Search

ICU patients with SARS-CoV-2-related pneumonia are at risk to develop a central dysautonomia which can contribute to mortality and respiratory failure. The pupillary size and its reactivity to light are controlled by the autonomic nervous system. Pupillometry parameters (PP) allow to predict outcomes in various acute brain injuries. We aim at assessing the most predictive PP of in-hospital mortality and the need for invasive mechanical ventilation (IV). We led a prospective, two centers, observational study. We recruited adult patients admitted to ICU for a severe SARS-CoV-2 related pneumonia between April and August 2020. The pupillometry was performed at admission including the measurement of baseline pupillary diameter (PD), PD variations (PDV), pupillary constriction velocity (PCV) and latency (PDL). Fifty patients, 90 % males, aged 66 (60–70) years were included. Seven (14 %) patients died in hospital. The baseline PD (4.1 mm [3.5; 4.8] vs 2.6 mm [2.4; 4.0], P = 0.009), PDV (33 % [27; 39] vs 25 % [15; 36], P = 0.03) and PCV (3.5 mm.s−1 [2.8; 4.4] vs 2.0 mm.s−1 [1.9; 3.8], P = 0.02) were significantly lower in patients who will die. A PD value <2.75 mm was the most predictive parameter of in-hospital mortality, with an AUC = 0.81, CI 95 % [0.63; 0.99]. Twenty-four (48 %) patients required IV. PD and PDV were significantly lower in patients who were intubated (3.5 mm [2.8; 4.4] vs 4.2 mm [3.9; 5.2], P = 0.03; 28 % [25; 36 %] vs 35 % [32; 40], P = 0.049, respectively). A reduced baseline PD is associated with bad outcomes in COVID-19 patients admitted in ICU. It is likely to reflect a brainstem autonomic dysfunction. [Display omitted] • Pupillary parameters PD, PDV and slope were found significantly lower in COVID-19 patients who will die. • A reduced baseline PD <2.75 mm is the most predictive parameter of in-hospital mortality. • PD and PDV were also associated with the need for invasive mechanical ventilation with greater significance for PD. • A reduced PD could result from a central autonomic impairment related to a brainstem dysfunction in severe COVID-19 patients. [ABSTRACT FROM AUTHOR]