Abstract
Background
Blood and pus culture are still considered standard methods for etiological investigations on osteoarticular infection (OAI) in children, but negative culture results are common and may increase the complexity of disease diagnosis and treatment. This study aims to explore the clinical characteristics and risk factors of culture-negative OAI and construct a prediction model for determining the likelihood of negative culture results.
Methods
This single-center retrospective analysis included pediatric OAI cases diagnosed between 2014 and 2023. Demographic data, clinical features, and laboratory parameters were systematically collected. Comparisons were made between culture-positive and culture-negative groups for both blood and pus cultures. Multivariate logistic regression analysis was performed to identify independent predictors of negative culture results, which were used to develop a nomogram-based predictive model. The discriminatory ability of the model was assessed using receiver operating characteristic curves, while its calibration was evaluated through the Hosmer–Lemeshow goodness-of-fit test and bootstrap calibration plots. The clinical utility of the model was determined using decision curve analysis.
Results
This study included 345 children diagnosed with OAI: blood culture was performed in 243 cases (38 culture-positive and 205 culture-negative), and pus culture was performed in 230 cases (115 culture-positive and 115 culture-negative). In the blood culture group, C-reactive protein (CRP) was identified as an independent predictor of negative blood culture. Receiver operating characteristic curve analysis based on CRP demonstrated a sensitivity of 73.7% and a specificity of 70.2% for predicting culture results. In the pus culture group, presence of septic arthritis, neutrophil percentage, and CRP level were identified as independent predictors of pus culture outcomes. The nomogram constructed using these predictors demonstrated good discriminative ability, calibration, and clinical utility.
Conclusion
CRP was identified as a key predictor of negative blood culture results. The presence of septic arthritis, neutrophil percentage, and CRP was identified as a key predictor of negative pus culture results. The nomogram constructed with pus culture predictors had good discrimination and clinical applicability and has potential for the early identification of culture-negative OAI cases in the clinic. This tool could optimize etiological investigation strategies and improve the efficiency of clinical diagnosis.
Supplementary Information
The online version contains supplementary material available at 10.1186/s40001-025-03773-z.
Keywords: Pediatric osteoarticular infections, Culture-negative, Nomogram, Risk factors
Introduction
Osteoarticular infections (OAI) in children primarily include osteomyelitis (OM), septic arthritis (SA), and concomitant septic arthritis and osteomyelitis (CSAO) [1, 2]. OAI is a serious condition that is primarily caused by bacterial and other pathogenic organisms [3, 4]. OAI has an incidence rate of 1–200 cases per 100,000 individuals worldwide [5, 6]. If not treated promptly or appropriately, OAI may lead to serious consequences, such as joint dislocation, growth disorders, limb deformities, and even life-threatening conditions [7–10]. Despite the continuous advancement of diagnostic technology, the etiological diagnosis of OAI in children is still challenging, with negative culture rates of 40–60% for pus cultures [11–13] and 80–90% for blood cultures [14, 15], which frequently complicates pathogen identification in clinical practice and consequently leads to prolonged empirical antibiotic therapy and delayed pathogen-directed treatment [16, 17].
In recent years, the application of molecular diagnostic techniques, such as polymerase chain reaction (PCR) and metagenomics next-generation sequencing (mNGS), in the detection of pathogens in children with OAI has gradually increased, and the positive detection rate is 10–20% higher compared with traditional culture [18, 19]. However, these methods are limited by their suboptimal sensitivity, high cost, and high technical requirements as a result of which traditional microbial culture is still regarded as the gold standard for etiological diagnosis [20, 21]. The characteristics of and factors associated with negative culture results have been studied in adults, but there is a paucity of systematic analysis of the clinical characteristics associated with negative culture results in children with OAI [22–24]. Therefore, this study tries to fill in this research gap by retrospectively analyzing 345 pediatric OAI cases over a span of 10 years to determine the clinical risk factors associated with blood culture and pus culture results and use the significant risk factors to construct a model to predict cases with a high likelihood of negative cultures. This model aims to support clinical decision-making by identifying cases in which conventional cultures are less likely to be informative and where molecular diagnostic techniques, such as PCR or mNGS, may be preferentially considered to improve etiological detection and diagnostic efficiency.
Methods
We retrospectively analyzed children who were hospitalized in Shenzhen Children's Hospital and considered to have a diagnosis of SA, OM, and CSAO between January 1, 2014, and December 31, 2023. Parents or legal guardians of all patients gave written informed consent.
The inclusion criteria for the study subjects were [25]: For SA, the diagnosis was made if one of the following criteria was met: (1) joint aspirate was purulent and/or Gram stain or culture detected bacteria; (2) there was clinical or imaging evidence suggesting arthritis (joint effusion) and blood culture was positive; or (3) synovial biopsy results supported the diagnosis of SA. For OM, the diagnosis was based on: (1) abnormal imaging (X-ray, ultrasound, computed tomography, or magnetic resonance imaging); or (2) positive bone tissue biopsy results accompanied by local swelling, heat, pain, and other clinical symptoms and signs consistent with osteomyelitis. Exclusion criteria were: (1) patients with chronic osteomyelitis (disease duration > 3 months) or infection after open fracture or surgery; (2) patients with diseases that may cause immune dysfunction (such as autoimmune diseases, blood system diseases, etc.); (3) patients with missing clinical data or incomplete medical records.
We extracted clinical data, including age, gender, onset time, site of onset, microbiological culture (blood and pus samples), and test indicators (white blood cells (WBC), neutrophil ratio, C-reactive protein (CRP), and erythrocyte sedimentation rate (ESR)) from the hospital's electronic medical record system [25].
The types of infectious diseases can be divided into SA, OM, and CSAO. The onset time is classified according to: within 2 weeks, 2 weeks to 3 months [26, 27].
Laboratory parameters, including WBC, CRP, and ESR, were measured at admission as part of routine clinical evaluation. CRP levels were determined using an immunoturbidimetric assay, with a reference interval of 0–20 mg/L. WBC was measured using an automated hematology analyzer, with a reference range of 5–15 × 109/L. ESR was determined using the Westergren method, with a reference range of 0–20 mm/h [28, 29]. All laboratory tests were performed in the central laboratory of our institution following standardized procedures.
Blood and pus specimens were obtained from children with suspected OAI by trained medical personnel after informed consent had been secured from parents or legal guardians. Blood cultures were routinely collected at the time of admission, particularly in patients presenting with systemic manifestations of infection, such as fever, or with elevated inflammatory markers. Pus samples were collected during surgical drainage or diagnostic aspiration. All specimens were immediately transported to the microbiology laboratory for processing. Blood cultures were performed using the BACT/ALERT VIRTUO automated blood culture system (bioMérieux, France) in accordance with the manufacturer’s instructions and institutional protocols. Pus specimens were inoculated onto sheep blood agar and China blue agar plates, followed by incubation at 35 °C in an atmosphere containing 5% CO₂ for 48 h. For slow-growing organisms or when fastidious pathogens were clinically suspected, incubation was extended for up to 5 days. Bacterial isolates from pus specimens were identified using the VITEK® 2 Compact system (bioMérieux, France) or matrix-assisted laser desorption/ionization–time-of-flight mass spectrometry (MALDI–TOF MS). Microbial identification was conducted in accordance with the Manual of Clinical Microbiology (11th edition). Blood and pus culture results were generally available within 2–5 days.
