Sure, the data isn't perfect, but data rarely is. You're also assuming that, because the data isn't perfect, that the effect measured would go away with better

Sure, the data isn't perfect, but data rarely is. You're also assuming that, because the data isn't perfect, that the effect measured would go away with better data. I actually think the opposite would happen. There is a correlation between the severity of self-reported sickness and cognitive decline. Those who ended up in a hospital show the most severe signs of cognitive decline. If they were actually in a hospital, they likely know for sure whether or not they were positive, compared to those who recovered at home and only suspected they had covid. There have also been visible changes to the brain when looking at MRIs before / after a person has covid.