Transthoracic echocardiography in children with Staphylococcus aureus bacteraemia: a multi-centre retrospective analysis Authors: Kirsty Gray, Farhana Ahad, Aubrey Cunnington Published: 2026 ABSTRACT AbstractBackgroundStaphylococcus aureus bacteraemia (SAB) is associated with risk of infective endocarditis (IE). IE is rare in children in the absence of known risk factors such as congenital heart disease. Routine transthoracic echocardiography (TTE) is often performed in children with SAB despite limited age-specific evidence.MethodsThis multi-centre retrospective cohort study included 101 children (aged ≤18 years) with SAB across six North West London hospitals (2018–2023). Data were collected from electronic patient records.Results71 children (70%) underwent TTE and 30 (30%) did not. IE was confirmed in four (5.6%) of those undergoing TTE. All four IE cases were right-sided and occurred in children with established risk factors: congenital heart disease (2/4), intravenous drug use (1/4) or central venous catheters (2/4). The TTE group had longer hospital stays than the no-TTE group (median 17 vs 4 days, p=0.0020). Acute in-hospital mortality occurred in 1/71 of the TTE group vs 0/30 of the no-TTE group (p=1.0). Among children undergoing TTE, features associated with IE included embolic phenomena (75% vs 1.5%, p<0.001), higher CRP (median 237 vs 124 mg/L, p=0.083), persistent bacteraemia at >72 hours (67% vs 4.9%, p=0.018) and polymicrobial growth on blood culture (50% vs 6%, p=0.033).ConclusionOur findings support a risk-stratified rather than universal approach to TTE in paediatric SAB. TTE could reasonably be omitted in children with a structurally normal heart, no established IE risk factor and no clinical feature suggestive of IE, but should be performed when embolic phenomena, persistent bacteraemia, polymicrobial growth, a pre-existing murmur or an established IE risk factor is present. Prospective multi-centre validation is needed before such an approach can be adopted in routine practice.Keywords: Health services research, Microbiology WHAT IS ALREADY KNOWN ON THIS TOPIC WHAT THIS STUDY ADDS In this multi-centre cohort of 101 children with SAB, infective endocarditis was diagnosed in 5.6% of children undergoing TTE, exclusively in those with established risk factors (congenital heart disease, intravenous drug use or central venous catheter); clinicians already appear to informally risk-stratify when deciding which children to image and which to investigate with repeat blood cultures, though this stratification is not applied consistently. In this multi-centre cohort of 101 children with SAB, infective endocarditis was diagnosed in 5.6% of children undergoing TTE, exclusively in those with established risk factors (congenital heart disease, intravenous drug use or central venous catheter); clinicians already appear to informally risk-stratify when deciding which children to image and which to investigate with repeat blood cultures, though this stratification is not applied consistently. HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY Our findings support the development of a structured risk-stratification approach to TTE in paediatric SAB, with prospective validation needed before routine omission of TTE in low-risk children can be recommended. Our findings support the development of a structured risk-stratification approach to TTE in paediatric SAB, with prospective validation needed before routine omission of TTE in low-risk children can be recommended. Introduction Staphylococcus aureus bacteraemia (SAB) is associated with increased risk of developing infective endocarditis (IE).1 In adults with SAB, transthoracic echocardiography (TTE) is recommended to identify or exclude IE, and this practice is frequently extrapolated to children.2 However, SAB-associated IE is less common in children and is particularly rare in the absence of known risk factors,3 meaning that TTE may be of limited value in this group. There is a significant variation in TTE practice in