Clinical Predictors of Cardiovascular Implantable Electronic Device-Related Infective Endocarditis KATHERINE Y. LE, M.D.,* MUHAMMAD R. SOHAIL, M.D.,† PAUL A. FRIEDMAN, M.D.,‡ DANIEL Z. USLAN, M.D.,§ STEPHEN S. CHA,¶ DAVID L. HAYES, M.D.,‡ WALTER R. WILSON, M.D.,† JAMES M. STECKELBERG, M.D.,† LARRY M. BADDOUR, M.D.,† and for the Mayo Cardiovascular Infections Study Group From the *Mayo School of Graduate Medical Education, Mayo Clinic, Rochester, Minnesota; †Divisions of Infectious Diseases and ‡Cardiovascular Diseases, Department of Medicine, Mayo Clinic, Rochester, Minnesota; §Division of Infectious Diseases, Department of Medicine, David Geffen School of Medicine, University of California Los Angeles, California; and ¶Division of Biostatistics and Informatics, Mayo Clinic, Rochester, Minnesota Background: Cardiovascular implantable electronic device (CIED)-related infective endocarditis (CIEDIE) is a serious complication of cardiac device infection and is associated with increased mortality. At present, there exist no criteria to predict CIED-IE in patients who present with CIED infection. Methods: We retrospectively reviewed all cases of CIED infection seen at Mayo Clinic Rochester between 1991 and 2008. CIED-IE was classified using pathologic and clinical criteria. Clinical predictors of CIED-IE were identified using logistic regression, and quantified using a summary score and plotted against the distribution of CIED-IE. Results: Ninety-three (22.4%) of the 416 patients with CIED infection had CIED-IE. Host factors including chronic immunomodulator therapy exclusive of corticosteroid (odds ratio [OR], 3.79 [confidence interval (CI) 1.10, 13.04]), chronic corticosteroid therapy (OR, 2.15 [CI 0.93, 5.00]), hemodialysis (OR, 3.24 [CI 1.39, 7.55]), or remote infection (OR, 1.77 [CI 0.99, 3.14]) were associated with increased odds of CIED-IE. Patients with CIED-IE were at increased odds of presenting with fever (OR, 3.78 [CI 1.93, 7.40]), or malaise (OR, 1.87 [CI 1.02, 3.41]), and have findings of leukocytosis (OR, 3.61 [CI 1.51, 8.62]). In marked contrast, they were at decreased odds of exhibiting signs/symptoms of infection at the generator pocket site (OR, 0.19 [CI 0.10, 0.36]). Summary scores of 6 and 11 predicted CIED-IE in approximately 50% and 90% of cases, respectively. Conclusions: Certain comorbid conditions and systemic manifestations of infection were associated with CIED-IE. In contrast, pocket site infection was negatively associated with CIED-IE. These findings should assist clinicians in identifying patients who would more likely benefit from further investigation of CIED-IE with transesophageal echocardiography. (PACE 2011; 34:450–459) endocarditis, predictors, infection, pacemaker, defibrillator, cardiovascular implantable electronic device Introduction Implantation of cardiovascular implantable Disclosures: All < $10,000. electronic devices (CIEDs), that include permaPAF: Honoraria/Consultant: Medtronic, Guidant, Astra Zeneca. nent pacemakers and implantable cardioverterSponsored research: Medtronic, Astra Zeneca via Beth Israel, Guidant, St. Jude, Bard. defibrillators, has rapidly increased in the United Intellectual property rights: Bard EP, Hewlett Packard, States over the past two decades.1,2 According Medical Positioning, Inc. to one estimate, the implantation rate of CIEDs DZU: Research: American Heart Association. increased by 42% among Medicare beneficiaries Honoraria/Consultant: Biotronik, Cubist, TyRx Pharma, Inc. from 1990 to 1999.3 Unfortunately, increasing DLH: Honoraria: Medtronic, Boston Scientific, St. Jude rates of CIED infection have also been observed, Medical, ELA Medical, Biotronik. and are much higher than expected based on the Royalty payments: UpToDate; Wiley-Blackwell. Medical advisory board: Boston Scientific, St. Jude Medical, implantation rates alone.3,4 Pixel Velocity. CIED infection can present as a pulseSteering committee member: Medtronic, St. Jude Medical. generator pocket infection or as bloodstream LMB: Royalty payments: UpToDate. Editorship: Massachusetts Medical Society (Journal Watch Infectious Diseases); ACP/PIER editorial consultant. MRS: Honoraria/Consultant: TyRx Pharma, Inc. All other authors: No disclosures. Address for reprints: Katherine Y. Le, M.D., M.P.H., Mayo School of Graduate Medical Education, 200 First Street SW, Rochester, MN 55905. Fax: 507-255-7767; e-mail: le.katherine1@mayo.edu Received July 19, 2010; revised October 12, 2010; accepted October 20, 2010. doi: 10.1111/j.1540-8159.2010.02991.x  C 2010 Wiley Periodicals, Inc. C 2010, The Authors. Journal compilation  450 April 2011 PACE, Vol. 34 PREDICTORS OF CARDIAC DEVICE-RELATED ENDOCARDITIS infection, or both, with or without CIED-related infective endocarditis (CIED-IE).5,6 Reported rate of CIED-IE in earlier publications ranges from 10% to 23% among patients with CIED infection.7,8 The prevailing hypothesis regarding pathogenesis of CIED-IE includes the initial development of infection at the device generator site