Prevention & Future Directions

1. Why Prevention Is the Highest-Value Intervention

The economics and biology of AD both push us towards prevention. Plaques and tangles accumulate for ~15–25 years before clinical symptoms (Part I), and even the best disease-modifying antibodies slow decline by only ~25–35% on the CDR-Sum-of-Boxes (Part VII). Once tangles spread out of the medial temporal lobe, the cortex is structurally disorganised — no antibody can rebuild it. Prevention works on the long preclinical window where the brain still has reserve.

Formally, if onset is shifted by τ years and dementia mortality is approximately exponential in age with rate μ, then prevalence P(age) under the shifted scenario satisfies:

\( P_\tau(a) \;\approx\; P_0(a-\tau)\,\frac{S(a)}{S(a-\tau)}, \quad \frac{P_\tau}{P_0} \;\approx\; e^{-\mu\tau} \)

For typical late-life mortality (μ ≈ 0.1 yr⁻¹) and τ = 5 yr, this gives a prevalence reduction of ~40–50%, matching the Brookmeyer estimate. By contrast, halving incidence with the same shift would only reduce prevalence by ~30%. The compounding from competing risk is what makes onset delay so powerful.

The number-needed-to-treat (NNT) calculus is also favourable. For a midlife population with 10-year dementia incidence of ~5%, an intervention that produces a relative-risk reduction of 20% has:

\( \text{NNT} \;=\; \frac{1}{\text{ARR}} \;=\; \frac{1}{0.05 \times 0.20} \;=\; 100 \)

An NNT of 100 over a decade for a disease that costs >$300,000 in lifetime care dwarfs almost every other midlife preventive intervention by cost-effectiveness. And many of the prevention levers are independently beneficial: blood-pressure control, exercise, hearing aids and smoking cessation save lives from cardiovascular causes regardless of their cognitive effect.

2. The Lancet Commission Framework

The Lancet Commission on Dementia Prevention, Intervention, and Care (Livingston et al., Lancet 2017, 2020, 2024) provides the dominant policy-level synthesis. The 2024 update identifies 14 modifiable risk factors across the life course; together they account for an estimated ~45% of all dementia cases in the population (population-attributable fraction, PAF). The remaining ~55% is driven by age, genetics (most prominently APOE-ε4), sex, and as-yet unidentified factors.

PAFs are not additive when factors share variance, so the headline 45% applies the Levin formula with adjustment for joint distributions:

\( \text{PAF} \;=\; \frac{P_e (\text{RR}-1)}{1+P_e(\text{RR}-1)}, \quad \text{Joint PAF} = 1 - \prod_i (1 - \text{PAF}_i\,w_i) \)

where P_e is exposure prevalence, RR the relative risk, andw_i a communality weight. The 45% figure is a midpoint estimate and is highly dependent on the assumed RRs — for instance halving the hearing-loss RR drops the total estimate to ~40%. The point is not the precise number but the order of magnitude: roughly half of dementia is, in principle, modifiable.

The Commission’s policy framework groups interventions by life stage: education and brain-healthy upbringing in childhood; cardiovascular, metabolic, and sensory care in midlife; social engagement and environmental health in late life. The cardiovascular cluster overlaps heavily with stroke and cardiovascular physiology.

3. Cardiovascular Risk and the Brain

The midlife cardiovascular cluster — hypertension, hyperlipidaemia, diabetes, smoking, obesity — is the single most actionable lever in dementia prevention. Vascular damage and AD pathology are not independent: small-vessel disease lowers the cognitive threshold at which a given amyloid/tau burden becomes symptomatic (Snowdon Nun Study, JAMA 1997 — lacunar infarcts dramatically increased risk of clinical dementia at any plaque load), and hypoperfusion may directly impair Aβ clearance via perivascular and glymphatic routes (Section 8).

Hypertension — the strongest midlife signal

Observational data are unequivocal: midlife (age 45–65) systolic BP >140 mmHg roughly doubles late-life dementia risk independent of stroke (HUNT, Whitehall II, ARIC, Honolulu–Asia Aging Study). Late-life BP associations are messier — in some cohorts low BP in the very old predicts dementia (reverse causality from frailty/preclinical disease).

Mechanistically, intensive BP control reduces white-matter hyperintensity progression on MRI (SPRINT-MIND MRI sub-study, JAMA 2019), and post-hoc analyses suggest particular benefit from agents crossing the BBB (ARBs, certain ACE-Is) over those that do not — though no head-to-head RCT has confirmed an antihypertensive class effect for cognition. See Pharmacology for the full antihypertensive armamentarium and Cardiovascular Physiology for the underlying haemodynamics.