paediatric SAB across centres.2 4 Established risk factors for the development of IE in children with SAB include congenital heart disease,2 3 the presence of prosthetic material or cardiac devices,3 previous IE,2 acquired valvular disease,3 presence of a central venous catheter (CVC) and intravenous drug use (in adolescents).5 6 Clinical features associated with IE include heart murmur,5 persistent bacteraemia,1 2 prolonged fever,3 embolic phenomena7 8 and absence of an alternative focus of infection.2 Despite these known associations, there remains significant variation in investigation practices, particularly regarding the use of TTE in children without identified risk factors. The identification and characterisation of a low-risk group of children who do not require TTE could have significant clinical and resource implications. This is particularly relevant in settings where TTE is not readily available. A risk-stratified approach in these settings could prevent unnecessary investigation, potentially reducing length of hospital stay and associated costs without compromising patient safety. We investigated the use and findings of TTE in a contemporary cohort of children with SAB across six hospitals in North West London (NWL), aiming to identify factors that might help to guide more targeted use of TTE in this group. Methods Study design and data source Infants and children (≤18 years old) diagnosed with SAB between 1 April 2018 and 31 March 2023 were identified from a centralised microbiology database at NWL Pathology.9 This included children treated at six hospitals providing secondary and tertiary level neonatal and paediatric care across NWL. Subjects were excluded if the index positive blood culture was collected after or within 24 hours of death. The study was approved by Imperial College Healthcare NHS Trust and London North West University Healthcare NHS Trust as a service improvement project and therefore did not require ethical approval. Data collection Data were collected retrospectively from electronic patient records, including age; sex; date of admission; date of discharge; whether bacteraemia was community-onset or hospital-onset; risk factors for cardiac complications of SAB; clinical signs; clinical management; laboratory results and complications of SAB. Classification of infection onset followed the Centres for Disease Control and Prevention (CDC) criteria for healthcare-associated infections.10 Bacteraemia was categorised as hospital-onset if the positive blood culture was collected on or after day 3 of hospitalisation (day of admission being day 1). Infections occurring on hospital day 1 or 2 were classified as community-onset. Features of SAB infection Data were collected on the initial presentation and course of SAB including persistence of fever (≥38 °C, measured as total days of fever from the time of initial positive blood culture), suspected primary focus of infection, peak C-reactive protein (CRP) during admission, and initial blood culture results, including species of organisms grown. Clinical management data included presence and drainage of focal collections, timing of drainage (measured in hours from the time of the first positive blood culture) and duration of antibiotic therapy (measured in days from the date of the first positive blood culture). Central venous catheter (CVC) management was documented, including whether the CVC was present at the time of positive blood culture and whether it was subsequently removed during the treatment course. Time to CVC removal was measured in hours from the time of the first positive blood culture. S. aureus can be a blood culture contaminant, but it is often difficult to exclude that SAB is clinically significant, and therefore recommended to treat all cases of SAB.2 We therefore included all patients with a positive blood culture for S. aureus. Suspected contaminants were identified based on documentation in the medical notes, typically involving a single positive blood culture bottle, rapid clinical improvement