Journal Logo


A Clinical Study of Culture-Negative Endocarditis

Werner, Maria MD; Andersson, Rune MD, PhD; Olaison, Lars MD, PhD; Hogevik, Harriet MD, PhD

Author Information
doi: 10.1097/
  • Free



Positive blood cultures are keystones in diagnosing infective endocarditis. However, blood culture-negative endocarditis (CNE) is detected in both population-based epidemiologic studies and hospital studies of infective endocarditis. The proportion of CNE among infective endocarditis patients in different studies has been estimated at 1%–55% (6,18,20,30,38,40). The marked difference between studies has been explained by inadequate blood culturing techniques, antibiotics preceding blood culture, and different criteria for diagnosis.

To our knowledge, only 1 clinical epidemiologic study (18) has concentrated on patients with CNE during the last decade, a nationwide survey of infective endocarditis patients in France describing 88 patients with CNE. Previous antibiotic treatment was found in 48% of all cases, mortality was 15%, and a causative agent was found in 17% of all episodes. Fifty-one percent of the episodes were defined as definite according to the Duke criteria (11).

Infective endocarditis has been prospectively studied at the Department of Infectious Diseases, Göteborg University, Sweden, since 1984. The proportion of CNE was estimated earlier (1984–1988) to be from 12% to 21% (20,27), depending on which inclusion criteria were used. Antibiotic treatment preceded blood culture in 85% of the CNE episodes, compared with 35% of the episodes in the control group of blood culture-positive episodes. The study was expanded to include the Department of Infectious Diseases, Borås, Sweden, since 1989.

The Duke criteria (11) for classifying patients with infective endocarditis have been established during the last few years and have nearly replaced the Beth Israel criteria (von Reyn criteria) (39), although the latter have been modified to include findings at echocardiography (20,27). These classification systems have not been compared in consecutively collected CNE episodes, to our knowledge.

We conducted the present study with the following objectives: 1) to estimate the proportion of CNE in a group of patients with infective endocarditis recruited in a secondary and a tertiary hospital in Sweden; 2) to describe and analyze data from patients with CNE regarding suspected risk factors and demographic data; and 3) to evaluate the Duke and modified Beth Israel criteria in CNE patients.



The study took place at the Departments of Infectious Diseases at Sahlgrenska University Hospital, Göteborg, and Borås Hospital, Borås, Sweden. The former is a university clinic and a tertiary referral center and serves an urban population of about 600,000 inhabitants. The latter is a second-level referral hospital that serves a mixed urban and rural population of about 250,000 inhabitants.


The patients were prospectively recruited at the Göteborg clinic between 1984 and 1996 and at the Borås clinic from 1989 through 1996. The patients were transferred from other departments as cases of suspected infective endocarditis or fever of unknown origin; were referred by their general practitioner; or attended the outpatient clinics. All patients were adults (>18 yr).

We prospectively studied 233 consecutive episodes (217 patients) at the Göteborg clinic between 1984 and 1988. Of these episodes, 161 (149 patients; 81 male, 80 female episodes) were diagnosed as apparent infective endocarditis (20). Suspected infective endocarditis was identified in 517 episodes (460 patients; 267 male and 250 female episodes) in Göteborg between 1989 and 1996: 145 episodes (89 male, 56 female episodes) were classified according to the Duke criteria (11) as definite, 111 (59 male, 52 female episodes) as possible, and 261 (119 male, 142 female episodes) as rejects. Antibiotic combination therapy was initiated in 363 of the 517 episodes. Records from 237 episodes, in patients discharged with the diagnosis of infective endocarditis (International Statistical Classification of Diseases, ninth revision [ICD-9]) at the Göteborg clinic from 1989 through 1996, were also evaluated to find CNE patients not registered in the prospective study. In Borås, 70 episodes in 39 men and 31 women with an infective endocarditis diagnosis (ICD-9) were found.

Study design

The study was a nonrandomized descriptive study of consecutive patients with CNE treated as infective endocarditis according to a study protocol with uniform antibiotic treatment regimens, clinical evaluation procedures, and collection of specimens. Echocardiography was regularly done in the first week and, if indicated, repeated at the fourth week of treatment. Transesophageal echocardiography, available since 1991, was used when applicable. The location of infective endocarditis was determined by surgery, autopsy, or echocardiography. Patients who attended the Göteborg clinic were regularly examined by a dentist but those in Borås, only sporadically. Demographic data and data concerning suspected risk factors and other preceding events were extracted from records and registered in a standardized questionnaire.

All CNE episodes were evaluated with modified Beth Israel criteria (20,28) further modified to define only patients with the absence of bacterial growth at blood culture as culture negative (Table 1). The episodes were also categorized according to the Duke criteria (11) (Tables 2 and 3). Only lesions verified by pathologic and anatomic diagnosis (PAD) or culture were accepted as definite by the Duke pathologic criteria.

