HLC 1533 No. of Pages 8
ORIGINAL ARTICLE
Heart, Lung and Circulation (2014) xx, 1–8 1443-9506/04/$36.00 http://dx.doi.org/10.1016/j.hlc.2014.02.010
Infective Endocarditis B. Dunne *, T. Marr, D. Kim, D. Andrews, M. Edwards, C. Merry, R. Larbalestier Department of Cardiothoracic Surgery, Royal Perth Hospital, Perth, Western Australia, Australia Received 15 December 2013; accepted 11 February 2014; online published-ahead-of-print xxx
Background
Infective endocarditis continues to pose a therapeutic challenge to treating clinicians. We believe that the successful management of endocarditis mandates a thorough understanding of the risk factors for adverse outcomes and a co-ordinated team approach.
Methods
Between the years 2000 and 2009, 85 patients required surgery for infective endocarditis, with a total of 112 infected valves being treated surgically. Data was analysed to determine factors significantly associated with morbidity and mortality.
Results
The mean age was 50.5 years. Nine (10.5%) of these patients had Prosthetic Valve Endocarditis, the remaining 76 (89.5%) had Native Valve Endocarditis. Twenty-nine percent of patients were NYHA 4 pre-operatively, 15% of patients were haemodynamically unstable requiring inotropic support, 34% were persistently febrile despite antibiotic therapy, and 48% had suffered any embolic event, 20% suffered cerebral emboli. The commonest causative organism in our series was Staphylococcus Aureus (54.1%) with 2.3% of cases being due to MRSA. The second commonest organism isolated was Streptococcus spp. at 21.1%. Operative mortality was 12.9%, of which on-table mortality was 2.2%. Mean follow-up was 56 months (range 1-151). Early recurrence rates (/< 65 years) Gender Intravenous drug abuse
Return to Theatre for Bleeding Tracheostomy Cerebrovascular Accident Acute Renal Failure
Operative technique
Causative organism Prosthetic Valve Endocarditis vs. Native Valve Endocarditis Valvular involvement (Pure Aortic vs. Pure Mitral vs. Multivalvular) Periannular extension Uncontrolled sepsis pre-operatively (defined as ongoing fevers >38 C despite antibiotics) NYHA Class Inotropic requirement Duration of pre-operative antibiotics Pre-operative cerebral emboli
To our knowledge, this article is the largest published series of surgically treated infective endocarditis to come from Australasia.
Materials and Methods Data collection Between the years 2000 and 2009, 85 patients underwent surgery for treatment of their infective endocarditis at Royal Perth Hospital. A total of 112 infected valves were treated surgically. These patients’ records were scrutinised by retrospective chart review. Long-term follow-up was conducted by telephone interview with the patients themselves, if contactable, and with their current general practitioner or cardiologist. All data was collected and entered into a database. All post-operative complications were recorded, the following were analysed based on their incidence and/or associated morbidity (Table 1). The following post-operative variables were also analysed to determine their association with In-hospital mortality
All operations were performed through a median sternotomy incision with cardiopulmonary bypass. Cardiac arrest was achieved using hyperkalaemic cold blood cardioplegia solution delivered using a combination of antegrade and retrograde delivery in a 50:50 ratio. Cardioplegia was maintained with intermittent retrograde cardioplegia at 20-30 minute intervals depending on surgeon preference or at any sign of cardiac electrical activity. The most commonly performed procedure was a mechanical valve replacement, which was the procedure of choice for 78 (69.6%) of the surgically treated valves. A bioprosthetic valve was used in 13 cases (11.6%). A total of 91 (81.2%) valve replacements were performed out of a total of 112 infected valves. As displayed in Table 2, 23 valves (20.5%) underwent repair by a variety of techniques, while six valves (5.3%) were treated solely by vegetectomy. A small proportion of patients required concomitant cardiac procedures unrelated to their infective endocarditis. Coronary artery surgery was performed in three (3.5%) patients. Ventricular septal defects were closed in two (2.3%) patients and one patient (1.1%) had a left ventricular myectomy to repair hypertrophic obstructive cardiomyopathy.
Statistical methods Non-parametric statistical analysis was applied in the analysis of the dataset that was predominantly comprised of nominal (binary) variables, with the exception of age as an interval variable. ICU stay was categorised as either three days, five days, seven days or more than seven days. Preoperative antibiotic administration was categorised as either seven days, 14 days or more than 14 days. Univariate tests of independence between variables were achieved with the McNemar test, and where relevant, Fisher’s exact test, based on the null hypothesis of a shared marginal distribution, with p-values reported for these tests in conjunction with a
Table 2 Distribution of procedures by type and valve.
