Bayesian group sequential designs for phase III emergency medicine trials: a case study using the PARAMEDIC2 trial

Ryan, Elizabeth G.; Stallard, Nigel; Lall, Ranjit; Ji, Chen; Perkins, Gavin D.; Gates, Simon

doi:10.1186/s13063-019-4024-x

Trials

Table 3 Operating characteristics for Bayesian group sequential designs for PARAMEDIC2

From: Bayesian group sequential designs for phase III emergency medicine trials: a case study using the PARAMEDIC2 trial

Design and scenarios^a	Average duration (weeks)	Average sample size (sd)	Proportion stopped early^b	Overall proportion declaring a difference^c	Proportion that did not declare a difference	Proportion flip-flop ^d	Average probability adrenaline superior
Bayesian Design 1 (B1)
Null: Placebo 6% vs Adrenaline 6%	168	7968 (390)	0.0074	0.0493	0.9506	0.0001	0.5004
Placebo 8% vs Adrenaline 6%	133	6100 (2075)	0.502	0.935	0.064	0.001	0.0072
Placebo 6% vs Adrenaline 8%	131	6019 (2107)	0.525	0.928	0.072	0	0.9932
Placebo 7% vs Adrenaline 6%	163	7676 (1149)	0.084	0.431	0.569	0	0.0995
Placebo 6% vs Adrenaline 7%	162	7654 (1174)	0.09	0.433	0.567	0	0.8961
Null: Placebo 3% vs Adrenaline 3%	168	7980 (293)	0.0053	0.044	0.956	0	0.5039
Placebo 5% vs Adrenaline 3%	109	4842 (1985)	0.779	0.994	0.006	0	0.0009
Placebo 3% vs Adrenaline 5%	103	4546 (1960)	0.828	0.995	0.004	0.001	0.9991
Placebo 4% vs Adrenaline 3%	158	7410 (1396)	0.17	0.659	0.341	0	0.0435
Placebo 3% vs Adrenaline 4%	155	7285 (1524)	0.202	0.663	0.337	0	0.9571
Null: Placebo 2% vs Adrenaline 2%	168	7987 (227)	0.004	0.0371	0.9629	0	0.5028
Placebo 4% vs Adrenaline 2%	96	4140 (1641)	0.908	0.999	0.001	0	0.0004
Placebo 2% vs Adrenaline 4%	91	3883 (1589)	0.934	1	0	0	0.9995
Placebo 3% vs Adrenaline 2%	153	7172 (1587)	0.234	0.792	0.208	0	0.0238
Placebo 2% vs Adrenaline 3%	150	6991 (1709)	0.288	0.814	0.186	0	0.9785
Bayesian Design 2 (B2)
Null: Placebo 6% vs Adrenaline 6%	168	7961 (440)	0.0121	0.0484	0.9516	0	0.5001
Placebo 8% vs Adrenaline 6%	134	6137 (1911)	0.706	0.92	0.08	0	0.0085
Placebo 6% vs Adrenaline 8%	128	5836 (1984)	0.79	0.943	0.057	0	0.9936
Placebo 7% vs Adrenaline 6%	163	7695 (978)	0.144	0.419	0.581	0	0.1132
Placebo 6% vs Adrenaline 7%	161	7584 (1217)	0.158	0.436	0.564	0	0.9039
Null: Placebo 3% vs Adrenaline 3%	168	7980 (296)	0.0075	0.0467	0.9533	0	0.5006
Placebo 5% vs Adrenaline 3%	110	4882 (1932)	0.901	0.991	0.008	0.001	0.0011
Placebo 3% vs Adrenaline 5%	106	4689 (1903)	0.938	0.996	0.004	0	0.9991
Placebo 4% vs Adrenaline 3%	156	7343 (1375)	0.282	0.658	0.341	0.001	0.0524
Placebo 3% vs Adrenaline 4%	155	7260 (1430)	0.313	0.665	0.335	0	0.9587
Null: Placebo 2% vs Adrenaline 2%	168	7984 (242)	0.0079	0.0411	0.9589	0	0.5014
Placebo 4% vs Adrenaline 2%	98	4248 (1706)	0.974	1	0	0	0.0004
Placebo 2% vs Adrenaline 4%	94	4019 (1674)	0.983	1	0	0	0.9996
Placebo 3% vs Adrenaline 2%	152	7106 (1460)	0.406	0.779	0.221	0	0.0257
Placebo 2% vs Adrenaline 3%	148	6928 (1621)	0.443	0.789	0.211	0	0.978
Bayesian Design 3 (B3)
Null: Placebo 6% vs Adrenaline 6%	167	7936 (492)	0.027	0.0515	0.9481	0.0004	0.5003
Placebo 8% vs Adrenaline 6%	123	5562 (1879)	0.827	0.935	0.065	0	0.0066
Placebo 6% vs Adrenaline 8%	118	5333 (1829)	0.87	0.945	0.055	0	0.9924
Placebo 7% vs Adrenaline 6%	159	7497 (1183)	0.235	0.424	0.574	0.002	0.0959
Placebo 6% vs Adrenaline 7%	157	7382 (1216)	0.303	0.435	0.562	0.003	0.9017
Null: Placebo 3% vs Adrenaline 3%	168	7957 (293)	0.0223	0.0492	0.9505	0.0003	0.5002
Placebo 5% vs Adrenaline 3%	101	4416 (1587)	0.976	0.995	0.005	0	0.011
Placebo 3% vs Adrenaline 5%	97	4186 (1535)	0.982	0.993	0.006	0.001	0.9985
Placebo 4% vs Adrenaline 3%	151	7052 (1475)	0.431	0.66	0.338	0.002	0.043
Placebo 3% vs Adrenaline 4%	146	6821 (1512)	0.53	0.678	0.32	0.002	0.9552
Null: Placebo 2% vs Adrenaline 2%	168	7970 (282)	0.0173	0.0415	0.9576	0.0009	0.4979
Placebo 4% vs Adrenaline 2%	90	3854 (1364)	0.996	0.999	0.001	0	0.0006
Placebo 2% vs Adrenaline 4%	87	3643 (1327)	0.996	0.999	0.001	0	0.9993
Placebo 3% vs Adrenaline 2%	144	6691 (1594)	0.564	0.788	0.21	0.002	0.0217
Placebo 2% vs Adrenaline 3%	139	6436 (1592)	0.66	0.80	0.198	0.002	0.9776

^aDifferent effect size scenarios that were simulated for each design are given as placebo 30-day survival rate vs adrenaline 30-day survival rate
^bProportion of simulations that stopped early and were declared to have a difference (in the correct direction) at the final analysis
^cThe simulated type I errors are italicised
^dThese simulations were stopped early for efficacy or harm, but they did not meet the critical values to declare a difference between the treatments at the final analysis once all patients were followed up (insufficient evidence of a difference)

Back to article page

ISSN: 1745-6215

Contact us

Submission enquiries: Access here and click Contact Us
General enquiries: info@biomedcentral.com