Lipids, diabetes, and the metabolic cluster

• Statins — observational data show modest dementia-incidence reductions, but RCTs (HPS, PROSPER) have not confirmed this for cognitive endpoints. Reasonable to use for vascular indication; not yet justified for cognition alone.
• Diabetes — ~1.5–2× dementia risk; intensive glycaemic control (ACCORD-MIND) did not improve cognition beyond standard care, but newer agents (GLP-1, SGLT2) are under active study.
• Smoking — ~1.3× risk; the gradient is reversible — ex-smokers’ risk approaches that of never-smokers within ~10 years.
• Atrial fibrillation — independent dementia risk factor; anticoagulation for stroke prevention also reduces dementia incidence in observational cohorts.

The clinical implication is clear: treat midlife vascular risk aggressively. The brain is just another vascular end-organ, and the same evidence base that supports BP control for stroke and myocardial infarction supports it for dementia — with cognitive endpoints now confirmed in SPRINT-MIND.

4. The FINGER Trial and Multidomain Prevention FLAGSHIP

FINGER (Finnish Geriatric Intervention Study to Prevent Cognitive Impairment and Disability; Ngandu et al., Lancet 2015) is the landmark trial of multidomain lifestyle intervention for dementia prevention — the first RCT to show that a structured combined-lifestyle bundle improves cognition in at-risk older adults.

Design

• N = 1,260 Finnish adults age 60–77 with elevated CAIDE dementia risk score and cognition at or below age-norm.
• 2-year randomised, controlled, single-blind trial.
• Intervention arm: structured multidomain bundle (diet + exercise + cognitive training + cardiovascular monitoring).
• Control arm: regular health advice (active control).
• Primary endpoint: change in modified Neuropsychological Test Battery (NTB) total score at 24 months.

Results

• +25% improvement in global cognition (NTB total) in intervention vs control (β = 0.022 / yr, 95% CI 0.002–0.042; p = 0.030).
• +83% improvement in executive function.
• +150% improvement in processing speed.
• Memory subdomain showed a non-significant trend.
• Effects were larger in APOE-ε4 carriers in pre-specified subgroup analysis — the genetically high-risk benefited most from lifestyle modification.

FINGER 2.0 and World-Wide FINGERS

FINGER’s bundle is being replicated in >55 countries under the World-Wide FINGERS consortium (Kivipelto, Mangialasche, Solomon). U.S. POINTER (JAMA 2025) tested a culturally adapted version in 2,111 ethnically diverse American adults and showed similar cognitive benefit, validating the cross-cultural generalisability. FINGER 2.0 extends the design to a 5-year endpoint and embeds biomarker readouts (plasma p-tau217, MRI volumetry) to ask whether multidomain lifestyle modifies the AD biological signature — not just performance.

The clinical implication: a structured, supervised multidomain programme is now evidence-based for at-risk older adults. The remaining policy question is how to deliver it at scale outside trial conditions — primary care, community centres, digital coaching.

5. Diet

Diet is the most heavily studied non-pharmacological exposure in dementia epidemiology and the most overstated. Observational signals are robust but confounded by socioeconomic and behavioural co-variates; RCTs are sparse and short. The pattern that consistently emerges is plant-forward, fish-inclusive, low-saturated-fat, low-ultraprocessed.

Mediterranean diet

Vegetables, legumes, fruits, whole grains, nuts, olive oil; moderate fish, poultry, dairy; minimal red and processed meat; moderate red wine. The PREDIMED trial (Estruch et al., NEJM 2013, 2018) was a 7,447-person RCT in Spain testing Mediterranean + extra-virgin olive oil (~50 g/d) or Mediterranean + nuts (~30 g/d) vs low-fat control. The primary CV-event endpoint was reduced by ~30%; in the PREDIMED-Navarra cognitive sub-study (Martinez-Lapiscina, J Neurol Neurosurg Psychiatry 2013) the Mediterranean arms showed better composite cognition than control over 6.5 yr (effect sizes 0.18 SD on MMSE-equivalent; modest but consistent).

MIND diet

The MIND diet (Mediterranean–DASH Intervention for Neurodegenerative Delay; Morris et al., Alzheimers Dement 2015) emphasises 10 brain-protective food groups (green leafy vegetables, other vegetables, nuts, berries, beans, whole grains, fish, poultry, olive oil, wine) and limits 5 unhealthy ones. Observational data showed dose-response cognitive protection: high adherence (top tertile) reduced AD incidence by ~53% vs low adherence (HR 0.47).

