We evaluated the reliability of clinical diagnoses using the recently
standardized criteria for the diagnosis of vascular dementia (VaD) dev
eloped by the National Institute of Neurological Disorders and Stroke
(NINDS) and the Association Internationale pour la Recherche et l'Ense
ignement en Neurosciences (AIREN). Two neurologists and two psychiatri
sts independently reviewed clinical data abstracted from those of 42 d
emented subjects participating in a longitudinal study of dementia at
the University of Pittsburgh. For each patient we abstracted the clini
cal data on a standardized form. Each physician diagnosed each case ac
cording to the NINDS-AIREN criteria, using both clinical information a
nd MRIs. We calculated the interrater agreement for all two-way combin
ations of clinicians with kappa statistics, which ranged from 0.46 (mo
derate agreement) to 0.72 (substantial agreement). The moderate reliab
ility observed in this study may be attributable to patient-, clinicia
n-, or criteria-centered sources of variance.