Aortic valve replacement for active infectious endocarditis in 108 patients

Aortic valve replacement for active infectious endocarditis in 108 patients A comparison offreehand allograft valves with mechanical prostheses and bioprostheses A total of 108 patients hospitalized with active (acute) endocarditis on either a native aortic valve (n = 66) or a previously inserted replacement device (n = 42) underwent aortic valve replacement

because they were too in for hospital discharge. A nonstented aortic allograft valve was used in 78 patients and prosthetic (mechanical or bioprosthetic) valves in 30 patients. The survival rate was 82 % at 1 month, 73% at 1 year, 64% at 5 years, and 36% at 15 years. It was better in patients with native valve endocarditis than prosthetic valve endocarditis. The incremental risk factors for death in the early phase postoperatively were older age at operation, higher New York Heart Association functional class, and a larger number of previous aortic valve procedures. There were 13 episodes of recurrent endocarditis, giving an actuarial freedom of 80% at 10 years. The hazard function for recurrent endocarditis had only a low constant phase when allograft valves were used, which contrasted with the existence of a high peaking early phase (in addition to the constant phase) when prosthetic devices were used. No risk factors for recurrent endocarditis were found in patients receiving a prosthesis, and "localized" versus "extensive" endocarditis was the only risk factor when an allograft was used. Reoperation was performed in 24 patients for a variety of reasons, and freedom from reoperation was 61 % at 10 years. It is concluded that the allograft valve is the valve of choice when aortic valve replacement is required for active endocarditis. (J THoRAe CARDIOVASC SURG 1992; 103:130-9)

David Haydock, MB, FRACS, Brian Barratt-Boyes, MB, ChM, FRACS, Terence Macedo, MB, Auckland, New Zealand, John W. Kirklin, MD, and Eugene Blackstone, MD, Birmingham, Ala.

Earlier studies have shown a difference between allograft valve replacement and prosthetic replacement of the aortic valve, in that allograft replacement is followed by only a low constant phase of hazard for infectious endocarditis whereas after prosthetic replacement an early high peaking phase of infectious endocarditis also is present. 1-4 It has been presumed, therefore, that infection present at the time of operation or more often accidentally introduced at operation was more likely to evolve into

From the Cardiothoracic Surgical Unit, Green Lane Hospital, Auckland, New Zealand, and the Department of Surgery, Universityof Alabama, Birmingham, Ala. Received for publication June 21, 1990. Accepted for publication Oct. 19, 1990. Address for reprints: Sir Brian Barratt-Boyes, Green Lane Hospital, Greenlane Rd., Auckland, New Zealand.

12/1/26455

130

active aortic valve endocarditis when prosthetic material was part of the replacement device. The present study was designed to compare the resistance to infection of the allograft valve with that of prosthetic devices in the most provocative setting, namely, in patients undergoing aortic valve replacement for acute (active) infectious endocarditis. Patients and methods Patients. Between 1965 and 1984, 348 patients undergoing aortic valve replacement had endocarditis as the cause of their aortic valve lesion. One hundred eight of these patients had acute endocarditis at the time of operation. There were 87 male and 21 female patients, aged 9 to 76 (mean 44) years. At the time of operation II patients were in New York Heart Association (NYHA) class II, 16 in class III, 40 in class IV, and 41 in class V (confined to bed with severe congestive heart failure resulting from low cardiac output or renal failure and requiring operation on an urgent or emergent basis). The definition of acute or active infectious endocarditis was

Volume 103 Number 1

Aortic valve replacement for endocarditis

January 1992

I3I

Table I. Recurrent endocarditis according to the type of replacement device inserted Recurrent endocarditis Type of replacement device

n

No.

