Changing Profile of Infective Endocarditis: Results of a 1-Year Survey in France

Context Since the first modern clinical description of infective endocarditis (IE) at the end of the 19th century, the profile of the disease has evolved continuously, as highlighted in epidemiological studies including a French survey performed in 1991.

Objective To update information gained from the 1991 study on the epidemiology of IE in France.

Design and Setting Population-based survey conducted from January through December 1999 in all hospitals in 6 French regions representing 26% of the population (16 million inhabitants).

Patients Three hundred ninety adult inpatients diagnosed with IE according to Duke criteria.

Main Outcome Measures Incidence of IE; proportion of patients with underlying heart disease; clinical characteristics; causative microorganisms; surgical and mortality outcomes.

Results The annual age- and sex-standardized incidence was 31 (95% confidence interval [CI], 28-35) cases per million, not including the region of New Caledonia, which had 161 (95% CI, 117-216) cases per million. There was no previously known heart disease in 47% of the cases. The proportion of prosthetic-valve IE was 16%. Causative microorganisms were: streptococci, 48% (group D streptococci, 25%; oral streptococci, 17%, pyogenic streptococci, 6%); enterococci, 8%; Abiotrophia species, 2%; staphylococci, 29%; and other or multiple pathogens, 8%. Blood cultures were negative in 9% and no microorganism was identified in 5% of the cases. Early valve surgery was performed in 49% of the patients. In-hospital mortality was 16%. Compared with 1991, this study showed a decreased incidence of IE in patients with previously known underlying heart disease (20.6 cases per million vs 15.1 cases per million; P<.001); a smaller incidence of oral streptococcal IE (7.8 cases per million vs 5.1 cases per million; P<.001), compensated by a larger proportion of IE due to group D streptococci (5.3 cases per million vs 6.2 cases per million; P = .67) and staphylococci (4.9 cases per million vs 5.7 cases per million; P = .97); an increased rate of early valve surgery (31.2% vs 49.7%; P<.001); and a decreased in-hospital mortality rate (21.6% vs 16.6%; P = .08).

Conclusion Although the incidence of IE has not changed, important changes in disease characteristics, treatment, and outcomes were noted.

Recent studies have shown that the annual incidence of infective endocarditis (IE) is stable between 15 and 60 cases per million.^1-4 However, since the first modern clinical description of IE by Osler,⁵ the profile of the disease has continuously evolved. This fact was highlighted in different epidemiological studies,^6,7 including a French survey of IE performed in 1991, in which the main results, compared with a similar study conducted in 1983,⁸ showed (1) stable annual incidence at 24 cases per million; (2) increased age of onset; (3) decreasing proportion of IE on abnormal native valve, compensated by an increased proportion of prosthetic valve IE and native valve IE in patients with previously unrecognized predisposing conditions; and (4) a changing microbiologic profile, with an increased proportion of group D streptococci and staphylococci.^1,9

To update the description of epidemiological, clinical, microbiologic, and outcome characteristics of IE in France, the same group who had conducted the 1991 survey carried out a new survey.

This population-based study was conducted prospectively in 1999 in 6 French regions: Ile de France, Lorraine, Rhône-Alpes, Franche-Comté, Marne, and New Caledonia. The first 3 had taken part in the previous study. The population of the 6 regions (16 million inhabitants) comprises 26% of the French population. Only those patients with a first hospitalization between January 1, 1999, and December 31, 1999, were included in this analysis.

The study was distributed by mail to all physicians working in hospitals of these regions who were either likely to take care of patients with IE (ie, specialists in infectious diseases, intensive care, internal medicine, cardiology, and cardiac surgery) or were echocardiographers or microbiologists. Physicians were asked to fax a notification form to the study coordinating center for each suspected case of IE. They were reminded of this study on a regular basis by mail. For each patient older than 18 years and living in one of the study regions, specific case report forms were sent to both the attending physician and the microbiologist.

On the clinical report form, the following information was collected: sex, date of birth, place of residence, date of first hospitalization, transfer from or to another facility, history of heart disease, procedures and situations at risk for IE, comorbidities, clinical signs and symptoms, laboratory and imaging examinations, echocardiographic data, microbiological data, medical and surgical treatment, and outcome. Investigators were provided with standard definitions of all variables. For quantitative parameters, the physicians were asked to provide the maximal or highest value recorded during the course of the disease. Location of IE was determined according to echocardiographic and/or surgical findings. Patients were informed of the study but did not need to give consent.

