Demographic Characteristics of the Cohorts
A total of 957 patients were included in the study between September 2015 and February 2020. Group 1 consisted of 893 hospitalized children (58% male) with a median age of 33.9 months (IQR 16-109), who were compared to 64 outpatients (64% male) with a median age of 20.3 months (IQR 14.2- 33.8) in Group 2 (p=0.001).
Group 2: oral treatment group
Age and clinical manifestations
Clinical data of patients are described in Table 1. None of the patients under 3 months of age received oral treatment, 56/64 (87.5%) were between 3 months and 5 years old, and 8 (12.5%) were older than 5 years. The clinical diagnosis was OM in 38/64 cases (59.4%), SA in 17/64 patients (26.5%), both OM-SA and SD in 4/64 cases (6.3%) and PSI in 1/64 (1.5%). Children over 5 were always diagnosed with OM. The median of days of symptom onset before diagnosis was 3.5 (IQR 2- 6.5). Almost half of the patients developed fever at some point during their clinical course (48.8%).
Table 1
Comparative data between oral and intravenous treatment groups
| Group 1 (n= 893) | Group 2 (n=64) | p |
Male % | 58.2 % | 64.1% | 0,431 |
Age in months (IQR) | 33.9 (16-109) | 20.3 (14.2-33.8) | 0.001 |
Age (%) <3 months 3 months- 5 years >5 years | 27(3%) 502 (56.7%) 357 (40.3%) | 0 (0%) 56 (87.5%) 8 (12.5%) | < 0.0001 |
Temperature >37.9ºC | 559 (63%) | 31 (48.4%) | 0.024 |
Maximum temperature °C (IQR) | 39 (38.5- 39.5) | 38.8 (38.5-39.4) | 0,722 |
Diagnostic delay, days (IQR) | 4 (2-8) | 3.5 (2-6.5) | 0.189 |
Diagnosis (%) - OM - SA - OM-SA - SD - PSI | 346 (38.7%) 280 (31.3%) 184 (20.6%) 62 (6.9%) 21 (2.3%) | 38 (59.4%) 17 (26.5%) 4 (6.3%) 4 (6,3%) 1 (1.5%) | 0.008 |
Leucocytes/mm3 (IQR) | 12,095 (9,220-15,320) | 11,900 (9,500-14,700) | 0.644 |
Neutrophils/mm3 (IQR) | 6,500 (4,530-9,300) | 5,186 (3,822-6,500) | <0.001 |
ESR: mm/h (IQR) | 50 (32-71) | 41.5 (30-65) | 0.535 |
C-reactive protein; mg/L (IQR) | 34.5 (10.2- 80.4) | 15.6 (7-30) | <0.001 |
ERS/CRP ratio | 1.4 (0.6-3.6) | 3.3 (1.7-5.7) | <0.001 |
Procalcitonin ng/ml (IQR) | 0.1 (0.1-0.3) | 0.1 (0.1-0.2) | 0.570 |
Leukocytes/mm3 in joint fluid (IQR) | 68,040 (38,400-107,520) | 96,000 (60,900- 193,000) | 0,015 |
Number of arthrocentesis (median) | 1 | 2 | 0.001 |
Microbiology (%) - No organism detected - MSSA - MRSA - K.kingella - S.pneumoniae - S. pyogenes - S. agalactiae - Others | 478 (53.5%) 208 (23.3%) 13 (1.4%) 103 (11.5%) 15 (1.7%) 34 (3.8%) 4 (0.4%) 38 (4.2%) | 44 (68.75%) 2 (3.1%) 0 18 (28.1%) 0 0 0 0 | 0.065 <0.001 1.000 0.001 0.616 0.161 1.000 1.000 |
Hip arthritis (%) | 119 (25.4%) | 1 (4.5%) | 0.026 |
Shoulder arthritis (%) | 35 (7.5%) | 2 (9.1%) | 0.777 |
Knee artritis (%) | 172 (36.7%) | 13 (59.1%) | 0.034 |
Total days of treatment (IQR) | 28 (22-38) | 25 (21-28) | |
Complications (%) | 164 (19.7%) | 0 | < 0.001 |
Sequelae (clinical or radiological) (%) | 72 (8.7%) | 1 (1.7%) | 0.052 |
Group 1= intravenous, Group 2= oral, OM= osteomyelitis, SA= septic arthritis, OM-SA: OM associated SA, SD= spondylodiscitis, PSI= pyogenic sacroiliitis, IQR= interquartile range, ERS= erythrocyte sedimentation rate, MRSA= methicillin-resistant S.aureus, MSSA= Methicilin-susceptible S. aureus |
Biomarkers
The median leukocyte count was 11,900/mm3 (IQR 9,500- 14,700), and ESR and CRP levels were 41.5 (IQR 30–65) mm/h and 15.6 (IQR 7–30) mg/L, respectively.
Microbiology
In 68.7% of the cases, the microbiological studies did not yield an isolate. Among children in whom the causative agent was identified, the most frequent etiologic agent was K. kingae, in 18/20 children (90%); 3 by blood culture, 6 by synovial fluid culture and 12 by specific PCR. Methicilin-susceptible S. aureus (MSSA) was isolated in 2/20 patients by blood culture (10%).