Allograft, freehand Allograft, stent mounted Hancock* Carpentier-Edwardst Starr-Edwardst Bjork-Shiley:j: Braunwald-Cutter§ Total

78 1 12 1 11 3 2 108

8 0 2 1 1 1 0 13

%

10 0 17

100 9

33 0 12

70% CL (%)

7-15 0-85 6-35 15-100 1-28 4-76 0-61 9-16

Note: One patient had the replacement device, used at the operation for infectious endocarditis, subsequently replaced and still later developed prosthetic valve endocarditis. He is not included as having experienced the event. • Johnson & Johnson Cardiovascular, King of Prussia, Pa. tBaxter Healthcare Corp., Edwards Division, Santa Ana, Calif. :j:Shiley, Inc., Irvine. Calif. §Cutter Biological, Berkeley, Calif.

based on the need to proceed with valve replacement during the hospital admission for acute endocarditis because of the severity of the patient's illness. In 44% of cases infection was uncontrolled by antibiotic treatment, in 90% there was progressive deterioration in valve function associated usually with heart failure, and in 31% embolization had occurred. All patients were on a program of intravenous antibiotics for varying periods preoperatively (mean 26 days), and these drugs were continued postoperatively for 4 weeks if cultures of tissue removed at operation were negative for infection and for 6 weeks if they were positive, but longer if evidence of infection persisted. The degree of surgical urgency varied, but in no case was the operation elective. The mean time from initial diagnosis of infectious endocarditis to the operation was 4 t/2 weeks (up to 32). In 95% of patients blood culture was positive and in 31% culture of the valve removed at operation was also positive; Gram stain was positive in 64%. The surgeon considered the disease process to be active in 88% and extensive in 34%. Vegetations were present in 77%. In only four instances was the diagnosis of active infection based solely on the appearance of the valve at operation. At operation the infective process was found to have spread beyond the immediate confines of the valve or anulus in 38 patients (35%). Eighteen of them had involvement of the base of the anterior mitral leaflet, often with spread to the adjacent membranous ventricular septum, 4 had frank abscess formation, and 17 had false aneurysms; one false aneurysm was associated with a fistulous connection to the left atrium and another with an infection-generated ventricular septal defect. Two of this group also had a congenital aneurysm of the sinus of Valsalva and 2 a mycotic aneurysm of the ascending aorta. The most common organisms were Streptococcusviridans in 30 patients, Staphylococcusaureus in 24, Streptococcusfaecalis in 12, and Staphylococcusepidermidis in 12. Only I patient had a fungal infection (Torula glabrata). No patients were known to be addicted to drugs. The endocarditis occurred on the native aortic valve in 66 patients and on a previously inserted replacement device in 42 patients (allograft in 27, mechanical prosthesis in 13, and bioprosthesis in 2). Among these 42 patients the procedure was a first repeat operation in 35, a second repeat operation in 6, and a third repeat operation in I.

Autopsy was undertaken in 19 of the hospital deaths and in 18 of the 30 late deaths. Surgical procedure. Cardiopulmonary bypass and mild hypothermia were used. Myocardial protection before 1978 consisted of continuous coronary artery perfusion and from 1978 onward, cold cardioplegic arrest with St. Thomas Hospital solution. Infected material was removed as completely as possible. False aneurysms were closed either by direct suture or with pericardial or Dacron patches before the allograft or prosthetic device was positioned. In some cases the sutures to the neck of the aneurysm were passed through the sewing ring of the device which, when seated, plugged the ostium. Interrupted sutures were used for all prosthetic valve replacements and continuous sutures for all freehand allograft valve replacements.t When replacement of the ascending aorta and sinuses of Valsalva was required, a standard Bentall procedure was performed with a Dacron tube graft containing a prosthetic valve except in one patient who had an allograft root replacement. The type of valve inserted at the operation for active endocarditis was an aortic allograft in 78, a stented aortic allograft in I, a mechanical prosthesis in 16, and a bioprosthesis in 13 (Table I). Allograft valves were harvested from cadavers, disinfected with antibiotic solution, and stored at 4 0 C in Hanks solution or nutrient medium until used," Eighteen patients had simultaneous mitral valve replacements (in 13 for active endocarditis) and 8 had repair or replacement of the ascending aorta for ascending arch aneurysm (2 aneurysms were mycotic). During the entire period under review a freehand allograft valve was the preferred replacement device, but implementation of this policy varied among the five surgeons involved. Thus the reason for not using an allograft was not stated in the operation report in 12 instances, and in 2 a suitable valve was not available. In 3 instances the surgeon considered a prosthetic aortic valve appropriate because a prosthetic mitral valve was also required. In 12 instances selection was biased toward a prosthesis either because a Bentall root replacement procedure was required (n = 4) or because the aortic root was too large for an allograft. Apart from this last category there were no major differences between the groups. Follow-up. The follow-up ranged up to 242 months and averaged 79 months. Follow-up was maintained for 5.9 years or more in 50% of the patients. No patient was completely lost to

