| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 2007;83:2074-2080
© 2007 The Society of Thoracic Surgeons
a Division of Cardiothoracic Surgery, Department of Surgery, Oregon Health and Sciences University, Portland, Oregon
b Division of Cardiothoracic Surgery, Department of Surgery, Portland Veterans Affairs Medical Center, Portland, Oregon
Accepted for publication January 23, 2007.
* Address correspondence to Dr Slater, Oregon Health and Science University, Division of Cardiothoracic Surgery, 3181 SW Sam Jackson Park Rd (L353), Portland, OR 97239-3098 (Email: slaterm{at}ohsu.edu).
Presented at the Fifty-third Annual Meeting of the Southern Thoracic Surgical Association, Tucson, AZ, Nov 8–11, 2006.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterial and MethodsResultsCommentDiscussionReferences |
|---|
Methods: The goal of this retrospective study was to identify independent variables for early and late mortality in 364 consecutive patients with endocarditis over a 10-year period.
Results: The mean age of patients was 48.2 years, 35% had a history intravenous drug use, 19.8% were reoperative, and 93% had native valve endocarditis. Fever (68%) and fatigue (36%) were the most common presenting symptoms, and congestive heart failure (52%), embolization (45%), and uncontrolled sepsis (36%) were the most common indications for surgery. Overall survival at discharge, 1, 5, and 10 years was 87%, 76%, 55%, and 31%, respectively. Survival at discharge, 5, and 10 years was 91%, 69%, and 41% for surgical patients and 85%, 60%, and 31% for medically treated patients, respectively. Surgery was associated with improved short-term and long-term survival (p < 0.0.01). Independent predictors of early death were hemodynamic instability (p = 0.013) and age older than 55 years (p < 0.025). Medical treatment (p = 0.005), age older than 55 years (p = 0.032), institution (p < 0.001), New York Heart Association functional class III or IV (p = 0.002), and hemodynamic instability (p = 0.044) were predictive of late death.
Conclusions: Short-term and long-term mortality from endocarditis remains high, although surgically treated patients had improved survival. Differing outcomes from two geographically similar institutions highlight the limitations of extrapolating risk factors between disparate patient populations.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionReferences |
|---|
The incidence of endocarditis in the United States is approximately 1.7 to 6.2 cases per 100,000 person years [1]. The American Heart Association issued a consensus statement on the diagnosis of infectious endocarditis in 1998 and estimated 15,000 to 20,000 cases of endocarditis per year in the United States [2].
Patients at increased risk for the development of endocarditis include those with congenital or acquired structural valvular abnormalities. Patients actively abusing intravenous drugs (IVDA) are also at risk for the development of endocarditis, presumably from previously unrecognized episodes of endocarditis that damage the valve and leave it prone to subsequent infection. Patients who have recently undergone implantation of prosthetic cardiac material, especially prosthetic valves, are at increased risk for the development of endocarditis in both the immediate and, to a lesser degree, later postoperative periods. Patients with permanently implanted transvenous pacing or defibrillation systems and those with long-term central venous access are at increased risk for endocarditis as well.
Although endocarditis comprises only 0.2% of all deaths caused by cardiovascular disease, the short-term and long-term mortality from endocarditis remains high. Short-term (discharge or 30-day) mortality rates of 9% to 38% have been reported, and extremely limited long-term survival rates, some as low as 12%, have also been reported [1, 3–7].
Cabel and colleagues [4] monitored 138 patients with endocarditis and reported 80% hospital survival and 70% survival at 10 years. Prosthetic valve endocarditis has been associated with increased morbidity and mortality compared with native valve endocarditis. Romano and colleagues [8] reported respective survival at 1, 5, 10, and 15 years of 97.5%, 91.4%, 80.5%, and 49.4% for patients after native valve endocarditis and 90.8%, 84.9%, 59.4%, and 43.4% after prosthetic valve endocarditis.
Most patients with endocarditis are successfully treated with antibiotics. Surgical treatment has been shown to be effective and in some cases is superior to medical therapy in specific subgroups of patients. Patients thought to derive the most benefit from surgical intervention are those with large vegetations, embolization, congestive heart failure (CHF), and those with associated perivalvar abscesses. In some cases, extensive débridement of all involved cardiac structure requires extensive cardiac reconstruction.
Recent studies have demonstrated improved survival in patients who undergo early operative intervention for the treatment of endocarditis [4, 7]. No consensus exists on which patient groups receive the most benefit from surgical intervention. In addition, the indications for and timing of surgery to reduce the incidence of embolus in patients with vegetations is not clearly defined [9].
