Developing a nomogram based on admission AGR and CLR to predict subsequent surgical intervention in pediatric acute hematogenous osteomyelitis

Abstract

Abstract Pediatric acute hematogenous osteomyelitis (AHO) can progress rapidly, but admission findings associated with subsequent operative management remain incompletely defined. In this two-center retrospective study, we evaluated the albumin-to-globulin ratio (AGR) and C-reactive protein-to-lymphocyte ratio (CLR) as admission biomarkers associated with surgical intervention during the index hospitalization. Because this outcome is partly clinician-determined, the model was designed to estimate subsequent surgical management rather than the biological necessity for surgery. We included 120 children admitted between 2015 and 2024 (conservative treatment, n = 58; surgical treatment, n = 62). Clinical characteristics, imaging findings, microbiological results, and laboratory variables measured within 2 h of admission and before antibiotic initiation were reviewed. AGR and CLR were modeled as continuous predictors in Firth’s penalized logistic regression and used to construct an admission-based nomogram. Lower AGR and higher CLR were independently associated with surgical intervention (AGR: OR 0.056, 95% CI 0.012–0.220, P  < 0.001; CLR: OR 1.008, 95% CI 1.002–1.018, P  = 0.008). The model showed moderate discrimination (AUC = 0.755; optimism-corrected AUC = 0.749) and satisfactory calibration within this cohort (Hosmer–Lemeshow P  = 0.922; Brier score = 0.198), with internal validation performed using stratified tenfold cross-validation and 1000 bootstrap resamples. At the Youden-derived cutoff of 0.552, sensitivity and specificity were 0.645 and 0.759, respectively. Decision curve analysis suggested potential net benefit mainly across threshold probabilities of approximately 10–60%. These findings indicate that admission AGR and CLR are associated with later clinician-determined surgical intervention and may help identify children who require closer observation, earlier imaging prioritization, and timely reassessment. The nomogram should be considered exploratory and hypothesis-generating; independent external validation is required before routine clinical use.