Addressing Time-Dependent Bias in Observational Studies of ECPR

1

Observational ECPR comparisons face resuscitation time bias from early ROSC in CCPR patients and immortal time bias before ECPR pump initiation.
2

ECPR recipients are a selected cohort that must survive logistical delays, while registries often cannot identify when the decision to pursue ECPR was made.
3

Time-zero synchronization and adjustment for low-flow duration and OHCA severity may partly reduce bias, but limited overlap restricts causal inference.
4

Landmark analyses and target trial emulation have limitations because they can exclude patients or require clinically and ethically implausible prolonged CCPR comparisons.
5

The authors propose standardized registry timestamps, prospective collection of time-varying data, and protocolized minimum CCPR durations for ECPR-eligible patients.

1