Valve Replacement
Aortic
Mitral
Tricuspid
Pulmonary
42 (37.5%)
43 (38.3%)
3 (2.6%)
3 (2.6%)
Bentall’s Vegetectomy
3 (2.6%) 1 (0.8%)
2 (1.7%)
3 (1.7%)
0
Suture Repair
0
3 (2.6%)
4 (3.5%)
0
Commisuroplasty
0
2 (1.7%)
2 (1.7%)
0
Annuloplasty
0
3 (2.6%)
0
0
Neochords
0
1 (0.8%)
1 (0.8%)
0
Quadrangular Resection
0
1 (0.8%)
0
0
Please cite this article in press as: Dunne B, et al. Infective Endocarditis. Heart, Lung and Circulation (2014), http://dx.doi. org/10.1016/j.hlc.2014.02.010
HLC 1533 No. of Pages 8
3
Infective
Mantel-Haenszel common Odds ratio p-value as a means of comparison. Close agreement is expected for statistically significant univariate relationships. The strength of significant univariate relationships were further analysed by means of the correlation-like measure (a proxy for R2), the f (phi) coefficient, with a range of 0 to 1, with strong relationships tending toward 1. Multivariate binary logistic regression modelling was performed to analyse more complex interrelationships amongst all sample variables and early post-operative mortality by means of a combined forward and backward stepwise conditional approach. This ensured a parsimonious outcome with variable selection. Results were considered significant where p < 0.05. Beta values were reported to indicate positive or negative relationships with respect to mortality, whilst a Nagelkerke R-statistic measured model fit. Odds ratios with respective p-values were reported as standard. Wald x2 values were also reported to quantify unique contributions by each predictor variable, with larger values suggestive of greater unique contributions. Statistical analysis was performed with SPSS 19.0. (http://www-01.ibm.com/support/ docview.wss?uid=swg21476197)
Results The mean age of the patients was 50.5 years (range 15 – 80). The sex distribution was 60 male patients (71%) and 25 female patients (29%). A history of IV drug abuse was obtained in 25 (29.4%) patients. Four patients (4.7%) were immunosuppressed and three patients (3.5%) had intracardiac foreign bodies such as a pacemaker. Nine (10.5%) of these patients had Prosthetic Valve Endocarditis, the remaining 76 (89.5%) had Native Valve Endocarditis. Twenty-five (29%) patients were NYHA 4 pre-operatively (mean NYHA Class 2.3), 13 (15%) patients were haemodynamically unstable requiring inotropic support, 29 patients (34%) were persistently febrile despite antibiotic therapy, 41 patients (48%) had suffered embolic events, 17 of these (20%) being cerebral emboli and 46 patients (54.1%) had vegetations >1 cm in largest diameter. Eighty-five patients underwent surgery for treatment of infective endocarditis, with a total of 112 valves being treated surgically. Twenty-six patients (30.5%) had multi-valve involvement. Mitral valve endocarditis was the most common isolated valve involved (29 cases, 34.1%) and also the most common valve involved in multi-valve endocarditis (24 cases). Aortic valve endocarditis was the second most common valve involved in isolated (22 cases, 25.8%) and multivalve disease (21 cases). In total left-sided valves were involved in 77 (90.7%) cases. Periannular extension was present in 22.3% of cases and 16.9% of all involved valves (n=19). This was more common in aortic valve endocarditis (20.9% of all involved aortic valves) than in mitral valve endocarditis (14.8% of all
involved mitral valves) or tricuspid valve endocarditis (16.6% of all involved tricuspid valves). All patients had multiple blood cultures prior to surgery. The commonest causative microbiological agent in our series was Staphylococcus Aureus with 44 cases (54.1%), two of these (2.3%) being due to MRSA. The second commonest organism isolated was Streptococcus spp. with 18 cases (21.1%). Seventeen cases (20%) were due to other organisms, nine (10.5%) of these being other Gram Positive Cocci (mostly Enterococcus) and eight (9.4%) being Gram Negative Bacilli (mostly Haemophilus). More than one organism was isolated in two cases (2.3%) and six cases (7.0%) had a diagnosis of ‘culture negative endocarditis’. The mean duration of pre-operative antibiotic treatment was 16 days, however, this was influenced significantly by a minority of patients with prolonged antibiotic treatment prior to surgery, the median duration of antibiotic treatment was only eight days, in keeping with our aggressive approach to surgical management.
Operative mortality On-table death occurred in two patients (2.2%). In one case, a Bentall’s procedure had to be abandoned due to gross destruction of the fibrous skeleton of the heart precluding any reconstructive surgery. A second patient failed to wean from bypass after mitral valve replacement on a background of rheumatic mitral stenosis and severe pulmonary hypertension complicated by endocarditis. Total operative mortality was 12.9% (n=11). Ongoing sepsis was the leading cause of death post-operatively (n=4, 4.7%), low cardiac output (n=3, 3.5%), respiratory failure (n=1, 1.1%) and haemorrhagic stroke (n=1, 1.1%) also contributing. Three of these deaths were in nine patients with PVE (33% mortality) and eight deaths were in 76 patients with NVE (10.5%).
