The aim of this study was to assess the impact of an operative protocol with a multidisciplinary approach on the outcome of patients with prosthetic valve endocarditis (PVE). A formal policy for the care of PVE was introduced at our hospital in 2003 in which patients were referred to and managed by a preexisting team involving a cardiologist, a specialist in infectious diseases, and a cardiac surgeon. All patients underwent transesophageal echocardiography as soon as clinical suspicion of PVE arose. If high-risk conditions such as heart failure, ring abscess, conditions associated with impending malfunctioning of the prosthesis, or vegetations at high risk for systemic embolization were found during the initial multidisciplinary evaluation (performed within 12 hours of admission), patients were operated on within 48 hours. Stable patients were evaluated weekly by the multidisciplinary team, and on-treatment surgery was performed whenever high-risk conditions developed or when there was persistent fever/bacteremia after 1 week of adequate antibiotic therapy. Comparing the period 2003 through 2009 with 1996 through 2002 (when a multidisciplinary policy was not followed), patients with PVE were more numerous (61 vs 38), older (mean age 68.3 vs 63.1, p = 0.01), and had more co-morbidities (mean Charlson index 3.15 vs 2.42, p = 0.03). The most frequent causative organisms were Staphylococci in both periods. In the second period, fewer patients had delayed diagnosis (39% vs 71%, p = 0.03), heart failure (20% vs 45%, p = 0.01), abscess (20% vs 39%, p = 0.04), culture-negative infective endocarditis (11% vs 29%, p = 0.03), and worsened renal function (21% vs 42%, p = 0.04). A significant reduction in in-hospital mortality (53% to 23%, p = 0.04) and 3-year mortality (60% to 28%, p = 0.001) was observed, driven by the increased number of patients successfully treated with medical therapy alone (44% vs 16%, p = 0.04). In conclusion, formalized, collaborative management led to significant improvement in PVE-related mortality.
Prosthetic valve endocarditis (PVE) has been estimated to occur with a relatively low but increasing incidence ranging from 0.1% to 2.3% per patient-year and to account for 10% to 30% of all cases of infective endocarditis (IE). Despite advances in diagnostic imaging and medical and surgical care interventions, PVE continues to be an extremely serious and potentially lethal complication of heart valve surgery, associated with a risk of morbidity and mortality ranging from 20% to 80%. Recommendations for management of patients with for PVE based on prospective randomized studies are still lacking, and therapeutic strategies are controversial. Even within the same hospital, patients may be managed in different ways, and deviations from the published guidelines, which have been demonstrated to have a negative impact on the patient outcomes, are common. To improve the outcome of patients with PVE, we implemented a new strategy based on the development of local diagnostic and therapeutic guidelines modeled on the 1998 American Heart Association guidelines and a local consensus on the management of IE formed among cardiologists, infectious disease specialists, microbiologists, and cardiac surgeons. The primary aim of this policy was to determine whether patients would benefit from receiving the same treatment regardless of the initial department of admission or the physician in charge. This study evaluated the impact of the formalized multidisciplinary management strategy on the outcome of patients with PVE.
Methods
Ca’ Foncello Hospital, Treviso, is a tertiary care hospital in northeastern Italy serving about 650,000 inhabitants. A formal policy for the care of IE was introduced in 2003 in which patients were referred to and managed by a preexisting team involving a cardiologist, a specialist in infectious diseases, a microbiologist, and a cardiac surgeon as soon as the diagnosis was considered. From 1996 to 2009, 418 consecutive patients with definite IE according to modified Duke criteria were identified at our institution. Clinical, microbiological, and imaging data were collected prospectively in a computerized database. The diagnosis of PVE was established when 2 main Duke criteria or 1 main and 2 minor criteria were fulfilled. Early PVE was defined when occurring within 2 months of operation.
Definition of the parameters and of the managing protocol has been illustrated in detail elsewhere. In brief, all patients with suspected PVE (defined as the presence of fever of unknown origin lasting ≥5 days and/or positive blood cultures, and/or embolic phenomena) underwent transesophageal echocardiography (TEE) as soon as possible. After PVE was diagnosed and a TEE study was available, the initial multidisciplinary evaluation was performed within 12 hours of admission/diagnosis by a cardiologist, a cardiac surgeon, and a specialist in infectious diseases. This collaborative initial evaluation was mostly aimed at assessing the presence of indications for urgent surgery—namely, conditions associated with impending hemodynamic impairment (refractory III or IV New York Heart Association class heart failure, rocking prosthesis, risk of complete block of mechanical prosthesis, severe left-sided regurgitation of the valve prosthesis due to either ring dehiscence or lacerated leaflets, aorto-cavitary fistula), and conditions in which antibiotic therapy had little chance of curing the infection or proved ineffective (i.e., fungal PVE, ring abscess or false aneurysm, persistent bacteremia, or persistent fever with leukocytosis after 1 week of adequate antibiotic therapy). Another indication for early surgery was the prevention of systemic embolic events in the presence of left-sided, mobile, >10-mm vegetations. In all these cases, patients underwent urgent surgery (i.e., within 48 hours). If the patients developed conditions at high risk or had persistent bacteremia/fever after 1 week of appropriate antibiotic therapy (defined as failed antibiotic therapy), on-treatment surgery was indicated. When none of these conditions was present or patients, although defined at high risk, were deemed unfit for surgery (in case of severe co-morbidities, active malignancy, age >90 years, logistic EuroSCORE >30, major neurologic deficit/coma due to ischemic stroke, hemorrhagic stroke), medical therapy was indicated. The patient’s clinical course was evaluated daily. C-reactive protein, blood cultures, and plasmatic concentrations of vancomycin, teicoplanin, and gentamicin were assessed every 3 days. Patients underwent repeat TEE and were evaluated by the multidisciplinary team every week.
