Non-COVID is one thing, but even among those with COVID, we may have a biased sample here. Suppose that 97% of the COVID athletes didn't have chest pain or oth

Non-COVID is one thing, but even among those with COVID, we may have a biased sample here. Suppose that 97% of the COVID athletes didn't have chest pain or other symptoms of myocarditis and thus had no reason to get an MRI, and that the remaining 3% is broken down into "2% got an MRI and didn't have myocarditis, 1% got an MRI and did have myocarditis"; that would match the reported 1/3 rate of myocarditis among those who got MRI'd, but it would be only 1% among those with COVID. We've seen this type of sampling bias before, where people having a problem makes them more likely to choose to get tested for it, which leads to an unrealistically high "x% of people have this problem" estimate.