| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
J Thorac Cardiovasc Surg 1995;110:1708-1724
© 1995 Mosby, Inc.
SURGERY FOR ACQUIRED HEART DISEASE |
Birmingham, Ala. and Brisbane, Queensland, Australia
Supported in part by a grant from The National Heart Foundation of Australia, G90B2990.
Address for reprints: David C. McGiffin, MD, University of Alabama at Birmingham, 619 19th St. South, Birmingham, AL 35294.
Abstract
Replacement valve endocarditis occurred in 3.7% of 2443 patients who underwent primary or redo aortic valve replacements at The Prince Charles Hospital between December 31, 1969, and January 1, 1992, based on a cross-sectional follow-up in 1992 which was 98.8% complete. Because some patients had re-replacements during the study period, a total of 2686 operations were considered for analysis. A variety of replacement devices were used, including 571 allografts (21%), 1152 xenografts (43%), and 880 mechanical valves (36%). Insertion of an allograft valve resulted in a constant risk of endocarditis which, by multivariable hazard function analysis, negated the effect of any early-phase risk factors (p <0.0001). With other replacement devices, the risk of infection peaked early after operation (9 weeks) and then gave way to a constant risk. Compared with the risk associated with allograft valves, constant risk was higher when the replacement device was a Carpentier-Edwards xenograft (n = 1021, p = 0.02) and lower when a St. Jude Medical mechanical valve was used (n = 505, p = 0.05). In nonallograft recipients, the presence of active preoperative endocarditis (p <0.0001) or a concomitant synthetic aortic root replacement (p = 0.0006) increased the magnitude of the early peaking risk. Regardless of replacement device, constant risk was increased in patients with renal dysfunction (p = 0.01), in younger patients (p <0.0001), and in those with active or healed preoperative endocarditis (p = 0.04). When preoperative endocarditis was caused by Staphylococcus aureus, risk was higher than when it was caused by other organisms (p = 0.04). A culture-positive postoperative wound infection was associated with increased risk of replacement valve infection (p <0.001) and when it occurred, the same organism was usually responsible (86%). Identification of patients at increased risk for replacement valve infection may lead to reduced morbidity through strategies such as selective use of replacement devices and antimicrobial prophylaxis. (J THORAC CARDIOVASC SURG 1995;110:1708-24)
Infective endocarditis is a serious and frequently fatal condition, particularly when it occurs on a valve replacement device. 
1 In patients with active endocarditis, the risk of recurrent infection may be lower when an allograft replacement valve is used. 1,2 However, in patients undergoing aortic valve replacement (AVR) for any reason, usually for reasons other than endocarditis, the prevalence of postoperative endocarditis and the impact of replacement with various modern devices has not been fully characterized. Therefore a study to determine the prevalence of replacement endocarditis and its risk factors was undertaken in a series of patients undergoing AVR for any reason, in which a variety of replacement devices were used.
PATIENTS AND METHODS
Study group
A total of 2443 patients had an isolated AVR at The Prince Charles Hospital (TPCH) from January 1, 1970, to January 1, 1992. Among these 2443 patients, 89 had their first AVR and 2 patients had their first two AVRs either before 1970 or at another institution. Among the 2443, 290 (11.9%) had a second AVR at TPCH during this period, 43 (1.8%) had a third, and 1 had a fourth, leading to a total of 2686 operations. Patients with previous mitral or tricuspid valve operations were not included.