Early post-operative complications Acute renal failure was the most commonly encountered postoperative complication with 22.3% (n=19) of patients affected. There was a 5.8% (n=5) incidence of post-operative cerebrovascular event and, unfortunately, a 4.7% (n=4) rate of early paravalvular leak. Permanent pacemakers were required in 3.5% (n=3) of patients for complete heart block post-operatively and 2.3% (n=2) of patients required tracheostomy insertion to aid prolonged wean for ventilatory support. Deep sternal wound infection occurred in 1.1% (n=1) of patients.
Late results There were two cases (2.3%) of early recurrence (at less than three months post-op). All cases occurred in patients who had native valve endocarditis. One case was a mechanical replacement of both aortic and mitral valves and one was a mechanical replacement of the tricuspid valve. Both were taking immunosuppressive medication prior to the time of surgery. One patient was managed non-operatively, as she was also diagnosed with inoperable breast cancer simultaneously and subsequently passed away and one was managed with replacement of his infected valve but also died postoperatively.
Please cite this article in press as: Dunne B, et al. Infective Endocarditis. Heart, Lung and Circulation (2014), http://dx.doi. org/10.1016/j.hlc.2014.02.010
HLC 1533 No. of Pages 8
4
B. Dunne et al.
Table 3 Univariate analysis of predictors of operative mortality. Predictors of early post operative mortality Variable
Death (n=11)
Survivors (n=74)
OR
RR
p
F
McNemar / Fisher’s Exact
Mantel-Haenszel
Pre-operative variables Age > 65
6 (55%)
13 (18%)
5.62
4.16
0.013a
0.011
0.298
Inotropic requirement
5 (45%)
8 (11%)
6.88
4.62
0.011a
0.007
0.323
Uncontrolled sepsis
8 (73%)
21 (28%)
6.73
5.15
0.006a
0.008
0.314
Cerebral emboli Male
6 (55%) 7 (64%)
11 (15%) 53 (72%)
6.87 0.69
4.80 0.73
0.007a 0.724
0.005 0.589
0.333 -0.059
Prosthetic valve
3 (27%)
6 (8%)
4.25
3.17
0.089
0.070
0.209
Staph A +
7 (64%)
40 (54%)
1.49
1.42
0.747
0.553
0.065
IVDU
0 (0%)
25 (34%)
n/a
n/a
0.029
n/a
-0.249
R endocarditis
0 (0%)
8 (11%)
n/a
n/a
0.589
n/a
-0.124
L endocarditis
8 (73%)
43 (58%)
1.92
1.78
0.513
0.300
0.100
Multivalvular
3 (27%)
21 (28%)
0.95
0.95
1.000
0.939
-0.008
Periannular extension NYHA 3/4
4 (36%) 5 (45%)
16 (22%) 43 (58%)
2.07 0.60
1.86 0.64
0.278 0.521a
0.289 0.433
0.117 -0.086
Pre-op ABRx 65, Pre-Operative Inotropic Requirement, Uncontrolled Sepsis or Cerebral Emboli and Post-Operative Acute Renal Failure all achieved statistical significance. Prosthetic Valve Endocarditis and Resternotomy for Bleeding trended towards significance (Table 3). These variables were then subjected to multivariate analysis in order to exclude the possibility of confounding. The
Please cite this article in press as: Dunne B, et al. Infective Endocarditis. Heart, Lung and Circulation (2014), http://dx.doi. org/10.1016/j.hlc.2014.02.010
HLC 1533 No. of Pages 8
5
Infective
Figure 1 Long-term survival.
multivariate analysis revealed only one variable (Age >65 years) that was predictive of operative mortality.
Factors affecting post-operative morbidity All pre-operative variables were also subjected to univariate analysis to determine their impact on post-operative morbidity. The significant results of this analysis are presented in Table 4.
Our results indicate that Age >65, Pre-Operative Cerebral Emboli and Inotropic Requirement are associated with PostOperative Acute Renal Failure. Pre-Operative Cerebral Emboli are significantly associated with Post-Operative Cerebrovascular Accident – most likely due to two patients who suffered post-operative haemorrhagic transformation of pre-operative embolic infarcts. Female sex and Left-Sided Endocarditis approach a statistically significant association with Post-Operative CVA.
Table 4 Univariate analysis of impact of pre-operative variables on postoperative morbidity. Relationship between pre-operative variables and post-operative morbiditya Pre-operative variable
Post-operative variable CVA
AF
ARF
Late Recurrence
n=6
n=14
n=20
n=7
n
p
n
Age>65
n 11
4
0.059b
L Endocarditis
6
b
0.077
Cerebral emboli
4
0.013b
Female
p
p
a
p
5
0.021b
0.012
IVDU
Inotropic requirement
n b
8
0.022b
7
0.01b
Hantel-Haenszel common odds ratio p-value reported.
b
Fisher’s Exact test p-value.
Please cite this article in press as: Dunne B, et al. Infective Endocarditis. Heart, Lung and Circulation (2014), http://dx.doi. org/10.1016/j.hlc.2014.02.010
HLC 1533 No. of Pages 8
6
B. Dunne et al.
Unsurprisingly, IV drug use prior to surgery is a known risk factor for ongoing IV drug use and has proven to be statistically significant as a risk factor for late recurrence.