Enumeration data were expressed as the number of cases (percentages), and measurement data were expressed as mean ± standard deviation if they were normally distributed, otherwise as median (interquartile range). The differences in characteristics between the positive and negative groups of venous blood and pus culture were compared using chi-square tests (categorical variables), independent sample t tests (normal continuous variables), or Wilcoxon rank-sum tests (non-normal continuous variables). Variables with P < 0.2 in univariate analysis were included in a multivariate logistic regression model (conditional forward) to identify independent risk factors for culture results [30]. The receiver operating characteristic (ROC) curve and the area under the curve (AUC) were used to evaluate the effectiveness of blood culture predictors. Diagnostic performance of CRP was additionally evaluated at multiple predefined thresholds (> 20, > 40, > 45.85, > 60, > 80, and > 100 mg/L). Sensitivity, specificity, PPV, NPV, likelihood ratios, odds ratios, and 95% confidence intervals were calculated for each cut-off to support different clinical decision-making scenarios. For pus culture with multiple risk factors, a nomogram model was constructed using R software version 4.2.2 based on the logistic regression results to predict the probability of a negative pus culture in children with OAI. The discrimination of the model was evaluated by the ROC curve. The calibration of the model was evaluated by the Hosmer–Lemeshow test and calibration curve (bootstrap method, 1000 resampling). Decision curve analysis (DCA) was used to evaluate the clinical net benefit of the model at different threshold probabilities. P < 0.05 was considered statistically significant. We performed statistical analysis using R software (version 4.3.1; R Foundation for Statistical Computing, Vienna, Austria).
Results
Clinical characteristics of children with OAI
A total of 345 OAI cases were included, including 207 males (60.00%) and 138 females (40.00%), with a mean age of 4.23 ± 3.93 years. Of the 345 total cases, 251 (72.75%) were treated within 2 weeks of onset, and 94 (27.25%) were treated between 2 weeks and 3 months of onset. With regard to disease type, 131 cases (37.97%) were classified as SA; 111 cases (32.17%) as OM; and 103 cases (29.86%) as CSAO.
Since some children had multiple sites of infection, only the top five sites with the highest level of involvement were considered. Based on this criterion, the top five most frequently involved sites were the knee joint (102 cases, 29.39%), the hip joint (96 cases, 27.67%), the femur (89 cases, 25.65%), the tibia (39 cases, 11.24%), and the humerus (21 cases, 6.05%).
In terms of laboratory test results at the initial diagnosis, the WBC count was 12.35 ± 5.68 × 109/L; the neutrophil percentage, 52.97% ± 19.02%; CRP, 37.75 ± 39.92 mg/L; and ESR, 54.08 ± 32.04 mm/h.
The average hospitalization time for all the children was 25.71 ± 14.94 days, and the average hospitalization cost was 19869.44 ± 15571.94 RMB (Table 1).
Table 1.
Clinical characteristics of 345 children with OAI
| Clinical characteristics | Value |
|---|---|
| Clinical | |
| Age (years) | 4.23 ± 3.93 |
| Male, n (%) | 207 (60.00%) |
| Female, n (%) | 138 (40.00%) |
| Onset time | |
| 0–2 weeks, n (%) | 251 (72.75%) |
| 2 weeks–3 months, n (%) | 94 (27.25%) |
| Infection type | |
| SA, n (%) | 131 (37.97%) |
| OM, n (%) | 111 (32.17%) |
| CSAO, n (%) | 103 (29.86%) |
| Top 5 Affected Sites | |
| Knee, n (%) | 102 (29.39%) |
| Hip, n (%) | 96 (27.67%) |
| Femur, n (%) | 89 (25.65%) |
| Tibia, n (%) | 39 (11.24%) |
| Humerus, n (%) | 21 (6.05%) |
| Laboratory | |
| WBC (× 10⁹/L) | 12.35 ± 5.68 |
| Neutrophil percentage (%) | 52.97 ± 19.02 |
| CRP (mg/L) | 37.75 ± 39.92 |
| ESR (mm/h) | 54.08 ± 32.04 |
| Microbiological results | |
| Positive blood culture, n/N (%) | 38/243 (15.64%) |
| Positive pus culture, n/N (%) | 115/230 (50.00%) |
| Other | |
| Length of hospital stay (days) | 25.71 ± 14.94 |
| Hospital cost (CNY) | 19,869.44 ± 15,571.94 |
Pathogenic results of blood culture
Blood culture was performed in 243 cases, of which 38 cases (15.64%) had positive results and 205 cases (84.36%) had negative results. The pathogens detected in the positive blood cultures were divided into Gram-positive bacteria (34, 89.47%) and Gram-negative bacteria (4, 10.53%) (Table 2).
Table 2.
Microbiological and pathogen distribution of 38 cases of blood culture-positive OAI
| Classification | Category | Pathogen | n (%) |
|---|---|---|---|
| Gram-positive | Staphylococcus spp. | MSSA | 20 (52.63%) |
| MRSA | 6 (15.79%) | ||
| Staphylococcus hominis | 1 (2.63%) | ||
| Streptococcus spp. | Streptococcus pneumoniae | 5 (13.17%) | |
| Streptococcus pyogenes | 2 (5.26%) | ||
| Gram-negative | Enterobacteriaceae | Escherichia coli | 1 (2.63%) |
| Pseudomonas spp. | Pseudomonas aeruginosa | 1 (2.63%) | |
| Haemophilus spp. | Haemophilus influenzae | 1 (2.63%) | |
| Brucella melitensis | Brucella melitensis | 1 (2.63%) |
Among the Gram-positive bacteria, Staphylococcus aureus was the most frequently isolated pathogen and was represented by 20 strains of methicillin-susceptible S. aureus (MSSA, 52.63%) and 6 strains of methicillin-resistant S. aureus (MRSA, 15.79%). In addition, a Staphylococcus hominis isolate was detected in 1 case (2.63%). Some other commonly detected Gram-positive organisms included Streptococcus pneumoniae (5 cases, 13.17%) and Streptococcus pyogenes (2 cases, 5.26%).
The Gram-negative bacteria included Escherichia coli, Pseudomonas aeruginosa, Haemophilus influenzae, and Brucella melitensis, which were detected in one case each, with each species contributing to 2.63% of the total Gram-negative bacteria.
Comparison of clinical characteristics between negative and positive blood culture groups
The demographic data, clinical characteristics, and laboratory indicators of the negative and positive blood culture groups are shown in Table 3. The mean age of the blood culture-positive group was 4.02 ± 3.93 years, which was slightly higher than that in the negative group (3.88 ± 3.78 years) without being significantly different (P > 0.05). In terms of sex distribution, male children accounted for 50.00% (19/38) of all cases in the positive group and 60.98% (125/205) in the negative group, and there was no statistically significant difference between the two groups (P > 0.05).
Table 3.
