Synchronizing Time Zero and Treatment Assignment in Target Trial Emulation of Left Ventricular Unloading

1

Requiring at least 48 hours of VA-ECMO while setting time zero at cannulation could exclude patients who died or were weaned earlier.
2

Classifying unloading by receipt within 24 hours after cannulation may create immortal-time bias because patients must survive until device placement.
3

Cloning, artificial censoring, and inverse-probability-of-censoring weighting can align treatment strategies during a grace period and assign early events appropriately.
4

Time-varying clinical indications for unloading may confound its association with mortality, and baseline propensity-score weighting cannot address this confounding.
5

The weighted risk difference was 1.4% (95% CI, −11.8% to 14.6%), indicating no detectable association but not establishing equivalence or excluding benefit or harm.

1