Microbiologists were asked to fill out a form that included information on the identification and susceptibility of causative microorganism (total number and number of positive blood cultures, results of valve cultures, and serological tests) and to send the isolated strains to a single centralized core laboratory (Laboratoire de Bactériologie, Centre Hospitalier Universitaire Bichat-Claude Bernard, Paris). Identification of streptococci and staphylococci was subsequently confirmed by the Centre National de Référence des Streptocoques (Paris) and the Centre National de Référence des Toxémies à Staphylocoques (Lyon), respectively. In addition, for patients whose first blood cultures remained negative, attending physicians were asked to complete specific analyses including additional sets of blood cultures on enriched media and serological tests for Coxiella burnetii, Bartonella species, chlamydiae, and Legionella, which were performed in national reference laboratories (Centre National de Référence des Rickettsies [Marseilles] and the Centre National de Référence des Legionella [Lyon]), and for Brucella, Mycoplasma pneumoniae, Candida, and Aspergillus. Results of all serological tests were validated by an expert microbiologist (F.V.) according to previously published recommendations.¹⁰

Once completed, all case report forms were checked and validated by 2 expert investigators who had not been involved in the care of the corresponding case. These investigators were responsible for validating the diagnosis according to the Duke criteria.¹¹ Only definite cases of IE were included in the study. This study was approved by the institutional review board of the Centre Hospitalier Universitaire de Besançon.

Incidence rates, expressed as number of cases per million inhabitants, were calculated by dividing the number of cases recorded within the study year by the number of persons residing in the study regions and aged 15 years or older. Population references were obtained for France and New Caledonia from the 1999¹² and 1996¹³ censuses, respectively. Results of this survey were compared with those of the 1991 survey within the 3 areas that were common to both studies in terms of crude and standardized incidences, group-specific incidences (underlying heart disease, causative microorganism), surgical treatment, and mortality rates. Comparison was restricted to patients older than 18 years and fulfilling the Beth Israel criteria modified with echocardiography for definite, probable, or possible endocarditis.¹ The latter were used because the Duke criteria were not available in 1991 and could not be applied retrospectively. In both periods, incidence was standardized to the sex-by-age distribution of the 1990 French population.¹⁴ The method described by Fay and Feuer¹⁵ was used for calculation and comparison of 95% confidence intervals (CIs) of standardized incidence rates.

Quantitative variables are expressed as their mean (SD) with 95% CIs or median (interquartile range). Qualitative variables are expressed as percentages. For intergroup comparison, we used ad hoc methods (1-way analysis of variance, Pearson χ² test). All statistical analyses were performed using SAS version 8.02 software (SAS Institute Inc, Cary, NC) and .05 was the level of significance.

A total of 925 notifications led to the identification of 819 patients with a putative diagnosis of IE. Among these, 429 were excluded for the following reasons: hospitalization outside the study period (n = 178); living outside the study regions (n = 131); failure to fulfill Duke criteria (n = 106); case report form not returned (n = 11); and younger than 18 years (n = 3). The present report is based on the remaining 390 eligible cases of definite IE. There were 277 men and 113 women with a mean (SD) age of 59.5 (17.2) years (range, 16-95 years).

The crude annual incidence of IE was 30 (95% CI, 27-33) cases per million population. The age- and sex-standardized annual incidence was 31 (95% CI, 28-35) cases per million for France not including New Caledonia, and 161 (95% CI, 117-216) for New Caledonia (mean, 44; 95% CI, 39-50 in men and mean, 17; 95% CI, 14-20 in women). Incidence increased dramatically in patients older than 50 years and peaked at 145 cases per million in men between 70 and 80 years (Figure 1).

The distribution of underlying heart disease is summarized in Table 1. Of note, 185 patients (47%) had no previously known heart disease. Thirty-five patients (9%) had a history of prior IE (mean [SD] time interval between the 2 episodes, 7.5 [4.0] years; range, 0-60 years). Twenty-eight patients (7%) had a pacemaker; 12 of these also had valvular disease, 10 of whom developed left-sided IE without pacemaker involvement. Among the 63 patients who had prosthetic valves, 10 had 2 prostheses. Of the 73 prosthetic valves, 33 were mechanical and 40 were biological.

Table 2 summarizes the distribution of the different IE locations. A major echocardiographic criterion was present in 355 cases (91%), including 336 cases with vegetations (86%), 66 cases with an abscess (17%), and 15 cases of prosthesis dehiscence (24% of the patients with valvular prosthesis). Ninety percent of the patients underwent both transthoracic and transesophageal echocardiography. The remaining 10% had only transthoracic echocardiography.

Fever was nearly universal (355/387, 92%) (missing data for 3 cases) while septic shock was reported in 9% of patients (32/371, missing data for 19 cases). Severe congestive heart failure was quite common (34%), with only over one third requiring diuretic therapy (36%). Ten of 355 patients (3%) had a Glasgow Coma Scale score of 8 or lower during the course of the disease. Serum creatinine level was greater than 2.04 mg/dL (>180 µmol/L) at least once in 102 patients (27%).