Comparison of clinical characteristics between children with negative and positive blood culture results
| Clinical characteristics | Negative cultures (n = 205) | Positive cultures (n = 38) | P |
|---|---|---|---|
| Clinical | |||
| Age (years) | 3.88 ± 3.78 | 4.02 ± 3.93 | 0.837 |
| Male, n (%) | 125 (60.98%) | 19 (50.00%) | 0.206 |
| Onset time | |||
| 0–2 weeks, n (%) | 151 (73.66%) | 34 (89.47%) | 0.036 |
| 2 weeks–3 months, n (%) | 54 (26.34%) | 4 (10.53%) | 0.036 |
| Infection type | |||
| SA, n (%) | 78 (38.05%) | 11 (28.95%) | 0.285 |
| OM, n (%) | 64 (31.22%) | 10 (26.32%) | 0.546 |
| CSAO, n (%) | 63 (30.73%) | 17 (44.73%) | 0.092 |
| Top 5 affected sites | |||
| Knee, n (%) | 63 (30.00%) | 9 (20.45%) | 0.382 |
| Hip, n (%) | 54 (25.71%) | 15 (34.09%) | 0.099 |
| Femur, n (%) | 54 (25.71%) | 12 (27.27%) | 0.505 |
| Tibia, n (%) | 26 (12.39%) | 4 (9.09%) | 0.71 |
| Humerus, n (%) | 13 (6.19%) | 4 (9.09%) | 0.353 |
| Laboratory | |||
| WBC (× 10⁹/L) | 12.37 ± 5.64 | 15.04 ± 7.59 | 0.012 |
| Neutrophil percentage (%) | 51.85% (39.30–68.08%) | 68.50% (52.25–76.60%) | 0.001 |
| C-reactive protein (CRP, mg/L) | 23.96 (6.80–54.30) | 69.55 (33.36–112.90) | < 0.001 |
| Erythrocyte sedimentation rate (ESR, mm/h) | 50.00 (23.50–79.00) | 80.00 (56.25–98.75) | < 0.001 |
| Other | |||
| Length of hospital stay (days) | 24.91 ± 15.11 | 28.37 ± 11.42 | 0.182 |
| Hospital cost (CNY) | 15790.60 (9863.38–22898.51) | 24471.00 (18924.34–32624.79) | < 0.001 |
In terms of the duration of onset, 89.47% (34/38) of the children in the culture-positive group had an onset time of less than two weeks, compared to 73.66% (151/205) in the culture-negative group. Conversely, the proportion of children with onset time between 2 weeks and 3 months was 10.53% (4/38) and 26.34% (54/205) in the positive and negative groups, respectively. These differences were statistically significant (P = 0.036).
The incidence of SA was significantly higher in the culture-negative group (38.05%, 78/205) than in the culture-positive group (28.95%, 11/38). The incidence of OM was also higher in the culture-negative group (31.22%, 64/205) than in the culture-positive group (26.32%, 10/38), but the difference was not significant (P > 0.05). In contrast, cases of CSAO were slightly more prevalent in the positive group (44.73%, 17/38) than in the negative group (30.73%, 63/205), although the difference did not reach statistical significance (P > 0.05).
As mentioned earlier, only the top five most frequently affected sites were considered in the assessment of involved sites, and this led to a total of 254 sites across both the positive and negative groups. In the negative group, the knee joint was most commonly affected (30.00%, 63/210), followed by the hip joint (25.71%, 54/210) and the femur (25.71%, 54/210), with the lowest incidence found in the tibia (12.39%, 26/210) and the humerus (6.19%, 13/210). In the positive group, the hip joint was mainly affected (34.09%, 15/44), followed by the femur (27.27%, 12/44) and the knee joint (20.45%, 9/44), with the tibia and humerus having the lowest incidence again (9.09%, 4/44 each). There was no significant difference in the distribution of affected sites between the two groups (P > 0.05).
WBC count, neutrophil percentage, CRP, and ESR in the positive blood culture group were significantly higher (P < 0.05 for all) than those in the negative group: WBC, 15.04 ± 7.59 × 109/L versus 12.37 ± 5.64 × 109/L; neutrophil percentage, 68.5% (52.25–76.60%) versus 51.85% (39.30–68.08%); CRP, 69.55 mg/L (33.36–112.90 mg/L) versus 23.96 mg/L (6.80–54.30 mg/L); ESR, 80.00 mm/h (56.25–98.75 mm/h) versus 50.00 mm/h (23.50–79.00 mm/h).
The hospitalization costs of the blood culture-positive group were significantly higher than those of the blood culture-negative group: 24,471.00 (18,924.34–32,624.79) RMB versus 15,790.60 (9,863.38–22,898.51) RMB (P < 0.001). However, the difference in hospitalization time was not statistically significant: 28.37 ± 11.42 days in the positive group versus 24.91 ± 15.11 days in the negative group (P > 0.05).
Identification of independent predictors of blood culture results
To identify the independent risk factors for positive blood culture, multivariate logistic regression with conditional forward selection was used to perform multivariate logistic regression analysis [30] for the following significant variables with P < 0.2 in univariate analysis: duration of onset (< 2 weeks and 2 weeks to 3 months), presence of CSAO, WBC count, neutrophil percentage, CRP, and ESR (Table 4). Only CRP emerged as an independent predictor, with a regression coefficient of 0.021 and an odds ratio (OR) of 1.021 (95% CI 1.013–1.029; P < 0.001), which indicates that the likelihood of a positive blood culture increased by approximately 2.1% for each 1 mg/L increase in CRP.
Table 4.
Logistic regression analysis for predicting blood culture results
| Variables | Positive cultures | Negative cultures | Standard error | χ2 | P | ||||
|---|---|---|---|---|---|---|---|---|---|
| β | Odds ratio | 95% CI for odds ratio | β | Odds ratio | 95% CI for odds ratio | ||||
| CRP | 0.021 | 1.021 | 1.013–1.029 | −0.021 | 0.979 | 0.972–0.987 | 0.004 | 24.77 | < 0.001 |
| Intercept | −2.795 | 0.061 | – | 2.795 | 16.393 | – | 0.326 | 73.7 | < 0.001 |
When negative blood culture was used as the dependent variable, the regression coefficient for CRP was −0.021 and the OR was 0.979 (95% CI 0.972–0.987, P < 0.001). This indicates that elevated CRP was negatively correlated with negative blood culture: that is, for every 1 mg/L increase in CRP, the likelihood of a negative blood culture result was reduced by about 2.1%. ROC curve analysis showed that CRP showed moderate discriminative ability for predicting negative blood culture results, with AUC = 0.747 (95% CI 0.659–0.836, P < 0.001). At the optimal cut-off value of 45.850 mg/L, sensitivity was 73.7% and specificity was 70.2% (Fig. 1).
Fig. 1.
ROC curve of the clinical prediction model for blood culture-negative OAI in 243 children
For the prediction of negative blood culture results, progressively higher CRP thresholds were associated with a trade-off between sensitivity and specificity, characterized by declining sensitivity and increasing specificity. Lower cut-off values (e.g., > 20 mg/L) yielded greater sensitivity, whereas higher thresholds (e.g., > 80 or > 100 mg/L) provided improved specificity and positive predictive value, thereby allowing tailored application across different clinical scenarios (Table 5).
Table 5.
Diagnostic reliability of different measures of CRP for predicting blood culture results
| CRP | Sensitivity | Specificity | PPV | NPV | LR + | LR− | OR | 95%Cl |
|---|---|---|---|---|---|---|---|---|
| > 20 | 0.816 | 0.459 | 0.218 | 0.931 | 1.507 | 0.402 | 3.75 | 1.579–8.907 |
| > 40 | 0.737 | 0.649 | 0.28 | 0.93 | 2.098 | 0.406 | 5.172 | 2.378–11.248 |
| > 45.85 | 0.737 | 0.702 | 0.315 | 0.935 | 2.476 | 0.375 | 6.61 | 3.025–14.443 |
| > 60 | 0.553 | 0.78 | 0.318 | 0.904 | 2.518 | 0.573 | 4.392 | 2.138–9.023 |
| > 80 | 0.421 | 0.878 | 0.39 | 0.891 | 3.453 | 0.659 | 5.236 | 2.430–11.286 |
| > 100 | 0.342 | 0.932 | 0.481 | 0.884 | 5.009 | 0.706 | 7.094 | 2.995–16.804 |
Pathogenic results of pus culture
Pus culture was performed in 230 cases, of which 115 cases (50.00%) had positive results and 115 cases (50.00%) had negative results. Among the 115 pus culture-positive cases, 3 were infected with multiple pathogens, leading to a total of 119 isolated pathogens (Table 6).
