Morris CD, Reller MD, Menashe VD. Thirty-Year Incidence of Infective Endocarditis After Surgery for Congenital Heart Defect. JAMA. 1998;279(8):599-603. doi:10.1001/jama.279.8.599
From the Congenital Heart Disease Research Center (Drs Morris, Reller, and Menashe), Division of Medical Informatics and Outcomes Research (Dr Morris), Division of Pediatric Cardiology (Drs Reller and Menashe), Oregon Health Sciences University, Portland.
Context.— The incidence of infective endocarditis after surgical repair of congenital
heart defects is unknown.
Objective.— To determine the long-term incidence of endocarditis after repair of
any of 12 congenital heart defects in childhood.
Design.— Population-based registry started in 1982.
Setting.— State of Oregon.
Participants.— All Oregon residents who underwent surgical repair for 1 of 12 major
congenital defects at the age of 18 years or younger from 1958 to the present.
Main Outcome Measure.— Diagnosis of infective endocarditis confirmed by hospital or autopsy
Results.— Follow-up data were obtained from 88% of this cohort of 3860 individuals
through 1993. At 25 years after surgery, the cumulative incidence of infective
endocarditis was 1.3% for tetralogy of Fallot, 2.7% for isolated ventricular
septal defect, 3.5% for coarctation of the aorta, 13.3% for valvular aortic
stenosis, and 2.8% for primum atrial septal defect. In the cohorts with shorter
follow-up, at 20 years after surgery the cumulative incidence was 4.0% for
dextrotransposition of the great arteries; at 10 years, the cumulative incidence
was 1.1% for complete atrioventricular septal defect, 5.3% for pulmonary atresia
with an intact ventricular septum, and 6.4% for pulmonary atresia with ventricular
septal defect. No children with secundum atrial septal defect, patent ductus
arteriosus, or pulmonic stenosis have had infective endocarditis after surgery.
Conclusion.— The continuing incidence of endocarditis after surgery for congenital
heart defect, particularly valvular aortic stenosis, merits education about
endocarditis prophylaxis for children and adults with repaired congenital
THE INCIDENCE OF infective endocarditis is rare and is estimated to
be 0.38 case per 10000 person-years.1 Over the
last 2 decades, a changing pattern of occurrence of infective endocarditis
has been noted,2,3 in part owing
to a decrease in the case-fatality rate, a shift to more uncommon causative
organisms, an improvement in diagnostic methods, and the growing proportion
of cases in individuals who have had surgery for congenital heart disease.
Although case series of children with endocarditis have been reported, these
studies could not estimate the incidence of endocarditis after surgery for
a congenital heart defect.2- 7
It is well understood that the presence of a heart defect increases this incidence,
particularly defects associated with high-velocity or turbulent flows, such
as ventricular septal defect (VSD), aortic stenosis, coarctation of the aorta,
and patent ductus arteriosus (PDA). Conversely, defects such as secundum atrial
septal defects (ASDs) are not associated with infective endocarditis.8,9
The risk of infective endocarditis should be eliminated or markedly
decreased for heart defects that can be completely repaired surgically such
as ASDs, VSDs, and PDA.8 For other defects,
such as aortic valve disease, that risk may not be markedly altered by surgery.
For some complex defects, the risk of infective endocarditis may actually
be increased by surgery by placement of a prosthetic valve or conduit or with
palliative shunt surgery for complex cyanotic heart disease.7
Over the last 15 years, we have prospectively followed up all Oregon
residents with any of 12 congenital heart defects for morbidity and mortality
as part of a registry.10 The purpose of this
analysis was to demonstrate the incidence of endocarditis in patients with
congenital heart disease after surgical repair in a population-based cohort.
The estimation of comparative risks among different defects will allow prevention
to focus on individuals who are at highest risk after surgery.
In 1982, we instituted a population-based registry to enroll all Oregon
residents who had surgical repair of major congenital heart defects at less
than 19 years of age from 1958 to the present.10
Surgery was performed at 5 Oregon hospitals; 0.7% of these procedures occurred
at hospitals outside Oregon. Over time, this registry has been expanded to
include 12 major heart defects: (1) tetralogy of Fallot; (2) isolated VSD;
(3) isolated secundum ASD; (4) coarctation of the aorta with or without VSD
or aortic valve disease; (5) aortic stenosis including isolated valvular,
subaortic, and supra-aortic stenosis; (6) isolated pulmonary valvular stenosis
with intact ventricular septum; (7) dextrotransposition of the great arteries;
(8) isolated PDA in children older than 3 months to exclude PDA associated
with prematurity; (9) primum ASD with cleft mitral valve; (10) complete atrioventricular
septal defect (AVSD); (11) pulmonary atresia with intact ventricular septum;
and (12) pulmonary atresia with VSD (tetralogy of Fallot with pulmonary atresia).