A concomitant procedure not involving valve replacement, most frequently coronary bypass grafting, was performed in 856 (32%) of the operations (AppendixTable 1). The average patient age at first TPCH operation was 56 years (±15 years, standard deviation), and 74% of patients were male. In 243 (9.0%) of the operations the patient had previous (preoperative) endocarditis and the organisms involved are outlined in Appendix Table 2; in 92 (38% of 243) of these cases active endocarditis was an indication for operation. Additional patient characteristics are outlined in AppendixTable 3.
|
|
|
A variety of replacement devices were used (AppendixTable 4), but biologic valves predominated (xenografts in 43% of the operations, allografts in 21%), and some patients early in the experience received what would now be regarded as obsolete valves (fascia lata valves or Braunwald-Cutter prostheses, Cutter Biological, Berkeley, Calif.). Mechanical and xenograft valves were inserted with an interrupted suture technique. Allograft valves were inserted by a variety of methods, 3 including the subcoronary technique, the cylindrical technique, and as an aortic root replacement. Valve selection was at the discretion of the surgeon, but an allograft aortic valve was more likely to have been inserted in the latter part of the experience. Abscess cavities were not specifically closed unless closure was necessary for anchoring of valve prostheses.
|
Definitions
Preoperative endocarditis
Patients were considered to have had preoperative endocarditis if morphologic, histologic, or bacteriologic evidence of aortic valve infection was present at the time of their valve replacement. Such patients were subdivided into two groups. Patients with positive cultures of operative specimens or positive blood cultures in the immediate preoperative period were considered to have active preoperative endocarditis. Patients lacking positive cultures, but with morphologic evidence of previous disease, were considered to have remote endocarditis, and an attempt was made to identify from the medical record the previous episode of endocarditis and the identity of the causative organism(s).
Replacement valve endocarditis
Replacement valve endocarditis was defined as infection occurring on the AVR device. Replacement valve infection was considered to be present (1) if typical clinical symptoms of endocarditis developed, including fever, splenomegaly, peripheral skin or mucocutaneous lesions, or a regurgitant murmur, along with positive blood cultures with no obvious extracardiac source of infection, (2) if blood cultures were positive for infection with consistent findings at reoperation or autopsy, or (3) if organisms were cultured from the removed prosthetic valve at the time of reoperation. 4
Two patients who had active native valve infection at the time of AVR continued to show clinical signs of valve infection in the early postoperative period, leading to an early diagnosis of replacement valve endocarditis. In these cases, the event endocarditis was considered to have occurred on the date of the first positive blood culture in the postoperative period.
Data collection and follow-up
Information was obtained from hospital and outpatient medical records and by direct contact with the patients' families, local physicians, and cardiologists. Information was collected from January 15, 1992, through November 9, 1992.
Of the 2443 patients, 2413 (98.8%) were either known to be dead or were contacted during the follow-up interval. The remaining 30 patients were lost to follow-up at some time in their postoperative period, but only three were not contacted after hospital dismissal. The mean period from operation to most recent follow-up was 6.9 years (83 months; standard deviation ± 5 years). Seventy-five percent of patients were followed up for more than 15 months, 50% for more than 5.7 years (68 months), and 25% for more than 14 years (172 months).
Data analysis
Simple contingency tables and comparisons of means were made for all variables. Actuarial analyses were performed by the Kaplan-Meier method. 5 The overall hazard function (instantaneous risk of endocarditis) was estimated by a generic three-phase model, 6 facilitating multivariable, parametric analysis ofrisk factors within individual phases of risk. 7 A p value of 0.1 was used as the criterion for retaining a variable in the model (this level was chosen because, despite the large number of operations, there were relatively few events), but for each identified risk factor the p valve is provided, allowing for independent assessment as to its believability.
For the Kaplan-Meier estimates and for the multivariable analysis, patients were censored at the time of their death or at the time of a heart transplant (one patient). Because of the prevalence of more than one AVR, two techniques were used for patients in the study group who had multiple AVRs. In one, each patient was entered into the data set once, at the time of the initial AVR procedure at TPCH. The patient was then followed up in the usual way (intention-to-treat analysis). In other words, crossovers were simply accepted. This method had the number of patients as the N. The second method took account of the time-varying covariate of repeated AVRs by using the modulated renewal process. 8 In this process patients who underwent a second AVR were removed from the study group (censored) at the time of the second replacement and were entered into a second study group in which time zero was the time of the second replacement. Those with a third AVR were censored and entered into a third study group, and so forth. This resulted in a complete depiction of the 2686 operations among the 2443 patients. The multivariable analysis was made of the composite group of 2686 operations by a modulated renewal process, which properly evaluated the possibility that the time-varying covariate "AVR number" 2 or 3 (or more) was itself an incremental risk factor (explanatory variable). In this, patients with an initial AVR elsewhere or before this study received a value of "2" for "number of previous AVRs" in their first entry in the study.