with subsequent extension of infection to involve the leads. Hematogenous seeding of the device from a distant infectious focus is thought to be a less common occurrence.9 In most cases, a diagnosis of CIED-IE requires transesophageal echocardiogram (TEE) to detect infected vegetations on electrode leads and/or cardiac valves. Transthoracic echocardiogram (TTE) has limited sensitivity (30% to 40%) when compared to TEE (90% to 95%) in detecting vegetations in adults.5,8,10–12 However, TEE is an invasive procedure with potential complications from conscious sedation, insertion, and manipulation of the TEE probe, and is not readily available at some medical centers.13 For these reasons, it is neither practical nor desirable for every patient who presents with CIED infection to undergo investigation with TEE. Consequently, it is imperative that patients who are more likely to have CIED-IE be identified so that timely investigation with TEE is done. Prompt identification of CIED-IE has important therapeutic and prognostic implications. Experts have recommended that patients with CIED-IE that involves cardiac valves receive early and complete explantation of the infected device with a prolonged duration of antimicrobial therapy.6,14,15 In addition, prompt and appropriate management of CIED-IE is crucial because patients with this complication can have worse outcomes.8 A systematic analysis to identify clinical predictors associated with CIED-IE has not been published. Moreover, earlier investigations that addressed the subject of CIED-IE have used modified Duke criteria16 that were originally intended to classify cases of valvular endocarditis and has limitations when applied to CIED-IE cases. The aim of our investigation was to identify patients who were more likely to have CIED-IE, so that optimal management, which includes the selective use of TEE, is achieved. Methods We retrospectively reviewed all cases of CIED infection at Mayo Clinic Rochester (MCR) between January 1, 1991, and December 31, 2008. Cases were identified in the Mayo Clinic Heart Rhythm Device Database, the surgical index, and the computerized central diagnostic index. All patients consented to use of their medical records for research purposes. The Mayo Foundation PACE, Vol. 34 Institutional Review Board approved the study protocol. Definitions CIED infection was previously defined by our group6,8 as clinical evidence of device infection at the generator pocket, or microbiologically confirmed device infection based on positive cultures from the generator pocket, lead(s), or blood (in the presence of local inflammatory signs at generator pocket or absence of another source of bloodstream infection and resolution of bloodstream infection after device explantation). CIED-IE cases were classified using pathologic evidence of CIED lead or valve infection for patients who underwent surgical removal, or using a clinical criterion for a patient who underwent percutaneous device removal. Clinically, patients were classified as having CIEDIE if they had positive echocardiographic findings and two or more positive blood cultures for typical skin organisms (coagulase-negative staphylococci [CoNS], Corynebacterium species, Propionibacterium species), or one positive blood culture for all other microorganisms. Positive echocardiographic findings for CIED-IE were defined as presence of an oscillating intracardiac mass on cardiac valve or supporting structures (in the path of regurgitant jets), or CIED leads in the absence of an alternative anatomic explanation, or visualization of a cardiac abscess, or new dehiscence of prosthetic valve. Patients with CIED-IE as defined by the clinical criteria were further classified as having left-sided CIED-IE if they had any positive left-sided echocardiographic findings, and those having right-sided CIEDIE if they had positive right-sided valvular or lead-associated echocardiographic findings. The remaining patients were rejected as having CIEDIE. Demographic and Device-Related Variables Demographic variables included the patient’s age in years, gender, and race/ethnicity. Body mass index (BMI) was calculated as kilograms per meter squares. Device characteristics included device type, device age, number of procedures, the indication for CIED implantation, procedure type, and anatomical site of implantation. Comorbid Conditions and Clinical Features at Presentation Patients’ medical records were reviewed for comorbid conditions, presenting systemic signs/symptoms, inflammatory signs at the generator pocket site, and laboratory parameters for potential predictors of CIED-IE. April 2011 451 LE, ET AL. Statistical Analyses Statistical analyses were conducted using the SAS version 9.1.3 software (SAS Institute Inc., Cary, NC, USA). Bivariate comparisons were performed using Student’s t-test for continuous and Pearson χ 2 or Fisher’s Exact test for categorical variables. Clinical predictors of CIED-IE were first identified using univariate logistic regression