153 S/F94 as a proxy for COVID-19 severity 7 There are multiple advantages of using S/F94 as an intermediate outcome measure in a phase II clinical trial in hospitalised patients. It is an easy, non-invasive measurement, using near-ubiquitous monitoring equipment. In contrast, daily PaO2 measurements (from an arterial blood sample) are time-consuming, require highlyskilled staff, and are burdensome for patients unless an indwelling arterial catheter is present (unusual outside of critical care areas). It is likely that the results of recent and ongoing clinical trials suggesting harm from hyperoxia will, in future, mean high SaO2 values a less common finding, particularly in the intensive care unit. In order to determine the utility of a surrogate outcome in clinical trials, a distinction can be made between “individual level surrogacy” and “trial level surrogacy”23. If there is an association between the surrogate and the outcome of interest in individual patients, the surrogate works on an individual level. If the effect that a treatment has on the surrogate can be used to predict the causal effect treatment has on the outcome, there is also trial level surrogacy. There are some scenarios, as explained by Buyse and colleagues23, in which there is individual level surrogacy but no trial level surrogacy, for example due to (known and unknown) confounders, or treatment being dependent on the surrogate (e.g. low S/F94 values could lead to additional interventions that influence the outcome, confounding the influence of treatment on outcome). Trial level surrogacy can be demonstrated with data from (multiple) randomised controlled trials. With the data we have available, we can thus only show individual level surrogacy and not trial level surrogacy. Determining whether S/F94 is also a trial level surrogate would be a desirable objective for further studies. Of the pragmatic endpoints available from routinely collected data, the WHO ordinal scale is the best-performing endpoint. In studies where clinical observations can be obtained, S/F94 is a robust measure of pulmonary oxygenation function, and is the best measure to optimise statistical power for comparisons. S/F94 is comparable to the P/F ratio as a measure of pulmonary oxygenation, and superior to SpO2/FIO2 ratio. Where protocolised measurements can be obtained, further improvements in statistical power are expected. S/F94 may have utility in clinical studies of other disease processes where pulmonary oxgenation failure contributes to mortality, such as influenza and ARDS24. In conclusion, S/F94 is a powerful and robust intermediate endpoint for clinical studies of COVID-19 and may have broad utility in forms of acute lung injury.
RkJQdWJsaXNoZXIy MTk4NDMw