141 S/F94 as a proxy for COVID-19 severity 7 Introduction Therapeutic research in COVID-19 depends on efficient, accurate assessment of therapeutic candidates in early-stage clinical studies. Efficacy measures should be “clinically meaningful”1 endpoints such as the WHO ordinal scale2. Intermediate endpoints for early phase trials, or severity measures for observational studies, must be modifiable by therapy and ideally should have a continuous numerical distribution to improve statistical power3. The endpoint should accurately predict the definitive outcome of interest and ideally should also be closely related to the causal pathway to this outcome. In COVID-19, efficacy measures such as the WHO ordinal scale, duration of hospitalisation, and viral load have been used widely5. Both the WHO ordinal scale and various alternative ordinal scales,7 rely on a complex clinical measure - the level of respiratory support received by a patient - as an indicator of illness severity. Viral load is a valid outcome for antiviral therapy, but it has not been shown to correlate with mortality benefit, and is not directly relevant to the effect of anti-inflammatory treatments10. In the RECOVERY trial, we identified a need for more powerful intermediate endpoints for early phase clinical trials. Impairment of the pulmonary oxygenation function indicates disease progression in COVID-19,11 and is strongly predictive of mortality12. Importantly, in COVID-19, failure of pulmonary oxygenation is likely to be mechanistically linked to death: patients at extreme risk of mortality12 have high survival rates if oxygenation is provided by extracorporeal membrane oxygenation (ECMO)13. Pulmonary oxygenation function, together with clinical decision-making and resource availability, determines movement between most of the stages of the WHO Ordinal Scale (WHO scale points 4-9)2. Oxygenation function is a key determinant of efficacy for immunosuppression with corticosteroids in COVID-199. It is likely that pulmonary oxygenation function lies on the causal pathway between the SARS-CoV-2 infection and death for many hospitalised patients. Peripheral oxygen saturation can be measured easily and non-invasively using a pulse oximeter (formally, arterial oxygen saturation measured by pulse oximetry, rather than direct measurement in blood, is SpO2). The ratio of SaO2 or SpO2 to inspired fraction of oxygen (FIO2), known as the S/F ratio, provides a continuous index of pulmonary oxygenation function which can be calculated without an arterial blood sample. S/F correlates well with the most widely-used arterial blood-derived measure of oxygenation - P/F ratio (PaO2/FIO2) 14. S/F under steady state conditions in humans can range from around 0.5 (severe oxygenation defect) to 4.8 (perfect oxygenation function). A major limitation of S/F is the ceiling effect: at high SaO2 values, SaO2 ceases to be dependent on pulmonary oxygenation function, because the blood is close to maximally oxygenated and the relationship between the P/F ratio and the S/F ratio is non-linear16. For example, a healthy patient with perfect lungs breathing 21% oxygen with SaO2=0.99 would have S/F=4.7, but the same patient breathing 100%
RkJQdWJsaXNoZXIy MTk4NDMw