Children with PDA or patent foramen ovale associated with these defects are
included in the cohort.
Children who had palliative surgery only are excluded from this cohort.
However, children who had placement of a systemic-to-pulmonary shunt or pulmonary
artery banding are included if subsequent definitive surgery was performed.
For inclusion in this registry, definitive surgical repair of tetralogy of
Fallot is defined as closure of the VSD and relief of pulmonary valvular and
right ventricular outflow tract obstruction. For VSD and primum and secundum
ASD, the defect was closed by a prosthetic patch or suture; repair of the
mitral cleft is usually included in repair of primum ASD. Repair for relief
of coarctation of the aorta has varied by aortic anatomy, surgical preference,
and year of surgery; a small number with balloon dilatation of native coarctation
has been included. Surgical procedures for coarctation have included end-to-end
anastomosis, bypass graft (ie, ascending-descending conduit), subclavian flap
aortoplasty, and prosthetic patch aortoplasty. Initial procedures for relief
of aortic stenosis included valvotomy in 96%, aortic valve replacement in
2%, and balloon valvotomy in 2%. For pulmonary stenosis, 82% had surgical
valvotomy, and 18% had balloon valvotomy. Repair of transposition of the great
arteries has changed from the Mustard repair, which was used until 1982, to
the Senning technique, and more recently to arterial switch. Patent ductus
arteriosus was repaired by ligation, division, or both. Complete AVSD repair
includes closure of the atrial and ventricular septa and repair of the mitral
valve. Repair of pulmonary atresia includes establishment of right ventricular
to pulmonary artery continuity; when associated with VSD, this also includes
closure of the defect.
To form the registry, medical records departments in all Oregon hospitals
that performed cardiac or thoracic surgery were asked to identify cases using
both procedure and diagnostic codes of hospital admissions. Computerized records,
card files of hospital admissions, and surgical logs were searched to identify
cases. Since 1982, data have been added to the registry prospectively with
yearly ascertainment of surgical cases. Chart abstraction was performed by
the principal investigator or a research assistant who had undergone training
and monitoring to maintain data quality and integrity.
To obtain long-term follow-up information, subjects were traced through
next of kin, physicians, employment records, Department of Motor Vehicles
registrations, city and telephone directories, and the National Death Index.
Follow-up status of all individuals in the registry was determined by a mailed
questionnaire every 2 years; data from the follow-up cycle that began in late
1993 are included in this analysis. Individuals who did not complete the mailed
questionnaire were contacted by telephone for a formatted interview. In addition
to assessing functional status, the questionnaire asked the individual or
their family about specific events such as endocarditis, recurrent surgery,
or any hospitalization. Affirmative answers to endocarditis and all hospitalizations
for cardiac or infection-related events were confirmed through medical records
or patient's physicians. Death certificates or hospital records of all deaths
were obtained. Prior to surgery, endocarditis was identified at the time of
inclusion into the registry. Endocarditis was considered to be present if
the medical records indicated (1) histologic evidence of endocarditis from
surgery or autopsy; (2) positive blood culture and new or changing regurgitant
murmur, development of congestive heart failure, vegetation on echocardiogram,
or vascular phenomena; or (3) negative blood culture with fever and new or
changing regurgitant murmur, vegetation on echocardiogram, or vascular phenomena.
Only 1 case of physician-diagnosed endocarditis did not meet any of these
The cumulative incidence of postoperative endocarditis for each defect
cohort was estimated by the Kaplan-Meier method. This was defined as elapsed
time from the date of surgery to the date of endocarditis or, in subjects
free of endocarditis, the date of death or the last date of contact with the
subject. Annualized risk was calculated by dividing the total number of cases
of endocarditis into the total years of follow-up after surgery for each cohort.
All statistics were calculated separately for each defect cohort and are presented
as the proportion (and SE) with endocarditis.
