on 09-May-2022 (Mon)

We detected 1515 patients with E-BSI and 65 with enterococcal IE (4.29% of all episodes of E-BSI, 16.7% of patients with E-BSI who underwent transthoracic echocardiography, and 35.5% of all patients with E-BSI who un- derwent TEE)

The percentage of patients with E-BSI who have infec- tive endocarditis (IE) is estimated to be between 3% and 10% [1–3]

From 2003 to 2007, patients with E-BSI were managed by the attending physician who requested co nsulta- tion with the infectious diseases department or the laboratory of echocardiography [3] according to his/her own criteria (pe- riod A). From 2008 to 2012 (period B), a physician from the infectious diseases department visited the patients with E-BSI and promoted the systematic use of echocardiography. We rec- omme nded the systematic performance of TEE in most pa- tients, p rovided the patient consented to and the attending physician agreed with the indication

All patients fulfilling the m odified Duke criteria [12] for IE were considered cases, and patients with E-BSI and a TEE result that ruled out IE were con sidered controls. Control patie nts were randomly se- lected from among patients with E-BSI and a negative TEE re- sult and no criteria for IE according to the modi fied Duke criteria [12 ]

Underlying diseases were classified according to the McCabe and Jackson scale [14]. Comorbidities were assessed using the Charlson comorbidity score [15]

During the study period (2003–2012), we detected 1515 epi- sodes of E-B SI. The annual distribution is shown in Tabl e 1.

Enterococcal IE was detected in 65 patients, who accounted for 4.29% of all patients with E-BSI (3.76% in period A and 4.54% in period B)

IE was diagnosed in 16.7% of patients who underwent TTE and 35.5% of the patients who underwent TEE. Of all the episodes of enterococcal IE, only 18 cases (27.7%) were detected by TTE; the remaining 47 (72.3%) were demonstrated only after TEE. Sensitivity of TTE and TEE for the diagn osis of enterococcal IE was 32% vs 95% (P < .01).

The epide miological, microbiological, and clinical characteristics of both gro ups are shown in Table 2. No differences were de tected in age or sex, but pati ents with IE more frequently presented with a history o f stroke (27. 7% vs 13.8%, P = .05), immunosuppressive therapy (24.2% vs 10.8%, P = .03), previous heart valve disease (63.0% vs 29.2%, P < .01), and previous heart valve surgery (44.6% vs 2 4.6%, P = .03). Malignancy, however, was more frequent in controls (23% vs 41.5%, P = .02)

Episodes of IE were caused mainly by Enterococcus faecalis (86.2% vs 58.5%, P <01).Inaddition,theywereassociated with continuous bacteremia (93.8% vs 69.2%, P < .01), commu- nity acquisition (43.1% vs 20%, P < .01), and unknown source of infection (38.4% vs 10.7%, P < .01)

In the control group, how- ever, E-BSI was mainly nosocomial (69.2% vs 46.2%, P =.01) and had a gastrointestinal origin (48.4% vs 13.8%, P < .01).

The multivariate analysis showed that enterococcal IE is 9-fold more probable in patients with positive blood cultures in all of 3 blo od cultures or the majority of more than 3 blood cultures (OR, 9.9; 95% CI, 2.2–40.6).

Other factors independently asso- ciated with enterococcal IE were a history of heart valve disease (OR, 3.7; 95% CI, 1.6–8.7) and an unknown source of bactere- mia(OR,7.7;95%CI,2.5–23.8)

The score, which we called the NOVA score, was based on the following variable s: number of positive blood cultures (N) suggestive of continuous bacteremia (3/3 blood cultures or the majority if more than 3), 5 points; unknown origin of bactere- mia (O), 4 points; prior valve disease (V), 2 points; and auscul- tation of a heart murmur (A), 1 point (Table 4)

The area under the ROC curve for the NOVA score was 0.829 (95% CI, .758–.901)

The best binary cutoff value for ruling out IE without the need for TEE was established at a NOVA score < 4 points (Figure 1)

The probability of enterococcal IE with different scores is as follows: 5 points, 23.3%; 6 points, 45.5%; 7 points, 82.4%; 8 points, 66.7%; 9 points, 60.0%; 10 points, 100%; 11 points, 83.3%; 12 points, 80%

According to this model, the percentage of patients with E-BSI who would not re- quire echocardiography (score < 4 points) ranged from 14.6% in a setting with a prevalence of endocarditis of 50%, such as our case-control study, to 27.7% in a setting with a 5% prevalence of endocarditis.

