Prosthetic Valve Endocarditis: Superiority of Surgical Valve Replacement Versus Medical Therapy Only Victor L. Yu, MD, Guodong D. Fang, MD, Thomas F. Keys, MD, Alan A. Harris, MD, Layne O. Gentry, MD, Peter C. Fuchs, MD, Marilyn M. Wagener, MPH, and Edward S. Wong, MD University of Pittsburgh and VA Medical Center, Pittsburgh, Pennsylvania; The Cleveland Clinic, Cleveland, Ohio; Baylor University, Houston, Texas; Rush Presbyterian-St. Luke's Medical Center, Chicago, Illinois; St. Vincent Hospital and Medical Center, Portland, Oregon; and Hunter-McGuire VA Medical Center, Richmond, Virginia The objective of our study was to assess the long-term outcome of patients with prosthetic valve endocarditis. We used a multicenter, prospective, observational study design. Six university teaching hospitals with high vol- ume cardiothoracic surgery participated. Seventy-four patients with prosthetic valve endocarditis as defined by explicit, objective criteria were selected for participation. All patients were followed up prospectively for 1 year. Thirty-one percent and 69% had development of endo- carditis within 60 days of valve insertion ("early") and after 60 days ("late"), respectively. The most common causes were Staphylococcus epidermidis (40%), Staphylo- coccus aureus (20%), streptococcal species (18%), and aerobic gram-negative bacilli (11%). Physical signs of endocarditis (new or changing murmur, stigmata, em- I t is well-accepted that the mortality of patients with prosthetic valve endocarditis is significant, although precise data are elusive. Studies addressing outcome [1-8] have been retrospective, have been from a single institu- tion reflecting the biases and practices of that institution, have been compiled over an extended period of 5 to 16 years with ongoing changes in surgical approaches and materials, have used nonuniform definitions of outcome (mortality versus "complications"), and have used varying times for outcome assessment. Most cases in larger series of prosthetic valve endocarditis were reported before 1985 [4,8,9]. We present a prospective study of outcome in patients with prosthetic valve endocarditis. Furthermore, our study was a multicenter study over a 2- to 3-year period with uniform data collection and strict, explicit definitions of endocarditis. Material and Methods We undertook a prospective, observational study of bacte- remic patients with preexisting prosthetic heart valves. The Accepted for publication Feb 18, 1994. Address reprint requests to Dr Yu, Division of Infectious Diseases, Uni- versity of Pittsburgh Medical Center, Montefiore W-931 , Pittsburgh, PA 15213. © 1994 by The Society of Thoracic Surgeons boli) were seen in 58%. At 6 months and 12 months, mortality was 46% and 47%, respectively. Surgical re- placement of the infected valve led to significantly lower mortality (23%) as compared with medical therapy alone (56%), as assessed by both univariate and multivariate analyses (p < 0.05). Improved outcome was seen for the surgical group even when controlling for severity of illness at time of diagnosis. From these findings we conclude that accurate assessment of outcome in pros- thetic valve endocarditis requires long-term follow-up of at least 6 months following diagnosis. Surgical therapy warrants greater scrutiny; evaluation in controlled clini- cal trials is appropriate. (Ann Thorae Surg 1994;58:1073-7) details of the protocol have been described previously [10]. All patients with positive blood cultures were evaluated for the presence of a prosthetic valve. Those patients with a valve were then monitored prospectively for 1 year. Those patients classified as having "prosthetic valve endo- carditis" by the definitions detailed below constituted the study group. All treatment decisions were made by the attending physicians without intervention by the investi- gators. Hospitals This study was conducted at six university teaching hos- pitals: The Cleveland Clinic, Cleveland, OH; St. Luke's Episcopal Hospital, Houston, TX; Rush-Presbyterian-St. Luke's Medical Center, Chicago, IL; St. Vincent Hospital and Medical Center, Portland, OR; Hunter-McGuire VA Medical Center, Richmond, VA; and Presbyterian Univer- sity Hospital and VA Medical Center, Pittsburgh, P A. The study was performed for 2 to 3 years at each hospital from 1986 through 1989. Definitions Endocarditis was classified as "definitive," "presump- tive," or "suspicious." "Definitive" denoted one of the following: (1) culture or pathology (or both) proved with demonstration of vegetations or isolation of the organism 0003-4975/94/$7.00
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Prosthetic valve endocarditis: Superiority of surgical valve replacement versus medical therapy onlyProsthetic Valve Endocarditis: Superiority of Surgical Valve Replacement Versus Medical Therapy Only Victor L. Yu, MD, Guodong D. Fang, MD, Thomas F. Keys, MD, Alan A. Harris, MD, Layne O. Gentry, MD, Peter C. Fuchs, MD, Marilyn M. Wagener, MPH, and Edward S. Wong, MD University of Pittsburgh and VA Medical Center, Pittsburgh, Pennsylvania; The Cleveland Clinic, Cleveland, Ohio; Baylor University, Houston, Texas; Rush Presbyterian-St. Luke's Medical Center, Chicago, Illinois; St. Vincent Hospital and Medical Center, Portland, Oregon; and Hunter-McGuire VA Medical Center, Richmond, Virginia
The objective of our study was to assess the long-term outcome of patients with prosthetic valve endocarditis. We used a multicenter, prospective, observational study design. Six university teaching hospitals with high vol ume cardiothoracic surgery participated. Seventy-four patients with prosthetic valve endocarditis as defined by explicit, objective criteria were selected for participation. All patients were followed up prospectively for 1 year. Thirty-one percent and 69% had development of endo carditis within 60 days of valve insertion ("early") and after 60 days ("late"), respectively. The most common causes were Staphylococcus epidermidis (40%), Staphylo coccus aureus (20%), streptococcal species (18%), and aerobic gram-negative bacilli (11%). Physical signs of endocarditis (new or changing murmur, stigmata, em-
I t is well-accepted that the mortality of patients with prosthetic valve endocarditis is significant, although
precise data are elusive. Studies addressing outcome [1-8] have been retrospective, have been from a single institu tion reflecting the biases and practices of that institution, have been compiled over an extended period of 5 to 16 years with ongoing changes in surgical approaches and materials, have used nonuniform definitions of outcome (mortality versus "complications"), and have used varying times for outcome assessment. Most cases in larger series of prosthetic valve endocarditis were reported before 1985 [4,8,9].
