AD8: scoring, cutoffs & interpretation

01 - What it measures

What this scale measures.

The AD8 is an 8-item informant-based screen for early dementia developed at Washington University in St. Louis (Galvin et al. 2005). Unlike performance tests such as the MMSE or SLUMS, it asks someone who knows the patient well whether there has been *change* over recent years in memory, orientation, judgement and daily function - each item answered yes/no, total 0-8. Because it measures intra-individual change rather than a snapshot of ability, it is less confounded by education, language and baseline ability, and it can be completed when the patient cannot be tested directly.

02 - Origin & purpose

Where it comes from.

The AD8 was developed at Washington University in St. Louis and first described by Galvin and colleagues in 2005 (Neurology). It was designed to address a gap that performance tests leave open: detecting early dementia when a knowledgeable informant can describe decline that a single bedside assessment might miss. The instrument is copyright Washington University in St. Louis.

03 - Scoring & cutoffs

How scoring works.

A score of 0-1 is considered normal cognition; 2 or more endorsed items suggests cognitive impairment and warrants further assessment. In the original validation (Galvin 2005), the >=2 cutoff gave an area under the curve of 0.834, sensitivity 74% and specificity 86% for very mild dementia (CDR-based), with sensitivity rising to 85% when milder-to-severe dementia cases were included. The 2019 Cochrane review (Hendry et al.) pooled seven studies at the >=2 cutoff: sensitivity 0.92 (95% CI 0.86-0.96) but specificity 0.64 (0.39-0.82); at >=3, sensitivity 0.91 and specificity 0.76. The pattern - high sensitivity, modest specificity - fits a screen: a positive AD8 identifies people who need proper assessment, it does not diagnose dementia.

04 - Validation evidence

How well it performs.

The AD8 was developed against Clinical Dementia Rating (CDR) staging at Washington University, reporting an AUC of 0.834 and sensitivity 74% with specificity 86% at the >=2 cutoff in the 2005 validation sample of 112 CDR 0 and 68 CDR 0.5 participants; sensitivity rose to 85% when 56 people with mild-to-severe dementia were added and prevalence increased. The 2006 reliability study in 255 patient-informant dyads is the source of most of the figures clinicians quote: internal consistency of 0.84, intra-rater stability of weighted kappa 0.67, inter-rater ICC of 0.80, and a correlation with the CDR of 0.75. Administration by telephone was as reliable as in person (weighted kappa 0.65). The 2019 Cochrane review and a 2023 meta-analysis of 11 informant studies both confirm the same shape of result - high sensitivity, modest specificity - with the 2023 pooled informant estimates for dementia being 91% sensitivity against 64% specificity. Heterogeneity between studies was substantial and every study included in the 2023 review carried high or unclear risk of bias in at least one QUADAS-2 domain, most often around blinding of the index test and reference standard.

05 - How it compares

How it compares to the alternatives.

06 - When to use it

Right tool, wrong tool.

The AD8 is best used when an informant is available and the patient cannot be tested directly, or when there is concern about decline despite normal performance-test scores in highly educated patients. It is also useful in primary-care triage before referral.

07 - Confidence & precision

Reading the score with care.

The AD8's strength is sensitivity: pooled estimates of around 0.91-0.92 at the >=2 cutoff mean it catches most cases, but specificity near 0.64 means false positives are common and a positive screen must always be followed by full assessment. The informant's familiarity matters - someone who sees the patient daily over years gives a more reliable picture than a distant contact. The self-rated version is materially weaker and should not be treated as interchangeable: pooled self-rated accuracy for mild cognitive impairment is roughly 57% sensitivity and 71% specificity, against 80% and 79% for the informant version, because loss of insight, low mood and anxiety all distort self-report. The instrument's originators suggested a lower cutoff of 1 for the self-rated form, so the familiar >=2 threshold should not be applied to it without thought.

08 - Limitations

What it cannot tell you.

The AD8 cannot diagnose dementia - it flags the need for assessment. It depends entirely on having a reliable informant, and the informant's judgment of change is subjective. It is less suited to monitoring change over short intervals because it asks about decline over an indefinite recent period, not a fixed window. A patient self-report version exists but performs less well than the informant version. The AD8 is sometimes described as validated for emergency-department use, but the 2023 systematic review found no studies conducted in emergency or preoperative settings, so that claim should not be relied on. The AD8 is frequently described in the literature and on third-party sites as free and open-access, which is inaccurate as to Washington University's current posted terms.

09 - Licensing, explained

How licensing works.

The AD8 is copyright Washington University in St. Louis. Reproduction is permitted without modification solely for clinical care (a clinician's own non-research patient care) and non-commercial, investigator-initiated research. Clinical-trial and commercial uses - including electronic publication - require Washington University's prior written permission.

Because of this, Aisel cannot host the instrument, offer a scorable version, or reproduce the 8 items on this page. This page is an information resource: it explains scoring, cutoffs and evidence, but contains no AD8 item text. Clinicians can obtain the form for their own practice via Washington University or the Alzheimer's Association.

Common questions from clinicians

FAQ

References

Source literature.

1. [1]Galvin JE, Roe CM, Powlishta KK, Coats MA, Muich SJ, Grant E, Miller JP, Storandt M, Morris JC The AD8: a brief informant interview to detect dementia. Neurology. 2005;65(4):559-564. (2005) ↩
2. [2]Galvin JE, Roe CM, Xiong C, Morris JC Validity and reliability of the AD8 informant interview in dementia. Neurology. 2006;67(11):1942-1948. (2006) ↩
3. [3]Galvin JE, Roe CM, Coats MA, Morris JC Patient's rating of cognitive ability: using the AD8, a brief informant interview, as a self-rating tool to detect dementia. Archives of Neurology. 2007;64(5):725-730. (2007) ↩
4. [4]Galvin JE, Fagan AM, Holtzman DM, Mintun MA, Morris JC Relationship of dementia screening tests with biomarkers of Alzheimer's disease. Brain. 2010;133(11):3290-3300. (2010) ↩
5. [5]Hendry K, Green C, McShane R, Noel-Storr AH, Stott DJ, Anwer S, Sutton AJ, Burton JK, Quinn TJ AD-8 for detection of dementia across a variety of healthcare settings. Cochrane Database of Systematic Reviews. 2019;3:CD011121. (2019) ↩
6. [6]Tanwani R, Danquah MO, Butris N, Saripella A, Yan E, Kapoor P, Englesakis M, Chung F Diagnostic accuracy of Ascertain Dementia 8-item Questionnaire by participant and informant - a systematic review and meta-analysis. PLoS One. 2023;18(9):e0291291. (2023) ↩