Introduction

Infective endocarditis (IE) is an infection of the cardiac endothelium that has low incidence in the pediatric population; however, it has an important morbidity and mortality (described mortality of 5–10% [1, 2]).

Congenital heart disease (CHD) is currently the main risk factor to develop IE in children (as much as 15–140 times higher compared to general population [3, 4]), due to the increase in survival in these patients that currently undergo surgical corrections that previously were not technically possible [5, 6] and the presence of prosthetic material after corrective surgery [7, 8]. However, with the increasing complexity in healthcare in pediatrics, a rise of IE in pediatric patients without CHD has been described [2], being other risk factors associated with pediatric IE prolonged indwelling central venous catheter (CVC) [9,10,11], immunocompromised patients (especially those receiving oncological treatment [12] or hematopoietic stem cell transplantation (HSCT) recipients [13]) or pediatric patients admitted to neonatal or pediatric intensive care units (NICU, PICU) that undergo invasive procedures [14, 15].

Staphylococcus aureus and viridans streptococci are still the leading causes of pediatric IE [2, 4, 5], nonetheless in recent years there has been an increase in pediatric IE caused by other less frequent microorganisms, such as coagulase-negative staphylococci (CoNS), Gram-negative bacteria (GNB), and fungi [11, 16]. Risk factors for CoNS IE are prosthetic material and prolonged indwelling CVC [5]. Neonatal period and prolonged indwelling CVC are risk factors for GNB IE [5]. NICU admission (especially in preterm infants), cancer patients, and prolonged indwelling CVC are well described risk factors for fungal IE [17, 18].

In pediatric IE, the complication with the highest morbidity is septic embolism, which can occur up to 40% of the cases (higher risk than adult population) [19]. The reported in-hospital mortality in pediatric series of IE ranges from 5 to 10%; with higher rates in patients with S. aureus IE, preterm infants (31%), patients with tetralogy of Fallot and pulmonary atresia (48%) and patients with prosthetic valve IE (8%) [1, 2, 10, 20].

It is important to have an updated knowledge of IE epidemiology, risk factors, clinical features, and complications in children in order to provide an adequate management. The aim of the present study is to describe these features of the IE episodes treated at a tertiary Spanish pediatric hospital in a 10-years’ period.

Materials and methods

Study design

An observational retrospective study was performed, describing the characteristics of pediatric patients (< 16 years old) with IE treated at a tertiary hospital in Madrid (Spain), national reference for complex CHD patients with advanced heart surgery, between January 2012 and December 2021.

Patients

Patients were eligible if they fulfilled all the following inclusion criteria: age < 16 years, treated at any unit at Hospital Universitario 12 de Octubre, with definite or possible IE (according to the ESC 2015 modified Duke criteria [21]).

Clinical data

Clinical data were obtained from the review of informatized clinical records and collected in a REDCap online database elaborated for this study. Patients’ date of birth, gender, date of diagnosis, fulfilled modified Duke criteria, and underlying conditions were noted. Also, previous surgical correction of CHD, presence of indwelling CVC, and multidrug-resistant bacteria (MDRB) colonization were recorded. Recorded data of the IE episodes included clinical presentation, echocardiographic findings, and microbiologic data. Also, medical, and surgical treatment, complications, length of hospitalization, and outcomes were noted.

Statistical analysis

Categorical variables were reported as frequencies and percentages, and continuous variables as median and interquartile range (IQR). To evaluate the possible association between microorganism and clinical presentation and complications, Fisher’s exact test and Wilcoxon-Mann-Whitney test were used. The significance level was α = 0.05. Statistical analysis was performed using IBM SPSS Statistics 25.

Results

Demographic features

During the study period, 32 episodes of IE were identified (demographic and clinical features shown in Table 1); considering the Duke modified criteria, 28 (87.5%) were definite IE and 4 (12.5%) possible IE.

