Approaches to Selecting "Time Zero" in External Control Arms with Multiple Potential Entry Points: A Simulation Study...

Area of Science:

• Health Services Research
• Biostatistics
• Pharmacoeconomics

Background:

• Estimating comparative effectiveness using external control arms requires defining "time zero," the eligibility point for a contemporary study.
• This choice is complex when patients have received multiple lines of therapy, offering alternative time zero points.

Purpose of the Study:

• To evaluate the performance of different methods for selecting "time zero" when using external control arms.
• To identify optimal strategies for defining the start of a study cohort in comparative effectiveness research.

Main Methods:

• A simulation study compared 8 methods for selecting "time zero" in external control arms.
• The simulation incorporated multiple prior lines of therapy and varied study attributes like sample size and covariate overlap.

Main Results:

• Five methods demonstrated good performance in accounting for patient inclusion differences: all lines, random line, and those based on mean absolute error, root mean square error, or propensity scores.
• The first eligible line can be statistically inefficient; all lines with censoring are unsuitable for survival outcomes; the last eligible line is not recommended.

Conclusions:

• Several viable methods exist for selecting "time zero" from external control arms, though some perform poorly.
• Analysts must carefully consider the study context and justify their chosen method for defining "time zero".