We present a prospective study of outcome in patients with prosthetic valve endocarditis. Furthermore, our study was a multicenter study over a 2- to 3-year period with uniform data collection and strict, explicit definitions of endocarditis.
Material and Methods
We undertook a prospective, observational study of bacte remic patients with preexisting prosthetic heart valves. The
Accepted for publication Feb 18, 1994.
Address reprint requests to Dr Yu, Division of Infectious Diseases, Uni versity of Pittsburgh Medical Center, Montefiore W-931 , Pittsburgh, PA 15213.
© 1994 by The Society of Thoracic Surgeons
boli) were seen in 58%. At 6 months and 12 months, mortality was 46% and 47%, respectively. Surgical re placement of the infected valve led to significantly lower mortality (23%) as compared with medical therapy alone (56%), as assessed by both univariate and multivariate analyses (p < 0.05). Improved outcome was seen for the surgical group even when controlling for severity of illness at time of diagnosis. From these findings we conclude that accurate assessment of outcome in pros thetic valve endocarditis requires long-term follow-up of at least 6 months following diagnosis. Surgical therapy warrants greater scrutiny; evaluation in controlled clini cal trials is appropriate.
(Ann Thorae Surg 1994;58:1073-7)
details of the protocol have been described previously [10]. All patients with positive blood cultures were evaluated for the presence of a prosthetic valve. Those patients with a valve were then monitored prospectively for 1 year. Those patients classified as having "prosthetic valve endo carditis" by the definitions detailed below constituted the study group. All treatment decisions were made by the attending physicians without intervention by the investi gators.
Hospitals This study was conducted at six university teaching hos pitals: The Cleveland Clinic, Cleveland, OH; St. Luke's Episcopal Hospital, Houston, TX; Rush-Presbyterian-St. Luke's Medical Center, Chicago, IL; St. Vincent Hospital and Medical Center, Portland, OR; Hunter-McGuire VA Medical Center, Richmond, VA; and Presbyterian Univer sity Hospital and VA Medical Center, Pittsburgh, PA. The study was performed for 2 to 3 years at each hospital from 1986 through 1989.
Definitions Endocarditis was classified as "definitive," "presump tive," or "suspicious." "Definitive" denoted one of the following: (1) culture or pathology (or both) proved with demonstration of vegetations or isolation of the organism
0003-4975/94/$7.00
1074 YU ET AL PROSTHETIC VALVE ENDOCARDITIS
(or both) from the prosthetic valve (taken at autopsy or open heart operation); (2) presence of vegetations as doc umented by echocardiogram. "Presumptive" included these with one of the following signs: (1) peripheral embolic signs (Osler nodes, Janeway lesions, Roth spot, petechiae, splenomegaly, or acute cerebrovascular acci dent); (2) new or changing murmur. "Suspicious" was defined as (l) sustained bacteremia as defined by 3/3 positive or at least 4 positive blood cultures within 72 hours, plus (2) no portal of entry identified.
Demonstration of vegetations by transthoracic echocar diography in a bacteremia patient with a prosthetic valve was arbitrarily considered as definitive evidence for endo carditis; transesophageal echocardiography was not rou tinely performed at the time of study. The above criteria are strict, especially when compared with these of Von Reyn and associates [11].
Severity of illness was assessed for each patient at the time of bacteremia using an index based on mental status, vital signs, the need for respiratory support, and the occurrence of cardiac arrest; this index has been found to be highly predictive of outcome in patients with bactere mia [12-14]. A score of 4+ or greater denotes critical illness.
All patients were prospectively monitored by review of clinic visits or follow-up by telephone. The end points for follow-up were survival, absence of signs of infection, and stable function of the prosthetic valve. Outcome was assessed at 30 days, 60 days, 6 months, and 12 months after bacteremia first was detected.
Clinical and laboratory data for analysis were entered into a computer database (Prophet Systems, Division of Research Resources, National Institutes of Health, Be thesda, MD). Categorial data were analyzed using Fisher's exact test. Continuous variables were compared using the t test or Mann-Whitney U test. A multiple regression model was also used to evaluate the mortality risks. Factors used in the regression model were those found to be significant by univariate analysis as well as factors hypothesized to have a serious effect on outcome. A Kaplan-Meirer probability plot was calculated for survival analysis. The beginning point was date of positive blood culture and the end point was date of death or 1 year follow-up. The Mantel-Cox log-rank test was used to compare the two survival curves.
Results
Seventy-four cases of prosthetic valve endocarditis were identified. Patient age ranged from 23 to 85 years (mean, 60; median, 63 years). 57% (42/74) were classified as definitive, 30% (22/74) as presumptive, and 13% (10/74) as suspicious. Twenty-three patients had vegetations detect able by transthoracic echocardiography; 12 patients with positive echocardiograms underwent valvular operation or autopsy, and all proved to have endocarditis by culture and histopathologic examination. In 5 patients other evi dence was present for endocarditis and in 6 patients echocardiography was the sole criterion used in diagnosis of endocarditis at the time bacteremia was discovered.