Patients who received allograft valves were found to have only a single constant phase of hazard (lacking the early phase present in patients who received other valves). In previous studies, discrepancies of this type precluded the inclusion of all operations in a single multivariable analysis, in part because of computational intractabilities introduced by the complete absence of a hazard phase. 2 In this study, estimates when the allograft and nonallograft recipients were combined were not found to be intractable, and the absence of an early phase in the former group was represented by a large negative coefficient. The assumption was that a previously undetectable early phase of hazard was present in the allograft group. The analyses were repeated with the allograft group evaluated independently and the results were found to be similar. In this manuscript results from the combined analyses are presented.
As a means of internal validation of the multivariable model, Kaplan-Meier estimates were made of several subgroups, and solutions for the same groups of patients were computed by averaging the individual patient-specific predictions from the multivariable equation. The comparison was made in the time-related domain and also by comparing the total number of events (Appendix Table 5).
|
|
Overall prevalence
The non-risk-adjusted actuarial freedom (Fig. 1, A) from endocarditis among the 2443 patients followed up after their first AVR at TPCH was 94.6% at 10 years and 89.8% at 20 years. The parametric estimate was similar, and its corresponding hazard function had two phases (Fig. 1, B), with an early peak at 9 weeks that gave way to a constant risk by about 6 months after operation.
|
Table I).
Prereplacement endocarditis
The prevalence of replacement valve endocarditis was higher by non-risk-adjusted analysis in those with preoperative endocarditis than in those without it (Fig. 2). In risk-adjusted analyses, preoperative endocarditis was associated with increased constant-phase hazard, but early-phase risk was increased only when there was active endocarditis at the time of AVR. The constant-phase risk was not equal for all patients with preoperative endocarditis; risk was higher in patients with Staphylococcus aureus endocarditis than in those with preoperative endocarditis caused by other organisms (p = 0.05) (Fig. 3).
|
|
Table I) as a large negative coefficient in the early phase.
Replacement with other devices
In non-risk- adjusted and risk-adjusted analyses, patients who received the now obsolete fascia lata valve had a higher prevalence of postoperative endocarditis at any time after operation. Late postoperative endocarditis was less prevalent (p = 0.05) after AVR with a St. Jude Medical mechanical valve (St. Jude Medical, Inc., St. Paul, Minn.). In risk-adjusted analysis only, other valve-specific differences in risk were identified, including an increased constant-phase risk for endocarditis in patients receiving the Carpentier-Edwards xenograft valve (Baxter Healthcare Corp., Edwards Division, Santa Ana, Calif.) (see Table I).
Figs. 4A, 4B, and 4C depict the estimated freedom from replacement valve endocarditis after replacement with the three most commonly used devices in patients with differing preoperative endocarditis histories.
|
|
|
Primary versus redo AVR
When examined in a non-risk-adjusted manner, primary AVR was followed by a lower prevalence of replacement valve endocarditis than was redo AVR (Wilcoxon p = 0.03). The small group (n = 43) of patients having a second redo operation (the third AVR operation) did not have a believably higher non-risk-adjusted prevalence of postoperative endocarditis than those having their first redo operation (Wilcoxon p = 0.76). However, redo AVR was not found to be a risk factor for replacement valve endocarditis in risk-adjusted analyses. This is likely a result of the differences in risk factors (explanatory variables) among those having primary and redo operations (Appendix Table 6).