The MIND-RCT (Barnes et al., NEJM 2023), however, was a 3-year RCT in 604 older adults at increased dementia risk and was negative on its primary endpoint — both MIND and control (mild calorie restriction with usual diet) showed similar small cognitive improvements over 3 yr. The most plausible interpretation is that 3 years is too short, the control diet was already healthy, and observational diet effects accrue over decades not years. Mediterranean and MIND patterns remain reasonable defaults.

Omega-3 / fish oil

Observational data link higher DHA/EPA intake (~2 fish meals/wk) to lower dementia incidence. RCT supplementation is mixed: AREDS2 was negative for cognition; MAPT showed signal only in APOE-ε4 carriers. Plausibly the dose, source (whole fish vs capsule), and timing (midlife vs late-life initiation) matter. Whole-fish consumption is recommended; capsule supplementation is not strongly supported by RCTs.

See Metabolism for the full nutritional and metabolic-pathway context.

Alcohol

The historical “J-shaped” observational curve (light drinkers fare best) has largely collapsed under Mendelian-randomisation analyses (Topiwala, Lancet 2017 MRI study; Anstey 2024 systematic review). Current evidence supports a monotonic dose-response: any alcohol increases dementia risk to some degree, with steep increase above ~14 units/wk. The Lancet Commission threshold of “>21 units/wk” is a policy compromise, not a safe lower limit.

6. Physical Exercise

Physical activity is the single behavioural exposure with the most robust dose-response cognitive evidence. Meta-analyses (Northey 2018, Erickson 2019) of ~50 trials show moderate effects on global cognition (g ~0.2–0.3) for both aerobic and resistance training, with combined regimens superior to either alone. The dose threshold appears to be ~150 min/week of moderate-intensity activity — the same as the cardiovascular guideline.

Aerobic exercise

Erickson et al. (PNAS 2011) randomised 120 sedentary older adults to 1 yr of walking (~3×/wk, 40 min) vs stretching control. The walking group showed a 2% increase in hippocampal volume while controls showed the expected ~1.4% age-related decline — a net effect of ~3.4%, equivalent to reversing 1–2 years of brain aging. BDNF rose proportionally to volume change.

Resistance training

The SMART trial (Mavros, J Am Geriatr Soc 2017) and SPRInG trial showed that progressive resistance training improved executive function and reduced white-matter lesion progression in older adults with MCI. Effects scaled with intensity (60–80% 1RM) and persisted at 18-month follow-up. The mechanism likely involves IGF-1 and direct effects on neurogenesis.

The BDNF mechanism

Brain-derived neurotrophic factor (BDNF) is the canonical mediator. Exercise induces BDNF expression via PGC-1α → FNDC5 → irisin → BDNF (Wrann et al., Cell Metab 2013) and via lactate–HCAR1 signalling. BDNF acts on TrkB receptors to promote synaptic plasticity, dendritic spine formation, and adult hippocampal neurogenesis. APOE-ε4 carriers may have impaired BDNF response, which makes the consistent observational finding that exercise benefits ε4 carriers more (not less) somewhat counterintuitive — possibly reflecting greater room for improvement.

\( \text{Exercise} \;\xrightarrow{\text{PGC-1}\alpha}\; \text{FNDC5} \;\to\; \text{irisin} \;\to\; \text{BDNF}\,/\,\text{TrkB} \;\to\; \text{synaptic plasticity} \)

Practical prescription: 150 min/wk moderate-intensity aerobic + 2×/wk resistance training is the consensus, with stronger evidence at 300 min/wk if tolerated. The intervention is essentially free of harm in screened populations and produces independent cardiovascular and metabolic benefits.

7. Cognitive Reserve and Education

Cognitive reserve (Stern, Lancet Neurol 2012, 2020) is the framework that explains why two patients with identical pathology can have very different clinical presentations. It is operationally defined as the gap between expected and observed cognitive performance for a given pathological burden — the brain’s ability to maintain function despite damage by recruiting alternative networks or operating its primary networks more efficiently.

Three constructs are distinguished:

• Brain reserve — structural: total brain volume, neuron number, synapse count. Largely fixed by adulthood but eroded by pathology.
• Cognitive reserve — functional: efficiency and adaptability of network use, modulated by life experience.
• Brain maintenance — the capacity to resist age- and pathology-related change at the molecular level.

