Infection reduces survival in cardiovascular implantable electronic device (CIED) recipients. However, the clinical predictors of short- and long-term mortality in patients with CIED infection are not well understood. We retrospectively reviewed all patients with CIED infection who were admitted to Mayo Clinic from January 1991 to December 2008. Survival data were obtained from the medical records and the United Sates Social Security Index. The purported risk factors for short-term (30-day) and long-term (>30-day) mortality were analyzed using univariate and multivariate models. Overall, 415 cases of CIED infection were identified during the study period. The mean follow-up duration for the 243 patients who were alive at the last follow-up visit was 6.9 years. In a multivariate model, heart failure (odds ratio 9.31, 95% confidence interval 2.08 to 41.67), corticosteroid therapy (odds ratio 4.04, 95% confidence interval 1.40 to 11.60), and presentation with CIED-related infective endocarditis (odds ratio 5.60, 95% confidence interval 2.25 to 13.92) were associated with increased short-term mortality. The factors associated with long-term mortality in the multivariate model included patient age (hazard ratio 1.20, 95% confidence interval 1.06 to 1.36), heart failure (hazard ratio 2.01, 95% confidence interval 1.42 to 2.86), metastatic malignancy (hazard ratio 5.99, 95% confidence interval 1.67 to 21.53), corticosteroid therapy (hazard ratio 1.97, 95% confidence interval 1.22 to 3.18), renal failure (hazard ratio 1.94, 95% confidence interval 1.37 to 2.74), and CIED-related infective endocarditis (hazard ratio 1.68, 95% confidence interval 1.17 to 2.41). In conclusion, these data suggest that the development of CIED-related infective endocarditis and the presence of co-morbid conditions are associated with increased short- and long-term mortality in patients with CIED infection.
Infection is a serious complication of cardiovascular implantable electronic device (CIED) implantation, requiring prompt, complete device removal and systemic antibiotic therapy. Moreover, CIED infection is associated with a significant increase in short- and long-term mortality and financial cost. However, our understanding of the factors associated with poor outcomes in patients with infected devices remains limited. In the present investigation, we identified the clinical predictors associated with increased short- and long-term mortality among patients with infection involving CIEDs.
Methods
We retrospectively reviewed all patients with CIED infection who were admitted to Mayo Clinic (Rochester, Minnesota) from January 1, 1991 to December 31, 2008. The cases of CIED infection were identified from the Mayo Clinic Heart Rhythm Device Database, the surgical index, and the computerized central diagnostic index. The Mayo Clinic institutional review board reviewed and approved the study proposal.
CIED infection was defined using criteria previously described by our group. Cases of CIED infection were further classified as pocket infection or endovascular infection (bloodstream infection or CIED-related infective endocarditis [CIED-IE]). The diagnosis of CIED-IE was determined from echocardiographic findings of an oscillating intracardiac mass on a heart valve or device lead, visualization of a cardiac abscess, or new dehiscence of a prosthetic valve.
The follow-up duration was calculated from the date of hospital admission to the date of death or the last follow-up visit. In addition to extracting the follow-up data from the available medical records, updated vital status and death dates (if available) were obtained from the United States Social Security Index on December 1, 2010. Because the Social Security Index data can have a lag time of ≤6 months, patients who were still alive according to this Index were assumed to be alive through June 1, 2010.
Short-term mortality was defined as death within 30 days of admission to the hospital. The associations of clinical features with short-term mortality were evaluated using Wilcoxon rank sum, chi-square, and Fisher’s exact tests. These associations were further evaluated using univariate and multivariate logistic regression models and summarized with odds ratios and 95% confidence intervals. A multivariate model was developed using a stepwise selection process with the p value for a feature to enter or leave the model set at 0.05. Model discrimination (i.e., how well the features in the model separated patients who died within 30 days of admission from the patients still alive at 30 days after admission) was summarized using the area under a receiver operating characteristic curve (AUC). The AUC can range from 0.5 to 1.0, with higher values indicating improved predictive ability. Model calibration (i.e., how well the predicted probabilities estimated by the model agreed with the observed short-term mortality) was summarized using the Hosmer and Lemeshow goodness-of-fit test. A statistically significant p value from this test would reject the null hypothesis that the features in the model fit the data well.
