I think this is wrong: Ioannidis's major point is that a lot of studies in medicine are underpowered (a similar issue to the replication crisis in psychology), leading to a lot of false positives. But there is no reason to think that this would be the case for a study that finds a vitamin deficiency in the population. We can expect no issues with p-hacking or garden of forking paths just because there is no place for them (the researchers are not looking for crazy-small interactions between vitamin levels and one of many possible outcomes, they're just recording the vitamin concentrations), and we know roughly the right concentrations for those vitamins too.

So I think this is misplaced skepticism, you shouldn't really dismiss studies so easily without thinking through the issues first. This would be much more relevant if the research was specifically about some random compound's small effect on Covid-19 without a prior reason to think it would be relevant at all - then the problems would indeed appear.