Prognostic Value of the Veno-Arterial Carbon Dioxide Difference in COVID-19-Associated Acute Hypoxemic Respiratory Failure: A Retrospective Cohort Study

Background/Objectives: In acute hypoxemic respiratory failure, hyperventilation and alterations in oxygenation can influence CO2 kinetics via the Haldane effect, potentially modifying the venous-to-arterial carbon dioxide partial pressure gradient (PvaCO2) irrespective of tissue perfusion, thereby complicating its prognostic significance. This study assessed the predictive importance of PvaCO2 in critically ill patients experiencing acute respiratory failure linked to COVID-19. Methods: This retrospective bicentric study included critically ill patients with hypoxemic COVID-19 (PaO2/FiO2 < 300 mmHg) admitted to the intensive care unit (ICU) during the first and second waves of the Belgium epidemic, with at least one PvaCO2 assessment within 5 days of admission. PvaCO2 was assessed early after ICU admission using paired arterial and central venous blood gases. The highest daily value was registered. The first registered value was considered as a baseline. The primary outcome was ICU death within 28 days. Results: Among 300 ICU admissions, 136 patients met inclusion criteria, with PvaCO2 measured within 48 h in 76%. Median PaO2/FiO2 was 123 (79–176) mmHg, with no significant differences in baseline severity between patients with normal or abnormal PvaCO2. Lactate levels were generally normal, and inotrope use was uncommon. Early abnormal PvaCO2 was not significantly associated with ICU death within 28 days in the competing-risk analysis (sHR 1.69, 95% CI 0.86–3.33; p = 0.13), while complementary Bayesian analysis indicated an 89.7% posterior probability of increased hazard. When stratified by hypoxemia severity—severe (PaO2/FiO2 ≤ 100 mmHg), moderate (>100–200 mmHg), and mild (>200 mmHg)—a stronger exploratory signal related to ICU mortality was observed in moderate hypoxemia; however, the formal interaction did not reach statistical significance (Pinteraction = 0.075). Conclusions: Early PvaCO2 was not significantly associated with ICU death within 28 days in patients with predominantly hypoxemic respiratory failure without major baseline circulatory shock. Exploratory subgroup analyses suggested that its association with outcome may vary according to arterial and central venous oxygenation. This hypothesis warrants targeted evaluation in future prospective studies.