Portland, Oregon, has a high incidence of IVDA and a correspondingly high incidence of endocarditis. We analyzed our experience with endocarditis during a 10-year period at both a tertiary care university hospital (UH) and a Veterans Affairs Medical Center (VAMC) to better understand the risk factors associated with short-term and long-term mortality.
|
Material and Methods |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionReferences |
|---|
Overall Study Population
We evaluated 364 patients (106 at the VAMC and 258 at the UH) of 418 admissions, of which 54 (12.9%) were readmissions. The mean age of patients was 48.2 years, 66% were men, and 14.4% of patients were at a mean New York Heart Association (NYHA) functional class of 1.6 on presentation and were considered to have CHF. A total of 35% had a history of IVDA, and 19.8% had undergone previous cardiac surgery. Fever (68%) and fatigue (36%) were the most common presenting symptoms. Embolic complications were eventually diagnosed in 45% of patients, and embolic findings were the presenting symptom in 18%.
Follow-Up
Patient follow-up was through the medical record. Patients were censored at the last clinical visit in the medical record. Follow-up averaged 712 days and was similar in the patients treated medically and surgically (728 versus 753 days). Follow-up was longer at the VAMC compared with the UH (866 versus 640 days, p = 0.022).
Statistical Analysis
Data analysis was done with SPSS 14.1 software (SPSS Inc, Chicago, IL). Univariate analysis was performed using the Student t test, and analysis of variance was used to assess factors associated with in-hospital and late death. Because Kaplan-Meier survival analysis delineated the institution as a significant variable in survival for both all-cause (p < 0.001) and endocarditis-related mortality (p = 0.048), subsequent univariate and multivariate analyses were performed separately for each institution.
Factors analyzed included age older than 55 years, male gender, history of CHF, NYHA functional class, history of IVDA, history of previous cardiac surgery, history of admission renal failure (serum creatinine 
1.4 mg/dL), hypertension, diabetes mellitus, vegetations on echocardiography, embolic phenomena, prosthetic valve, valve affected, organism involved, and whether surgical intervention was performed. The age cutoff of 55 was selected based on a receiver-operating curve that suggested that a cutoff of 55 maximized specificity and sensitivity (0.70 for both).
Factors identified on univariate analysis to be associated with mortality (p < 0.10) were used to construct a binomial logistic regression model to determine independent predictors for in-hospital and late death (p < 0.05). Survival analysis was performed using standard Kaplan-Meier analysis, and comparisons were determined using Mantel-Cox tests.
|
Results |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionReferences |
|---|
|
|
Microbiology
Overall, 86% had positive blood cultures, of which 74% were gram-positive cocci. Staphylococcus spp predominated at 49%. Of these, 22% were methicillin-resistant S aureus and 65% were methicillin-sensitive S aureus. S epidermidis comprised 12% of all staphylococci infections. Streptococcus spp represented 14% of positive cultures and Enterococcus spp, 7%.
Haemophilus aphrophilus, H paraphrophilus, and H parainfluenzae; Actinobacillus actinomycetemcomitans, Cardiobacterium hominis, Eikenella corrodens, and Kingella kingae (HACEK organisms) were found in 1.8% of patients, fungal organisms in 0.6%, and the organism was not clearly isolated or identified in 2%. Multiple organisms were present in 6% of patients.
Operations were performed in 22% of patients. The leading indications were CHF (52%), embolization (45%), and uncontrolled sepsis (36%). Patients undergoing medical and surgical treatment had similar embolic findings (43.6% versus 42.4%, p = 0.47) and rates of diabetes (13.5% versus 14.1%, p = 0.5) and hypertension (21.2% versus 20.7%; p = 0.52; (Table 2). Patients treated medically were older (49.2 versus 44.7 years; p = 0.043), were at a lower NYHA class (1.6 versus 1.9, p < 0.001), and were more frequently women (38.3% versus 22.5%, p = 0.004). Follow-up duration was similar in patients treated medically and surgically (728 versus 753 days; p = 0.8)
|
Readmission
The rate of endocarditis-related readmission was 14.8% and represented either recurrence of the index infection, infection of the repaired or replaced valve with a new organism, or infection of a different valve. Because of the relatively small numbers of patients within each of these subgroups, meaningful analysis was not possible. Readmissions rates for patients treated medically (16%) and surgically (14%) were nearly identical (p = 0.7), as were the rates at the VAMC (15 %) and UH (16 %).
Embolic Complications
Embolic complications were diagnosed in 43% of our study population. In a significant proportion of patients (17.5%), left-sided emboli (peripheral arterial emboli, 11.5%; stroke, 6%) were the presenting symptom. Emboli were cited as the primary or contributing indication for surgical intervention in 45% of patients.