The cohorts included in this analysis comprise 3860 individuals who
had surgery for 1 of 12 heart defects. Secundum ASD was the most common defect,
and pulmonary atresia the least common. The sample size, median age at operation,
and total patient-years of follow-up after surgery are shown in Table 1. In this population, 90.8% were non-Hispanic white, 4.0%
Hispanic, 2.3% Asian or Pacific Islander, 2.1% African American, and 0.8%
Native American. The median age at operation ranged from 0.005 year (2 days)
for pulmonary atresia to 7.0 years for aortic valve stenosis; however, the
age of surgery has decreased over time, as previously described for this cohort.10 Most recently, the median age at reparative surgery
for tetralogy of Fallot has been 7.2 months, 10.8 months for VSD, 3.0 months
for coarctation of the aorta, 4.9 years for aortic stenosis, 8.4 months for
pulmonary valve stenosis, and 11 days for transposition of the great arteries.
Follow-up data for endocarditis were available from 88% of this population
through 1993. The highest rate of follow-up was in the cohort with transposition
of the great arteries (95%), and the lowest rate was for those with PDA (84%).
The median follow-up duration was longest for the cohort with tetralogy of
Fallot at 159 months and shortest for the cohort with pulmonary atresia at
8 months, owing in part to a high rate of operative mortality in this latter
Prior to definitive surgery, infective endocarditis occurred most often
in children with VSD and tetralogy of Fallot with a systemic-to-pulmonary
shunt. In the 98 children with tetralogy of Fallot who had a palliative shunt,
4 cases occurred for a risk of 8.2 cases per 1000 patient-years (Table 2); no cases occurred in the first
year after shunt placement. Prior to surgical closure of the VSD, the incidence
(SE) of infective endocarditis was 1.1% (0.7%) at 5 years of age; overall,
the risk was 3.8 cases per 1000 patient-years prior to surgical closure (Table 2). Single cases of endocarditis were
noted in the cohort with valvular aortic stenosis, pulmonary atresia with
VSD, and PDA prior to definitive surgery. No children with primum or secundum
ASD, coarctation of the aorta, pulmonary valve stenosis, transposition of
the great arteries, complete AVSD, or pulmonary atresia had infective endocarditis
prior to operative repair.
In the 30 years after repair of secundum ASD, pulmonary valve stenosis,
and PDA, no one has developed infective endocarditis (Table 2 and Table 3).
After surgery, the highest incidence of infective endocarditis has been
in the cohort with aortic valve stenosis (Table 3). This rate excludes individuals with isolated supravalvular
(n=18) or subvalvular aortic stenosis (n=36) in whom there were no cases of
infective endocarditis either before or after surgery. Thirteen cases of infective
endocarditis were diagnosed after surgery for aortic valve stenosis. The incidence
of infective endocarditis appears to increase more rapidly after 5 years of
follow-up after surgery, and by 25 years the cumulative incidence is 13.3%
(3.8%) (Table 3). Considered over
all years of follow-up, the risk of infective endocarditis is 7.2 cases per
1000 patient-years (Table 2).
To determine if valve replacement increased the risk of endocarditis
in the cohort with aortic stenosis, we compared this with the risk with a
native valve (Figure 1). The incidence
with a native valve was computed using the time of valvotomy to time of valve
replacement or the last date of observation. Incidence with a prosthetic valve
was computed from time of valve replacement to the last observation. In total,
16% of the cohort had aortic valve replacement either as their initial surgery
(n=2) or at reoperation (n=26). With a prosthetic valve, there were 3 cases
of endocarditis with a 10-year incidence of 26% (13%). With a native valve,
there were 10 cases with a 10-year incidence of 5% (2%), 20-year incidence
of 11% (4%), and 25-year incidence of 15% (6%).
In the cohort with coarctation, infective endocarditis occurred in 8
individuals after surgery. In all cases, the infection occurred either at
the site of repair or at an associated abnormal aortic valve, with or without
stenosis; in 3, infective endocarditis occurred immediately after surgery
for coarctation. As with valvular aortic stenosis, the risk appears to increase
with age or time after surgery. By 30 years after surgery, the cumulative
incidence of endocarditis is 3.5% (Table
3). Over all years of follow-up, the risk is 1.2 cases per 1000
patient-years (Table 2).