Discussion Ever since its recognition as a pathological process bacterial endocarditis has proven to be a challenge to diagnose and treat. The mortality associated with this condition approached 100% in the pre-antibiotic era [1] but improved dramatically with the introduction of antibiotic therapy and fell to between 30 and 40% [1,2]. The first cases of bacterial endocarditis successfully treated with surgery began to emerge in the early 1960’s [3–5] and the first series of five patients was reported in 1967 [6]. Surgical management of bacterial endocarditis decreased mortality rates to approximately 20% initially [1] and there have been significant improvements since then with some groups reporting mortality in native valve endocarditis as low as 4-6% [7–10]. However, most series still report an overall mortality of 12-21% [15,17–19]. Furthermore, the mortality associated with prosthetic valve endocarditis remains even higher, ranging from 16% to 45% [7,8,11,12,16]. Repeated comparisons with medical therapy alone (although never in a randomised controlled trial) have shown the short and long-term survival benefit of surgery. Early comparisons from the 1960s and 1970s showed improved survival for both native valve and prosthetic valve endocarditis [13] in the shorter term and later publications have demonstrated convincingly that surgery improves long-term survival [14,15]. More recent data has shown encouraging results with valve-preserving repair techniques in infective endocarditis [23–25]. The most recent debate surrounding the management of endocarditis revolves around the timing of surgical intervention. A recent RCT comparing very early surgery to traditional approaches showed improved outcomes in low-risk patients with large left-sided vegetations [20]. Recent review articles have shown increased support for early surgical intervention [21,22] and have suggested a gradual decrease in mortality over time that mirrors the increasingly aggressive approach to surgical intervention [21], although this may well be a reflection of improved peri-operative care. The patient with infective endocarditis and cerebral emboli is almost certainly the most difficult clinical scenario and timing of surgery is often a fine balance between the risk of haemorrhagic transformation of the infarct and the risk of further embolic events. The traditional approach has been to delay surgery for at least one to two weeks in the case of an ischaemic stroke [26], but even this delay has been questioned with recent evidence showing no increase in risk when surgery is performed less than one week post-ischaemic stroke [27]. A multi-disciplinary approach to these patients is also a concept we are embracing. The concept of the ‘‘Endocarditis Team’’ (made up of cardiac surgeons, cardiologists, infectious
diseases physicians and, especially important in cases with cerebral emboli, neurologists) and based on the concept of the ‘‘Heart Team’’ born out of the SYNTAX trial, has been mentioned in the literature [21] and, we feel, is central to a successful outcome in these complex cases. This paper was written using data from a decade’s experience of surgically managed infective endocarditis. These patients were predominantly managed along the traditional paradigm rather than with very early surgical intervention. But this policy has been modified over that time to adopt a more aggressive surgical approach with earlier intervention. We hoped to identify factors in these patients associated with adverse peri-operative outcomes and so use the identified factors as indicators that aggressive early intervention is warranted in the future. Age over 65 years, Pre-Operative Inotropic Requirement, Uncontrolled Sepsis or Cerebral Emboli were the factors that we found to be significantly associated with increased mortality. As such, certainly in cases with patients over 65 years of age, or with pre-operative inotrope/vasopressor requirement we would encourage very early (48 hours of intravenous antibiotics. Those patients with pre-operative cerebral emboli pose the greatest dilemma and must be managed in close liaison with a neurologist to ensure the earliest possible surgical intervention. The significant association between Age >65 and Pre-Operative Cerebral Emboli and Inotropic Requirement with PostOperative ARF is unsurprising. Interestingly, neither the Causative Organism nor the Duration of Pre-Operative Antibiotic Therapy had a significant impact on any post-operative outcome, but most relevantly had no impact on rates of early recurrence or paravalvular leak. However, low numbers of early recurrence (n=2) made accurate statistical analysis difficult. However, we did record six cases of late recurrence (>3 months). Analysis of these patients revealed that the only factor significantly associated with late recurrence is a history of intravenous drug use. A previous paper, by the primary investigator in this study, on surgically treated infective endocarditis demonstrated poorer outcomes in non-streptococcal NVE [8] and advocated early aggressive surgical intervention, particularly in cases of Staph aureus endocarditis. As such, our unit approach has been to intervene earlier in Staphylococcal endocarditis and as such the similar outcomes in Staphylococcal and Streptococcal endocarditis may be due to a more aggressive approach to the former.
