Cardiac implantable electronic devices and bloodstream infections: management and outcomes

Abstract

Background and Aims

Bloodstream infection (BSI) of any cause may lead to device infection in cardiac implantable electronic device (CIED) patients. Aiming for a better understanding of the diagnostic approach, treatment, and outcome, patients with an implantable cardioverter defibrillator (ICD) and cardiac resynchronization therapy and defibrillator (CRT-D) hospitalized with BSI were investigated.

Methods

This is a single-centre, retrospective, cohort analysis including consecutive ICD/CRT-D patients implanted between 2012 and 2021. These patients were screened against a list of all hospitalized patients having positive blood cultures consistent with diagnosed infection in any department of a local public hospital.

Results

The total cohort consisted of 515 patients. Over a median follow-up of 59 months (interquartile range 31–87 months), there were 47 BSI episodes in 36 patients. The majority of patients with BSI (92%) was admitted to non-cardiology units, and in 25 episodes (53%), no cardiac imaging was performed. Nearly all patients (85%) were treated with short-term antibiotics, whereas chronic antibiotic suppression therapy (n = 4) and system extraction (n = 3) were less frequent. Patients with BSI had a nearly seven-fold higher rate (hazard ratio 6.7, 95% confidence interval 3.9–11.2; P < .001) of all-cause mortality.

Conclusions

Diagnostic workup of defibrillator patients with BSI admitted to a non-cardiology unit is often insufficient to characterize lead-related endocarditis. The high mortality rate in these patients with BSI may relate to underdiagnosis and consequently late/absence of system removal. Efforts to increase an interdisciplinary approach and greater use of cardiac imaging are necessary for timely diagnosis and adequate treatment.

Structured Graphical Abstract

Mortality in patients with and without BSI. Blue line, patients without BSI; red line, patients with BSI. BSI, bloodstream infection; CI, confidence interval; CRT-D, cardiac resynchronization therapy and defibrillator; HR, hazard ratio; ICD, implantable cardioverter defibrillator.

Open in new tabDownload slide

See the editorial comment for this article ‘Blood stream infection in defibrillator recipients: cardiac imaging for all patients or sometimes skillfull neglect?’, by M.G. Bongiorni and G. Zucchelli, https://doi.org/10.1093/eurheartj/ehae025.

Introduction

Infection in cardiac implantable electronic devices (CIED) is rare, but the severity of complications and healthcare costs are substantial.^1–6 Device infection typically manifests either as a pocket or systemic infection. The early post-operative period is the time of the highest risk for device infection; thus, current preventive strategies focus mainly on reducing the risk at that time.^7,8

Knowledge about the lifelong risk of device infection, patient-related risk factors, and mitigation strategies to influence device or lead bacterial seeding is limited. The diagnosis of systemic infection without signs of local inflammation is challenging.⁹ Symptoms may be non-specific, and a long period of time may elapse between exposure, bacterial seeding, onset of symptoms, and, finally, diagnosis and treatment of CIED infection. Late device infection can arise from various non-cardiac sources, as, for example, pneumonia and urogenital tract, gastrointestinal, or orthopaedic infection.¹⁰ As primary care physicians are frequently managing out-of-hospital infections in device patients, limited data exist on the accuracy and timing of the diagnosis of CIED infection in these patients. Furthermore, there is inadequate understanding of the initial infections most likely resulting in a CIED infection.

The 2019 International CIED Infection Criteria¹¹ recommend a second and, if necessary, a third level of diagnostic imaging when a positive blood culture is found independently of clinical signs of pocket infection and, eventually, referring the patient to a centre with expertise in CIED infection and extraction. At present, there is little knowledge of the diagnostic workflow followed in CIED patients with positive blood cultures admitted to or hospitalized in a non-cardiology unit.

Aiming for a better understanding of the diagnostic workup leading to the diagnosis of device infection and outcome of CIED patients who have bloodstream infection (BSI), we investigated patients with an implantable cardioverter defibrillator (ICD) or cardiac resynchronization therapy with defibrillation (CRT-D) capability when hospitalized and diagnosed for BSI.

