Delirium is the most common cause of acute cognitive change in hospitalised older patients (affecting approximately 30% of older inpatients) and is frequently misdiagnosed as dementia in both hospital and primary care settings. The key distinction is the time course: dementia develops over months to years; delirium develops over hours to days. The confusion in delirium is characterised by: fluctuating consciousness (worse at night — 'sundowning'), inattention (the hallmark — patient cannot focus or follow a conversation), and disorientation. Any elderly patient who was previously cognitively intact (or had mild dementia) and develops acute confusion should be assumed to have delirium until a precipitant is identified and treated — not assumed to have worsened dementia. The most common delirium precipitants in the community: urinary tract infection (but ASB in elderly should not be assumed to cause confusion without specific urinary symptoms), chest infection, constipation/faecal impaction, acute pain (unrecognised — especially hip fracture), dehydration, medication changes (especially anticholinergics, opioids, benzodiazepines), and urinary retention. GPs who identify delirium should always ask: 'What is different today compared to yesterday?' — this question identifies the precipitant in the majority of cases.

Lewy body dementia (LBD) is the most important dementia subtype for GPs to recognise correctly because of the potentially fatal adverse effect of antipsychotics in this condition — approximately 50% of patients with LBD develop severe neuroleptic sensitivity reactions when given antipsychotics (including haloperidol, chlorpromazine, and risperidone), characterised by: acute Parkinsonism, profound sedation, autonomic instability, and sometimes death. The mortality from neuroleptic sensitivity in LBD is approximately 3× that of antipsychotic use in Alzheimer's disease. The clinical features of LBD that allow diagnosis before antipsychotics are inadvertently prescribed: (1) fluctuating cognition (lucid periods alternating with confusion — may appear to have delirium); (2) persistent, well-formed visual hallucinations (often of small animals or people — the patient may describe them clearly and without distress); (3) features of parkinsonism (at least one of tremor, rigidity, bradykinesia); (4) REM sleep behaviour disorder (acting out dreams — punching/kicking in sleep — precedes LBD by years). Any patient with these features who requires sedation should be managed with non-antipsychotic agents (benzodiazepines, small doses of quetiapine — the lowest-risk antipsychotic for LBD). Document LBD prominently as a contraindication to standard antipsychotics in the medical record.

The MoCA has largely superseded the MMSE in primary care for dementia screening because it has significantly better sensitivity for mild cognitive impairment — the MMSE was designed to screen for moderate-severe dementia and misses approximately 30% of mild dementia cases (patients with MCI or early dementia often score 27–30 on MMSE, falsely reassuring). The MoCA has a 90% sensitivity for MCI vs 18% for MMSE. However, MoCA scores are affected by education level — the original MoCA study showed that one point should be added to the total score for patients with ≤12 years of education. The standardised MoCA administration is important: quiet room, patient wearing hearing aid and glasses if required, no coaching, specific instructions read as written on the administration sheet. GPs administering the MoCA should be trained (free online certification at www.mocatest.org). The most important collateral history question for detecting early dementia in primary care: 'Does your family member ever ask the same question or tell the same story twice in the same conversation?' — this autobiographical repetition is highly specific for early Alzheimer's disease and is the most sensitive single-question discriminator.

The distinction between dementia and depression ('pseudo-dementia') is clinically important because depression is entirely reversible while dementia is not — making it the single most important reversible cause of apparent cognitive decline. Features suggesting depression rather than dementia: patient complains of memory problems (patients with early dementia typically underestimate their deficit, while depressed patients overestimate it); prominent depressed mood, anhedonia, early morning wakening, reduced appetite; neuropsychological testing shows better performance with encouragement and cues (whereas true dementia shows inconsistent cuing response); cognitive symptoms parallel mood symptoms (both worsen and improve together); history of previous depressive episodes; onset in context of identifiable life stressor. The clinical rule: treat depression first in any patient where depression is a plausible explanation for cognitive symptoms — use PHQ-9 to confirm and track, start antidepressant ± IAPT, and reassess cognition at 3 months. A patient who was 'demented' before treatment and cognitively normal after antidepressant treatment had pseudodementia, not dementia.

