In reply:

Abstract

We appreciate the letter by Yu and Li1 in regard to the study design and statistical analyses in our study titled “Risk of Venous Thromboembolism After Carbon Monoxide Poisoning: A Nationwide Population-Based Study.” They mentioned that one of the limitations of the cohort-crossover design is that there may be covariates that change over time, such as ICU admission. We also agree that cohort-crossover designs may be affected by time-varying within-person confounding. To overcome this limitation, we performed a sensitivity analysis by adjusting for hospitalization with a condition increasing the risk of venous thromboembolism.2 As a result, the risk of venous thromboembolism was observed to be increased after carbon monoxide poisoning, as in the main analyses.