Cognitive assessment tools: choose the right screen, use it properly
Domains of cognition, SCD → MCI → dementia, GPCOG, MMSE plus clock, MoCA (encoding vs retrieval), RUDAS, and when a “normal” screen is a red flag.
- Dr Rebecca Moore
- Introduced in the Australian Dementia Network context. Credentials beyond the transcript are not invented here.
Big take-homes
Choose the right tool for the right person, and use it properly. Cognitive screens are helpful for diagnosis in context — but they are not diagnostic tests on their own, and they are not population screening tools.
In the room, about a third to half of GPs said they perform a cognitive assessment about once a week; roughly half do it themselves and half use the practice nurse. Formal training in which tool and how to interpret it is often thin — this talk fills that gap.
What cognition actually is
Memory is what most people think of first. The brain is bigger than that. A useful screen looks across several domains so you get a clearer picture of what is going on.
- Visuospatial — where we are in space, navigation, judging distance.
- Recall / encoding — Alzheimer’s pathology often starts in hippocampal regions that encode new memory, so short-term memory problems appear early.
- Executive function (more frontal) — organisation, planning, judgment, insight, rationalising, mental flexibility.
- Attention — multitasking and sustained focus.
- Language — understanding and expressing; new word-finding difficulty, new grammatical/syntax errors, or (in some dementias) difficulty forming the words themselves.
- Motor / praxis — using tools, motor control, praxis.
- Social / personality — behaviour and personality change; easy to miss if you only chase memory scores.
Normal aging → SCD → MCI → dementia
Normal aging is a bit like a computer slowing down. Processing and multitasking take longer, but the information is still there.
Subjective cognitive decline (SCD) — you feel word-finding slips or more careless errors, but others have not noticed, and formal screens are normal. Flag and review these people: there is a small associated risk of later dementia, but most do not progress.
Mild cognitive impairment (MCI) is where we want the diagnosis made. Others notice problems; screens are worse than expected for that person; but function remains intact (ADLs done normally — maybe with more lists and calendar prompts). That intact function is the distinguishing line from dementia.
Dementia — function is impaired.
MCI has many causes: mood, grief, medications, metabolic issues, sleep apnoea, through to a prodrome of dementia. Some MCI reverses, many stay stable, and about 40–50% progress to dementia — that is the cohort we most want to identify.
Patterns by dementia type
| Type | What often shows early |
|---|---|
| Alzheimer’s | Hippocampal / short-term memory; often early language fluency, executive, and visuospatial change too. |
| Vascular | Can hit anywhere — mixed picture on assessment. |
| Lewy body | Memory components often relatively OK; visuospatial often affected early. |
| Frontotemporal (FTD) | Younger onset; behaviour and personality. May pass cognitive screens — delayed diagnosis and false reassurance. Behaviour change + family concern + good scores = red flag → refer. |
Tools are not diagnostic
Good tools are objective and reproducible over time and between operators. They are not diagnostic by themselves, and they should not be used as carte blanche population screening. Best use: someone with cognitive concerns, or case-finding in that context.
GPCOG
Takes about five minutes, scored out of nine. Starts with remembering a name and address; includes a plot-drawing task. It is the only validated screening tool in the over-75 population — useful in the over-75 health check or driving assessments when you expect someone to be well. Nine out of nine is fairly reassuring. If the score is worse than expected, use the informant questionnaire, follow the investigation suggestions, and move to a more in-depth screen.
MMSE and clock drawing
The MMSE is embedded in software and familiar from medical school — but map it against the domain slide and the gaps show: little on visuospatial (overlapping pentagons only), little executive beyond the three-stage command, and very little language fluency.
It is not particularly sensitive to subtle change. People with higher education often score well even with significant underlying impairment. For subtle change, prefer MoCA — and if you do MMSE, always add a clock drawing task.
Clock drawing hits visuospatial domains and the abstract concept of drawing time (executive). Choose a time that uses both sides of the clock (excludes hemispheric neglect / field loss). 10 past 11 is among the hardest: you must recode “10” as the 2 position, and 10 and 11 sit close together — a stimulus-response trap for executive dysfunction. Poor clocks are also visually persuasive for sceptical family.
One practical scoring approach (many exist): point for the circle; point for the right numbers; point for spacing (break into four quadrants — right numbers in each); point for hands in correct position and length. Example from the talk: two out of three when hand lengths were wrong.
MoCA deep dive
MoCA is much more sensitive for mild cognitive impairment. Training was historically encouraged and is expensive; it is not really mandatory now. Download as PDF; use different versions if testing more often than three-monthly to reduce practice effect. Multiple languages and a blind version exist. It takes a little longer until you are fluent with it.
Trail-making
Complex: visuospatial + executive + attention — all relevant to driving. A larger trail-making PDF up to 13 is helpful for impaired cognition and driving assessment. Patient joins number–letter in ascending order (1-A, 2-B…).
Serial 7s, abstraction, schooling
- Serial 7s on MoCA are only out of three (not five like MMSE) — some errors still leave a decent score.
- Abstraction: banana/orange → category “fruit”; train/bicycle → category of transport (not “both have wheels” — that is zero). Cue the first category then ask the second.
- Add +1 if ≤12 years of schooling.
Five-word memory — encoding vs retrieval
Harder than MMSE: five words, longer delay. Present about one per second; free recall twice after learning (instructions may be repeated only once). After intervening tasks, free recall: only uncued words count toward the final score. Then category cue (e.g. colour, flower), then multiple-choice if needed.
RUDAS
Especially helpful for lower education or English as a second language — more everyday tasks. Shopping list remembered with up to five repetitions; more praxis (show me your right foot; which is my left knee); animal fluency in one minute. Validated for use with an interpreter — sit the interpreter beside you so non-verbal cues reach the participant.
Other tools worth knowing
- Snellgrove maze — can help with driving assessment.
- Addenbrooke’s / ACE — more comprehensive; used a lot in specialty settings.
- Frontal Assessment Battery (FAB) — if concerned about frontotemporal dementia.
- SLUMS — sensitive for MCI (less used by this speaker).
- KICA — for Aboriginal and Torres Strait Islander populations, particularly rural communities.
Cases: Sarah and Greg
Sarah — anxiety masking early Alzheimer’s
Well-educated woman, 67, new to the area, hearing loss, significant anxiety. Increasingly anxious socially; hard to follow conversations; couldn’t remember names; had to concentrate hard on simple tasks. MMSE scored well — anxiety was treated. She worsened. Neuropsychiatry later diagnosed stage 1 Alzheimer’s. Mood disorders (anxiety or depression) often appear as a prodrome. Lesson: good history and collateral; a MoCA would have been more helpful than MMSE alone.
Greg — poor Trail B, good RUDAS, not dementia
Geriatrics registrar case: diligent GP referred Greg after a terrible Trail Making B on a driving assessment. Neither he nor his wife had cognitive concerns. Left school at year 8; lifelong builder’s labourer. Scored well on RUDAS. Conclusion: likely underlying learning disability affecting trail-making — not cognitive impairment. Right tool, plus history and examination, still matter.
Choose the right tool for the job, in the context of a good history and examination. Cognitive tools themselves are not diagnostic tests.
DTA resources
Dementia Training Australia (DTA) was flagged with a QR code in the room: podcast and further resources for clinicians wanting deeper training.
drkotha.com · cognitive-assessment.drkotha.com