|
The association of postoperative complications to replacement valve endocarditis
The presence of a postoperative wound infection or postoperative cardiac dysfunction was associated with an increased prevalence of postoperative endocarditis (Appendix Table 7), even when adjustments were made for identified risk factors (Table II). Among the seven patients who had wound infections and later had endocarditis, the mean period from wound infection to diagnosis of endocarditis was 6 weeks. In six of the seven patients (86%) the organism cultured from the infected wound was subsequently responsible for the valve infection, and in the remaining patient endocarditis occurred many months after operation. Patients having an early reoperation for excessive bleeding (reentry) had a higher prevalence of replacement valve endocarditis, but this finding could be due to chance alone (p = 0.2).
|
|
Table III). Staphylococcal endocarditis tended to occur early after operation, whereas the risk of streptococcal infection was constant across time (Fig. 5). The risk of endocarditis caused by other organisms was similar to that for staphylococcal organisms. In a separate risk-adjusted analysis of the event replacement valve staphylococcal endocarditis,patients with preoperative endocarditis were found to be at increased risk of staphylococcal endocarditis (p < 0.001) only if their preoperative infection was caused by a staphylococcal organism (Table IV).
|
|
|
DISCUSSION
Method
This analysis is made more specific by the use of parametric, phase-specific, multivariable methods. These techniques have been used previously for the analysis of endocarditis after AVR. 9,10 In an attempt to increase clarity, but also for reasons of mathematical tractability, we have designated some characteristics as negative risk factors,instead of using positive coefficients and the absence of a characteristic. For example, we have designated use of allograft valve as a negative risk factor instead of nonuse of an allograft valve as a positive risk factor.
Patients with previous nonaortic valve operations were not included in the study group. This exclusion was an attempt to eliminate ambiguity regarding the location of postoperative endocarditis in patients with clinical evidence of valve infection. For similar reasons, we excluded those with double valve operations (concomitant non-AVRs) and censored patients at the time of any subsequent valve replacement.
This analysis does not address risk factors for mortality, reoperation, or morbidity (other than endocarditis). Admittedly, such information would be prerequisite to definitive recommendations regarding treatment strategies, such as valve selection for particular patients, because replacement valve endocarditis is a relatively uncommon cause of morbidity or mortality.
Prevalence of replacement valve endocarditis
The overall estimate of incidence of prosthetic valve endocarditis in this report is low (3.8%) and consistent with data from previous reports. 11-14 In actuarial estimates, we censored patients at the time of death, reoperation, or non-AVR. It is important to remember that the reported actuarial estimates cannot be used independently to estimate a percentage of patients in whom endocarditis will develop. 15 For example, although the actuarial occurrence of endocarditis at 20 years is 9.7%, the expected number of endocarditis cases for a large group of patients followed up for 20 years would be less than 9.7%, because some patients will die and still others will have reoperations before endocarditis develops.
Recurrent endocarditis
Previous work has shown that patients with preoperative (native or replacement) endocarditis are at increased risk for postoperative endocarditis, 11,15 and this analysis further stratifies this risk by identifying two important characteristics of patients with preoperative endocarditis: the causative organism (Staphylococcus aureus or other) and the infection status at the time of valve replacement (active or remote). These characteristics lead to four possible risk categories for those with preoperative infection: (1) active Staphylococcal aureus infection, (2) active infection with an organism other than Staphylococcus aureus,(3) remote Staphylococcus aureus infection, and (4) remote infection with an organism other than Staphylococcus aureus (see Fig. 3).
That active valve infection, but not remote, increases the early-phase risk of replacement valve infection suggests that in some patients with active preoperative endocarditis the infection persists into the postoperative period. These findings support the concept that an allograft valve is indicated in the face of active endocarditis. 2
Constant-phase risk of recurrent endocarditis was found to be increased in all patients with previous endocarditis, but the magnitude was dependent on the infecting organism. Preoperative Staphylococcus aureus infection increased the constant risk more than infection with other organisms. From a separate analysis of risk factors for staphylococcal endocarditis, we found that another staphylococcal infection tended to recur. This apparent predisposition to valve infection, most pronounced in those who had staphylococcal endocarditis, has been suggested previously. 1 Such a recurrence tendency might be expected in patients who are prone to recurrent bacteremias, such as persons who abuse intravenous drugs or those with chronic skin infections (e.g., carbunculosis). Alternatively, this tendency to recurrence might reflect subtle difference in immunologic susceptibility, 16 perhaps otherwise clinically insignificant, which predispose some patients to recurrent valve infections.
