Evaluation of point-of-care multiplex polymerase chain reaction in guiding antibiotic treatment of patients acutely admitted with suspected community-acquired pneumonia in Denmark: A multicentre randomised controlled trial

Abstract

Background Rapid and accurate detection of pathogens is needed in community-acquired pneumonia (CAP) to enable appropriate antibiotics and to slow the development of antibiotic resistance. We aimed to compare the effect of point-of-care (POC) polymerase chain reaction (PCR) detection of respiratory pathogens added to standard care with standard care only (SCO) on antibiotic prescriptions after acute hospital admission. Methods and findings We performed a superiority, parallel-group, open-label, multicentre, randomised controlled trial (RCT) in 3 Danish medical emergency departments (EDs) from March 2021 to February 2022. Adults acutely admitted with suspected CAP during the daytime on weekdays were included and randomly assigned (1:1) to POC-PCR (The Biofire FilmArray Pneumonia Panel plus added to standard care) or SCO (routine culture and, if requested by the attending physician, target-specific PCR) analysis of respiratory samples. We randomly assigned 294 patients with successfully collected samples (tracheal secretion 78.4% or expectorated sputum 21.6%) to POC-PCR (n = 148, 50.4%) or SCO (146, 49.6%). Patients and investigators owning the data were blinded to the allocation and test results. Outcome adjudicators and clinical staff at the ED were not blinded to allocation and test results but were together with the statistician, blinded to data management and analysis. Laboratory staff performing standard care analyses was blinded to allocation. The study coordinator was not blinded. Intention-to-treat and per protocol analysis were performed using logistic regression with Huber–White clustered standard errors for the prescription of antibiotic treatment. Loss to follow-up comprises 3 patients in the POC-PCR (2%) and none in the SCO group. Intention-to-treat analysis showed no difference in the primary outcome of prescriptions of no or narrow-spectrum antibiotics at 4 h after admission for the POC-PCR (n = 91, 62.8%) odds ratio (OR) 1.13; (95% confidence interval (CI) [0.96, 1.34] p = 0.134) and SCO (n = 87, 59.6%). Secondary outcomes showed that prescriptions were significantly more targeted at 4-h OR 5.68; (95% CI [2.49, 12.94] p < 0.001) and 48-h OR 4.20; (95% CI [1.87, 9.40] p < 0.001) and more adequate at 48-h OR 2.11; (95% CI [1.23, 3.61] p = 0.006) and on day 5 in the POC-PCR group OR 1.40; (95% CI [1.18, 1.66] p < 0.001). There was no difference between the groups in relation to intensive care unit (ICU) admissions OR 0.54; (95% CI [0.10, 2.91] p = 0.475), readmission within 30 days OR 0.90; (95% CI [0.43, 1.86] p = 0.787), length of stay (LOS) IRR 0.82; (95% CI [0.63, 1.07] p = 0.164), 30 days mortality OR 1.24; (95% CI [0.32, 4.82] p = 0.749), and in-hospital mortality OR 0.98; (95% CI [0.19, 5.06] p = 0.986). Conclusions In a setting with an already restrictive use of antibiotics, adding POC-PCR to the diagnostic setup did not increase the number of patients treated with narrow-spectrum or without antibiotics. POC-PCR may result in a more targeted and adequate use of antibiotics. A significant study limitation was the concurrent Coronavirus Disease 2019 (COVID-19) pandemic resulting in an unusually low transmission of respiratory virus. Trial registration ClinicalTrials.gov (NCT04651712).