International guidelines recommend the initiation of empirical antibiotherapy for possible associated bacterial pneumonia in COVID-19 critically ill yet further suggesting a rapid reassessment upon source documentation [1]. In this prospective cohort analysis, we investigated the respiratory co-infection rate in COVID-19 critically ill through the use of rapid molecular testing and measured its impact on antibiotic management.

This preliminary analysis was conducted over a 1-month period at the intensive care unit (ICU) of the Cliniques universitaires Saint-Luc and included all COVID-19 adult patients from whom a lower respiratory tract sample could be obtained. Specimens were conveyed to the microbiology laboratory where a FilmArray Pneumonia Panel plus test (FA-PNEU, BioFire Diagnostics, Salt Lake City, UT, USA) was performed. The FA-PNEU is an automated multiplex PCR test allowing direct detection of 15 bacteria with a semi-quantitative value, 3 atypical bacteria, 9 viruses, and 7 antimicrobial resistance genes within 1 h and 15 min [2]. FA-PNEU testing was done 24/7, and results were immediately called to the intensive care physician pursuing antimicrobial optimization.

Forty-one COVID-19 patients were admitted to ICU, and 32 could be included upon respiratory sample availability. The study population was comparable to previously described COVID-19 critically ill in terms of age, sex ratio, severity scores, comorbidities, and symptoms [3]. FA-PNEU was performed within a mean of 10 days following symptoms’ onset and a mean of 1 day following ICU admission. FA-PNEU results identified 13/32 (40.6%) patients with a bacterial co-infection as detailed in Table 1. Staphylococcus aureus, Haemophilus influenza, and Moraxella catarrhalis were the principal bacteria identified with significant genome copies. None of the 32 FA-PNEU tests identified atypical bacteria neither other respiratory viruses. Direct communication of FA-PNEU results led to speeded-up antibiotic modifications in 15/32 (46.9%) patients.

Table 1 FA-PNEU results and antibiotic management (this table should appear after the result section)
Full size table

It is a known difficulty to adjudicate on the presence of a co-infection in COVID-19 patients particularly in critically ill. Clinical presentation, inflammatory markers, and bilateral radiological infiltrates lead to misperception and cannot be used in the diagnosis of a bacterial superinfection. As a consequence, empirical antibiotherapy is quasi-systematically initiated until microbiological documentation of co-infecting pathogens. Yet, current data on co-infections is limited. With the focus on intensive care settings, a case series in February 2020 analyzing 21 COVID-19 ICU patients reported no bacterial respiratory co-infections but 3 influenza infections [4]. A similar case series investigated in March 2020 stated none of the 15 COVID-19 critically ill had a bacterial co-infection neither were they tested positive for respiratory viruses [5]. No information however was available on how patients were tested neither on treatment strategy. In our setting applying generalized molecular screening for co-infection, the rate was 40.6% and the main detected pathogens were causal agents of community-acquired pneumonia.

As rapid molecular testing was performed within the shortest possible time following ICU admission, a majority of our patients did not receive empirical antibiotherapy while awaiting FA-PNEU result. Ultimately one third of the patients remained antibiotic-free over the entire process, and 5 patients had their antibiotics stopped following a negative FA-PNEU result. These antibiotic savings are crucial for COVID-19 critically ill known to have a long ICU stay with reported nosocomial infection rates as high as 31% [6].

To conclude, bacterial documentation is essential to assess co-infection in COVID-19 critically ill. The use of molecular diagnostic tools and the initiation of narrow-spectrum antibiotics are key elements of COVID-19 antimicrobial stewardship guidelines in critically ill. Studies on larger populations and in different geographical areas should be performed to outline analogous antibiotic saving strategies.