Overall survival was estimated using the Kaplan-Meier method. The follow-up duration was calculated from the date of admission to the date or death or the last follow-up visit. Associations of the features with the interval to death were evaluated using univariate and multivariate Cox proportional hazards regression models and summarized with hazard ratios and 95% confidence intervals. A multivariate model was developed using stepwise selection with the p value for a feature to enter or leave the model set at 0.05. Model discrimination (i.e., how well the features in the model separated patients who died from those who were censored at the last follow-up visit) was summarized using a concordance index (c index). The c index corresponds to the proportion of all usable pairs of patients in whom the observed and predicted survival times were concordant. Similar to the AUC, the c index can range from 0.5 to 1.0, with higher values indicating improved predictive ability.
Statistical analyses were performed using the SAS software package (SAS Institute, Cary, North Carolina). All statistical tests were 2-sided, and p <0.05 was considered statistically significant.
Results
Overall, 415 patients with CIED infection were identified from January 1, 1991 to December 31, 2008. The clinical patient characteristics are summarized in Table 1 . Of the 415 patients, 1 patient who was alive at hospital discharge was excluded from the analysis of short-term mortality because the 30-day follow-up data were not available. Of the remaining 414 patients, 23 (5.6%) died within 30 days after admission and 391 (94.4%) were alive at 30 days after admission. The univariate associations of the candidate predictors with short- and long-term mortality are summarized in Table 1 .
| Short-term Mortality ∗ | Long-term Mortality † | ||||
|---|---|---|---|---|---|
| No (n = 391) | Yes (n = 23) | p Value | HR (95% CI) | p Value | |
| Age at admission (yrs) | 0.83 | 1.24 (1.11–1.39; 10-y increments) | <0.001 | ||
| Median | 71 | 74 | |||
| Range | 17–95 | 25–91 | |||
| Gender | 0.55 | ||||
| Female | 97 (25%) | 7 (30%) | 1.0 (reference) | 0.76 | |
| Male | 294 (75%) | 16 (70%) | 1.06 (0.74–1.50) | ||
| Race and ethnicity | 0.05 | 0.53 | |||
| White | 357 (91%) | 18 (78%) | 1.0 (reference) | ||
| Other | 34 (9%) | 5 (22%) | 0.83 (0.46–1.49) | ||
| Year of device placement | 0.77 | 1.05 (0.90–1.23; 5-y increments) | 0.51 | ||
| Median | 2001 | 2001 | |||
| Range | 1975–2008 | 1988–2008 | |||
| Age of device (yrs) | 0.046 | 1.03 (0.99–1.08; 1-y increments) | 0.14 | ||
| Median | 1 | 3 | |||
| Range | 0–24 | 0–15 | |||
| Device type | 0.10 | ||||
| Permanent pacemaker | 241 (62%) | 14 (61%) | 0.94 | 1.0 (reference) | |
| Implantable cardioverter defibrillator | 150 (38%) | 9 (39%) | 1.30 (0.95–1.79) | ||
| Procedure (n = 393) | |||||
| De novo implant | 162 (43%) | 13 (65%) | 0.38 | 2.33 (1.59–3.41) | <0.001 |
| Generator replacement | 95 (25%) | 2 (10%) | — | ||
| System revision/upgrade | 83 (22%) | 4 (20%) | 1.62 (1.03–2.56) | −0.038 | |
| Lead revision/insertion | 30 (8%) | 1 (5%) | — | ||
| Other | 3 (1%) | 0 | 1.0 (reference) | ||
| Generator site (n = 376) | 0.89 | ||||
| Pectoral | 329 (92%) | 18 (95%) | 1.0 | 1.0 (reference) | |
| Abdominal | 28 (8%) | 1 (5%) | 1.04 (0.59–1.83) | ||
| Chambers | 0.87 | ||||
| Single | 82 (21%) | 6 (26%) | 0.60 | 1.0 (reference) | |
