Methods: We conducted a single center, retrospective cohort study of all patients admitted to the PICU from January 2013 to June 2015 when PCT was available. Patients were included if they were <18 years of age, had a PCT level drawn by the managing medical team and were on <48 hours of antibiotic therapy at time of PCT. Presence of SBI was determined by expert retrospective review of the electronic record. The discriminatory ability of PCT to determine SBI was examined with AUC-ROC plots and sensitivity, specificity and predictive values. Antibiotic length of therapy, completion of antibiotic course, and economic comparisons were made between SBI and non-SBI groups.
Results: Of the 75 patients included, 28 (37%) had a SBI and 47 (63%) did not. Median PCT values were higher in the SBI group (6.48 ng/mL) than the non-SBI group (0.23 ng/mL) [p<0.0001] and PCT had a good ability to predict bacterial infection (AUC, 0.83, 95% CI: 0.74 – 0.93; p<0.0001). A PCT ≥ 1.0 ng/mL was the ideal threshold to detect SBI with a sensitivity of 82% (95% CI, 63% -94%), specificity of 75% (60% – 86%), positive predictive value of 65% (48% – 81%) and negative predictive value of 88% (73% – 96%). Antibiotic length of therapy was longer in the SBI group versus the non-SBI group (5 [3 – 8] days vs 2 [1 -6] days, p<0.01) and completion of antibiotic course was more likely in those with a SBI (96% vs 60%, p=0.001). Total costs of antibiotics and PCT testing in the non-SBI group were $3,172 and $10,046, respectively.
Conclusion: PCT was able to adequately predict bacterial infection in a pragmatic PICU setting and may be useful in minimizing antibiotic consumption in those without a SBI if available in real-time. Cost of the assay does not financially outweigh savings in antibiotic use.
S. Chen, None
N. Fusco, None
A. Hassinger, None
See more of: Poster Abstract Session