This is disheartening. The dose used seems to be the one used by
Fowler et al. (50mg-C/kg-weight) giving rise to a clear benefit by day 4 (the maximum duration of the treatment):

I have taken a look at
the second paper and I don't see any critically concerning point except for some perhaps minor points.
1.- Vitamin C used:
All sites used locally available vitamin C formulations
no one of them (page 6 of the Supplemental Appendix 1) was the one used by Fowler et al. I am beginning to think that quality (ratio of ascorbic to dehydroascorbic perhaps) could be an issue.
2.- Looking at the mortality results only the LOVIT trial, taking place in Canada, was positive but without statistical significance (too low number of patients). From eTable 12 on page 26 of Supplementala Appendix 1, 24 deaths from 84 on the treatment arm of LOVIT critically ill patients, 33 deaths from 97 on the control arm. This gives rise to a
relative risk of death ratio of 0.8754 but with a very wide 95% confidence interval of 0.5528 to 1.3864. Meanwhile taking both trials together there are 395 deaths from 1037 in the treatment arm, 189 deaths from 531 in the control arm (same table, with numbers somewhat different than in the main paper), giving rise to a relative risk of death ratio of 1.0508 with 95% confidence interval of 0.9059 to 1.2190.
3.- Things doesn't seem to be favorable for vitamin C even for the first 4 days (treatment duration) from Figure 2.c. That's strange given the results of Fowler
et al. shown just above.
PS: Owen, I hope you are doing better.