Infection and migration incidence of cardiac implantable electrical devices in Japan: Web‐based survey results

Abstract Web‐based survey was conducted for experiences of cardiac implantable electrical device (CIED) infection and migration in Japan. A total of 155 cardiologists’ answer was collected in January, 2018. CIED includes pacemaker (PM), implantable cardioverter defibrillator (ICD), cardiac resynchronization therapy pacemaker (CRT‐P), and cardiac resynchronization therapy ICD (CRT‐D) and total of 10,499 CIEDs’ experiences of within previous twelve months were reported. CIED includes pacemaker (PM), implantable cardioverter defibrillator (ICD), cardiac resynchronization therapy pacemaker (CRT‐P), and cardiac resynchronization therapy ICD (CRT‐D.). The infection rate of PM, ICD, CRT‐P, and CRT‐D was 0.79%, 0.81%, 0.45%, and 2.0%, respectively, and the device migration rate was 0.68%, 0.64%, 0.45%, and 0.93%, respectively. The overall infection rate was 0.85% and migration rate was 0.68%.


Introduction
Cardiac implantable electronic devices (CIED) are safe and effective for treating cardiac arrhythmias with increase in survival and quality of life. [1][2][3] Implant rate is increasing due to widening of indications and increasing number of generator replacements. 4 Device-related infections (DRI) are amongst the most dreaded complications causing significant morbidity and mortality and generating substantial healthcare costs. [5][6][7] It has been claimed that DRI rate has risen disproportionally to the rate of implantations. 8,9 Increasing proportions of implantable cardioverterdefibrillator (ICD) and cardiac resynchronization therapy (CRT) systems as well as implantations in patients at higher risk of infection may explain these findings.
Reported DRI rates vary considerably depending on definition of device infection, type of operation, device type, and the duration of follow-up. The risk of infection is highest early after the operation, estimated to 0.5-1% for PM patients within the first 12 months, 10-12 0.7-1.7% for ICD patients within 6 months, 13,14 and up to 4.3% within the first 2.5 years of CRT implantation. 15,16 Long-time risk of DRI has been reported sparsely in the literature and often with mixed types of operations, making it difficult to interpret the results. We therefore examined the incidence and incidence rates of DRI in different types of CIED with lifelong follow-up. We also investigated patient-, operation-, and device-related risk factors to identify patients with the highest risk of DRI.

Study population and event definition
All Danish patients who received a CIED from January 1982 to April 2018 were identified in the Danish Pacemaker and ICD Register (DPIR) and included. Type of operation (first implant, replacement, or up-/downgrade) and all reoperations with changes in implanted hardware were recorded. In cases with CIED system removal, the reason was specified. We recorded DRI as CIED removals due to either device pocket infection or systemic infection. The patients were followed until DRI or right censored at death, emigration or lost to follow-up. Patients with more than one operation were censored at time of reoperation and had a new entry for time of risk with the new device. The study was approved by the Danish Data Protection Agency [Jrn. 18/20048(16/39520)] and performed according to declaration of Helsinki. According to Danish law, Ethics Committee approval is not required for registry-based studies.

Data sources
In Denmark, CIED operations are centralized to 14 centres. Pacemaker implantations are performed in all centres while ICD and CRToperations are undertaken only in six, respectively five high-volume centres. The DPIR is a national clinical quality database founded in 1982 by physicians from all implanting centres. Clinical and technical information of all CIED operations in Denmark have been recorded prospectively by the treating physicians. The registry is co-financed by the device industry, but it is controlled by a steering committee of physicians appointed by the Working Group of Arrhythmia under the Danish Society of Cardiology. Data are audited regularly and data quality is considered of high standard. 17 The entire Danish population is registered in the Danish Civil Registrations System and is continuously updated on vital status. 18

