Use of carbon dioxide production to detect bacterial superinfections in mechanically ventilated patients with acute respiratory distress syndrome: an exploratory prospective cohort study.

Background: Bacterial superinfections are common in patients with acute respiratory distress syndrome (ARDS) but diagnosing them is challenging. Exhaled carbon dioxide (V'CO2) may be increased during bacterial infection, suggesting a potential marker for detecting bacterial superinfections in ARDS patients.

Methods: In a prospective cohort study of mechanically ventilated adult patients with ARDS due to SARS-CoV-2 in a tertiary intensive care unit, we assessed V'CO2 measurements from continuous volumetric capnography and calculated daily median V'CO2 levels. The primary outcome was to determine if a first substantial increase in daily median V'CO2 was associated with a first bacterial superinfection. Protocolised microbiological sampling and adjudicated clinical interpretations were used to determine the onset of a first superinfection.

Results: A total of 150 days of continuous volumetric capnography were analysed in 31 mechanically ventilated adult patients with ARDS due to SARS-CoV-2. We observed 10 patients (32%) with a first episode of substantial increase of daily median V'CO2, and 12 (39%) patients with a first bacterial superinfection. A V'CO2 increase was not associated with a superinfection on the same day (OR 3.47, 95% CI 0.64 to 18.92, p=0.15, adjusted for age and gender). Investigating all 150 test days of median V'CO2 revealed a poor sensitivity (17%, 95% CI 2% to 48%) for detecting superinfections. However, a first V'CO2 increase indicated superinfection with high specificity (94%, 95% CI 89% to 98%). Patients with superinfections showed higher daily median V'CO2 levels (210 mL/min) than those without (176 mL/min, p<0.001), even after adjusting for age and gender (OR 1.56, 95% CI 1.16 to 2.08, p=0.003).

Conclusions: A sudden increase in daily median V'CO2 did not reliably detect bacterial superinfections, which was reflected in a poor sensitivity and inability to rule out superinfections in patients without V'CO2 increase. Nevertheless, high specificity suggests that V'CO2 may be useful to rule in superinfections in patients with ARDS.

Trial Registration Number: NCT04410263.