without targeted antimicrobials, absence of clinical signs of infection and a suitable alternative explanation for the patient’s presentation. Complications of SAB were documented including: requirement for intensive care, clinical suspicion of meningitis (and subsequent CSF results), presence of multifocal infection and diagnosis of vertebral osteomyelitis. Acute in-hospital mortality during the index SAB admission was recorded. Features associated with increased risk of IE Data were collected on established risk factors for IE including underlying cardiac conditions (congenital heart disease, prosthetic valves/material, intracardiac devices, previous infective endocarditis, hypertrophic cardiomyopathy and acquired valvular disease), central venous catheters and history of intravenous drug use. Any history of pre-existing heart murmurs (diagnosed prior to admission for investigation of SAB) was documented. Clinical features at presentation associated with IE were recorded, including presence of new onset heart murmurs, immunological phenomena (such as Osler’s nodes and Roth’s spots), suspected and proven embolic phenomena (such as cerebral emboli, arterial emboli, septic emboli, conjunctival haemorrhage or painless skin lesions) and persistence of bacteraemia on repeat blood cultures (<48 hours, between 48 and 72 hours and >72 hours after initial culture). Where performed, TTE results were collected and categorised according to the Modified Duke Criteria.11 Statistical analysis All analyses were conducted in R (version 4.4.2). Due to the low incidence of infective endocarditis (n=4) and low prevalence of several key variables, we carried out univariable analyses using the elrm package for exact logistic regression.12 Firth’s penalised likelihood regression (logistf package) was used for variables with small cell counts (congenital heart disease and repeat blood culture performance).13 The outcome variable for the univariable logistic regression analyses was whether transthoracic echocardiography was performed. We refrained from calculating ORs when comparing children diagnosed with IE on TTE to children with a negative TTE, as these would likely be misleading due to the small group size. Key variables (such as total days of fever and antibiotic duration) were not normally distributed. Non-parametric tests were therefore used for group comparisons. For continuous variables (age, CRP etc.), the Wilcoxon rank-sum test (Mann-Whitney) was used to compare between groups. Fisher’s exact test was used for categorical variables. Continuous variables are presented as median (range), and categorical variables as numbers (percentages) with 95% CIs calculated using Wilson’s method where appropriate. Univariable ORs are reported with 95% CIs for variables with p≤0.1. This threshold was chosen to identify potentially important risk factors given the small numbers of children per group. P-values are presented to two significant figures, with values below 0.001 reported as p<0.001. For analyses involving subgroups, denominators represent only the relevant eligible population. For example, when considering percentages of children in table 1 who underwent CVC removal, the denominator (N) represents only the subset of children with a CVC-associated focus of infection, while the numerator (n) represents those who had their CVC removed during the course of treatment. Similarly, analyses of persistent bacteraemia at specific time points include only those children who had repeat blood cultures performed within the specified timeframe. Table 1Characteristics and outcomes of children with SAB, grouped by TTE statusCharacteristicsAll children (n=101)TTE not performed (n=30)TTE performed (n=71)p- valueUnivariable ORAge at admission - median years (range)4.3 (0.0–17.6)8.2 (0.0–17.2)4.1 (0.0–17.6)0.35Sex - n (%)(95% CI)1.0Male60 (59.4) (49.7 to 68.5)18 (60.0) (42.3 to 75.4)42 (59.2) (47.5 to 69.8)Female41 (40.6) (31.5 to 50.3)12 (40.0) (24.6 to 57.7)29 (40.8) (30.2 to 52.5)Neonate (age≤28 days) - n (%)(95% CI)10 (9.9) (5.5 to 17.3)3 (10.0) (3.5 to 25.6)7 (9.9) (4.9 to 19.0)1.0Congenital heart disease - n (%)(95% CI)17 (16.8) (10.8 to 25.3)2 (6.7) (1.8 to 21.3)15 (21.1) (13.2 to 32.0) 0.088 3.71 (0.78 to 35.71)IV drug use - n (%)(95% CI)1 (1.0) (0.2 to 5.4)0 (0.0) (0.0 to 11.4)1 (1.4) (0.2 to 7.6)1.0CVC present - n (%)(95% CI)25 (24.8) (17.4 to 34.0)6 (20.0) (9.5 to 37.3)19 (26.8) (17.9 to 38.1)0.62New-onset heart murmur - n (%)(95% CI)7 (6.9) (3.4 to 13.6)0 (0.0) (0.0 to 11.4)7 (9.9) (4.9 to 19.0) 0.10 NAPre-existing heart murmur - n (%)(95% CI)7 (6.9) (3.4 to 13.6)0 (0.0) (0.0 to 11.4)7 (9.9) (4.9 to 19.0) 0.10 NAEmbolic phenomena - n (%)(95% CI)6 (5.9) (2.8 to 12.4)2 (6.7) (1.8 to 21.3)4 (5.6) (2.2 to 13.6)1.0Community-acquired - n (%)(95% CI)77 (76.2) (67.1 to 83.5)26 (86.7) (70.3 to 94.7)51 (71.8) (60.5 to 81.0)0.13Fever duration - median days (range)1.0 (0.0–14.0)0.0 (0.0–14.0)1.0 (0.0–14.0) 0.087 NAPeak CRP mg/L - median (range), n=90120 (0.3–403)44 (0.3–365), n=25125 (0.8–403), n=65 0.013 NARequirement for intensive care - n (%)(95% CI)19 (18.8) (12.4 to 27.5)3 (10.0) (3.5 to 25.6)16 (22.5) (14.4 to 33.5)0.17Vertebral osteomyelitis - n (%)(95% CI)3 (3.0) (1.0 to 8.4)2 (6.7) (1.8 to 21.3)1 (1.4) (0.2 to 7.6)0.21Meningitis suspected - n (%)(95% CI)9 (8.9) (4.8 to 16.1)1 (3.3) (0.6 to 16.7)8 (11.3) (5.8 to 20.7)0.27Multifocal infection - n (%)(95% CI)9 (8.9) (4.8 to 16.1)1 (3.3) (0.6 to 16.7)8 (11.3) (5.8 to 20.7)0.27CVC removal - n/N (%)(95% CI)14/25 (56.0) (37.1 to 73.3)3/6 (50.0) (18.8 to 81.2)11/19 (57.9) (36.3 to 76.9)1.0Time to CVC removal - median hours (range)72 (2–801)72 (24–801)67.5 (2–314)0.67Focal collection requiring drainage - n (%)(95% CI)7 (6.9) (3.4 to 13.6)0 (0.0) (0.0 to 11.4)7 (9.9) (4.9 to 19.0) 0.10 NASAB and other bacteria grown - n (%)(95% CI)9 (8.9) (4.8 to 16.1)3 (10.0) (3.5 to 25.6)6 (8.5) (3.9 to 17.2)1.0MRSA infection - n (%)(95% CI)3 (3.0) (1.0 to 8.4)1 (3.3) (0.6 to 16.7)2 (2.8) (0.8 to 9.7)1.0Antibiotic duration - median days (range), n=9314 (2–77)14 (2–56), n=2815 (7–77), n=65 0.0060 NALength of stay - median days (range)13 (0–749)4 (0–749)17 (0–459) 0.0020 NAAcute in-hospital mortality - n (%)(95% CI)1 (1.0) (0.2 to 5.4)0 (0.0) (0.0 tp 11.4)1 (1.4) (0.2 to 7.6)1.0Bold p-values indicate p≤0.10.CRP, C-reactive protein; CVC, central venous catheter; IVDU, intravenous drug use; MRSA, methicillin-resistant Staphylococcus aureus; OR, odds ratio; SAB, Staphylococcus aureus bacteraemia ; TTE, transthoracic echocardiography. Characteristics and outcomes of children with SAB, grouped by TTE status Bold p-values indicate p≤0.10. CRP, C-reactive protein; CVC, central venous catheter; IVDU, intravenous drug use; MRSA, methicillin-resistant Staphylococcus aureus; OR, odds ratio; SAB, Staphylococcus aureus bacteraemia ; TTE, transthoracic echocardiography. When subgroup analyses resulted in zero events in one group, statistical comparisons could not be performed and p-values are reported as ‘not applicable’ (NA). Missing data Data were available for all key clinical variables and outcome measures. We found no systematic patterns of missing data across the TTE grouping categories. We therefore used an available case analysis approach, using all observed data for each variable. For variables with missing data exceeding 5% (peak CRP: 10.9%; antibiotic duration: 7.9%), the number of available observations is reported alongside the results. Patient and public involvement This was a retrospective analysis of clinical data. Patients and the public were not involved in the design, conduct, reporting, or dissemination of this research. Results A total of 109 children with SAB meeting eligibility criteria were identified. Eight children were excluded from the study because their electronic records were not available, positive blood culture was obtained post-mortem, blood culture was positive for a different micro-organism or they died shortly after the positive blood culture was taken (within 24 hours) (figure 1). 