Modified Beth Israel criteria*, further modified so only patients with absence of bacterial growth at blood culture were accepted as culture negative
Duke criteria*
Definitions of terminology used in the Duke criteria*


Patients with absence of bacterial growth in blood culture and apparent infective endocarditis (1984–1988, Göteborg) were defined as having CNE.

Patients with absence of bacterial growth in blood culture and classified as definite or possible according to the Duke criteria (1989–1996, Göteborg) were defined as having CNE.

Patients with absence of bacterial growth in blood culture and discharged with an infective endocarditis diagnosis (ICD-9) and not registered in the prospective studies were defined as having CNE.

An episode was defined as an admission to hospital for treatment of infective endocarditis, irrespective of the time interval since any previous episode.

The first day of an episode was the first day of intravenous antibiotic combination treatment.

Heart failure was defined as an increased demand for diuretics.

Treatment mortality was defined as all deaths during treatment and up to 1 month posttreatment.

Blood culture and serology

According to the protocol, a standardized series of 3 blood cultures was drawn with 3 separate venous punctures with intervals of at least 20 minutes. In Göteborg, 1 aerobic and 1 anaerobic bottle (the biphasic model modified Castaneda system) were used at every blood culture. The bottles contained 2.5 mL of blood each in the first 21 months of the study (altogether 5 mL). This procedure was changed to an increased blood volume of 5 mL per bottle (altogether 10 mL). The BACT/ALERT System (Organon Teknika Corp., Durham, NC) was introduced for specimens from most Göteborg patients in the study in 1993. At every blood culture, 20 mL of blood was divided into 1 aerobe and 1 anaerobe bottle. FAN—aerobic and anaerobic media composed of brain heart infusion broth and Ecosorb (contains absorbent charcoal and Fullers earth)—media were introduced in 1995 instead of standard media. All blood cultures collected in Göteborg were incubated for 10 days.

Blood from the Borås patients was cultured in the SIGNAL system of OXOID (Basingstoke, UK). In this system, only 1 bottle containing 10 mL of blood was used on every blood culture occasion. The incubation time was 7 days. Coxiella burnetii serology was performed in some suspected cases, but not regularly.

Statistical methods

Values of quantitative variables were expressed as means, medians, and ranges. Confidence limits (95% CI) were given. The chi-square test was used to test the difference between proportions. A significance level of 0.05, 2-tailed test, was used. Yates correction was used when applicable.



During the study period there were 116 episodes of CNE in 115 patients (mean age, 63 yr; median age, 67 yr; range, 19–85 yr). Nineteen episodes were identified at the Borås clinic and 97 episodes (95 patients) at the Göteborg clinic. Three patients were referrals to the Göteborg clinic from other parts of western Sweden. The series consisted of 58 episodes in men (mean age, 63 yr; median age, 64 yr; range, 38–83 yr) and 58 episodes in women (mean age, 63 yr; median age, 70 yr; range, 19–85 yr). The actual proportions of CNE in the prospectively studied populations with infective endocarditis are presented in Table 4. Fifteen episodes identified retrospectively from records in Göteborg from 1984 through 1996 were not included in this calculation. The male/female ratio was 1.0 in the CNE group (Duke reject group excluded).

The proportion of CNE in the studied IE populations


The 116 episodes of CNE were divided into 20 definite, 80 possible, and 16 reject according to the Duke criteria (11) and were also evaluated with the modified Beth Israel criteria (20,28) (Table 5). The alternative diagnoses in the Duke reject episodes are shown in Table 6.

Duke* compared with modified Beth Israel criteria in 116 episodes of culture-negative infective endocarditis
Alternative diagnosis in 16 CNE patients classified as reject by the Duke criteria*

The proportion of CNE was estimated to be 6.4% (16/251) (95% CI 3.8–9.9) in the Duke definite group and 40% (65/163) (95% CI 33–47) in the Duke possible group (1984–1996, Göteborg prospective episodes). The distribution of major and minor criteria in the Duke possible group is presented in Figure 1. All consecutive prospective episodes of suspected infective endocarditis in Göteborg, 1989–1996, were subdivided into an annual incidence and sorted with the Duke criteria (Figure 2).

Distribution of major and minor criteria in 80 Duke possible episodes.
Göteborg, Sweden, 1989–1996: 517 consecutive episodes of suspected infective endocarditis evaluated with the Duke criteria (CNE = blood culture negative-endocarditis).


Data concerning symptom duration, doctor’s delay, and time to treatment are presented in Table 7. The median highest recorded temperature before infective endocarditis treatment was 39 °C (range, 36.8°-40.3 °C).

Days with symptoms until hospitalization and start of IE treatment

Predisposing factors and risk factors

Data concerning suspected risk factors are presented in Table 8. Thirty percent of the episodes occurred after previous open heart surgery. The median time between surgery and the start of infective endocarditis symptoms was 26 months (mean, 62 mo; range, 0–385 mo).

