Table 3.

Estimated unadjusted treatment effects (odds ratio (OR) and p value) on 14 day mortality with different approaches taking into account between-centre differences

Approach	Model	OR unadjusted	P value tx effect
-Uniform treatment effect over centres -No adjustment for between-centre differences	Fixed effect logistic regression	1.22	0.00010
-Uniform treatment effect over centres -Adjustment for between-centre differences	Random effect logistic regression with random intercept	1.24	0.00003
-Varying treatment effect over centres -Adjustment for between-centre differences	Random effect logistic regression with random slope	1.22 (95% range: 1.17-1.26)	0.00029