Medical Treatments and Therapies for Cognitive Decline

The World Health Organization describes dementia as a syndrome caused by diseases that damage the brain; it is not an inevitable part of ageing.¹ But not every memory lapse is dementia, and not every confused person has a slowly progressive disorder. The time course changes the priority.

Rapid decline over days or weeks, a marked change after a medicine was started or stopped, fever, hallucinations with acute illness, or inability to manage food, fluids or essential treatment also warrants timely professional review. When in doubt, use local urgent-care advice. “Cognitive decline” should never become a reason to normalise a new medical emergency.

A short cognitive score documents part of the picture. It cannot reveal the cause, determine capacity or select treatment alone.

Assessment begins with the person’s history and, with permission, observations from someone who knows them. Clinicians connect the time course and affected abilities to daily function, then use examination, validated testing, medicine review, targeted laboratory tests and structural brain imaging as appropriate.⁴

1. NoticeRecord the change and concrete daily examples.
2. TriageSeparate an emergency or delirium from a gradual pattern.
3. AssessHistory, function, examination, cognition and context.
4. InvestigateTargeted tests and imaging where they can answer a question.
5. ExplainCause, uncertainty, stage and alternative explanations.
6. ChooseMatch one or more options to an agreed goal.
7. ReviewBenefit, adverse effects, burden and changed priorities.

NICE advises checking delirium, depression, sensory impairment and medicines with anticholinergic burden. A normal brief score does not rule out dementia when history and function remain concerning.⁵ Treatable contributors may worsen cognition without explaining an entire progressive syndrome.

For mild cognitive impairment, the American Academy of Neurology advises evaluating modifiable contributors, reviewing cognitively impairing medicines and monitoring cognition over time. It found no high-quality evidence that drug treatment changes MCI generally; counselling should be honest about that uncertainty.⁶

Care may combine approaches, but each needs a job. A group-average scale change, making breakfast with less help and reduced distress are different outcomes.

Activity, vascular-risk care, smoking cessation, social participation and hearing care support health. WHO’s risk-reduction guidance does not promise reversal of established dementia.⁷

Donepezil, galantamine and rivastigmine are acetylcholinesterase inhibitors; memantine works differently. They have roles in defined dementias and stages—not as general “brain boosters”—and do not remove disease.

A Cochrane review of donepezil found small average benefits over about six months, alongside more adverse events and withdrawals than placebo.⁸ A separate review found a small benefit from memantine in moderate to severe Alzheimer’s disease but no benefit in mild disease.⁹ These are averages, not forecasts for one person.

In MCI, acetylcholinesterase inhibitors have not convincingly prevented dementia or improved cognition and increase adverse effects.¹⁰ Stimulants for ADHD, wakefulness disorders or selected severe apathy do not form a generic cognitive-decline regimen.

Lecanemab and donanemab reduce brain amyloid and are authorised in several regions for selected early symptomatic Alzheimer’s disease. The FDA gave lecanemab traditional approval in July 2023 for initiation in amyloid-confirmed MCI or mild dementia due to Alzheimer’s.¹¹ These drugs do not treat normal ageing, concern alone, later dementia or another cause.

What “slowing” meant in the pivotal trials

These group differences mean modest slowing, not stopped disease or restored ability. Whether the average difference is noticeable is debated, and trials cannot forecast one person.

The current US lecanemab label reports any ARIA in 21% of treated trial participants versus 9% receiving placebo; risk was substantially higher in people with two APOE ε4 copies. It requires baseline and serial MRI, symptom-triggered assessment and genetic-risk counselling.¹⁵ Donanemab carries the same class warning and its own eligibility, titration, MRI and stopping instructions.¹⁶

EU authorisations limit both drugs to amyloid-confirmed early symptomatic Alzheimer’s in people with one or no APOE ε4 copies, with controlled access and product-specific monitoring.¹⁷¹⁸ US labels instead use APOE testing for risk information without automatic exclusion. There is no universal eligibility or MRI timetable.