Patient-specific risk factors for replacement valve endocarditis
The identified increased risk of postoperative endocarditis in the younger valve recipient likely results from the broader environmental exposure to potential pathogens experienced by this group. In support of this hypothesis, we found a correlation between younger age and endocarditis with Coxiella burnetii (Q fever), an organism likely to infect workers in slaughter houses, usually young men. The association of renal dysfunction, commonly seen in patients with dysfunction of other organ systems, with increased constant-phase risk for endocarditis probably represents the reduced resistance to infection in this group of patients.
Operative risk factors for replacement valve endocarditis
Of particular interest to the surgeon is the possibility that choice of a valve replacement device can favorably modulate the effects of existing (patient-specific) risk factors. Our finding that those who received allograft valves had only a single constant-phase risk for replacement valve infection is consistent with previous reports. 2,11,14 This negation of early-phase risk provides the greatest potential for benefit when early risk is expected to be the highest, as in the patient with active preoperative endocarditis or the patient who requires an aortic root replacement. However, in a patient with a remote episode of endocarditis (elevated constant risk), the allograft valve is not predicted to prevent recurrent endocarditis, because its beneficial effects are confined to the early phase.
Although several previous studies of replacement valve endocarditis have grouped different valves on the basis of the predominant material in their construction (i.e., mechanical, xenograft), in our analysis this approach would have masked valve-specific differences in risk, perhaps as a result of an insufficient number of observations. Indeed, the estimated constant-phase risk for valves grouped in this manner is not believably different. However, when the risk was entered into the analysis individually, we found, in addition to differences in now "historical"mechanical valves, that those receiving the St. Jude Medical valve (n = 505) were at lower risk and those receiving Carpentier-Edwards xenograft valves (n = 1021) were at a higher constant risk. The lower constant-phase risk in the St. Jude Medical recipients was unexpected and perhaps explains why, in a previous analysis of mortality (based on many of the same patients in this study group), AVR with a St. Jude Medical valve was found to result in a lower constant-phase risk of death. 1
That aortic root replacement with a synthetic graft was associated with a greatly increased early-phase risk, but aortic root replacement with tissue (allograft aortic root replacement) was not, suggests that it is the placement of the synthetic material, not progression of the disease process alone, which predisposes patients to postoperative valve infection. With this evidence, the need to replace the aortic root would favor the use of an allograft valve with contiguous replacement of the aortic root, especially in the setting of preoperative infection.
Longer bypass time, although identified previously as a correlate for postoperative endocarditis, 1,11 was found in this analysis to be associated with increased constant-phase risk for replacement valve endocarditis, whereas the relationship of bypass time to early risk of infection, the usual experience in surgery, was weak and not believable. Longer bypass time may be a surrogate for a group of individually weak associations that indicate poor patient condition or more extensive disease, such as concomitant procedures, preoperative and postoperative organ system dysfunction, number of previous aortic valve operations, and New York Heart Association class. The increase in risk expected for patients with longer bypass time probably represents the grouped influence of several individually weak risk factors.
Association of postoperative complications to replacement valve endocarditis
Although the association of major postoperative wound infections with subsequent endocarditis was strong (large coefficient) and believable (p < 0.0001), it is unclear whether this represents a causal relationship. Regardless, the patient with a wound infection is at a greatly increased risk for subsequent diagnosis of prosthetic valve endocarditis, and when it occurs, the valve infection is usually caused by the same organism. These findings indicate that a patient who has a postoperative wound infection after AVR should receive aggressive antimicrobial therapy selected on the basis of positive cultures from the infected wound.
