Skip to content
Umbrella Aged Care
All topics

Medicines in dementia

GPs & clinicians Aged care staff Residents & families 13 min read

Medication in dementia is an area where the gap between what is prescribed and what the evidence supports remains wide. The symptomatic cognitive medicines have modest effects. The psychotropics used for behaviour have modest effects and substantial harms. And the medicines that quietly worsen cognition — anticholinergics, benzodiazepines, sedating antihistamines — are often the ones nobody is reviewing.

Numbers in the text link to the 16 sources listed at the foot of this page.

Cholinesterase inhibitors and memantine

Donepezil, galantamine and rivastigmine are cholinesterase inhibitors, subsidised in Australia for Alzheimer’s disease. Memantine, an NMDA receptor antagonist, is subsidised for moderately severe to severe Alzheimer’s disease. Rivastigmine also has randomised evidence in Parkinson’s disease dementia — a 541-patient trial found moderate cognitive and global improvement, at the cost of more nausea, vomiting and tremor 92, and cholinesterase inhibitors are generally the more useful class in dementia with Lewy bodies.

What they do: on average, a modest symptomatic improvement or slowing of decline in cognition and global function, sustained for months to a couple of years in responders. What they do not do: modify the underlying disease, prevent progression, or help everybody. A meaningful proportion of people gain nothing measurable.

PBS subsidy in Australia requires the diagnosis to involve a specialist, and continuation beyond the initial treatment period depends on documented response. The listing structure was changed on 1 May 2023, consolidating the initial treatment phases and moving initial authority from written to telephone or online 56. Criteria change — check the current PBS listing before prescribing rather than relying on memory or a summary such as this one.

Practical points on the cognitive medicines

  • Adverse effects are dose-related and predictable. Nausea, vomiting, diarrhoea, anorexia and weight loss, vivid dreams, bradycardia and syncope. In a frail resident already losing weight, a cholinesterase inhibitor may be doing net harm.

  • Check the pulse. Bradycardia and syncope are genuine causes of falls in this group, and the association is frequently missed because the medication has been in place for years.

  • Avoid the anticholinergic own goal. Prescribing a cholinesterase inhibitor while the resident is on oxybutynin for bladder instability is pharmacologically self-defeating, and it remains a common finding.

  • Review whether it is still doing anything. In advanced dementia, where the person no longer recognises family and requires full assistance, the case for continuing is weak. Deprescribing should be considered and discussed rather than continued by default until death.

  • Withdrawal can unmask decline. If stopping, taper and monitor. Some people deteriorate noticeably and warrant recommencement — that is useful information, not a failure.

Antipsychotics: the harms are not theoretical

Antipsychotics have a modest effect on aggression and psychosis in dementia. The effect size is small. The harms are large, well replicated, and quantified.

A population-based matched cohort study of 173,910 people with dementia in English primary care found that, compared with non-use, antipsychotic use was associated with increased risk of pneumonia (hazard ratio 2.19), acute kidney injury (1.72), venous thromboembolism (1.62), stroke (1.61), fracture (1.43), myocardial infarction (1.28) and heart failure (1.27) 50. The risks were highest soon after initiation. In the first 90 days, the cumulative incidence of pneumonia was 4.48% in antipsychotic users versus 1.49% in matched non-users 50.

On mortality: a large retrospective study calculated numbers needed to harm of 26 for haloperidol, 27 for risperidone, 40 for olanzapine and 50 for quetiapine over 180 days, with a dose–response relationship 51. A 2026 systematic review and meta-analysis of 45 studies covering two million participants found a pooled mortality hazard ratio of 1.32 71.

On stroke specifically: a 2025 matched cohort study of 28,403 risperidone users with dementia found an adjusted hazard ratio for stroke of 1.28, and the increase applied to people with and without prior cardiovascular disease 52. An individual participant data meta-analysis of six randomised trials found the median onset of cerebrovascular events was 4.3 weeks and of major cardiovascular events 4.8 weeks after starting 53 — that is, harms cluster in the first month, which is exactly the period in which many prescriptions are written and then never reviewed.