Ann Thorac Surg 1994;58:1073-7
Table 1. Early (within 60 Postoperative Days) Versus Late (More than 60 Postoperative Days) Endocarditis
Type of Endocarditis
Characteristics of Early (%) Late (%) Endocarditis (n = 23) (n = 51) P Value
Clinical signs Emboli 400/23) 20/51) NS Stigmata 13 (3/23) 2402/51) Splenomegaly 40/23) 6 (3/51) New or changing 30 (7/23) 31 06/51)
murmur Organism"
Staphylococcus 13 (3/23) 24 (12/51) NS aureus
Staphylococcus 57 (13/23) 3307/51) NS (0.08) epidermidis
Enterococcus faecalis 17 (4/23) 6 (3/51) NS (0.19) Streptococcus 40/23) 2402/51) NS (0.05)
species Gram-negative 22 (5/23) 6 (3/51) NS (0.10)
bacilli Candida species 13 (3/23) 4 (2/51) NS (0.17) Polymicrobial 22 (5/23) 4 (2/51) 0.03
Mortality' Survived 35 (8/23) 63 (32/51) 0.04 Died 65 OS/23) 3709/51)
a p Value obtained using Fisher's exact test (two-tailed). b Polymicrobial organisms included. 'Six-month and 12-month mortality.
NS = not significant (p ~ 0.05; if P < 0.20, exact values are given in parentheses).
Valve Type There were 76 valves in 74 patients. One patient had two mechanical valves and 1 patient had a mechanical and a biological valve. Of the 41 biological valves, 29 were porcine, (9 were Hancock, 7 were Carpentier Edwards, and 13 were unspecified), 7 were bovine Ionescu-Shiley valves, and there were 5 "other." Of the 32 mechanical valves, 17 were Starr-Edwards, 7 were St. Jude, 3 were Bjork-Shiley, and 5 were "other." Three valves were of an unknown type.
Physical signs of endocarditis (embolic event, stigmata, new or changing murmur) were documented at the time of bacteremia or during the hospital course in 58% (43/74) of patients. They included peripheral embolism (cerebral emboli or cerebrovascular accident) in 3% (2/74), stigmata (Osler's nodes, Janeway lesions, Roth spot, or petechiae) in 20% (15/74), splenomegaly in 5% (4/74), and new or changing murmur in 31% (23/74).
Prosthetic valve endocarditis was arbitrarily classified as early when it occurred within 60 days of valve insertion or late when valve insertion occurred more than 60 days previously [15]. Thirty-one percent (23/74) were classified as early and 69% (51/74) as late (Table 1). Streptococcal species (but not enterococcus) occurred more frequently in cases of late endocarditis (Table 1). Calderwood and co workers [16] have argued that early endocarditis would be more accurately defined at a 1 year breakpoint from date of
Ann Thorac Surg 1994;58:1073-7
Late endocarditis
Early endocarditis
Days post bacteremia
a Evaluated using Fisher's exact test (2-tailed). b Polymicrobial organ isms were included, but three miscellaneous organisms are not. C Ex cluding Enterococcus. d Includes 1 patient with both mechanical and biological valve.
Early = onset of endocarditis within 60 days of valve insertion; Late =
onset of endocarditis 60 days or longer after valve insertion; NE = not evaluable; NS = not significant (if p < 0.20, the exact value is given in parentheses).
the investigators. Patients in the two treatment groups were comparable as assessed by the following parameters: peripheral emboli, stigmata, new or changing murmur, microbial cause, portal of entry, and valve type (data not shown). For patients who survived, the duration of antibi otic therapy ranged from 4 to 12 weeks. Thirty percent (22/74) also underwent operation to replace the infected valve. Surgical replacement of the valve was significantly associated with improved outcome (23% mortality) com pared with medical therapy only (56% mortality; p < 0.01; Table 2 and Fig 2). The surgical group was less ill than the medical only group (mean degree ill for surgical therapy versus medical therapy only, 1.0 and 1.9, respectively). Moreover, when outcome was stratified by degree of illness, surgical therapy remained superior to medical therapy for those patients not critically ill (Table 2). Mor tality was lower for the surgical group when subclassified by microbial agent; however, sample sizes were small (Table 2). Of 15 patients with S aureus endocarditis, mor tality was 0% (0/4) with operation versus 82% (9/11) with medical therapy only (p = 0.01; Table 2).
Table 2. Mortality (6 months) of Patients with Prosthetic Valve Endocarditis Receiving Surgical Versus Medical Therapy Only Assessed by Risk Factor
Therapy
All pa tients 23 (5/22) 56 (29/52) 0.01
Demographics 1-3+ ill 28 (5/22) 51 (24/47) 0.04 4+ ill 100 (5/5) NE
Early 50 (4/8) 73(11/15) NS
Late 70/14) 4908/37) 0.01 Organism"
Staphylococcus aureus 0(0/4) 82(9/11) 0.01 Staphylococcus 33 (4/12) 5600/18) NS
epidermidis Streptococcus species" 0(0/3) 40 (4/10) NS Enterococcus faecalis 43 (3/7) NE Gram-negative bacilli 500/2) 91 00/11) NS
or Candida Polymicrobial 100 (7/7) NE
Valve typed Mechanical 140/7) 63 OS/24) 0.04 Biological 25 (4/16) 4802/25) NS (0.19)
Valve location Mitral 27 (3/11) 5404/26) NS (0.17)
Aortic 18 (3/17) 54 OS/28) 0.03
1
0.9
0.8
0.7
0.2
0.1
O-'-"'--"'---'----''--'----i.--'---'---'--'-~......._'__-'--"'--"'---.......... o
Microbial Cause For the purposes of this study, Staphylococcus classified as coagulase-negative were considered as Staphylococcus epi dermidis. Organisms included Staphylococcus epidermidis, 40% (30/74); Staphylococcus aureus, 20% (15/74); Enterococ cus faecalis, 9% (7/74); viridans streptococci, 7% (5/74); other streptococcal species, 14% 00/74); Pseudomonas aeu ruginosa, 3% (2/74); other aerobic gram-negative bacilli, 5% (4/74); miscellaneous bacteria, 4% (3/74); and Candida albicans, 7% (5/74). The total is greater than 100% because 9% (7/74) were polymicrobial.
