TABLE 2.

Sequential analysis results for the rofecoxib vs ns-NSAID comparison. This application has two different outcomes, myocardial infarction (w₁ = 2.2), and gastrointestinal bleeding (w₂ = 0.04). The tuning parameters used are N = 1000, α = 0.05, and power-type Type I error spending with ρ = 0.5. The critical values are shown on the scale of the ratio between weighted sums from exposure A to exposure B populations (T = S_i,A/S_i,B)

Cumulative data($\hat{R}$)				Critical value		Alpha spending
Test	Gastro. bleed.	Myoc. inf.	T	Lower	Upper	Target	Actual
1	22 (0.85)	24 (1.20)	0.85	0.34	2.97	0.0107	0.0089
2	45 (1.03)	59 (1.14)	1.04	0.48	2.07	0.0161	0.0161
3	59 (1.20)	90 (1.11)	1.19	0.55	1.80	0.0193	0.0191
4	74 (1.38)	133 (1.11)	1.37	0.62	1.60	0.0227	0.0227
5	92 (1.36)	170 (0.88)	1.36	0.65	1.56	0.0256	0.0255
6	108 (1.28)	196 (0.86)	1.27	0.68	0.148	0.0276	0.0275
7	122 (1.25)	225 (0.94)	1.25	0.69	1.44	0.0295	0.0294
8	135 (1.27)	252 (0.99)	1.27	0.71	1.42	0.0311	0.0310
9	146 (1.23)	286 (1.00)	1.27	0.72	1.38	0.0329	0.0328
10	156 (1.25)	319 (1.05)	1.24	0.74	1.36	0.0345	0.0344
11	168 (1.24)	340 (1.05)	1.24	0.74	1.34	0.0356	0.0356
12	177 (1.23)	353 (1.08)	1.23	0.75	1.34	0.0364	0.0364
13	181 (1.27)	368 (1.06)	1.27	0.65	1.34	0.0370	0.0370
14	189 (1.30)	384 (1.03)	1.30	0.76	1.32	0.0378	0.0378
15	199 (1.32)	395 (1.03)	1.32	0.76	1.32	0.0385	0.0385
16	206 (1.32)	408 (1.04)	1.32	na	na	na	na
17	218 (1.31)	428 (1.02)	1.31	na	na	na	na
18	236 (1.27)	457 (0.98)	1.27	na	na	na	na
19	247 (1.24)	478 (0.96)	1.24	na	na	na	na
20	258 (1.24)	485 (0.96)	1.23	na	na	na	na
21	265 (1.23)	491 (0.92)	1.23	na	na	na	na