Predictive Value of White Blood Cell Counts for Infection in Patients With Diabetic Ketoacidosis and Rational Use of...

Nonspecific clinical presentations, absence of timely available paraclinical indicators of infection, and increasing concerns for antimicrobial resistance complicate decision-making in initiation of antibiotic therapy in patients with diabetic ketoacidosis (DKA). The authors have aimed to explore the role of white blood cell (WBC) counts as an indicator of infectious disease in patients with DKA.

A retrospective observational study was conducted using data from patients hospitalized with DKA in a tertiary care center in Iran. Patient data was retrieved from the hospital information system (HIS). Associations between leukocytosis and leukopenia along with several paraclinical tests (including ESR and CRP levels) and a final diagnosis of infectious disease were examined.

Data from 129 patients were analyzed. Patients diagnosed with infection were significantly older (p < 0.001) and spent more time in-hospital (p = 0.008). Mean WBC counts, ESR and CRP were elevated in patients with and without infectious disease. There was a significant association between leukocytosis and infection (p < 0.001). ESR (p = 0.01) and CRP (p = 0.002) were also significantly higher in patients with an infection. Leukocytosis had a sensitivity of 82.3%, and a specificity of 49.2% (AUC = 0.665, 95% CI 0.571-0.759) for infection detection. Excluding cases with leukopenia, leukocytosis had a negative predictive value of 81.8%-91.4% for predicting infectious disease.

We discourage the use of leukocytosis, per se, as a marker of infection in the setting of DKA. We found the absence of leukocytosis or leukopenia of high predictive value in exclusion of the possibility of infectious disease in DKA cases.