Outcome After diagnosis, follow-up information was available for all 74 patients. Mortality was 31% (23/74) 30 days after onset of bacteremia, 41% (30/74) by 60 days, 46% (34/74) by 6 months, and 47% (35/74) by 12 months. Of the 35 deaths at 12 months, 31 were directly attributable to infection.
valve replacement. Fifty-one percent (38/74) and 49% (36/74) were early and late, respectively, by Calderwood's definition; more cases due to Staphylococcus epidermidis would have been included in the early group (55%; 21/38) with this definition.
Surgical Versus Medical Therapy Only All patients received appropriate antimicrobial agent ther apy as defined by in vitro susceptibility tests and review by
UNIVARIATE ANALYSIS. Univariate analyses showed that mortality at 6 months was significantly associated with a polymicrobial cause 000%; 7/7) versus a single microbial cause (40%; 27/67; P = 0.003), early endocarditis (65%; 15/23) versus late endocarditis (37%; 19/51; P = 0.04; Fig 1), and increasing severity of illness ("4+" ill) 000%; 5/5) versus non- "4+" ill (42%; 29/69; P = 0.02). Nonsignificant correlations for mortality at 6 months included portal of entry, valve location, single versus multiple valves, biolog ical versus mechanical valves, and immunosuppression (data not shown).
Fig 1. Kaplan-Meier curves comparing survival of patients with early onset (:560 days after valve replacement) and late onset (>60 days) of prosthetic valve endocarditis (p = 0.02 by Mantel-Cox test).
MULTIVARIATE ANALYSIS. In addition to surgical therapy, the following other factors were entered into the stepwise
1076 YU ET AL PROSTHETIC VALVE ENDOCARDITIS
Ann Thorac Surg 1994;58:1073--7
whether there have been notable improvements in mortal ity [18].
On the other hand, surgical replacement of the affected valve with newer techniques and improved graft materials appears to offer an improved prognosis, but data address ing this point are largely circumstantial. The possible superiority of surgical replacement over medical therapy only for prosthetic valve endocarditis is controversial [19]. Retrospective studies are clouded by the potential for bias; perhaps less ill patients are selected for surgical therapy or conversely, patients who fail to respond to medical therapy are more likely to undergo operation.
The overall mortality at 6 months in our study was 46% (34/74). Of the 74 patients, 30% (22/74) underwent valve replacement. Valve replacement was significantly associ ated with a favorable outcome at 6 months with a 23% (5/22) mortality compared with a 56% (29/52) mortality rate for those treated with medical therapy only (p < OJll, Table 2 and Fig 2). This favorable trend for surgical therapy, as assessed by both univariate and multivariate analysis, was maintained even for high-risk subgroups, various microbial causes, and severity of illness (Table 2).
Based on our previous study, in which duration of antibiotic therapy played little role in preventing pros thetic valve endocarditis [10], we believe that the duration of antibiotic therapy preoperatively or postoperatively is not a critical factor in improving outcome, although con siderable attention has been directed at this issue [20].
Based on the results of this prospective study, we suggest that more patients with prosthetic valve endocar ditis should be considered for surgical therapy. The cur rent indications for operation include valve dysfunction, congestive heart failure, invasive infection, persistent bac teremia, early endocarditis, and more virulent organisms, especially fungi and gram-negative bacilli [18]. We point out that onset of congestive heart failure, valvular dysfunc tion, and conduction abnormalities represent the progres sion of infection to such a degree that valvular replacement becomes technically more difficult at these points. The advantages of earlier operation would be the minimization of destruction of the annulus, prevention of burrowing of the organism through the myocardium, and prevention of hemorrhagic complications from peripheral emboli. The superiority of surgical over medical therapy for prosthetic valve endocarditis also has been suggested in numerous retrospective studies [2-4,7, 16,21-26].
What are the weaknesses of our study? First, our study was an observational study and allocation of therapies was not randomized. Thus, bias in selection of therapies was present. For example, no critically-ill patient (4+ ill or greater in Table 2) underwent operation. Yet it should be noted that all of these patients died receiving medical therapy only and that surgical therapy was more success ful than medical therapy alone in those less severely ill (l to 3+ ill in Table 2). Surgical therapy also was more successful than medical therapy as assessed by multivari ate analyses. We also analyzed therapy for SUbgroups of patients with risk factors for poor outcome, and surgical therapy remained consistently superior to medical therapy for all subgroups (Table 2).
365300
Days post bacteremia
Fig 2. Comparison of Kaplan-Meier curves shows that the survival of patients treated with surgical valve replacement was significantly bet ter than that of patients treated with medical therapy only (p = 0.01 by Mantel-Cox test).
logistic regression analysis in various permutations be cause they had been identified by univariate analysis as potentially significant in affecting outcome: polymicrobial cause (yes, no), severity of illness (3+ ill or less, 4+ ill), staphylococcal disease (yes, no), valvular location (espe cially aortic), composition of valve (mechanical, biological), timing of onset (early, late), and therapy (surgical, medical therapy only). The significant factors adversely affecting 6-month outcome were early onset (odds ratio 1.4; 95% confidence interval, 1.1 to 1.8; P = 0.003), 4+ severity of illness (odds ratio 1.6; 95% confidence interval, 1.1 to 2.4; P = 0.03), and medical therapy only (odds ratio 1.4; 95% confidence interval, 1.1 to 1.7; P = 0.01).
