The paper is available on Sci-hub, but the paper itself is pretty light, referring primarily to the supplemental material[1] for details about the tests. I can't judge about the tests, but their control was "Negative human serum(IgA/IgM/IgG)" according to the supplemental materials, and if I understand correctly this is simply a basic serum that would always test negative and so I think this adds some credence to the idea that it might not be specific enough. Maybe it wasn't possible to get a control with an older coronavirus?

In any case, I agree that it's surprising, because knowing what we do about the virus now, if it was in the population back in September and assuming that the numbers in the general population were a lot higher than 23 in September, I think something would have been noticed earlier, although I guess it's somewhat plausible that it could have been masked by the normal autumn/winter flu hospitalizations. I also think it would have been caught in the research that the Italian NIH did earlier this year when they looked at wastewater[2] and found that it was there in mid-December 2019 but tested from October. Given these, plus this paper being both light on details and being published in what I can tell as a small journal makes me skeptical.

1: https://journals.sagepub.com/doi/suppl/10.1177/0300891620974... 2: https://www.reuters.com/article/us-health-coronavirus-italy-...