Statistical analysis
All statistical analyses were performed using Stata software package (Stata/MP version 14.2. for Windows). Categorical variables are presented with number and/or frequencies while age is presented as median age with interquartile range (IQR). Absolute incidence and incidence rates were calculated for each device and operation type as number of DRI per device operation and over time as DRI per 1000 years of followup and presented with 95% confidence interval (CI). Median time to infection was calculated and presented with IQR.
Multiple-record and multiple-event per subject proportional hazards analyses were used to identify independent patient-, surgery-and devicerelated risk factors of DRI and were presented as hazard ratios (HR) with 95% CI. A surrogate variable (complexity) for the duration and complexity of the operation was constructed by counting the number of pieces of hardware implanted and removed during the same operation. Exact duration of surgery and body mass index (BMI) was available for a subset of patients. We stratified continuous numeric variables in groups and divided implantation centres into PM-centres and ICD-centres. [The three major university centres  and centres implanting ICDs from the year starting ICD implantations] Total number of operations and any previous infection were counted for each patient and included in the analysis. The multivariable regression analysis included sex, device type, operation type, age, implant year, prior infection, centre volume, and complexity. P-values of <0.05 were considered statistically significant. Most patients (81.2%) had PM, 12.3% had ICD, and 6.5% had CRTsystem with-or without ICD. The PM patients had a nearly even gender distribution (males 55.2%) and a median age of 77, while the ICD and CRT populations had predominance of males (70-80%) and were younger ( Table 1). The combined incidence of infection during the device lifetime was 1.19% for PM, 1.91% for ICD, 2.18% for CRT-P, and 3.35% for CRT-D. The risk was lowest in de novo implantations (PM = 0.94%, ICD = 1.66%, CRT-P = 1.60%, and CRT-D = 2.42%) and considerably higher in all other types of operations ( Table 2). The incidence of early infection ranged from 0.16% to 0.30% at 30 days and 0.31% to 0.86% at 90 days for PM and CRT-D, respectively ( Figure 1).

Incidence
The incidence rate of DRI after first implantations was 2.04, 3.84, 4.38, and 6.76/1000 DY for PM, ICD, CRT-P, and CRT-D, respectively. The incidence rate was considerable higher in reoperations with rates at 4.62-5.14/1000 DY for PM replacement and up-/ downgrade operations and as high as 18.94/1000 DY for CRT-D replacement operations ( Table 2 and Supplementary material online, Table S2). The lifetime risk of infection for a patient living with a device through several years is illustrated in Take home figure. .

Risk factors
The multiple proportional hazard analysis showed that device type was associated with a higher risk of infection with uni-and multivariable HR of 1.83 5.29), respectively. We did not find any significant difference between upgrades and replacement operations, whatever the type of device, but the HR increased with each operation summing up in HR of 61.35 for five or more operations as compared to de novo implantations.
When we examined device indication, we found that PM and CRT-P patients that received their device due to AV-block (HR 1.47, CI 1.31-1.67) had the highest risk of DRI, while the risk was highest in those ICD and CRT-D patients that received a primary prophylactic defibrillator unit (HR 1.48, CI 1.23-1.79) (Supplementary material online, Table S3). In univariate analysis, risk of DRI was significantly higher in ICD-centres, but in the multivariable analysis, the risk was significantly lower in ICD-centres.
Male sex, young age, prior DRI, longer operation time, and higher complexity of the operation as well as BMI > 30 were all associated with higher risk of DRI in univariate analysis; however, BMI and operation time were not significantly associated with higher risk of DRI in the multivariable analysis ( Table 3 and Supplementary material online, Table S3). Cumulative incidence function curves of the time to infection with death as competing risk were calculated and graphically illustrated in Figure 2.

Discussion
We found lowest incidence of DRI in PM patients and considerable higher in all other types of CIEDs. The risk was lowest in de novo implantations and several folds higher in all types of reoperations. In multivariable analysis, both ICD and CRT systems were statistically significant independent risk factors as compared to pacemakers. Other independent risk factors were young age, male sex, and prior DRI.

Incidence
We observed a combined incidence of DRI of 1.19% for PM patients with an incidence rate of 2.59/1000 DY. The incidence was lowest in de novo implantation and considerable higher in all types of reoperations. These findings are in line with our previous study 10 but lower than other studies. 20 In ICD patients, the combined incidence of DRI was 1.91% during the device lifetime. A recent study 13 of 200 909 North American ICD and CRT-D patients reported a combined incidence of 1.7% for ICD and CRT-D patients within 6 months of follow-up. Although the risk is highest within the first 12 months after implantation, these numbers are higher than ours, likely because this study included ICD and CRT-D systems. Our incidence also is lower than the 2.2% after 3 months reported in an earlier retrospective study. 21 However, this study included patients who were not operated by specialized electrophysiologist. In our population, all ICD implantations are performed in six centres by specialized electrophysiologist with a high annual number of operations which may add to explain the low incidence of DRI.
The risk of DRI in CRT patients has been reported sparsely in the literature and often as subgroups in mixed device populations. Other studies 15,16 reported combined DRI incidence of about 4.2-4.8% in CRT patients. These studies report few infections, a short follow-up and no stratification for operation type. In our dataset with lifelong follow-up, we observed lower DRI risk with a combined incidence of 2.18% and 3.35% for CRT-P and CRT-D, respectively and even lower after de novo implantation.
Selected CRT patients have been upgraded from a non-CRT device, and the longevity of a CRT-device is considerably shorter than for a PM. Therefore, CRT patients experience more reoperations and a higher total number of operations when compared to PM and ICD patients. These factors might partly explain the increased incidence of DRI, however the incidence is still higher in de novo implantations as compared to pacemakers. This indicates that device type is an independent risk factor, likely because CRT-implantations are more technical challenging and time-consuming. Other risk factors related to comorbidity, which have not been examined in this study, may also be important.