Postoperative cardiac dysfunction was also associated with replacement valve endocarditis. This finding may reflect the decreased immunologic tolerance of these patients or the increased bacteriologic challenge imposed by invasive monitoring and frequent diagnostic procedures.
SUMMARY
Overall, the risk of replacement aortic valve endocarditis was low. When the replacement device was an allograft valve the risk was constant; with other devices there was an initial peaking risk that gave way to a constant risk by about 6 months. Compared with the risk of replacement with an allograft valve, the constant risk of endocarditis was found to be lower when a St. Jude Medical (mechanical) valve was used and higher when replacement was with a Carpentier-Edwards (xenograft) valve. Active preoperative endocarditis increased both the early and the constant risk of recurrent infection, whereas remote endocarditis increased only the constant-phase risk. When preoperative endocarditis was caused by Staphylococcus aureus, patients were at greater risk of recurrent infection than when the preoperative infection was caused by other organisms. Patients receiving a concomitant aortic root replacement were at increased early risk of endocarditis when a synthetic graft was used. A culture-positive postoperative wound infection was associated with an increased risk of subsequent diagnosis of endocarditis; both infections were usually caused by the same organism.
Appendix: DISCUSSION
Dr. Robert B. Karp (Chicago, Ill.).
The large number of patients in this study allows the analysis to approach the concept of "truth." The incidence of replacement valve endocarditis over a 23-year period was 3.7%, with a mean of about 7 years' follow-up. This is a laudably low figure, and as the authors point out, it should not be taken as the exact figure because of statistical methods that allow censoring and reinclusion of patients.
Some of the concepts elaborated should be reiterated. The allograft neutralized the early risk factors for replacement valve endocarditis. However, what is abundantly clear is that the patient is a risk factor, particularly the patient with acute or even remote bacterial endocarditis. Constant risks were renal dysfunction, younger age, and, as noted, healed endocarditis. Disturbingly, the use of a synthetic valve conduit was a risk factor, along with the well-known factor of Staphylococcus aureus native valve endocarditis.
The authors found that wound infection was associated with replacement valve endocarditis. This poses one question: What comes first—a bacteremia caused by the mediastinitis that infects the replacement device or a bacteremia from other sources that seeds both the wound and the replacement device? As noted by the authors, one third of patients having bacterial endocarditis as a risk factor for replacement device endocarditis had active bacterial endocarditis. However, importantly, the remaining two thirds had remote or inactive endocarditis. Thus this prompts a second question. What are the rules and inferences for dealing with patients with remote endocarditis as to the most effective replacement device?
Young age was a risk factor for replacement valve endocarditis. Drug abuse occurs in the younger age groups, but this was not an issue in the Brisbane experience. There was, however, an association with young age and Coxiella burnetii or Q fever. For the North American surgeon, who is not dealing with endemic Q fever, do the authors believe young age would remain a risk factor?
These data suggest again that young patients should receive a homograft aortic valve both for the well-known advantage of anticoagulation-free, thromboembolism-free existence, but also to negate the early risk of replacement valve endocarditis.
Finally, many surgeons are not homograft enthusiasts. Can the authors speculate on the reason for the superiority of the St. Jude Medical prosthesis over the various porcine xenografts with regard to freedom from replacement valve endocarditis?
Dr. Agnihotri.
Thank you, Dr. Karp, for your questions. I will address your comments sequentially.
We did do a separate multivariable analysis in which we included some postoperative variables. Many of you may be familiar with the statistical pitfalls of including covariates that occur after time zero, which in our case was the time of AVR. When these variables are identified as risk factors, the question of causality will typically arise, as it does in the identified association of postoperative wound infection and replacement valve endocarditis. I cannot say whether the wound infection led to the valve infection or whether they were both a result of bacteremia from some third source. Nevertheless, we believe that the observation does have clinical significance. If a major wound infection develops in a valve recipient, our data do indicate an important increase in risk for a subsequent diagnosis of endocarditis, and we would take aggressive steps to prevent the appearance of valve infection.