Predisposing factors in 116 episodes of culture-negative infective endocarditis

Twenty-four percent of the episodes occurred in prosthetic valve carriers. The median time from surgery to first symptom of infective endocarditis was 23 months (mean, 34 mo; range, 0–162 mo). Six of the episodes could be defined as early prosthetic valve endocarditis (<2 mo after surgery), and 23 as late prosthetic valve endocarditis (>2 mo after surgery).

Noncardiac surgery had been performed less than 90 days before the first symptom of infective endocarditis in 6 episodes. The surgical procedures were hip surgery, neural decompression, fasciotomy of the thigh, incision of a vaginal hematoma, urethra dilatation, and drainage of a peritoneal abscess. Four patients developed infective endocarditis during hospital treatment; the reasons for their initial admission were valvular surgery, ventricular septum defect repair, coronary by-pass, and childbirth, respectively. In 9 episodes, patients presented with wounds or abscesses. Two patients had postoperative infections after thoracotomy, 6 had superficial wounds, and 1 had a deep abscess. One patient had a urinary tract infection preceding the first symptoms of infective endocarditis.

Dental status

In 22 episodes (19%), the patient had visited the dentist less than 90 days before the first day of illness (see Table 8). The dentist examined the patients during 74 episodes and found abnormal dental status in 39 (52%).

Cardiac murmurs, heart failure, and embolism

Data concerning heart failure are shown in Tables 8 and 11. In 46 (40%) episodes heart failure was noted at the start of infective endocarditis therapy, with deterioration in 27 (23%) of the episodes. In another 15 (13%) of the episodes, heart failure developed during infective endocarditis therapy (see Table 11). Data concerning murmurs are presented in Table 9; 89% of patients presented a cardiac murmur at admission.

Clinical and diagnostic events and surgery during IE treatment (n = 116)
Cardiac murmurs (systolic and/or diastolic) (n = 116)

In 19 episodes (16%) clinical embolism was present at the start of infective endocarditis therapy (see Table 8); 11 of the emboli passed to the central nervous system. In another 9 (8%) episodes emboli occurred after the initiation of infective endocarditis therapy (see Table 11).


Left-sided infective endocarditis was diagnosed in 94 episodes (aortic valve, 49; mitral valve, 35; combined aortic and mitral valve, 9; and combined aortic and patch infection, 1) (Table 10). One patient had a pacemaker infection, and 2 patients had mural infections. In the remaining 19 episodes no apparent location was found at echocardiography. Transesophageal echocardiography was used in 6 of these episodes.

Location of infective endocarditis determined by echocardiography, surgery, or autopsy (n = 116)


Transesophageal echocardiography was performed in 39 episodes and transthoracic echocardiography in 106 episodes. Altogether 114 episodes were investigated with echocardiography during the first week of treatment. The use of transesophageal echocardiography and transthoracic echocardiography was studied especially in the Duke possible group since transesophageal echocardiography became available in 1991. Five (38%) of the 13 episodes with 2 minor criteria were investigated with transesophageal echocardiography, compared with 19 of 24 (79%) in the 1 major and 2 minor criteria group during this period (p = 0.03).

Blood culture, serology, and antibiotic therapy

A median of 3 blood cultures was done (mean, 5; range, 2–16). The exact number of blood cultures before infective endocarditis treatment was unknown in 7 patients. One patient had serologically verified Q-fever endocarditis. In 34 (29%) episodes the patients were on antibiotic treatment, and in another 19 (16%) episodes antibiotic pretreatment was given less than 60 days (mean, 11 d; median, 7 d; range, 1–41 d) before the first blood culture was done. The median duration of antibiotic treatment in the pretreated episodes was 10 days (mean, 9 d; range, 1–33 d).

The proportion of episodes with ongoing or previous antibiotic treatment at the time of the first blood culture in all CNE episodes was 45% (95% CI 37–55). The proportion of antibiotic pretreatment at blood culture was 45% (95% CI 24–69) in the definite group, 46% (95% CI 35–57) in the possible group, and 44% (95% CI 20–64) in the reject group (Duke criteria). The Duke possible group was further divided into 6 subgroups. Antibiotic pretreatment at blood culture was present in 43% (17/39) of episodes in the 1 major and 2 minor criteria group, in 100% (4/4) of episodes in the 1 major and 1 minor criteria group, in 14% (1/7) of episodes in the 3 minor criteria group, and in 55% (15/27) of the episodes in the 2 minor criteria group. There was no antibiotic pretreatment in the groups with 1 minor or 4 minor criteria.

Valvular culture was performed in 9 of 16 surgically removed valves. One valve was culture positive (Capnocytophaga canimorsus). One valve was cultured postmortem with growth of Klebsiella and E. coli (see Table 11).


