This study is based on patients recruited in the EUCLIDS (www.euclids-project.eu), funded by the European Union's Seventh Framework Programme (FP7). EUCLIDS was a prospective, observational, multicenter study aiming to determine the genetic basis of life-threatening bacterial infections of childhood. The network, consisting of 194 hospitals in nine European countries and 2 hospitals in The Gambia, Africa, recruited patients from July 2012 to May 2017. Harmonized procedures for inclusion and exclusion criteria as well as clinical syndrome definitions are published elsewhere (17). The study protocol was approved by the Ethical Committee of the Medical University of Graz (24-116 ex 11/12) and by the local Ethical Committees from all participating centers. The study was conducted adherent to the Declaration of Helsinki principles and Good Clinical Practice guidelines. Written informed consent was obtained from parents or legal guardians for each child before study inclusion.

Children and young adolescents from 1 month to 18 years of age admitted to hospital with sepsis (or suspected sepsis) and/or severe focal infection were eligible for EUCLIDS. Sepsis was defined as a systemic inflammatory response syndrome (SIRS) plus suspected infection according to Goldstein et al. (18). For this study, prospectively recruited patients with bone and joint infections were selected from the EUCLIDS database. Clinical definitions for POAIs can be found in the Supplementary File A. Fifty-two patients from the Swiss center were excluded due to divergent inclusion criteria, together with 27 patients from the non-European arm The Gambia. Two additional patients were excluded, one suffering from chronic recurrent multifocal osteomyelitis (CRMO) and one with insufficient clinical information. In line with other POAI studies, patients were subdivided into three groups of ages for data analysis: infants (≤ 3months), young children (>3–60 months), and older children (>60 months) (14, 19, 20).

For the present study, the EUCLIDS database was queried for demographics, illness severity, therapies, blood count, microbiology, operations, and outcome at hospital discharge. Orthopedic operations were defined as an open surgery, including arthroscopy. Procedures such as needle puncture or drainage alone were not classified as orthopedic surgeries.

For a better overview, localizations affected by osteoarticular infections (OAIs) were grouped into regions. In OM, the skull region also includes the frontal sinuses and temporomandibular joint, the shoulder region includes clavicle and scapula, and the lower arm region includes radius and ulna. The hand region in SA only contains carpal/metacarpal joints, whereas in OM, it describes phalanges, wrist, and certain fingers. In SA, the pelvis describes only the sacroiliac joint, whereas in OM, it includes sacral, iliac, ischial, and pubic bone. The foot region in OM includes os calcalneum, os talus, joints distal to ankle joints, tarsal/metatarsal bone, metatarsophalangeal joints, and toes. In SA, it describes only subtalar joints.

Microbiological diagnosis was performed according to locally available microbiological investigations including bacterial culture from sterile sites (blood, cerebrospinal fluid, and invasive diagnostic samples), and from unsterile sites (throat and wound swabs).

Identification and resistance testing were performed according to standard procedures and, where available, these included molecular techniques such as PCR. Primers used for PCR diagnostics are exemplary as shown in Supplementary Table 1 and account for at least 60% (15/25) of all PCRs performed in our cohort (8, 21).

Bacteria that were rarely detected (in only one to a maximum of three patients) were classified into a group named “other bacteria” and consisted of Actinomyces oris, Fusobacterium necrophorum, Group B Streptococcus, Mycobacterium kansasii, Mycobacterium tuberculosis, Staphylococcus epidermidis, Staphylococcus hominis, Haemophilus influenzae, Streptococcus milleri, Staphylococcus warneri, and other unidentified Streptococci.

At discharge, patient outcomes were categorized according to the Pediatric Overall Performance Category (POPC) score (22), determined either by direct observation or by chart review (Supplementary Table 2).

