Abstract
BACKGROUND: A high proportion of coronavirus disease 2019 (COVID-19) survivors may develop long-term cognitive impairment. We aimed to develop a multivariate causal model exposing the links between COVID-19-associated biomarkers, illness-related variables, and their effects on cognitive performance. METHODS: In this prospective study, we assess the potential drivers for the development of cognitive impairment in patients with severe COVID-19 pneumonia aged ≥ 18 years at 6-month follow-up after hospital discharge, using the Montreal Cognitive Assessment (MoCA). Patients with pre-existing cognitive impairment were excluded. Laboratory results at hospital admission were clustered by principal component analysis (PCA) and included in a path analysis model evaluating the causal relationship between age, comorbidities, hypoxemia, invasive mechanical ventilation (IMV) requirement, in-hospital delirium, and cognitive performance. RESULTS: We studied 92 patients: 54 (58.7%) men and 38 (41.3%) women, with median age of 50 years (interquartile range 42-55), among whom 50 (54.4%) tested positive for cognitive impairment at 6-month follow-up. Path analysis revealed a direct link between the thrombo-inflammatory component of PCA (C-reactive protein, fibrinogen, and neutrophils) and hypoxemia severity at hospital admission. Our model showed that low PaO(2)/FiO(2) ratio values, unlike the thrombo-inflammatory component, had a direct effect on cognitive performance, independent from age, in-hospital delirium, and invasive mechanical ventilation. CONCLUSION: In this study, biomarkers of thrombo-inflammation in COVID-19 and low PaO(2)/FiO(2) had a negative effect on cognitive performance 6 months after hospital discharge. These results highlight the critical role of hypoxemia as a driver for impaired cognition in the mid-term.