Chemical restraint and the law

In Australian residential aged care, a psychotropic medicine used to influence a person’s behaviour — rather than to treat a diagnosed mental disorder, physical illness or end-of-life distress — is a chemical restraint, and it is a regulated restrictive practice. This is not a matter of clinical style; it is a legal obligation with reporting and compliance consequences.

Five restrictive practices are regulated: chemical, environmental, mechanical and physical restraint, and seclusion 54. A restrictive practice may only be used as a last resort, to prevent harm, after alternatives have been tried and documented, in the least restrictive form, and for the shortest necessary time. Informed consent must be obtained from the person or, if they lack capacity, from a restrictive practices substitute decision-maker under a defined hierarchy 54. A behaviour support plan must be in place.

The prescriber’s obligations are explicit: assess the risk of harm, try and document alternatives, discuss benefits and risks, obtain informed consent, and record all of it 5455. A telephone order for risperidone with no documentation of alternatives, no consent and no review date is not just poor medicine — it is a compliance failure that sits with both the prescriber and the provider.

Other psychotropics used in dementia — the honest position

  • Benzodiazepines. Increase falls, fractures, confusion and mortality, and produce tolerance within weeks 66. There is a narrow role in acute severe distress and in terminal care. There is no role for regular long-term use in behaviour management.

  • Antidepressants. Reasonable for genuine comorbid depression, which is common and under-treated — half of Australian residents with dementia have a recorded depression or mood disorder 1. Evidence for treating agitation itself is much weaker. Avoid tricyclics and paroxetine because of anticholinergic load; watch for hyponatraemia with SSRIs.

  • Sodium valproate. Should not be used for agitation in dementia. Trials show no benefit and clear harm, including sedation, falls and accelerated brain volume loss 8182.

  • Melatonin. Widely used for sleep–wake disturbance and generally well tolerated. Reasonable to trial; unreasonable to continue indefinitely without a benefit somebody has actually assessed.

  • Cannabinoids. Trials to date show effects on behavioural symptoms without measurable cognitive benefit, and the evidence base remains small 37. Not a first-line option, and prescribing in Australia carries its own regulatory requirements.

Anticholinergic burden — the quiet problem

Medicines with anticholinergic activity worsen cognition, precipitate delirium, and are associated in observational data with increased dementia risk. They are everywhere: oxybutynin and solifenacin for bladder, amitriptyline for pain or sleep, promethazine and other sedating antihistamines, older antipsychotics, some antiemetics, and tricyclics.

Australian general practice data covering 400 practices show progress but not enough: the proportion of patients with dementia carrying a mean daily anticholinergic cognitive burden score of 3 or more fell from 26.1% in 2013 to 19.2% in 2020 62. Nearly one in five is still too many.

The intellectually honest caveat is that Cochrane found insufficient randomised evidence to confirm that deprescribing anticholinergics improves cognitive outcomes — three small trials, very low certainty 63. That is an argument for better trials, not for continuing to prescribe oxybutynin to a person with dementia. The observational association with dementia risk, the established causal link with delirium, and the immediate harms of dry mouth, constipation, urinary retention and falls are sufficient reason to reduce burden where a safer alternative exists.

A practical deprescribing sequence for a resident with dementia

  1. Start with the anticholinergics. Calculate the burden formally. Replace or stop the worst offenders — oxybutynin, amitriptyline, promethazine — first.

  2. Then the benzodiazepines and Z-drugs. Taper slowly. Abrupt cessation causes rebound insomnia and agitation that is then misread as treatment failure.

  3. Then antipsychotics without a current indication. Most residents on long-term antipsychotics for behaviour can be withdrawn without symptom return 66. Taper and monitor rather than stopping abruptly.