Surgery data are presented in Table 11. The median time to surgery from start of infective endocarditis therapy was 9 days (mean, 15 d; range, 0–40 d). Seventeen patients underwent early open heart surgery in 18 episodes, with a male/female ratio of 0.89 (8/9).

Mortality and follow-up

Eight patients (7%) died during infective endocarditis therapy or less then 1 month posttreatment (see Table 11). The median time from start of infective endocarditis therapy to death was 29 days (mean, 38 d; range, 5–80 d). One of the deceased was surgically treated. Three patients had autopsies, and all had PAD-proven endocarditis. Four of the 8 patients who died were defined as Duke definite, including the 3 autopsied cases.

Four patients suffered a new episode of culturepositive infective endocarditis during follow-up. Ninety-seven of 106 surviving patients had a median follow-up time of 42 months (mean, 49 mo; range, 1–180 mo).



The rather high proportion of CNE in the studied population corresponds with studies from Finland (Duke criteria, definite 23% versus possible 66%) (16), Denmark (Beth Israel definite and possible 12%) (3), and Sweden (Duke criteria, definite and possible 23%) (23). A study from the Netherlands found 1% CNE in the studied infective endocarditis group (38). This low proportion is due to selection bias, the participants being collected on the basis of positive blood cultures. Studies from developing countries such as India and Morocco present much higher proportions of CNE (55%–58%) (4,6). The relatively high rate of CNE in the cited studies from developed countries does not support the statement that CNE is a rare diagnosis (<5%) in modern infective endocarditis studies (37). The low proportion (6.4%) of CNE among Duke definite episodes in the present study may be an underestimation due to the classification system.

There was no trend in the annual incidence of CNE among the patients with infective endocarditis, even though transesophageal echocardiography and better blood culturing techniques became accessible during the study period (see Figure 2). The use of echocardiography on wide indications may add to a still high proportion of CNE among infective endocarditis patients even though the sensitivity and specificity of the method is high (7,26). Treatment and suspicion of infective endocarditis on wide indications probably increase the proportion of CNE in western countries, but early case detection is important to attain a low mortality from infective endocarditis (3,20).

The male/female ratio in the CNE episodes (1.0) differs from the French CNE study (18) by a ratio of 3 (which may be due to selection bias of the reporting system in the study) and from other general clinical and population-based studies of infective endocarditis by a ratio of 1.3–2.5 (3,6,8,16,25,35,38,40). The even sex distribution in our study corresponds with the series described by von Reyn (39) and the Duke Endocarditis Service (11).

The median age of all patients in the present study (67 yr) was higher than that in the French CNE study (18) (53 yr). The median age in women was even higher (70 yr). These age and gender differences may be due to the prospective character of our study, with early diagnostic procedures in suspected cases of infective endocarditis. An aged population with a majority of elderly women with good access to hospital care, as in Sweden during the studied period, probably contributes to the group of infective endocarditis patients with an increase of the median age and a decrease of the male/female ratio


The modified Beth Israel criteria and the Duke criteria were compared (see Table 5) (11,39). Only 20 episodes fulfilled the criteria for Duke definite, and 16 for Duke reject. The small number of Duke definite episodes, although more than the episodes classified as Beth Israel definite, may partly depend on strict use of the Duke criteria in patients who have undergone surgery or died. Valvular culture and/or PAD were not performed in 4 patients who underwent surgery, showed macroscopic signs of infective endocarditis, and were classified as possible infective endocarditis. Thus the Duke criteria may be insensitive when used strictly, not accepting macroscopic evidence of infective endocarditis.

The Duke possible group is the largest, accounting for 80 (69%) of the 116 CNE episodes. It is heterogeneous, including patients with a high likelihood of infective endocarditis together with patients with a very low likelihood but with infective endocarditis treatment >4 days and no other obvious diagnosis. In our study, 39 of the episodes in the Duke possible group fulfilled 1 major and 2 minor criteria. Authors of a Swiss study (33) estimated the power of every Duke criterion. The major microbiologic criterion had the greatest impact (53%), followed by the major endocardial criterion (34%), when Duke definite and possible episodes were distinguished. In CNE the major endocardial criterion is very important in the absence of positive blood cultures. Only 38 of the CNE episodes in our study were investigated with transesophageal echocardiography. This more sensitive method may have been able to detect more episodes with the major endocardial criterion. The interobservational variation in echocardiographic investigations also stresses the risk of misclassification (15). Investigation of immunologic phenomena such as rheumatoid factor was not done, according to the study protocol. A positive rheumatoid factor test could have had an impact, classifying more episodes as Duke definite (33). Neither Coxiella burnetii nor Bartonella serology was done routinely as the incidences of Q-fever and Bartonella endocarditis were presumed to be low in the studied population.