I 32

The Journal of Thoracic and Oardiovascular

Haydock et at.

Surgery

100

90 80 70 10

>

...>

...c

::J In

...C1Iv C1I

c.

60

50 (6)

40

30 20

Years

% Survival

1/12 1 5 10 15

82% 73% 64% 52% 36%

10

a

E 0

2

4

6

8

10

12

14

16

18

20

Years After Operation Fig. 1. Survival after aortic valve replacement for acute infectious endocarditis. The circles, representing deaths, are positioned at the time of death along the horizontal axis. The vertical bars represent 70% confidence limits (approximately 1 standard deviation) of the estimates. The numbers in parentheses are the numberof patients at riskat that time after operation. The solid line is the parametric estimate of survival and is enclosed within its 70% confidence limits (dashed line). (See Appendix B for details.)

follow-up, but 6 were unable to be contacted in the formal follow-up. Follow-up of 2, 5, II, 37, 77,and 80 months wasavailable in these 6. Methodsof analysis. Events after valve replacement (death, recurrent endocarditis, reoperation) were analyzed in a timerelated manner withboth nonparametric and completely parametric methods" to obtainestimates of freedom from the event and the instantaneous risk of the event (hazard function)." 8 Risk factors for the event weredetermined in the parametric hazard function domain (seeAppendix A forthe variables considered in the analyses). A p valueof 0.1 wasused as the criterion for retaining a variable in the model.

Results Survival. The survival and hazard function are presented in Figs. I and 2. The primary cause of late death was myocardial infarction in 10, reoperation in 6, documented sudden death in 3, a cerebrovascular event in 2, congestive heart failure in I, and carcinoma in 2. In 6 of the patients dying late, recurrent endocarditis was present. Mortality in relation to the type of device inserted is shown in Table II. Older age at operation, higher NYHA functional class, and the number (zero or more) of previous aortic valve

replacements were the risk factors for death in the early peaking phase of hazard, and none was present in the constant phase (see Appendix A and Appendix B for variables entered into analysis, shaping parameters, and p values). The simple actuarial depiction of survival according to whether the acute endocarditis was on the native valve (zero previous valve replacements) or on a replacement device (Fig. 3) validates that part of the multivariate analysis. Also, survival with an allograft used to replace the infected valve and with a prosthetic valve (mechanical or bioprosthetic) were similar (P for difference [Gehan-Wilcoxon] = 0.5), a fact that validates the absence of the replacement device from the risk factors. Recurrent endocarditis. Time-related freedom from recurrent endocarditis is given in Fig. 4. The hazard functions were qualitatively different, according to the device used (Fig. 5), there being only a constant phase of hazard when an allograft valve was used to replace the infected valve, but in addition an early peaking phase and a late phase (beyond 10 years) when a prosthetic device was used. (See Appendix C for equations.) When translated to a representation of freedom from recurrent pros-

Volume 103 Number 1

Aortic valve replacement for endocarditis

January 1992

13 3

0.050 0.045 0.040 0.035 ';"

..c: 0.030 +'" c::: 0

~ 0.025

• ..c: +'" ro 0.020 0.015

I

", I 1\

0.005 0.000

2.7 4.8 7.3

15

,

I

0.010

46 4.0

1/12 1 5 10

C1I

0

Hazard (x 1000)