The DVLA notification requirement for dementia is a legal obligation that GPs must communicate to patients — any person diagnosed with dementia must notify the DVLA. The DVLA will assess fitness to drive on an individual basis (there is no blanket automatic withdrawal of driving licence — driving ability is assessed through a standardised assessment and on-road driving test in many cases). The GP's role: (1) advise the patient of the notification requirement at or shortly after diagnosis; (2) document this advice in the clinical record; (3) advise cessation of driving if the patient poses a clear safety risk and has not notified the DVLA. GPs have a legal pathway for direct DVLA notification if a patient continues to drive despite clear safety concerns and has been told to stop — this must be documented and the patient must have been warned of the intention to notify. The RCGP has published guidance on this ('Protecting patients — guidance on confidentiality, disclosure and sharing information'). Power of attorney (Lasting Power of Attorney — LPA) should be discussed early in the dementia journey, while the patient still has legal capacity — once capacity is lost, a court-appointed deputy is required (more complex and expensive).

The antipsychotic prescribing caution in dementia is one of the most important patient safety issues in primary care prescribing for elderly patients — a 2009 meta-analysis showed that atypical antipsychotics (olanzapine, risperidone, quetiapine) increase mortality in elderly patients with dementia by approximately 60% compared to placebo (primarily from cardiovascular events and infections). The MHRA issued a specific warning in 2004 that risperidone is not licensed for dementia in the elderly and significantly increases stroke risk. Current NICE NG97 guidance states that antipsychotics should only be offered to people with dementia if: (1) they are at risk of harming themselves or others; (2) they are experiencing agitation, hallucinations, or delusions causing them severe distress; (3) the expected benefit outweighs the risk; (4) the decision is taken with specialist input (old age psychiatry, care of the elderly). Antipsychotics in dementia must be time-limited (review at 6 weeks, maximum 12 weeks where possible), and the indication, risks, and patient/family agreement must be documented. For LBD specifically, any antipsychotic prescription is a serious patient safety event.

Admiral Nurses (Dementia UK) are specialist nurses who provide intensive support to families of people with dementia — named after Admiral, the Labrador owned by Joseph Levy (who founded the service in memory of his father). Admiral Nurses provide: expert guidance to families on managing challenging behaviours, emotional support during the journey, help with advance care planning discussions, and clinical case management. They are increasingly embedded in primary care settings (some practices have dedicated Admiral Nurses) and community memory services. GPs should know how to refer to the local Admiral Nurse service (or to Dementia UK's helpline: 0800 888 6678) for families who are struggling with a dementia diagnosis. The value is enormous: carer breakdown is the most common precipitant of emergency hospital admission and care home placement for dementia patients — Admiral Nurse support reduces carer stress, delays care home placement, and reduces emergency hospital admissions.

The 12 modifiable risk factors for dementia identified by the Lancet Dementia Commission (Livingston et al., Lancet 2020) represent one of the most important prevention frameworks in contemporary medicine — they include factors that are directly addressable in primary care consultations. Hearing loss is the most impactful single factor (population attributable fraction 8.2% — more than any other factor) and is highly treatable with hearing aids. GPs often overlook hearing loss in consultations about memory concerns, focusing entirely on cognitive testing. A simple whisper test at 60 cm distance (or pure tone audiogram if available) should be performed at every memory concern consultation. The evidence linking untreated hearing loss to dementia risk is compelling: a 2019 meta-analysis showed that hearing aid use reduced 5-year cognitive decline by approximately 75% compared to untreated hearing loss in participants with moderate-severe hearing impairment. The mechanism is plausible: hearing loss causes social isolation, reduces cognitive stimulation (less auditory input to process), and may accelerate neurodegeneration through shared pathophysiology. This is a genuinely actionable intervention that costs nothing for the GP to screen for.

The post-delirium cognitive assessment principle is important and frequently omitted — acute delirium 'unmasks' pre-existing but previously compensated cognitive reserve deficits. Approximately 30–40% of patients who develop delirium during acute illness have underlying MCI or mild dementia that was not previously identified. After delirium resolves (typically 2–4 weeks), formal cognitive assessment should be performed — not during the delirium phase (when confusion is acute and testing is invalid) but at a follow-up appointment 6–8 weeks after the acute illness. A MoCA score below 26 at this point suggests underlying MCI or early dementia that should be investigated and monitored through the memory clinic pathway. This follow-up cognitive assessment is a quality standard that prevents the common scenario of discharging a post-delirium patient without recognising the underlying cognitive vulnerability.