Five individuals with tetralogy of Fallot had infective endocarditis
after reparative surgery. All cases occurred within the first decade after
surgery; at 10 years after surgery the cumulative incidence of endocarditis
is 1.3% (0.6%), which remains constant through 30 years (Table 3). Three of the 5 patients with infective endocarditis had
a residual VSD; one of these individuals was presumed to have acquired infective
endocarditis after a cardiac catheterization. One case occurred in the immediate
postoperative period, and another had an infected prosthetic right outflow
In the group with pulmonary atresia with VSD, there were 3 episodes
of infective endocarditis after reparative surgery; 2 of the 3 had a pulmonary
homograft. At 10 years after surgery, 6.4% of the cohort had an episode of
infective endocarditis (Table 3);
2 episodes occurred immediately postoperatively. One episode occurred in a
child with pulmonary atresia without VSD 1 month after surgery.
After surgery for VSD, infective endocarditis occurred in 4 individuals.
At 30 years after surgery, the cumulative incidence is 4.1% (Table 3). This risk appears to increase 20 years after surgical
closure of the defect. Two of these 4 individuals had a residual VSD; 2 others
had no residual VSD but 1 had a bicuspid aortic valve and the other developed
aortic insufficiency after VSD repair. No one with a closed VSD in the absence
of other anomalies developed endocarditis.
Two cases of infective endocarditis occurred after repair of primum
ASD, 1 in the immediate postoperative period. At 20 years after surgery, the
incidence was 2.8% (Table 3). Only
1 child developed infective endocarditis after complete AVSD repair; the 15-year
incidence rate is 1.1% (Table 3).
After repair of dextrotransposition of the great arteries, 1 individual developed
endocarditis, which resulted in his death 16 years after a Mustard procedure.
No cases of endocarditis occurred after Senning or arterial switch procedures
in this cohort.
The causative organism was identified in 92% of cases. A positive blood
culture was present in 33 of 38 cases of endocarditis that occurred after
surgery; 3 were consistently culture negative, and in 2 the diagnosis was
first considered at autopsy. Because endocarditis was diagnosed as early as
1962 in this cohort, the method of diagnosis has differed over time. In 26%
of the cases, echocardiography was not used. Of those with an echocardiogram,
findings indicative of endocarditis were present in 54%. The most common infecting
organisms were Streptococcus viridans and Staphylococcus aureus (each 23% of total); coagulase-negative staphylococcus, β-hemolytic
streptococcus, or Staphylococcus epidermidis infection
(each 9%); and single cases of Candida, Serratia, and both S aureus and S viridans infection.
Overall, of the 38 cases of endocarditis after surgical repair, there
were 7 deaths from endocarditis (18%). These deaths were distributed among
different heart defects. Endocarditis occurred in the immediate postoperative
period in 22% and from infected patch material in 6%. The infection was presumed
by the treating physician to be of dental origin in 14% based on a recent
dental procedure or poor oral hygiene. Single episodes of endocarditis were
presumed to be from a ventriculoperitoneal shunt, a Pott shunt, cardiac catheterization,
skin laceration, acne, and intravenous drug abuse. The route of infection
was unclear in 32%.
In contrast to previous studies that have analyzed the frequency of
heart defects in a case series of individuals with endocarditis,2- 6
our analysis prospectively determined the occurrence of endocarditis in groups
with specific heart defects after surgery. This approach allows computation
of the long-term risk of endocarditis for specific heart defects and a direct
comparison of this risk among subpopulations. As the population of adults
with congenital heart defects increases in number owing to the dramatic improvement
in operative survival over the last 40 years, it is likely that endocarditis
will play a significant role in morbidity and mortality in this population.
Table 2 summarizes the risk
of developing endocarditis categorized by the clustering of risk apparent
in this cohort. This reflects risk over all years after surgical repair. It
should be noted that endocarditis within the immediate postoperative period
explained 22% of the cases, occurring in children with tetralogy of Fallot,
primum ASD, coarctation, pulmonary atresia, and pulmonary atresia with intact
septum. Endocarditis in this period does not markedly alter the risk over
time or the comparative risk among these groups, except for pulmonary atresia,
as endocarditis continued to occur after this period. This emphasizes the
importance of antibiotic prophylaxis. Because of the frequent use of prosthetic
conduits with pulmonary atresia, these individuals likely continue at high
risk of endocarditis after the immediate operative period.
