Device infections related to cardiac resynchronization therapy in clinical practice–An analysis of its prevalence, risk factors and routine surveillance at a single center university hospital

Abstract Background The implantation rates of cardiac implantable electronic devices have steadily increased, accompanied by a steeper rise of device related infections (DRI). Hypothesis The prevalence of DRI for cardiac resynchronization therapy (CRT) is higher in clinical practice than reported previously, even at a university hospital, and likely higher than reported to the national device registry. Methods Electronic medical records of consecutive patients undergoing a CRT procedure between January 2016 and December 2017 were analyzed. Clinical history, procedure related variables and complications were reviewed by specialists in cardiology and infectious diseases. Results A total of 171 patients, mean aged 74 years, 138 males (80.7%) were included. Twelve DRI occurred in 10 patients during mean 2.5 years follow‐up, giving a prevalence of 7% (incidence of 29/1000 person‐years). Reoperation, pocket haematoma, ≥3 procedures, previous device infection and indwelling central venous line were the strongest predictive factors according to univariate analysis. Out of 63/171 (36.8%) major complications, 31(49.2%) were lead‐related. There were 49/171 (28.7%) reoperations and 15/171 (8.8%) minor complications. The number major complications and DRI reported to the national device registry were 7/171 (4.1%) and 2/171 (0.6%), respectively, reflecting a 5‐fold underreporting. Conclusions The high rate of CRT device infections is in sharp contrast to those reported by others and to the national device registry. Although a center specific explanation cannot be excluded, the high rates highlight a major issue with registries, reinforcing the need for better surveillance and automatic reporting of device related complications.


| INTRODUCTION
The implantation rates of cardiac implantable electronic devices (CIED) have steadily increased along with their expanding indications 1 and have been accompanied by an even steeper rise of device related infections (DRI). 2 Although the reason for the increased infection rates remains unclear it may be related to the aging population with more comorbidities and fragility, and more complex devices, particularly for cardiac resynchronization therapy (CRT). [3][4][5] The prevalence of infections is related to the complexity of the device and ranges between 1.6%/2 years follow-up and 1.9%/3.4 years follow-up for CRT-P (P = pacemaker) and 3.1%/ 3.4 years follow-up and 8.6%/2.6 years follow-up for CRT-D alone (D = defibrillator). 3,6 DRI are devastating complications resulting in increased morbidity, mortality and high costs for the national health care systems. 7 Most of the reported DRI rates from registries and randomized clinical trials (RCTs) have ranged between 0.6% and 1.3% while those from prospective studies have reported somewhat higher rates ranging between 2.3% and 3.4%. [7][8][9] The real-world prevalence of DRI, as reflected by figures from device registries, has probably been underestimated due to a significant underreporting. 10,11 A meta-analysis comparing complication rates for implantable cardiac defibrillator (ICD) procedures in RCTs versus the US National Cardiovascular Data Registry found an almost 3-fold lower complication rate in registry data. 11 The National Swedish Pacemaker and ICD Registry (www. pacemakerregistret.se) collect all CIED implantation procedures and related complications occurring during the first year after implantation from all 43 implanting sites in Sweden. According to the annually published report, the prevalence of CRT infections in Sweden and at our clinic has been 1.1%-1.2% and 0.6%-0.7%, respectively. Given the observed underreporting in other registries from previous publications 8 our aim was to perform a retrospective, single-center study of CRT-device related infections at our university center and identify the most important predictive factors.

| Study population
All consecutive patients who underwent a CRT procedure (de novo, upgrade, revision or generator change) at Uppsala university hospital, which is a tertiary-care hospital between January 2016 and December 2017 were included retrospectively to ensure at least 150-200 patients. There were no exclusion criteria. The 10 digit identity number of each patient, the type of CRT procedure and the reported complications were obtained from The National Swedish Pacemaker and ICD Registry.
The study protocol was approved by the Swedish Ethical Review Authority.

| Study design
Patient's medical records were then analyzed in detail and obtained from the regional electronic medical record system (Cosmic, Cambio ® ), unified for recording of healthcare provided by all caregivers to patients in the referral area of Uppsala university hospital. Medical information from other regions were obtained from the responsible cardiologist by standardized telephone interviews. A thorough review of all patients' medical records was performed by two independent physicians and started from the index procedure, defined as the earliest CRT- The presence of a post-procedure complication was analyzed and defined as major if entailing a significant risk leading to reoperation, blood transfusion, prolonged hospital stay >24 h, intervention or readmissions for management, or death, and as minor if associated with patient discomfort, spontaneously resolving or treated on an outpatient basis.
A reoperation was defined as any surgical procedure performed to manage a postoperative complication during the follow-up period.
The index procedures were thus not classified as reoperations. Chest tubes or pericardial drainages were not classified as reoperations.

