The methodology mentions nothing about randomization so its unclear if there was any selection bias here. If you look a table 2, the control group was on average 6 years older and had more co-morbidities. This suggests that there was some selection bias and that other undocumented factors could have greatly affected this experiment. Doing multi-variate analysis with such a small sample size gives rise to huge uncerta…
It's an observational study, which by definition is not randomized. It's pure association, it may very well be pure confounding. In favor of the authors: The only thing they really claim is that an RCT should be done. Which I guess is an okay'ish conclusion from such a study.
Larger randomized trials are in progress [1][2] but the results won't be published until July.
[1] https://clinicaltrials.gov/ct2/show/NCT04344041 [2] https://clinicaltrials.gov/ct2/show/NCT04351490.