In 43% (17/39) of the Duke possible episodes with 1 major and 2 minor criteria, ongoing or previous antibiotic treatment was found at the first blood culture. There was no significant difference in antibiotic pretreatment in this group compared with the other studied groups or subgroups. Our study did not add support to the suggested need for a new criterion of antibiotic pretreatment at blood culture in CNE patients with 1 major and 2 minor criteria to classify these episodes as Duke definite. It is necessary to have a larger study group with CNE patients with either surgical valve replacement or autopsy where PAD and valvular culture is performed consistently, to study the influence of antibiotic pretreatment in CNE.

In 27 of the 80 possible episodes, only 2 minor criteria were present (fever >38.0 °C and a predisposing heart condition). Valvular prosthesis, an earlier episode of infective endocarditis, atherosclerotic valves or congenital heart disorder, and fever were noted. Transesophageal echocardiography was done (since it became available in 1991) in 5 of 13 patients in the 2 minor criteria group compared with 19 of 24 in the 1 major 2 minor group during this period (p = 0.03). It is too simple to classify patients with a valvular prosthesis and fever as Duke possible. We agree with the proposed modifications of the Duke criteria to increase the specificity of the possible class and that at least 3 minor criteria or 1 major and 1 minor criterion are necessary to classify an episode as possible infective endocarditis (24). Transesophageal echocardiography should be used when diagnosing suspected CNE to distinguish patients in the possible and reject groups.

Earlier evaluations of the Duke criteria have studied the sensitivity and the specificity in different patient groups (9,11,17,34,36). In our study the Duke criteria were more sensitive compared with the modified Beth Israel criteria in identifying the definite episodes (20 versus 13 episodes, respectively). However, it is more useful to compare the modified Beth Israel criteria definite and probable group (the latter including patients who did not undergo surgery or die) with the Duke definite group. This comparison, on the other hand, shows an increased sensitivity in the Beth Israel criteria (28 versus 20 episodes, respectively).

In the present study, the modified Beth Israel criteria distinguish 62 episodes as reject, indicating a better specificity compared with 16 reject episodes identified with the Duke criteria. This better discriminating ability is due to the fact that the Beth Israel criteria disregard the time of infective endocarditis treatment (more or less than 4 d) and do not require a firm alternative diagnosis to reject infective endocarditis, which sometimes is impossible in a clinical situation. Studies of the positive predictive value of the Duke criteria in different study populations are wanted.

We have not calculated the sensitivity and the specificity of the Duke criteria, as different methods were used to collect the patients. The alternative diagnosis in 16 reject episodes is presented (see Table 6) just to stress the importance of trying to rule out other important diagnoses in patients with suspected infective endocarditis (13).


The mean duration from the first symptom to admission was 23 days, which is shorter than in earlier reports studying CNE (18,30) (see Table 7). The median time to admission was considerably shorter, 10 days. A more important variable is the median time from first symptoms to infective endocarditis treatment, 14 days in this study. This is comparable to data from modern studies of infective endocarditis in general, with a delay of 11–13 days (3,20).

Cardiac murmurs and embolism

In 89% of episodes a cardiac murmur was present (see Table 9). Murmurs have a great influence in prompting for diagnostic procedures, especially in episodes with a new murmur, 37% of the cases in our study. A new cardiac murmur is a major endocardial criterion that is of great importance when classifying infective endocarditis episodes with the Duke criteria.

In 16% of episodes, embolism preceded attendance to hospital. More than 50% of these emboli went to the central nervous system (CNS). CNS emboli often have serious symptoms and may have increased the suspicion of infective endocarditis in those patients.

Predisposing factors and risk factors

Data concerning earlier prosthetic surgery, earlier infective endocarditis, and location (see Table 8) correspond with previous studies (18). Alcohol abuse has been proposed as a risk factor in infective endocarditis (1). In the present study, 8% involved habitual drinkers, a higher proportion than earlier described in the Göteborg study (19). Homeless alcoholics with suspected or verified body lice infestation have been associated with Bartonella quintana infective endocarditis (5). No patient in this study was homeless or was noted to have body lice. The antibiotic effect of neuroleptic drugs has been shown in in vitro and in vivo experiments (22). The influence of neuroleptic drug treatment in CNE is probably low, as only 2% of the CNE patients used this medication.

Antibiotic treatment and blood culture

In the present study, antibiotic treatment preceded blood culture in 45% (95% CI 36–55) of all episodes of CNE, which corresponds with other studies that report previous antibiotic treatment in 48%–50% of episodes (18,23). It differs from a study from Iowa (30), where antibiotic pretreatment was present at blood culture in 62% (52 episodes) of CNE episodes. We analyzed the episodes with ongoing and completed antibiotic pretreatment together in relation to blood culture because the influence of antibiotic pretreatment has been found to be important (29). The median time from withdrawal of antibiotic treatment to blood culture was 7 days in the patients with antibiotic pretreatment, who accounted for 16% of the episodes. The length of antibiotic pretreatment has been proposed to have a great influence on inhibition of bacterial growth on blood culture (37). The median duration of antibiotic therapy was 10 days in our study.

