. 2022 Sep 18;41(28):5517–5536. doi: 10.1002/sim.9580

TABLE 3.

For Scenario 2 with binary outcome, interim analysis performance using each estimator with O'Brien‐Fleming and Pocock stopping boundaries under (a) the null hypothesis $β = 0$ and (b) the alternative $β = \log (0.247 / 0.33) = - 0.290$ , with maximum sample size $n_{\max} = 900$ and $t_{e n d} = 330$ days

P(reject)

MC E(SS)

MC E(Stop)

P(reject)

MC E(SS)

MC E(Stop)

(a) Null hypothesis

O'Brien‐Fleming

Pocock

{\hat{β}}_{𝒯_{F}} (t)

0.024

900.0 (2.3)

329.4 (6.5)

0.022

896.3 (33.6)

327.4 (19.8)

{\hat{β}}_{I P W} (t)

0.024

898.8 (15.8)

328.0 (14.5)

0.023

894.3 (42.3)

326.5 (23.7)

{\hat{β}}_{A I P W 1} (t)

0.023

899.0 (14.0)

328.1 (13.8)

0.025

894.1 (42.8)

326.3 (24.3)

{\hat{β}}_{A I P W 2} (t)

0.024

899.0 (14.9)

328.0 (14.2)

0.026

894.1 (42.7)

326.2 (24.4)

(b) Alternative hypothesis

O'Brien‐Fleming

Pocock

{\hat{β}}_{𝒯_{F}} (t)

0.770

887.5 (44.3)

285.9 (44.0)

0.690

827.2 (122.4)

264.4 (67.2)

{\hat{β}}_{I P W} (t)

0.767

808.3 (121.4)

241.3 (61.3)

0.700

744.9 (157.6)

228.3 (76.7)

{\hat{β}}_{A I P W 1} (t)

0.806

801.8 (122.5)

236.1 (59.6)

0.746

733.2 (157.3)

221.4 (74.7)

{\hat{β}}_{A I P W 2} (t)

0.809

799.9 (123.4)

235.5 (59.7)

0.748

731.6 (157.3)

220.6 (74.7)

Note: Entries are as in Table 2. The standard error for entries for P(reject) in (a) is $\approx 0.0016$ .