Conclusion This review demonstrated that our results on mortality and early post-operative outcomes compare well with other
Please cite this article in press as: Dunne B, et al. Infective Endocarditis. Heart, Lung and Circulation (2014), http://dx.doi. org/10.1016/j.hlc.2014.02.010
HLC 1533 No. of Pages 8
7
Infective
published data [15,17–19]. However, this review was undertaken to identify weaknesses in our approach and help guide our emerging policy of earlier surgical intervention by identifying risk factors for poor outcomes. As already stated, over the period of this review our unit policy has changed, guided by previous author experience, and has moved towards a policy of early surgical intervention (
ORIGINAL ARTICLE
Heart, Lung and Circulation (2014) xx, 1–8 1443-9506/04/$36.00 http://dx.doi.org/10.1016/j.hlc.2014.02.010
Infective Endocarditis B. Dunne *, T. Marr, D. Kim, D. Andrews, M. Edwards, C. Merry, R. Larbalestier Department of Cardiothoracic Surgery, Royal Perth Hospital, Perth, Western Australia, Australia Received 15 December 2013; accepted 11 February 2014; online published-ahead-of-print xxx
Background
Infective endocarditis continues to pose a therapeutic challenge to treating clinicians. We believe that the successful management of endocarditis mandates a thorough understanding of the risk factors for adverse outcomes and a co-ordinated team approach.
Methods
Between the years 2000 and 2009, 85 patients required surgery for infective endocarditis, with a total of 112 infected valves being treated surgically. Data was analysed to determine factors significantly associated with morbidity and mortality.
Results
The mean age was 50.5 years. Nine (10.5%) of these patients had Prosthetic Valve Endocarditis, the remaining 76 (89.5%) had Native Valve Endocarditis. Twenty-nine percent of patients were NYHA 4 pre-operatively, 15% of patients were haemodynamically unstable requiring inotropic support, 34% were persistently febrile despite antibiotic therapy, and 48% had suffered any embolic event, 20% suffered cerebral emboli. The commonest causative organism in our series was Staphylococcus Aureus (54.1%) with 2.3% of cases being due to MRSA. The second commonest organism isolated was Streptococcus spp. at 21.1%. Operative mortality was 12.9%, of which on-table mortality was 2.2%. Mean follow-up was 56 months (range 1-151). Early recurrence rates (/< 65 years) Gender Intravenous drug abuse
Return to Theatre for Bleeding Tracheostomy Cerebrovascular Accident Acute Renal Failure
Operative technique
Causative organism Prosthetic Valve Endocarditis vs. Native Valve Endocarditis Valvular involvement (Pure Aortic vs. Pure Mitral vs. Multivalvular) Periannular extension Uncontrolled sepsis pre-operatively (defined as ongoing fevers >38 C despite antibiotics) NYHA Class Inotropic requirement Duration of pre-operative antibiotics Pre-operative cerebral emboli
To our knowledge, this article is the largest published series of surgically treated infective endocarditis to come from Australasia.
Materials and Methods Data collection Between the years 2000 and 2009, 85 patients underwent surgery for treatment of their infective endocarditis at Royal Perth Hospital. A total of 112 infected valves were treated surgically. These patients’ records were scrutinised by retrospective chart review. Long-term follow-up was conducted by telephone interview with the patients themselves, if contactable, and with their current general practitioner or cardiologist. All data was collected and entered into a database. All post-operative complications were recorded, the following were analysed based on their incidence and/or associated morbidity (Table 1). The following post-operative variables were also analysed to determine their association with In-hospital mortality
All operations were performed through a median sternotomy incision with cardiopulmonary bypass. Cardiac arrest was achieved using hyperkalaemic cold blood cardioplegia solution delivered using a combination of antegrade and retrograde delivery in a 50:50 ratio. Cardioplegia was maintained with intermittent retrograde cardioplegia at 20-30 minute intervals depending on surgeon preference or at any sign of cardiac electrical activity. The most commonly performed procedure was a mechanical valve replacement, which was the procedure of choice for 78 (69.6%) of the surgically treated valves. A bioprosthetic valve was used in 13 cases (11.6%). A total of 91 (81.2%) valve replacements were performed out of a total of 112 infected valves. As displayed in Table 2, 23 valves (20.5%) underwent repair by a variety of techniques, while six valves (5.3%) were treated solely by vegetectomy. A small proportion of patients required concomitant cardiac procedures unrelated to their infective endocarditis. Coronary artery surgery was performed in three (3.5%) patients. Ventricular septal defects were closed in two (2.3%) patients and one patient (1.1%) had a left ventricular myectomy to repair hypertrophic obstructive cardiomyopathy.
Statistical methods Non-parametric statistical analysis was applied in the analysis of the dataset that was predominantly comprised of nominal (binary) variables, with the exception of age as an interval variable. ICU stay was categorised as either three days, five days, seven days or more than seven days. Preoperative antibiotic administration was categorised as either seven days, 14 days or more than 14 days. Univariate tests of independence between variables were achieved with the McNemar test, and where relevant, Fisher’s exact test, based on the null hypothesis of a shared marginal distribution, with p-values reported for these tests in conjunction with a
Table 2 Distribution of procedures by type and valve.