  4. Then reconsider preventive medicines. Statins, bisphosphonates and tight glycaemic or blood pressure targets have time-to-benefit horizons measured in years. In advanced dementia with a prognosis measured in months, they are burden without benefit. Hypoglycaemia and postural hypotension are immediate harms.

  5. Then review the cognitive medicines. As above — is this still achieving anything the family or staff can identify?

  6. Document the reasoning and the plan. Deprescribing without documentation looks like neglect in a file audit and like abandonment to a worried family. Deprescribing with documented reasoning is good geriatric medicine.

Sources cited on this page

  1. 1 Australian Institute of Health and Welfare. Dementia in Australia — Residential aged care. (2021–22 ACFI data: 54% of ~242,000 permanent residents had dementia.) View source
  2. 37 Kumari A, et al. Dietary bioactives in Alzheimer’s disease: a critical appraisal of clinical trials and future nutritional strategies. Nutrients. 2026.
  3. 50 Mok PLH, et al. Multiple adverse outcomes associated with antipsychotic use in people with dementia: population based matched cohort study. BMJ. 2024;385:e076268.
  4. 51 Maust DT, et al. Antipsychotics, other psychotropics, and the risk of death in patients with dementia. JAMA Psychiatry. 2015;72:438–45.
  5. 52 Choma J, et al. Risk of stroke associated with risperidone in dementia with and without comorbid cardiovascular disease: population-based matched cohort study. Br J Psychiatry. 2025.
  6. 53 Le HT, et al. Multiple adverse outcomes associated with risperidone in people with dementia: an individual participant data meta-analysis. CNS Drugs. 2026.
  7. 54 Australian Government Department of Health, Disability and Ageing. Restrictive practices in aged care — a last resort. View source
  8. 55 Aged Care Quality and Safety Commission. Strengthened Aged Care Quality Standards (commenced 1 November 2025). View source
  9. 56 Pharmaceutical Benefits Scheme. Changes to treatment phases and authority levels in medicines for the treatment of Alzheimer’s disease (effective 1 May 2023). View source
  10. 62 Bezabhe WM, et al. Trends in anticholinergic drug exposure and associated risk factors in older Australian patients with dementia. J Psychiatr Res. 2026. (19.2% of Australian patients with dementia had a mean daily ACB score ≥3 in 2020.)
  11. 63 Taylor-Rowan M, et al. Anticholinergic deprescribing interventions for reducing risk of cognitive decline or dementia in older adults. Cochrane Database Syst Rev. 2022.
  12. 66 Australian Prescriber. Supporting the appropriate use of psychotropic medicines in aged and disability care. View source
  13. 71 Thant PE, et al. The consequences of antipsychotic medication use for people living with dementia: a systematic review and meta-analysis. Front Psychiatry. 2026. (Pooled mortality HR 1.32, 95% CI 1.12–1.56.)
  14. 81 Tariot PN, et al. Chronic divalproex sodium to attenuate agitation and clinical progression of Alzheimer disease. Arch Gen Psychiatry. 2011. (n=313; no delay in agitation or psychosis; greater hippocampal and whole-brain volume loss.)
  15. 82 Fleisher AS, et al. Chronic divalproex sodium use and brain atrophy in Alzheimer disease. Neurology. 2011.
  16. 92 Emre M, et al. Rivastigmine for dementia associated with Parkinson’s disease. N Engl J Med. 2004;351:2509–18. (n=541; moderate improvement on ADAS-cog, with more nausea, vomiting and tremor.) View source

See every source cited across the dementia section →

General information only, reflecting the interpretation of Umbrella Aged Care’s GPs of the published evidence as at September 2026. It is not individual medical advice, does not create a doctor–patient relationship, and must not be used to start, stop or change any treatment. Evidence and Australian regulatory and PBS arrangements change — always confirm current advice with the treating GP, pharmacist or specialist.

Enquire about care