Data are presented as medians with interquartile ranges (IQRs) or numbers (percentage). For readability, percentages in all tables were rounded. Characteristics were compared across disease and age groups, using Kruskal–Wallis test with Dunn's correction or Mann–Whitney test (continuous data) as well as Fisher's exact tests or Pearson's chi-squared test (categorical data), as appropriate. Statistical significance was set at 0.05. To reduce the likelihood of false-positive results caused by multiple testing, α was adjusted with Bonferroni correction. Nonsignificant analyses are not shown.

For all analyses, a two-sided p value < 0.05 was considered significant. Statistical analyses were conducted in R (version 3.5.1) (23) and with the GraphPad Prism Software Version 8 (GraphPad Software, La Jolla, CA, USA).

We analyzed a cohort of 380 patients with POAIs, including 203 (53.4%) with OM, 158 (41.6%) with SA, and 19 (5%) with both OM and SA, recruited in five European countries (Figure 1). The median age at admission was 60 months (IQR, 17–122), and 162 (42.6%) of them were female (Table 1). Only 2.7% (10/380) of the whole cohort were infants (<3 months), whereas young and older children were almost evenly distributed [47.6% (181/380) and 49.7% (189/380), respectively] (Table 2). Young children were more often diagnosed with SA compared to OM [101/158 (63.9%) vs. 74/203 (36.4%), (p < 0.0001)] and OM + SA [101/158 (63.9%) vs. 6/19 (31.5%), (p = 0.011)]. Older children predominantly suffered from OM [OM vs. SA: 123/203 (60.5 %) vs. 54/158 (34.2%), (p < 0.0001)] (Table 2).

In total, the number of patients diagnosed with a systemic infection was 66 of which 19 (28.7%) were admitted to the pediatric intensive care unit (PICU) (Table 1). Two patients suffered from septic shock and four from toxic shock syndrome. No patient died, one needed an amputation of the right foot and four left toes, and two received skin transplantation. No infant (<3 months) suffered from systemic infection (Table 2).

Osteomyelitis mainly affected the femur [36/202 (17.8%)] and tibia [31/202 (15.3%)]. Patients with SA predominantly showed infections of the hip [62/151 (41.1%)] and the knees [53/151 (35.1%)] (Figure 2). Multifocal bone or joint infection sites were identified in 12 patients (9 with OM and 3 with SA). Thirty-one patients suffered from chronic OM (defined as beginning of symptoms >1 month before admission).

Compared to patients with SA, patients with OM received longer intravenous antibiotic therapies (median of 11 days vs. 5, p < 0.0001) and had longer hospital stays (median of 10 days vs. 7, p = 0.0005).

Fifty-six percent (213/380) of patients needed orthopedic surgery. Patients with SA were more often submitted to operations than patients with OM [118/158 (74.6%) vs. 82/203 (40.3%), (p < 0.0001)] (Table 1) and showed a reduced mobility at discharge [SA vs. OM: 36/58 (62.1%) vs. 53/125 (42.4%), (p = 0.01)].

According to the Pediatric Overall Performance (POPC) Score (Supplementary Table 1), 301 of 340 (88.5%) patients with POAIs showed a good overall performance at discharge, 36 (10.5 %) a mild overall disability, 3 (0.8%) a moderate, and 1 (0.2%) a severe overall disability. More young children (3–60 months) with OM + SA showed a POPC score of two compared to the same age group with OM [OM + SA vs. OM: 3/5 (60%) vs. 6/64 (9.3%), (p = 0.013)] or SA [OM + SA vs. SA: 3/5 (60%) vs. 2/97 (2.1%), (p = 0.0006)].

A causative organism was detected in 247/380 (65%) of patients. S. aureus was identified in 141/247 (57.1%) of all microbiological confirmed cases, followed by GAS in 45 (18.2%), K. kingae in 22 (8.9%), and S. pneumoniae in 16 (6.5%) cases (Table 1).

Staphylococcus aureus was more frequently detected in patients with OM compared to patients with SA [92/122 (75.4%) vs. 39/110 (35.4%), (p < 0.0001)] (Table 1).