101 children remained eligible for inclusion in the study. Flow diagram of study subjects showing identification, eligibility, and grouping of children with S. aureus bacteraemia (SAB) from April 2018 to March 2023 across six hospitals in North West London. Created by author. Subject characteristics and comparison of children undergoing TTE versus no TTE Across the cohort of 101 children, the median age at admission was 4.3 years (range 0.0–17.6), with a male:female ratio of 1.5:1 (60:41) (table 1). TTE was performed in 71 (70.3%) children and not performed in 30 (29.7%). No significant differences were observed in demographic characteristics between groups. Methicillin-sensitive S. aureus accounted for 88.1% (89/101) of isolates, methicillin-resistant S. aureus for 3.0% (3/101), and 8.9% (9/101) had mixed infections with MSSA and other bacteria. The additional bacteria isolated with MSSA included Streptococcus pyogenes, Staphylococcus hominis, Enterobacter cloacae, Staphylococcus epidermidis, Streptococcus milleri, Streptococcus vestibularis, beta-haemolytic Streptococcus (species not reported), and coagulase-negative Staphylococcus (species not reported). The majority of infections were community-acquired (76.2%, 77/101). Polymicrobial growth on initial blood culture was no more common in CVC-associated cases than non-CVC cases (8.0%, 2/25, vs 9.2%, 7/76; Fisher’s exact p=1.0). Neonates (≤28 days at initial blood culture) accounted for 7/71 (9.9%) of the TTE group and 3/30 (10.0%) of the no-TTE group, with no significant difference between groups (Fisher’s exact p=1.0). No primary focus was identified in 27.7% (28/101). Among children with an identified focus, the most common were musculoskeletal (24/101, 23.8%), line-associated (21/101, 20.8%), cutaneous (12/101, 11.9%), respiratory (7/101, 6.9%), infective endocarditis (4/101, 4.0%), and suspected contaminants (3/101, 3.0%). Only one child (1.0%) had a history of intravenous drug use. The prevalence of established IE risk factors was lower in the no-TTE group than the TTE group. Congenital heart disease was present in 16.8% (17/101) of children overall and was less common in the no-TTE group (6.7% vs 21.1%, p=0.088). The most common cardiac diagnoses were persistent ductus arteriosus (n=7), atrial septal defect (n=3) and ventricular septal defect (n=2). The prevalence of IE was substantially higher in children with congenital heart disease (11.8%, 2/17) compared with those without (2.4%, 2/84). No children had prosthetic valves, intracardiac devices, previous infective endocarditis or acquired valvular disease. A pre-existing murmur was documented in 6.9% (7/101), and an additional 7.4% (7/94) of children presented with a new-onset murmur; in both groups, all documented murmurs occurred in children who underwent TTE. The prevalence of CVCs was similar between groups (20.0% vs 26.8%, p=0.62). No children had immunological phenomena at presentation. Of the nine children (8.9%) in whom meningitis was clinically suspected, none had positive CSF cultures for S. aureus or CSF pleocytosis (WCC>5 × 10⁶/L). Children who underwent TTE had features indicative of more severe illness. Peak CRP was higher in the TTE group (median 125 vs 44 mg/L, p=0.013), with a non-significant trend towards longer fever duration (1 vs 0 days, p=0.087). Surgical drainage of focal collections was performed in 7/101 children, all in the TTE group (9.9% vs 0%, p=0.10). Children in the TTE group received longer antibiotic courses (median 15 vs 14 days, p=0.0060), with prolonged courses (>21 days) given to 41.5% in the TTE group versus 28.6% in the no-TTE group. Repeat blood cultures were more commonly performed in the TTE group than the no-TTE group (88.7% vs 53.3%, OR 6.73, 95% CI 2.20 to 22.11, p<0.001). Length of hospital stay was substantially longer in the TTE group (median 17 days, range 0–459) than the no-TTE group (median 4 days, range 0–749, Mann-Whitney p=0.0020). The wide range in both groups reflects a heterogeneous mix of straightforward short admissions and prolonged complex stays. Acute in-hospital