Users can't really tell that an email is styled or plain before opening it, so if the opened numbers were too different, I'd be concerned that something was up.
And with the open rate that significantly affected, all other metrics would be skewed as well.