Table 6.
Microbiological and pathogen distribution of 115 cases of pus culture-positive OAI
| Classification | Category | Pathogen | n (%) |
|---|---|---|---|
| Gram-positive | Staphylococcus spp. | MSSA | 54 (45.38%) |
| MRSA | 19 (15.97%) | ||
| Staphylococcus epidermidis | 1 (0.84%) | ||
| Staphylococcus haemolyticus | 1 (0.84%) | ||
| Streptococcus spp. | Streptococcus pneumoniae | 9 (7.56%) | |
| Streptococcus pyogenes | 4 (3.36%) | ||
| Streptococcus agalactiae | 2 (1.68%) | ||
| Streptococcus intermedius | 1 (0.84%) | ||
| Bacillus spp. | Bacillus pumilus | 1 (0.84%) | |
| Gram-negative | Pseudomonas spp. | Pseudomonas aeruginosa | 4 (3.36%) |
| Enterobacteriaceae | Escherichia coli | 3 (2.53%) | |
| Serratia marcescens | 2 (1.68%) | ||
| Klebsiella pneumoniae | 1 (0.84%) | ||
| Escherichia hermannii | 1 (0.84%) | ||
| Salmonella spp. | Salmonella typhimurium | 2 (1.68%) | |
| Salmonella enteritidis | 2 (1.68%) | ||
| Salmonella typhi | 1 (0.84%) | ||
| Salmonella Group B | 1 (0.84%) | ||
| Salmonella Group C1 | 1 (0.84%) | ||
| Salmonella Group C2 | 1 (0.84%) | ||
| Salmonella Group E1 | 1 (0.84%) | ||
| Haemophilus spp. | Haemophilus influenzae | 2 (1.68%) | |
| Stenotrophomonas spp. | Stenotrophomonas maltophilia | 1 (0.84%) | |
| Other | Chryseobacterium indologenes | 1 (0.84%) | |
| Fusobacterium nucleatum | 1 (0.84%) | ||
| Fungi | Candida spp. | Candida albicans | 1 (0.84%) |
| Candida parapsilosis | 1 (0.84%) |
Gram-positive bacteria were predominant among the isolated pathogens, accounting for 92 cases (77.31% of the total). The most frequently identified species was S. aureus, which comprised 54 MSSA strains (45.38%) and 19 MRSA strains (15.97%). In addition, single isolates of Staphylococcus epidermidis and Staphylococcus haemolyticus (each contributing to 0.84%) were also identified. Common streptococcal species included S. pneumoniae (9 cases, 7.56%), S. pyogenes (4 cases, 3.36%), Streptococcus agalactiae (2 cases, 1.68%), and Streptococcus intermedius (1 case, 0.84%). Among Bacillus species, Bacillus pumilus was identified in one case (0.84%).
Gram-negative bacteria were detected in 25 cases (21.01%), with a relatively discrete distribution. The most frequently detected species were P. aeruginosa (4 cases, 3.36%) and Escherichia coli (3 cases, 2.53%). Additionally, Serratia marcescens, Salmonella typhimurium, Salmonella enteritidis, and H. influenzae were isolated in 2 cases each (1.68%). Other Gram-negative organisms, which were detected in one case each (contributing to 0.84% each), included Klebsiella pneumoniae; Escherichia hermannii; Salmonella typhi; Salmonella Group B, Group C1, Group C2, and Group E1; Stenotrophomonas maltophilia, Chryseobacterium indologenes, and Fusobacterium nucleatum.
Finally, fungi were isolated in 2 cases (1.68%), which included one case each (0.84%) of Candida albicans and Candida parapsilosis.
Comparison of clinical characteristics between negative and positive pus culture groups
The demographic data, clinical characteristics, and laboratory indicators of the pus culture-negative and pus culture-positive groups are presented in Table 7.
Table 7.
Comparison of clinical characteristics between children with negative and positive pus culture results
| Clinical characteristics | Negative cultures (n = 115) | Positive cultures (n = 115) | P |
|---|---|---|---|
| Clinical | |||
| Age (years) | 3.66 ± 3.66 | 4.49 ± 4.05 | 0.105 |
| Male, n (%) | 74 (64.35) | 66 (57.39) | 0.280 |
| Onset time | |||
| 0–2 weeks, n (%) | 80 (69.57) | 85 (73.91) | 0.464 |
| 2 weeks–3 months, n (%) | 35 (30.43) | 30 (26.09) | 0.464 |
| Infection type | |||
| SA, n (%) | 67 (58.26) | 37 (32.17) | < 0.001 |
| OM, n (%) | 17 (14.78) | 35 (30.43) | 0.005 |
| CSAO, n (%) | 31 (26.96) | 43 (37.39) | 0.090 |
| Top 5 Affected sites | |||
| Knee, n (%) | 53 (41.41%) | 32 (27.59%) | 0.004 |
| Hip, n (%) | 36 (28.13%) | 30 (25.86%) | 0.382 |
| Femur, n (%) | 27 (21.09%) | 27 (23.28%) | 1.000 |
| Tibia, n (%) | 9 (7.03%) | 14 (12.06%) | 0.272 |
| Humerus, n (%) | 3 (2.34%) | 13 (11.21%) | 0.010 |
| Laboratory | |||
| WBC (× 10⁹/L) | 12.19 ± 4.42 | 14.61 ± 7.36 | 0.030 |
| Neutrophil percentage (%) | 50.70 (38.93–65.63) | 63.00 (47.30–72.30) | < 0.001 |
| C-reactive protein (CRP, mg/L) | 33.07 ± 33.27 | 55.28 ± 46.48 | < 0.001 |
| Erythrocyte sedimentation rate (ESR, mm/h) | 56.34 ± 30.67 | 62.43 ± 33.99 | 0.155 |
| Other | |||
| Length of hospital stay (days) | 27.07 ± 15.95 | 29.41 ± 15.90 | 0.267 |
| Hospital cost (CNY) | 21221.05 ± 9223.37 | 25544.37 ± 9223.37 | 0.049 |
The mean age of the positive pus culture group was 4.49 ± 4.05 years, which was slightly higher than that of the negative group (3.66 ± 3.66 years); however, the difference was not statistically significant (P > 0.05). In terms of sex distribution, males comprised 64.35% (74/115) of the positive group and 57.39% (66/115) of the negative group, with no significant difference observed between the two groups (P > 0.05).
With regard to the duration of onset, 73.91% (85/115) of children in the culture-positive group had an onset time of less than two weeks, compared to 69.57% (80/115) in the culture-negative group. Conversely, the proportion of children with an onset time between two weeks and three months was 26.09% (30/115) and 30.43% (35/115) in the positive and negative groups, respectively. These differences were not statistically significant (P > 0.05).
The incidence of SA was significantly higher in the culture-negative group (58.26%, 67/115) than in the culture-positive group (32.17%, 37/115; P < 0.001). In contrast, OM was more common in the culture-positive group (30.43%, 35/115) than in the negative group (14.78%, 17/115; P = 0.005), and CSAO was also slightly more prevalent in the positive group (37.39%, 43/115) than in the negative group (26.96%, 31/115). No significant inter-group differences were observed in the incidence of OM or CSAO (P > 0.05).
Based on the top five most frequently affected sites, 244 sites were included from both the positive and negative groups. In the positive group, the knee joint was the most commonly affected (27.59%, 32/116), followed by hip joint (25.86%, 30/116), femur (23.28%, 27/116), tibia (12.06%, 14/116), and humerus (11.21%, 13/116). In the negative group, too, knee joint was the most commonly affected (41.41%, 53/128), followed by hip joint (28.13%, 36/128), femur (21.09%, 27/128), tibia (7.03%, 9/128), and humerus (2.34%, 3/128). Statistically significant differences were observed between the two groups in terms of the involvement of the knee joint (P = 0.004) and the humerus (P = 0.010), with knee joint involvement being significantly higher in the pus culture-negative group and humerus involvement being significantly higher in the pus culture-positive group.
