Impact of a Multiplex PCR Assay for Rapid Diagnosis and Antibiotic Utilization in Trauma Intensive Care Unit Patients with Ventilator-Acquired Pneumonia.

Ventilator-associated pneumonia (VAP) is a frequent complication in injured patients. Multiplex polymerase chain reaction (PCR) facilitates rapid identification of many respiratory pathogens prior to formal culture results. Our objective was to evaluate the effect of multiplex PCR implementation in a trauma intensive care unit (TICU) on antibiotic utilization and de-escalation. Injured adult patients admitted to the TICU with quantitative respiratory cultures were included. Patients were dichotomized into two groups, before (PRE) or after (POST) implementation of the pneumonia (PNA) panel. The PRE cohort included all patients meeting study criteria from January to June 2021, and the POST cohort included all patients meeting study criteria from January to June 2022, Patients were excluded if there was any documented infection requiring antibiotics other than a respiratory source. During the study period, 60 patients met criteria for inclusion, 30 PRE and 30 POST. Diagnosis of VAP was confirmed in 43.3% PRE and 50% POST patients. The time to antibiotic change was substantially shorter in the POST group (23 h vs. 61 h, p < 0.001). In the POST cohort, 83% of initial antibiotic regimens were eligible for change on the basis of PNA panel. Of these, 88% were changed in a median time of 15.4 h. In all patients, total days of antibiotic therapy (DOT) were not different (9 vs. 10, p = 0.207); however, vancomycin DOT was less in the POST group (2 d vs. 3 d, p ≤ 0.001). In those patients diagnosed with VAP, the total antibiotic (10 vs. 12 d p = 0.008), vancomycin (2 vs. 3 d p = 0.003), and cefepime DOT (3 vs. 4 d 0.029) were substantially less in the POST group. Utilization of multiplex PCR in addition to bacterial culture substantially reduced time to achieve targeted antibiotic therapy in suspected pneumonia. Furthermore, it reduced the number of days of vancomycin therapy.