At least 1 vascular phenomenon (systemic arterial embolism, 126; septic pulmonary infarct, 43; intracranial hemorrhage, 9; mycotic aneurysm, 12; or Janeway lesion, 9) occurred in 172 patients (44%). Distribution of systemic arterial emboli was as follows: central nervous system, 60 (48%); limbs, 38 (30%); liver, 4 (3%); spleen, 40 (32%); kidney, 8 (6%); coronary, 8 (6%); and others not specified, 16 (13%). Central nervous system emboli were recorded in 24 (21%) of mitral valve IE, 20 (15%) of aortic valve IE, and 10 (18%) of combined aortic and mitral IE cases. Immunologic manifestations were observed in 104 patients (27%) and included glomerulonephritis, 43 (41%); Osler nodes, 17 (16%); Roth spots, 5 (5%); and rheumatoid factor, 53 (51%). At least 1 extracardiac phenomenon was present in 78% of the right-sided IE cases (mainly pulmonary embolism, 68%) and in 52% of left-sided IE cases.

A total of 143 patients (37%) had experienced a medical or surgical procedure (n = 49) and/or a situation at risk of bacteremia (n = 107) within the month prior to hospitalization. Portals of entry were dental (poor dental condition or dental procedure, n = 34), cutaneous (n = 76), and miscellaneous other conditions (n = 46). Cutaneous portals of entry included traumatic or chronic wounds (n = 23), intravenous drug use (n = 22), and percutaneous iatrogenic procedure (n = 31). The latter were distributed as follows: chronic hemodialysis, 13; indwelling catheters, 11; pacemakers, 5; and percutaneous balloon valvuloplasty, 2.

The median (interquartile range) number of blood cultures drawn was 5 (3-7); positive blood cultures was 4 (3-6) and the mean proportion of positive blood cultures was 78%. Blood cultures remained negative in 9% of the cases. Overall, streptococci remained the most common causative agent of IE. The distribution of causative microorganisms is displayed in Table 3. Group D streptococci and Staphylococcus aureus accounted for a quarter of the cases each, whereas oral streptococci were responsible for less than 20% of the cases. Nosocomial and/or iatrogenic IE were more often due to staphylococci than were community-acquired IE (17/49 [35%] vs 73/341 [21%]; P = .04).

Table 4 shows a comparison of patient characteristics across the 3 main groups of microorganisms. Ages were significantly different within the 3 groups, group D streptococci IE patients being older and IE patients with S aureus being younger. Patients with group D streptococci and S aureus IE had less known underlying valve disease. Patients with S aureus IE had a higher mortality rate and a lower surgery rate.

The causative microorganisms were identified by blood culture in 356 patients (91%) and by other methods in 15 additional patients (4%) (6 cultures of valve or pacemaker leads, 8 serological tests, 1 16S ribosomal RNA gene amplification that led to the identification of Staphylococcus cohnii in a peripheral embolus). No microorganism could be identified in 19 patients (5%), 12 of whom received antibiotic treatment prior to blood samplings. Two or more microorganisms were identified concomitantly in 13 patients.

Mean (SD) duration of hospital stay was 47 (33) days (range, 0-268; median [interquartile range], 42 [28-59] days). During this period, 191 patients (49%) underwent valve surgery after a median (interquartile range) 21 (9-37) days after admission. The diagnosis of IE was confirmed by macroscopic, histological, or bacteriologic findings in 179 of the 191 operated patients (94%). In the remaining 12 patients, gross anatomic examination was not conclusive, histological examination was not performed in 11 cases, and bacteriologic examination was negative in 9. Sixteen patients (4%) underwent pacemaker lead removal (thoracotomy, 12; percutaneous removal, 4). Sixty-two patients died, which accounted for a 16% in-hospital mortality rate. The surgery and mortality rates according to various conditions are displayed in Table 5.

Compared characteristics of IE in the 2 study periods are displayed in Table 6.^15,16 The overall crude incidence of IE has not changed over time. However, the standardized incidence of IE has decreased in patients with previously known heart disease and in patients with prosthetic valves. The incidence of IE due to oral streptococci has decreased while it has remained stable for group D streptococci and S aureus. The rate of surgical treatment has significantly increased and the mortality rate tended to decrease.

The annual IE incidence rate of 31 cases per million found in this study is in the range of those from other recent studies (ie, comprised between 15 and 60 cases per million).^1-4 Only 1 recent study has shown a dramatically higher annual incidence rate of 116 per million,¹⁷ which was partially explained by a high proportion associated with intravenous drug use. By contrast, only 6% of patients in our population were intravenous drug users. The high incidence of IE observed in New Caledonia can be explained by the persistence of rheumatic fever and the low socioeconomic status.¹⁸

One major strength of our study is the low influence of referral bias, compared with most previous studies, which are mainly based on series collected in a single tertiary-care center. It is well known that the clinical features of IE substantially differ between patients from referral centers and those seen in primary-care centers.¹⁹ In the present study, the 390 cases were reported by 193 physicians from 91 unselected medical facilities.