Methods

Study population

This is a single-centre, retrospective, cohort analysis including consecutive patients implanted with a de novo ICD, CRT-D, or subcutaneous ICD (S-ICD) between January the 1st 2012 and December the 31st 2021 at Istituto Cardiocentro Ticino—Ente Ospedaliero Cantonale in Lugano, Switzerland. Exclusion criteria were patients under the age of 18 at the time of implantation. The study was approved by the local ethics committee (2022-00809, Rif CETI 4102).

Data source

All patients receiving a CIED are included in a prospective national registry managed by the Working Group of Cardiac Pacing and Electrophysiology of the Swiss Society of Cardiology. In the Ticino canton, ICD, CRT-D, or S-ICD are implanted exclusively at the Istituto Cardiocentro Ticino, whereas follow-up may take place at one of the eight regional hospitals (three hospitals in Lugano, Bellinzona, Acquarossa, Faido, Locarno, or Mendrisio) or at the outpatient clinic of a local cardiologist who is part of the patient’s local management network. The regional hospital sites share a common database containing patients’ electronic medical records including laboratory and microbiology findings. A list containing all patients with a de novo ICD, CRT-D, or S-ICD implanted in Ticino from January the 1st 2012 until December the 31st 2021 was generated from the national database and screened against a list of all hospitalized patients who had positive blood cultures at any local public hospital department at any time after device implantation. Each electronic medical record was assessed to determine patients’ characteristics, and each case was reviewed with an infectious disease specialist (E.B.) to exclude patients with positive blood cultures due to contamination (such as coagulase-negative staphylococci on a single culture).

Event definition

Bloodstream infection was defined as the presence of blood cultures with growth of a pathogen consistent with the patient’s infection and clinical signs of infection (e.g. fever, change in laboratory parameters, such as elevated C-reactive protein or leucocytosis). A device infection was considered definite, possible, or rejected according to the 2019 International CIED Infection Criteria published in the European Heart Rhythm Association (EHRA) international consensus document on how to prevent, diagnose, and treat CIED infections.¹¹

Definition of comorbidities

The age-adjusted Charlson Comorbidity Index (CCI) was used.¹² It is a predictor of 10-year survival in patients with multiple comorbidities, using age, sex, and comorbidity categories of the patient including history of myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular disease (including transient ischaemic attack), dementia, chronic obstructive pulmonary disease, connective tissue disease, peptic ulcer disease, liver disease (divided into severe, defined as cirrhosis with portal hypertension and variceal bleeding history; moderate, defined as cirrhosis with portal hypertension but no variceal bleeding history; or mild, defined as chronic hepatitis or cirrhosis without portal hypertension), diabetes mellitus (divided into uncomplicated or with end-organ damage), hemiplegia, moderate to severe chronic kidney disease (defined as creatinine level > 3 mg/dL, presence of signs of uraemia, and patient on dialysis or with previous kidney transplant), solid tumour (with or without metastasis), leukaemia, lymphoma, and acquired immunodeficiency syndrome. Renal function was assessed by calculating the estimated glomerular filtration rate (eGFR) according to the 2009 Chronic Kidney Disease Epidemiology Collaboration equation.^13,14

Statistical analysis

All analyses were performed using the Stata software (release 17, StataCorp, College Station, TX, USA). A two-sided P-value < .05 was considered statistically significant, and there was no adjustment for multiple comparisons. Continuous data were described as mean and standard deviation if normally distributed and as median and 25th–75th percentiles otherwise. Categorical data were reported as counts and percentage. The incidence of BSI was computed as the number of affected patients over the total person-years, together with its 95% confidence interval (CI). The mortality rate per 1000 person-years was recorded, and Kaplan–Meier cumulative survival curves were plotted. A time-dependent Cox model was used to assess the association of BSI with subsequent mortality. Similarly, an event-free survival method was used to identify potential correlates of BSI development. Non-collinear variables with a P-value < .1 at univariable analysis were candidates for inclusion in the multivariable model. However, given the low number of patients with BSI (n = 36), and to avoid overfitting, the number of covariates to include into the multivariable model was limited. For this reason, candidate predictors considered as most clinically relevant were identified among the potential candidates: age at implant, atrial fibrillation, history of myocardial infarction, chronic obstructive pulmonary disease, and eGFR < 60 mL/min. Hazard ratios (HR) and 95% CI were reported, and Harrell’s c statistic for model discrimination was calculated.

