Background: PCT has a high sensitivity and specificity for bacterial infections; low results allow clinicians to safely discontinue or de-escalate antibiotics. Trials demonstrate high overall concordance between PCT algorithms and antibiotic use. However, results stratified by low versus high PCT are lacking, especially among inpatients in the United States (US). This study aimed to determine the effect of low PCT results on antibiotic use at a US tertiary care center with an existing PCT guideline.
Methods: Retrospective review of all hospitalized patients (n=181) with a PCT result during a 2 month period. PCT levels <0.25 were labeled low. Antibiotic discordance was defined as antibiotic continuation without de-escalation in patients with low PCT levels or discontinuation in patients with high PCT levels. The odds of discordance based on PCT result, and based on patient- and provider-level factors within the low PCT subgroup were calculated by univariate analysis.
Results: Antibiotics were continued in 55% of patients with low PCTs and discontinued in 9% of patients with high PCTs (Figure, Pearson's chi-squared p<0.001). The odds ratio of discordant antibiotic use for low versus high PCT results was 12.5 (95% CI 5.4-28.8, p<0.001). Age, sex, suspected infection site (pneumonia vs. SIRS/sepsis), patient complexity as measured by Charlson Comorbidity Index, and disease severity measured by All Patient Refined DRGs were not statistically significant predictors of antibiotic discordance in patients with low PCTs. However, critical care was more likely than medicine services to use antibiotics discordantly (OR 4.4, 95% CI 1.4- 14.0, p= 0.011) and all patients with low PCTs on immunocompromised services received antibiotics.
Conclusion: Antibiotic discontinuation or de-escalation occurred in less than half of the patients with low PCTs, suggesting further studies should stratify based on PCT result in addition to reporting an overall concordance rate. Discordance among patients with low PCT results could not be explained by included measures of patient complexity or severity. Differences between medicine and critical care services may reflect variation in provider decision making worthy of further exploration.
L. Schulz, None
A. Lepak, None