The distinction between active and remote endocarditis as patient-specific risk factors is important. Patients with active preoperative endocarditis, defined as culture-positive valve infection at operation, were at high risk for recurrence early after operation. Patients with healed endocarditis did not have evidence of this early increase in risk. These findings, in addition to the valve-specific difference in risk, lead us to the inference that patients with active preoperative endocarditis should receive an allograft replacement device. In the patient with remote infection, early risk is not increased, and I think that factors other than risk of recurrent infection determine appropriate valve selection.
As Dr. Karp points out, one of the interesting aspects of this data set, which originated in Australia, was that there was a higher prevalence of Coxiella than one would have found in a similar American data set. Coxiella is the organism responsible for Q fever, a public health problem in Australia that is primarily seen in young men, typically workers in slaughter houses. The number of patients in our study group with preoperative Coxiella endocarditis was 21, or roughly 8% of the total group of preoperative valve infections. Among these 21 patients there were only three recurrences. These numbers are sufficiently low that I do not believe they influenced our results or conclusions.
The last question regarding why we found modern mechanical valves to have a lower long-term resistance to infection in some patients than did the allograft valve is a fascinating one. We have discussed this issue among ourselves at length. What we do know from previous work is that the allograft valve is not static over time. There are structural changes, and one could hypothesize that these could lead to disturbances of flow and surfaces properties that could increase the likelihood of bacterial adherence, an essential step in the conversion of a bacteremia into a localized valve infection. These changes do not occur in mechanical valves, a fact that may explain the observed differences in late risk.
Appendix: APPENDIXES
Appendix I. Variables entered into the multivariate analysis
Demographic variables.
Demographic variables were age at operation and gender.
Clinical variables.
Clinical variables included the following: preoperative endocarditis status (yes/no; if yes, native valve or prosthetic valve endocarditis), causative organisms in patients with preoperative endocarditis, New York Heart Association functional class immediately before operation, organ system dysfunction before operation (hepatic, pulmonary, cardiac, renal), and number of previous AVRs.
Surgical variables.
Surgical variables were as follows: date of operation, surgeon, valve replacement device, size of replacement valve, concomitant procedures, and duration of cardiopulmonary bypass.
Postoperative variables.
Postoperative variables were as follows: wound infection (with fever, pain, discharge, inflammation, and positive results of cultures), postoperative organ system dysfunction (cardiac, renal, hepatic, pulmonary), returns to the operating room for postoperative bleeding, and any postoperative urinary tract infection.
Note: Hepatic dysfunction was defined as a bilirubin level greater than or equal to 35 mmol/L, and renal dysfunction was defined as a creatinine level greater than or equal to 03 mmol/L or a urea concentration greater than or equal to 15 mmol/L (or both). Pulmonary dysfunction refers to the need for ventilation in the preoperative period or the inability to normally discontinue (wean from) ventilatory support in the postoperative period.
Appendix II. Parameter estimates from multivariable analyses.
In the multivariable analysis for replacement valve endocarditis in which demographic, clinical history, and surgical variables were entered, shaping parameters for the hazard function were as follows: 
= 2.7025, 
= 0, 
= 0.2010, m = 1 (corresponding risk factor estimates are provided in Table I). In an independent analysis for the event replacement valve endocarditis, in which demographic, clinical history, surgical, and postoperative variables were entered, shaping parameters were as follows: = 2.7025, = 0, = 0.2010, m = 1 (corresponding risk factor estimates are provided in Table II). In a multivariable analysis for the event staphylococcal endocarditis, in which demographic and clinical history variables were entered, shaping parameters were as follows: p = 1.951, 
= 0, v = 1.042, m = 0 (corresponding risk factor estimates are provided in Table IV). Standard errors of estimates and variance-covariance estimates are available from us on request.
Acknowledgments
We thank Dr. John Kirklin for his review of the data and manuscript and Dr. Eugene Blackstone for his assistance with the data analysis. We acknowledge the assistance of Lesley Early, RN, and Laurie Kear, RN, for their superb work in data collection. We also thank Dr. E. Gregory Stafford, Dr. Michael A. H. Gardner, Dr. Peter Pohlner, and Dr. Peter J. Tesar for permitting us to include their patients in the study.
