TABLE 5

Summary of the accuracy of DDI predictions using different rifampicin models (A–G)

	Observed	Model A	Model B	Model C	Model D	Model E	Model F	Model G
		Ind_max 8,^{^a}	Ind_max 8 liver, 16 gut,	Ind_max 16	Ind_max 22.7	Ind_max 29.9	Ind_max 12	Ind_max 20
		IndC₅₀ 0.32^{^a}	IndC₅₀ 0.32	IndC₅₀ 0.32	IndC₅₀ 0.30	IndC₅₀ 0.71	IndC₅₀ 0.32	IndC₅₀ 0.32
Rifampicin, i.v. MDZ
Geometric mean fold induction	1.99	1.71	1.72	2.04	2.13	2.11	1.96	2.15
GMFE		1.21	1.21	1.16	1.18	1.18	1.16	1.20
RMSE		0.51	0.47	0.36	0.38	0.38	0.38	0.40
% Within acceptance limits^{^b}		83.3	100	100	83.3	83.3	100	83.3
Rifampicin, oral MDZ
Geometric mean fold induction	18.1	6.47	9.69	17.1	29.3	26.6	10.9	23.7
GMFE		3.26	2.21	1.70	1.96	1.85	1.99	1.75
RMSE		27.0	24.8	21.6	21.5	20.7	24.1	20.6
% Within acceptance limits^{^b}		27.3	45.5	72.7	36.4	36.4	63.6	63.6
Rifampicin, all MDZ (i.v. and oral)
Geometric mean fold induction	n/a
GMFE		2.30	1.79	1.48	1.64	1.58	1.65	1.53
RMSE		21.7	20.0	17.4	17.3	16.7	19.4	16.5
% Within acceptance limits^{^b}		47.1	58.8	82.4	52.9	52.9	76.5	70.6
Rifampicin, all victims (i.v.)
Geometric mean fold induction	n/a
GMFE		1.24	1.24	1.15	1.16	1.13	1.16	1.17
RMSE		0.53	0.56	0.35	0.36	0.35	0.42	0.37
% Within acceptance limits^{^b}		90.0	100	100	90.0	90.0	100	90.0
Rifampicin, all victims (oral)
Geometric mean fold induction	n/a
GMFE		2.81	2.12	1.69	1.91	1.80	1.96	1.72
RMSE		21.5	19.9	17.7	20.5	19.2	19.1	18.9
% Within acceptance limits^{^b}		26.3	26.3	68.4	52.6	52.6	73.7	68.4
Rifampicin, all victim drugs
Geometric mean fold induction	n/a
GMFE		2.12	1.77	1.48	1.61	1.53	1.63	1.51
RMSE		17.4	16.1	14.4	16.5	15.5	15.5	15.3
% Within acceptance limits^{^b}		48.3	51.7	79.3	65.5	65.5	79.3	75.9

GMFE, geometric mean fold error; n/a, not applicable; RMSE, root mean square error.
↵^a Default rifampicin induction parameters (V12). Geometric mean fold induction for observed data were calculated in a meta-analysis using published methodology (Einolf, 2007; Cubitt et al., 2011; Ghobadi et al., 2011; Barter et al., 2013; Supplemental Table 3).
↵^b Acceptance limits proposed by Guest et al. (2011).