The highest incidence of endocarditis after long-term follow-up is for
those with left-sided outflow obstruction; for example, for aortic valvular
stenosis the incidence rises to 20.6% after 30 years (Table 3). The risk of endocarditis in this group has been cited
as 1.6 cases per 1000 patient-years in those treated medically, but 4.1 cases
per 1000 patient-years with surgical treatment.11,12
In the present study, the annualized risk with aortic valve stenosis is somewhat
higher at 7.2 cases per 1000 patient-years (Table 2). This study appears to confirm a higher rate of infective
endocarditis with a prosthetic valve as compared with a native valve with
aortic stenosis.8 However, this finding must
be interpreted with caution as individuals contributed time to both cohorts.
For coarctation of the aorta, the risk is more moderate at 1.2 cases per 1000
patient-years (Table 2). In the
majority of these individuals, the site of infection was an anatomically abnormal
aortic valve, a well-recognized association.13
In contrast, the occurrence of endocarditis is low with right-sided
heart defects, particularly tetralogy of Fallot. In all individuals who developed
infective endocarditis after repair of tetralogy, either a residual VSD was
present or the prosthetic patch or a Pott shunt was the source of infection.
Surgery is known to reduce the risk of infective endocarditis for children
with tetralogy of Fallot9; these data indicate
that with complete closure of the VSD and without systemic-to-pulmonary shunt,
this risk may be very small. The higher occurrence of endocarditis with pulmonary
atresia with VSD at 11.5 cases per 1000 patient-years (Table 2) likely results from the placement of a prosthetic conduit.
This estimate must be interpreted with caution given the small cohort size
and relative paucity of years of follow-up.
The incidence of infective endocarditis in this population-based cohort
is likely to be representative of the larger population after definitive surgery
for these defects. However, this may underestimate the true incidence given
the reliance on self-reporting from mailed questionnaires or interviews. We
have taken great efforts to minimize this by following a high percentage of
the cohort, and by obtaining information from multiple sources through inquiries
about endocarditis, hospitalizations, and valve replacement; some cases may
have been missed because no autopsy was performed. Conversely, it is also
possible that the incidence is overestimated, given a higher diagnostic suspicion
in a patient with a heart defect who presents with bacteremia. Nonetheless,
we do not believe that the risks of infective endocarditis in this study substantially
misrepresent the true occurrence, particularly in the cohorts with the largest
With changes in diagnostic methods over the time of this study, it is
impossible to determine the endocarditis risk of specific anatomic or physiologic
variants, such as with mitral regurgitation after primum ASD repair or bicuspid
aortic valve with coarctation. It is possible that endocarditis risks for
a cohort may be dependent on such variants. These data are also generalized
from surgical procedures performed over more than 35 years; it is possible
that changes in surgical strategy may alter these risks somewhat. Despite
this, the strength of these data is in the population database, prospective
determination, and long period of follow-up.
Unfortunately, this study cannot address the issue of antibiotic prophylaxis
in a population with repaired congenital heart defects. Given the limitations
of observational data, it is impossible to determine if the use or lack of
use of antibiotic prophylaxis had any effect on incidence. The American Heart
Association recommendations support use of endocarditis prophylaxis for individuals
in all heart defect cohorts included in this study except for isolated secundum
ASD and for VSD and PDA more than 6 months after surgical repair without residua.14 The European consensus statement is in agreement except
that it does not include pulmonic stenosis on the list of defects for which
endocarditis prophylaxis is recommended.15 These
data support the risks and concomitant recommendations for prophylaxis suggested
in both statements with regard to no risk associated with secundum ASD and
with VSD and PDA after surgical repair and the European consensus as to no
risk with pulmonic stenosis. Longer follow-up from this and larger data sets
will need to confirm no risk associated with pulmonic stenosis.
Perhaps the most important outcome of this study is the recognition
of a high risk of endocarditis with aortic stenosis. As such, greater efforts
should be made to educate individuals with defects that place them at high
risk for endocarditis, as well as others participating in their health care.
In a previous study of parental knowledge of endocarditis prophylaxis, 98%
of parents knew the name of their child's heart condition but only 56% of
parents with at-risk children knew measures to prevent endocarditis.16 In a survey of adults with congenital heart disease
seen in a special cardiology clinic, 68% could correctly identify their heart
condition and 79% knew that they needed to take an antibiotic prior to dental
work.17 Education should begin with the parents
of children as to the risk of infective endocarditis and the appropriate use
of antibiotic prophylaxis. As individuals become adolescents, they should
receive adequate information so that as adults, they will have a greater understanding
of their heart disease and the importance of prevention of endocarditis.