Assessed at 30 days or 60 days, the mortality for surgical therapy was 14% (3/22) and 18% (4/22), respectively, whereas the mortality for medical therapy only was 38% (20/52) and 50% (26/52), respectively. The superiority of surgical therapy remained statistically significant at both 30 days and 60 days (p < 0.05). Outcome assessed at 12 months was essentially identical to that at 6 months.
Comment
We prospectively followed up 74 patients with prosthetic valve endocarditis as defined by strict, objective criteria. The fact that follow-up was maintained for 1 year in our prospective study was important and revealing in that complications and mortality attributable to endocarditis often became manifest as late as 6 months after discharge. Thus, we recommend that assessment of therapy for pros thetic valve endocarditis should be continued for at least 6 months after hospital discharge.
The short-term mortality rate for prosthetic valve endo carditis is high (about 50% to 65%) [8, 17]. In an attempt to improve prognosis, more potent antibiotic agents, antibi otic prophylaxis during valve replacement, and an in creased duration of antibiotic therapy have been em ployed. However, despite these measures, it is unclear
Ann Thorae Surg 1994;58:1073-7
Second, we failed to assess, in a standardized fashion, the cardiac status at the time of discovery of bacteremia. All patients who underwent operation fulfilled the classic indications for surgical valvular replacement. However, because there was no objective evaluation of cardiac status, we were unable to precisely delineate how many patients receiving medical therapy only would have fulfilled the same criteria.
Assessment of cardiac status by New York Heart Asso ciation classification, results of transesophageal echocardi ography, and hemodynamic data obtained by cardiac catheterization would seem reasonable. We concede that development of uniform, objective criteria would require a major effort; nevertheless, our study suggests that this would be a worthwhile undertaking. This issue should be addressed by a consensus committee before embarking on future prospective studies of prosthetic valve endocarditis.
In conclusion, we raise the suggestion that current antimicrobial agent therapy has achieved its maximal benefit. We also suggest that an immediate improvement in mortality for prosthetic valve endocarditis might be feasible by liberalizing current indications for surgical therapy at earlier stages of infection. This hypothesis requires confirmation in controlled clinical trials.…
The objective of our study was to assess the long-term outcome of patients with prosthetic valve endocarditis. We used a multicenter, prospective, observational study design. Six university teaching hospitals with high vol ume cardiothoracic surgery participated. Seventy-four patients with prosthetic valve endocarditis as defined by explicit, objective criteria were selected for participation. All patients were followed up prospectively for 1 year. Thirty-one percent and 69% had development of endo carditis within 60 days of valve insertion ("early") and after 60 days ("late"), respectively. The most common causes were Staphylococcus epidermidis (40%), Staphylo coccus aureus (20%), streptococcal species (18%), and aerobic gram-negative bacilli (11%). Physical signs of endocarditis (new or changing murmur, stigmata, em-
I t is well-accepted that the mortality of patients with prosthetic valve endocarditis is significant, although
precise data are elusive. Studies addressing outcome [1-8] have been retrospective, have been from a single institu tion reflecting the biases and practices of that institution, have been compiled over an extended period of 5 to 16 years with ongoing changes in surgical approaches and materials, have used nonuniform definitions of outcome (mortality versus "complications"), and have used varying times for outcome assessment. Most cases in larger series of prosthetic valve endocarditis were reported before 1985 [4,8,9].
We present a prospective study of outcome in patients with prosthetic valve endocarditis. Furthermore, our study was a multicenter study over a 2- to 3-year period with uniform data collection and strict, explicit definitions of endocarditis.
Material and Methods
We undertook a prospective, observational study of bacte remic patients with preexisting prosthetic heart valves. The
Accepted for publication Feb 18, 1994.
Address reprint requests to Dr Yu, Division of Infectious Diseases, Uni versity of Pittsburgh Medical Center, Montefiore W-931 , Pittsburgh, PA 15213.
© 1994 by The Society of Thoracic Surgeons
boli) were seen in 58%. At 6 months and 12 months, mortality was 46% and 47%, respectively. Surgical re placement of the infected valve led to significantly lower mortality (23%) as compared with medical therapy alone (56%), as assessed by both univariate and multivariate analyses (p < 0.05). Improved outcome was seen for the surgical group even when controlling for severity of illness at time of diagnosis. From these findings we conclude that accurate assessment of outcome in pros thetic valve endocarditis requires long-term follow-up of at least 6 months following diagnosis. Surgical therapy warrants greater scrutiny; evaluation in controlled clini cal trials is appropriate.
(Ann Thorae Surg 1994;58:1073-7)
details of the protocol have been described previously [10]. All patients with positive blood cultures were evaluated for the presence of a prosthetic valve. Those patients with a valve were then monitored prospectively for 1 year. Those patients classified as having "prosthetic valve endo carditis" by the definitions detailed below constituted the study group. All treatment decisions were made by the attending physicians without intervention by the investi gators.
Hospitals This study was conducted at six university teaching hos pitals: The Cleveland Clinic, Cleveland, OH; St. Luke's Episcopal Hospital, Houston, TX; Rush-Presbyterian-St. Luke's Medical Center, Chicago, IL; St. Vincent Hospital and Medical Center, Portland, OR; Hunter-McGuire VA Medical Center, Richmond, VA; and Presbyterian Univer sity Hospital and VA Medical Center, Pittsburgh, PA. The study was performed for 2 to 3 years at each hospital from 1986 through 1989.