Survival
Overall survival at discharge, 1, 5, and 10 years was 87%, 76%, 55%, and 31%, respectively (Fig 2). Overall survival was after medical and surgical treatment was 65.0% and 82.6% (p = 0.001). Improved survival was observed after surgery at all time points, including discharge (91% versus 85%, p < 0.05), 5 years (69% versus 60%, p < 0.01), and 10 years (41% versus 31%, p < 0.01; Fig 3). Survival of 70% for patients with NVE was better than the 56% observed in patients with PVE, and this was statistically significant (p = 0.034).
|
|
|
|
|
|
Comment |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionReferences |
|---|
A similar proportion of patients in this study were treated medically (78%) compared with previously reported data. A longitudinal study by Tleyjeh and colleagues [13] of patients in Olmstead County, Minnesota, reported a rate of surgery that remained relatively constant at 10% to 23% during a 30-year span. Chu and colleagues [12] observed an operative rate of 27% in a series of 267 hospitalized patients with the diagnosis of endocarditis.
Gram-positive organisms predominated, and most were Staphylococcus spp. The predominance of gram-positive organisms and, in particular, the emergence of S aureus as the most common single organism causing endocarditis has been described previously. Chu and colleagues [12] reported a 44% incidence of S aureus in a series of hospitalized patients, whereas Romano [8] and Grunenfelder [11] and their colleagues reported high incidences of Streptococcus spp. Gram-negative and fungal cultures were rare in all series.
Outcomes
Readmission
The number of readmissions identified in this study is almost certainly an underestimation of the true magnitude of endocarditis-associated readmissions. Factors contributing to this underestimate of readmission are that admission data were only available for the two index hospitals in the study and the transient nature of the patient population associated with endocarditis. In addition, the large number of hospitals within the study area and the average follow-up duration of only 2 years could magnify this underestimation.
Emboli
Emboli were observed in 43% of patients in this study and were identified clinically or on imaging studies. Thuny and colleagues [9] evaluated the role of echocardiography in predicting embolism in a series of 383 patients with endocarditis. They observed embolus in 34.1% of their patients and reported that vegetation size exceeding 10 mm and severe vegetations mobility predicted subsequent embolism. They also found that IVDA and right-sided vegetations were predictive of subsequent emboli [9].
Survival
Overall mortality was 31% and was similar to that observed in other studies of infectious endocarditis [7, 13]. Improved survival was noted in patients treated surgically (82.6% versus 65%, p = 0.001) a finding reported in previous studies. By using propensity analysis, Cabell and colleagues [4] analyzed demonstrated a significant survival advantage in 1516 patients with NVE treated surgically compared with those treated medically (88.8% versus 62.0%, p < 0.001).
The 70% survival for patients with NVE was better than the 56% observed in patients with PVE, and this was statistically significant (p = 0.034). Increased mortality has been reported with PVE by many authors. Romano and colleagues [8] reported a mortality of 6.6% for NVE and 24.2% for PVE (p < 0.0001) in a cohort of 353 patients. Grunenfelder and colleagues [11] analyzed outcomes for 171 patients diagnosed with endocarditis. They reported lower mortality for the patients with NVE compared with those with PVE at discharge (6.1% versus 15.1%) and at 10 years (47% versus 63%) [11]. Conversely, Jassal and colleagues [14] found no difference in survival when they analyzed outcomes of 78 patients with NVE and 13 with PVE (n = 91).
Independent predictors of early death included hemodynamic instability and age older than 55 years. Independent predictors of late death included medical treatment, age older than 55 years, institution, NYHA class III or IV, and hemodynamic instability. Many of these risk factors have been previously identified. Thuny and colleagues [9] identified congestive heart failure, female gender, S aureus infection, renal insufficiency, vegetation length, and age to be independently associated with mortality [9, 14]. Balasubramanian and colleagues [3] analyzed 61 patients undergoing operations for endocarditis and identified poor left ventricular function, heart failure, and extensive infection to be independent predictors of mortality.
Limitations
The retrospective nature of this study constitutes a clear limitation. Patients with endocarditis may not have been identified, and specific patient populations may have been under or overrepresented. In addition, because patients were not followed up except through the medical record, this study may have underestimated the true mortality of this cohort. Using Kaplan-Meier survival analysis where patients were censored at their last known follow-up minimized this potential problem. Ideally, patient survival could be better evaluated by cross-referencing with the Social Security Death Index.