Valve Replacement
Aortic
Mitral
Tricuspid
Pulmonary
42 (37.5%)
43 (38.3%)
3 (2.6%)
3 (2.6%)
Bentall’s Vegetectomy
3 (2.6%) 1 (0.8%)
2 (1.7%)
3 (1.7%)
0
Suture Repair
0
3 (2.6%)
4 (3.5%)
0
Commisuroplasty
0
2 (1.7%)
2 (1.7%)
0
Annuloplasty
0
3 (2.6%)
0
0
Neochords
0
1 (0.8%)
1 (0.8%)
0
Quadrangular Resection
0
1 (0.8%)
0
0
Please cite this article in press as: Dunne B, et al. Infective Endocarditis. Heart, Lung and Circulation (2014), http://dx.doi. org/10.1016/j.hlc.2014.02.010
HLC 1533 No. of Pages 8
3
Infective
Mantel-Haenszel common Odds ratio p-value as a means of comparison. Close agreement is expected for statistically significant univariate relationships. The strength of significant univariate relationships were further analysed by means of the correlation-like measure (a proxy for R2), the f (phi) coefficient, with a range of 0 to 1, with strong relationships tending toward 1. Multivariate binary logistic regression modelling was performed to analyse more complex interrelationships amongst all sample variables and early post-operative mortality by means of a combined forward and backward stepwise conditional approach. This ensured a parsimonious outcome with variable selection. Results were considered significant where p < 0.05. Beta values were reported to indicate positive or negative relationships with respect to mortality, whilst a Nagelkerke R-statistic measured model fit. Odds ratios with respective p-values were reported as standard. Wald x2 values were also reported to quantify unique contributions by each predictor variable, with larger values suggestive of greater unique contributions. Statistical analysis was performed with SPSS 19.0. (http://www-01.ibm.com/support/ docview.wss?uid=swg21476197)
Results The mean age of the patients was 50.5 years (range 15 – 80). The sex distribution was 60 male patients (71%) and 25 female patients (29%). A history of IV drug abuse was obtained in 25 (29.4%) patients. Four patients (4.7%) were immunosuppressed and three patients (3.5%) had intracardiac foreign bodies such as a pacemaker. Nine (10.5%) of these patients had Prosthetic Valve Endocarditis, the remaining 76 (89.5%) had Native Valve Endocarditis. Twenty-five (29%) patients were NYHA 4 pre-operatively (mean NYHA Class 2.3), 13 (15%) patients were haemodynamically unstable requiring inotropic support, 29 patients (34%) were persistently febrile despite antibiotic therapy, 41 patients (48%) had suffered embolic events, 17 of these (20%) being cerebral emboli and 46 patients (54.1%) had vegetations >1 cm in largest diameter. Eighty-five patients underwent surgery for treatment of infective endocarditis, with a total of 112 valves being treated surgically. Twenty-six patients (30.5%) had multi-valve involvement. Mitral valve endocarditis was the most common isolated valve involved (29 cases, 34.1%) and also the most common valve involved in multi-valve endocarditis (24 cases). Aortic valve endocarditis was the second most common valve involved in isolated (22 cases, 25.8%) and multivalve disease (21 cases). In total left-sided valves were involved in 77 (90.7%) cases. Periannular extension was present in 22.3% of cases and 16.9% of all involved valves (n=19). This was more common in aortic valve endocarditis (20.9% of all involved aortic valves) than in mitral valve endocarditis (14.8% of all
involved mitral valves) or tricuspid valve endocarditis (16.6% of all involved tricuspid valves). All patients had multiple blood cultures prior to surgery. The commonest causative microbiological agent in our series was Staphylococcus Aureus with 44 cases (54.1%), two of these (2.3%) being due to MRSA. The second commonest organism isolated was Streptococcus spp. with 18 cases (21.1%). Seventeen cases (20%) were due to other organisms, nine (10.5%) of these being other Gram Positive Cocci (mostly Enterococcus) and eight (9.4%) being Gram Negative Bacilli (mostly Haemophilus). More than one organism was isolated in two cases (2.3%) and six cases (7.0%) had a diagnosis of ‘culture negative endocarditis’. The mean duration of pre-operative antibiotic treatment was 16 days, however, this was influenced significantly by a minority of patients with prolonged antibiotic treatment prior to surgery, the median duration of antibiotic treatment was only eight days, in keeping with our aggressive approach to surgical management.
Operative mortality On-table death occurred in two patients (2.2%). In one case, a Bentall’s procedure had to be abandoned due to gross destruction of the fibrous skeleton of the heart precluding any reconstructive surgery. A second patient failed to wean from bypass after mitral valve replacement on a background of rheumatic mitral stenosis and severe pulmonary hypertension complicated by endocarditis. Total operative mortality was 12.9% (n=11). Ongoing sepsis was the leading cause of death post-operatively (n=4, 4.7%), low cardiac output (n=3, 3.5%), respiratory failure (n=1, 1.1%) and haemorrhagic stroke (n=1, 1.1%) also contributing. Three of these deaths were in nine patients with PVE (33% mortality) and eight deaths were in 76 patients with NVE (10.5%).
Early post-operative complications Acute renal failure was the most commonly encountered postoperative complication with 22.3% (n=19) of patients affected. There was a 5.8% (n=5) incidence of post-operative cerebrovascular event and, unfortunately, a 4.7% (n=4) rate of early paravalvular leak. Permanent pacemakers were required in 3.5% (n=3) of patients for complete heart block post-operatively and 2.3% (n=2) of patients required tracheostomy insertion to aid prolonged wean for ventilatory support. Deep sternal wound infection occurred in 1.1% (n=1) of patients.