Results

Study population

A total of 515 patients were included in the study. Baseline characteristics are summarized in Table 1. The majority of patients was male (79%) with a median age of 69 years and a history of ischaemic cardiomyopathy (54%). Most patients received a device for primary prevention of sudden cardiac death (73%). Just over a third of the patients had a CRT-D. The most common comorbidities were arterial hypertension (55%), atrial fibrillation (30%), and diabetes (23%).

Bloodstream infection

During a median follow-up of 59 months [interquartile range (IQR) 31–87 months], there were 64 cases of positive blood cultures. After review with the infectious disease specialist, 17 cases were dismissed as contaminants, leading to a total of 47 BSI episodes in 36 patients. The most common presumed BSI source was the urinary tract (27%) followed by pulmonary and venous catheter infections with 15% and 13%, respectively (Table 2). In about a fifth of patients, infection source remained unknown. The most common pathogens were Gram-negative bacteria (41%) with Escherichia coli being the most frequent one (14%), followed by coagulase-negative staphylococci (17%). In contrast, Staphylococcus aureus (12%), Enterococcus (10%), and Streptococcus species (9%) were less frequent (Table 3).

The median time to first BSI was 25 months (IQR 5.4–55.9 months). The BSI incidence rate was 22.8 per 1000 patient-years (95% CI 14.1–34.8) for CRT-D and 22.1 per 1000 patient-years (95% CI 14.4–32.4) for transvenous ICDs. For subcutaneous ICDs, BSI incidence rate was 0 per 1000 patient-years (95% CI 0–24.9).

Diagnostic workup

Ninety-two per cent of patients were primarily admitted to a non-cardiology unit, 37 patients (79%) to an internal medicine department or intensive care unit, and 6 patients (13%) to a surgical unit (general surgery, orthopaedics, or urology). The remaining 8% of patients (n = 4) were directly admitted from the emergency room to the cardiology department. Additional six patients (13%) were transferred to a cardiology department after a median stay of 6 days (IQR 2–6) at an internal medicine unit. Cardiac imaging was not performed in 25 patients (53%); in the remaining patients, a transthoracic echocardiography (TTE) was performed in 22, a transoesophageal echocardiography (TEE) in 16, and a cardiac positron emission tomography with a computed tomography (PET-CT) in 4 cases (Figure 1).

According to the Novel 2019 International CIED Infection Criteria, seven patients had a definite device infection (Figure 1).

Therapeutic management and outcome

The majority of BSI cases (85%, n = 40) was treated with short-term antibiotics only (i.e. 1–6 weeks); three patients underwent a system extraction procedure (6.4%), and four patients were put on chronic antibiotic suppression therapy (8.5%). All patients on chronic antibiotic therapy either refused an extraction procedure, or the comorbidities deemed extraction too risky.

Sixty-five patients of the entire study population died over a median follow-up of 59 months (IQR 31–87 months); the overall mortality was 4.1 per 100 person-years (95% CI 3.3–5.0). In the time-dependent Cox model, all-cause mortality was nearly 7 times higher in patients with BSI (Figure 2) as compared with patients without BSI (HR 6.7, 95% CI 3.9–11.2; P < 0.001). After adjustment for CCI, patients with BSI still showed a six-fold higher mortality rate (HR 5.8, 95% CI 3.5–9.9, P < .001).

