Clinical features, procalcitonin concentration, and bacterial infection in febrile hospitalized cancer patients: a descriptive study and association analysis.

Abstract

Background: Fever in cancer patients can occur for reasons other than bacterial infections. Without practical tools to distinguish actual infections, treatment delays may occur, reducing effectiveness and increasing antibiotic resistance. This study aimed to identify clinical features and procalcitonin (PCT) levels as indicators of bacterial infection in fevers among cancer patients.

Methods: This retrospective study enrolled 225 patients with cancer and fever at the Maharat Nakhon Ratchasima Hospital. Data on the clinical characteristics, laboratory results, and bacterial cultures were collected. Associations were analyzed using logistic regression, and the appropriate PCT cutoff point was determined using ROC analysis.

Results: Of 225 cancer patients with fever, 54 (24%) had positive bacterial cultures, with Klebsiella pneumoniae being the most common pathogen. Significant clinical features included age (OR 1.06, 95% CI 1.01-1.12), increased heart rate (OR 1.05, 95% CI 1.02-1.08), and localizing symptoms (OR 7.62, 95% CI 2.49-22.70). Key laboratory findings were absolute neutrophil count (OR 1.15, 95% CI 1.03-1.28) and PCT level (OR 1.39, 95% CI 1.07-1.80). Appropriate PCT cutoff points for predicting bacterial infection were analyzed using various methods, resulting in values of 1.045, 0.546, 0.546, and 0.4025 ng/ml. The concordance probability and closest to the point (0,1) methods suggested a rounded cutoff point of 0.5 ng/ml, which provided a sensitivity of 61% and a specificity of 78%. The AUC for PCT was 0.731, indicating moderate accuracy.

Conclusion: Procalcitonin, in conjunction with clinical features, may be used to classify the cause of fever in cancer patients. Therefore, a clinically predictive model would be useful.