Our results confirm that, as indicated in the American guidelines, TEE should be the test of choic e when the indication is to detect IE, especially if the pretest probability is high, such as in patients with staphylococcal bacteremia, fun- gemia, prosthetic heart valve, or intracardiac device [21].

Al- though systematic performance of TEE is not recommended in patients with enterococcal bacteremia in current guidelines, our relatively s mall percentage of patients wi th enterococcal bacteremia who underwent TEE (25.6%) reflects the real daily practice and, to date, is the only figure in the literature.

In many institutions, Enterococcus species is the third most common cause of BSI. The main origins are the gastrointestinal tract and catheter-related infections [22].

Our study suggests that most patients with E-BSI (those with a score ≥4 points) should undergo TEE (16.7% positive). We also show that TTE misses more than 70% of episodes of IE and that those patients should undergo TEE

Our study is subject to a series of limitations. First, as it was performed in a single center, and the sample siz e was not as large as it might have been

Second, since an infectious diseases specialist previously evaluated patients with bacteremia, selection bias should be taken into consideration. Interestingly, despite the intervention of the in- fectious diseases department, the rate of compliance with echo- cardiography recommendations remains low (34.4%), and a significant number of patients (1127) with enterococcal bacter- emia did not undergo TEE, thus limiting the ability of the study to estimate the real prevalence of IE.

Finally, ours is a case- control study, and the results should be validated in a second cohort and/or prospective study

Overall, the NOVA score is particularly useful for identifying a subgroup of patients with enterococcal bacteremia who may not need to undergo TEE (sensitivity 100%) because of an ex- tremely low risk of endocarditis

In conclusion, our study shows that the prevalence of entero- coccal IE depends on whether the sample comprised all cases among those with E-BSI (4.3%), only patients undergoing echo- cardiography (16.7%), or only patients undergoing TEE (35.5%). Use of TEE in all patients with E-BSI is difficult, costly, time consuming, and subject to complications. Depending on the local prevale nce of endocarditis, appli cation of the N OVA bedside prediction score could safely obviate echocardiography in 14%–27% of patients with enterococcal bacteremia

Enterococci are part of the normal human gut flora and can cause severe infections including urinary tract infections, gastrointestinal infections, catheter-related infections, and infective endocarditis (IE)

Enterococcal bacteremia is a relatively common condi- tion and has a mortality rate of around 20% [2, 3]

A sub- stantial proportion of patients with enterococcal bacteremia are diagnosed with IE [2, 4–6], and recent population-based studies suggest that up to 25% of patients with community- acquired Enterococcus faecalis bacteremia (EFsB) have this condition [5, 6]

Transesophageal echocardiography (TEE) is the method of choice to diagnose IE and identifies changes compatible with IE with a higher sensitivity than transthoracic echo- cardiography (TTE) [4]. However, TEE is a semi-invasive, resource-demanding procedure which is not readily available at all sites and it is not without discomfort for the patient.

The NOVA score was later validated in an adapted form and was found to have a high sensitivity (97%) but limited specificity (23%) [5].

All consecutive blood cultures positive for E. faecalis from January 2012 through December 2016 were obtained from the database of the Laboratory for Clinical Microbiology in Skåne County, in southern Sweden. This is the only labo- ratory in the region and has a catchment area of 1.3 mil- lion inhabitants and nine hospitals.

The 362 included patients had 397 epi- sodes of MEFsB, of which 44 episodes in 44 patients (11%) fulfilled the Duke criteria for IE (Supplementary material 2). Four episodes, in which the Dukes criteria were not fulfilled, were treated as possible IE by the physician

In the final multivariate model, symptom duration, embo- lization, number of positive blood cultures, unknown ori- gin of infection, and heart murmur were associated with IE (Table 2).

