Table 2.
Outcome measures at 180 days
| Dexamethasone 12 mg | Dexamethasone 6 mg | Adjusted risk difference or adjusted mean differences (99% CI) | P valuea | |
|---|---|---|---|---|
| Mortality | ||||
| Death by 180 days no./total no. (%) | 164/486 (33.7) | 184/477 (38.6) | − 4.3 (− 11.7 to 3) | 0.13b |
| Health-Related Quality of Lifec medians (IQRs) | ||||
| EQ-5D-5L index values | 0.80 (0–0.97) | 0.68 (0–0.95) | 0.06 (− 0.01 to 0.12) | 0.10d |
| Survivors onlye | 0.93 (0.81–1) | 0.92 (0.77–1) | 0.02 (− 0.02 to 0.07) | 0.39 |
| EQ VAS | 65 (0–90) | 55 (0–85) | 4 (− 3 to 10) | 0.22f |
| Survivors onlye | 80 (65–95) | 80 (65–90) | 0 (− 4 to 4) | 0.49 |
CI confidence interval; EQ-5D-5L EuroQol-5 domains-5 levels; IQR interquartile range; VAS visual analogue scale
aThe analyses were adjusted for the stratification variables being trial site, age below 70 years and the use of invasive mechanical ventilation at baseline
bP = 0.12 by Fisher’s exact test
cPatients who died within 180 days after randomisation were assigned the value zero corresponding to a health state as bad as being dead for EQ-5D-5L index values and the worst possible value for EQ VAS. Data from non-responders were multiply imputed (n = 60 for the index values and n = 58 for EQ VAS scores); all health-related quality of life data in this table were calculated using the multiply imputed datasets. The data for each domain and the analyses using the Danish value set only and those of the adjusted median differences, the best–worst and worst–best imputed dataset and the complete case dataset (including the descriptive data) are presented in Fig. 3 and in Tables 1 and 2 in ESM 3. Higher values indicate better quality of life
dP = 0.04 based on the analyses of bootstrapped adjusted mean differences because of markedly skewed data
ePost hoc analyses as these were not predefined in the statistical analysis plan. Descriptive data and adjusted analyses in survivors only were calculated using the multiply imputed datasets to match the primary analyses
fP = 0.14 based on the analyses of bootstrapped adjusted mean differences because of markedly skewed data