Late results There were two cases (2.3%) of early recurrence (at less than three months post-op). All cases occurred in patients who had native valve endocarditis. One case was a mechanical replacement of both aortic and mitral valves and one was a mechanical replacement of the tricuspid valve. Both were taking immunosuppressive medication prior to the time of surgery. One patient was managed non-operatively, as she was also diagnosed with inoperable breast cancer simultaneously and subsequently passed away and one was managed with replacement of his infected valve but also died postoperatively.
Please cite this article in press as: Dunne B, et al. Infective Endocarditis. Heart, Lung and Circulation (2014), http://dx.doi. org/10.1016/j.hlc.2014.02.010
HLC 1533 No. of Pages 8
4
B. Dunne et al.
Table 3 Univariate analysis of predictors of operative mortality. Predictors of early post operative mortality Variable
Death (n=11)
Survivors (n=74)
OR
RR
p
F
McNemar / Fisher’s Exact
Mantel-Haenszel
Pre-operative variables Age > 65
6 (55%)
13 (18%)
5.62
4.16
0.013a
0.011
0.298
Inotropic requirement
5 (45%)
8 (11%)
6.88
4.62
0.011a
0.007
0.323
Uncontrolled sepsis
8 (73%)
21 (28%)
6.73
5.15
0.006a
0.008
0.314
Cerebral emboli Male
6 (55%) 7 (64%)
11 (15%) 53 (72%)
6.87 0.69
4.80 0.73
0.007a 0.724
0.005 0.589
0.333 -0.059
Prosthetic valve
3 (27%)
6 (8%)
4.25
3.17
0.089
0.070
0.209
Staph A +
7 (64%)
40 (54%)
1.49
1.42
0.747
0.553
0.065
IVDU
0 (0%)
25 (34%)
n/a
n/a
0.029
n/a
-0.249
R endocarditis
0 (0%)
8 (11%)
n/a
n/a
0.589
n/a
-0.124
L endocarditis
8 (73%)
43 (58%)
1.92
1.78
0.513
0.300
0.100
Multivalvular
3 (27%)
21 (28%)
0.95
0.95
1.000
0.939
-0.008
Periannular extension NYHA 3/4
4 (36%) 5 (45%)
16 (22%) 43 (58%)
2.07 0.60
1.86 0.64
0.278 0.521a
0.289 0.433
0.117 -0.086
Pre-op ABRx 65, Pre-Operative Inotropic Requirement, Uncontrolled Sepsis or Cerebral Emboli and Post-Operative Acute Renal Failure all achieved statistical significance. Prosthetic Valve Endocarditis and Resternotomy for Bleeding trended towards significance (Table 3). These variables were then subjected to multivariate analysis in order to exclude the possibility of confounding. The
Please cite this article in press as: Dunne B, et al. Infective Endocarditis. Heart, Lung and Circulation (2014), http://dx.doi. org/10.1016/j.hlc.2014.02.010
HLC 1533 No. of Pages 8
5
Infective
Figure 1 Long-term survival.
multivariate analysis revealed only one variable (Age >65 years) that was predictive of operative mortality.
Factors affecting post-operative morbidity All pre-operative variables were also subjected to univariate analysis to determine their impact on post-operative morbidity. The significant results of this analysis are presented in Table 4.
Our results indicate that Age >65, Pre-Operative Cerebral Emboli and Inotropic Requirement are associated with PostOperative Acute Renal Failure. Pre-Operative Cerebral Emboli are significantly associated with Post-Operative Cerebrovascular Accident – most likely due to two patients who suffered post-operative haemorrhagic transformation of pre-operative embolic infarcts. Female sex and Left-Sided Endocarditis approach a statistically significant association with Post-Operative CVA.
Table 4 Univariate analysis of impact of pre-operative variables on postoperative morbidity. Relationship between pre-operative variables and post-operative morbiditya Pre-operative variable
Post-operative variable CVA
AF
ARF
Late Recurrence
n=6
n=14
n=20
n=7
n
p
n
Age>65
n 11
4
0.059b
L Endocarditis
6
b
0.077
Cerebral emboli
4
0.013b
Female
p
p
a
p
5
0.021b
0.012
IVDU
Inotropic requirement
n b
8
0.022b
7
0.01b
Hantel-Haenszel common odds ratio p-value reported.
b
Fisher’s Exact test p-value.
Please cite this article in press as: Dunne B, et al. Infective Endocarditis. Heart, Lung and Circulation (2014), http://dx.doi. org/10.1016/j.hlc.2014.02.010
HLC 1533 No. of Pages 8
6
B. Dunne et al.
Unsurprisingly, IV drug use prior to surgery is a known risk factor for ongoing IV drug use and has proven to be statistically significant as a risk factor for late recurrence.