Education and occupation

Each additional year of education reduces dementia incidence by ~7% in meta-analysis. The Nun Study (Snowdon, JAMA 1997) famously showed that linguistic complexity in 22-year-old autobiographies predicted AD pathology and clinical dementia 60 years later — idea density at young adulthood was a powerful proxy for lifelong cognitive richness. Cognitively complex occupations (e.g., physicians, mathematicians, teachers) confer additional protection beyond formal education years.

Bilingualism and engagement

Bialystok’s Toronto cohort showed lifelong bilinguals presented with AD symptoms ~4–5 yr later than matched monolinguals despite similar clinical severity at presentation — consistent with greater compensatory capacity at a given pathology load. Engagement-style activities (reading, music, games, social interaction) consistently associate with reduced incident dementia in cohorts spanning 20–30 yr.

Limits and caveats

Cognitive reserve delays clinical presentation but, paradoxically, accelerates decline once symptoms emerge: the high-reserve patient has accumulated more pathology by the time the network can no longer compensate, and decompensates faster. Reserve is also poorly modifiable in late life — most observational evidence concerns childhood education and lifelong cognitive engagement, not short-term “brain-training” apps. The ACTIVE trial (Rebok, J Am Geriatr Soc 2014) showed targeted speed-of-processing training reduced 10-yr dementia incidence by ~29%, but most commercial brain-training products show domain-specific improvement without far transfer to global cognition.

8. Sleep, the Glymphatic System, and Aβ Clearance

Sleep emerged as a major dementia risk axis after the discovery of the glymphatic system: the perivascular CSF–ISF exchange pathway through which solutes including Aβ are cleared from the brain interstitium (Iliff et al., Sci Transl Med 2012; Nedergaard lab). Glymphatic flow is gated by aquaporin-4 (AQP4) channels on astrocytic endfeet and is dramatically upregulated during sleep — in mice, the interstitial space expands by ~60% during sleep, allowing CSF–ISF convective exchange to clear Aβ ~2× faster than during wake (Xie et al., Science 2013).

\( J_{\text{clearance}} \;\propto\; \Delta P_{\text{perivascular}} \cdot k_{\text{AQP4}} \cdot V_{\text{ISF}}, \quad V_{\text{ISF}}^{\text{sleep}} \approx 1.6 \, V_{\text{ISF}}^{\text{wake}} \)

The clinical consequence is that sleep restriction acutely raises CSF Aβ42 in humans (Ju, JAMA Neurol 2017; Lucey et al., Sci Transl Med 2018: one night of sleep deprivation increased CSF Aβ by ~30%) and chronic poor sleep is associated with greater amyloid-PET burden on cross-sectional and longitudinal studies (Spira; Sprecher; Mander). The dose-response is U-shaped: both <6 hr and >9 hr predict elevated dementia risk in cohort studies, though long-sleep associations may reflect reverse causality (preclinical disease → longer sleep).

Obstructive sleep apnoea (OSA)

OSA produces intermittent hypoxia, sleep fragmentation, and disrupted slow-wave sleep — all plausibly damaging to the glymphatic system. Untreated OSA is associated with ~26% increased dementia risk in meta-analysis (Bubu, Sleep 2017), and CPAP treatment in observational cohorts attenuates this. RCT evidence for CPAP and cognition is mixed; the SAVE trial in CV outcomes did not show cognitive benefit, but adherence was poor. Practical recommendation: screen and treat OSA in any cognitively concerning older adult.

Slow-wave sleep and orexin

Slow-wave (NREM) sleep is the phase of maximum glymphatic flow. Aging selectively erodes slow-wave activity, and pharmacological enhancement (e.g., suvorexant orexin antagonists) is under early investigation as an Aβ-clearance strategy (Lucey lab pilot data 2023). Whether this translates to disease modification remains untested.

9. Hearing Loss Treatment FLAGSHIP

Hearing loss is the single largest population-attributable midlife dementia risk factor (PAF ~7% in the Lancet 2024 update). Each 10 dB of hearing loss associates with ~20% increased dementia risk (Lin et al., JAMA Intern Med 2011, 2013). The biological hypothesis is multipronged: sensory deprivation accelerates auditory-cortex atrophy, social withdrawal reduces cognitive engagement, and the increased cognitive load of degraded auditory processing steals resources from other networks.

