Table 1.

Models fitted to endpoints for the simulated data.

Model	Endpoint	Endpoint Variable Type	Coefficient Interpretation
Simple Logistic	E1: Proportion of patients hospitalized or dead on day 7	Binary	Odds ratio of discharged from the hospital versus not discharged on day 7
Sliding Dichotomy	E2: Proportion of patients moving to less severe categories from day 0 to day 7	Binary	Odds ratio of moving versus not moving to a less severe category from day 0 to day 7
Win Ratio	E3: Winners versus losers between IVIG and placebo on day 7	Binary	For all possible comparisons of patients in IVIG versus placebo, the number of IVIG winners divided by the number of IVIG losers
Proportional Oddsa	E4: Day 7 ordinal endpoint	Ordinal	Average odds ratio of being in a less versus more severe category on day 7
Longitudinal Ordinal Outcome	E5: Distribution of the ordinal endpoint over the seven days of follow-up	Ordinal Longitudinal	Average multiplicative increase in the odds ratio of being in a less versus more severe category across the follow-up period
Cox Proportional Hazards	E6: Number of days to first hospital discharge	Time-to-Event	Hazard ratio of time-to-hospital discharge
Accelerated Failure Time (Exponentiala and Weibull distributions)	E6: Number of days to first hospital discharge	Time-to-Event	Reduction in quantiles of time-to-hospital discharge

^aModel was used in the analytic setting, displayed in Fig. 1, Fig. 2.