Table 1 Demographic and clinical features of the 32 episodes of infective endocarditis (IE)
Full size table

The median age at diagnosis was 9.1 (4.2–129.8) months; 17 (53.1%) episodes were diagnosed in infants younger than 1 year of age, 3 (9.4%) in patients between 1–5 years, 4 (12.5%) in patients between 6–10 years and 8 (25.0%) in patients between 11–15 years.

Regarding underlying conditions, 28 (87.5%) patients had CHD (Table 2), 8 (25.0%) were preterm infants (median gestational age at birth: 27.5 weeks, IQR: 25.0–32.8 weeks), 1 (3.1%) was immunocompromised (metastatic medulloblastoma) and 6 (18.8%) had other chronic conditions (3 had a polymalformative syndrome, 1 short bowel syndrome, 1 esophageal atresia, and 1 Down syndrome); in 11 (34.4%) episodes, more than one underlying condition was associated (6 episodes: CHD plus preterm birth, 5 episodes: CHD plus other chronic conditions).

Table 2 Congenital heart disease (CHD) diagnosis
Full size table

Regarding the 28 episodes of IE diagnosed in patients with CHD, a corrective cardiac surgery or therapeutic cardiac catheterization was previously performed in 20 (71.4%). The median time of IE diagnosis after the cardiac procedure was 47.5 (18.3–348.0) days. At diagnosis, in 20 (62.5%) episodes the patients had an indwelling CVC, and the median time of IE diagnosis after CVC placement was 8.5 (4.0–14.3) days. Eight out of 8 (100%) preterm infants, 1 out of 1 (100%) immunocompromised patient and 5 out of 6 (83.3%) patients with other chronic conditions had an indwelling CVC at diagnosis (i.e., 14 out of 15, 93.3%), compared to 6 out of 17 (35.3%) patients with isolated CHD that had and indwelling CVC at diagnosis (p < 0.001).

Clinical manifestations

Presenting signs were fever in 27 (84.4%) episodes, shock in 16 (50.0%), heart failure in 1 (3.1%), vascular phenomena in 14 (43.8%) (11 (34.4%) septic embolisms, 2 (6.3%) intracranial hemorrhage, 1 (3.1%) Janeway’s lesions), immunological phenomena in 1 (3.1%) (glomerulonephritis and rheumatoid factor elevation in the same episode). The presenting signs in preterm infants compared to other patients were different: fever was only present in 4 out of 8 (50.0%) compared to 23 out of 24 (95.8%) of the non-preterm patients (p < 0.001) and shock was present in 7 out of 8 (87.5%) compared to 9 out of 24 (37.5%) of the non-preterm patients (p < 0.001).

Echocardiogram

An echocardiogram was performed in all episodes, transthoracic echocardiography was used in 29 (90.6%) episodes and both transthoracic and transesophageal echocardiography was used in 3 (9.4%). There were echocardiographic findings in 27 (84.4%), which are shown in Table 3.

Table 3 Echocardiographic findings
Full size table

Microbiology

Thirty-six microbiological isolates were obtained in the 32 episodes (Table 4).

Table 4 Isolated microorganisms of the 32 episodes of infective endocarditis
Full size table

Blood cultures were positive in 30 (93.8%) episodes; in 4 (12.5%) episodes, more than one microorganism was isolated (2 microorganisms isolated per episode). In those 4 episodes, the isolated microorganisms were: Staphylococcus aureus and Enterococcus faecalis, Enterococcus faecalis and Staphylococcus epidermidis, Enterococcus faecalis and extended spectrum beta-lactamase (ESBL) Escherichia coli, and Stenotrophomonas maltophilia and carbapenemase-producing Klebsiella pneumoniae. In those 2 episodes without positive blood cultures, Trichosporon inkin was isolated at surgical specimen in one and serology for Coxiella burnetii was positive (confirmed by PCR in surgical specimen and blood) in another case.

