TABLE 6.

Credibility metrics from literature (used in Q-Finder) computed on Virtual Twins results.

Subgroup Ranking a	Subgroup description	Coverage Discovery/Test	Adjusted odds-ratios for differential treatment effect (IC95%) Discovery b	p-value for differential treatment effect Discovery	Adjusted odds-ratios for differential treatment effect (IC95%) Test b	p-value for differential treatment effect Test	Adjusted p-value for differential treatment effect Test c
S1	# of OGLD $\ge 2$AND # of different lipid-lowering agents for dyslipidemia $\ge 1$	21%/21%	2.00 [1.0; 3.8]	3.27E-2	1.37 [0.8; 2.2]	2.21E-1	3.32E-1
S2	# of OGLD $\le 2$ AND Duration of insulin therapy (year) $\ge 3$ AND # of different lipid-lowering agents for dyslipidemia $\ge 1$	47%/44%	1.77 [1.2; 2.7]	9.01E-3	1.37 [0.9; 2.0]	8.80E-2	2.64E-1
S3	Receives oral glycaemic lowering drugs = Yes AND Total serum triglycerides (mg/dL) $\ge 169.7$ AND # of different lipid-lowering agents for dyslipidemia = 0	4%/1%	0.53 [0.2; 1.5]	2.51E-1	0.68 [0.1; 4.5]	7.03E-1	7.03E-1

^aSubgroup ranking is the same as in Table 4.

^bOdds-ratios are adjusted for confounding factors through multiple regression model

^cAdjusted p-values for multiple testing are based on a Benjamini-Hochberg procedure on the test dataset