Fig. 2: Evaluation of S/F₉₄ in observational data.

a Smoothed distributions of \(S/{F}_{94}\) values in survivors and non-survivors during the first \(12\) days of the study, not including assigned minimum/maximum values (restricted to \(39\),\(765\) patients aged \(20-75\), oxygen therapy within \(3\) days). b Histogram showing distribution of \(S/{F}_{94}\) values on day 5 as used for subsequent analyses (in purple). Patients discharged home before day 5 are assigned the maximum value (4.78), and patients who died before day 5 are assigned to an arbitrary minimum of 0.5 (in black). c Distribution of \(S/{F}_{94}\) values day \(5\) compared with WHO ordinal scale² value at the same time point, in patients who met our inclusion criteria (aged \(20-75\), oxygen therapy within \(3\) days). No assigned minimum or maximum values are included in this figure. Hosp = hospitalised, no oxygen support; Ox = Hospitalised, oxygen by mask or nasal prongs; CPAP/HFNO = Hospitalised, oxygen by continuous positive airway pressure; high-flow nasal oxygen or non-invasive ventilation; IMV = Intubation and mechanical ventilation; IMV \(S/F\le 2=\) Mechanical ventilation; \(S/F\le 2\) or vasopressors; MOF = Multi-organ failure & mechanical ventilation & \(S/F\le 2\) & ECMO or renal replacement therapy. d Logistic regression analysis with \(95\%\) confidence interval, using both \(S/{F}_{94}\) on day \(0\) and \(S/{F}_{94}\) on day \(5\) as covariates, showing a clear association between mortality at \(28\) days and \(S/{F}_{94}\) value on day \(5\).