Further- more, we removed age from the model since this variable resulted in very large standard errors of other estimates indi- cating a numerical problem

These variables were included in a novel score, and despite not being significantly correlated with IE in mul- tivariate analysis, we chose to include also the valve disease criterion of NOVA in our new score to conform with earlier studies [4, 5]. The novel score was termed DENOVA and each variable was given one point to make it easy to use (Table 2)

The DENOVA score was tested for its ability to separate episodes with IE from episodes without IE and the resulting receiver operator characteristics (ROC) curve had an area under the curve (AUC) of 0.95 (95% CI, 0.94–0.97) com- pared to the adapted NOVA score which had a significantly lower AUC of 0.91 (95% CI, 0.89–0.95) (p = 0.001 for dif- ference) (Fig. 1)

With a cutoff at ≥ 3 points DENOVA had a sensitivity of 100% and a specificity of 83% to identify episodes of IE in MEFsB, while at the suggested cutoff of 4 points [4], the NOVA score had a sensitivity of 100% and a specificity of 29%

When the DENOVA score was applied to the valida- tion cohort, the resulting ROC curve had an AUC of 0.95 (95% CI, 0.93–0.98) which was significantly higher than that of the adapted NOVA (AUC 0.92 (95% CI 0.88–0.95), p = 0.007). At the predefined cutoff of three points, DEN- OVA had a sensitivity of 100% and a specificity of 85% compared to the NOVA score (with cutoff 4 [4]) with a sen- sitivity of 100% and a specificity of 35%

Combining the calibration and validation cohorts and analyzing the subgroup of patients where echocardiography had been performed, DENOVA had a sensitivity of 100% and a specificity of 62%.

The numbers needed to screen to find one case of IE was 2.4 for DENOVA (≥ 3) and 7 for NOVA (> 4)

Finally, our design avoids the bias introduced when patients subjected to TEE with a finding of IE are compared to patients subjected to TEE without a finding of IE (a type of over-matching) such as in the study behind the NOVA score [4].

When we applied the DENOVA score only to patients that had undergone echocardiography, the sensitivity was still 100%, whereas the specificity dropped to 62%. This was expected since the decision by the treating physician to perform TEE was likely based on a perceived increased risk for IE

However, there are several limitations to this study. First, the DENOVA score cannot be applied to E. faecium bac- teremia and not to polymicrobial EFsB. In both these con- ditions, IE is uncommon [2, 4, 6] and separate very large studies would be needed to adapt the score to these condi- tions.

Secondly, the retrospective design of our study makes it sensitive to misclassification, potentially non-differential. For example, a clinician who suspects IE might be more prone to take additional blood cultures, make a more thor- ough heart auscultation, or look for emboli, which will potentially increase the DENOVA score.

Thirdly, a limited number of patients underwent TEE in our study. This is a common problem in this type of studies [4, 5, 14, 15] and a consequence of the retrospective design. Even in prospective studies, however, a high frequency of TEE in bacteremia is not always obtained [16] and, impor- tantly, a negative TEE does not rule out the presence of IE or that a visible vegetation can develop after TEE

To assess the risk of misclassification, we followed the patients for 360 days. Patients with a non-IE MEFsB were rarely treated more than 14 days and it is thus unlikely that a misclassi- fied IE would have been without a relapse.

However, a large number of patients died within 360 days after a non-IE epi- sode (142 and 62 in the calibration and validation cohorts, respectively). We cannot rule out that some of these patients died from a missed IE, yet the analysis of patient records did not indicate that this was the cause of death.

Non- β-hemolytic streptococci (NBHS) have been the dominant cause of IE historically, and are still responsible for a large pro- portion (13%–44%) of cases [3–8]

In addition to IE, NBHS are known to cause other types of invasive infections such as abscesses [13, 16, 17], neutropenic fever [18, 19], and bacteremia in neonates [20]

The risk fac- tors for IE in patients with NBHS bacteremia have not been studied systematically, although prior dental surgery has been associated with a higher risk of IE and neutropenia with a lower likelihood of IE [21].

Two cohorts of patients with NBHS bacteremia were stud- ied retrospectively. A list of blood cultures positive for NBHS from 977 individual, consecutive patients was received from the Department of Clinical Microbiology in Lund, Sweden

The inclusion was made into 1 of 2 cohorts, the first with patients cul- tured between 1 January 2012 and 30 June 2013, the second with patients cultured between 1 July 2013 and 31 December 2014. The first group was used to assess general patient characteristics and outcomes, and to generate the scoring system. The second group of patients was used to validate the scoring system.

Patients were considered to have IE if they fulfilled the modified Duke criteria [30] or were diagnosed with IE at autopsy. Patients were placed in the negative group if (1) TEE had been per- formed without signs of IE; (2) if they received <sy

Between 1 January 2012 and 30 June 2013, blood cultures from 446 patients with growth of NBHS were recorded. After exclud- ing persons <d.