Years

\

I

- --

-- - -

\

\

---------

o

2

4

6

8

10

12

14

16

18

20

Years After Operation Fig. 2. Hazard function (instantaneous riskat each moment in time after the operation) fordeath after operation. The solid line is a parametricdepiction enclosed within the 70% confidence limits (dashed lines). (See Appendix B for'equation.) thetic valve endocarditis, the advantage of the allograft valve was mainly in the early years after the operation (Fig. 6). Among the 13 patients experiencing recurrent endocarditis, 5 underwent reoperation with 2 perioperative deaths. Overall, 6 of the 13 patients died, 5 within 2 months of the appearance of recurrent endocarditis. By multivariate analysis, no risk factors for recurrent endocarditis were found in the group receiving a prosthesis, and "localized" versus "extensive" endocarditis at the operation for acute infectious endocarditis was the only risk factor (P = 0.07) in the group receiving an allograft. The site of the infectious endocarditis (the native aortic valve versus a replacement valve) was not a risk factor. Freedom from reoperation. Time-related freedom from reoperation for any reason is depicted in Fig. 7, and the single gradually rising phase of hazard is depicted in Fig. 8. (See Appendix D for parameter estimates.) Younger age at the operation for infectious endocarditis was the only risk factor (P = 0.07) for reoperation of any kind. Among the 24 patients undergoing reoperation on the aortic valve replacement device, 19 had received an allograft valve at the operation for infectious endocarditis and 5 a bioprosthesis. In 17 patients the reoperation was required because of primary tissue failure, in 5

because of incompetence owing to recurrent endocarditis, in 1 because of a perivalvular leak, and in 1 because of a dissection of the ascending aorta. Discussion This article specifically addresses the frequency of recurrent endocarditis in relation to the device used for aortic valve replacement in a group of patients presumably at high risk for this event. It coincidentally provides data on the operative and subsequent survival of patients with acute endocarditis being referred for operation because they were too symptomatic for hospital discharge. It does not, however, assess the overall survival of patients with acute endocarditis, because those treated medically, which includes patients dying before operation could be performed and those with severe embolic cerebral damage, have been excluded. The data analysis is made more specific by the use of actuarial, multivariate, and hazard function techniques, but it suffers from the deficiency of relatively small numbers, a problem common to other published assessments of similar material. Prosthetic valve endocarditis, this term being used to encompass infection on any type of replacement device (allograft, bioprosthesis, and mechanical prosthesis), is a major complication of valve replacement carrying a mor-

The Journal of

I 34

Haydock et al.

Thoracic and Cardiovascular Surgery

100

90 80 70

ro

> 60

...>

:J V'I +J

t: C1I

... \"I

50 (4)

40

C1I

c...

30 1/12 1 5 10 15

20 10 0

0

2

74% 59% 57% 41% 16%

86% 80% 72% 50% 50%

4

6

(2) P(Gehan-Wllcoxon)=.03

8

10

12

14

16

20

18

Years After Operation Fig. 3. Stratified actuarial depictions of survivalaccording to initial native valve endocarditis (NVE) or initial prosthetic valve endocarditis (PVE). The actuarial depictions are as in Fig. I.

Table II. Death after aortic valve replacement for acute infectious endocarditis, according to type of replacement

device inserted Deaths Type of replacement device inserted Allograft, freehand Allograft, stent mounted Hancock Carpentier-Edwards Starr-Edwards Bjork-Shiley

Braunwald-Cutter Total

Hospital

Total

n

No.

%

70% CL (%)

No.

%

78 1 12 I 11 3 2 108

13 I 3 0 2 I 0 20

17 100 25 0 18 33 0 19

12-22 15-100 11-44 0-85 6-38 4-76 0-61 15-23

38

49 100 33 0 36 33 100 46

I

4 0 4 I 2 50

For legend see Table I.

tality of up to 65%.3 Its reported incidence varies and is related to the type of device. Provided there is high quality control of the antibiotic treatment used to disinfect allograft valves and, in particular, contamination by fungi is stringently avoided and the valve is inserted without the addition of any foreign material other than sutures (for example a stent), the incidence of subsequent infection is low." It would also appear that infection of an allograft is rare in the early phase postoperatively, perhaps in part because the antibiotic used perioperatively

can penetrate the entire device. Thus our own data I and that of O'Brien and associates- show a low constant hazard function identical to that found in the present study. The only exception to this observation in the literature has been the report by Penta and colleagues,'? in which the majority of early infections were from preoperative fungal contamination of the graft, and our own material between 1962 and 1966, which included 5 early infections among 254 patients (1.9%).9 We have no explanation for this finding.