Clinical conditions associated with bloodstream infection occurrence after device implantation

In a univariable analysis (see Supplementary data online, Table S1), age, arterial hypertension, history of atrial fibrillation, history of acute myocardial infarction, and chronic obstructive pulmonary disease were associated with the occurrence of BSI. In the multivariable analysis, however, only acute myocardial infarction remained an independent predictor, with a two-fold higher rate of BSI (Table 4).

Discussion

This study addresses gaps in the cardiac diagnostic workflow to assess for possible device infection in hospitalized patients with BSI. We found that over the lifetime of their devices, 92% of patients hospitalized with BSI were hospitalized in non-cardiological units. However, 53% of them were never investigated using cardiac imaging and 13% underwent TTE only to exclude possible device infection. The lack of more advanced imaging, i.e. TEE, PET-CT, single-photon emission computed tomography, or cardiac CT, left uncertainty about whether there was a device infection in 66% of our defibrillator patients with BSI, since a major criterion of the 2019 International CIED Infection Criteria was not sufficiently assessable.

An in-depth analysis of the comorbidities of each of the 25 patients with a BSI, in whom a cardiac imaging was not performed, showed that the number and characteristics of the comorbidities remained mostly unchanged. The modest, yet not significant, increase in CCI (median CCI at baseline: 5, IQR 4–7 vs. median CCI at BSI hospitalization: 7, IQR 4–9) was mostly due to aging of the population.

Cumulative mortality of those with BSI was ∼50% at 4 years (Figure 2), being nearly seven times higher compared with that of patients who never experienced a BSI (Structured Graphical Abstract).

There is uncertainty about the optimal diagnostic tool for device endocarditis. Although the Novel 2019 International CIED Infection Document presented additional criteria for device endocarditis and merged the modified Duke criteria with the 2015 European Society of Cardiology (ESC) criteria, none represent a validated and standardized tool for the diagnosis in this specific setting. A recent study by Boix-Palop et al.¹⁰ evaluated patients with late-onset BSI, defined as episodes of clinically significant bacteraemia occurring more than 1 year after implantation or device manipulation. They reported 187 (2.9%) cases of clinically significant bacteraemia out of 6313 patients. In this patient cohort, 41% never underwent cardiac imaging, 41% had a TTE, 18% had both TTE and TEE, and <10% a PET-CT scan. Unlike the study by Boix-Palop, which included a large group of patients with prosthetic valves (57%), in which the medical community is significantly more attentive to endocarditis or systemic infection, our study cohort consisted exclusively of patients with de novo ICD or CRT-D implantation and BSI. In line with past literature,^10,15 the rate of cardiac imaging in our cohort was as low as 47% and, when performed, TTE was prevalent. This is a surprisingly low rate, especially considering that over half of patients had a Gram-positive pathogen, well-known for resulting in a higher risk for device infection as compared with other pathogens.^16,17 On the other hand, recommendations for a more systematic use of second- and third-level cardiac imaging was given only in 2020.¹¹ The lack of cardiac imaging was likely a factor in missing the diagnosis of endocarditis related to the need for other than microbiological criteria for a device infection/endocarditis as recently indicated by the Novel 2019 International CIED Infection Document.¹¹

Four out of 16 patients (25%) with a BSI who underwent TEE had a device infection confirmed. Transoesophageal echocardiography is known to have a good sensitivity and specificity for native valve endocarditis; however, its accuracy for lead vegetation is significantly lower.^18,19 Therefore, in the vast majority of cases, it was not possible to exclude a device infection. In such cases, a PET-CT scan should be considered.²⁰ Indeed, in four patients with negative or inconclusive echocardiographic examinations, a PET-CT was performed, leading to confirmation of device infection in two cases.