| Dual | 309 (79%) | 17 (74%) | 0.97 (0.68–1.40) | ||
| Previous procedures (n) | 2 (1–8) | 1 (1–5) | 0.06 | 0.88 (0.77–1.01; 1-U increase) | 0.06 |
| Leads in place (n) | 2 (1–8) | 2 (1–4) | 0.17 | 0.94 (0.79–1.11; 1-U increase) | 0.46 |
| Coronary artery disease | 220 (56%) | 16 (70%) | 0.21 | 1.61 (1.18–2.21) | 0.003 |
| Coronary artery bypass grafting | 111 (28%) | 6 (26%) | 0.81 | 1.07 (0.77–1.49) | 0.68 |
| Heart failure | 211 (54%) | 21 (91%) | <0.001 | 2.34 (1.70–3.23) | <0.001 |
| Ejection fraction (%; n = 375) | 45 (10–75) | 37.5 (15–70) | 0.12 | 0.79 (0.72–0.87) (10% increments) | <0.001 |
| Atrial fibrillation | 138 (35%) | 11 (48%) | 0.22 | 1.40 (1.04–1.90) | 0.029 |
| Antiplatelet therapy (n = 396) | 175 (47%) | 8 (40%) | 0.57 | 0.98 (0.72–1.33) | 0.88 |
| Anticoagulation (warfarin or heparin) | 154 (39%) | 9 (39%) | 0.98 | 1.12 (0.82–1.51) | 0.50 |
| Statin therapy (n = 394) | 116 (31%) | 6 (32%) | 0.95 | 1.16 (0.82–1.63) | 0.41 |
| Diabetes mellitus | 112 (29%) | 3 (13%) | 0.10 | 1.50 (1.09–2.08) | 0.014 |
| Renal disease | 71 (18%) | 6 (26%) | 0.40 | 2.92 (2.05–4.15) | <0.001 |
| Hemodialysis | 24 (6%) | 2 (9%) | 0.65 | 3.52 (2.13–5.80) | <0.001 |
| Chronic obstructive pulmonary disease | 59 (15%) | 8 (35%) | 0.020 | 1.60 (1.11–2.30) | 0.012 |
| Prosthetic heart valve | 56 (14%) | 5 (22%) | 0.36 | 1.44 (0.97–2.14) | 0.07 |
| Peripheral vascular disease | 40 (10%) | 3 (13%) | 0.72 | 1.64 (1.06–2.55) | 0.028 |
| Cerebrovascular disease | 41 (10%) | 4 (17%) | 0.30 | 1.05 (0.65–1.72) | 0.84 |
| Chronic skin conditions | 33 (8%) | 1 (4%) | 0.71 | 1.74 (1.07–2.80) | 0.024 |
| Implanted central venous catheter | 26 (7%) | 1 (4%) | 1.0 | 1.96 (1.13–3.41) | 0.017 |
| Active malignancy | 46 (12%) | 4 (17%) | 0.50 | 1.79 (1.21–2.66) | 0.004 |
| Metastatic malignancy | 2 (1%) | 2 (9%) | <0.001 | 7.09 (2.24–22.48) | <0.001 |
| Autoimmune disease | 26 (7%) | 1 (4%) | 1.0 | 0.66 (0.31–1.40) | 0.27 |
| Corticosteroid therapy | 31 (8%) | 7 (30%) | 0.003 | 2.62 (1.68–4.09) | <0.001 |
| Body mass index, (kg/m 2 ) | 0.043 | 0.92 (0.80–1.05; 5-U increase) | 0.22 | ||
| Median | 27.7 | 25.0 | |||
| Range | 17.0–63.7 | 19.5–43.3 | |||
| Body mass index (kg/m 2 ) | 0.13 | ||||
| <30 | 275 (70%) | 18 (78%) | 0.42 | 1.0 (reference) | |
| ≥30 | 116 (30%) | 5 (22%) | 0.77 (0.54–1.08) | ||
| Charlson co-morbidity score | |||||
| ≥1 | 319 (82%) | 21 (91%) | 0.40 | 3.46 (2.00–6.00) | <0.001 |
| ≥2 | 211 (54%) | 17 (74%) | 0.06 | 2.31 (1.68–3.19) | <0.001 |
| ≥3 | 125 (32%) | 12 (52%) | 0.045 | 2.36 (1.74–3.22) | <0.001 |
| Initial presentation with device infection | <0.001 | ||||
| Mayo | 150 (38%) | 10 (43%) | 0.62 | 1.0 (reference) | |
| Outside Mayo | 241 (62%) | 13 (57%) | 0.60 (0.45–0.81) | ||
| Symptom onset (d; n = 413) | 7 (0–1750) | 3 (0–30) | 0.008 | 0.98 (0.93–1.03) (30-d increments) | 0.42 |
| Infection classification | |||||
| Limited to generator pocket | 210 (54%) | 4 (17%) | <0.001 | 0.45 (0.33–0.61) | <0.001 |
| Infective endocarditis related | 75 (19%) | 14 (61%) | <0.001 | 1.72 (1.22–2.41) | 0.002 |
| Metastatic focus | 17 (4%) | 3 (13%) | 0.09 | 2.15 (1.22–3.79) | 0.008 |
| Leukocytosis (n = 413) | 148 (38%) | 14 (61%) | 0.029 | 1.61 (1.19–2.16) | 0.002 |
| Anemia (n = 413) | 208 (53%) | 12 (52%) | 0.91 | 1.81 (1.33–2.47) | <0.001 |
| Elevated serum creatinine | 135 (35%) | 16 (70%) | <0.001 | 2.53 (1.87–3.44) | <0.001 |