Underreporting is a potential limitation that relies on physicians reporting and not on active search of cases by investigators. In our study, underreporting was minimized in that we relied on 3 separate sources of notification (physicians, echocardiographists, and microbiologists) that were relatively independent, though not completely, of one another. In addition, the proportion of case report forms that were not returned was as low as 1.3% (11 of 819). Finally, the proportion of patients living within 1 of the study regions and treated for IE in hospitals outside these regions was lower than 5% by analysis of the 1999 French national database file of diagnosis related groups at hospitalization discharge.²⁰

Some of our findings depict "classic" features of IE including predominance of men and prevalence in the elderly. In contrast, we recorded several new trends: (1) an increasing percentage of IE in patients with no previously known heart disease; (2) important changes in the distribution of causative microorganisms; (3) an increase of performance valve surgery during the initial hospitalization; and (4) a trend toward decreased in-hospital mortality.

The proportion of IE patients without previously known cardiac disease increased from 34% in 1991 to 47% in 1999. This resulted mainly from a significantly decreased incidence of IE in patients with previously known valve disease, especially those with prosthetic valves. This trend was also evidenced by Tornos et al.²¹ In their study of native valve IE in nonaddicts, the proportion of patients with no previously known heart disease significantly increased from 22% in the 1975-1983 period to 46% in the 1984-1992 period. These figures can be interpreted in several ways, including the dramatic decrease of postrheumatic fever valvular complications, the improved practice of antibiotic prophylaxis in patients with valve disease, and an underrecognition of degenerative valve lesions in elderly patients.²²

We observed major changes in the distribution of causative microorganisms. The proportion of group D streptococci markedly increased from 14% in 1991 to 25% in 1999. However, this almost 2-fold increase of group D streptococcal IE is explained less by an increased incidence of group D streptococcal IE than by a sharp decrease in the incidence of oral streptococcal IE (Table 6). Altogether, the decreased incidence of IE in patients with known valvular disease as well as the decreased incidence of IE due to oral streptococci may have resulted from improved dental hygiene and prophylaxis in at-risk patients, although our study was not equipped to determine the veracity of this.

Another trend observed in this study as well as in others^21,23 was the increased proportion of staphylococci, both S aureus and coagulase-negative staphylococci. This can neither be attributed to an overrepresentation of intravenous drug users nor to prosthetic valve endocarditis since only 13% of S aureus IE occurred in patients with prosthetic valves, whereas this percentage was 18% in IE caused by oral streptococci (Table 4). By contrast, 15% of all cases of staphylococcal IE had a nosocomial or iatrogenic origin. We also showed that nosocomial or iatrogenic IE were more often due to staphylococci than were community-acquired IE. In a recent study, hospital-acquired IE not related to cardiac surgery represented about 10% of a series of 248 cases of IE.²⁴ In 2 recent series of nosocomial IE, staphylococci were responsible for about 80% of the cases, with a S aureus coagulase-negative staphylococci ratio of about 3:1.^25,26

Compared with the 1991 survey, this study also showed important changes in terms of outcome and prognosis. Overall in-hospital mortality rate decreased from 21.6% to 16.6%, although the percentage of early surgical treatment increased from 31.2% to 49.7%. As expected, patients with aortic valve IE were operated on more often than patients with IE in other locations. By contrast, staphylococcal IE, which is usually regarded as more destructive and requiring more aggressive and earlier surgical treatment, was found to be operated on less frequently than enterococcal and streptococcal IE. Although not fully understood, this trend has also been found in other recent studies.^27,28

Improvement of outcome may result to a large extent from improved surgical management, as emphasized by Tornos et al²¹ who showed a decreased overall mortality rate from 19% in 1975-1983 to 12% in 1984-1992 concurrently with a significantly decreased surgical mortality rate from 43% to 18%. They attributed this reduction in surgical mortality rate to better timing of surgery in cases of IE complicated with congestive heart failure. We also showed a significantly lower mortality rate in operated patients compared with those treated only medically. Even if some of the most severely ill patients may have been denied surgery, these results do support early surgical treatment of IE, especially in patients who do not improve satisfactorily under medical treatment alone, as previously advocated.²⁹

Infective endocarditis has long been described as a disease with polymorphic manifestations.³⁰ This study contributes to reminding physicians that while its incidence has not decreased, its clinical and microbiologic profile is changing continuously and rapidly. This emphasizes the need for close epidemiological surveillance of this disease.