All S. aureus isolates were methicillin-sensitive and all E. faecalis isolates were ampicillin-sensitive. Among those 10 non-HACEK GNB, 5 (13.9%) were MDRB: 1 MDR Pseudomonas aeruginosa, 1 ESBL Escherichia coli, 1 carbapenemase-producing Klebsiella pneumoniae, and 2 AmpC beta-lactamase producing GNB (Enterobacter cloacae and Serratia marcescens).

In 10 (31.3%) episodes, patients were previously colonized by MDRB (clinical features of those patients and comparison with non-colonized patients in Table 5). Of those 10 patients, the colonizing bacteria were isolated in 3 (30.0%) of the patients’ blood cultures: 1 MDR Pseudomonas aeruginosa, 1 ESBL Escherichia coli, and 1 AmpC beta-lactamase producing Enterobacter cloacae. MDRB colonization was associated with MDRB IE (p = 0.007), with the presence of indwelling CVC (p = 0.030), and with longer time of PICU/NICU stay (p = 0.005).

Table 5 Clinical features of multidrug-resistant bacteria (MDRB) colonized patients and comparison with non-colonized patients
Full size table

Regarding the 4 episodes of fungal IE, one Candida albicans IE occurred in a premature newborn (25 weeks of gestational age at birth) with a patent ductus arteriosus and a 20-days’ CVC, the other in an infant with esophageal atresia and an 8-days’ CVC, and both Trichosporon inkin IE occurred in an adolescent with pulmonary atresia without CVC but with a prosthetic pulmonary conduit (relapse). All of them where right IE, located at native tricuspid valve, interventricular septum (right ventricle), and prosthetic pulmonary conduit respectively. Both patients with C. albicans IE where previously admitted to NICU/PICU. Both the patient with Trichosporon inkin IE and the infant with esophageal atresia and C. albicans IE had pulmonary embolisms and both required surgical treatment.

Treatment

The median of total antibiotic treatment duration was 47.0 (IQR 34.8–55.5) days, with a median duration of empirical antibiotic treatment of 4.5 (IQR 3.0–12.0) days. The median delay in initiating antibiotic treatment after the symptoms’ onset was 2.0 (IQR 0.0–5.0) days. The most used empirical antibiotic regimen was meropenem plus a glycopeptide (vancomycin or teicoplanin) with or without an aminoglycoside (gentamicin or amikacin) in 6 (18.8%) episodes, followed by a glycopeptide (vancomycin or teicoplanin) plus an aminoglycoside (gentamicin or amikacin) in 5 (15.6%).

Surgical management was required in 11 (34.4%) episodes, with a median time from diagnosis to surgical intervention of 5.0 (IQR 2.0–22.0) days. The indication of surgery was the difficult to eradicate isolated microorganism in 6 (18.8%) episodes (3 (9.4%) S. aureus, 2 (6.3%) P. aeruginosa, 1 (3.1%) C. albicans; with persistent bacteriemia in 5 (15.6%)), embolism prevention in 2 (6.3%), severe valve regurgitation in 2 (6.3%), and severe pulmonary conduit obstruction in 1 (3.1%).

Complications

PICU/NICU admission related to IE or its complications was required in 26 (81.3%) episodes, with a median stay of 26.0 (6.5–87.3) days. Complications secondary to IE were present in 25 (78.1%) episodes, 17 (53.1%) of the episodes with more than one complication. Registered complications were septic embolisms in 11 (34.4%) (8 (25.0%) pulmonary, 1 (3.1%) pulmonary and soft tissue, 1 (3.1%) cerebral, 1 (3.1%) cerebral and hepatic), shock in 8 (25.0%), heart failure in 7 (21.9%), valve regurgitation in 7 (21.9%), acute renal failure in 4 (12.5%), neurological complications in 3 (9.4%), local extension of infection in 3 (9.4%), and heart rhythm disturbances in 1 (3.1%). In-hospital mortality occurred in 2 (6.3%) episodes which was not attributed to IE or its complications, but to complications related to the patients’ underlying condition (pulmonary atresia and polymalformative syndrome (CHARGE syndrome) respectively).