| Elevated sedimentation rate (n = 152) | 89 (61%) | 3 (60%) | 1.0 | 1.62 (0.97–2.72) | 0.07 |
| Elevated C-reactive protein (n = 34) | 14 (44%) | 2 (100%) | 0.21 | 2.04 (0.66–6.34) | 0.22 |
| Positive pocket culture (n = 369) | 265 (75%) | 10 (67%) | 0.55 | 0.73 (0.50–1.06) | 0.10 |
| Positive lead culture (n = 277) | 178 (67%) | 8 (67%) | 1.0 | 0.86 (0.56–1.32) | 0.49 |
| Positive blood culture | 175 (45%) | 17 (74%) | 0.006 | 2.10 (1.55–2.85) | <0.001 |
| Microorganisms | |||||
| Staphylococci (coagulase-negative Staphylococcus , methicillin-sensitive or methicillin-resistant Staphylococcus aureus ) | 269 (69%) | 18 (78%) | 0.34 | 1.14 (0.81–1.59) | |
| Other | 122 (31%) | 5 (22%) | 1.0 (reference) | 0.45 | |
| Staphylococcus aureus (methicillin-sensitive or methicillin-resistant) | 116 (30%) | 13 (57%) | 0.007 | 1.85 (1.36–2.51) | <0.001 |
| Other | 275 (70%) | 10 (43%) | 1.0 (reference) | ||
| Hardware removed | 373 (95%) | 19 (83%) | 0.027 | — | — |
| Subset of cases with hardware removed (n = 392) | |||||
| Timing of hardware removal | |||||
| At initial presentation | 349 (94%) | 16 (84%) | 0.13 | 1.0 (reference) | |
| After failure of conservative treatment | 24 (6%) | 3 (16%) | 1.56 (0.87–2.82) | 0.14 | |
| Lead extraction procedure (n = 391) | |||||
| Percutaneous | 346 (93%) | 17 (89%) | 0.43 | 1.0 (reference) | |
| Surgical | 24 (6%) | 2 (11%) | 1.30 (0.77–2.19) | 0.32 | |
| Other or not removed | 2 (1%) | 0 | |||
| Complications of extraction | 37 (11%) | 4 (24%) | 0.11 | 0.76 (0.44–1.32) | 0.33 |
| Successful extraction | 323 (93%) | 13 (76%) | 0.030 | 1.20 (0.64–2.24) | 0.57 |
∗ Associations of clinical features with short-term mortality were evaluated using Wilcoxon rank sum, chi-square, and Fisher’s exact tests.
† Associations of clinical features with interval to death were evaluated using Cox proportional hazards regression models and summarized with hazard ratios and 95% confidence intervals.
Patients who had CIEDs implanted for a longer duration and those who presented with a more acute onset of symptoms had greater short-term mortality. Several co-morbid conditions, including chronic obstructive pulmonary disease, malignancy, chronic corticosteroid therapy, and a lower body mass index, were also associated with greater short-term mortality. Patients with greater short-term mortality were more likely to present with leukocytosis, worsening renal function, positive blood culture findings, Staphylococcus aureus infection, and echocardiographic findings indicative of CIED-IE ( Table 1 ).
The results of a multivariate model developed using the candidate predictors from the univariate analysis are summarized in Table 2 . Heart failure, chronic corticosteroid therapy, and CIED-IE were all significantly associated with increased short-term mortality in this multivariate model. Patients with heart failure, for instance, were >9 times more likely to die within 30 days of admission than were patients without heart failure. Similarly, patients receiving chronic corticosteroid therapy had a 4-fold and those with CIED-IE a 5.6-fold increase in short-term mortality. The AUC for these features in this model was 0.83 (95% confidence interval 0.75 to 0.91). The p value from the Hosmer and Lemeshow goodness-of-fit test was 0.76, indicating that the features in the model fit the data well and that the model was well calibrated.
