Dear editor,

Patients with COVID-19 induced respiratory failure are initially treated with lung-protective controlled mechanical ventilation to prevent ventilator-induced lung injury.[1] Adequate timing of the switch to a support ventilation mode seems critical. Prolonged sedation and neuromuscular blocking may increase ICU-acquired weakness, whereas premature transition to spontaneous breathing may result in self-inflicted lung injury (P-SILI).[2] Currently, there are no data to predict a successful switch from controlled mechanical ventilation to spontaneous breathing in COVID-19 patients.
The oxygenation index (OI) was originally developed to measure severity of illness and predict outcome in neonatal respiratory failure. Previously, OI was also found to predict mortality in adult patients with ARDS.[3] The OI has not been used before to predict the moment of weaning from mechanical ventilation. In contrast to the PaO2 /FiO2 (P/F) ratio, the OI is not only based on the fraction of inspired oxygen (FiO2 ) and partial pressure of oxygen in arterial blood (PaO2 ), but also integrates the mean airway pressure, which corresponds with static lung compliance.[4] We hypothesise that OI is a predictor for successful transition to spontaneous breathing.

Methods

This retrospective single-centre study was performed in a closed-format 29-bed mixed ICU in a tertiary teaching hospital. All patients ≥18 years admitted to the ICU from 15 March to 30 April 2020 with PCR-confirmed COVID-19, and at least one attempt to switch from controlled mechanical ventilation to spontaneous breathing, were included in the study. The study was performed in accordance with the Declaration of Helsinki. According to applicable laws, the need for individual consent was waived by a local ethics committee.

Protocol
All patients were treated in accordance with a local COVID-19 protocol, including haemodynamic monitoring with pulsecontour analysis (PiCCO®, Getinge AB, Gothenburg, Sweden). In the first week of ICU admission pressure regulated volume control was the preferred ventilator setting. Sedation with midazolam/fentanyl was titrated to a Richmond AgitationSedation Scale level -4 to -5.[5] Additional muscle paralysis was used if deemed necessary to facilitate lung-protective ventilation with tidal volumes of 6 ml/kg ideal body weight and a driving pressure 30/min, minute volume was >15 l/min, tidal volume >8 ml/kg ideal body weight, P0.1 >5 cm H2 O, presence of physical signs of respiratory distress or a decrease in P/F ratio.

Data collection
Demographic characteristics, severity of illness scores over the first 24 hours following ICU admission and the maximum SOFA score during admission were extracted from the patient data management system (Epic®, Verona, Wisconsin, USA). C-reactive protein (CRP), EVLWi, Pulmonary Vascular Permeability Index (PVPi ) and ventilation parameters, including mode of mechanical ventilation, static compliance (Cstat) and P/F ratio, were recorded once daily. OI was calculated for the day of admission and prior to the moment that the first attempt to switch to spontaneous breathing was made. The formula FiO2 (%) x mean airway pressure (mmHg)/PaO2 (mmHg) was used for calculation.

Statistical analysis
The Statistical Package for Social Sciences (SPSS 24 for Windows, Chicago, IL, USA) was used for statistical analysis. Comparison between the groups with a successful and an unsuccessful first switch to spontaneous breathing was performed with the Mann-Whitney U test. A p value <0.05 was considered statistically significant. For distinctive variables a ROC analysis was executed to test performance and to determine an optimal cut-off value.