Footnotes
From the Division of Cardiothoracic Surgery, Department of Surgery, the University of Alabama at Birmingham Medical Center, Birmingham, Ala., and the Departments of Cardiac Surgery a and Cardiology, b The Prince Charles Hospital, Brisbane, Queensland, Australia. 
Read at the Seventy-fifth Annual Meeting of The American Association for Thoracic Surgery, Boston, Mass., April 23-26, 1995.
References
This article has been cited by other articles:
|
|
 |
|
  S. Pagni, A. Dempsey, and E. H. Austin III Tricuspid and Aortic Valve and Ventricular Septal Defect Endocarditis: An Unusual Presentation of Acute Q Fever Ann. Thorac. Surg., December 1, 2009; 88(6): 2027 - 2029. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. C. Stamou, G. Petterson, and A. M. Gillinov Surgical Treatment of Mitral Valve Endocarditis Card. Surg. Adult, January 1, 2008; 3(2008): 1069 - 1078. [Full Text]
|
|
|
|
|
|
J. P.A. Puvimanasinghe, J. J.M. Takkenberg, M. J.C. Eijkemans, E. W. Steyerberg, L. A. van Herwerden, G. L. Grunkemeier, J. D. F. Habbema, and A. J.J.C. Bogers Prognosis After Aortic Valve Replacement With the Carpentier-Edwards Pericardial Valve: Use of Microsimulation Ann. Thorac. Surg., September 1, 2005; 80(3): 825 - 831. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Mylonakis and S. B. Calderwood Infective Endocarditis in Adults N. Engl. J. Med., November 1, 2001; 345(18): 1318 - 1330. [Full Text] [PDF]
|
|
|
|
 |
|
 T. Mihaljevic, J. G. Byrne, L. H. Cohn, and S. F. Aranki Long-term results of multivalve surgery for infective multivalve endocarditis Eur. J. Cardiothorac. Surg., October 1, 2001; 20(4): 842 - 846. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. G. Seipelt, J. F. Vazquez-Jimenez, I. M. Seipelt, A. Franke, K. Chalabi, F. A. Schoendube, and B. J. Messmer The St. Jude ""Silzone"" valve: midterm results in treatment of active endocarditis Ann. Thorac. Surg., September 1, 2001; 72(3): 758 - 762. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. M. Gillinov, R. Diaz, E. H. Blackstone, G. B. Pettersson, J. F. Sabik, B. W. Lytle, and D. M. Cosgrove III Double valve endocarditis Ann. Thorac. Surg., June 1, 2001; 71(6): 1874 - 1879. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. R. Moon, D. C. Miller, K. A. Moore, P. E. Oyer, R. S. Mitchell, R. C. Robbins, E. B. Stinson, N. E. Shumway, and B. A. Reitz Treatment of endocarditis with valve replacement: the question of tissue versus mechanical prosthesis Ann. Thorac. Surg., April 1, 2001; 71(4): 1164 - 1171. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
O. Lund, S. L. Nielsen, H. Arildsen, L. B. Ilkjaer, and H. K. Pilegaard Standard aortic St. Jude valve at 18 years: performance profile and determinants of outcome Ann. Thorac. Surg., May 1, 2000; 69(5): 1459 - 1465. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 O. Lund, V. Chandrasekaran, R. Grocott-Mason, H. Elwidaa, R. Mazhar, A. Khaghani, A. Mitchell, C. Ilsley, and M. H. Yacoub PRIMARY AORTIC VALVE REPLACEMENT WITH ALLOGRAFTS OVER TWENTY-FIVE YEARS: VALVE-RELATED AND PROCEDURE-RELATED DETERMINANTS OF OUTCOME J. Thorac. Cardiovasc. Surg., January 1, 1999; 117(1): 77 - 91. [Abstract] [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 |