A timely diagnosis of cardiac device infection is of paramount importance since system extraction is important for cure rather than antibiotic therapy alone. A recent meta-analysis of 11 studies²¹ found that system extraction was associated with significantly lower mortality [odds ratio (OR) 0.22, 95% CI 0.09–0.50, I² = 62.8%] compared with antibiotic therapy alone. A delay of system extraction (>7 days after diagnosis) was associated with significantly increased in-hospital mortality. Using the nationally representative, all-payer, Nationwide Readmissions Database, Lee et al.²² evaluated patients undergoing transvenous lead removal for device infection. In 12 999 patients who underwent lead removal for device infection, for 8834 (68%), the removal was done early (≤7 days), and for 4165 (32%) patients, the lead removal was delayed (>7 days). Delayed lead removal was associated with a significant increase in in-hospital mortality (8.3% vs. 3.5%; adjusted OR 1.70; 95% CI 1.43–2.03; P < 0.001).

This is in keeping with recent US data by Sciria et al.,²³ who showed that only 11.5% of the 25 000 CIED patients admitted with endocarditis between 2016 and 2019 were managed with system extraction, leading to a significantly increased mortality (9.5% vs. 6.0%, P < .001). Similarly, in a more recent study by Pokorney et al., in 11 304 patients with CIED infection, system extraction was performed within 30 days in only 18.6% of cases.²⁴

In line with these studies, only a minority of our patients (6, or 13%) admitted to a non-cardiology unit were referred to an extraction centre for further assessment of a possible device infection and, on average, 6 days after initial admission.

Therefore, a high mortality of ∼50% at 4 years in medically treated ICD/CRT-D patients with BSI is not surprising. Delay in transfer may be related to the patient’s critical condition, the desire to withhold further treatment, or other circumstances which we were unable to assess and, in a retrospective analysis, are frequently difficult to evaluate.

One reason for delayed transfer may be an identification barrier of device infection, related to the lack of physician knowledge and awareness, as was described by Lakkireddy et al.²⁵ and by Sood and colleagues.²⁶ One strategy to minimize this barrier is by education and sharing of evidence. Another strategy could be the use of an automatic system triggering an alert to a clinical expert in the hospital whenever a positive blood culture is registered in a device patient’s medical record. A recent observational study by Rios et al.²⁷ analysed the impact of such an alert system and found a significant reduction of time to echocardiography (25.7 vs. 83.7 h; P = .023), to infectious disease (22.6 vs. 81.2 h; P = .050), and to cardiology consultation (28.8 vs. 91.5 h; P = .019) as compared with before its introduction. As for time to device removal (121.2 vs. 169.7 h; P = .260) and to antibiotic therapy initiation (3.0 vs. 3.8 h; P = .620), there was a trend to time reduction. Further evidence is necessary to determine if an automatic alert system may lead to faster medical and extraction therapy to improve outcome.

The majority of previous studies has investigated the outcome of device patients with a definite diagnosis of systemic device infection. In contrast, we investigated device patients hospitalized with BSI. Although BSI may be related to an intravascular device infection, this is not always the case, and if there is no evidence of endocarditis with advanced imaging, with certain pathogen strain characteristics and immunological competence, effective antibiotic therapy alone may be a reasonable therapy. A systemic device infection is believed to develop due to haematogenous seeding occurring considerably later compared with pocket infection or device manipulation. Typically non-pathogenic microorganisms such as coagulase-negative staphylococci may adhere to the device and establish a focus of infection. However, their frequency varies by country and even hospital.^16,28–31 Although a substantial proportion of our patients presented with coagulase-negative staphylococci and other Gram-positive bacteria, we found that patients who ultimately died of sepsis often had multiple pathogens. That may indicate multiple entry points, a temporary deficit of immunocompetency, or the presence of more severe comorbidities, resulting in a greater failure rate of both medical and non-pharmacological therapy. The incidence of patients diagnosed with severe BSI leading to hospitalization was around 22 per 1000 patient-years, hence considerably higher compared with the incidence rate of 2.5 per 1000 device-years after device implantation reported by Olsen et al.,³² who collected data from three nationwide registries: the Danish National Patient Registry, the Danish National Prescription Registry, and the Danish Pacemaker and ICD Register. Considering that patient characteristics at first device implant were similar between our population and the Danish study, it is reasonable to assume that around 1 out of 10 device patients admitted to a hospital with BSI will have a systemic device infection or endocarditis.