Because one of the study institutions was a VAMC, a disproportionate number of patients were men. Follow-up was also longer in the VAMC patients (866 versus 640 days) and was statistically significant (p = 0.022). However, data were analyzed at specific time points to avoid potential bias introduced by the discrepancy in follow-up duration. Because only two institutions, both in the same metropolitan area and both serving as referral hospitals were included in this study, the sample population evaluated may not be representative of patients with endocarditis in general. As a result of these limitations and the composition of this study population, it may not be possible to extrapolate the findings to other clinical environments.
Future Directions
To definitively identify risk factors associated with mortality in patients with endocarditis and to determine optimal strategies for treatment, a prospective randomized trial comparing medical and surgical treatment strategies would be ideal. A prospective study might also be better able to determine the true embolic risk posed by valvular vegetations of differing size and composition and define the optimal timing for surgical treatment of these patients to avoid subsequent embolic complications. A multicenter design could provide adequate numbers of patients to determine the impact of less common variables, specific organisms, and differing surgical techniques.
We chose to evaluate and analyze standard risk factors that have been associated with cardiovascular outcomes for this study. Most of these risk factors have been found to be predictive of morbidity and mortality in patients undergoing cardiac surgery but are not specific to endocarditis. It is conceivable that nonclassical risk factors that were not evaluated in this study are important in determining outcomes in patients with endocarditis and should be evaluated in future studies.
Conclusions
Despite advances in diagnosis, medical treatment, surgical technique, and critical care, short-term and long-term mortality from endocarditis remains high. Survival for surgical patients at discharge, 5, and 10 years was 91%, 69%, and 41%, respectively. Survival for medically treated patients at discharge, 5, and 10 years was 85%, 60%, and 31%, respectively. Predictors of early death included hemodynamic instability on admission (p = 0.013) and age older than 55 years (p < 0.025). Independent predictors of late death included medical treatment (p = 0.005), age older than 55 (p = 0.032), institution (p < 0.001), NYHA class III or IV (p = 0.002), and hemodynamic instability on admission (p = 0.044).
The differences in outcomes from two geographically similar institutions presented in this study highlight the potential limitations of extrapolating risk factors from one patient population to another.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionReferences |
|---|
The questions I have are related to the indications for surgery and a little bit about the detail of surgery that I would be interested in what your group’s approach is. You mentioned congestive heart failure, emboli, and persistent sepsis as indications for surgery, which I think most of us would certainly agree with. But I would like you to maybe comment on the type of organism, bacterial or fungal, in terms of an indication for how you approach this earlier cautiously or very early, the question of annular invasion, how about the size of the vegetations on the valves by echocardiographic studies? And then finally, if you have replaced a valve, what is your long-term treatment in terms of antibiotic, antifungal, short- and long-term treatment, and also the role of aortic homografts for patients with aortic valvular endocarditis? Thank you.
DR SLATER: Thank you, Dr Grover. Regarding the indications and timing of surgery, early intervention before the patients become chronically hospitalized and malnourished is best. The earlier you can make the decision to operate, the better, and intervene before they have become subjected to multiple IVs, colonized with hospital flora, and their pulmonary status deteriorates from lying in bed. Aside from obvious indications for surgery that we talked about, congestive heart failure and such, patients who are failing to normalize their white blood cell count and resolve their sepsis, and those patients that are continuing to embolize, should be taken to the operating theater.
You also asked about size of vegetation, and this is an area of continued controversy. Some authors have proposed that vegetations over 1 cm or those with excess mobility should warrant a more aggressive surgical approach, and I would agree with this. Patients who have already embolized, but still have residual vegetations would be operated on in our practice. We have taken patients who are actively embolizing and who have central nervous system involvement to the operating room. We CT scan them, and if they do not have active bleeding, we will remove their vegetation, and so far we have had good luck avoiding intracranial hemorrhage at the time of surgery.
You then asked what our surgical approach to treatment of these patients is. We are fairly aggressive in terms of débriding these patients. It is important to get back to absolutely normal tissue, because coming back, again, especially if you are going to replace the valve with a prosthetic valve, you don’t want to come back and turn your native valve endocarditis into prosthetic valve endocarditis. I think this is more easily accomplished in the aortic root, where you can take out the entire annulus and then reconstruct it with a homograft. Even if you have to take out the intervalvar fibrosa and detach the anterior mitral valve leaflet, you can then bring that back together with a homograft and some pericardium. Our preference is to use homograft to reconstruct these more extensive resections.
I think the problem of annular involvement is more difficult in the mitral position where there is not a good biological or non-prosthetic substitute, and I don’t have a good answer for when you have to extensively débride those patients except to reconstruct the annulus with pericardium. You then have to put a prosthetic valve in, because we have not been enamored with either the results of mitral homografts or Ross II type operations for this problem.