The observation of no difference in the proportions of antibiotic pretreatment at blood culture between the Duke definite, possible, and reject groups underlines that antibiotic pretreatment is not the sole explanation for blood culture negativity in infective endocarditis. The thesis of continuous bacteremia in infective endocarditis as stated by Beeson and colleagues (2) in the classical work from the 1940s, where 6 patients with terminal infective endocarditis disease (5 patients with symptoms for 4–6 months and 1 patient with symptoms for 4 days) were investigated, needs to be reconsidered when patients with infective endocarditis are diagnosed earlier.

In the beginning of the present study, the blood volume used for blood culture was rather small (5 mL per puncture), but it has increased steadily and is now always 20 mL per culture. In spite of this, there was no significant change in the proportion of CNE among infective endocarditis patients in Göteborg during the years 1989–1996 (blood volume change from 10 to 20 mL). Problems in blood culturing, such as time from venipuncture to start of incubation, have not been analyzed but could have had an impact on the number of CNE patients.

The blood cultures in this study were analyzed in routine practice and were not always incubated more than 1 week. Fastidious bacteria such as the HACEK (Hemophilus aphrophilus, Actinobacillus acinetocomitans, Cardiobacterium hominis, Eikenella, Kingella kingae) group and Granulicatella may have been undetected. α-Streptococci may be inhibited from growing in blood culture by antibiotic pretreatment (20,29,43). The probability of α-streptococci as etiologic agents in some of the episodes of CNE is rather high, as the patients were on antibiotic treatment in more than 40% of the CNE episodes.

The FAN media have an antimicrobial neutralizing effect (41). The observation time in the present study was too short to evaluate a decrease in the proportion of CNE episodes after the introduction in 1995. To improve the diagnostic accuracy it is important that excised valves in all cases of suspected infective endocarditis are subjected to PAD investigation and that the necessary staining is performed to detect, for example, Whipple disease, Bartonella infection, and Q-fever (31). Tropheryma whippelii and Bartonella have been described as etiologic agents in infective endocarditis (10,14,32). The excised valves in the present study were not investigated for Whipple disease. New techniques such as the polymerase chain reaction (PCR) may improve our understanding of CNE and have been helpful in diagnosing agents such as Tropheryma whippelii and Bartonella in valves (12,21,42).


The rate of early valvular surgery up to 1 month after the end of antibiotic infective endocarditis treatment was rather low in the CNE group (14%) and in the Duke definite group (15%) compared with the previously mentioned French study (18). Our study is a prospective study including all suspected cases, and thus differs from the French study which is based on reporting from physicians which may include an increased reporting of surgically proven cases. The low rate of surgery perhaps reflects the fact that patients with suspected infective endocarditis were treated following a scheme which earlier had yielded good results (20,28).


The mortality during treatment was low (7%) in our study as a whole compared with earlier series (14%–32%) (18,30). The mortality was 20% in the Duke definite episodes. There was an overrepresentation of Duke definite episodes among those who died (4/8). More episodes may have been defined as Duke definite if autopsies had been done regularly, as only 3 of 8 (38%) patients who died had autopsies.


The proportion of CNE is about 20% in a large group of prospectively investigated episodes of infective endocarditis in Scandinavia. Fewer than 50% of all episodes of CNE can be explained by antibiotic pretreatment. Among 116 episodes of CNE studied, 20 were definite, 80 possible, and 16 reject by the Duke criteria versus 13 definite, 15 probable, 27 possible, and 61 reject by the modified Beth Israel criteria. The Duke criteria probably both underestimate the real frequency of definite CNE episodes and overestimate the frequency of possible CNE episodes. This study supports the importance of performing PAD and valvular culture in every surgically treated episode or autopsy of suspected infective endocarditis, especially in episodes that are blood culture negative, to refine the diagnosis and treatment of infective endocarditis.


We acknowledge the help of Per Nordin, Skaraborg Institute, Skövde, Sweden, for statistical review.


