Laboratory Markers Associated with Deep Neck Infection in Emergency Department Patients with Throat or Neck Symptoms.

Background/

Objectives

Deep neck infection (DNI) can be difficult to distinguish from localized upper aerodigestive disease during emergency department (ED) evaluation. We assessed routinely available laboratory markers in adults presenting with throat or neck symptoms and examined whether the associations were reproduced in an imaging-defined subgroup.

Methods

This single-center retrospective cohort included 1640 adults with prespecified throat- or neck-related presentations between January 2021 and May 2026. The primary outcome was DNI confirmed using prespecified diagnostic evidence from contrast-enhanced neck CT, endoscopic/fibroscopic evaluation, or ultrasonographic evaluation with aspiration or drainage when applicable. A clinical model incorporating demographics, presenting symptoms, symptom duration, and initial vital signs was compared with the same model after addition of inflammatory laboratory markers. A secondary analysis was restricted to 662 patients who underwent contrast-enhanced neck CT.

Results

DNI was confirmed in 273 patients (16.6%). In 1166 complete cases (268 events), the 10-fold cross-validated AUC increased from 0.644 for the clinical model to 0.788 for the clinical model augmented with ANC and CRP (ΔAUC, 0.144; 95% CI, 0.109-0.179; likelihood-ratio p < 0.001). In this augmented model, ANC (aOR per 5 × 109/L, 1.49; 95% CI, 1.25-1.79) and CRP (aOR per two-fold increase, 1.51; 95% CI, 1.37-1.65) remained associated with DNI. In the CT-defined subgroup, the corresponding cross-validated AUC increased from 0.562 to 0.685 (ΔAUC, 0.123; 95% CI, 0.077-0.167). Derived inflammatory indices showed no clear advantage over conventional markers.

Conclusions

ANC and CRP remained associated with confirmed DNI after adjustment for clinical characteristics, and their addition improved discrimination beyond structured clinical assessment. More complex derived indices offered no clear advantage over conventional markers. Model performance remained insufficient for stand-alone diagnostic or imaging decisions.