When analyzing the remaining patients, 26 cases of IE and 197 cases of non-IE were identified; the remain- der were unknown (Figure 1)

Patients with IE had experienced symptoms for a significantly longer period at the time when the blood culture was taken (P < .0001)

Some factors were significantly more common in the group with IE including community-acquired infection (P = .02), preexisting heart valve disease (P < .001), and heart murmur upon auscul- tation (P < .001). Embolic events were more common in the IE group, but this difference was not significant (P = .2). The pres- ence of fever was similar in those with confirmed or excluded IE

Streptococci of the S. sanguinis group were the most common cause of IE (11 of the 26 confirmed cases), followed by S. bovis group (5 cases), S. mutans group (4 cases), S. mitis group (4 cases), and S. sali- varius group (2 cases).

No IE cases in this cohort were caused by S. anginosus group isolates. Compared to the non-IE group, S. sanguinis (P = .001), S. bovis (P = .03), and S. mutans (P = .007) group streptococci were overrepresented in the IE group, and S. anginosus group streptococci were underrepresented (P < .001)

The number of positive blood cultures was higher in the IE group (P < .001), with a median of 2 positive compared with 1 in the non-IE group. The presence of continuous bacteremia was also significantly higher (P = .002) in the group with confirmed IE

Neither the 30-day all-cause mortality nor the 6-month all- cause mortality differed significantly between cases with con- firmed or excluded IE. A higher proportion of patients in the IE group had undergone (TEE) or trans-thoracic echocardiog- raphy (TTE)

Using such variables, the HANDOC risk score (HANDOC = Heart murmur or valve disease; Aetiology with the groups of Streptococcus mutans, Streptococcus bovis, Streptococcus sanguinis, or Streptococcus anginosus; Number of positive blood cultures ≥2; Duration of symptoms of ≥7 days; Only 1 species growing in blood cultures; and Community- acquired infection) was chosen, with parameters that were common and differed significantly between patients with and without IE. The score is presented in Table 4.

With a sensitivity of 100% (95% CI, 88%–100%) and specificity of 73% (95% CI, 67%–80%), the cutoff was set between 2 and 3 points.

When the HANDOC score was tested against the whole cohort (including also the unknown category), the performance of the score was similar (75% specificity; area under the ROC curve, 0.96) to that when applied only to IE and non-IE cases. The resulting negative predictive value was 100% and the positive predictive value was 23% with the prevalence of 7.6% as in the main cohort.

Between 1 July 2013 and 31 December 2014, blood cultures with NBHS from 522 patients were received. The inclusion and exclusion of patients is presented in Supplementary Appendix 3. HANDOC was applied to the patients with (n = 37) and with- out (n = 264) IE; using a cutoff score of ≥3, the resulting sen- sitivity was 100% (95% CI, 91%–100%) and the specificity was 76% (95% CI, 71%–81%)

When HANDOC was applied to the entire validation cohort, including also the unknown group, 77% of the cases without confirmed IE had a score of ≤2 points

As a subpopulation analysis, we applied the HANDOC score in patients where echocardiography of any kind was performed; the resulting sensitivity was 100% and the specificity 62%. Including only the cases where TEE was performed, the sensi- tivity was 100% and the specificity was 47%

With the established cutoff of 3 points, HANDOC had excellent sensitivity (100%) and good specificity (74%) in the first cohort. Importantly, the specificity was similar for men and women and was also unaffected by inclusion of the group of patients where IE could formally not be ruled out.

The fact that both the score creation and the score vali- dation cohorts consisted of patients from the same geographical area and from the same hospitals is a limitation of the study and makes it difficult to draw definite conclusions about the suitabil- ity of the score in other settings

The “heart murmur or valvular disease” and “aetiology” criteria are similar to those included in the Duke criteria, and the “number of cultures” criterion of HANDOC is included in the Duke criteria. However, “duration of symptoms,” “only 1 species,” and “community acquired” are parameters not part of the Duke criteria.

In the univariable analysis, both the presence of heart mur- mur and underlying heart disease were associated with IE, but we chose to combine these variables as they are strongly interconnected mechanistically and were significantly corre- lated (P < .0001 using 2-tailed Pearson’s test).

A long duration of symptoms is a textbook description of NBHS IE and was found to be highly indicative of IE in our investigation and should clearly be included in a scoring system.