1. Bayliss R, Clarke C, Oakley CM, Somerville W, Whitfield AG, Young SE. Incidence, mortality and prevention of infective endocarditis. J R Coll Physicians Lond 20: 15–20, 1986.
2. Beeson PB, Brannon ES, Warren JV. Classics in infectious diseases. Observations on the sites of removal of bacteria from the blood of patients with bacterial endocarditis. Rev Infect Dis 7: 565–73, 1985.
3. Benn M, Hagelskjaer LH, Tvede M. Infective endocarditis, 1984 through 1993: A clinical and microbiological survey. J Intern Med 242: 15–22, 1997.
4. Bennis A, Zahraoui M, Azzouzi L, Soulami S, Mehadji BA, Tahiri A, Chraibi N. [Bacterial endocarditis in Morocco]. Ann Cardiol Angeiol Paris 44: 339–44, 1995.
5. Brouqui P, LaScola B, Roux V, Raoult D. Chronic Bartonella quintana bacteremia in homeless patients. N Engl J Med 340: 184–9, 1999.
6. Choudhury R, Grover A, Varma J, Khattri HN, Anand IS, Bidwai PS, Wahi PL, Sapru RP. Active infective endocarditis observed in an Indian hospital 1981–1991. Am J Cardiol 70: 1453–8, 1992.
7. Daniel W, Mugge A, Grote J, Nonnast-Daniel B. Evaluation of endocarditis and its complications by biplane and multiplane transesophageal echocardiography. Am J Cardiac Imaging 9: 100–5, 1995.
8. Delahaye F, Goulet V, Lacassin F, Ecochard R, Selton-Suty C, Hoen B, Etienne J, Briancon S, Leport C. Characteristics of infective endocarditis in France in 1991: A 1-year survey. Eur Heart J 16: 394–401, 1995.
9. Dodds GA, Sexton DJ, Durack DT, Bashore TM, Corey GR, Kisslo J. Negative predictive value of the Duke criteria for infective endocarditis. Am J Cardiol 77: 403–7, 1996.
10. Drancourt M, Mainardi JL, Brouqui P, Vandenesch F, Carta A, Lehnert F, Etienne J, Goldstein F, Acar J, Raoult D. Bartonella (Rochalimae) quintana endocarditis in three homeless men. New Engl J Med 332: 419–23, 1995.
11. Durack DT, Lukes AS, Bright DK. New criteria for diagnosis of infective endocarditis: Utilization of specific echocardigraphic findings. Am J Med 96: 200–9, 1994.
12. Goldenberger D, Kunzli A, Vogt P, Zbinden R, Altwegg M. Molecular diagnosis of bacterial endocarditis by broad-range PCR amplification and direct sequencing. J Clin Microbiol 35: 2733–9, 1997.
13. Goodfield NE, Bhandari S, Plant WD, Morley-Davies A, Sutherland GR. Cardiac involvement in Wegener’s granulomatosis. Br Heart J 73: 110–5, 1995.
14. Gubler JG, Kuster M, Dutly F, Bannwart F, Krause M, Vogelin HP, Garzoli G, Altwegg M. Whipple endocarditis without overt gastrointestinal disease: Report of four cases. Ann Intern Med 131: 112–6, 1999.
15. Heinle S, Wilderman N, Harrison JK, Waugh R, Bashore T, Nicely LM, Durack D, Kisslo J. Value of transthoracic echocardiography in predicting embolic events in active infective endocarditis. Duke Endocarditis Service. Am J Cardiol 74: 799–801, 1994.
16. Heiro M, Nikoskelainen J, Hartiala JJ, Saraste MK, Kotilainen PM. Diagnosis of infective endocarditis. Sensitivity of the Duke vs von Reyn criteria. Arch Intern Med 158: 18–24, 1998.
17. Hoen B, Beguinot I, Rabaud C, Jaussaud R, Selton-Suty C, May T, Canton P. The Duke criteria for diagnosing infective endocarditis are specific: Analysis of 100 patients with acute fever or fever of unknown origin. Clin Infect Dis 23: 298–302, 1996.
18. Hoen B, Selton-Suty C, Lacassin F, Etienne J, Briancon S, Leport C, Canton P. Infective endocarditis in patients with negative blood cultures: Analysis of 88 cases from a one-year nationwide survey in France. Clin Infect Dis 20: 501–6, 1995.
19. Hogevik H. Infective endocarditis: Epidemiology and clinical presentation [dissertation]. Goteborg, Sweden: Goteborg University, p 35, 1996.
20. Hogevik H, Olaison L, Andersson R, Lindberg J, Alestig K. Epidemiologic aspects of infective endocarditis in an urban population. A 5-year prospective study. Medicine (Baltimore) 74: 324–39, 1995.
21. Jalava J, Kotilainen P, Nikkari S, Skurnik M, Vanttinen E, Lehtonen OP, Eerola E, Toivanen P. Use of the polymerase chain reaction and DNA sequencing for detection of Bartonella quintana in the aortic valve of a patient with culture-negative infective endocarditis. Clin Infect Dis 21: 891–6, 1995.
22. Kristiansen JE. The antimicrobial activity of psychotherapeutic drugs and stereo-isomeric analogues. Dan Med Bull 37: 165–82, 1990.
