TABLE 2.

Effect size scenarios explored for the Bayesian adaptive design case study

Scenario	$P_{\text{Control}}$ (%)	$P_{\text{CPAP}}$ (%)	$P_{\text{NIPPV}}$ (%)
(1) Null	16%	16%	16%
(2) One intervention superior	16%	16%	9.5%
(3) Both interventions superior	16%	9.5%	9.5%
(4) NIPPV > CPAP > Control a	16%	12.5%	9.5%
(5) Both interventions have small improvement	16%	12.5%	12.5%
(6) Harm	16%	18%	18%

Note: $P_{\text{Control}},P_{\text{CPAP}}$, $P_{\text{NIPPV}}$ are the 7‐day mortality rates for each arm; Note that $P_{\text{CPAP}}$ and $P_{\text{NIPPV}}$ could be interchangeable here, depending on whether the clinicians thought that continuous positive‐pressure ventilation (CPAP) or non‐invasive intermittent positive‐pressure ventilation (NIPPV) were more likely to be superior.

^a Here ‘>’ means ‘better than’, that is, have lower 7‐day mortality.