| Outcomes
The primary endpoint was occurrence of DRI during the follow-up period (prevalence in percentage of patients and incidence per 1000 person-years). Conventional risk factors for device infection were tested and grouped as patient-, procedure-and device-related risk factors (Table S6) in accordance with previous reports. 7 Patients with and without DRI were compared with regard to risk factors in order to identify those of greatest importance. Secondary endpoints were occurrence of other complications of the device surgery during follow-up. For comparisons of complications between patient groups, only those occurring prior to the infection were included in the DRI group. Deaths directly related to the infection or indirectly due to withdrawal of required CRT were defined as DRI-related mortality.
The prevalence of infections and other complications were compared with those reported to the National Swedish Pacemaker and ICD Registry for the study population.

| Statistical analysis
Appropriate parametric statistical tests were used for the analysis of data. All continuous data were examined for normal distribution by

| Study population
A total of 171 patients, median age 74 (range 15-95) years, were included, 21 of whom were referrals from regional hospitals due to temporary lack of CRT-implanters or were concomitant to a cardiac surgery or ablation. The patient characteristics are shown in Table 1.
All patients had symptomatic heart failure and fulfilled conventional criteria for CRT.

| Pre, peri-, and intraprocedural routines
All patients had normal white blood cell count and C-reactive protein prior to surgery.
A whole-body shower with Chlorhexidine soap/shampoo twice with 6 h interval was undertaken prior to surgery. Standard intravenous (iv) antibiotic prophylaxis was Cloxacillin 2 g or Clindamycin 600 mg (in case of allergy) administered 0.5-1 h prior to incision, reiterated at 2 and 8 h after the initial dose for de novo/upgrade/revision operations.
Antiplatelet therapy except for acetylsalicylic acid was withdrawn 5 days before the implantation if not absolutely indicated. Treatment with vitamin K antagonists was continued uninterrupted with an international normalized ratio (INR) <2.5 on the day of surgery while DOACs were interrupted for at least 24 h before surgery. One patient with left ventricular assist device (LVAD) had ongoing heparin treatment.
The hair in the surgical field was cut with clippers. Residuals of monitoring electrodes were removed and one additional wash with Chlorhexidine soap was performed. The skin was then painted with a Chlorhexidine-alcohol solution. Adhesive sheets were used to cover the patient. All operators had full aseptic body gowning and double gloving.
All procedures were performed under local anesthesia and conscious sedation, using axillary/subclavian vein access. A venography of the coronary sinus was obtained in all patients before implanting the left ventricular lead. Each device was secured with a suture and the wound was closed with multiple layer sutures.

| CIED infections
A total of 12 DRI occurred in 171 patients during a mean of 29.6 ± 11 months follow-up (414 person-years). The 12 DRI (10 endocarditis/ sepsis and 2 pocket infections) occurred in 10 patients giving a prevalence of 7%, translating to an incidence of 29/1000 person-year. Nine DRI occurred within 1 year, directly related to the culprit procedure.
Of the 12 DRI, six were classified as possible and six as definite, according to the EHRA criteria for device related infections. 7 Nine of 12 blood cultures were positive and S aureus was the most common agent (42%) ( Table 2). Of the three DRI with negative blood cultures, two had bacterial growth from the pocket alone and one with sepsis clinically was analyzed during ongoing treatment with antibiotics. Lead/valve vegetation was seen in 3 of 11 (27%) patients undergoing transesophageal echocardiography.
All DRI patients were treated with antibiotics according to ESC guidelines. 7 Hardware removal was performed in 9 of 12 (75%) DRI while salvage antibiotic therapy was used alone in the remaining three due to severe comorbidities.

| Risk factors for device infection
The predefined risk factors in infected and non-infected patients are shown in Tables 3 and 4. Time from the start of the skin incision until the wound closure. d The procedure type refers to the culprit procedure in the DRI-group while it represents the last type of procedure in an individual patient at the end of follow-up in the non-DRI group. This means that "de novo" here are de novo index procedures with no reoperation during follow-up. e Requiring reoperation or blood transfusion.

| Comparison with national device registry data
A total of 9/171 (5.3%) DRI occurred within 1 year from the culprit procedure but only 2/171 (1.2%) were reported to the national device registry, reflecting a 4.5-fold underreporting. No procedure or complication related deaths were reported to the national registry, although five DRI-related deaths occurred. There were 59/171 (34.5%) overall major complications during 1 year from the procedure although only 9/171 (5.3%) were reported to the national registry (Table 5).

