Likelihood-Ratio-Test Methods for Drug Safety Signal Detection from Multiple Clinical Datasets

Table 6

Type-I error (or FDR) for cases with data generated under null hypothesis (all relative risks to be 1) and power (%) for cases with data generated under alternative hypothesis (varying relative risks).

Cases	Description	LRT methods
		Pooled	wLRn	MMLR
Type-I error
sim0allocc	With and from real data	0.055	0.047	0.050
	Allocc case in Table 2, rr1
sim01occ	With and from real data	0.073	0.047	0.049
	1st occurrence case in Table 2, rr1
sim0LIP	With and from real Lipiodol	0.046	0.045	0.038
	Data, 13 studies, 27 rows, rr1
Power
Simaallocc	rr2	86.7	67.2	48.9
Simaallocc	rr3	99.3	98.4	92.8
Simaallocc	rr4	100	100	99.9
sima1occ	rr2	74.3	37.2	30.1
sima1occ	rr3	97.5	77.0	66.5
sima1occ	rr4	100	98.2	96.9
simaLIP	rr2	38.3	23.4	11.1
simaLIP	rr3	95.0	86.3	63.0
simaLIP	rr4	100	100	99.3
Simaallocc	rr21	32.9	100	99.9
sima1occ	rr21	16.7	11.5	8.0

Note. rr1, rr2, rr3, and rr4 are for cases having the 1st row with relative risk () to be 1, 1.5, 2, and 3, respectively (all other rows have relative risk as 1). In rr1, rr2, rr3, and rr4 cases, is the same for all 6 studies. In rr21 case, for 1st row in studies 1 to 6, respectively. For case sim01occ, ) are the counts of the 1st AEOSET events for PLandPPI group from studies 1 to 6, respectively. ) are the exposure for PLandPPI group from studies 1 to 6, respectively. ) are the counts of 1st AEOSET events for PL group from studies 1 to 6, respectively. ) are the exposure for PL group from studies 1 to 6, respectively. The and for case sim0allocc can also be found in Table 2 (for repeated occurrences of AEOST). For case sim0LIP, and and ) cannot be listed here due to space limitation. The values of range from 0 to 128, and the values of range from 11 to 257.