The technical issue is whether we need a one-sided (leron better than placebo) or two sided (leron not equal to placebo) confidence interval. A 95% interval (p<.05) one-sided is a 90% interval (p<.1) two sided. The latter makes sense with a drug that might do damage.
So, given 17 placebo and 34 leronlimab patients, my table could read
___________Maximum Leronlimab deaths for significance (p<.05):
If Placebo____One-sided test_____Two sided deaths are