Table 3.

Estimated Subdistribution Hazard Ratio of Time to Clinical Stability and Time to Hospital Discharge, and Estimated Cause-Specific Hazard Ratio of Time to in-Hospital Death (Non-Steroid User vs Steroid Users)^a

	Estimated Hazard Ratio, No (95% CI)
Time to clinical stability, subdistribution hazard ratio	0.89 (0.78–1.0)
Time to hospital discharge, subdistribution hazard ratio	1.21 (0.81–1.8)
Time to in-hospital death, hazard ratio	0.93 (0.92–0.94)

Notes: ^aThe results were calculated according to the following steps: multiple imputation of original data, calculating propensity score for steroid use within each dataset using the potential confounding factors (age, sex, hospitals, use of home oxygen therapy, activity of daily living, altered mental status, respiratory rate, heart rate and blood urea nitrogen), propensity score matching, survival analyses within each dataset using the potential confounding factors (Fine and Gray model for time to clinical stability and time to hospital discharge, and Cox proportional hazards model for time to in-hospital death), and combining the results.

Abbreviation: CI, confidence interval.