In May 2025, the FDA cleared the Lumipulse G pTau217/β-amyloid 1-42 plasma ratio to aid assessment of Alzheimer-related amyloid pathology in selected symptomatic people. It is not a population-screening or stand-alone diagnostic test.¹⁹

The Alzheimer’s Association’s 2025 guideline applies after comprehensive evaluation in specialised memory care for objective impairment. Its assay-specific triage and substitution thresholds are conditional and supported by low-certainty evidence.²⁰ Tests are not interchangeable.

Age, symptoms, coexisting disease and setting change predictive value. False results can misdirect care; indeterminate zones are an expected safety feature.

More than one process may contribute. Alzheimer’s pathology can coexist with vascular injury, Lewy bodies, sleep disruption, sensory loss, mood symptoms or medicine effects; care must follow the whole person.

Behaviour is information. Agitation may express pain, fear, constipation, overstimulation, loneliness or a medicine effect. Ask what happened and which need is unmet. Medication may sometimes be needed for severe distress or risk, but should not replace assessment, proportionate support and review.

The target changes from broad engagement, to a practised task, to a personally meaningful daily activity.

A rehabilitation plan begins with an activity

1. Name the task. “Join the weekly video call without another person setting it up” is observable; “improve executive function” is not yet a daily-life goal.
2. Locate the breakdown. Does the person forget the time, lose the device, struggle with a password, misread an icon, become fatigued or need hearing support?
3. Choose a strategy. Practise one sequence, simplify the interface, use spaced retrieval, place a visible cue, reduce steps or assign a human prompt.
4. Measure transfer. Can the person do the actual activity with less help, greater confidence or fewer errors—and does that last outside the session?

Occupational, speech and language, physical or neuropsychological support can address routines, communication, mobility and strategies. Choose the team required by the goal.

Exercise supports cardiovascular health, mobility, sleep, mood and participation. In people already living with dementia, a Cochrane review found possible benefit for everyday activities but no clear cognitive effect, with substantial variation between programmes.²⁵ Prescribe activity for a real person—considering falls, heart and lung disease, pain, preferences and access—not as a guaranteed disease-modifying dose.

Supportive care is active treatment. Mood, sleep, communication, pain, activity and relationships affect daily cognition and quality of life, even when biomarkers do not change.

Sensory support reduces communication load and improves access. NICE suggests regular audiology review in dementia or MCI.²⁹ ACHIEVE found no total-population cognitive benefit over three years, though a higher-risk subgroup appeared to benefit.³⁰ Treat hearing because it matters—not as a dementia-prevention promise.

Keep the person with cognitive impairment central. Offer information in accessible pieces, allow time, ask permission before involving others and revisit choices. A diagnosis does not automatically remove decision-making ability; capacity is decision-specific and supported participation remains important even when another person has formal authority.

Computerised training can be engaging, repeatable and accessible at home. It can also become a polished way to practise the same narrow task. For people with MCI, a Cochrane review found very-low-certainty evidence from small trials and no evidence about whether training reduced future dementia incidence.³² A higher score inside an app is not proof of safer medicine use, better conversation or delayed disease.

Virtual-reality and “exergaming” programmes combine cognitive challenge with movement or immersion. Evidence is low or very low certainty, with possible benefit against inactive controls but not active treatments; studies varied and were supervised.³⁴ Tolerance, balance, vision, seizures, space and support matter.

Digital tools can still be useful as reminders, communication aids, accessible activities or rehabilitation supports. Judge them by the intended outcome, burden, data handling and failure plan. A useful assistive tool does not need to claim it rewires the brain.

1. PopulationDiagnosis, stage, biomarkers and exclusions define who was studied.
2. ComparatorPlacebo, usual care and active treatment answer different questions.
3. OutcomeBiomarker, test, daily function and quality of life are not equivalents.
4. MagnitudeFind absolute change, scale range and uncertainty.
5. BurdenHarms, withdrawals, scans, cost and care work shape value.

Trial registration separates a planned experiment from a completed result. Check intervention, sponsor, identifier, status, eligibility and posted outcomes on a registry such as ClinicalTrials.gov.³⁵ A protocol is not a positive result, and a press release is not peer review.

