The present study analyzed consecutive patients with CIED infections treated between 1985 and 2015 in our center. From 1985 to 1998, the patients were referred from the clinic to the infectious disease service. In 1999, a specific intervention program developed by a multidisciplinary team was implemented for these types of infections. When a CIED infection was suspected, the patients (from the hospital itself or transferred from other centers) were evaluated by a team member (A. de Alarcón González) to determine the type of infection and the priority for action. Data were prospectively and systematically collected according to a preestablished protocol.

Two batches of serial blood cultures (baseline and at 30 and 60minutes) were obtained from all patients at a 12-hour interval. If patients showed local signs of infection, the exudate was sampled for Gram staining and aerobic and anaerobic cultures; if there was no external discharge, sampling was performed using needle aspiration of the generator pocket. During the surgery, samples were obtained from the pocket and the region 4cm distal to the leads; some of the samples were cultured while the remainder was stored at –20°C. From 2005, negative cultures were additionally analyzed with genomic sequencing techniques, specifically, 16S rRNA polymerase chain reaction (PCR). If systemic infection was suspected and/or the blood cultures were positive, cardiac ultrasound was performed (in transesophageal mode from 1999 onward).

Infections were classified according to location as:

• •

  Local infection: infections were considered local when patients showed no systemic symptoms (fever, shock, embolisms, or remote infectious complications), blood cultures were negative, and there were signs of infection in the region of the generator pocket, such as pain, erythema, and purulent material.

• •

  Systemic infection: infections were considered systemic when patients showed systemic symptoms and blood cultures were repeatedly positive. Negative blood cultures required the presence of vegetation in the lead or right-sided cardiac structures.

Depending on the time of onset, infections were classified as: a) acute: those appearing before 1 month after device implantation or manipulation; b) delayed: those appearing between 2 and 12 months after implantation or manipulation; and c) late: those appearing after 12 months.

Antibiotic therapy was always performed in accordance with the antibiogram of the causative agent isolated in culture. After device removal, oral antibiotic therapy continued for 1 or 2 weeks for local infections. If the device was not removed, the duration varied from patient to patient.

Initial treatment of systemic infections always involved intravenous antibiotics before device removal; except in patients with severe sepsis, the device was not removed until the blood cultures were negative (typically at 1 week). If the removal was complete, the therapy was maintained for 2 further weeks and, after the first week, could be switched to oral therapy. If valves were affected (eg, the tricuspid), the antibiotic therapy was maintained for at least 4 weeks.

The surgical procedures are detailed in the supplementary material.

Device replacement was performed in 1 or 2 stages, depending on the type of device, as described in the supplementary material.

Patients of our hospital underwent a specific follow-up for 1 year in a clinic specializing in infectious diseases and then for life in a pacemaker assessment clinic. Patients referred from other centers underwent a minimum follow-up for 3 months in our hospital; all such patients also underwent follow-up blood cultures for systemic infections and stayed in contact with the referral hospital for 1 year.

Infection relapse was considered to have occurred if, during the 1-year follow-up, the newly implanted device was infected by the same microorganism isolated from the originally extracted system. In contrast, the infection was considered to be a reinfection if a different microorganism was isolated; it was then recorded as a new episode. Persistence of the original signs and symptoms or continued isolation of the same microorganism after a specific therapeutic intervention was defined as treatment failure.

A descriptive analysis with measures of central tendency and dispersion was performed for continuous variables, as well as a frequency distribution for qualitative variables. For the comparison of continuous variables, the Student t test was used (after analysis of normality with the Kolmogorov-Smirnov test); the Mann-Whitney U test was used if the distribution was not normal. For categorical variables, the chi-square test or Fisher exact test was used as appropriate. P < .05 was considered statistically significant.

During the study period, 325 infections were recorded in 314 patients, 226 (72%) men and 88 women; 28.5% of the patients had been referred from other centers. The median patient age was 69 [60-76] (range, 14-90) years and 32% had 2 or more concomitant chronic diseases.

The infection was related to a pacemaker in 229 patients (71%), to an implantable cardioverter defibrillator in 91, and to cardiac resynchronization therapy in 27 (22 had received an implantable cardioverter defibrillator). The main characteristics of the patients are summarized in Table 1.

The infections were categorized as local in 196 episodes (60.3%) and systemic in 129 (39.7%). In 160 patients (49%), the infection developed after further manipulation (generator replacement in 80% of these infections). The infections were classified as acute (n = 131 [40.3%]), delayed (n = 100 [30.8%]), and late (n = 94 [28.9%]).

The most frequently isolated microorganisms were coagulase-negative staphylococci (42.8%) and Staphylococcus aureus (16.9%). A more detailed description of the microbial etiology can be found in Table 2A and Table 2B.

When there were local symptoms, the culture yield of the generator region was high (75.0%-85.3% of isolations); also high were the culture positivity of the leads in systemic infections (75%) and even in local infections (52.6%). A more detailed summary of these results is shown in Table 3.

An echocardiographic study before device removal was performed in 116 patients (90%) with systemic infection: 45 patients underwent a transthoracic study, 42 underwent a transthoracic and transesophageal study, and 29 proceeded directly to a transesophageal study. Of the 87 transthoracic ultrasound studies, vegetation was visible in 24 (27.5%) vs 63.3% (n = 45) of the 61 transesophageal studies.

