Table 2.

Example data in contrast-level format.

Study	Arm.1	Arm.2	Arm.3	Number of arms	lor 2	lor 3	se 2	se 3	Arm1	Arm2	Arm3	V	PLA lo
1	A	B	C	3	−1.82	−1.83	0.42	0.41	1	2	3	0.10	0.45
2	A	B		2	−2.46		0.53		1	2			1.97
3	A	B		2	−2.10		0.65		1	2			1.15
4	A	B		2	−3.58		0.69		1	2			1.39
5	A	B		2	−1.64		0.42		1	2			1.52
6	A	E		2	−2.43		0.23		1	5			1.17
7	A	E		2	−2.23		0.18		1	5			0.35
8	A	E		2	−2.65		0.62		1	5			1.75
9	B	E		2	−0.07		0.17		2	5
10	B	E		2	−1.51		0.31		2	5
11	B	E		2	−0.54		0.40		2	5
12	B	E		2	−1.14		0.47		2	5
13	B	E		2	0.31		0.34		2	5
14	A	B		2	−3.39		0.48		1	2			1.66
15	A	C		2	−0.92		0.25		1	3			−0.34
16	A	C		2	−2.38		0.40		1	3			0.58
17	A	C		2	−0.88		0.26		1	3			0.15
18	A	C		2	−2.22		0.30		1	3			1.15
19	A	C		2	−2.51		0.44		1	3			1.25
20	A	C		2	−0.70		0.34		1	3			0.82
21	A	C		2	−3.36		0.52		1	3			1.82
22	A	C		2	−1.67		0.83		1	3			2.08
23	A	B		2	−1.22		0.27		1	2			0.30
24	A	D		2	−1.92		0.18		1	4			0.75
25	C	E		2	−0.63		0.17		3	5
26	B	C		2	−0.19		0.21		2	3

“lor 2” is the column of log odds ratio of “Arm 2” to “Arm 1.” “se 2” shows the corresponding within-trial standard error. The column labeled “V” contains the variance of the log odds of “Arm 1” only if the trial has more than two arms, as discussed in the section “Multi-arm trials.” The column labeled “PLA lo” contains the log odds for the baseline treatment.