The study was divided into 2 consecutive periods of 7 years: period 1 before implementation of the standardized protocol (1996 through 2002), and period 2 after the implementation of our protocol (2003 through 2009). The investigation protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki and was approved by the Ethical Committee of the Hospital.
Continuous variables were compared with Student’s t test and Mann-Whitney U test for nonnormal variables, and categorical values were analyzed with chi-square test or Fisher’s exact text, as appropriate. The comparative distribution of clinical, microbiological, and echocardiographic variables and of clinical events was assessed in the 2 periods in a bivariate fashion. The impact of the management protocol on in-hospital mortality was assessed using the Cox proportional hazard model. All tests were 2-sided, and differences were considered statistically significant at p <0.05. Statistical analysis was carried out using also SPSS software V 12.0 (SPSS Inc., Chicago, Illinois).
Results
During the entire study period, 99 patients had definite PVE (24%), 297 patients had native valve endocarditis, and 22 patients had infection of the pacemaker lead. Thirty-eight cases of PVE were included during period 1 (1996 through 2002), and 61 cases of PVE were included in period 2 (2003 through 2009), following the implementation of the management protocol. Comparing period 2 with period 1 ( Table 1 ), the number of patients increased, with more patients referred from community hospitals; patients were older with more co-morbidities (mean Charlson co-morbidity index 2.42 vs 3.15, p = 0.03). The time interval between the onset of symptoms and definite diagnosis of PVE was shorter in period 2. All patients in period 2 underwent ≥1 TEE studies with the first examination performed at a mean 7 ± 3 days interval from the onset of symptoms, whereas this interval was significantly longer in period 1 (27 ± 14 days). In period 1, 31 of 38 underwent ≥1 TEE study. The most common infecting organisms were Staphylococcus aureus and coagulase-negative Staphylococci in both periods. Culture-negative endocarditis was significantly more frequent in period 1. As for the pattern of valve infection PVE involved, there were more mechanical valves in period 1 and more bioprostheses in period 2. No cases of PVE involved the right heart valves, and the distribution between aortic and mitral PVE was similar in the 2 periods. Ring abscess were identified more frequently during period 1. Consultation rate with infectious diseases specialists was higher in period 2 (100% vs 79%, p = 0.0003). The adherence to guidelines on antibiotic therapy was more frequent in period 2 (97% vs 74%, p = 0.001), with longer duration of antibiotic treatment observed in period 2 (53 ± 9 days vs 35 ± 14 days, p = 0.001). The use of rifampin and aminoglycosides did not differ between periods.