Outcomes

There were 6 (18.8%) recurrences: 4 (12.5%) in-hospital relapses and 2 (6.3%) outpatient recurrences (1 relapse, 1 reinfection). At discharge, there were sequelae in 10 (31.3%) episodes, 3 (9.4%) of the episodes with more than one sequela: valvular regurgitation in 7 (21.9%) (2 (6.3%) of them associating heart failure), neurological sequelae in 3 (9.4%), heart rhythm disturbances in 1 (3.1%), valvular double lesion in 1 (3.1%), and chronic renal failure in 1 (3.1%). At follow-up (median time 3.4 years, IQR 0.7–7.2 years), 6 (18.8%) of those patients had persistent sequelae (1 (3.1%) with more than one sequela): valvular regurgitation in 4 (12.5%), heart failure in 2 (6.3%), neurological sequelae in 1 (3.1%), heart rhythm disturbances in 1 (3.1%), and chronic renal failure in 1 (3.1%).

Comparison of clinical features and complications depending on causative microorganism of IE

Clinical features and complications of IE episodes caused by non-HACEK GNB were compared to those episodes caused by GPB, and no statistically significant differences were found between them (Table 6).

Table 6 Comparison of clinical features and complications depending on causative microorganism of IE
Full size table

Although children with non-HACEK GNB IE exhibited a higher proportion of PICU admission (100% vs. 70.6%), shock (77.8% vs. 47.1%), heart failure (33.3% vs. 11.8%), neurological complications (22.2% vs. 5.9%), more than one underlying condition (55.6% vs. 29.4%), and presence of indwelling CVC (88.9% vs. 58.8%), these distinctions did not reach statistical significance. On the other hand, children with GPB IE showed a trend towards higher rate of residual valve regurgitation (29.4% vs 0.0%).

Discussion

Although IE is a rare disease in childhood, its incidence is increasing in the last years due to increase in health-care complexity [2]. For this same reason, epidemiology, risk factors and causative microorganisms are changing [11], so it is of paramount importance to have an updated knowledge of them in our setting to manage this disease adequately. We designed our study to describe the characteristics of the pediatric IE episodes treated in a cardiology reference center in a 10 years’ period.

As shown in previous reports [5], the main underlying condition was CHD (87.5%), suggesting that CHD is the main risk factor to develop IE in the pediatric population. The second risk factor for developing IE (25.0%), was being preterm (especially those extremely preterm, with a median of 27.5 weeks of gestational age at birth in our study); which has been also previously published as risk factor due to the prolonged indwelling CVC [9] and the fact that they undergo invasive procedures [14, 15]. In our study, only 3.1% episodes occurred in immunocompromised children, a population that may have been underrepresented due to the characteristics of our center, where HSCT (a well described risk factor for developing IE in children [13]) is not performed. Additionally, indwelling CVC was present in 93.3% of the recorded IE episodes in patients that were preterm, immunocompromised or had other chronic conditions compared to 35.3% episodes where CHD was the only underlying condition. This finding is in line with other studies where indwelling CVC is the main risk factor to develop an IE in these 3 groups of patients [9,10,11]. In our series, there were two incidence peaks regarding age at presentation: 53.1% presented in patients younger than 1 year and 25.0% in patients older than 10 years, similar to what is reported in previous literature where episodes peak in the first year of life (where surgical correction of the more severe CHD is performed) and adolescence [2].

Regarding clinical manifestations, we found that the preterm infant group had significantly different clinical presentation compared to the other patients: only 50.0% of preterm infants in our study had fever (vs. 95.8% in non-preterm children) and 87.5% presented with shock (vs. 37.5% in non-preterm children). These findings suggest that maybe other diagnostic criteria of IE should be applied in preterm patients, as previously has been proposed by other group [22].

