Direction of effect | Analyses including all trials, no. (%) n = 61 | Analyses including only trials reporting both outcomes, no. (%) n = 55* |
---|---|---|
Concordant | ||
Total no. of comparisons | 44 (72.1) | 33 (60.0) |
Benefit to both outcomes | 35 (57.4)† | 25 (45.4)‡ |
Harm to both outcomes | 9 (14.8) | 8 (14.5) |
Discordant | ||
Total no. of comparisons | 17 (27.9) | 21 (38.2) |
Benefit to mortality, harm to admission | 8 (13.1)† | 7 (12.7) |
Benefit admission, harm to mortality | 9 (14.8) | 14 (25.4) |
Comparisons with non-overlapping CIs of treatment effects | 5 (8.2)† | 5 (9.1) |
Note: CI = 95% confidence interval.
↵* In 1 comparison, the odds ratio was exactly 1; thus, concordance or discordance could not be stated.
↵† Three comparisons with statistically significant treatment effects for both outcomes.
↵‡ One comparison with statistically significant treatment effects for both outcomes.