mortality occurred in 1/71 (1.4%) of the TTE group and 0/30 (0%) of the no-TTE group (Fisher’s exact p=1.0). Of 25 children with CVC-associated SAB, 14 had the line removed and 11 had the line retained. IE was diagnosed in 2/11 (18%) of children with retained lines compared with 0/14 of children whose line was removed. Persistent bacteraemia at ≥72 hours was similar between these groups (1/9 vs 1/14). Six children had embolic phenomena at presentation, of whom four underwent TTE and two did not. The two children with embolic phenomena who did not undergo TTE received antibiotic durations of 14 and 42 days respectively, and no further complications were noted. Of the seven children with established IE risk factors who did not undergo TTE (two with CHD, six with CVC (one also having CHD), none with IVDU), none had positive blood cultures at ≥72 hours, none had multifocal infection and none had vertebral osteomyelitis. One had embolic phenomena (described above) and one was admitted to ICU. Comparison of children with IE on TTE versus negative TTE Of the 71 children who had TTE performed, IE was confirmed in four (5.6%). All four IE cases were right-sided. The small number of children diagnosed with IE means that statistical comparisons between these groups should be interpreted with caution. All children with IE had established risk factors: congenital heart disease (2/4, 50.0%), previous history of intravenous drug use (1/4, 25.0%) or CVCs (2/4, 50.0%). Features significantly associated with IE compared with negative TTE included embolic phenomena (75% vs 1.5%, p<0.001), pre-existing heart murmur (50% vs 7.5%, p=0.046) and polymicrobial growth on initial blood culture (50% vs 6.0%, p=0.033) (table 2). Persistent bacteraemia was also more common in children diagnosed with IE: positive cultures at <48 hours occurred in all three IE cases with repeat cultures performed, 3/3 (100%) versus 12/36 (33.3%) in the negative TTE group (p=0.050); at 48–72 hours, 2/3 (66.7%) versus 5/32 (15.6%, p=0.095); and at >72 hours, 2/3 (66.7%) versus 2/41 (4.9%, p=0.018). Table 2 Characteristics of children with negative TTE vs IE on TTE CharacteristicNegative TTE (n=67)IE on TTE (n=4)p-valueAge at admission - median years (range)4.1 (0.0–17.4)5.5 (0.0–17.6)0.86Sex - n (%) (95% CI)0.64Male39 (58.2) (46.3 to 69.3)3 (75.0) (30.1 to 95.4)Female28 (41.8) (30.7 to 53.7)1 (25.0) (4.6 to 69.9)Congenital heart disease - n (%) (95% CI)13 (19.4) (11.7 to 30.4)2 (50.0) (15.0 to 85.0)0.19IV drug use - n (%) (95% CI)0 (0.0) (0.0 to 5.4)1 (25.0) (4.6 to 69.9) 0.056 CVC present - n (%) (95% CI)17 (25.4) (16.5 to 36.9)2 (50.0) (15.0 to 85.0)0.29New-onset heart murmur - n (%) (95% CI)6 (9.0) (4.2 to 18.2)1 (25.0) (4.6 to 69.9)0.35Pre-existing heart murmur - n (%) (95% CI)5 (7.5) (3.2 to 16.3)2 (50.0) (15.0 to 85.0) 0.046 Embolic phenomena - n (%) (95% CI)1 (1.5) (0.3 to 8.0)3 (75.0) (30.1 to 95.4) <0.001 Community-acquired - n (%) (95% CI)49 (73.1) (61.5 to 82.3)2 (50.0) (15.0 to 85.0)0.31Fever duration - median days (range)1.0 (0.0–14.0)6.0 (0.0–14.0)0.98Peak CRP mg/L - median (range)124 (0.8–403), n=61237 (124–363), n=4 0.083 Requirement for intensive care - n (%) (95% CI)14 (20.9) (12.9 to 32.1)2 (50.0) (15.0 to 85.0)0.22Vertebral osteomyelitis - n (%) (95% CI)1 (1.5) (0.3 to 8.0)0 (0.0) (0.0 to 49.0)1.0Meningitis suspected - n (%) (95% CI)8 (11.9) (6.2 to 21.8)0 (0.0) (0.0 to 49.0)1.0Multifocal infection - n (%) (95% CI)7 (10.4) (5.2 to 20.0)1 (25.0) (4.6 to 69.9)0.39CVC removal - n/N (%) (95% CI)11/17 (64.7) (41.3 to 82.7)0/2 (0.0) (0.0– to 65.8)NATime to CVC removal - median hours (range)67.5 (2–314)NANAFocal collection drained - n (%) (95% CI)7 (10.4) (5.2–20.0)0 (0.0) (0.0–49.0)1.0MRSA infection - n (%) (95% CI)2 (3.0) (0.8–10.2)0 (0.0) (0.0–49.0)1.0SAB and other bacteria grown - n (%) (95% CI)4 (6.0) (2.3–14.4)2 (50.0) (15.0–85.0) 0.033 Antibiotic duration - median days (range)14 (7–77), n=6142 (14–42), n=40.41Bold p-values indicate p≤0.10. CRP, C-reactive protein; CVC, central venous catheter; IE, infective