Incorporating individual historical controls and aggregate treatment effect estimates into a Bayesian survival trial: a simulation study

Table 1 Summary of the 32 scenarios considered for the simulation of the historical controls and new trial data

Scenario	Survival distribution of historical controls^a	Generation of new data
		Survival distribution^b	Parameters
		Survival distribution^b	Control arm	Treatment effect^c
S1	Weibull	Weibull	Commensurate controls	Null
S2				Disappointing
S3				Historical
S4				Anticipated
S5	Weibull	Weibull	Negative prior-data conflict	Null
S6				Disappointing
S7				Historical
S8				Anticipated
S9	Weibull	Weibull	Positive prior-data conflict	Null
S10				Disappointing
S11				Historical
S12				Anticipated
S13	Piecewise exponential	Piecewise exponential	Commensurate controls	Null
S14				Disappointing
S15				Historical
S16				Anticipated
S17	Piecewise exponential	Piecewise exponential	Negative prior-data conflict	Null
S18				Disappointing
S19				Historical
S20				Anticipated
S21	Piecewise exponential	Piecewise exponential	Positive prior-data conflict	Null
S22				Disappointing
S23				Historical
S24				Anticipated
S25	Weibull	Piecewise exponential	Commensurate controls	Null
S26				Disappointing
S27				Historical
S28				Anticipated
S29	Piecewise exponential	Weibull	Commensurate controls	Null
S30				Disappointing
S31				Historical
S32				Anticipated

^aSurvival distribution used to generate individual historical controls
^bSurvival distribution used to generate individual patient data for the control arm of the new trial
^cNull, Disappointing, historical and anticipated effects correspond to a hazard ratio of 1, 0.886, 0.786, and 0.55 in the new trial, respectively

ISSN: 1471-2288