This methodological study compared the effect estimates among different time-zero settings in a comparative effectiveness study with a non-user comparator using actual real-world data.

The treatment effects, based on the adjusted hazard ratio for the outcome, varied among the six time-zero setting methods, even though the same dataset was used for the same research question.

An incorrectly set time zero can result in a biased result and mislead the conclusion; therefore, researchers should carefully select the time zero to minimize those biases when designing a study.