Exact confidence limits after a group sequential single arm binary trial

摘要: Group sequential single arm designs are common in phase II trials as well attribute testing and acceptance sampling. After the trial is completed, especially if recommendation to proceed further testing, there interest full inference on treatment efficacy. For a binary response, potential construct exact upper lower confidence limits, first published method for which Jennison Turnbull (1983). We place their within modern theory of limits provide new general result that ensures consistent with test result, an issue has been largely ignored literature. Amongst methods based minimal sufficient statistic, we propose two out-perform Turnbull's across 10 selected designs. One these prefer recommend practical theoretical reasons. also investigate inverting Fisher's combination test, pure tie-breaking variant it. range considered, neither large enough improvements efficiency justify violation sufficiency principle. any nonadaptive design, R-package provided select compute from given realization.