and later summarized by multivariate logistic regression using stepwise elimination. Using the findings from our multivariate logistic regression models, we then devised a scoring system containing both statistically and clinically meaningful predictors to better quantify their individual and cumulative effects on the probability of CIED-IE. The resulting distribution of the summary score was plotted against the distribution of CIED-IE cases. To evaluate our CIED-IE classification scheme’s ability to discriminate outcomes, survival curves were generated. For the survival analysis, we considered two primary end points: infection-related death during index hospitalization and all-cause mortality. For infection-related death, survival was considered from time of presentation to MCR to time of hospital discharge, and data were censored at the time of infectionrelated death. For all-cause mortality, survival was considered from time of presentation to MCR to time of last follow-up, and data were censored at time of death. Kaplan-Meier survival curves were plotted and log-rank test was used to determine the univariate significance of CIEDIE on outcome of interest. A two-sided P value of 0.05 or less was considered to indicate statistical significance. Results We identified 416 patients who met our case definition for CIED infection (Table I) and elderly white males were the predominant group within this cohort. Sixty percent of device-related infections involved permanent pacemakers. The subset of patients with CIED-IE had devices implanted for a longer period of time than did the overall cohort. Device placement indications included heart block, sinus node disease, or ventricular arrhythmia in the bulk of patients. Most were initial implants or system revisions/upgrades and were located at the original site of implantation on the left chest wall. Diagnostic echocardiography was performed in 82% of patients. Of those who underwent echocardiography, 22.8% underwent TTE alone, while 77.1% had TEE (47.2% TEE only, 29.9% both). An echocardiographic study was obtained 452 in 96.9% of patients who had bloodstream infection. Of the 416 patients with CIED infection, 47 underwent surgical (27 immediate and 20 after failed percutaneous extraction), 368 underwent percutaneous device removal, and 21 were managed conservatively without device removal. Among patients who underwent immediate surgical device removal, only two were due to concerns regarding vegetation size. Many of these patients had additional indications for surgical removal of devices via median sternotomy. Lead vegetation specimens were available for 23 of 47 patients who underwent surgical removal. Of these, 18 demonstrated histopathologic evidence of CIEDIE. Twenty-four patients who underwent cardiothoracic surgery for lead removal did not have lead specimens submitted for histopathologic evaluation. Clinical criteria were used in these 24 cases, the 348 cases that underwent percutaneous device removal, and the 21 cases that did not undergo device removal to define CIEDIE. Of these 393 patients, 26 (6.6%) had leftsided valvular IE and 49 (12.4%) had right-sided valvular IE or had lead-associated IE (Fig. 1). When combined with pathologic criteria, 93 (22.4%) were accepted and 323 (77.6%) were rejected cases of CIED-IE (Fig. 1, Table II). CoNS (38.2%) and Staphylococcus aureus (31.0%) caused the large majority of CIED infections; gram-negative bacilli and nonstaphylococcal gram-positive organisms were less commonly identified. Patients with CIED-IE were at increased odds of infection by S. aureus (43.0% vs 27.5%), or gram-negative bacilli (11.8% vs 5.8%). There was no difference in the distribution of CoNS between the two groups (Fig. 2). In the univariate setting, host factors including chronic immunomodulator therapy exclusive of corticosteroid (>1 month prior to onset of infection), chronic corticosteroid therapy (>1 month prior to onset of infection), hemodialysis, implanted central venous catheter, and the presence of remote focus of primary infection were positively associated with CIED-IE (Table II). Similarly the presenting systemic signs/symptoms of fever, tachycardia, chills, diaphoresis, malaise, nausea, anorexia, hypotension, congestive heart failure symptoms, and metastatic infectious foci were also associated with CIED-IE. In contrast, inflammatory signs/symptoms at the generator pocket site were less commonly associated with CIED-IE. Of the laboratory parameters, leukocytosis, anemia, elevated serum creatinine, and high erythrocyte sedimentation rate (ESR) were positively associated with diagnosis of CIED-IE (Table II). April 2011 PACE, Vol. 34 PREDICTORS OF CARDIAC DEVICE-RELATED ENDOCARDITIS Table I. Comparison of Patient Demographics and CIED Features in Cases with and Without Complicating Infective Endocarditis* Demographics Age (years), mean ± standard deviation (SD) Male Race/ethnicity (white) BMI, mean ± SD Device characteristics Permanent pacemakers Single chamber