In orally treated patients, no other germens were isolated.
Therapeutic approach
Only one patient was admitted due to oral intolerance with vomiting in the context of an intercurrent acute gastroenteritis. The most common oral antibiotic therapy was cefuroxime (43/64 71.7%), amoxicillin-clavulanate (11/64, 18.3%) and cefadroxil in 3 cases (5%). Three patients (4.6%) required a change of antibiotic therapy: one patient to narrow the spectrum according to the antibiogram and two patients due to poor oral tolerance. The median of length of therapy in Group 2 was 25 days (IQR 21-28 days).
Almost all patients received at least one arthrocentesis. This procedure was not performed in two cases of OM-SA (ankle and knee, respectively) because there was a minimal effusion and the decision was made to maintain an expectant attitude, showing posterior complete clinical remission. The rest of the 19 patients with arthritis (17/17 SA and 2/4 OM-SA) had the procedure carried out, and 18 patients underwent a joint lavage (94.7%). None of the patients required surgery.
Group 1 vs group 2
Age and clinical manifestations
The most relevant findings are shown in Table 1. The median age at diagnosis was higher in Group 1 than in Group 2 (34 vs 20 months; p=0.001). Group 1 had a higher percentage of arthritis at diagnosis (including SA and OM-SA) (51.9% vs 32.8%), whereas OM was the most frequent diagnosis in Group 2 (40.4% vs 59.4%, p = 0.011).
The duration of symptoms before diagnosis was longer for patients diagnosed with OM in Group 1 (4.5 days vs 3.5; p=0.031), however there were not significant differences when SA and OA were analyzed.
Patients in Group 1 presented fever more frequently (63% vs 48.4%; p=0.024), although the maximum temperature was comparable in both groups.
Some risk factors, such as previous trauma, injury or skin infection were present more frequently present in Group 1 (p <0.05). Regarding the anatomic locations of OAI, hip involvement was less frequent in Group 2 whereas knee location was more frequently observed in these patients.
Biomarkers
ESR and leukocyte count were comparable in both groups. However, both neutrophil count and CRP values were higher in children from Group 1 (p<0.001). The median ESR/CRP ratio was significantly higher in Group 2 (1.4 IQR 0.6-3.6 vs 3.3 IQR 1.7-5.7, p <0.001) and it was a useful parameter to distinguish K. kingae etiology from the rest of pathogens (p<0.01). A ratio <0.67 made it possible to rule out, with 91% sensitivity, the presence of K. kingae. Nevertheless, no good optimal cut-off point for CRP was identified in the logistic regression analysis, with sensitivity and specificity values that would allow, by itself, to rule out the appearance of complications.
Microbiology
Blood cultures were mostly sterile of both groups; however this was more frequently observed in Group 2 (74.8% vs 87.5%, p=0.036). As expected, among microbiological confirmed isolations, MSSA was more frequent in Group 1 (23.3% vs 3.1%, p <0.001) and K. kingae in Group 2 (11.5% vs 28.1%, p=0.001). Methicillin-resistant S. aureus (MRSA) was only present in Group 1.
The distribution of agents by age in the entire cohort is shown in Figure 1. Overall, S. aureus affected preferably older children and K. kingae was isolated more frequently in young patients, especially under 3 years of age. Overall, 76 patients with OM in Group 1 underwent bone puncture. The rate of bacterial isolation in bone puncture was 63.2%, most commonly S. aureus (33 cases; 30 MSSA, 3 MRSA) and K. kingella (6 cases).
Therapeutic approach
Joint lavage was performed in 194/464 (41,8%) children with arthritis in Group 1 and in 18/21 (85.7%) (16 SA and 2 OM-SA) in Group 2 (p=0.003). The median number of arthrocentesis performed per patient was higher in Group 2 (1 IQR 1-1 vs 2 IQR 1-2, p 0.001), as it was the cellularity of the joint fluid (median 68,040/mm3, IQR 38,400- 107,520 vs median 96,000/mm3 IQR 60,900-193,000, p 0.015).
Some type of surgical procedure was performed in 288 patients (34.9%) from Group 1, being elective in 137/288 (47.5%) of children and secondary to poor outcome and complications in 16.3% and 19.8%, respectively. None of the patients of Group 2 required surgery.
Outcome
There were no complications or clinical long-term sequelae described in Group 2. In contrast, some complications were described in Group 1: pyomyositis (7.7%), soft tissue and/or subperiosteal abscess (22.3%), deep venous thrombosis (4%), septic emboli (5%) and others (4.2%). One of the 6 SD affecting the cervical spine (all from Group 1) associated soft tissue abscess as a local complication (16.6%).
Clinical sequelae were identified in 70 patients (7.9%), especially dysmetria, mobility limitation, pain and limping, all from Group 1. Furthermore, radiological sequelae were seen in 51 patients (5.7%) from Group 1. Only one patient (1.7%) from Group 2 developed a radiological sequela, identifying fragmentation of the ossification nucleus without lytic lesion or impact on mobility.
