We also repeated the individual- and family-level analyses for any AD excluding individuals diagnosed before 2001, date from which outpatients were first included in the register, to ensure that more severe patients (i.e., inpatients) were not biasing the results.