. 2022 Feb 27;22:54. doi: 10.1186/s12874-022-01513-z

Table 6.

HIGH clinical trial: Detection of treatmen-by-subset interaction when K=2

	Interaction method	$\hat{θ}$ global CI95%	$\hat{θ_{A}}$ CI95%	$\hat{θ_{B}}$ CI95%	Decision	n_A	n_B	Probability of interaction effect*
Age Partition			Age≥65	Age<65
1er interim analysis	Millen		1.228[0.721-1.988]	0.913[0.548-1.412]	Go with entire population	87	107	0.56
	Gail and Simon		1.228[0.721-1.988]	0.913[0.548-1.412]	Go with entire population	87	107	0.02& 0.09
2nd interim analysis	Millen		1.367[0.931-1.957]	0.868[0.586-1.241]	Go with entire population	178	210	0.80
	Gail and Simon		1.367[0.931-1.957]	0.868[0.586-1.241]	Go with entire population	178	210	0.02& 0.08
3th interim analysis	Millen		1.184[0.879-1.568]	0.745[0.529-1.011]	Go with entire population	265	320	0.85
	Gail and Simon		1.184[0.879-1.568]	0.745[0.529-1.011]	Go with entire population	265	320	0.01& 0.00
Final analysis	Millen	0.989[0.819-1.182]	1.245[0.956-1.597]	0.780[0.580-1.017]	Go with entire population	360	416	0.9
	Gail and Simon	0.989[0.819-1.182]	1.245[0.956-1.597]	0.780[0.580-1.017]	Go with entire population	360	416	0.01& 0.00
SOFA neurological Partition			SOFA neuro≥1	SOFA neuro<1
1er interim analysis	Millen		0.799[0.391-1.349]	1.136[0.750-1.655]	Go with entire population	27	167	0.66
	Gail and Simon		0.799[0.391-1.349]	1.136[0.750-1.655]	Go with entire population	27	167	0.02& 0.02
2nd interim analysis	Millen		0.633[0.341-1.026]	1.252[0.916-1.677]	Enrichment in subset A	54	334	0.94
	Gail and Simon		0.633[0.341-1.026]	1.252[0.916-1.677]	Enrichment in subset A	54	334	0.08& 0.00
3th interim analysis	Millen		0.753[0.477-1.107]	1.252[0.916-1.679]	Enrichment in subset A	77	334	–
	Gail and Simon		0.753[0.477-1.107]	1.252[0.916-1.679]	Enrichment in subset A	77	334	–
Final analysis	Millen	1.156[0.890-1.477]	0.858[0.350-1.688]	1.252[0.916-1.677]	Enrichment in subset A	102	334	–
	Gail and Simon	1.156[0.890-1.477]	0.858[0.350-1.688]	1.252[0.916-1.677]	Enrichment in subset A	102	334	–
*PaO₂/FiO₂Partition*			*PaO₂/FiO*₂<100	*PaO₂/FiO*₂≥100
1er interim analysis	Millen		1.380[0.852-2.131]	0.848[0.498-1.342]	Go with entire population	80	114	0.78
	Gail and Simon		1.380[0.852-2.131]	0.848[0.498-1.342]	Go with entire population	80	114	0.07& 0.28
2nd interim analysis	Millen		1.164[0.790-1.659]	1.050[0.717-1.475]	Go with entire population	152	236	0.30
	Gail and Simon		1.164[0.790-1.659]	1.050[0.717-1.475]	Go with entire population	152	236	0.00& 0.03
3th interim analysis	Millen		0.973[0.694-1.319]	0.957[0.711-1.254]	Go with entire population	219	366	0.16
	Gail and Simon		0.973[0.694-1.319]	0.957[0.711-1.254]	Go with entire population	219	366	0.00& 0.00
Final analysis	Millen	0.998[0.824-1.196]	0.983[0.731-1.292]	1.017[0.809-1.263]	Go with entire population	278	498	0.09
	Gail and Simon	0.998[0.824-1.196]	0.983[0.731-1.292]	1.017[0.809-1.263]	Go with entire population	278	498	0.00& 0.00

The reported intervals are 95% credibility intervals, defined as [quantile(2.5%), quantile(97.5%)] of the posterior distribution.

* In case of Millen’s criterion, this refers to the posterior probability that P_2K (equation (2)). In case of Gail & Simon’s criterion, it refers to the posterior probabilities P_quali and P_quanti, respectively, as described in equations (5) & (6).