Definitions Endocarditis was classified as "definitive," "presump tive," or "suspicious." "Definitive" denoted one of the following: (1) culture or pathology (or both) proved with demonstration of vegetations or isolation of the organism
0003-4975/94/$7.00
1074 YU ET AL PROSTHETIC VALVE ENDOCARDITIS
(or both) from the prosthetic valve (taken at autopsy or open heart operation); (2) presence of vegetations as doc umented by echocardiogram. "Presumptive" included these with one of the following signs: (1) peripheral embolic signs (Osler nodes, Janeway lesions, Roth spot, petechiae, splenomegaly, or acute cerebrovascular acci dent); (2) new or changing murmur. "Suspicious" was defined as (l) sustained bacteremia as defined by 3/3 positive or at least 4 positive blood cultures within 72 hours, plus (2) no portal of entry identified.
Demonstration of vegetations by transthoracic echocar diography in a bacteremia patient with a prosthetic valve was arbitrarily considered as definitive evidence for endo carditis; transesophageal echocardiography was not rou tinely performed at the time of study. The above criteria are strict, especially when compared with these of Von Reyn and associates [11].
Severity of illness was assessed for each patient at the time of bacteremia using an index based on mental status, vital signs, the need for respiratory support, and the occurrence of cardiac arrest; this index has been found to be highly predictive of outcome in patients with bactere mia [12-14]. A score of 4+ or greater denotes critical illness.
All patients were prospectively monitored by review of clinic visits or follow-up by telephone. The end points for follow-up were survival, absence of signs of infection, and stable function of the prosthetic valve. Outcome was assessed at 30 days, 60 days, 6 months, and 12 months after bacteremia first was detected.
Clinical and laboratory data for analysis were entered into a computer database (Prophet Systems, Division of Research Resources, National Institutes of Health, Be thesda, MD). Categorial data were analyzed using Fisher's exact test. Continuous variables were compared using the t test or Mann-Whitney U test. A multiple regression model was also used to evaluate the mortality risks. Factors used in the regression model were those found to be significant by univariate analysis as well as factors hypothesized to have a serious effect on outcome. A Kaplan-Meirer probability plot was calculated for survival analysis. The beginning point was date of positive blood culture and the end point was date of death or 1 year follow-up. The Mantel-Cox log-rank test was used to compare the two survival curves.
Results
Seventy-four cases of prosthetic valve endocarditis were identified. Patient age ranged from 23 to 85 years (mean, 60; median, 63 years). 57% (42/74) were classified as definitive, 30% (22/74) as presumptive, and 13% (10/74) as suspicious. Twenty-three patients had vegetations detect able by transthoracic echocardiography; 12 patients with positive echocardiograms underwent valvular operation or autopsy, and all proved to have endocarditis by culture and histopathologic examination. In 5 patients other evi dence was present for endocarditis and in 6 patients echocardiography was the sole criterion used in diagnosis of endocarditis at the time bacteremia was discovered.
Ann Thorac Surg 1994;58:1073-7
Table 1. Early (within 60 Postoperative Days) Versus Late (More than 60 Postoperative Days) Endocarditis
Type of Endocarditis
Characteristics of Early (%) Late (%) Endocarditis (n = 23) (n = 51) P Value
Clinical signs Emboli 400/23) 20/51) NS Stigmata 13 (3/23) 2402/51) Splenomegaly 40/23) 6 (3/51) New or changing 30 (7/23) 31 06/51)
murmur Organism"
Staphylococcus 13 (3/23) 24 (12/51) NS aureus
Staphylococcus 57 (13/23) 3307/51) NS (0.08) epidermidis
Enterococcus faecalis 17 (4/23) 6 (3/51) NS (0.19) Streptococcus 40/23) 2402/51) NS (0.05)
species Gram-negative 22 (5/23) 6 (3/51) NS (0.10)
bacilli Candida species 13 (3/23) 4 (2/51) NS (0.17) Polymicrobial 22 (5/23) 4 (2/51) 0.03
Mortality' Survived 35 (8/23) 63 (32/51) 0.04 Died 65 OS/23) 3709/51)
a p Value obtained using Fisher's exact test (two-tailed). b Polymicrobial organisms included. 'Six-month and 12-month mortality.
NS = not significant (p ~ 0.05; if P < 0.20, exact values are given in parentheses).
Valve Type There were 76 valves in 74 patients. One patient had two mechanical valves and 1 patient had a mechanical and a biological valve. Of the 41 biological valves, 29 were porcine, (9 were Hancock, 7 were Carpentier Edwards, and 13 were unspecified), 7 were bovine Ionescu-Shiley valves, and there were 5 "other." Of the 32 mechanical valves, 17 were Starr-Edwards, 7 were St. Jude, 3 were Bjork-Shiley, and 5 were "other." Three valves were of an unknown type.
Physical signs of endocarditis (embolic event, stigmata, new or changing murmur) were documented at the time of bacteremia or during the hospital course in 58% (43/74) of patients. They included peripheral embolism (cerebral emboli or cerebrovascular accident) in 3% (2/74), stigmata (Osler's nodes, Janeway lesions, Roth spot, or petechiae) in 20% (15/74), splenomegaly in 5% (4/74), and new or changing murmur in 31% (23/74).