Implanted at MCR Age of device (years), mean ± SD Number of procedures, mean ± SD Indication for CIED implantation Heart block Sinus node disease Ventricular arrhythmia Hypertrophic cardiomyopathy Supraventricular tachycardia Syncope Other Procedure type Initial implant System revision/upgrade Lead revision/insertion Generator replacement Other Site of generator At original site of implantation† Left chest CIED-IE (n = 93) CIED-I (n = 323) P-Value 67.6 ± 13.5 65 (69.8) 85 (91.4) 29.3 ± 5.9 69.2 ± 15.4 249 (77.0) 292 (90.4) 28.2 ± 6.2 0.334 0.155 0.771 0.128 61 (65.5) 25 (26.8) 26 (27.9) 3.0 ± 2.8 1.9 ± 1.1 196 (60.6) 65 (20.1) 106 (32.8) 2.1 ± 2.8 2.1 ± 1.3 0.390 0.163 0.374 0.023 0.163 0.682 30 (32.2) 29 (31.1) 20 (21.5) 0 (0) 0 (0) 2 (2.1) 12 (12.9) 97 (30.0) 89 (27.5) 83 (25.7) 4 (1.2) 2 (0.6) 14 (4.3) 34 (10.5) 44 (49.4) 19 (21.3) 7 (7.8) 19 (21.3) 4 (4.3) 131 (42.8) 68 (22.2) 24 (7.8) 80 (26.1) 20 (6.1) 62 (69.6) 27 (51.9) 223 (70.5) 113 (58.8) 0.696 0.868 0.395 *All values are expressed as number (percentage), unless indicated otherwise. † Most recent device-related procedure before presenting with CIED infection. After adjusting for age, gender, race/ethnicity, and BMI, we incorporated significant associations demonstrated by univariate analyses of comorbid conditions, systemic signs/symptoms, and laboratory parameters in multivariable logistic regression models. The odds of CIEDIE remained elevated among patients receiving chronic immunomodulator therapy exclusive of corticosteroid (odds ratio [OR] 3.79 [1.10, 13.04]), chronic corticosteroid therapy (OR 2.15 [0.93, 5.00]), hemodialysis (OR 3.24 [1.39, 7.55]), or had remote infection (OR 1.77 [0.99, 3.14]). These patients were at increased odds of presenting with systemic signs/symptoms of fever (OR 3.78 [1.93, 7.40]), or malaise (OR 1.87 [1.02, 3.41]), but in the absence of infectious signs/symptoms at the generator pocket site (OR 0.19 [0.10, 0.36]). The most meaningful laboratory parameter remained leukocytosis (OR 3.61 [1.51, 8.62]) (Table III). PACE, Vol. 34 The predictive summary score reflecting the negative and positive effects of significant host factors, presenting sign/symptoms and laboratory predictors on CIED-IE diagnosis, ranged from −5 to +21 (Table III). When plotted against the distribution of CIED-IE, at a summary score of 6, approximately 50% of patients had CIED-IE. A sigmoidal relationship was seen with 75% of cases classified as CIED-IE at a summary score of 9, and 90% at a summary score of 11 (Fig. 3). Patients were hospitalized for an average of 17 days, during which time 5.7% of them experienced CIED infection-related death. After hospital discharge, patients were followed for an average of 2 years. At the time of last follow-up, the all-cause mortality rate was 33.5%. Survival curves for those classified as having leftsided valvular IE and right-sided valvular IE or lead-associated CIED-IE were similar. Likewise, April 2011 453 LE, ET AL. CIED infection N=416 Immediate surgical extraction n=27 Percutaneous extraction n=368 No device removal n=21 (Clinical criteria) Surgical extraction after failed percutaneous extraction n=20 Successful percutaneous extraction n=348 Lead specimen n=23 No lead specimen n=24 (Pathologic criteria) (Clinical criteria) (Clinical criteria) Left-sided valvular IE n=26 Lead-associated CIED-IE n=18 Right-sided valvular IE, or lead-associated CIED-IE n=49 No lead-associated CIED-IE n=5 Rejected cases of CIED-IE n=318 Figure 1. Schema outlining device removal technique, availability of a histopathologic specimen, and criteria applied in cases of CIED infection. survival was similar among patients with implantable cardioverter defibrillators when compared to those with permanent pacemakers. Both in-hospital (P = 0.005) and overall survival (P = 0.025) were higher among patients who did not have CIED-IE (Figs. 4 and 5). We also examined survival in the 75 patients who did not undergo echocardiography as part of the diagnostic evaluation for CIED infection during hospital admission and observed no significant differences in index hospitalization (96.0% vs 92.0%, P = 0.235) or overall survival (73.3% vs 64.5%, P = 0.144) among them as compared to that of patients who underwent echocardiography. Discussion The current investigation includes one of the largest cohorts of CIED-IE cases analyzed to date, and offers several key contributions to a contemporary clinical profile of CIED infection. First, we propose clinical criteria to better define CIED infection cases that are complicated by CIED-IE, and examine short- and long-term patient outcomes under this classification scheme. Second, we identify clinical predictors of CIED-IE, and develop a summary score to quantify these predictors in our study population. Third, our predictive model may provide valuable insights into CIED-IE disease pathogenesis. 