Regarding microbiological isolates, the most frequent isolates were CoNS (19.4%) and S. aureus (16.7%) in contrast with what is reported in the literature [2, 4, 5, 20] and other recent series [23, 24] where the most frequently isolated microorganisms are S. aureus and viridans streptococci (which in our series were only isolated in 8.3% of the episodes). Remarkably, our study showed an important proportion of non-HACEK GNB (27.8%) which is significantly above what has been previously reported (0–9.1%) [23, 25, 26]. The proportion of fungal IE in our series (11.1%) was similar to other Spanish series [25, 27], being C. albicans the most frequent isolate [5].

Of note, we registered the MDRB colonization. Among 10 MDRB colonized children with IE, 3 (30.0%) had the same bacteria in blood cultures. This significantly high proportion of MDRB IE has not been previously reported in pediatric IE literature. Due to this finding, it might be advisable to consider empirical antibiotic coverage of these MDRB when suspecting IE in colonized patients. MDRB colonization was also associated with the presence of CVC and with longer NICU/PICU stay. Children with severe CHD spend long periods admitted to NICU/PICU and those with chronic conditions need a prolonged indwelling CVC, both situations with high risk for MDRB colonization and therefore possibility for MDRB IE.

Regarding IE complications, the main complication was septic embolism, present in 34.4% (n = 11) of the episodes (28.1% pulmonary embolisms and 6.3% cerebral embolisms). Proportion of embolism was similar to that reported in other pediatric study (40%) [18], although in our study the proportion of cerebral embolisms was lower (6.3% vs 20%) and that of pulmonary embolisms was higher (28.1% vs 8%), probably due to a lower proportion of left IE in our cohort (18.8% vs 55.9%). Previous studies showed that left sided IE have a higher risk for cerebral embolism compared to right IE (32% vs 2.8%) [28].

In-hospital mortality in our study was 6.3% (n = 2), like that reported in other studies [1, 2]. None of the deaths was due to IE or its complications. One of the patients had pulmonary atresia (a described risk factor for higher mortality -48%- [2]) and the other patient had a polymalformative syndrome. In our study, 18.8% (n = 6) of the patients had persistent sequelae at follow-up. Mortality and sequelae rates are consistent with previous literature [1, 8, 11, 23,24,25]. Despite IE is a rare disease, long-term sequelae in children with IE are still common, which may be relevant in populations with higher risk of IE, like patients with CHD or preterm infants.

We did not find statistically significant differences in clinical features or in complication rates between non-HACEK GNB and GPB IE, probably due to a limited sample size, but we could observe some interesting trends in this study (Table 6). There was a higher proportion of non-HACEK GNB IE in children with more than one underlying condition (55.6%) and children with indwelling CVC (88.9%), probably due to frequent contact with the health-care system. Non-HACEK GNB IE showed a higher proportion of PICU admission (100%), shock (77.8%), heart failure (33.3%) and neurologic complications (22.2%), which may be due to their higher virulence. On the other hand, GPB IE showed a higher proportion of residual valve regurgitation (29.4%), which may be caused by tendency of biofilm formation in GPB with higher potential for valve disruption.

Limitations of our study are those inherent to retrospective studies (missing data, prone to biases). Limited sample size is a major limitation of the study as IE is a rare condition in children, even in a reference center with advanced heart surgery like in this study.

In conclusion, risk factors for developing IE, the proportion of embolic complications, and mortality rate were consistent with previously published findings. Notably, the proportion of IE cases attributed to non-HACEK GNB was higher than previously reported, suggesting an evolving epidemiology of IE. It is noteworthy that one-third of children colonized with MDRB subsequently developed IE caused by the same MDRB strains; this emphasizes the potential importance of considering empiric antibiotic coverage against colonizing MDRB in those patients. While no statistically significant differences in clinical features and complications were observed when comparing IE episodes caused by non-HACEK GNB and those caused by GPB, larger cohort studies are needed for further investigation.