The number of positive cultures and number of cultures taken correlated significantly in our study (P < .0001 with 2-tailed Pearson correlation). We therefore compared the number of positive cultures in a subgroup where 2 cultures had been taken (n = 180). The number of positive blood cultures in this subgroup was significantly higher (P < .0001, Fisher exact test) in the group with IE than in the group where IE had been excluded.

We did not find cases of IE with growth of NBHS in only 1 flask but the HANDOC score was ≥3 in only 18 patients with growth in one bottle only, the majority of which had long dur- ation of symptoms, heart murmur on auscultation, or preex- isting heart valve disease. In our experience, IE occurs also in patients with a single positive flask and such a finding should not preclude the patient from echocardiography guided by a risk stratification using HANDOC

A limitation of this study is that the group of patients with IE was too small to allow multivariable analysis. Thus, it may well be possible that the variables associated with IE in our anal- yses are not truly directly linked to the outcome

The microbiological variables “aetiology,” “number of cultures,” and “only 1 species” may typ- ically not all be independently associated with IE, but as they are easily accessible and work well in the model, we find it rea- sonable to include them anyway.

If echocardiography had only been performed in the cases with a HANDOC score of ≥3, the total number of investigations would have been decreased from 307 to 225. In our study, the number needed to screen to find 1 case of IE was 3.6.

Among the 2008 patients enrolled, 221 (11.0%) had definite IE of whom 39 (17.6%) un- derwent valve surgery, 25% of them within 6 days of SAB diagnos is.

Ten predictors indepen- dently associat ed with IE were used to build up the prediction score: intracardiac device or previous IE, n ative valve dis ease, intravenous drug use, community or no n-nosocomial- acquisition, cerebral or extracerebral emboli, vertebral osteomyelitis, severe sepsis, meningi- tis, C-reactive protein above 190 mg/L, and H48-persistent bacteremia.

Patients with a score 2(nZ 792, 39.4%) were at low IE-risk (1.1%; negative predictive value: 98.8% (95% CI, 98.4 e99.4)) compared to those 3 who were at higher risk (17.4%)

Staphylococcus (S ) aureus is among the most frequent causes of both healthcare-associated and community- acquired bloodstream infections worldwide, with incidence rates of between 20 and 50 cases/100,000 population per year in industrialized countries. 1,2

One of the most severe complications of Staphylococcus aureus bloodstream infec- tion (SAB) is infective endocarditis (IE), reported to occur in 5e17% of cases.

Echocardiography plays a key role in IE diagnosis. 4 Rec- ommendations on the systematic use of echocardiography in SAB patients are not consistent in the literature. 4e

in a recent pooled analysis of SAB pro- spective studies, echocardiography was performed in only 56% of patients despite being strongly recommended to in- vestigators. 11

VIRSTA is an observational prospective cohort study previ- ously described 15 conducted between April 2009 and January 2012 which included all consecutive adult patients having at least one blood culture positive for S. aureus in 8 tertiary-care university hospitals in France. Patients with catheter colonization without SAB, defined as positive blood cultures only through vascular access device spec- imen and those referred to the hospitals for the manage- ment of IE were excluded

Persistent bacteremia was defined as positive blood cultures more than 48 h after the first posi- tive blood culture result (Fig. 1)

The primary endpoint was the diagnosis of definite IE according to modified Duke classification 17 within 12 weeks established by a local adjudication committee made up of cardiologists, infectious diseases specialists and bacteriologists.

First, a descriptive analysis of patients was performed. Potential IE predictors were then selected based on the literature, 6,13,14,18e21 categorized as 1/patient background characteristics, 2/initial SAB presentation characteristics, and 3/early extracardiac events (Table 1)

Only SAB charac- teristics and extracardiac events present and/or occurring during the first 48 h following the T0 blood sample collec- tion were considered, as well as the result of the T48 h blood sample collection

To assess the robustness of the model, three sensitivity analyses were performed. The first one was performed in the subpopulation of patients, who underwent echocardiography, or in whom echocardiography was not performed but for whom the addition of an echocardiographic major criterion would not have upgraded the modified Duke classification to a definite IE case (hereafter referred to as “echo sensitivity analysis”).

n the second one, we excluded all definite IE in which definite classification was based on the presence of modified Duke Criteria which were included in the statistical model as potential pre- dictors of IE to avoid the resulting tautology (hereafter referred to as “modified Duke criteria sensitivity analysis”

A third sensitivity analysis was performed to evaluate the performance of the score among patients with highest diagnostic uncertainty. We excluded from our population patients with definite IE, based on duke criteria determined within the first 48 h (patients with the microbiological major criteria AND 3 minor criteria)

Echocardiography was performed in 1348 patients (67.1%), and 605 patients (30.1%) underwent TEE

The adjudication committee categorized 221 (11.0%) (95% CI 9.6%e12.4%) patients as definite IE cases.

