Grades are shown as the source states them. GRADE: 1 = strong, 2 = weak recommendation; quality of evidence A (high) to C (low). The source also gives the percentage of experts who endorsed each recommendation.
Key recommendations
Do not use cognitive tests to screen adults without symptoms for MCI or dementia — including those with risk factors such as a family history or vascular risk.
1CTable 3
Be alert to early signs in older people: cognitive complaints from the person or an informant, unexplained decline in instrumental activities, missed appointments, trouble following instructions or taking medicines, reduced self-care, falling for financial scams, or new depression or anxiety in later life. If concerned, assess cognition, daily function and neuropsychiatric symptoms with validated tools.
1C; 1ATable 3
A report from a reliable informant is an essential part of assessing suspected neurocognitive disorders in every setting.
1CTable 3 (informant)
Never base a diagnosis of MCI or dementia solely on an impaired cognitive screening result.
1BTable 3 (in-depth assessment)
Minimise exposure to strongly anticholinergic medicines in older people, using alternatives for indications such as depression, neuropathic pain and urge incontinence.
1BTable 6 (medications)
Diagnosis
Distinguish MCI from dementia by clinical assessment of cognition and function, exploring the complaint with the patient and a family member.
1ATable 3
For a rapid objective check, use a brief tool such as the Mini-Cog, the Memory Impairment Screen with clock drawing, the AD8, the GPCOG or a four-item version of the MoCA.
2BTable 3
With more time, use a fuller screen (3MS, MMSE or RUDAS). The MoCA is more sensitive to MCI and is preferred when MCI is suspected or when the MMSE is in the normal range (24 or more).
1BTable 3
Reserve the biological (ATN biomarker) definition of Alzheimer's disease for research, not routine care; amyloid or tau imaging is discouraged outside research in people without memory decline.
1B; 1ATable 1
Use standardised criteria (for example VASCOG or DSM-5) to diagnose vascular MCI and vascular dementia.
1CTable 2
Investigations
Structural brain imaging is recommended in most situations — including symptom onset within the past 2 years, unexpected decline, recent significant head injury, new neurological signs, cancer with risk of brain metastases, bleeding risk, features of normal-pressure hydrocephalus, or significant vascular risk factors.
1CTable 4
MRI is preferred to CT, especially for vascular lesions. If CT is used, a non-contrast scan with coronal reformats helps assess hippocampal atrophy.
2C; 1CTable 4
FDG-PET (or SPECT perfusion imaging if PET is not practical) helps differential diagnosis when the cause remains unclear after specialist evaluation; amyloid PET should be ordered only by dementia experts.
1A; 1BTable 4
CSF analysis is not routine. Consider it when the diagnosis is uncertain and onset was before 65, or when language, visuospatial, dysexecutive or behavioural features predominate.
1CTable 4
For persistent subjective cognitive complaints with normal testing and intact daily function, carry out the standard work-up for reversible causes and assess for depression and anxiety.
1BTable 3 (subjective cognitive decline)
Treatment
Recommend exercise (group or individual) for people living with dementia.
1BTable 7
Consider group cognitive stimulation therapy for mild to moderate dementia.
2BTable 7
Consider psychosocial and psychoeducational interventions for caregivers, and case management for people living with dementia.
2C; 2BTable 7
Cholinesterase inhibitors and memantine may be considered for vascular cognitive impairment in selected patients.
2BTable 2
Where a vascular contribution is known or suspected, strongly consider antihypertensive treatment for average blood pressure of 140/90 mmHg or above.
1BTable 2
Aspirin is not recommended for MCI or dementia when imaging shows only white-matter lesions and there is no history of stroke or brain infarcts.
2CTable 2
After more than 12 months on a cholinesterase inhibitor or memantine, consider stopping if there has been meaningful worsening over 6 months with no other cause, no benefit at any point, severe or end-stage dementia, intolerable side effects, or adherence too poor for safe use. Stop them in MCI and when prescribed for other conditions such as frontotemporal dementia.
1B; 1CTable 8
Do not stop a cholinesterase inhibitor while clinically meaningful psychosis, agitation or aggression persists, unless the drug appears to have worsened those symptoms.
2BTable 8
Important geriatric considerations
Risk reduction: recommend a Mediterranean-style diet and physical activity of at least moderate intensity; ask about difficulty hearing in everyday life and confirm suspected hearing loss with audiometry.
1B; 1ATable 6
Take a sleep history that includes symptoms of sleep apnoea; adults with sleep apnoea should be treated with CPAP.
1CTable 6
Assess frailty as a marker of future dementia, and manage frailty to reduce the burden of dementia.
1BTables 5–6
Measure gait speed in people with cognitive complaints when time allows: slow gait (below 0.8 m/s) together with cognitive impairment signals higher risk of dementia.
1BTable 5
Assess caregiver burden regularly — it is a major determinant of hospitalisation and nursing-home placement.
1CTable 3 (follow-up)
Review people with dementia typically every 6–12 months, with a reliable informant; cover cognition, function, behaviour and caregiver burden at least once a year.
1CTable 3 (follow-up)
Drug doses
Deprescribing a cholinesterase inhibitor or memantine: reduce the dose by 50% every 4 weeks down to the starting dose, then stop after 4 weeks at the starting dose. Restart if clinically meaningful worsening follows.
2C; 1BTable 8
When to refer
Consider memory-clinic referral for sustained neuropsychiatric symptoms that start in later life, and sleep-clinic referral when the history suggests a sleep disorder.
2B; 1CTable 5
Refer for audiometry when hearing loss is reported, and to ENT for chronic otitis media or a failed otoscopic examination.
1ATable 6
Red flags
A first psychiatric episode in later life (psychosis, anxiety, depression or mania) should raise suspicion of a neurocognitive disorder.
1BTable 5
Unexpected, unexplained decline in someone already known to have dementia is an indication for brain imaging.
1CTable 4
Recent updates
2020: the 5th consensus conference addressed risk reduction and, for the first time, convened a working group on psychosocial and non-pharmacological interventions; it also added recommendations on deprescribing dementia medicines.
Sections 3.6–3.8