The clinical signs and symptoms are shown in Table 4. Of the patients with systemic infection, 77 (57.7%) had local signs; because these signs were present a long time before the onset of the systemic infection in 32 (25%) patients, they were given standard therapy for local infection (typically with local septic surgery) before the disease progressed to a systemic infection.

In 102 local infections (55.6%), the therapeutic approach of choice involved various types of percutaneous traction (PCT). Unsatisfactory results were obtained in 75 (73.5%). Local surgery (generator removal with the leads left in situ, pocket debridement and cleaning, and placement of a new generator) had a clinical success rate of just 20%, with a median of 2.5 [1-3] (1-12) procedures per patient. The cables were left in place on 9 occasions, even though a system was implanted contralaterally to the infected generator; in all patients, extrusion and discharge were witnessed in the same region months afterward. Isolated antimicrobial therapy was successful on 48.3% of occasions. A comparison of the patients successfully treated with this therapy (n = 15) with those who were not (n = 16) revealed the following: a) the patients had a significantly shorter median clinical course: 10 [3-15] (2-60) days vs 35.5 [16-187] (11-474) days (P < .01), and b) the patients less frequently developed skin erosion: 4 occasions (28.6%) vs 11 (64.7%; P = .07). In both situations, the prescribed antimicrobial therapy was prolonged—60 [40-90] days—and was always guided by the antibiogram.

In contrast, PCT ultimately achieved clinical success in 96.3% of complete removals (n = 94) and in 86.2% of incomplete removals (n = 20) due to retained intravascular fragments after lead rupture. In 3 cases, the patients developed, months later, a systemic infection that ultimately led to open heart surgery.

The antimicrobial therapy had poor efficacy for systemic infections (an 85.1% failure rate) and a high mortality rate (11.1%). Complete PCT achieved a clinical success rate of 89.3%, whereas incomplete PCT had a failure rate of 60% (n = 12), leading to cardiac surgery in 11 patients and subsequent femoral snare removal in the remainder. Of the 21 patients who underwent cardiac surgery in our center, 11 required it for PCT failure; the remainder had been referred from other centers that had tried simple manual traction. The results of the various approaches are detailed in Table 5.

Percutaneous traction was performed for a total of 280 episodes (166 local infections and 114 systemic infections); complete removal was achieved in 231 patients (82.5%) and a cure was achieved in 246 (89%). The success rate of this technique was related to the age of the implanted CIED (Table 6), with no influence of other factors such as the number of leads, implantation side, and type of device.

In the entire series, there were 12 deaths: 2 patients (1%) with local infection and 10 (7.7%) with systemic infection. Only 2 of these deaths were related to the PCT technique (superior vena cava tear: 1 in a patient with local infection and 1 with systemic infection). In another 7 episodes, the cause was the sepsis itself: 3 of these patients were undergoing antimicrobial therapy, 2 underwent PCT (1 with incomplete removal and 1 with complete removal), and 2 underwent cardiac surgery. Another 2 patients died several days after PCT completion from unrelated causes (stroke, 1 of them with local infection) and, finally, another patient with dilated cardiomyopathy who underwent resynchronization therapy but did not have system replacement died months afterward due to heart failure.

The various PCT-related complications are described in Table 7.

Device replacement after its removal was not performed in 13 patients (4%) in this series.

In 9 patients with local infections who underwent PCT with complete device removal, a contralateral implantation was first performed and the generator alone was removed; in our center, the remaining leads were removed because the patients continued to show signs of infection. Of the 128 remaining patients, the replacement was performed in the same surgical session in 96 (75%) and in 2 stages in 32, with no significant differences in contralateral infection frequency.

In the 4 patients with systemic infection who underwent PCT, the removal was performed after placement of a contralateral system. Of the 81 remaining replacements, the replacement was performed at the same time as the removal in 56 (69%) and at a later stage in 25. There was 1 infection by the same microorganism and another by a different bacterium (3.5%) in 1-stage procedures, whereas there were 4 infections (16%) by a different microorganism in 2-stage procedures (P = .04).

The differences between the 2 procedures are shown in Table 8.

Of the 304 surviving patients, 13 (4.3%) did not complete the minimum follow-up (3 months). The others were followed up for at least 1 year—median 36 [9.5-70.5] (6-309) months—, although 12 patients had not reached this time point by study closure. There were 31 deaths (10%) due to causes unrelated to device problems.

Our work has some limitations: a) this study was performed in a referral center and there may have been an excessive number of local septic surgery failures because most patients referred to us had already undergone this failed approach, which means that the data fail to reflect its real efficacy; however, the analysis was repeated with patients from our center alone (data not shown) and the results were identical; b) because a marked percentage of patients were referred from other centers, the positive results of the microbiological studies shown here might be even lower than the actual figures, given that in many cases the cultures were not correctly obtained in the original hospital and the patients were already undergoing antimicrobial therapy, greatly complicating the identification of the causative agent; similarly, because a transesophageal echocardiographic study was not performed in all patients with systemic infections, the number of vegetations might actually be higher than that reported here; however, we do not believe that this would markedly affect the fundamental crux of the study, which is analysis of the various therapeutic approaches; and c) replacement in 1 or 2 stages was not random, with a greater number of defibrillator and resynchronization implants in the 2-stage replacement group; due to their greater size and the increased implantation difficulty, the surgical procedure is longer and favors a greater number of infections. However, this does not invalidate the conclusion that the 1-stage replacement is equally as safe and avoids the inconvenience of the “bridge interval” in pacemaker-dependent patients (bedridden and long-term hospitalized).