| Summary: | Choosing optimal outcome measures maximizes statistical power, accelerates discovery and improves reliability in early-phase trials. We devised and evaluated a modification to a pragmatic measure of oxygenation function, the [Formula: see text] ratio. Because of the ceiling effect in oxyhaemoglobin saturation, [Formula: see text] ratio ceases to reflect pulmonary oxygenation function at high [Formula: see text] values. We found that the correlation of [Formula: see text] with the reference standard ([Formula: see text]/[Formula: see text] ratio) improves substantially when excluding [Formula: see text] and refer to this measure as [Formula: see text]. Using observational data from 39,765 hospitalised COVID-19 patients, we demonstrate that [Formula: see text] is predictive of mortality, and compare the sample sizes required for trials using four different outcome measures. We show that a significant difference in outcome could be detected with the smallest sample size using [Formula: see text]. We demonstrate that [Formula: see text] is an effective intermediate outcome measure in COVID-19. It is a non-invasive measurement, representative of disease severity and provides greater statistical power.
|
|---|