection. Antibiotic prophylaxis was found to be protective (OR 0.0087, P = 0.005). In our retrospective comparison of the total device population vs. those with CDI, only a limited number of variables could be reliably obtained from procedural logbooks (age, sex, screening time, device type, number of leads implanted). In our model, only the implantation of a biventricular device appeared to be an independent predictor (OR 7.57, 95% CI 2.4–23.7, P = 0.0001) of device infection. This emphasizes the importance of high standards of sterility, antibiotic prophylaxis, and implantation technique for patients with the co-morbidity of heart failure undergoing biventricular device implantation.

No prospective studies have assessed conservative antibiotic therapy vs. combination device extraction and antibiotic therapy. Some investigators have advocated the conservative approach, with antibiotics and pocket debridement.20 However, the high rate of uncontrolled or relapsing bacteraemia, even after prolonged medical therapy, makes hardware removal the preferred option. Reported mortality rates for CDI endocarditis without device extraction ranges from 31 to 66%, compared with 18% in patients in whom the hardware is extracted, followed by prolonged antibiotic therapy.21 In our study, complete device extraction was undertaken in 82%. Of these patients, none had a subsequent relapse, and their mortality was 7.4%. Of those managed with partial removal or conservative therapy, relapse occurred in 67% (n = 8/12) of these cases, with mortality occurring in 8.4% (n = 1/12). Based on these observations, and current literature,12,13 we recommend complete hardware extraction.

In patients who are pacemaker-dependent, it is our practice to place a transvenous temporary pacing wire if require symptomatically, and to allow at least 7 days until another permanent system is implanted.

Eighty-four percent of patients undergoing device extraction in our study had successful percutaneous removal of device hardware, without complication. In cases where extraction was not initially performed, the most common reason was physician concern about removing LV pacing lead in cases of biventricular pacing systems, and the ability to re-access the LV coronary venous branch again in the future. After complete hardware extraction and control of infection, the need for a replacement device must be carefully assessed. Re-implantation should be at a new site. Timing of re-implantation is still hotly debated. Some authors recommend delaying re-implantation up to 6 weeks, although the consensus currently suggests that re-implantation can take place when patients are no longer bacteraemic.12 In our series, the median time to device re-implantation was 28 days (range 1–74 days), without any further relapse. Of note, in published series, up to 50% of pacemaker patients do not require further implantation after device removal.13

The duration of antibiotic therapy depends largely on the infective organism and the clinical presentation, and it has varied widely in the published literature.12,13 Some series have recommended 6 weeks of therapy after device removal. However, with most cases of device-related endocarditis limited to the right heart, some authors have suggested 4 weeks of therapy should be adequate.12,13 In our study, the median duration of antibiotics in total was 42 days, with a median of 14 days of intravenous antibiotics, followed a median of 14 days of oral antibiotics. In our series, local generator site infections were treated for a minimum of 2 weeks, while confirmed device-related endocarditis was treated for a full 6-week period with intravenous antibiotics.

Antibiotic prophylaxis was administered to all patients in our study prior to device implantation. Debate surrounds the administration of prophylactic antibiotics. A previous meta-analysis has suggested that prophylactic antibiotic therapy reduces the incidence of potentially serious device infection after pacemaker implantation, P = 0.0046, OR 0.256 95% CI 0.1–0.656.21 In the prospective PEOPLE study, use of prophylactic parenteral antibiotic therapy was associated with lower rates of device infection P = 0.02, OR 0.4 95% CI 0.18–0.86.19 Duration of antibiotics should be <48 h. Previous studies have shown no benefit to topical povidine–iodine or antibiotic installation into the generator pocket at the time of implantation.13,22

Study limitations

Our study has some limitations. First, retrospective observational studies have well-documented biases, although in conditions as rare as CDI, a retrospective observational study allows capture of the largest possible number of cases. Secondly, the incidence and number of cases may be underestimated due to referral bias, as our institution is a tertiary referral unit and perhaps some less severe cases were managed locally, without reaching our attention. Thirdly, device infection lacks clear definitions, making its diagnosis challenging. Recently, a large retrospective analysis of CDI from the Mayo Clinic suggested diagnostic criteria, which we have adopted.12 The univariate comparisons drawn are limited by their small numbers. The multivariate comparisons are limited by the small number of variable inputted into the model, as only five variables were reliably recorded in procedural logbooks on every device implant recipient. Finally, as device infection is a rare occurrence, their small numbers limit all studies in this area.


Cardiac device infections remain a rare but potentially fatally complication of device implantation. As the numbers of devices implanted rises, prospective studies of the optimal treatment for patients with this complication are urgently required. Delay in total device extraction results in worse outcomes. Our study supports a strategy of prompt and total device extraction, which can be safely performed percutaneously, combined with an appropriate duration of antibiotics.

Conflict of interest: none declared.


R.M.'s research position is funded generously through unrestricted educational grants from the Health Service Executive of Ireland, the Irish Heart Foundation, and Medtronic Ireland.


View Abstract