Risk factors
Implantable cardioverter-defibrillator patients had a slightly higher risk of DRI as compared to PM patients, whereas CRT-P and CRT-D systems were associated with considerably higher risk of DRI. De novo implantations had lower risk for DRI in all device types, while any reoperation was associated with a significantly increased risk of DRI, independent of the type of device and the type of reoperation.
It is thus evident that patients undergoing reoperations are highrisk patients, and whenever surgery is performed, it is important to take appropriate actions to prevent infections. Currently, we do not have evidence of the best way to reduce DRI risk in these patients, but a recently published randomized trial 22 has demonstrated that an antibiotic envelope can reduce pocket infections in high-risk patients. Our findings emphasize the importance of considering choice of device type thoroughly when planning CIED therapy in order to minimize risk of early reoperation. Furthermore, the device industry must focus on increasing device longevity. Even though all CIED operations in Denmark are performed by specialized cardiologist, we still find a significant lower risk in ICD-centres. This emphasizes the importance of operators keeping a high annual number of operations.
Young age and male sex were independent patient-related risk factors associated with higher DRI risk. These findings have been described by others in the field of cardiology 23,24 as well as in prosthetic joint infections. 25 There are no obvious reason for this diversity amongst gender; however, bacterial skin colonization differs among genders 26 which may reflect differences in host response to bacterial exposure. Risk of infection in the oldest patients was low compared to younger patients. As we only included DRI with hardware removal, patients, with unrecognized DRI, who died before removal or were conservatively treated, were not included. In the early part of the study period, there was a higher risk related to device removal procedures, and a proportion of elderly fragile patients may not have been considered for device removal procedures. This should however not be an issue within the most recent decade. It is also plausible that a small proportion of elderly patients had unrecognized DRI, since such patients are more often admitted to non-cardiology departments where the infection might remain undiagnosed. Still it is very likely from these data that the DRI risk is considerable higher in younger patients. The reason remains unclear and should be investigated further.
Device indication as a marker of underlying heart disease also seemed to play a role but could not be included in the   27 Obesity often lead to surgical complications 28 and BMI > 30 correlated to a higher risk of DRI in our data, but not statistically significant in the multivariable model. Long duration of surgery was also expected associated with increased DRI risk, but this could not be confirmed, probably due to collinearity with device type. We observed the highest risk of infection in the early period ( Table 3), although rate of infection has increased in recent years. This increase may be due to a higher number of CRT systems and reoperations. Nevertheless, these findings persisted in the multivariable model, suggesting that there are more patient-related risk factors, yet to be revealed.

Strengths and limitations
This study is a retrospective study and therefore conclusions about causality cannot be drawn. Data were entered prospectively in DPIR, and the patients were followed prospectively. With more than 35 years of reporting and follow-up it would be unlikely that the entered data are flawed by any systematic bias. All registers are prone to underreporting of complications as well as misclassifications, but this register is audited regularly and considered complete with respect to changes in hardware with a high validity and quality in registration of complications. 17 The observed incidence of DRI represents the lowest estimate since DRI was counted only for patients who had their system removed due to infection. However, the large number of consecutive patients in a nationwide setting with lifelong follow-up is likely to provide reliable minimal estimates for the absolute risk of DRI.

Conclusion
In this nationwide cohort of the Danish CIED population with a lifelong follow-up, we found the lowest incidence and incidence rate of DRI in de novo PM implantations. The risk was slightly higher in ICD systems and considerable higher in CRT systems. Majority of risk factors for DRI were patient-and operation-related such as total number of operations, type of operation, young age, and male sex.
It is important to be aware of these risk factors and consider the device indication and the choice of device type thoroughly when planning CIED therapy in order to modify adjustable risk factors and minimize risk of early reoperation. Furthermore, the device industry should focus on increasing device longevity to reduce the number of replacement operations.

Supplementary material
Supplementary material is available at European Heart Journal online.

Figure 2
Cumulative incidence function with death as competing risk of the time to first infection in different device types after first implants (Fi) and replacement (Re) operations. The competing risk analysis is adjusted for sex, age, year of implantation, centre, and complexity (number of hardware pieces implanted and removed during the same operation).