Discussion Ever since its recognition as a pathological process bacterial endocarditis has proven to be a challenge to diagnose and treat. The mortality associated with this condition approached 100% in the pre-antibiotic era [1] but improved dramatically with the introduction of antibiotic therapy and fell to between 30 and 40% [1,2]. The first cases of bacterial endocarditis successfully treated with surgery began to emerge in the early 1960’s [3–5] and the first series of five patients was reported in 1967 [6]. Surgical management of bacterial endocarditis decreased mortality rates to approximately 20% initially [1] and there have been significant improvements since then with some groups reporting mortality in native valve endocarditis as low as 4-6% [7–10]. However, most series still report an overall mortality of 12-21% [15,17–19]. Furthermore, the mortality associated with prosthetic valve endocarditis remains even higher, ranging from 16% to 45% [7,8,11,12,16]. Repeated comparisons with medical therapy alone (although never in a randomised controlled trial) have shown the short and long-term survival benefit of surgery. Early comparisons from the 1960s and 1970s showed improved survival for both native valve and prosthetic valve endocarditis [13] in the shorter term and later publications have demonstrated convincingly that surgery improves long-term survival [14,15]. More recent data has shown encouraging results with valve-preserving repair techniques in infective endocarditis [23–25]. The most recent debate surrounding the management of endocarditis revolves around the timing of surgical intervention. A recent RCT comparing very early surgery to traditional approaches showed improved outcomes in low-risk patients with large left-sided vegetations [20]. Recent review articles have shown increased support for early surgical intervention [21,22] and have suggested a gradual decrease in mortality over time that mirrors the increasingly aggressive approach to surgical intervention [21], although this may well be a reflection of improved peri-operative care. The patient with infective endocarditis and cerebral emboli is almost certainly the most difficult clinical scenario and timing of surgery is often a fine balance between the risk of haemorrhagic transformation of the infarct and the risk of further embolic events. The traditional approach has been to delay surgery for at least one to two weeks in the case of an ischaemic stroke [26], but even this delay has been questioned with recent evidence showing no increase in risk when surgery is performed less than one week post-ischaemic stroke [27]. A multi-disciplinary approach to these patients is also a concept we are embracing. The concept of the ‘‘Endocarditis Team’’ (made up of cardiac surgeons, cardiologists, infectious
diseases physicians and, especially important in cases with cerebral emboli, neurologists) and based on the concept of the ‘‘Heart Team’’ born out of the SYNTAX trial, has been mentioned in the literature [21] and, we feel, is central to a successful outcome in these complex cases. This paper was written using data from a decade’s experience of surgically managed infective endocarditis. These patients were predominantly managed along the traditional paradigm rather than with very early surgical intervention. But this policy has been modified over that time to adopt a more aggressive surgical approach with earlier intervention. We hoped to identify factors in these patients associated with adverse peri-operative outcomes and so use the identified factors as indicators that aggressive early intervention is warranted in the future. Age over 65 years, Pre-Operative Inotropic Requirement, Uncontrolled Sepsis or Cerebral Emboli were the factors that we found to be significantly associated with increased mortality. As such, certainly in cases with patients over 65 years of age, or with pre-operative inotrope/vasopressor requirement we would encourage very early (48 hours of intravenous antibiotics. Those patients with pre-operative cerebral emboli pose the greatest dilemma and must be managed in close liaison with a neurologist to ensure the earliest possible surgical intervention. The significant association between Age >65 and Pre-Operative Cerebral Emboli and Inotropic Requirement with PostOperative ARF is unsurprising. Interestingly, neither the Causative Organism nor the Duration of Pre-Operative Antibiotic Therapy had a significant impact on any post-operative outcome, but most relevantly had no impact on rates of early recurrence or paravalvular leak. However, low numbers of early recurrence (n=2) made accurate statistical analysis difficult. However, we did record six cases of late recurrence (>3 months). Analysis of these patients revealed that the only factor significantly associated with late recurrence is a history of intravenous drug use. A previous paper, by the primary investigator in this study, on surgically treated infective endocarditis demonstrated poorer outcomes in non-streptococcal NVE [8] and advocated early aggressive surgical intervention, particularly in cases of Staph aureus endocarditis. As such, our unit approach has been to intervene earlier in Staphylococcal endocarditis and as such the similar outcomes in Staphylococcal and Streptococcal endocarditis may be due to a more aggressive approach to the former.
Conclusion This review demonstrated that our results on mortality and early post-operative outcomes compare well with other
Please cite this article in press as: Dunne B, et al. Infective Endocarditis. Heart, Lung and Circulation (2014), http://dx.doi. org/10.1016/j.hlc.2014.02.010
HLC 1533 No. of Pages 8
7
Infective
published data [15,17–19]. However, this review was undertaken to identify weaknesses in our approach and help guide our emerging policy of earlier surgical intervention by identifying risk factors for poor outcomes. As already stated, over the period of this review our unit policy has changed, guided by previous author experience, and has moved towards a policy of early surgical intervention (
![[Infective endocarditis].](/assets/img/hold.png)
