Background
Randomized-controlled trials have shown that procalcitonin-guided algorithms can reduce antibiotic duration for lower respiratory tract infections (LRTIs). The goal of this study was to compare antibiotic duration for LRTIs with and without procalcitonin testing in real-life practice.
Methods
This retrospective cohort study included all acute-care hospital admissions for presumed LRTIs between 1/2018-12/2021 at 81 Veterans Affairs facilities with on-site procalcitonin testing. The exposure was procalcitonin testing; the primary outcome was antibiotic duration. We used 1:1 nearest-neighbor propensity score matching to estimate the difference in outcome between procalcitonin-tested and non-tested patients.
Results
35,610 patients with LRTIs were included (6,015 [16.9%] with procalcitonin testing; 29,595 [83.1%] without testing). In tested patients, the median number of procalcitonin levels checked was 2 (interquartile range, 1-3). Mean antibiotic duration was 10.0 days in the procalcitonin group compared to 8.3 days in non-tested patients (unadjusted difference 1.7 days; p < 0.0001). After propensity-score matching with 3,903 pairs, antibiotic duration remained greater in the procalcitonin group (9.6 days vs. 9.2 days; p < 0.0001). In a sub-group analysis of 2,241 tested-patients with a procalcitonin value at the standard threshold for antibiotic discontinuation, antibiotic duration was shorter in tested vs. non-tested patients with a mean difference of 0.1 days (p<0.01).
Conclusions
In this retrospective propensity-matched cohort of patients with presumed LRTIs across a geographically-diverse group of hospitals, patients who underwent procalcitonin testing did not have a meaningful reduction in antibiotic duration compared to those who were not tested. Poor implementation of procalcitonin testing may have undermined its effectiveness.