The rapid availability of third world originated studies to internet searches by those living elsewhere, which has happened because of internet publishing and especially in the last three years because of openly searchable preprint sites such as medrxiv, is a new enough thing in science that I doubt there has been enough time to get a valid such likely future failure rate, especially if you are including preprints.

To get an approximate base rate you would need to check the years prior to open internet medical paper publishing. At minimum you would need to read the older years of paper versions of those nation's journals, likely in their medical schools, and count the studies fitting your criteria over the past ten years or more, ideally even including the studies rejected by such journals if you are including preprints in your questions about the Covid-19 studies. Not an easy project, so best to take the studies you mention as proposing a hypothesis, perhaps to be tested in the future via reproducibility. Just like most other clinical studies published by anyone else anywhere.