454 While our CIED-IE rate of 22.4% appears similar to reported rates ranging from 10% to 23%,7,8,14 meaningful comparisons across study populations have been difficult to perform due to lack of a uniform CIED-IE classification scheme. Previous investigators have utilized either the original17 or modified Duke criteria16 to classify CIED-IE. However, Duke criteria were not developed to define CIED-IE17,18 and alternative criteria designed specifically for CIED infection are needed. The majority of infected CIED leads in contemporary practice are removed percutaneously. During percutaneous device removal, vegetations usually become dislodged from the lead19–21 and are not available for histopathologic confirmation of CIED-IE. This is in contrast to valvular endocarditis where surgically resected tissue from valves, or less often emboli, are available for histopathologic examination6,8 and can be used to validate proposed clinical criteria. Some investigators have attempted to overcome this limitation by using positive lead tip cultures (as equivalent of valve tissue cultures in valvular endocarditis) to define CIED-IE.8,10,22,23 However, lead tips can become contaminated when extracted through infected generator pockets8 and subsequent cultures may only be reliable when leads are removed via other techniques that are less frequently utilized for lead removal, such as via femoral vein incision24 or median sternotomy. April 2011 PACE, Vol. 34 PREDICTORS OF CARDIAC DEVICE-RELATED ENDOCARDITIS Table II. Univariate Comparisons of Comorbid Conditions and CIED Features in Cases with and Without Complicating Endocarditis* CIED-IE (n = 93) CIED-I (n = 323) Comorbid conditions Active malignancy 12 (12.9) 39 (12.0) Autoimmune disease 10 (10.7) 17 (5.2) Immunomodulator therapy 8 (8.6) 5 (1.5) Corticosteroid therapy 13 (13.9) 17 (5.2) Splenectomy 3 (3.2) 4 (1.2) Diabetes mellitus 29 (31.1) 86 (26.6) Hemodialysis 13 (13.9) 13 (4.0) CAD 52 (55.9) 185 (57.2) Status-post CABG 26 (27.9) 91 (28.1) CHF 59 (63.4) 175 (54.1) EF 44.7 ± 16.9 42.5 ± 16.6 Atrial fibrillation 31 (33.3) 120 (37.1) COPD 19 (20.4) 48 (14.8) Cirrhosis 1 (1.0) 3 (0.9) Chronic skin condition 8 (8.6) 22 (6.8) Implanted CVC 11 (11.8) 16 (4.9) Prosthetic heart valve 15 (16.1) 46 (14.2) Vascular graft 3 (3.2) 6 (1.8) Previous device infection 11 (11.8) 37 (11.4) Remote infection 26 (27.9) 51 (15.7) Presenting signs/symptoms – systemic Fever 74 (79.5) 94 (29.1) Tachycardia 15 (16.1) 18 (5.5) Chills 54 (58.0) 73 (22.6) Diaphoresis 25 (26.8) 32 (9.9) Malaise 57 (61.2) 79 (24.4) Nausea 15 (16.1) 15 (4.6) Anorexia 19 (20.4) 30 (9.2) Hypotension 21 (22.5) 25 (7.7) CHF symptoms 33 (35.4) 78 (24.1) Metastatic infectious foci 11 (11.8) 9 (2.7) Presenting signs/symptoms – inflammatory signs at generator pocket site Erythema 20 (21.5) 190 (58.8) Pain 16 (17.2) 103 (31.8) Swelling 19 (20.4) 169 (52.3) Warmth 14 (15.0) 73 (22.6) Tenderness 15 (16.1) 119 (36.8) Drainage 2 (6.0) 16 (11.5) Purulent drainage 11 (11.8) 88 (27.2) Skin ulceration 7 (7.5) 61 (18.8) Erosion of lead/generator 3 (3.2) 91 (28.1) Intraoperative signs of infection 30 (32.6) 225 (70.0) Laboratory parameters 13.7 ± 7.9 9.2 ± 4.6 WBC (×109 cells/L), mean ± SD Leukocytosis 57 (61.2) 108 (33.4) Hematocrit (%), mean ± SD 34.8 ± 5.5 36.9 ± 5.8 Anemia 70 (75.2) 195 (60.3) OR [95% CI] P-Value 1.07 [0.54, 2.15] 2.16 [0.95, 4.91] 5.98 [1.90, 18.75] 2.92 [1.36, 6.27] 2.65 [0.58, 12.09] 1.24 [0.75, 2.06] 3.87 [1.72, 8.68] 0.94 [0.59, 1.50] 0.98 [0.59, 1.65] 1.46 [0.91, 2.36] 1.01 [0.99, 1.02] 0.84 [0.52, 1.37] 1.47 [0.81, 2.65] 1.15 [0.11, 11.27] 1.28 [0.55, 2.99] 2.57 [1.15, 5.75] 1.15 [0.61, 2.18] 1.76 [0.43, 7.18] 1.03 [0.50, 2.12] 2.07 [1.20, 3.56] 0.830 0.063 0.002 0.005 0.206 0.387 0.001 0.815 0.967 0.113 0.286 0.500 0.199 0.898 0.557 0.021 0.650 0.429 0.920 0.008 9.48 [5.42, 16.58] 3.25 [1.57, 6.75] 4.74 [2.91, 7.72] 3.34 [1.86, 6.00] 4.89 [3.00, 7.96] 3.94 [1.85, 8.42] 2.50 [1.33, 4.70] 3.47 [1.84, 6.55] 1.72 [1.05, 2.83] 4.68 [1.87, 11.67] <0.001 0.001 <0.001 <0.001 <0.001 <0.001 0.004 <0.001 0.030 <0.001 0.19 [0.11, 0.33] 0.44 [0.24, 0.79] 0.23 [0.13, 0.40] 0.60 [0.32, 1.13] 0.33 [0.18, 0.59] 0.49 [0.10, 2.25] 0.35 [0.18, 0.70] 0.35 [0.15, 0.79] 0.08 [0.02, 0.27] 0.20 [0.12, 0.33] <0.001 0.006 <0.001 0.117 <0.001 0.360 0.002 0.011 <0.001 <0.001 1.13 [1.08, 1.18] 3.15 [1.95, 5.07] 0.93 [0.89, 0.97] 1.99 [1.18, 3.36] <0.001 <0.001 0.002 0.009 Continued. PACE, Vol. 34 April 2011 455 LE, ET AL. Table II. Continued. Creatinine (mg/dL), mean ± SD ESR (mm/h), mean ± SD† CIED-IE (n = 93) CIED-I (n = 323) OR [95% CI] 1.8 ± 1.5 49.5 ± 31.6 1.4 ± 0.7 32.6 ± 28.2 1.34 [1.09, 1.65] 1.01 [1.00, 1.03] P-Value 0.004 0.004 *All values are expressed as number (percentage) unless indicated otherwise. CAD = coronary artery disease; CABG = coronary artery bypass grafting; CHF = congestive heart failure; EF = ejection fraction; COPD = chronic obstructive pulmonary disease; CVC = central venous catheter; WBC = white blood cell count; ESR = erythrocyte sedimentation rate; SD = standard deviation. Immunomodulator therapy (for >1 month preceding CIED infection, exclusive of corticosteroid); corticosteroid therapy (prednisone use for >1 month preceding CIED infection); chronic skin condition (dermatitis, psoriasis, pressure ulcers); remote infection (concomitant or preceding infection at a distant site such as skin and soft tissue infection, deep abscess, dental infection); hypotension (systolic blood pressure <90 or diastolic blood pressure <60 mmHg); CHF symptoms (diagnosed by a physician); leukocytosis (peripheral WBC count >10×109 cells/L); anemia (hematocrit <40% for men and 36% for women). † ESR n = 36 for CIED-IE; n = 116 for CIED-I. Therefore, we developed clinical criteria consisting of bloodstream infection and echocardiographic findings to define cases of CIEDIE. Indirect support for the accuracy of our classification scheme is evident in short- and longterm survival data. Patients who were classified as having CIED-IE had decreased survival, both during the index hospitalization period and overall, when compared to patients who were rejected as cases of CIED-IE (Figs. 4 and 5). Having established the clinical relevance of our diagnostic criteria, we next sought to identify clinical predictors for CIED-IE. Host factors including receipt of immunomodulator therapy, corticosteroid therapy, hemodialysis, or presence of remote primary sources of infection increased the odds of having CIED-IE. Patients with these risk factors, particularly when presenting with 50 p=0.004 Distribution percentage (%) CIED-IE CIED-I Overall p<0.001 p=0.271 40 30 p=0.002 20 p=0.050 10 0 CoNS Staph aureus GNB Others Organism Figure 2. Distribution of pathogens in cases of CIED infection and CIED-IE. 456 fever, malaise, or with leukocytosis, should undergo early evaluation for endocarditis with TEE. Our data are supported by a recent investigation in which the absence of fever, leukocytosis, and elevated ESR had a negative predictive value of 100% in ruling out non-staphylococcal CIEDIE.25 Although complete removal of an infected CIED system is a requisite for cure, regardless of the clinical manifestations of infection (pocket infection vs CIED-IE), patients with device-related endocarditis have higher in-hospital mortality and the diagnosis should prompt urgent removal of the infected device and longer duration of antimicrobial therapy to achieve cure.6,15,26 Expedited diagnosis and appropriate management is crucial due to the decreased in-hospital and overall survival in this group of patients (Figs. 4 and 5). When these clinical predictors were quantified alone and through a summary predictive score and plotted against the distribution of CIEDIE cases, the contribution of each variable was remarkable. At relatively low predictive summary scores of 6, 9, and 11, substantial portions (50%, 75%, and 90%) of the study population had CIED-IE. Our work also prompts a reevaluation of the current understanding of CIED-IE pathogenesis. At present, direct extension of infection from the generator pocket to CIED leads and/or cardiac valves27–30 is thought to be operative in most CIED infection cases. However, the low likelihood of CIED-IE among patients with clinical findings of pocket site infection was striking and suggests that novel pathogenic mechanisms, possibly related to hematogenous seeding, may be operative. Moreover, the finding of a negative association may be extremely valuable in defining patient April 2011 PACE, Vol. 34 PREDICTORS OF CARDIAC DEVICE-RELATED ENDOCARDITIS Table III. Multivariable Logistic Regression Models Predicting CIED-IE (Adjusted for Age, Gender, Race/Ethnicity, and BMI) OR [95% CI] Model 1—Comorbid conditions Immunomodulator therapy Corticosteroid therapy Hemodialysis Remote infection 3.79 [1.10, 13.04] 2.15 [0.93, 5.00] 3.24 [1.39, 7.55] 1.77 [0.99, 3.14] Model 2—Presenting signs/symptoms Fever Malaise Metastatic infectious foci Signs/symptoms at generator pocket Model 3—Laboratory parameters Leukocytosis Anemia Serum creatinine (mg/dL) ESR (mm/h) +4 +2 +3 +2 3.78 [1.93, 7.40] 1.87 [1.02, 3.41] 1.92 [0.65, 5.67] 0.19 [0.10, 0.36] <0.001 0.041 0.234 <0.001 +4 +2 3.61 [1.51, 8.62] 2.29 [0.78, 6.67] 0.92 [0.61, 1.40] 1.01 [0.99, 1.02] 0.003 0.127 0.717 0.323 100 CIED-IE (%) 75 50 25 0 -1 1 3 5 7 9 11 13 15 17 19 Predictor summary score Figure 3. Association between predictor summary score and probability of CIED-IE diagnosis. PACE, Vol. 34 −5 +4 −5–21 management schemes that include the optimal use of TEE. Our observations support limited preliminary data that described discordant findings between signs of generator pocket infection and echocardiographic evidence of lead infection among patients with S. aureus bacteremia,31,32 and observations that implicate soft tissue infections and intravascular catheters as sources of bloodstream infection that resulted in hematogenous seeding of device lead(s) and cardiac valves(s).32 Our investigation has several limitations inherent in retrospective study designs, including biases. We attempted to minimize bias by using objective criteria and standardized definitions to -3 Predictor Score 0.034 0.073 