"3 (1.2%) vs 10 (4%) 28-day in-hospital deaths...Mechanical ventilation occurred in 4 (1.7%) vs 10 (4.0%)...intensive care unit admission in 6 (2.4%) vs 8 (3.2%)" (ivermectin vs. non-ivermectin)
These authors made a huge mistake. They didn't check the odds of all these secondary differences being slanted towards ivermectin at once. A series of low-but-above-.05 P values in the same direction show correlation even more strongly than a single near-.05 P value does.
Also, the principal result is more narrowly-scoped than reported. Though within its scope, the study's accurate: When high-risk patients wait 4-7+ days since first COVID symptoms to decide to visit a clinic and for a PCR test to come back positive, ivermectin doesn't reduce "progression to severe COVID". But ivermectin nevertheless prevents many cases of secondary COVID effects in high-risk patients, like death and need for ventilator.
Among the 490 patients, 95 (19.4%) progressed to severe disease during the study period; 52 of 241 (21.6%) received ivermectin plus standard of care, and 43 of 249 (17.3%) received standard of care alone...
The 28-day in-hospital mortality rate was similar for the ivermectin and control groups (3 [1.2%] vs 10 [4.0%]...
If 52 received ivermectin and 3 died, that's about 5.8%. If 43 received standard care alone and 10 died, that's about 23.3%. So the actual mortality difference was 5.8% vs 23.3%.