Volume 103 Number 1 .Ianuarv 1992

Aortic valve replacement for endocarditis

135

100 VI

+'"

90

I'll

...

80

"lJ

70

"lJ

v 0

c

w

(9)

+'"

c: 60

......::JOJ

-... v

OJ

cc:: 0

OJ OJ u,

50 40 30

+'"

Years

% Free

1/12 1 5 10 15

100% 94% 88% 80% 65%

(2)

c: 20

...OJv OJ

e,

10 0

0

2

4

6

8 10 12 14 Years After Operation

16

18

20

Fig. 4. Freedom from recurrent endocarditis after aortic valve replacement in the group as a whole. The actuarial and parametic depictions are as in Fig. I. (See Appendix C for details.)

The resistance of the allograft valve to infection in the early phase postoperatively is in marked contrast to bioprostheses and mechanical prostheses, in which infection is most common in the first 6 weeks postoperatively.' This is again shown in the parametric and hazard function analyses of the present series in almost identical fashion to tha t of!vert and coworkers.' In fact, the only difference between this and the previous series from Green Lane Hospital' and from the University of Alabama.' is that the hazard at I month for infection of mechanical prostheses and bioprostheses was more than twice in the present series (0.025 j month) than in the previous two series (0.0085 jmonth). This could be explained by the fact that in the present series all aortic valve replacements were for acute (active) endocarditis, which makes early recurrence even more likely than when the host valve was usually sterile. This observation correlates with the information that the allograft valve is probably even more protective against recurrent infection when active infection is extensive (as judged by the surgeon) at the time of valve replacement. The 30-day mortality rate of 18% in this series of patients undergoing aortic valve replacement for acute endocarditis was considerably higher than for aortic valve replacement for all causes at Green Lane Hospital during the same time frame.v II It is comparable to those

reported by Richardson, 12 Lewis.P and their associates in similar groups of patients who were not drug addicts (intravenous heroin users). However, it should be noted that 42 patients (39%) were undergoing a first, second, or third reoperation, 26 (24%) required a simultaneous operation for mitral valve disease or an ascending arch aneurysm, and 80 (74%) were in functional class IV or V. Late survival was little different from that after aortic valve replacement and multiple valve replacement for other causes, although recurrent endocarditis may have been a more common cause of death. The risk factors for death were no different from those of valve replacement performed in other circumstances. Risk factors could be identified only for the early phase postoperatively. The number of late deaths resulting from coronary artery disease is a disturbing feature and raises the question of whether coronary angiography should be performed routinely in patients older than 40 years of age, despite the risk of embolization from vegetations on the aortic valve, so that coronary grafting can be added when appropriate. The reports from Welton,I4 Mills,15 and their colleagues would suggest that embolization is a rare complication of catheterization. In this series of patients allograft aortic root replacement, as distinct from freehand allograft insertion, was used only once. In fact, when root replacement was man-

136

The Journal of Thoracic and Cardiovascular Surgery

Haydock et al.

0.06

0.05

~..!: +"

c:

o :2:

Hazard (x 1000)

...