endocarditis; MRSA, methicillin-resistant Staphylococcus aureus; SAB, Staphylococcus aureus bacteraemia ; TTE, transthoracic echocardiography. Characteristics of children with negative TTE vs IE on TTE Bold p-values indicate p≤0.10. CRP, C-reactive protein; CVC, central venous catheter; IE, infective endocarditis; MRSA, methicillin-resistant Staphylococcus aureus; SAB, Staphylococcus aureus bacteraemia ; TTE, transthoracic echocardiography. Apparent associations whose interpretation was limited by the small IE group size included higher peak CRP (median 237 vs 124 mg/L, p=0.083), higher prevalence of congenital heart disease (50% vs 19.4%, p=0.19) and history of intravenous drug use (25% vs 0%, p=0.056). Of the four IE cases, two had retained CVCs that were not removed during admission. No significant differences were observed for other clinical variables, including requirement for intensive care, vertebral osteomyelitis, suspected meningitis, multifocal infection and focal collection drainage (all p>0.10). Discussion In this study, IE was uncommon among children diagnosed with SAB, occurring in only 4 (5.6%) of 71 children who underwent TTE. All four cases of IE were diagnosed in children with established risk factors (congenital heart disease, previous intravenous drug use, or presence of a central venous catheter). Children who underwent TTE had features indicative of more severe illness (higher CRP, longer fever duration, longer hospital stay, longer duration of antibiotic therapy and more frequent repeat blood cultures). This suggests that clinicians already informally risk-stratify when deciding to request TTE for children with SAB. A similar pattern of selective TTE has been previously described by Bui et al.4 However, this stratification was not consistently applied: two children with congenital heart disease and two children with embolic phenomena at presentation did not undergo TTE, despite both being well-recognised indications.2 Clinicians also appeared to risk stratify when choosing which children should have repeat blood cultures: nearly half of children in the no-TTE group did not have any repeat cultures performed, despite this being recommended in all children with SAB.2 Several established risk factors for IE were more common in the positive-TTE group than the negative-TTE group: congenital heart disease (50.0% vs 19.4%, p=0.19), pre-existing heart murmur (50.0% vs 7.5%, p=0.046) and a history of intravenous drug use.3 6 14 The higher prevalence of pre-existing murmurs in the positive-TTE group may reflect previously unrecognised structural cardiac lesions in some of these children. There was a higher incidence of embolic phenomena (75% vs 1.5%) in children with a positive TTE: this association has been described previously in the literature.7 8 Children with a positive TTE were more likely to have persistent bacteraemia at every time point examined, while children with negative TTE showed progressively lower rates of bacteraemia over time.2,4 This pattern supports current guidance which states that repeat blood cultures should be performed to document bacteraemia clearance in all children with SAB.2 In children with CVC-associated SAB, IE was identified only in those whose lines were retained (2/11 vs 0/14 in the line-removed group). Although the small numbers limit interpretation, this finding supports current recommendations to remove infected central venous catheters where feasible in children with SAB.2 Peak CRP was nearly twice as high in IE compared with negative TTE (237 vs 124 mg/L, p=0.083), consistent with McNeil et al.’s suggestion to use CRP >100 mg/L as a diagnostic adjunct when IE is suspected.3 The use of CRP as part of a risk-stratification tool would be particularly appealing given its widespread availability and routine measurement in children with suspected bacteraemia. However, CRP is often raised in children with SAB in the absence of IE and is therefore unlikely to be useful in isolation. One finding that does not have a clear precedent in the