0.006 0.050 Predictor summary score -5 P-Value categorize cases. However, lack of a uniform and consistent case definition and diagnostic criteria for CIED-IE in earlier publications limits our ability to make direct comparisons. No patient was excluded based on age, gender, ethnicity, or severity of illness. Laboratory and echocardiographic investigations were not uniformly applied in all cases, but were conducted at the discretion of the primary physician. However, our subgroup analysis suggests that this had a nondifferential impact on short- and long-term patient outcomes. As a tertiary care center, referral bias could have impacted our clinical data. Patients in our cohort, for example, could have more comorbid conditions as compared to that in the general population. Thus, selection biases might limit the generalization of our findings to other populations. We believe, however, that many of the factors related to the development of CIED infection are operative in all patients, and this notion is supported by findings from other medical centers.9,31,32 Nevertheless, validation of both our modified CIED-IE case definition and the predictive scoring system in cohorts at other institutions is needed. In conclusion, we defined CIED-IE cases by using clinical criteria that accommodated the uniqueness of lead-related endocarditis; in addition, a scoring system was developed that predicted patients with CIED-IE. Appropriate and judicious use of invasive diagnostic procedures, such as TEE, is more relevant than ever with the current focus on cost-effectiveness and optimal April 2011 457 LE, ET AL. 100 CIED-I Hospital Survival (%) 80 CIED-IE 60 40 20 Log Rank p=0.005 0 0 20 40 60 80 100 Time (days) Figure 4. Unadjusted CIED infection-related in-hospital survival curves. utilization of health-care resources. Our investigation, which included outcomes data, suggests that TEE may not be needed in patients with negative blood cultures and clinical evidence of CIED infection limited to the generator pocket site. Moreover, the negative correlation between pocket site changes and likelihood of CIED-IE challenges our current dogma that the generator pocket site serves as the predominant source of lead infection and prompts us to reevaluate alternative mechanisms of infection pathogenesis. The results of our investigation, however, should prompt additional study to define the optimal use of TEE in the management of CIED infection. At present, it would be premature to provide specific recommendations in the management of CIED infection based on the findings demonstrated in our current investigation. 100 Overall Survival (%) 80 60 CIED-I CIED-IE 40 20 Log Rank p=0.025 0 0 2 4 6 8 10 Time (years) Figure 5. Unadjusted overall survival curves at the time of last follow-up. 458 April 2011 PACE, Vol. 34 PREDICTORS OF CARDIAC DEVICE-RELATED ENDOCARDITIS Acknowledgments: This investigation utilized resources of the Mayo Cardiovascular Infections Study Group (Le KY, Sohail MR, Baddour LM, Steckelberg JM, Wilson WR, Enzler MJ, Chung HH, Friedman PA, Hayes DL, Anavekar NS, Nkomo VT, Dib C, Madhavan M, Thomas JM, Raza SS, Correa de Sa DD, Bachuwar A, Sultan OW, Abou Ezzeddine OF, Habib A, Patel R, Williamson EE, Kalra M, Edwards WD, Maleszewski JJ [Mayo Clinic Rochester]; Hellinger WC, Lynady CJ, Kusumoto F [Mayo Clinic Florida]; Virkram HR, and Keckich DW [Mayo Clinic Arizona]). This study was supported, in part, by “Small Grants” award from the Division of Infectious Diseases, Department of Medicine, Mayo Clinic College of Medicine. We wish to thank Joanne E. Spencer for her important contribution in data collection. References 1. Goldberger Z, Lampert R. Implantable cardioverter-defibrillators: Expanding indications and technologies. JAMA 2006; 295:809–818. 2. Birnie D, Williams K, Guo A. Reasons for escalating pacemaker implants. Am J Cardiol 2006; 98:93–97. 3. Cabell CH, Heidenreich PA, Chu VH, Moore CM, Stryjewski ME, Corey GR, Fowler VG Jr. Increasing rates of cardiac device infections among Medicare beneficiaries: 1990–1999. Am Heart J 2004; 147:582–586. 4. Uslan DZ, Tleyjeh IM, Baddour LM, Friedman PA, Jenkins SM, St Sauver JL, Hayes DL. Temporal trends in permanent pacemaker implantation: A population-based study. Am Heart J 2008; 155:896– 903. 5. Klug D, Balde M, Pavin D, Hidden-Lucet F, Clementy J, Sadoul N, Rey JL, et al.; PEOPLE Study Group. Risk factors related to infections of implanted pacemakers and cardioverter-defibrillators: Results of a large prospective study. Circulation 2007; 116:1349– 1355. 6. Sohail MR, Uslan DZ, Khan AH, Friedman PA, Hayes DL, Wilson WR, Steckelberg JM, et al. Management and outcome of permanent pacemaker and implantable cardioverter-defibrillator infections. J Am Coll Cardiol 2007; 49:1851–1859. 7. Arber N, Pras E, Copperman Y, Schapiro JM, Meiner V, Lossos IS, Militianu A, et al. Pacemaker endocarditis. Report of 44 cases and review of the literature. Medicine 1994; 73:299–305. 