Prosthetic valve endocarditis was arbitrarily classified as early when it occurred within 60 days of valve insertion or late when valve insertion occurred more than 60 days previously [15]. Thirty-one percent (23/74) were classified as early and 69% (51/74) as late (Table 1). Streptococcal species (but not enterococcus) occurred more frequently in cases of late endocarditis (Table 1). Calderwood and co workers [16] have argued that early endocarditis would be more accurately defined at a 1 year breakpoint from date of
Ann Thorac Surg 1994;58:1073-7
Late endocarditis
Early endocarditis
Days post bacteremia
a Evaluated using Fisher's exact test (2-tailed). b Polymicrobial organ isms were included, but three miscellaneous organisms are not. C Ex cluding Enterococcus. d Includes 1 patient with both mechanical and biological valve.
Early = onset of endocarditis within 60 days of valve insertion; Late =
onset of endocarditis 60 days or longer after valve insertion; NE = not evaluable; NS = not significant (if p < 0.20, the exact value is given in parentheses).
the investigators. Patients in the two treatment groups were comparable as assessed by the following parameters: peripheral emboli, stigmata, new or changing murmur, microbial cause, portal of entry, and valve type (data not shown). For patients who survived, the duration of antibi otic therapy ranged from 4 to 12 weeks. Thirty percent (22/74) also underwent operation to replace the infected valve. Surgical replacement of the valve was significantly associated with improved outcome (23% mortality) com pared with medical therapy only (56% mortality; p < 0.01; Table 2 and Fig 2). The surgical group was less ill than the medical only group (mean degree ill for surgical therapy versus medical therapy only, 1.0 and 1.9, respectively). Moreover, when outcome was stratified by degree of illness, surgical therapy remained superior to medical therapy for those patients not critically ill (Table 2). Mor tality was lower for the surgical group when subclassified by microbial agent; however, sample sizes were small (Table 2). Of 15 patients with S aureus endocarditis, mor tality was 0% (0/4) with operation versus 82% (9/11) with medical therapy only (p = 0.01; Table 2).
Table 2. Mortality (6 months) of Patients with Prosthetic Valve Endocarditis Receiving Surgical Versus Medical Therapy Only Assessed by Risk Factor
Therapy
All pa tients 23 (5/22) 56 (29/52) 0.01
Demographics 1-3+ ill 28 (5/22) 51 (24/47) 0.04 4+ ill 100 (5/5) NE
Early 50 (4/8) 73(11/15) NS
Late 70/14) 4908/37) 0.01 Organism"
Staphylococcus aureus 0(0/4) 82(9/11) 0.01 Staphylococcus 33 (4/12) 5600/18) NS
epidermidis Streptococcus species" 0(0/3) 40 (4/10) NS Enterococcus faecalis 43 (3/7) NE Gram-negative bacilli 500/2) 91 00/11) NS
or Candida Polymicrobial 100 (7/7) NE
Valve typed Mechanical 140/7) 63 OS/24) 0.04 Biological 25 (4/16) 4802/25) NS (0.19)
Valve location Mitral 27 (3/11) 5404/26) NS (0.17)
Aortic 18 (3/17) 54 OS/28) 0.03
1
0.9
0.8
0.7
0.2
0.1
O-'-"'--"'---'----''--'----i.--'---'---'--'-~......._'__-'--"'--"'---.......... o
Microbial Cause For the purposes of this study, Staphylococcus classified as coagulase-negative were considered as Staphylococcus epi dermidis. Organisms included Staphylococcus epidermidis, 40% (30/74); Staphylococcus aureus, 20% (15/74); Enterococ cus faecalis, 9% (7/74); viridans streptococci, 7% (5/74); other streptococcal species, 14% 00/74); Pseudomonas aeu ruginosa, 3% (2/74); other aerobic gram-negative bacilli, 5% (4/74); miscellaneous bacteria, 4% (3/74); and Candida albicans, 7% (5/74). The total is greater than 100% because 9% (7/74) were polymicrobial.
Outcome After diagnosis, follow-up information was available for all 74 patients. Mortality was 31% (23/74) 30 days after onset of bacteremia, 41% (30/74) by 60 days, 46% (34/74) by 6 months, and 47% (35/74) by 12 months. Of the 35 deaths at 12 months, 31 were directly attributable to infection.
valve replacement. Fifty-one percent (38/74) and 49% (36/74) were early and late, respectively, by Calderwood's definition; more cases due to Staphylococcus epidermidis would have been included in the early group (55%; 21/38) with this definition.
Surgical Versus Medical Therapy Only All patients received appropriate antimicrobial agent ther apy as defined by in vitro susceptibility tests and review by
UNIVARIATE ANALYSIS. Univariate analyses showed that mortality at 6 months was significantly associated with a polymicrobial cause 000%; 7/7) versus a single microbial cause (40%; 27/67; P = 0.003), early endocarditis (65%; 15/23) versus late endocarditis (37%; 19/51; P = 0.04; Fig 1), and increasing severity of illness ("4+" ill) 000%; 5/5) versus non- "4+" ill (42%; 29/69; P = 0.02). Nonsignificant correlations for mortality at 6 months included portal of entry, valve location, single versus multiple valves, biolog ical versus mechanical valves, and immunosuppression (data not shown).
Fig 1. Kaplan-Meier curves comparing survival of patients with early onset (:560 days after valve replacement) and late onset (>60 days) of prosthetic valve endocarditis (p = 0.02 by Mantel-Cox test).
MULTIVARIATE ANALYSIS. In addition to surgical therapy, the following other factors were entered into the stepwise
1076 YU ET AL PROSTHETIC VALVE ENDOCARDITIS
Ann Thorac Surg 1994;58:1073--7
whether there have been notable improvements in mortal ity [18].