23. Kurland S, Enghoff E, Landelius J, Nystrom S-O, Hambraeus A, Friman G. A 10-year retrospective study of infective endocarditis at a university hospital with special regard to the timing of surgical evaluation in S. viridans endocarditis. Scand J Infect Dis 31: 87–91, 1999.
24. Li JS, Nettles RE, McCarty DE, Fowler V, Roe M, Ryan T, Sexton DJ, Corey GR. Clarifications of the Duke criteria for Infective endocarditis. In: Fourth International Symposium on Modern Concepts in Endocarditis and Cardiovascular Infections, Yverdon-les-Bains, Switzerland, 1997.
25. Lien EA, Solberg CO, Kalager T. Infective endocarditis 1973–1984 at the Bergen University Hospital: Clinical feature, treatment and prognosis. Scand J Infect Dis 20: 239–46, 1988.
26. Lowry RW, Zoghbi WA, Baker WB, Wray RA, Quinones MA. Clinical impact of transesophageal echocardiography in the diagnosis and management of infective endocarditis. Am J Cardiol 73: 1089–91, 1994.
27. Olaison L, Hogevik H. Comparison of the von Reyn and Duke criteria for the diagnosis of infective endocarditis: A critical analysis of 161 episodes. Scand J Infect Dis 28: 399–406, 1996.
28. Olaison L, Hogevik H, Myken P, Oden A, Alestig K. Early surgery in infective endocarditis. QJM 89: 267–78, 1996.
29. Pazin GJ, Saul S, Thompson ME. Blood culture positivity; suppression by outpatient antibiotic therapy in patients with bacterial endocarditis. Arch Intern Med 142: 263–8, 1982.
30. Pesanti EL, Smith IM. Infective endocarditis with negative blood cultures. An analysis of 52 cases. Am J Med 66: 43–50, 1979.
31. Raoult D. Afebrile blood culture-negative endocarditis [editorial; comment]. Ann Intern Med 131: 144–6, 1999.
32. Raoult D, Fournier PE, Drancourt M, Marrie T, Etienne J, Cosserat J, Cacoub P, Poinsignon Y, Leclerq P, Sefton A. Diagnosis of 22 new cases of Bartonella endocarditis. Ann Intern Med 125: 645–52, 1996.
33. Rognon R, Kehtari R, Francioli P. Individual value of each of the Duke criteria for the diagnosis of infective endocarditis. Clin Microbiol Infect 5: 396–403, 1999.
34. Sekeres MA, Abrutyn E, Berlin JA, Kaye D, Kinman JL, Korzeniowski OM, Levison ME, Feldman RS, Strom BL. An assessment of the usefulness of the Duke criteria for diagnosing active infective endocarditis. Clin Infect Dis 24: 1185–90, 1997.
35. Steckelberg JM, Melton 3rd, LJ Ilstrup DM, Rouse MS, Wilson WR. Influence of referral bias on the apparent clinical spectrum of infective endocarditis. Am J Med 88: 582–8, 1990.
36. Stockheim JA, Chadwick EG, Kessler S, Amer M, Abdel-Haq N, Dajani AS, Shulman ST. Are the Duke criteria superior to the Beth Israel criteria for the diagnosis of infective endocarditis in children? Clin Infect Dis 27: 1451–6, 1998.
37. Tunkel AR. Evaluation of culture-negative endocarditis. Hosp Pract (Off Ed) 28( 2a): 59–62, 65–6, 1993.
38. van der Meer JT, Thompson J, Valkenburg HA, Michel MF. Epidemiology of bacterial endocarditis in The Netherlands. I. Patient characteristics. Arch Intern Med 152: 1863–8, 1992.
39. Von Reyn CF, Levy BS, Arbeit RD, Friedland G, Crumpacker CS. Infective endocarditis: An analysis based on strict case definitions. Ann Intern Med 94: 505–18, 1981.
40. Watanakunakorn C, Burkert T. Infective endocarditis at a large community teaching hospital, 1980–1990. A review of 210 episodes. Medicine (Baltimore) 72: 90–102, 1993.
41. Weinstein MP, Mirrett S, Reimer LG, Wilson ML, Smith-Elekes S, Chuard CR, Joho KL, Reller LB. Controlled evaluation of BacT/Alert standard aerobic and FAN aerobic blood culture bottles for detection of bacteremia and fungemia. J Clin Microbiol 33: 978–81, 1995.
42. Wendler D, Mendoza E, Schleiffer T, Zander M, Maier M. Tropheryma whippelii endocarditis confirmed by polymerase chain reaction. Eur Heart J 16: 424–5, 1995.
43. Werner AS, Cobbs CG, Kaye D, Hook EW. Studies on the bacteremia of bacterial endocarditis. JAMA 202: 199–203, 1967.

Cited By

This article has been cited 1 time(s).

Culture-Positive and Culture-Negative Endocarditis in Patients With Cancer: A Retrospective Observational Study, 1994-2004
Yusuf, S; Ali, SS; Swafford, J; Durand, J; Bodey, GP; Chemaly, RF; Kontoyiannis, DP; Tarrand, J; Rolston, KV; Yeh, E; Raad, II; Safdar, A
Medicine, 85(2): 86-94.
PDF (110) | CrossRef
© 2003 Lippincott Williams & Wilkins, Inc.