WBC count, neutrophil percentage, CRP, and ESR in the pus culture-positive group were higher than those in the negative group: WBC, 14.61 ± 7.36 × 109/L versus 12.19 ± 4.42 × 109/L; neutrophil percentage, 63.00% (47.30–72.30%) versus 50.70% (38.93–65.63%); CRP, 55.28 ± 46.48 mg/L versus 33.07 ± 33.27 mg/L; ESR, 62.43 ± 33.99 mm/h versus 56.34 ± 30.67 mm/h. The differences between the two groups in WBC (P = 0.030), neutrophil percentage (P < 0.001), and CRP (P < 0.001) were statistically significant.
The average hospitalization time of the pus-positive group was 29.41 ± 15.90 days, which was slightly longer than that of the negative group (27.07 ± 15.95 days), but the difference was not statistically significant (P > 0.05). The average hospitalization cost of the positive group (25544.37 ± 9223.37 RMB) was significantly higher than that of the negative group (21221.05 ± 9223.37 RMB) (P = 0.049).
Identification of independent predictors of pus culture results
To identify the independent risk factors for positive pus culture, a multivariate logistic regression model with conditional forward selection was used to analyze the variables with a P value less than 0.2 in the univariate analysis, namely age, presence of SA, presence of OM, knee joint involvement, humeral involvement, WBC count, neutrophil ratio, CRP, and ESR [30] (Table 8).
Table 8.
Logistic regression analysis for predicting pus culture results
| Variables | Positive cultures | Negative cultures | Standard error | χ2 | P | ||||
|---|---|---|---|---|---|---|---|---|---|
| β | Odds ratio | 95% CI for odds ratio | β | Odds ratio | 95% CI for odds ratio | ||||
| Septic arthritis | −1.349 | 0.259 | 0.141–0.477 | 1.349 | 3.859 | 2.088–7.092 | 0.313 | 18.602 | < 0.001 |
| Neutrophil ratio | 0.029 | 1.029 | 1.011–1.048 | −0.029 | 0.972 | 0.954–0.989 | 0.009 | 9.074 | 0.003 |
| CRP | 0.012 | 1.012 | 1.004–1.020 | −0.012 | 0.988 | 0.980–0.996 | 0.004 | 7.601 | 0.006 |
| Intercept | −1.461 | 0.232 | – | 1.461 | 4.312 | – | 0.51 | 8.211 | 0.004 |
The results showed that presence of SA, neutrophil percentage, and CRP levels were retained in the final logistic regression model for predicting positive pus culture. SA had a regression coefficient of −1.349 (OR: 0.259, 95% CI 0.141–0.477, P < 0.001), which indicates that children with SA had a significantly lower likelihood of positive pus culture, having only 25.9% of the likelihood observed in children with other types of osteoarticular infections. The regression coefficient for neutrophil percentage was 0.029 (OR: 1.029, 95% CI 1.011–1.048, P = 0.003), which suggests that each 1% increase in neutrophil proportion was associated with an approximately 2.9% increase in the likelihood of a positive culture. CRP had a regression coefficient of 0.012 (OR: 1.012, 95% CI 1.004–1.020, P = 0.006): this implies that for every 1 mg/L increase in CRP, the likelihood of a positive pus culture increased by approximately 1.2%.
Conversely, in the logistic regression model using negative pus culture as the dependent variable, presence of SA was associated with a regression coefficient of 1.349 and an OR of 3.859 (95% CI 2.088–7.092, P < 0.001). These values indicate that children with SA have a significantly increased likelihood of negative culture results—approximately 3.86 times higher than those with other types of osteoarticular infections. The regression coefficient for neutrophil percentage was –0.029 (OR: 0.972, 95% CI 0.954–0.989, P = 0.003), indicating that each 1% decrease in neutrophil proportion was associated with a 2.9% increase in the likelihood of a negative culture. Similarly, CRP exhibited a regression coefficient of −0.012 and an OR of 0.988 (95% CI 0.980–0.996, P = 0.006), which indicates that each 1 mg/L decrease in CRP was linked to an approximately 1.2% higher likelihood of a negative pus culture.
Construction of a prediction model for negative pus culture results
Based on multivariable logistic regression analysis, presence of SA, neutrophil percentage, and CRP levels were identified as independent predictors of pus culture-negative OAI. A nomogram prediction model for pus cultures was subsequently developed, incorporating these variables (Fig. 2). The neutrophil percentage ranged from 0 to 100%, and CRP levels ranged from 0 to 200 mg/L. The total point score of the nomogram ranged from 0 to 260, corresponding to a predicted probability of culture negativity between 0.1 and 0.9.
Fig. 2.
Nomogram of the clinical prediction model for pus culture-negative OAI
The nomogram demonstrated good discriminative ability. ROC curve analysis yielded an AUC of 0.750 (95% CI 0.687–0.813) (Fig. 3), indicating that the model performed well in distinguishing between pus culture-negative and pus culture-positive OAI. The optimal cut-off value was 0.599, with a sensitivity of 56.5% and a specificity of 85.2%. The concordance index (C index) of the model was 0.750. Internal validation using 1000 bootstrap resamples produced a corrected C index of 0.742 (95% CI 0.713–0.771), further supporting the model’s robust discriminative performance.
Fig. 3.
ROC curve of the clinical prediction model for pus culture-negative OAI in 230 children
Model calibration was evaluated using the Hosmer–Lemeshow test and a bootstrap-corrected calibration curve. The Hosmer–Lemeshow test yielded a χ2 value of 5.843 with a P value of 0.665, indicating good agreement between the predicted probabilities and observed outcomes. The calibration curve generated from 1000 bootstrap resamples (Fig. 4) demonstrated close concordance between predicted and actual values, with a mean absolute error of 0.016, further supporting the model’s excellent calibration performance.
Fig. 4.
Calibration curve of the clinical prediction model for pus culture-negative OAI
DCA was performed to assess the clinical net benefit of the model across a range of risk thresholds (Fig. 5). The results demonstrated that the model consistently provided a positive net benefit within the 0–80% threshold range, outperforming both the “all” and “none” strategies. These findings suggest that the model holds substantial clinical utility for early diagnosis and decision-making in children with culture-negative OAI, while striking a balance between diagnostic benefit and the avoidance of unnecessary interventions.
Fig. 5.
DCA of the nomogram for pus culture-negative OAI
Based on the clinical prediction model we developed, we designed a convenient online tool to predict the likelihood of negative pus culture results in children with OAI (https://oaiculturerisk.shinyapps.io/DynNomapp/). This tool can be used to calculate the likelihood of negative culture results in children with OAI; moreover, it provides an intuitive and easily interpretable way to display the prediction results. Using this tool, doctors can screen cases of OAI with the likelihood of negative culture results early on and consider (and apply) molecular diagnostic techniques (such as PCR or mNGS) in such cases to increase the detection rate of pathogens and improve the efficiency of diagnosis and treatment (Supplemental Fig. 1).