A 2026 Cochrane review pooled anti-amyloid trials and judged clinical effects trivial while ARIA increased; interpretation is debated because unsuccessful antibodies were combined with newer agents.³⁶ Pivotal lecanemab and donanemab trials found modest slowing in selected populations, while serious risks remain and clinical meaningfulness is contested.

A good plan is not a stack of interventions. It explains what may be happening, what matters, what each option is for, what it demands and when to reconsider.

1. Name the target.

   Diagnosis, pathology, symptom or activity; an unnamed target cannot be judged.

2. Record a baseline.

   Use a daily example alongside any scale: help needed, errors, distress or participation.

3. Set expectations.

   Could it improve, stabilise, slow average decline, compensate or reduce distress?

4. Count the load.

   Adverse effects, scans, travel, cost, practice and work transferred to others.

5. Create a safety route.

   Know which symptoms need routine, urgent or emergency contact.

6. Choose a review point.

   Agree what justifies continuing, adapting, pausing or stopping.

Consent is a process, not a signature. Explain risk accessibly, check understanding and revisit choices. APOE testing can inform ARIA risk but also reveals genetic information with implications for relatives; it deserves counselling, privacy and genuine choice.

“No medicine indicated” is not “nothing can be done.” Rehabilitation, safer routines, sensory care, symptom treatment, planning and carer support can materially change daily life and deserve serious measurement.

Medicine for cognitive change has entered a more complex era. Anti-amyloid treatments and blood biomarkers are genuine developments, but precision does not come from a novel molecule or test alone. It comes from asking a precise question, using evidence in the population where it applies, explaining absolute benefit and serious risk, and respecting the person’s priorities.

Symptom-targeting medicines, rehabilitation, psychological care, activity, sensory support and environmental adaptation can each have a place. None should be inflated into a universal cure, and non-drug care should not be demoted to an accessory for pharmaceuticals. Different approaches often address different outcomes.

Begin with urgency and diagnosis. Name the treatment goal. Separate slowing from improvement and biomarkers from lived function. Count the monitoring, cost and care work. Build in review and permission to change course. The most intelligent plan is not the most technologically impressive; it is the one that makes a meaningful difference without asking the person to surrender more safety, energy or autonomy than the possible benefit can justify.

1. Dementia.
2. Sudden confusion (delirium).
3. Symptoms of a stroke.
4. DETeCD-ADRD clinical practice guideline.
5. Dementia: assessment, management and support.
6. Practice guideline update summary: Mild cognitive impairment.
7. Risk reduction of cognitive decline and dementia.
8. Donepezil for dementia due to Alzheimer’s disease.
9. Memantine for dementia.
10. Cholinesterase inhibitors for mild cognitive impairment.
11. FDA converts lecanemab to traditional approval.
12. Lecanemab in early Alzheimer’s disease.
13. Donanemab in early symptomatic Alzheimer disease.
14. Earlier MRI monitoring with Leqembi.
15. Leqembi prescribing information.
16. Kisunla prescribing information.
17. Leqembi: European public assessment report.
18. Kisunla: European public assessment report.
19. FDA clears first blood test used to aid Alzheimer’s diagnosis.
20. Clinical practice guideline for blood-based biomarkers.
21. Revised criteria for diagnosis and staging of Alzheimer’s disease.
22. Cognitive stimulation for people with dementia.
23. Cognitive training for mild to moderate dementia.
24. Cognitive rehabilitation for mild to moderate dementia.
25. Exercise programmes for people with dementia.
26. Reminiscence therapy for dementia.
27. Music-based therapy for people with dementia.
28. Psychological treatments for depression and anxiety.
29. Hearing loss in adults: assessment and management.
30. Hearing intervention and cognitive decline.
31. START coping intervention for family carers.
32. Computerised cognitive training in MCI.
33. FDA authorises a game-based therapeutic for paediatric ADHD.
34. Exergaming for dementia or MCI.
35. ClinicalTrials.gov.
36. Anti-amyloid drugs: pooled benefit and harm review.