Table 1.

Analytically weighted regression results testing the quadratic interaction effects between treatment and baseline incidence for primary outcome

	Change in incidence of primary outcome	P	Change in incidence of unexpected cardiac arrests	P	Change in incidence of unplanned ICU admissions	P	Change in incidence of unexpected deaths	P
MET versus control	-4.461	0.168	-2.630	0.090	-1.185	0.504	-1.637	0.004**
Baseline incidence of the outcome (per 1000 admissions)	-1.168	0.144	-2.345	0.053	-1.051	0.056	-2.076	<0.001**
Linear interaction effect between MET and baseline incidence of outcome	1.797	0.043*	2.114	0.088	0.964	0.147	1.658	0.006**
Baseline incidence of outcome squared	0.125	0.014*	0.262	0.226	0.063	0.090	0.256	0.036*
Quadratic interaction effect between MET and baseline incidence of outcome	-0.139	0.012*	-0.378	0.104	-0.097	0.039*	-0.322	0.021*
Constant	6.632	0.035*	3.201	0.039*	2.397	0.113	2.046	<0.001**
Observations	23		23		23		23
R-squared	0.725		0.873		0.618		0.898

* P < 0.05;

† Changes in incidence for primary outcome were calculated as events per 1000 admissions.