DR CONSTANTINE MAVROUDIS (Chicago, IL): That was a very nice presentation with a very large database. I wonder, what was the incidence of congenital heart disease in your patients? For instance, there is a high incidence of endocarditis in the tricuspid valve in your group. That could be, as you noted, either from drug abuse or it could in fact be from a residual small ventricular septal defect. Do you know what the incidence of congenital heart disease in those patients was? And I don’t think, unless I missed it, that you included in your multivariate analysis the presence of arrhythmias and whether endocarditis plus additive arrhythmias had any effect on mortality. I can see you presenting a few more papers out of this database. It is very interesting material. Thank you very much.
DR SLATER: Thank you for those comments, Dr Mavroudis. In terms of congenital heart disease, we had 14 patients under the age of 10 and another 14 patients age 11 to 20. Some of these patients certainly have congenital heart disease, and I think some of that depends on what you consider congenital heart disease. Certainly older patients after tetralogy having subsequent operations are still congenital heart disease, but included in this are also bicuspid valve disease, which some people do and don’t consider congenital heart disease. We did not stratify the data, though, by congenital heart disease, nor did we stratify it by arrhythmias.
DR MARC MOON (St. Louis, MO): That was a nice study, but there is always a difficulty when you are comparing medical and surgical treatment groups for endocarditis because there is always a selection bias, and we have all seen patients that we don’t think can survive surgery, and of course they get triaged into the medical therapy failure group. So my first question, I have got two short ones, did you consider looking at patients that were referred for surgery and seen by a surgical team but declined surgery for any reason to see if they could be separated out from the medical treatment group?
DR SLATER: We actually tried to do that by going through and seeing if there was a chart note by the surgical team, but because of the way our database is set up, we had very incomplete data in that regard and therefore could not draw conclusions. I agree; it is an important question. Are these patients dying because they are not referred, and if so, is it really a fair comparison? We were not able to extract that data, again pointing to the need to really prospectively look at patient characteristics and outcomes in this patient group.
DR MOON: And my last question is, I think in your abstract you said nobody with prosthetic valve endocarditis survived 10 years, but it looked like that wasn’t the case here. Did you have survivors with prosthetic valves?
DR SLATER: I think there is a patient that made it almost to 10 years, but not quite. Only 7% of our patients had prosthetic valve endocarditis, and out at the end, most of the patients have been censored by time. The length of the study was only 10 years, and most patients do not have 10-year follow-up.
DR MOON: I think the dismal prognosis is consistent with other papers showing the same thing of prosthetic valve endocarditis, and I was wondering if that impacts your valve choice as to whether you put in a mechanical or a bioprosthetic valve in these patients. I tend to put a bioprosthetic in everybody with that. Is that your policy as well?
DR SLATER: I think there are two issues there. One is these patients don’t live that long. They have complicated courses, and therefore, avoiding anticoagulation is nice because you don’t want to deal with that added complication with a group of patients that tend to be very sick. Additionally, many of our patients are IV drug abuse patients and those patients can’t safely take anticoagulation.
In terms of reoperation, whatever type of prosthetic material you put in there is prone to reinfection. Whether that is a sewing ring on a prosthetic or a mechanical valve, I think it is probably the same. Therefore in the aortic position I am quite prone to use homografts, and I don’t have a great solution for the mitral position.
|
References |
|---|
|
TopAbstractIntroductionMaterial and MethodsResultsCommentDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  G. Nadji, D. Rusinaru, J.-P. Remadi, A. Jeu, C. Sorel, and C. Tribouilloy Heart failure in left-sided native valve infective endocarditis: characteristics, prognosis, and results of surgical treatment Eur J Heart Fail, July 1, 2009; 11(7): 668 - 675. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Bannay, B. Hoen, X. Duval, J.-F. Obadia, C. Selton-Suty, V. Le Moing, P. Tattevin, B. Iung, F. Delahaye, F. Alla, et al. The impact of valve surgery on short- and long-term mortality in left-sided infective endocarditis: do differences in methodological approaches explain previous conflicting results? Eur. Heart J., February 9, 2009; (2009) ehp008v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. G. Daniel and F. A. Flachskampf CHAPTER 22 Infective Endocarditis ESC Textbook of Cardiovascular Medicine, January 1, 2009; 2(1): med-9780199566990-chapter - med-9780199566990-chapter. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. H. Rahimtoola The Year in Valvular Heart Disease. J. Am. Coll. Cardiol., February 19, 2008; 51(7): 760 - 770. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