ACHIEVE is a milestone for three reasons:

• Effect size — a 48% relative reduction in 3-yr cognitive decline is larger than any anti-amyloid antibody on the same timeframe, achieved by a non-pharmacologic, non-invasive intervention with low cost and zero ARIA risk.
• Modifiability — it shifts hearing loss from “risk factor” to “treatable risk factor.”
• Targeting — effect concentrated in the at-risk subset suggests population-level deployment may need risk stratification.

The cost-effectiveness implications are profound. Hearing aids cost ~$1,000–$5,000; over-the-counter devices (FDA approval October 2022) bring costs to ~$200–$800. If the ARIC effect generalises even partially, the NNT to prevent one MCI/dementia event over 5 yr in at-risk older adults is plausibly ~10–20 — the most cost-effective intervention in the entire AD pipeline.

Practical recommendation: screen all adults aged ≥50 for hearing loss; treat clinically significant loss aggressively; remove cost and access barriers via over-the-counter availability. The ACHIEVE-2 follow-up will extend the endpoint and population.

10. Air Pollution and Brain Health

Particulate matter <2.5 μm in diameter (PM_2.5) crosses the olfactory epithelium and the lung–blood barrier, reaching the brain via direct and systemic routes. Cohort studies in the U.S. (WHIMS-MRI), UK (Biobank), Sweden (Betula), and Mexico City consistently link long-term PM_2.5 exposure to accelerated cognitive decline, smaller brain volumes (especially hippocampal), and increased incident dementia (Chen, Lancet 2017; Wilker, BMJ 2023 meta-analysis). The Lancet Commission added air pollution to its modifiable list in 2020.

Mechanistic findings include:

• Direct neuroinflammation — PM_2.5 activates microglia and elevates IL-6, TNF-α in CSF.
• Aβ deposition — Mexico City autopsy series (Calderón-Garcidueñas) show Aβ plaques and hyperphosphorylated tau in children chronically exposed to high pollution.
• Vascular damage — endothelial dysfunction, BBB disruption, accelerated atherosclerosis.
• Olfactory route — particles deposited in olfactory mucosa translocate via olfactory nerve to entorhinal cortex — the same region first affected by AD.

The dose-response is approximately linear above background:

\( \text{HR}_{\text{dementia}} \;\approx\; \exp\bigl[\beta \cdot (\text{PM}_{2.5} - \text{PM}_{2.5,\text{ref}})\bigr], \quad \text{HR per 2 } \mu\text{g/m}^3 \approx 1.04 \)

Mitigating strategies operate at three levels: policy (clean-air regulations, electric-vehicle adoption, industrial limits — the U.S. Clean Air Act has been associated with measurable cognitive benefits in cohort re-analyses); environmental (HEPA filtration in indoor spaces lowers indoor PM_2.5 by ~50%); and individual (avoiding peak-traffic exposure, mask use during pollution events).

The clinical implication is that public-health intervention on air quality is also a dementia-prevention intervention, and the cognitive benefits of clean-air regulations are likely to be undervalued in cost-benefit analyses that focus only on cardiovascular and respiratory outcomes.

11. Future Directions

The next decade of AD therapeutics will be defined by four shifts: from dementia-stage to preclinical-stage trials, from amyloid-only to multi-target approaches, from population-average dosing to genotype-stratified prescribing, and from delivery-limited (IV) to systemic (oral, subcutaneous) modalities.

Primary prevention with anti-amyloid agents

• A4 study (Sperling) — solanezumab in cognitively normal amyloid-PET-positive older adults; reported 2023, primary endpoint negative but informative on natural history of preclinical AD.
• AHEAD 3-45 — lecanemab in preclinical (3) and intermediate-amyloid (45) cohorts; results due ~2027–2028.
• Generation S1 / S2 (CAD106 vaccine, BACE inhibitor) — primary prevention in APOE-ε4 homozygotes; halted on safety/futility but framework continues.
• DIAN-TU — gantenerumab, solanezumab, lecanemab in autosomal-dominant FAD mutation carriers, pre-symptomatic.

Gene therapy for APOE

The APOE-ε2 allele is protective (~50% risk reduction per copy vs ε3); the Christchurch mutation (APOE3-R136S) in a famous PSEN1-E280A kindred from Colombia delayed dementia onset by ~30 yr despite high amyloid burden (Arboleda-Velasquez, Nat Med 2019). Lexeo Therapeutics and others are pursuing AAV-mediated APOE-ε2 vector delivery to convert ε4 carriers to a functional ε2-like phenotype. The first intracisternal AAV2/APOE-ε2 trial (LX1001) entered Phase 1 in 2022. Safety, durability, and immunogenicity remain to be established.