Discussion
Blood and pus culture are still considered standard methods for etiological investigations on OAI in children. We explored the clinical characteristics and risk factors of culture-negative OAI and constructed a prediction model. In the blood culture group, CRP was identified as an independent predictor of negative blood culture. ROC curve analysis based on CRP demonstrated a sensitivity of 73.7% and a specificity of 70.2% for predicting culture results. In the pus culture group, presence of SA, neutrophil percentage, and CRP level were identified as independent predictors of pus culture outcomes. The nomogram constructed using these predictors demonstrated good discriminative ability and calibration, as well as exhibited consistent clinical net benefit in decision curve analysis. These findings offer valuable guidance for the early identification of culture-negative cases in clinical practice.
For children with a high likelihood of negative blood or pus cultures, repeated conventional cultures may offer limited additional diagnostic value and may inadvertently prolong dependence on empirical antibiotic therapy [20]. In such settings, early consideration of molecular diagnostic techniques, such as PCR or mNGS, may improve pathogen detection, shorten the time to etiological diagnosis, and facilitate more targeted antimicrobial treatment [19]. Accordingly, the proposed models may serve as a useful adjunct to routine microbiological testing, facilitating more efficient and individualized clinical management of pediatric OAI.
Pathogen culture results in children with OAI
This study summarized 345 cases of childhood OAI over the last 10 years and found that the positive rate of blood culture was 15.64% (38 cases/243 cases) and the positive rate of pus culture was 50.00% (115 cases/230 cases). While other rare pathogens were detected, S. aureus was still the main causative pathogen of OAI infection in children, accounting for 68.42% (26/38) of the pathogens detected in blood culture and 61.35% (73/119) of the pathogens in pus culture.
The microbiological culture positivity rates observed in the present study were generally consistent with those reported by Ulziibat who documented blood culture positivity rates ranging from 9.9% to 12.7% and pus culture positivity rates ranging from 39.4% to 53.5% in pediatric osteoarticular infections [15]. The relatively low culture positivity may be partially attributable to fastidious pathogens, such as Brucella spp. or Mycobacterium tuberculosis, which are not routinely detected by conventional culture methods. To address this limitation, Feng et al. [31] demonstrated that an enhanced diagnostic protocol incorporating routine aerobic and anaerobic culture, Gram stain microscopy, acid-fast bacilli staining, and blood culture bottle enrichment significantly improved pathogen detection in pus specimens, achieving an overall positivity rate of 81.9%, representing a 20.1 percentage point increase compared with conventional culture methods. Future investigations in pediatric osteoarticular infections may benefit from adopting similar multimodal diagnostic strategies to enhance microbiological yield.
In terms of pathogen distribution, too, S. aureus was the most commonly identified organism in several regional studies, with detection rates as high as 60%–83% [15, 32–35], which are in agreement with the rates observed in our result (68.42% in blood cultures and 61.35% in pus cultures). Although Kingella kingae has been increasingly recognized as a significant pathogen in OAI [36–38], it was not detected in our results. This may reflect regional differences in pathogen epidemiology and could also be attributed to the fastidious growth requirements of this organism and the absence of species-specific PCR testing. Future studies incorporating molecular diagnostic techniques are warranted to better assess the prevalence of K. kingae in Chinese children with OAI.
Risk factors associated with a negative blood culture
Among the 243 cases with blood culture results, univariate analysis revealed that WBC, neutrophil percentage, CRP, and ESR were all significantly lower in the culture-negative group than in the culture-positive group (P < 0.05). Additionally, a higher proportion of children in the culture-negative group presented with a disease duration of 2 weeks–3 months. Multivariate logistic regression analysis identified CRP as an independent risk factor for negative blood culture results (OR = −0.021, 95% CI 0.972–0.987, P < 0.001), and ROC curve analysis further demonstrated that CRP had moderate discriminatory ability in predicting blood culture positivity, with an AUC of 0.747. The optimal cut-off value was 45.850 mg/L, yielding a sensitivity of 73.7% and a specificity of 70.2%. Similarly, Kheir et al. [39] reported that patients with culture-negative periprosthetic joint infections had significantly lower WBC and CRP levels than those with culture-positive infections. This finding is in alignment with our results and further supports the utility of inflammatory markers in predicting culture results.
Although the sensitivity and specificity of CRP for predicting negative blood culture were relatively modest in our findings, the primary aim of this study was not to position CRP as a standalone or definitive diagnostic test. Instead, it sought to explore whether CRP could contribute to identifying children with a higher probability of negative culture results. From this perspective, even a moderate discriminative performance may have clinical relevance, as it could assist clinicians in anticipating situations in which conventional blood cultures are less likely to yield informative results. Accordingly, CRP should be viewed as a complementary marker to inform subsequent diagnostic strategies—such as the earlier and more selective use of molecular diagnostic techniques—rather than as an independent determinant of clinical decision-making.
Risk factors associated with a negative pus culture
Among the 230 cases with pus culture results, univariate analysis indicated that the presence of SA, WBC, neutrophil percentage, and CRP levels was significantly associated with negative culture outcomes. Multivariate logistic regression further identified SA (OR = 3.859, 95% CI 2.088–7.092, P < 0.001), neutrophil percentage (OR = 0.972, 95% CI 0.954–0.989, P = 0.003), and CRP (OR = 0.988, 95% CI 0.980–0.996, P = 0.006) as independent predictors of negative pus culture. A nomogram prediction model was constructed based on these three variables, and it demonstrated good discriminative ability, with an AUC of 0.750 (95% CI 0.687–0.813). At the optimal cut-off of 0.599, its sensitivity and specificity were 56.5% and 85.2%, respectively. Calibration was assessed using the Hosmer–Lemeshow test (χ2 = 5.843, P = 0.665) and a bootstrap-corrected calibration curve based on 1,000 resamples, which yielded a mean absolute error of 0.016. Both tests indicated good agreement between the predicted and observed outcomes. Further, DCA showed that the model provided a net clinical benefit across a wide range of threshold probabilities (0%–80%), outperforming both the “all” and “none” strategies. These findings suggest that the model may have meaningful clinical utility in the early identification and management of culture-negative OAI in children.
Limitations
This study has several limitations. First, this was a single-center retrospective study, and a substantial proportion of children with OAI had received oral or intravenous antibiotics prior to hospital admission. Although the overall culture positivity rate was comparable to that reported in previous studies [15, 40, 41], the potential impact of prior antibiotic exposure on culture results cannot be completely excluded.
With regard to pus cultures, several prior studies have consistently shown that preoperative or pre-sampling antibiotic administration does not significantly diminish culture positivity in osteoarticular or other deep tissue infections [11, 42, 43]. Accordingly, prior antibiotic exposure is unlikely to have materially influenced the predictive model derived from pus culture results in the present study. In contrast, existing evidence indicates that antecedent antibiotic use can substantially reduce the yield of blood cultures. For instance, Hirosawa et al. reported that prior antibiotic exposure was significantly more common among patients with negative than positive blood cultures (26.0% vs. 13.5%), implying that blood culture positivity may be underestimated by approximately 10% in routine clinical practice [44]. On this basis, the blood culture positivity rate observed in our cohort is likely lower than the true rate of pathogen detection. Importantly, such potential misclassification of blood culture results is expected to be non-differential with respect to most baseline clinical characteristics and laboratory parameters. As a result, the observed predictive ability of CRP for negative blood culture outcomes is more likely to be attenuated rather than exaggerated. As such, the true discriminative performance of CRP may be modestly stronger than that observed in the present analysis.
Second, molecular diagnostic techniques, such as PCR or mNGS, which are known to enhance the sensitivity of pathogen detection [45–47], were not incorporated in this study. A final limitation was that the sensitivity of the model was relatively low (56.5%), which may limit its applicability in clinical scenarios where minimizing missed diagnoses is critical. Future studies should aim to validate the model’s performance and generalizability through prospective, multicenter designs incorporating both conventional cultures and molecular diagnostics.