.;, 0.04

Replacement

Years

Prosthetic Valves (n=30)

"'C "-

III

(:) 0.03

c:

LLJ

c: 0.02 QI

1.8 1.8

32

.83 .83 6.0

5

+"

=

5.8

1/12 3/12 1

"'C

Allografts (n 78)

10

1.8

1.8 1.8

"-

"-

~

'"' c:::

QI

0.01

12

6

A

18

24

30

36

42

48

54

60

Months After Operation

0.06

:c +"

0.05

c:

o :2:

•

:-e "'C III

0.04

Hazard (x 1000) Replacement

Years

"-

III

Prosthetic Valves (n=30)

o'"' 0.03

"'C

e

+"

c:

QI ""-

0.02

~

'"' c:::

Allografts (n=78)

5.8

1/12 3/12 1 5 10

LLJ

_

32

.83 .83 6.0

QI

0.01

, ,

: ~ I

-»

~

\

• • _ .....

.:.

4

B

sr;

6

_

8

10

12

14

16

18

20

Months After Operation

Fig. 5. A, Stratified hazard function depiction of recurrent endocarditis in patients with freehand allografts implanted compared with those with prosthetic mechanical or bioprosthetic valves. Patients with originally localized endocarditis had a constant hazard rate of 3.1 events per 1000 patients per month compared with 0.45 events per 1000 per month in patients with originally extensive endocarditis. (See Appendix C for details.) B, Same depiction, extending to 20 years.

Volume 103 Number 1

Aortic valve replacement for endocarditis

January 1992

100 III

90

...

80

".iJ "'C ttl

v

0 "'C

c:

--- -----

".

70

w

'

....c: 60

......

QI ~

-... v

Percent

50

Replacement

QI

0::

0

QI QI

Years

40 30

1/12 1 5 10 15

....c: 20 v ...

LL.

QI

QI

e,

10 0

.

0

2

4

Allografts (n =78)

Prosthetic Valves (n = 30)

100%

100%

90% 81% 72%

24%

(0)

87% 85% 73%

98%

6

8

10

12

14

16

18

20

Years After Operation Fig. 6. Stratified depiction of recurrent endocarditis according to whether the replacement device was a freehand allograft (circles) or a prosthetic valve (squares). (See Appendix C for details.) The depiction is as in Fig. I.

100

90 c::: 80

....o

70

~

Q)

c..

al 0::

,

60

0 50

QI QI

~

....c..

Years

40

...


100%

1/12 1

~30

c,

% Free 97%

5 10 15

20

83% 61% 36%

0

,

(12)

'I " ,

0',

;;,'

, , , , , , ,

10

o

10.---'--_ ...J_--'-----l._-'-----L_'---L.--'_...L-----'-_L---'-----l_--'-----L_.L..----L---JL....----.J

o

2

4

6

8 10 12 14 Years After Operation

16

18

20

Fig. 7. Freedom from reoperation after operation for acute endocarditis. The depiction is as in Fig. I. (See Appendix D for details.)

13 7

I 38

The Journal of Thoracic and Cardiovascular

Haydock et al.

Surgery

c.oso 0.045 0.040 ..c: ....

0.035

0

0.030

•

0.025

l:

~ l:

0

'';:; ttl

~

Hazard (x 1000)

Years

1/12 1 5 10 15

0.020

,,

2.4 2.6 4.0 6.7 11

QJ

c-

o 0.015 QJ a:: 0.010

0.005 0.000

- - -0

2

4

--6

8

10

12

14

16

18

20

Years After Operation Fig. 8. Hazard function for reoperation of any kind.

datory, and this in most instances was necessary for nonmycotic ascending arch and sinus of Valsalva aneurysms, a standard Bentall operation was used with a Dacron graft and prosthetic valve, rather than an allograft root replacement. Extensive destruction of the aortic root with abscess formation and false aneurysms is uncommon, particularly now that valve replacement is practiced early in the clinical course; and in our institution, fever that persists for more than 7 days from the commencement of antibiotics in association with valvular incompetence is an indication for operation." Accordingly, freehand allograft aortic valve replacement can usually be performed in a routine manner.f which involves first excising infected tissue. When a false aneurysm is present its mouth is closed either by direct suture or with a patch of pericardium. The allograft is then seated on top, with more than the usual amount of aortic wall left above the cusp hinge to cover the repair.'? In those instances in which there is extensive tissue destruction with partial or complete aortic-left ventricular discontinuity or ascending arch aneurysm, we would now use allograft root replacement rather than a composite prosthetic graft, using the technique described by Lau and associates.l'' but wrapping the lower suture line with pericardium rather than Teflon felt. Now that a very favorable longterm follow-up is available for this procedure,'? it would appear superior to other more complex reconstructions that use prosthetic materia1. 20, 21 Allograft root replace-

