Table 3.

Analysis of the WHO OSCI during the dosing period

	Intention-to-treat population		Per-protocol population
	Ratio (95% CI)	p value	Ratio (95% CI)	p value
Odds of improvement on the OSCI	OR 2·32 (1·07–5·04)	0·033	OR 2·80 (1·21–6·52)	0·017
Time to severe disease or death (OSCI ≥5)	HR 0·50 (0·18–1·38)	0·179	Not calculated	..
Odds of severe disease or death (OSCI ≥5)	OR 0·28 (0·07–1·08)	0·064*	OR 0·18 (0·04–0·93)	0·041*
Time to intubation or death (OSCI ≥6)	HR 0·38 (0·09–1·65)	0·198	Not calculated	..
Odds of intubation or death (OSCI ≥6)	OR 0·42 (0·09–1·83)	0·246*	OR 0·31 (0·05–1·79)	0·189*
Time to recovery	HR 2·19 (1·03–4·69)	0·043	HR 2·29 (1·07–4·91)	0·033
Odds of recovery	OR 3·19 (1·24–8·24)	0·017	OR 3·18 (1·21–8·39)	0·019
Time to hospital discharge	HR 1·37 (0·85–2·20)	0·196	HR 1·53 (0·96–2·42)	0·072
Odds of hospital discharge	OR 1·63 (0·61–4·35)	0·330	OR 2·14 (0·64–7·12)	0·215

ORs relate to the end-of-treatment visit on day 15 or 16. Time-to-event analyses include all data up to and including the end-of-treatment visit. Recovery was defined as a post-baseline OSCI score of 0 or 1, which does not rise above 1 at any subsequent visits. Hospital discharge was defined as a post-baseline OSCI score of 2 or less, which does not rise above 2 at any subsequent visits. Three patients died during the study; all deaths occurred in patients randomly assigned to placebo, so no modelling analysis was done. OSCI=Ordinal Scale for Clinical Improvement. OR=odds ratio. HR=hazard ratio.

^*Post-hoc analysis done by use of Firth logistic regression analysis.