Conclusion
In summary, this study conducted a comprehensive analysis of risk factors associated with both blood and pus culture results in childhood OAI and developed a predictive tool for determining the likelihood of negative pus culture results. For blood cultures, CRP was identified as a key predictive marker, demonstrating high sensitivity and specificity. For pus cultures, a nomogram incorporating SA, neutrophil percentage, and CRP was constructed, which showed good discriminatory performance and clinical applicability. These findings offer valuable guidance for the early identification of culture-negative cases in clinical practice. In children with a high likelihood of negative culture results, molecular diagnostic techniques such as PCR or mNGS may be considered to enhance pathogen detection rates and, thereby, improve the efficiency of diagnosis and treatment.
Supplementary Information
Supplementary Material 1. 1a Step 1 in using the online risk prediction tool for culture-negative OAI: Select the type of osteoarticular infection. 1b Step 2: Slide to select the values for neutrophil percentage and CRP. 1c Step 3: Click “Predict” to estimate the probability of a negative pus culture result. The predicted probability will then be displayed on the interface.
Acknowledgements
We would like to thank all individuals and institutions who contributed to this work.
Abbreviations
- OAI
Pediatric osteoarticular infections
- ROC
Receiver operating characteristic curve
- DCA
Decision curve analysis
- CRP
C-reactive protein
- SA
Septic arthritis
- OM
Osteomyelitis
- CSAO
Concomitant septic arthritis and osteomyelitis
- PCR
Polymerase chain reaction
- mNGS
Metagenomics next-generation sequencing
- WBC
White blood cell count
- ESR
Erythrocyte sedimentation rate
- AUC
Area under the curve
Author contributions
Jianlin Chen and Ruitao Lu were responsible for drafting the manuscript and conducting data analysis. Haoran Feng provided additional contributions to data analysis. Zilong Huang, Zhongjian Xie, Huifeng Yang, Jiahui Li, and Gen Liu were involved in data acquisition and contributed to the data analysis. Xin Qiu conceptualized and designed the overall project framework. Gianfilippo Caggiari, Guibing Fu, and Hansheng Deng oversaw project coordination and supervision. All authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
Funding
This study was supported by Shenzhen Clinical Research Center (20220819113341005), Science and Technology Innovation Commission of Shenzhen (KJZD20230923114002005), Longhua District Science, Technology and Innovation Bureau (10162A20221027B1FA526), Guangdong High-level Hospital Construction Fund and National Natural Science Fund of China (12274197), as well as Shenzhen Fund for Guangdong Provincial High-level Clinical Key specialties (No. SZXK035).
Data availability
The datasets generated or analyzed during the study are available from the corresponding author on reasonable request.
Declarations
Ethics approval and consent to participate
This study was approved by the Ethics Committee of Shenzhen Children's Hospital (approval number: 202405802). Given that this study was a retrospective clinical data analysis and all research methods complied with the guidelines of the Ethics Committee, the Ethics Committee agreed to waive the informed consent of the parents/guardians of underage patients. All clinical data were de-identified and anonymized prior to analysis. Identifiable personal information was removed or coded to ensure patient confidentiality, and no individual could be directly or indirectly identified from the dataset. The anonymized data were stored on secure, password-protected institutional servers with access restricted to authorized study personnel only. All study procedures were conducted in accordance with the Declaration of Helsinki and applicable institutional data protection policies.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Jianlin Chen, Ruitao Lu and Haoran Feng have contributed equally to this work.
Contributor Information
Gianfilippo Caggiari, Email: gcaggiari@uniss.it.
Guibing Fu, Email: fgbmd@163.com.
Hansheng Deng, Email: hanshengdeng@126.com.
Xin Qiu, Email: qiuxinfrank@163.com.
References
- 1.Saad L, Hupin M, Buteau C, et al. Late sequelae of osteoarticular infections in pediatric patients: a single-center study. Medicine (Baltimore). 2021;100(8):e23765. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Tobudic S, Forstner C, Burgmann H, et al. Real-world experience with dalbavancin therapy in gram-positive skin and soft tissue infection, bone and joint infection. Infection. 2019;47(6):1013–20. [DOI] [PubMed] [Google Scholar]
- 3.Alvares PA, Mimica MJ. Osteoarticular infections in pediatrics. J Pediatr (Rio J). 2020;96:58–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Zhang T, Yu S, Lv X, et al. Paediatric osteomyelitis and septic arthritis pathogen distribution and antimicrobial resistance in a single centre: a 15-year retrospective analysis. J Trop Pediatr. 2022. 10.1093/tropej/fmac038. [DOI] [PubMed] [Google Scholar]
- 5.De Marco G, Cochard B, Di Laura Frattura G, et al. Reflection on osteoarticular infections in children. Front Pediatr. 2023;11:1280878. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Dartnell J, Ramachandran M, Katchburian M. Haematogenous acute and subacute paediatric osteomyelitis: a systematic review of the literature. J Bone Joint Surg Br. 2012;94(5):584–95. [DOI] [PubMed] [Google Scholar]
- 7.Benvenuti M, An T, Amaro E, et al. Double-edged sword: musculoskeletal infection provoked acute phase response in children. Orthop Clin North Am. 2017;48(2):181–97. [DOI] [PubMed] [Google Scholar]
- 8.Ilharreborde B. Sequelae of pediatric osteoarticular infection. Orthop Traumatol Surg Res. 2015;101:S129–37. [DOI] [PubMed] [Google Scholar]
- 9.Pendleton A, Kocher MS. Methicillin-resistant staphylococcus aureus bone and joint infections in children. J Am Acad Orthop Surg. 2015;23(1):29–37. [DOI] [PubMed] [Google Scholar]
- 10.Bar-On E, Weigl DM, Bor N, et al. Chronic osteomyelitis in children: treatment by intramedullary reaming and antibiotic-impregnated cement rods. J Pediatr Orthop. 2010;30(5):508–13. [DOI] [PubMed] [Google Scholar]
- 11.Section J, Gibbons SD, Barton T, et al. Microbiological culture methods for pediatric musculoskeletal infection: a guideline for optimal use. J Bone Joint Surg. 2015;97(6):441–9. [DOI] [PubMed] [Google Scholar]
- 12.Chang WS, Chiu NC, Chi H, et al. Comparison of the characteristics of culture-negative versus culture-positive septic arthritis in children. J Microbiol Immunol Infect. 2005;38(3):189–93. [PubMed] [Google Scholar]
- 13.Chen X, Zhuang T, Zou C, et al. Analysis of pathogen distribution and antimicrobial resistance in bone and joint infections among young children. Clin Lab. 2024. 10.7754/Clin.Lab.2024.240333. [DOI] [PubMed] [Google Scholar]
- 14.Olijve L, Amarasena L, Best E, et al. The role of Kingella kingae in pre-school aged children with bone and joint infections. J Infect. 2021;83(3):321–31. [DOI] [PubMed] [Google Scholar]
- 15.Ulziibat M, Buettcher M, Altankhuyag U, et al. Microbiological spectrum of osteoarticular infections and their management in Mongolian children. Glob Pediatr Health. 2024;11:2333794x241298801. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Spyridakis E, Gerber JS, Schriver E, et al. Clinical features and outcomes of children with culture-negative septic arthritis. J Pediatr Infect Dis Soc. 2019;8(3):228–34. [DOI] [PubMed] [Google Scholar]