In 66 cases with systemic infections, S. aureus was found in 35 (53%) patients, with 4 of them suffering from toxic shock syndrome (Figure 3). One patient was identified with MRSA, 6 with a PVL- positive S. aureus, and one with a toxic-shock-syndrome-toxin 1 (TSST-1) producing S. aureus.

Group A Streptococcus was found in 15/66 (22.7%) of patients with systemic infections and with high predominance in young children with systemic infections and SA [7/15 (56.6%)] (Figure 3 and Table 2).

Kingella kingae was only detected in young children with focal SA [22/101 (21.7%), (p < 0.0001)] (Table 2). S. pneumoniae was more frequently identified in patients with SA compared to patients with OM [13/110 (11.8%) vs. 3/122 (2.4%), (p = 0.011)] (Table 1). No differences in the number of identified pathogens were found between the age groups.

Blood was the major source of causative pathogen detection in OM [73/141 (60%)], while this was the case for joint aspirate [70/126 (55.5%)] in SA (Figure 4A). The methods used for the identification of organisms were mainly culturing techniques in OM and SA [106/110 (96.3%) and 87/106 (82%)], followed by PCRs in SA [19/106 (17.9%)] (Figure 4B). K. kingae and GAS were predominantly identified in joint aspirate of patients with SA, while the main method of detection for K. kingae was PCR and culture for GAS (Figure 4B). Culture yielded positive results from blood more often in patients with OM [72/114 (63.2%)] than in patients with SA [42/114 (36.8%)]; the latter showed culture-positive pathogen detection more often when taken from joint aspirate [51/60 (85%) vs. OM 9/60 (15%)]. PCR-positive pathogen detection from joint aspirate was only seen in patients with SA (Figure 4C).

Past medical history showed five patients with sickle cell disease [3/203 (1.5%) with OM and 2/158 (1.3%) with SA]. In three of these cases, no pathogen was found, whereas in one case S. pneumoniae was identified and in one case S. aureus.

Neutrophils were not significantly different between the disease groups (Figures 5A,C). Patients with SA showed higher median CRP serum levels than patients with OM [82 mg/L vs. 52 mg/L, (p = 0.017)] (Figure 5B). Comparisons between the age groups revealed that young children with SA showed higher serum CRP levels than young children with OM [(63 mg/L vs. 38 mg/L, (p = 0.031)] (Figure 5D).

Cefuroxime [168/380 (44.2%)], clindamycin [109/380 (28.6%)], and flucloxacillin [96/380 (25.2%)] were most frequently prescribed among all patients with OAIs and all countries included in the consortium (Figures 1, 6). The main three substances were the same in patients with OM—cefuroxime [78/203 (38.4%)], clindamycin [67/203 (33%)], and flucloxacillin [51/203 (25.1%)]; in patients with SA—cefuroxime [58/158 (36,7%)], flucloxacillin [39/158 (24.6%)], and clindamycin [32/158 (20.3%)]; and in patients with both OM and SA—clindamycin [10/19 (52.6%)], flucloxacillin [6/19 (31.5%)], and cefuroxime [5/19 (26.3%)].

Comparison of antimicrobial differences between the countries revealed that the UK and the Netherlands showed higher administrations of flucloxacillin [UK, 54/204 (26.5%); the Netherlands, 37/57 (64.9%)] and ceftriaxone [UK, 52/204 (25.5%); The Netherlands, 11/57 (19.3%)]. In Austria, patients with OM and SA are mainly treated with clindamycin [25/34 (73.5%) vs. 6/8 (75%)] and cefuroxime [23/34 (67.6%) vs. 8/8 (100%)]. In Germany as well, patients with OM and SA are mainly treated with clindamycin [18/25 (72%) vs. 3/4 (75%)] and cefuroxime [14/25 (56%) vs. 3/4 (75%)].