Objective To report the diagnostic accuracy of AD8, a brief informant screening questionnaire for the detection of dementia and cognitive impairment.

Methods Prospective observational study, Cognitive Function Clinic.

Results Of 153 consecutive patients seen between July-December 2013, 100 (M:F=55:45; age range 16–92 years, median 63 years) attended with a suitable informant. No informant declined when invited to perform AD8. Of 100 patients, 33 were diagnosed with dementia (DSM-IV-TR criteria), 26 had MCI (Petersen criteria). Using the cutoff ≥2/8 (“cognitive impairment is likely to be present”; higher AD8 scores are worse) as defined in the index paper, AD8 score was sensitive (0.97) but not specific (0.15) for cognitive impairment. Looking at all AD8 scores, the most accurate cutoff (0.74) was ≥6/8, with sensitivity and specificity for cognitive impairment of 0.76 and 0.70 respectively, with area under the ROC curve of 0.77. AD8 scores showed no correlation with patient age (r=0.14) and low correlations with cognitive screening instrument scores (MMSE and 6CIT: r=−0.24 and 0.39 respectively).

Conclusions AD8 is simple to administer and score. With appropriate adjustment of the cutoff, accounting for the pragmatic clinical sample, AD8 has value in identifying and ruling out cognitive impairment.