I could see it happening if your data source works of raw test result statistics, since you don't know how often the same person was tested. But that doesn't so

I could see it happening if your data source works of raw test result statistics, since you don't know how often the same person was tested. But that doesn't sound like it is the case here, since they reference a specific database they use "at individual patient level".