| DISCUSSION
The observed prevalence of 7% DRI in CRT recipients (incidence of 29/1000 person-years) after mean 2.5 years follow-up is higher than previously reported. 3,6,7,9 Lower rates (1%-3%) have been reported from registries 8 and randomized trials 12,13 while higher (4.2%-4.8%) from observational studies. 6 To the best of our knowledge, in the few studies of CRT procedures using data from case records 6  prone to underreporting of complications as well as misclassifications, the Danish registry is regularly audited and considered to have a high validity.
Although one cannot exclude a center-or surgical technique specific explanation for the high prevalence of DRI in the present study, one likely explanation is the meticulous data review of electronic charts capturing all DRI and thus overcoming cases with incorrect or missing ICD-10 codes. As compared to other studies, patient demography seemed similar, which argues against a selection of patients with more severe comorbidities or anatomical/surgical difficulties.
Still, the present study confirmed several previously described modifiable risk factors, the strongest being reoperations, pocket haematoma, cardiac perforation and indwelling central venous line.
Reoperation, particularly multiple reoperations, is a well-known strong risk factor for DRI. 7,9 Pockets can be colonized by bacteria even after minor reoperations. The total rate of observed reoperations (28.7%) after all CRT procedures and the rate after de novo/upgrade/ revision procedures (36.4%) was unexpectedly high in the present study as compared to prior reports with the highest figure of 9%. 15 The observed rate of major complications (36.8%) was higher than reported previously for CRT-procedures, in the Danish registry 6.7%-11%, and was strongly associated with DRI. 8 The vast majority of major complications were lead revisions mainly due to lead dislodgements, which is a well-known and strong risk factor for DRI. 9 The 18.1% observed frequency of lead-related complications in the present study is high in comparison with previous reports, including a Danish registry in which the risk for any lead complication was 3.6%. 15 A Danish registry study further demonstrated that the operator inexperience was an independent risk factor with an adjusted risk ratio of 1.9 (95% confidence interval 1.4-2.6) for any complication if performed by an operator with an annual volume < 50 procedures. 8 In the present study, the three main operators met the criteria for being experienced based on their annual volume of procedures. The possibly higher rate of DRI for 2 of the operators may be related to differences in procedure time, uneven load of case complexity, variable skills or to nosocomial infection, the latter not analyzed. A low annual hospital device implant volume was also reported as a risk factor for DRI by the Danish registry study with higher complication rates in centers with less than 750 implantations annually. 8 Given the high volume of annual CIED procedures in our hospital, this can unlikely explain the high rate of DRI. Out-of-hours procedures increases the risk by 1.5, 8 but was not a risk factor in the present study.
Post-procedural pocket haematoma, known to increase the risk of infection by nine-fold, 16 was a strong risk factor in our study as well. Pocket haematoma can be prevented by avoiding bridging heparin/low molecular weight heparin therapy, known to increases the risk for haematoma by 5-fold, 17 minimizing broad incisions and withdrawing dual platelet inhibitors if possible a week before surgery. 7 Although oral anticoagulation and antiplatelets were widely used in the present study, bridging was never adapted and heparin was limited to one patient with LVAD. Antibacterial envelopes can prevent pocket infection, 13 but were not routinely used in the present study.
Procedure duration, known to be correlated with DRI, 7,9 has been shown to increase the risk of infection stepwise as compared to durations <30 min, and increases 1.5 times for durations 60-90 min and 2.4 times for those exceeding 120 min. 5 Procedure duration was not a risk factor in the present study possibly due to the wide variation of procedure times even for the same procedure types. Longer proce- CRT-D, being a larger and bulkier device, has been associated with a double risk 3 or even nearly 5.5-fold higher risk for DRI as compared with CRT-P. 9 The present study confirmed these findings as all patients with DRI had CRT-D implanted. crucial to reconsider an indication for the reoperation, it remains unproven if such precautions could have reduced the infection rate in the present study.
The 2.9% device related mortality in the present study, being entirely due to DRI, was higher than in a previous report of one procedure-related death in 5942 patients 8 and 0.8% in another study. 3 The inconsistencies in reports and definitions of mortality rates after CIED procedures makes comparisons difficult. DRI are associated with high fatality rates ranging from 6%-15% at 1 year to 14%-33% at 3 years, 4 even after device system extractions. 7,20 In the present study, the 50% mortality for DRI, mainly related to the salvage therapy, is high compared to previous reports 7,20 with only one study reporting similar rates. 14 Even though the small size of the cohort is a limitation and a center specific explanation for the high rates of infection cannot be excluded, they could be explained by the extensive chart review performed by specialists in electrophysiology and infectious diseases.
The present findings of underreporting to the national device registry are in line with many others, 10,11 and reflect the problems with registries being voluntary, lacking monitoring routines and procedures for continuous feedback. At the hospital level, regular and shared responsibilities for internal quality controls may further increase awareness and reduce a reluctance to report complications.
The findings in the present study should not be interpreted as a questioning of the benefits of CRT. On the contrary, the results emphasize the importance of both recognizing the risks of DRI and of allocating resources towards their prevention.

| CONCLUSIONS
The high prevalence of device infections is in sharp contrast to those reported previously, and to those reported to the national device registry.
Although a center specific explanation cannot be excluded, the high rates could also be due to the extensive chart review and thereby highlight a major issue with present registries, reinforcing a need for automatic and mandatory systems for reporting of device related complications. Hopefully it will give others an incentive to perform similar surveillances.

| STUDY LIMITATIONS
The small cohort of CRT implantations collected retrospectively, may have missed minor complications not recorded in the medical records and precluded a multivariate analysis.