The issue is the people (and the computer systems) it passes through for peer review.

There's a small number of publishers who put out most journals and not that many peer reviewing scientists per subfield. If you didn't publish preliminary results like this, those systems would be high value bottlenecks through which market sensitive data would flow all the time and a lot of effort would be put into breaking into them.

(edit: I do think it would be better to put out a preprint on Medrxiv at the same time though)