The validity of the diagnosis of mild dementia

Five operational methods for clinical diagnosis of mild dementia were compared to find out their diagnostic concordance when applied to a single group of seventy-five subjects. The clinical validity of the diagnoses was assessed in terms of their capacity to predict continued cognitive deterioration over three years after diagnosis and their capacity to reject the diagnostic influence of 'non-dementia' factors (that is, the cognitive consequences of depression, poor intellect, limited education and non-neurological physical illness). By all criteria of clinical validity the diagnostic methods for mild dementia performed poorly. Kappas measuring agreement between methods averaged only 0.15 and up to 57% of diagnostic deviance was explained by 'non-dementia' factors. Prediction of continued deterioration was poor, with a false positive rate which was too high for the diagnoses to be clinically usable. By no criterion of validity did the diagnostic methods exceed the performance of a clinician's judgement of the presence of pathological cognitive impairment or diagnosis by a cutpoint on the Mini-Mental State Examination.