existing literature is the polymicrobial-IE association we observed; this was not explained by CVC presence alone and merits further investigation in larger cohorts. Strengths of our study include its multi-centre design across six hospitals and the collection of detailed clinical, microbiological and outcome data for each child. Several limitations do however warrant consideration. The study’s retrospective design introduces selection bias, in that severely ill children were more likely to undergo TTE and repeat blood culture testing. This is likely to have inflated the apparent prevalence of IE among children who underwent TTE, and we cannot exclude undiagnosed IE in children who did not have TTE performed. The small number of IE cases (n=4) limits statistical power and the external validity of comparisons involving this group. We did not collect per-patient transfer data or data on SAB recurrence. In our experience working in the North West London sector, availability of on-site paediatric echocardiography is variable and dependent on which staff are present at the time, which may have influenced clinicians’ threshold for requesting TTE. Children who died within 24 hours of the index blood culture were excluded from the cohort by design, so the mortality figures we report reflect mortality among children who survived the immediate post-culture period. We did not have data to assess the quality of TTE examinations. In conclusion, our findings support a risk-stratified approach to TTE in paediatric SAB. While clinicians already appear to informally select a lower-risk group of children when deciding to omit TTE and repeat blood cultures, this stratification is not consistently carried out. Specifically, we propose that TTE could reasonably be omitted in children with SAB who have a structurally normal heart, no established risk factors for IE, and no clinical features suggestive of IE. TTE should be performed regardless of other risk factors when any of the following are present: embolic phenomena, persistent bacteraemia at >72 hours, polymicrobial growth on blood culture or a pre-existing heart murmur. Our data support the continued use of TTE in children with established risk factors for IE, congenital heart disease, intravenous drug use or central venous catheters, given that all four (5.6%) IE cases in our cohort occurred in this group.3 14 These suggestions need prospective validation. Larger multi-centre studies should test a structured risk-stratification tool against routine TTE to assess definitively whether selective omission of TTE in low-risk children is associated with worse outcomes. This approach has the potential to reduce unnecessary investigation without compromising patient safety, particularly in settings where on-site paediatric echocardiography is not readily available. Acknowledgements We thank Ola Czyz for her valuable support in facilitating data collection for the study. Footnotes Funding: This research received no funding from any agency in the public, commercial, or not-for-profit sectors. Infrastructure support was provided by the NIHR Imperial Biomedical Research Centre. Data availability free text: The data that support the findings of this study are available from the corresponding author upon reasonable request, subject to approval by the relevant NHS Trusts (Imperial College Healthcare NHS Trust and London North West University Healthcare NHS Trust) and in compliance with information governance regulations. Data are not publicly available because they contain information that could potentially compromise patient confidentiality, given the rarity of infective endocarditis in children. Patient consent for publication: Not applicable. Provenance and peer review: Not commissioned; externally peer reviewed. Patient and public involvement: Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research. Data availability statement Data are available upon reasonable request. References Data Availability Statement Data are available upon reasonable request.