Clinical outcomes
| Characteristic . | Fresher (N = 12 308) . | Older (N = 20 651) . | Effect estimate∗ (95% CI) . | P value (overall) . | P value† (trials) . |
|---|---|---|---|---|---|
| Primary outcome | |||||
| 28-d mortality, n (%) | 1446/12 236 (11.8) | 1984/20 572 (9.6) | OR, 1.06 (0.98 to 1.14)‡ | .124 | .645 |
| Secondary outcomes | |||||
| Time until death, d | 7.0 (3.0-14.0) (n = 1446) | 7.0 (2.5-14.0) (n = 1984) | MD, 0.00 (−0.92 to 0.92)§ | .999 | .285 |
| Patients discharged alive from the hospital at day 28, n (%) | 8894/12 233 (72.7) | 15 790/20 571 (76.8) | SHR, 0.99 (0.97 to 1.02)|| | .500 | .522 |
| Hospital length of stay, d | 8.0 (4.0-13.0) (n = 8894) | 7.0 (4.0-12.0) (n = 15790) | MD, 0.00 (−0.48 to 0.48)§ | .999 | .741 |
| Characteristic . | Fresher (N = 12 308) . | Older (N = 20 651) . | Effect estimate∗ (95% CI) . | P value (overall) . | P value† (trials) . |
|---|---|---|---|---|---|
| Primary outcome | |||||
| 28-d mortality, n (%) | 1446/12 236 (11.8) | 1984/20 572 (9.6) | OR, 1.06 (0.98 to 1.14)‡ | .124 | .645 |
| Secondary outcomes | |||||
| Time until death, d | 7.0 (3.0-14.0) (n = 1446) | 7.0 (2.5-14.0) (n = 1984) | MD, 0.00 (−0.92 to 0.92)§ | .999 | .285 |
| Patients discharged alive from the hospital at day 28, n (%) | 8894/12 233 (72.7) | 15 790/20 571 (76.8) | SHR, 0.99 (0.97 to 1.02)|| | .500 | .522 |
| Hospital length of stay, d | 8.0 (4.0-13.0) (n = 8894) | 7.0 (4.0-12.0) (n = 15790) | MD, 0.00 (−0.48 to 0.48)§ | .999 | .741 |
Data are median (quartile 25th-quartile 75th) or n/N (%).
ICU, intensive care unit; MD, median difference; SHR, subdistribution hazard ratio.
All models include trials as a fixed effect.
P value estimated from a fixed interaction term between treatment and trial in the model.
Effect estimate is OR; 95% CIs and P values calculated with a logistic regression model with trial as a fixed effect.
Effect estimate is median difference; 95% CI and P values calculated with a median regression model using an interior point algorithm and standard errors calculated using bootstrap with 1000 replications.
Effect estimate is SHR; 95% CI and P values calculated with a Fine-Gray competing risk model with death before discharge treated as competing event.