Fig. 4

Example visualisations illustrating how visual presentation of days alive without life support or similar outcome data may aid interpretation, using days alive without life support after 28 days from the COVID STEROID 2 trial [15]. Upper left sub-plot: cumulative percentage of patients (vertical axis) with less than or as many days alive without life support as listed on the horizontal axis in each group, after assigning -1 days to non-survivors. This sub-plot shows that patients in the control group had less days alive without life support over the full range of values. Upper right sub-plot:”heat map” visualising the proportion of patients (horizontal axis) with each value in each group after assigning 0 days to non-survivors; this figure is similar to an overturned stacked bar plot, displaying the entire distributions with each unique value having its own colour in a red-to-blue gradient (with colours corresponding to the number of days illustrated in the legend below the plot). Similar to the first sub-plot, this sub-plot shows that, over the full range of values, patients in the control group had fewer days alive without life support. Lower sub-plots: distributions of patients in each possible state (vertical axes: alive and home; alive and in hospital; alive, in hospital and on life support; dead) on each day after randomisation (horizontal axes) separately in the two treatment groups. This sub-plot shows additional details compared to the first two: mortality was higher in the control group with more early deaths, and survivors were on life support longer compared to the intervention group