Hard to rationalise that.
Is the data all day of sample/admission/death, or day of reporting?
spec date, admission date, death date
i think it's a mixture of Marvin's theory that more 85+ can be accepted to hospital (which strikes a sad note for what has passed) and those positive tests are coming in hospital itself and the admission is separate? but then would the admission even be recorded.....
building on @Marvin 's suggestion, it could be that now hospital's can accept many more patients, oldies are now quickly removed from their circumstances to prevent spread? this is all baseless theory.
i think it's a bit strange myself.