Study limitations

The retrospective nature of the study carries several limitations in data collection, including possible incomplete data and irregular follow-up visits. However, considering the local healthcare network (general cardiologists and local hospitals), as well as the availability of a centralized microbiological laboratory of all public hospitals, that seems unlikely. We are reporting clinical management of persistent BSI in patients with an ICD or CRT-D. It is unknown whether our findings are equally applicable to patients with transvenous or leadless cardiac pacemakers; this issue may require additional investigation. We focused on ICD and CRT-D patients who are known to have a higher risk of systemic infection. While all high-voltage devices were implanted at a single centre, the data concerning diagnostic management were collected from eight hospitals across an entire Swiss region (Canton Ticino), thus encompassing the collective experience of these facilities. To establish whether the diagnostic approach is similar in other regions of Switzerland or in different countries, further studies are warranted. Although patients who did not undergo any imaging despite BSI had similar clinical characteristics and number of comorbidities at the time of device implantation and hospitalization for BSI, the futility of further cardiac diagnostic imaging or patient’s refusal as a reason to withhold imaging, albeit unlikely, cannot be excluded completely. Finally, the relatively low overall event rate observed in our study raises the potential for bias, as the limited number of events could have influenced the stability and precision of the presented multivariable regression model. Consequently, the results warrant cautious interpretation.

Conclusions

The diagnostic workup of device patients with BSI, mainly hospitalized in non-cardiology units, is frequently insufficient to characterize possible lead-related endocarditis. The high mortality risk observed may be related to the patient’s underlying condition and multimorbidity. However, it is also likely that underdiagnosis or delayed diagnosis of device infection/endocarditis and/or late system removal may contribute to poor outcomes. Efforts to establish a systematic, interdisciplinary approach with greater use of cardiac imaging, including TEE and PET-CT scans, linked to timely triage and referral of these patients to a cardiology department with extraction capabilities, may provide an important opportunity to improve outcome of this population.

Supplementary data

Supplementary data are available at European Heart Journal online.

Declarations

Disclosure of Interest

T.Ö. received educational grants from Biotronik and Abbott, research grants from Medtronic and Philips, and speaker fees from MicroPort. E.B. declares no conflict of interests in relation to the present manuscript. E.B.’s institution received payments for the participation of E.B. to advisory boards or travel grants from MSD, Gilead Sciences, ViiV Healthcare, Moderna, AstraZeneca, Pfizer AG, AbbVie, and Eli Lilly. C.B.G. receives consulting fees for the following companies: AbbVie, Abiomed, Alnylam Pharmaceuticals, Anthos, Bayer Corporation, Boehringer Ingelheim, Boston Scientific Corporation, Bristol Myers Squibb, Cardionomic, CeleCor Therapeutics, Janssen Pharmaceuticals, Merck, Novo Nordisk, Novartis Pharmaceutical Company, Pfizer, and Philips. He also has salary funded by Duke grants sponsored by Boehringer Ingelheim, Bristol Myers Squibb, Daiichi Sankyo, Food and Drug Administration, Janssen Pharmaceuticals, Novartis Pharmaceutical Company, Pfizer, and Philips. He has equity in Tenac.io. A.A. is consultant to Boston Scientific, Cairdac, Medtronic, EP Solutions, Philips, Radcliffe Publishers, and XSpline; he received speaker fees from Boston Scientific, Medtronic, MicroPort, and Philips; he participates in clinical trials sponsored by Boston Scientific, Medtronic, and XSpline and has intellectual properties with Boston Scientific, Biosense Webster, and MicroPort CRM. All other authors have no conflicts to disclose.

Data Availability

The data underlying this article cannot be shared publicly for privacy reasons. The data may be shared on reasonable request to the corresponding author.

Funding

This work has been supported by Philips, the Netherlands.

Ethical Approval

The study was approved by the local Ethics Committee (2022-00809, Rif CETI 4102).

Pre-registered Clinical Trial Number

None supplied.

References