Objective:

The article examines the objective, findings, and clinical context described in Addressing Time-Dependent Bias in Observational Studies of ECPR.

Approach:

• Bias mechanisms: The article describes resuscitation time bias among CCPR patients and immortal time bias among patients classified as receiving ECPR.
• Methods considered: It discusses time-dependent adjustment, propensity scores, landmark analyses, target trial emulation, instrumental variables, and interrupted time-series analyses.
• Proposed research priorities: The authors propose improved registry timestamps, a multicenter prospective observational study with time-varying data, and protocolized minimum CCPR durations among ECPR-eligible patients.

Key Findings:

• Early ROSC in the CCPR group can bias estimates against ECPR because these patients may never reach the point of ECPR eligibility.
• ECPR patients must survive the interval from the treatment decision through transport and cannulation to pump initiation; misclassifying this interval can bias estimates in favor of ECPR.
• The decision point for ECPR is often difficult to identify retrospectively and is infrequently recorded in registries.
• Adjustment methods may partly reduce bias, but require adequate pre-exposure comparability in low-flow duration and OHCA severity; overlap is limited when low-flow duration is prolonged.
• Landmark analyses can exclude early CCPR survivors but may leave too few comparable patients. Target trial emulation and quasi-experimental approaches also face clinical, ethical, or methodological constraints.

Interpretation:

The authors state that structural limitations constrain observational ECPR–CCPR comparisons and that statistical adjustment alone cannot fully overcome them. Few clinically comparable CCPR patients remain at prolonged low-flow durations, limiting causal inference.

Limitations:

• The exact time when CCPR becomes a bridge to ECPR is often unrecorded, making time-zero specification difficult.
• Few clinically comparable CCPR patients may remain after prolonged resuscitation, restricting overlap and causal inference.
• Landmark analyses may exclude relevant patients or leave insufficient numbers for comparison.
• Instrumental-variable analyses require a valid, relevant instrument; transport distance and time may directly relate to OHCA mortality.
• Interrupted time-series analyses may be confounded by concurrent improvements in OHCA care.

Conclusion:

The authors prioritize standardized registry timestamps for the ECPR decision, cannulation, ECMO flow, and resuscitation termination; prospective multicenter collection of time-varying data; and protocolized minimum CCPR durations among ECPR-eligible patients.