Comparison between the treatments adjusted by age
Since the most of Group 2 children were between 3 months and 5 years old, an age adjusted comparison was performed, obtaining results very similar to the comparison of the entire cohort (Table 2). Fever was identified slightly more frequently in Group 1 (54% vs 41%, p= 0.047) as well as higher levels of CRP (23 IQR 9-53 vs 14.9 IQR 6-24.8 mg/L, p= 0.004).
Table 2
Comparison between the treatments adjusted by age
| Group 1 (n= 502) | Group 2 (n=56) | p |
Male % | 52 % | 57 % | NS |
Age in months (IQR) | 18.6 (13-27) | 17.54 (14-27) | NS |
Fever | 54 % | 41 % | 0.047 |
Diagnostic delay, days (IQR) | 4 (2-10) | 3 (1.2- 7) | NS |
Skin infections or recent surgery | 66 (13.1%) | 0 | 0.004 |
Diagnosis (%) - OM - SA - OM-SA - SD - PSI | 153 (30.5 %) 206 (41 %) 88 (17.5 %) 49 (9.8 %) 6 (1.2 %) | 30 (53.6 %) 18 (32.1 %) 4 (7.1 %) 4 (7.1 %) 1 (1.8 %) | <0.001 NS NS NS NS |
ESR: mm/h (IQR) | 51 (32-74) | 41.5 (30-72) | NS |
C-reactive protein; mg/L (IQR) | 23 (9- 53) | 14.9 (6-24.8) | 0.004 |
ERS/CRP ratio | 2 (0.9-5) | 3.6 (1.9-5.8) | 0.003 |
Microbiology (%) - No organism detected - MSSA - MRSA - K.kingella - S.pneumoniae - S. pyogenes - S. agalactiae - Others | 313 (62.3 %) 35 (7 %) 4 (0.8 %) 97 (19.3 %) 13 (2.6 %) 17 (3.4 %) 2 (0.4 %) 21 (4.2 %) | 37 (66 %) 1 (1.8 %) 0 18 (32.1 %) 0 0 0 0 | NS NS NS 0.02 NS NS NS NS |
Total days of treatment (IQR) | 28 (21-35) | 25 (21-28) | NS |
Surgery | 148 (29.8%) | 0 | <0.001 |
Complications (%) | 66 (13.1 %) | 0 | 0.001 |
Sequelae (clinical or radiological) (%) | 15 (2.9 %) | 0 | NS |
Group 1= intravenous, Group 2= oral, OM= osteomyelitis, SA= septic arthritis, OM-SA: OM associated SA, SD= spondylodiscitis, PSI= pyogenic sacroiliitis, IQR= interquartile range, ERS= erythrocyte sedimentation rate, MRSA= methicillin-resistant S.aureus, MSSA= Methicilin-susceptible S. aureus |
Table 3: Proposed criteria for initiating oral antibiotic therapy (all must be fulfilled) *
- Good general condition
- Appropriate oral tolerance
- No underlying disease
- Age > 6 months - < 3 years
- CRP < 80 mg/L
- ESR/CRP ratio ≥ 0,67**
- No history of injury, skin infections or recent surgery
- No local complications at onset
- No cervical spondylodiscitis
- Possibility of being closely monitored
- Patient's legal guardians provided informed consent.
* Individualize in case of hip arthritis
**ESR measured in mm/h. CRP measured in mg/L
ESR: erythrocyte sedimentation rate; CRP: C reactive protein
The diagnosis of OM was more frequent in Group 2 (30.5% vs 53.6%, p <0.000). Group 2 also presented a higher ESR/CRP ratio (2 IQR 0.9-5 vs 3.6 IQR 1.9-5.8, p= 0.003) and K.kingae identification (19.3% vs 32.1%, p = 0.02). Surgery, complications and clinical sequelae were only present in Group 1. No statistically significant differences in the rest of clinical data were found between Group 1 and 2.
Predictors of complications and sequelae
Exclusively oral antibiotic treatment, adjusted for age, sex, pathology, pathogens and risk factors, was not a statistically significant risk factor for functional or radiological sequelae [OR 0.28 (0.04-2.12); p 0.219]. In the multivariate analysis, the parameters associated with functional or radiological sequelae were infection caused by MRSA (versus sterile) [OR 6,91 (2.03-23.48); p 0.002], by “other pathogens” (versus sterile) [OR 2.32 (1.16-4.65); p 0.017] and a history of wound or skin infection [OR 2.08 (1.09-3.95); p 0.026].
Low – risk criteria proposal
New and stricter low risk criteria are proposed with the aim of identifying patients with low risk of complications and sequelae, in which K. kingae is very probably the causative agent (Table 3). Among the 172/957 patients who received adequate treatment and fulfilled the rest of the new low risk criteria: only one patient developed long-term sequelae (not specified) and S.aureus was isolated in four patients (2,3%), all of them with favorable outcomes.