To evaluate the diagnostic utility of leukocyte count (WBC), erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP) for distinguishing between septic and aseptic nonunions.
A single-gate (cohort) design was employed using one set of eligibility criteria applied to a consecutive sample of nonunions.
Private quaternary referral center.
Inclusion criteria were consecutive patients (>18 years) with a nonunion requiring surgery that allowed for direct or medullary canal tissue sampling from the nonunion site. The cohort included 204 subjects with 211 nonunions.
Blood samples were drawn for laboratory analysis of WBC, ESR, and CRP prior to surgery.
The reference standard used to define infection was the fracture-related infection confirmatory criteria. Measures of diagnostic accuracy were calculated. In order to assess the additional diagnostic gain of each index lab test while simultaneously considering the others, logistic regression models were fit.
The prevalence of infection was 19% (40 of 211 nonunion sites). The positive likelihood ratios (95% CI) for WBC, ESR, and CRP were 1.07 (0.38-3.02), 1.27 (0.88-1.82) and 1.57 (0.94-2.60), respectively. Multivariable modeling adjusted for the effect of preoperative antibiotics showed that WBC (p = 0.42), ESR (p = 0.48), and CRP (p = 0.23) were not significant predictors of infection.
In this consecutive sample of 211 nonunions in whom standard clinical practice would be to obtain index lab tests, our findings showed that WBC, ESR, and CRP were not significant predictors of infection.
Diagnostic Level II. See Instructions for Authors for a complete description of levels of evidence.