- 17.Calvo C, NúñEZ E, Camacho M, et al. Epidemiology and management of acute, uncomplicated septic arthritis and osteomyelitis: Spanish multicenter study. Pediatr Infect Dis J. 2016;35(12):1288–93. [DOI] [PubMed] [Google Scholar]
- 18.Samara E, Spyropoulou V, Tabard-Fougère A, et al. Kingellakingae and osteoarticular infections. Pediatrics. 2019;144(6):e20191509. [DOI] [PubMed] [Google Scholar]
- 19.Ramchandar N, Burns J, Coufal NG, et al. Use of metagenomic next-generation sequencing to identify pathogens in pediatric osteoarticular infections. Open Forum Infect Dis. 2021;8(7):ofab346. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Woods CR, Bradley JS, Chatterjee A, et al. Clinical practice guideline by the pediatric infectious diseases society and the infectious diseases society of America: 2021 guideline on diagnosis and management of acute hematogenous osteomyelitis in pediatrics. J Pediatr Infect Dis Soc. 2021;10(8):801–44. [DOI] [PubMed] [Google Scholar]
- 21.Rosey AL, Abachin E, Quesnes G, et al. Development of a broad-range 16S rDNA real-time PCR for the diagnosis of septic arthritis in children. J Microbiol Methods. 2007;68(1):88–93. [DOI] [PubMed] [Google Scholar]
- 22.Watanabe S, Kobayashi N, Tomoyama A, et al. Clinical characteristics and risk factors for culture-negative periprosthetic joint infections. J Orthop Surg Res. 2021;16(1):292. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Wang J, Wang Q, Shen H, et al. Comparable outcome of culture-negative and culture-positive periprosthetic hip joint infection for patients undergoing two-stage revision. Int Orthop. 2018;42(3):469–77. [DOI] [PubMed] [Google Scholar]
- 24.Lee HT, Sebro R. Predictors of positive bone cultures from CT-guided bone biopsies performed for suspected osteomyelitis. J Med Imaging Radiat Oncol. 2020;64(3):313–8. [DOI] [PubMed] [Google Scholar]
- 25.Jaybhaye A, Lg S, Dash N, et al. Clinical spectrum and microbial etiology of bone and joint infections in children: a retrospective analysis from South India. Am J Trop Med Hyg. 2023;108(5):936–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.El-Sobky T, Mahmoud S. Acute osteoarticular infections in children are frequently forgotten multidiscipline emergencies: beyond the technical skills. EFORT Open Rev. 2021;6(7):584–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Arnold JC, Bradley JS. Osteoarticular infections in children. Infect Dis Clin North Am. 2015;29(3):557–74. [DOI] [PubMed] [Google Scholar]
- 28.Pääkkönen M, Kallio MJ, Kallio PE, et al. C-reactive protein versus erythrocyte sedimentation rate, white blood cell count and alkaline phosphatase in diagnosing bacteraemia in bone and joint infections. J Paediatr Child Health. 2013;49(3):E189-92. [DOI] [PubMed] [Google Scholar]
- 29.Lorrot M, Fitoussi F, Faye A, et al. Laboratory studies in pediatric bone and joint infections. Arch Pediatr. 2007;14(Suppl 2):S86-90. [DOI] [PubMed] [Google Scholar]
- 30.Feng L, Shao L, Sun S, et al. Analysis of the efficacy and influencing factors of preoperative P-SOX neoadjuvant chemotherapy regimen for progressive gastric cancer-construction of a clinical prediction model. Cancer Med. 2023;12(12):13031–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Feng Q, Ha X, Guo W, et al. Optimization of pathogen detection in abscess specimens: a 6-year retrospective study. Eur J Med Res. 2025;30(1):619. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Chen WL, Chang WN, Chen YS, et al. Acute community-acquired osteoarticular infections in children: high incidence of concomitant bone and joint involvement. J Microbiol Immunol Infect. 2010;43(4):332–8. [DOI] [PubMed] [Google Scholar]
- 33.Yamagishi Y, Togawa M, Shiomi M. Septic arthritis and acute hematogenous osteomyelitis in childhood at a tertiary hospital in Japan. Pediatr Int. 2009;51(3):371–6. [DOI] [PubMed] [Google Scholar]
- 34.Chen X, Zhuang T, Zou C, et al. Analysis of antimicrobial resistance and clinical features of Staphylococcus aureus-infected bone and joint infections in children. BMC Pediatr. 2025;25(1):83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Trobisch A, Schweintzger NA, Kohlfürst DS, et al. Osteoarticular infections in pediatric hospitals in Europe: a prospective cohort study from the EUCLIDS consortium. Front pediatr. 2022;10:744182. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Dodwell ER. Osteomyelitis and septic arthritis in children: current concepts. Curr Opin Pediatr. 2013;25(1):58–63. [DOI] [PubMed] [Google Scholar]
- 37.Juchler C, Spyropoulou V, Wagner N, et al. The contemporary bacteriologic epidemiology of osteoarticular infections in children in Switzerland. J Pediatr. 2018;194:190-6.e1. [DOI] [PubMed] [Google Scholar]
- 38.Coulin B, Demarco G, Vazquez O, et al. Osteoarticular infections in children: accurately distinguishing between MSSA and Kingella kingae. Microorganisms. 2022. 10.3390/microorganisms11010011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Kheir MM, Tan TL, Shohat N, et al. Routine diagnostic tests for periprosthetic joint infection demonstrate a high false-negative rate and are influenced by the infecting organism. J Bone Joint Surg. 2018;100(23):2057–65. [DOI] [PubMed] [Google Scholar]
- 40.Ferroni A, Al Khoury H, Dana C, et al. Prospective survey of acute osteoarticular infections in a French paediatric orthopedic surgery unit. Clin Microbiol Infect. 2013;19(9):822–8. [DOI] [PubMed] [Google Scholar]
- 41.Moumile K, Merckx J, Glorion C, et al. Bacterial aetiology of acute osteoarticular infections in children. Acta Paediatr. 2005;94(4):419–22. [DOI] [PubMed] [Google Scholar]
- 42.van der Merwe M, Rooks K, Crawford H, et al. The effect of antibiotic timing on culture yield in paediatric osteoarticular infection. J Child Orthop. 2019;13(1):114–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Zhorne DJ, Altobelli ME, Cruz AT. Impact of antibiotic pretreatment on bone biopsy yield for children with acute hematogenous osteomyelitis. Hosp Pediatr. 2015;5(6):337–41. [DOI] [PubMed] [Google Scholar]
- 44.Hirosawa T, Sakamoto T, Hanai S, et al. Effect of prior antibiotic treatment on blood culture in an outpatient department of general internal medicine: a retrospective case-control analysis. Int J Gen Med. 2023;16:2709–17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Zhang Q, Wang Q, Zhang F, et al. Diagnostic value of metagenomic next-generation sequencing in patients with osteoarticular infections: a prospective study. Microbiol Spectr. 2025;13(5):e0106424. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Zhang K, Bai YZ, Liu C, et al. Composition of pathogenic microorganism in chronic osteomyelitis based on metagenomic sequencing and its application value in etiological diagnosis. BMC Microbiol. 2023;23(1):313. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.AGUDELO-PéREZ S, FERNáNDEZ-SARMIENTO J, RIVERA LEóN D, et al. Metagenomics by next-generation sequencing (mNGS) in the etiological characterization of neonatal and pediatric sepsis: a systematic review. Front Pediatr. 2023;11:1011723. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Material 1. 1a Step 1 in using the online risk prediction tool for culture-negative OAI: Select the type of osteoarticular infection. 1b Step 2: Slide to select the values for neutrophil percentage and CRP. 1c Step 3: Click “Predict” to estimate the probability of a negative pus culture result. The predicted probability will then be displayed on the interface.
Data Availability Statement
The datasets generated or analyzed during the study are available from the corresponding author on reasonable request.