On the other hand, surgical replacement of the affected valve with newer techniques and improved graft materials appears to offer an improved prognosis, but data address ing this point are largely circumstantial. The possible superiority of surgical replacement over medical therapy only for prosthetic valve endocarditis is controversial [19]. Retrospective studies are clouded by the potential for bias; perhaps less ill patients are selected for surgical therapy or conversely, patients who fail to respond to medical therapy are more likely to undergo operation.
The overall mortality at 6 months in our study was 46% (34/74). Of the 74 patients, 30% (22/74) underwent valve replacement. Valve replacement was significantly associ ated with a favorable outcome at 6 months with a 23% (5/22) mortality compared with a 56% (29/52) mortality rate for those treated with medical therapy only (p < OJll, Table 2 and Fig 2). This favorable trend for surgical therapy, as assessed by both univariate and multivariate analysis, was maintained even for high-risk subgroups, various microbial causes, and severity of illness (Table 2).
Based on our previous study, in which duration of antibiotic therapy played little role in preventing pros thetic valve endocarditis [10], we believe that the duration of antibiotic therapy preoperatively or postoperatively is not a critical factor in improving outcome, although con siderable attention has been directed at this issue [20].
Based on the results of this prospective study, we suggest that more patients with prosthetic valve endocar ditis should be considered for surgical therapy. The cur rent indications for operation include valve dysfunction, congestive heart failure, invasive infection, persistent bac teremia, early endocarditis, and more virulent organisms, especially fungi and gram-negative bacilli [18]. We point out that onset of congestive heart failure, valvular dysfunc tion, and conduction abnormalities represent the progres sion of infection to such a degree that valvular replacement becomes technically more difficult at these points. The advantages of earlier operation would be the minimization of destruction of the annulus, prevention of burrowing of the organism through the myocardium, and prevention of hemorrhagic complications from peripheral emboli. The superiority of surgical over medical therapy for prosthetic valve endocarditis also has been suggested in numerous retrospective studies [2-4,7, 16,21-26].
What are the weaknesses of our study? First, our study was an observational study and allocation of therapies was not randomized. Thus, bias in selection of therapies was present. For example, no critically-ill patient (4+ ill or greater in Table 2) underwent operation. Yet it should be noted that all of these patients died receiving medical therapy only and that surgical therapy was more success ful than medical therapy alone in those less severely ill (l to 3+ ill in Table 2). Surgical therapy also was more successful than medical therapy as assessed by multivari ate analyses. We also analyzed therapy for SUbgroups of patients with risk factors for poor outcome, and surgical therapy remained consistently superior to medical therapy for all subgroups (Table 2).
365300
Days post bacteremia
Fig 2. Comparison of Kaplan-Meier curves shows that the survival of patients treated with surgical valve replacement was significantly bet ter than that of patients treated with medical therapy only (p = 0.01 by Mantel-Cox test).
logistic regression analysis in various permutations be cause they had been identified by univariate analysis as potentially significant in affecting outcome: polymicrobial cause (yes, no), severity of illness (3+ ill or less, 4+ ill), staphylococcal disease (yes, no), valvular location (espe cially aortic), composition of valve (mechanical, biological), timing of onset (early, late), and therapy (surgical, medical therapy only). The significant factors adversely affecting 6-month outcome were early onset (odds ratio 1.4; 95% confidence interval, 1.1 to 1.8; P = 0.003), 4+ severity of illness (odds ratio 1.6; 95% confidence interval, 1.1 to 2.4; P = 0.03), and medical therapy only (odds ratio 1.4; 95% confidence interval, 1.1 to 1.7; P = 0.01).
Assessed at 30 days or 60 days, the mortality for surgical therapy was 14% (3/22) and 18% (4/22), respectively, whereas the mortality for medical therapy only was 38% (20/52) and 50% (26/52), respectively. The superiority of surgical therapy remained statistically significant at both 30 days and 60 days (p < 0.05). Outcome assessed at 12 months was essentially identical to that at 6 months.
Comment
We prospectively followed up 74 patients with prosthetic valve endocarditis as defined by strict, objective criteria. The fact that follow-up was maintained for 1 year in our prospective study was important and revealing in that complications and mortality attributable to endocarditis often became manifest as late as 6 months after discharge. Thus, we recommend that assessment of therapy for pros thetic valve endocarditis should be continued for at least 6 months after hospital discharge.
The short-term mortality rate for prosthetic valve endo carditis is high (about 50% to 65%) [8, 17]. In an attempt to improve prognosis, more potent antibiotic agents, antibi otic prophylaxis during valve replacement, and an in creased duration of antibiotic therapy have been em ployed. However, despite these measures, it is unclear
Ann Thorae Surg 1994;58:1073-7
Second, we failed to assess, in a standardized fashion, the cardiac status at the time of discovery of bacteremia. All patients who underwent operation fulfilled the classic indications for surgical valvular replacement. However, because there was no objective evaluation of cardiac status, we were unable to precisely delineate how many patients receiving medical therapy only would have fulfilled the same criteria.
Assessment of cardiac status by New York Heart Asso ciation classification, results of transesophageal echocardi ography, and hemodynamic data obtained by cardiac catheterization would seem reasonable. We concede that development of uniform, objective criteria would require a major effort; nevertheless, our study suggests that this would be a worthwhile undertaking. This issue should be addressed by a consensus committee before embarking on future prospective studies of prosthetic valve endocarditis.
In conclusion, we raise the suggestion that current antimicrobial agent therapy has achieved its maximal benefit. We also suggest that an immediate improvement in mortality for prosthetic valve endocarditis might be feasible by liberalizing current indications for surgical therapy at earlier stages of infection. This hypothesis requires confirmation in controlled clinical trials.…
Related Documents