ment can also be considered in preference to freehand insertion when there is excessive root distortion or the root is too large for a freehand allograft and not suitable for root tailoring. Interpretation of the significance of the reoperation incidence in this series is difficult because there were many variables. Five of the 22 reoperations were performed for recurrent endocarditis (in four others who died from this cause reoperation would have been appropriate) and in three others persistent large false aneurysms were repaired at the time of valve replacement. At reoperation a surprising number of patients required concomitant mitral valve replacements, only two of which were the result of earlier infection, most of the others being due to progression of rheumatic valvular disease. The majority of reoperations were unrelated to the initial endocarditis being required for primary leaflet failure of allograft or xenograft valves, a mean of 7 years postoperatively. This would be the only discouragement to allograft valve usage. The future incidence of primary tissue failure in allografts is expected to decrease now that the risk factors are better understood.r 6 REFERENCES 1. Kirklin JW, Barratt-Boyes BG. Cardiac surgery. New York: John Wiley, 1986:409. 2. O'Brien MF, Stafford EG, Gardner MAH, Pohlner PG, McGiffin DC. A comparison of aortic valve replacement

Volume 103 Number 1 January 1992

with viable cryopreserved and fresh allograft valves with a note on chromosomal studies. J THORAC CARDIOVASC SURG 1987;94:812-24. 3. Ivert TSA, Dismukes WE, Cobbs LG, Blackstone EH, Kirklin JW, Bergdahl LAL. Prosthetic valve endocarditis. Circulation 1984;69:223-32. 4. Blackstone EH, Kirklin JW. Death and other time-related events after valve replacement. Circulation 1985;72:75367. 5. Barratt-Boyes BG. A method for preparing and inserting a homograft aortic valve. Br J Surg 1965;52:847. 6. Barratt-Boyes BG, Roche AHG, Subramanyan R, Pemberton JR, Whitlock RML. Long-term follow-up of patients with the antibiotic sterilized aortic homograft valve inserted freehand in the aortic position. Circulation 1987; 75:768-77. 7. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53: 457. 8. Blackstone EH, Naftel DC, Turner ME Jr. The decomposition of time-varying hazard into phases, each incorporating a separate stream of concomitant information. J Am Stat Assoc 1986;81:615-24. 9. Clarkson PM, Barratt-Boyes BG. Bacterial endocarditis following homograft replacement of the aortic valve. Circulation 1970;42:987-91. 10.. Penta A, Zureshi S, Radley-Smith R, Yacoub MH. Patient status 10 or more years after "fresh" homograft replacement of the aortic valve. Circulation I984;70(Pt 2)1182-6. II. Barratt-Boyes BG. The timing of operation in valvular insufficiency. J Card Surg 1987;2:435-52. 12. Richardson JV, Karp RB, Kirklin JW, Dismukes WE. Treatment of infective endocarditis: a 10 year comparative analysis. Circulation 1978;58:589-97. 13. Lewis BS, Agathangelou NE, Colsen PR, Atunes M, Kinsley RH. Cardiac operation during active infective endocarditis: results of aortic, mitral, and double valve replacement in 94 patients. J THoRAc CARDIOVASC SURG 1982;84:579-84. 14. Welton DE, Young J B, Raizner AE, et al. Value and safety of cardiac catheterization during active infective endocarditis. Am J Cardiol 1979;44:1306-1O. 15. Mills J, Abbott J, Utley JR, Ryan C. Role of cardiac catheterization in infective endocarditis. Chest 1977;72:576-82. 16. Ormiston JA, NeutzeJM, AgnewTM, LoweJB, Kerr AR. Infective endocarditis: a lethal disease. Aust NZ J Med 1981;11:620. 17. Kirklin JK, Kirklin JW, Pacifico AD. Aortic valve endocarditis with aortic root abscess cavity: surgical treatment with aortic valve homograft. Ann Thorac Surg 1988; 45:674-7. 18. Lau JKH, Robles A, Cherian A, Ross DN. Surgical treatment of prosthetic endocarditis: aortic root replacement using a homograft. J THORAC CARDIOVASC SURG 1984; 87:712-6. 19. Yacoub MH. Allograft aortic root replacement. In: Yankah AC, Hetzer R, Miller DC, Ross DN, Somerville J,