8. Sohail MR, Uslan DZ, Khan AH, Friedman PA, Hayes DL, Wilson WR, Steckelberg JM, et al. Infective endocarditis complicating permanent pacemaker and implantable cardioverter-defibrillator infection. Mayo Clin Proc 2008; 83:46–53. 9. Camus C, Leport C, Raffi F, Michelet C, Cartier F, Vilde JL. Sustained bacteremia in 26 patients with a permanent endocardial pacemaker: Assessment of wire removal. Clin Infect Dis 1993; 17:46–55. 10. Cacoub P, Leprince P, Nataf P, Hausfater P, Dorent R, Wechsler B, Bors V, et al. Pacemaker infective endocarditis. Am J Cardiol 1998; 82:480–484. 11. Victor F, De Place C, Camus C, Le Breton H, Leclercq C, Pavin D, Mabo P, et al. Pacemaker lead infection: Echocardiographic features, management, and outcome. Heart 1999; 81:82–87. 12. Vilacosta I, Sarriá C, San Román JA, Jiménez J, Castillo JA, Iturralde E, Rollán MJ, et al. Usefulness of transesophageal echocardiography for diagnosis of infected transvenous permanent pacemakers. Circulation 1994; 89:2684–2687. 13. Peterson GE, Brickner ME, Reimold SC. Transesophageal echocardiography: Clinical indications and applications. Circulation 2003; 107:2398–2402. 14. Chua JD, Wilkoff BL, Lee I, Juratli N, Longworth DL, Gordon SM. Diagnosis and mangement of infections involving implantable electrophysiologic cardiac devices. Ann Intern Med 2000; 133:604– 608. 15. Baddour LM, Epstein AE, Erickson CC, Knight BP, Levison ME, Lockhart PB, Masoudi FA, et al. Update on cardiovascular implantable electronic device infections and their management. A scientific statement from the American Heart Association. Circulation 2010; 1:1–20. 16. Li JS, Sexton DJ, Mick N, Nettles R, Fowler VG Jr, Ryan T, Bashore T, et al. Proposed modifications to the Duke criteria for the diagnosis of infective endocarditis. Clin Infect Dis 2000; 30:633–638. 17. Durack DT, Lukes AS, Bright DK. New criteria for diagnosis of PACE, Vol. 34 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. April 2011 infective endocarditis: Utilization of specific echocardiographic findings. Duke Endocarditis Service. Am J Med 1994; 96:200– 209. Klug D, Wallet F, Kacet S, Courcol RJ. Detailed bacteriologic tests to identify the origin of transvenous pacing system infections indicate a high prevalence of multiple organisms. Am Heart J 2005; 149:322– 328. Novaro GM, Saliba W, Jaber WA. Images in cardiovascular medicine. Fate of intracardiac lead vegetations after percutaneous lead extraction. Circulation 2002; 106:e46. Rizzello V, Dello Russo A, Casella M, Biddau R. Residual fibrous tissue floating in the right atrium after percutaneous pacemaker lead extraction: An unusual complication early detected by intracardiac echocardiography. Int J Cardiol 2008; 127:e67–e68. Thuny F, Le Dolley Y, Mancini J, Casalta J, Gouriet F, Riberi A, Avienrinos J, et al. Ghost of infected leads: A new critrion of cardiac device-rated infective endocarditis. Poster presented at the 10th International Symposium on Modern Concepts in Endocarditis and Cardiovascular Infections, April 26–28, 2009, Naples, Italy, 2008. Massoure PL, Reuter S, Lafitte S, Laborderie J, Bordachard P, Clementy J, Roudaut R. Pacemaker endocarditis: Clinical features and management of 60 consecutive cases. Pacing Clin Electrophysiol 2007; 30:12–19. Sohail MR. Concerning diagnosis and management of pacemaker endocarditis. Pacing Clin Electrophysiol 2007; 30:829. Jarwe M, Klug D, Beregi JP, Le Franc P, Lacroix D, Kouakam C, Guédon-Moreau L, et al. Single center experience with femoral extraction of permanent endocardial pacing leads. Pacing Clin Electrophysiol 1999; 22:1202–1209. Viola GM, Awan LL, Darouiche RO. Nonstaphylococcal infections of cardiac implantable electronic devices. Circulation 2010; 121:2085–2091. Darouiche RO. Treatment of infections associated with surgical implants. N Engl J Med 2004; 350:1422–1429. Bluhm, G. Pacemaker infections. A clinical study with special reference to prophylactic use of some isoxazolyl penicillins. Acta Med Scand Suppl 1985; 699:1–62. Da Costa A, Lelièvre H, Kirkorian G, Célard M, Chevalier P, Vandenesch F, Etienne J, et al. Role of the preaxillary flora in pacemaker infections: A prospective study. Circulation 1998; 97:1791–1795. Klug D, Wallet F, Lacroix D, Marquié C, Kouakam C, Kacet S, Courcol R. Local symptoms at the site of pacemaker implantation indicate latent systemic infection. Heart 2004; 90:882–886. Golzio PG, Vinci M, Anselmino M, Comoglio C, Rinaldi M, Trevi GP, Bongiorni MG. Accuracy of swabs, tissue specimens, and lead samples in diagnosis of cardiac rhythm management device infections. Pacing Clin Electrophysiol 2009; 32(Suppl 1):S76–S80. Greenspon AJ, Rhim ES, Mark G, Desimone J, Ho RT. Leadassociated endocarditis: The important role of methicillin-resistant Staphylococcus aureus. Pacing Clin Electrophysiol 2008; 31:548– 553. Chamis AL, Peterson GE, Cabell CH, Corey GR, Sorrentino RA, Greenfield RA, Ryan T, et al. Staphylococcus aureus bacteremia in patients with permanent pacemakers or implantable cardioverterdefibrillators. Circulation 2001; 104:1029–1033. 459