| Variable | Period 1 (1996–2002), n = 38 | Period 2 (2003–2009), n = 61 | p Value |
|---|---|---|---|
| Mean age (yrs) | 63.1 ± 12.2 | 68.3 ± 15.7 | 0.01 |
| Patients aged >70 yrs | 20 (53%) | 45 (74%) | 0.04 |
| Male patients | 23 (60%) | 39 (64%) | 0.83 |
| Early PVE | 7 (18%) | 5 (8%) | 0.20 |
| Type of prosthetic valve | |||
| Mechanical | 20 (52%) | 19 (30%) | 0.03 |
| Bioprosthesis | 16 (42%) | 39 (64%) | 0.04 |
| Aortic homograft | 1 (3%) | 1 (2%) | 1 |
| Mitral reconstruction | 1 (3%) | 2 (4%) | 1 |
| Referred from other hospitals | 8 (21%) | 31 (51%) | 0.03 |
| Diagnosis >30 days | 27 (71%) | 24 (39%) | 0.03 |
| Diabetes mellitus | 7 (18%) | 21 (34%) | 0.11 |
| Chronic renal failure | 4 (10%) | 10 (16%) | 0.56 |
| Coronary artery disease | 5 (13) | 9 (15) | 1 |
| Neoplasm | 2 (5%) | 6 (10%) | 0.70 |
| Charlson co-morbidity index | 2.42 ± 1.61 | 3.15 ± 1.85 | 0.03 |
| Echocardiography | |||
| No. of studies performed | 2.1 ± 1.1 | 3.5 ± 0.9 | 0.004 |
| Interval onset of symptoms—TEE (days) | 27 ± 14 | 7 ± 3 | <0.001 |
| Localization of infection | |||
| Aortic | 22 (58%) | 37 (61%) | 0.83 |
| Mitral | 14 (37%) | 21 (34%) | 0.83 |
| Aortic and mitral | 2 (5%) | 3 (5%) | 1 |
| Initial regurgitation 2+/4+ | 20 (52%) | 26 (43%) | 0.41 |
| High-risk vegetations | 10 (26%) | 12 (20%) | 0.46 |
| Abscess | 15 (39%) | 12 (20%) | 0.04 |
| Left ventricular ejection fraction <0.45 | 14 (37%) | 32 (52%) | 0.15 |
| Causative agents | |||
| Oral Streptococci | 2 (10%) | 3 (5%) | 1 |
| Group D Streptococci | 2 (10%) | 7 (11%) | 0.47 |
| Staphylococcus aureus | 11 (34%) | 24 (39%) | 0.39 |
| Coagulase-negative Staphylococci | 6 (24%) | 14 (23%) | 0.45 |
| Enterococci | 2 (10%) | 3 (5%) | 1 |
| Culture-negative IE | 11 (29%) | 7 (11%) | 0.03 |
In period 2, fewer patients had worsened renal function (21% vs 42%, p = 0.04) and heart failure (20% vs 45%; Table 2 ). Medical therapy succeeded in eradicating the infection in more patients during period 2 (44% vs 16%, p = 0.04). Timing of surgery was significantly more anticipated in period 2 with more patients undergoing urgent surgery. Indications for urgent surgery in period 2 were as follows: prosthesis dehiscence with heart failure (n = 5), abscess (n = 3), persistent infection (n = 3), prevention of embolic risk (n = 2), and combined indications (5). Overall surgical mortality was not different, whereas overall in-hospital mortality declined from 53% to 23% (p = 0.04). No differences in mortality rate were observed between mechanical valves and bioprostheses or between prostheses in the mitral and aortic position.
| Variable | Period 1 (1996–2002), n = 38 | Period 2 (2003–2009), n = 61 | p Value |
|---|---|---|---|
| Heart failure | 17 (45%) | 12 (20%) | 0.01 |
| Embolic phenomena | 11 (29%) | 14 (23%) | 0.63 |
| Metastatic infections | 6 (16%) | 9 (15%) | 1 |
| Multiorgan failure | 7 (18%) | 8 (13%) | 0.56 |
| Worsened renal function | 16 (42%) | 13 (21%) | 0.04 |
| Medical therapy | 26 (68%) | 41 (67%) | 1 |
| Failed antibiotic therapy | 20 (52%) | 14 (23%) | 0.04 |
| Successful medical therapy | 6 (16%) | 27 (44%) | 0.04 |
| Surgery during hospitalization | 12 (32%) | 20 (33%) | 1 |
| Urgent surgery | 0 | 18 (29%) | 0.0001 |
| On-treatment surgery | 12 (31%) | 2 (3%) | 0.0001 |
| Type of surgery | |||
| Aortic homograft | 3 (25%) | 7 (35%) | 0.70 |
| Mechanical prosthetic valve | 2 (17%) | 0 | 0.13 |
| Biological prosthetic valve | 7 (58%) | 13 (65%) | 0.72 |
| Surgical mortality | 5 (42%) | 6 (30%) | 0.70 |
| Early postsurgical recurrence | 2 | 1 | 0.54 |
| Overall in-hospital mortality | 20 (53%) | 14 (23%) | 0.04 |
| 3-yr mortality | 23 (60%) | 17 (28%) | 0.001 |
Four patients (2 in period 1 and 2 in period 2) were lost at follow-up. Mortality occurring after hospital discharge did not significantly differ between the 2 periods ( Figure 1 ). In the multivariate analysis advanced age, heart failure, failed antibiotic therapy, the onset of multiorgan failure, and the presence of abscess at echocardiography resulted independent predictors of 3-year mortality ( Table 3 ). After adjustment for relevant prognostic factors (gender, heart failure, delayed diagnosis, diabetes mellitus, failed antibiotic therapy, oral Streptococci , Staphylococcus aureus ) and patient characteristics that had significantly changed between the 2 periods, the treatment during period 2 remained independently predictive of 3-year survival (odds ratio: 0.55, 95% confidence interval: 0.22 to 0.98, p = 0.04). The beneficial effect remained when calendar year was added to the Cox model.
