Diagnostic utility of procalcitonin for bacterial infections in diabetic ketoacidosis.

ABSTRACT

Procalcitonin is a widely used infection biomarker; however, its utility in identifying bacterial infection in diabetic ketoacidosis (DKA) is unclear. We aimed to evaluate its diagnostic performance for detecting DKA cases triggered by bacterial infections. We reviewed 303 case records of patients aged ≥ 13 years with DKA admitted to the emergency department, PGIMER (Chandigarh), between 2017 and 2022. Baseline procalcitonin was measured by electrochemiluminescence immunoassay, and a value > 0.5 ng/mL was considered elevated. Both microbiological reference standard (MRS) and composite reference standard (CRS) were used to evaluate the diagnostic performance of procalcitonin. 151/303 (49.8%) DKA cases had infection precipitations. Bacterial infections were present in 98 patients (53 microbiologically confirmed), of which urinary tract infection (n = 42), pneumonia (n = 19), skin and soft-tissue infection (n = 13), and bacteremia (n = 11) were common. The median value of procalcitonin was higher with bacterial infections than in patients without (3.68 vs. 1.00, P-value < 0.001). An elevated procalcitonin to detect bacterial infections in DKA had sensitivity 84.69%, specificity 34.15%, positive likelihood ratio (LR +) 1.29, and negative likelihood ratio (LR -) 2.44, against CRS. Against MRS, both LR + and LR - further decreased to 1.23 and 1.81, respectively. Using the receiver-operating-characteristic curve, an optimal cut-off of procalcitonin was calculated at 1.775 ng/ml against both CRS (area under curve 0.655, sensitivity 68.37%, specificity 59.02%, LR + 1.67, LR - 1.86, Yoden's index 0.274) and MRS (area under curve 0.616, sensitivity 67.92%, specificity 59.02%, LR + 1.66, LR - 1.84, Yoden's index 0.269). Procalcitonin does not help detect bacterial infections in patients with DKA at admission.