Part of the difficulty is that pathology departments are over burdened right now and autopsy specimens are backed up.
That may be the issue in the US and, if so, will permit retrospective correction of the numbers, up or down, on the assumption that whatever mechanism is eventually used to examine the specimens produces tolerably accurate results (which in vivo nasopharyngeal swabbing dismally fails to do).
It's not the case in the UK, however, where many of the certified deaths are no more than guesses by those with no ability to make even an educated guess, and where as far as I know no specimens have been taken from the individuals, ante- or post-mortem.
Jeremy