The rate of definite IE was 15.6% (n Z 211) in the 1348 subjects in whom echocardiography was per- formed.

The rate of definite IE was 15.6% (n Z 211) in the 1348 subjects in whom echocardiography was per- formed. Among them, echocardiography revealed one ma- jor criterion of IE in 80.6% (n Z 170), including vegetation in 139 patients (65.9%), abscess in 32 patients (15.1%), and new dehiscence of a prosthetic valve in 13 pa- tients (6.2%).

Ten predictive factors were independently associated with definite IE (Table 3).

The rate of endocarditis increased significantly from 1.1% (9/792) when the score was 2e17.4% (212/1216) when the score was 3 and up to 70.8% (63/89) when the score was 10

Score performance according to different cutoffs is presented in Table 4. For a score 2, the negative predictive value was 98.8% (95% CI 98.4; 99.4) and the sensitivity was 95.8% (95% CI 94.3; 97.8). After exclusion of (n Z 28) patients with definite IE as established within the first 48 h (third sensitivity analysis), the performance of the VIRSTA score was similar (95.3% for sensitivity and 98.9% for negative predictive value).”

Although echocardiography use differs between studies, impacting IE diagnosis rate, the 15.6% or 11% endocarditis rates (according to restriction of analysis to patients with echocardiography), are also similar to those reported in the literature. 11

Our rate of 67.1% echocardiography (1348 of our 2008 patients underwent echocardiography), is among the highest reported in the literature (43e79%). 6,10,12,23

Our study shows several readily available predictors independently associated with IE such as patient back- ground characteristics, including predisposing cardiac con- ditions. These findings agree with previous reports who proposed criteria to assess predictors of IE among SAB patients.

Interestingly, CRP, another predictor of IE in our popula- tion, has been proposed to be an additional minor criteria of Duke modified classification by Lamas and colleagues 27 because it improved Duke classification sensitivity in pa- tients with pathologically proven IE. To our knowledge, our VIRSTA study is the first who reported an independent association between CRP level and IE in patients with sus- pected IE in multivariable analysis. To our knowledge, this study is the first to report the CRP level as an independent predictor of IE.

Some early extracardiac events such as embolic events, vertebral osteomyelitis, meningitis and se- vere sepsis or shock were associated with a higher fre- quency of IE.

Persistent bacteremia was strongly associated with definite IE, and has been shown to be a pre- dictor of complicated SAB.

The proportion of IE increases concomitantly with the score, from 1% for a score 2 to more than 70% for a score 10. For a score 2, the negative predictive value was excellent (98.8%).

The threshold of 2 is somewhat arbitrary but combines the assets of a high negative predictive value (>95%) and a small number of undiagnosed IE cases (1.1%)

The score can be applied as soon as SAB is diagnosed and echocardiography indicated if the score is above 2. Other- wise, occurrence of complications and/or of a positive T48 h blood culture must lead to recalculation of the score and to indication for echocardiography in those with a score which has risen to above 2

In patients with a score 3, echocardiography should be performed urgently using the most sensitive method, i.e. TEE. According to the positive predictive value, more than one in 5 patients in this group would be classified as definite IE.

The use of this score in our cohort would have permitted the avoidance of urgent TEE in 792 patients (39.4%). Of note, among the 9 patients with definite IE and a score 2 (false-negative), none would have had an increased IE prediction score 3 if SAB complications that occurred later than 48 h had been considered (data not shown)

We acknowledge several limitations to our study. First, as with other large observational prospective studies and despite recommendations to the contrary, a substantial proportion of our patients’ cohort did not un- dergo echocardiography in our cohort.

Second, patients were enrolled only in tertiary care centers. This probably led to the recruitment of more severe patients with a higher prevalence of comorbidities. However, we minimized the referral bias by excluding pa- tients referred from other hospitals for the management of IE.

Early TEE should be performed ur- gently in patients with a score 3, and repeated if initially negative, while in the large subgroup of patients with a score 2, urgent TEE is not needed at an early stage, although the indications depend on the clinical context