Tau immunotherapy and ASOs

Anti-tau antibodies have been disappointing in AD (gosuranemab, semorinemab, tilavonemab all negative in Phase 2/3) but remain interesting in 4R tauopathies (PSP, CBD). Antisense oligonucleotides (ASOs) targeting MAPT mRNA (BIIB080) showed a ~50% reduction in CSF total tau in early studies and entered Phase 2 in 2023. ASOs may succeed where antibodies have failed because they reduce total tau production rather than clearing extant tangles — a more upstream intervention.

Microglial modulators — TREM2 agonists

TREM2 loss-of-function variants (R47H) increase AD risk ~3× (Part V). AL002 (Alector/AbbVie) is a humanised TREM2-agonist antibody designed to enhance microglial Aβ phagocytosis and the disease-associated microglial (DAM) protective phenotype. The INVOKE-2 Phase 2 trial in early AD reported in late 2024 (mixed results, not yet replicating the CDR-SB benefit of anti-amyloid antibodies). Other targets include CD33 (variant ATM-reducing) and CSF1R.

GLP-1 agonists

GLP-1 receptor agonists (semaglutide, liraglutide, tirzepatide) cross the BBB and have neuroprotective effects in preclinical AD models — reducing Aβ, suppressing neuroinflammation, and improving insulin signalling. EVOKE / EVOKE+ are 1,840-patient Phase 3 trials of oral semaglutide in early AD (read-outs ~2026). Observational data from diabetes cohorts already show ~30–50% reduced dementia incidence with GLP-1 use vs other glucose-lowering classes. If positive, semaglutide would become the first orally bioavailable disease-modifying therapy — a major access shift. See Pharmacology for full GLP-1 agonist pharmacology.

Other emerging directions

• Anti-inflammatory — NLRP3 inhibitors, IL-6 blockade.
• Synaptic / neurotrophic — sigma-1 agonists, BDNF mimetics, fornix DBS (ADvance trials).
• Mitochondrial — CoQ10 derivatives, NAD+ precursors.
• Senolytics — dasatinib + quercetin Phase 1 reported in 2024.
• Digital therapeutics — FDA-cleared cognitive-training and remote-monitoring platforms.
• Plasma p-tau217 — will dominate routine AD diagnosis by ~2027 (Part VI).

12. Living with the Diagnosis

The clinician’s temptation in a course on biology is to leave the patient at the door of the trial. That is wrong. The lived experience of AD — for the person diagnosed, for their family, and for the carers who absorb most of the global cost — is the part of the disease our therapeutics still touch least. Three components matter:

The patient perspective

Most patients with mild AD retain insight, agency, and emotional life for years. The early-disease tasks are: financial and legal planning while capacity is intact (durable power of attorney, advance directive, will), driving review, and identification of preferences for future care intensity. Books such as Still Alice (Genova, 2007) and The Long Hello (Lloyd-Williams) do more for clinical empathy than most journal articles.

The caregiver burden

~80% of AD care is delivered by unpaid family caregivers, predominantly middle-aged daughters and spouses. Caregiver depression rates exceed 30%; caregiver mortality is elevated (Schulz, JAMA 1999). Effective interventions include respite care, structured psychoeducation (REACH II, NYU Caregiver Intervention), and pharmacotherapy for sleep and depression. The caregiver is the patient’s immune system: protect them.

Advance directives and end-of-life

Conversations should be initiated at MCI or mild dementia, before capacity erodes. Key decisions: artificial nutrition / PEG tubes (no survival benefit in advanced AD — AGS guidelines); hospitalisation thresholds; antibiotic use; CPR status. The American Geriatrics Society and the Alzheimer’s Association recommend hospice for FAST stage 7 (loss of speech, ambulation, continence; median survival ~6 months) — and yet AD remains under-referred to hospice relative to malignancy of comparable prognosis.

Comfort-focused care in late AD includes: management of agitation (non-pharmacologic first; consider trazodone, low-dose atypical antipsychotic only with informed consent given mortality signal); pain treatment (under-recognised in non-verbal patients; PAINAD scale); aspiration-pneumonia management with comfort-focused goals; and bereavement support for family. The Conversation Project and Five Wishes are useful tools.

Where to Next

This is the final part of the course. From here you can return to the start, jump to related material, or explore other Courseshub courses that intersect heavily with AD prevention.