The Association Between Serum Inflammatory Markers and Pathogen Profile in Confirmed Fracture-related Infection.

Abstract

Objectives: To determine whether serum inflammatory markers, Erythrocyte Sedimentation Rate (ESR) and C-reactive protein (CRP), varied according to pathogen profile in patients with confirmed fracture-related infection (FRI).

Methods: Design: Retrospective Cohort Study.

Setting: Single, multi-site urban academic institution.

Patient selection criteria: Patients treated between January 2011 and July 2024 were included if they met FRI Consensus Group criteria for confirmed FRI, underwent surgical irrigation and debridement with intraoperative deep tissue sampling, and had ESR (mm/hr) or CRP (mg/L) values documented within 7 days of operative debridement. Cases were classified into four mutually exclusive pathogen groups: gram-positive (GP) monomicrobial, gram-negative (GN) monomicrobial, polymicrobial (PM), and culture-negative (CN).

Outcome measures and comparisons: ESR (reference range: 0-22 mm/hr) and CRP (reference range: <5.1 mg/L) were compared across pathogen groups, with direct comparison of gram-positive versus gram-negative infections. Receiver operating characteristic (ROC) analysis evaluated each marker's discriminatory ability.

Results: Among 359 patients meeting inclusion criteria, 331 had ESR and 343 had CRP values available. Gram-positive monomicrobial infections were identified in 143 (39.8%), patients (median [IQR] age: 49.0 [35-64], 71.3% male), gram-negative in 69 (19.2%) patients (median [IQR] age: 54.5 [39-65], 44.9% male), polymicrobial in 114 (31.8%) patients (median [IQR] age: 46.8 [37-65], 64.0% male) and culture-negative in 33 (9.2%) patients (median [IQR] age: 65.0 [39-69], 30.3% male). Baseline demographics were similar across groups (median age: 46.8-65.0, median BMI 26.4-27.5, median CCI 0-0; all p>0.05), except for sex distribution, with male sex most common in gram-positive infections (71.3%, p<0.001). ESR did not differ significantly across pathogen groups (median 53.0 GP, 57.0 GN, 45.0 PM, 48.5 CN mm/hr, p=0.728). CRP differed significantly across groups (median 17.4 GP, 7.0 GN, 14.0 PM, 10.8 CN mg/L, p=0.013); post hoc analysis identified gram-positive versus gram-negative as the only significant pairwise difference, with CRP approximately 2.5-fold greater in gram-positive than gram-negative infections (median 17.4 vs 7.0 mg/L, adjusted p=0.012). ROC analysis demonstrated modest ability of CRP to discriminate gram-positive from gram-negative FRIs (AUC=0.634, sensitivity 72.5%, specificity 49.2%), while ESR demonstrated no meaningful discrimination (AUC=0.482).

Conclusions: ESR did not differ by pathogen profile in confirmed FRI. CRP was significantly higher in gram-positive infections, with a median value approximately 2.5-fold greater than gram-negative infections. However, the modest AUC and poor specificity indicate CRP cannot reliably classify pathogen type at the individual patient level. Among patients with a diagnosis of confirmed FRI, relative CRP elevation may raise clinical suspicion for gram-positive etiology.

Level of evidence: III.