Aortic valve replacement for endocarditis

139

Yacoub M H, eds. Cardiac valve allografts 1962-1987. New York: Springer-Verlag, 198:149. 20. Buckley MJ, Mundth ED, Daggett WM, Austen WG. Surgical management of the complications of sepsis involving the aortic valve, aortic root and ascending aorta. Ann Thorac Surg 1971;12:391. 21. Frantz PT, Murray G F, Wilcox BR. Surgical management of left ventricular-aortic discontinuity complicating bacterial endocarditis. Ann Thorac Surg 1980;29:1-7.

Appendixes Appendix A: Variables entered into all risk factor analyses. Demographic variables. Age, gender, race. Previous operative variables: number of previous aortic valve procedures, number of previous aortic valve replacements. Clinical variables: New York Heart Association class (I to V), interval from onset of infection to operation, duration of preoperative antibiotics, blood urea nitrogen level, native versus prosthetic valve endocarditis. Morphologic variables: localized endocarditis, extensive endocarditis. Surgical variables: surgeon, date of operation, type of valve replacement device (mechanical prosthesis, xenograft, bioprosthesis, allograft [freehand insertion D, concomitant mitral valve replacement. Appendix B: Survival equations I. The parameter estimates (Figs. I and 2) for the early phase were as follows.u, = 0.6359,0 = O,p = 2.161,11 = 4.995, m = 0; and for the late phase: ~3 = 0.00001032, T = I, a = I, I' = I, '1/ = 2.114. 2. The shaping parameter estimates and the regression coefficients, standard deviations, and P values of the multivariate risk factor equation for death were as follows: Early phase: 0=0, p = 2.915, II = 5.028, m = 0; intercept = -4.909, age (years) at operation = 0.0230 ± 0.0111 (P = 0.04), NYHA class (I to V) = 0.7142 ± 0.25 (P = 0.005), number of previous aortic valve replacement operations = 0.8038 ± 0.24 (P = 0.0008); Late phase: ~3 = 0.00001530, T = 1, a = 1, I' = I, '1/ = 2.034 (no riskfactors). Appendix C: Equations for recurrent endocarditis I. The parameter estimates overall for recurrent endocarditis (Fig. 4) were as follows: Early phase: ~I = 0.08362, 0 = 0, p = 5.009, II = I, m = 0; and for the late phase: ~3 = 0.04464, T = 82.14, I' = I, a = 0, '1/ = 1. 2. The parameter estimate for recurrent endocarditis after freehand homograft insertion (Figs. 5 and 6) was as follows: Constant phase: ~2 = 0.001792. 3. The parameter estimates for recurrent endocarditis after insertion of other prosthetic devices (Figs. 5 and 6) were as follows: Early phase: ~I = 0.1393,0 = 0, p = 2.248, II = 0.4645, m = 0; and for the late phase: ~3 = 0.003467, T = 30.35, a = 0, 1'=1,'1/=1. 4. The regression coefficients, standard deviation, and P values of the multivariate risk factor equation for recurrent endocarditis after freehand allograft insertion were as follows: Constant phase: intercept = - 7.701, localized endocarditis = 1.925 ± 1.07 (P = 0.07). No risk factors were found in the prosthetic valve insertion group. Appendix D: Reoperation. The parameter estimates for reoperation (Fig. 7) were as follows: Late phase: ~3 = 0.2734, T = 115.8, a = 0, I' = I, '1/ = 1.