If someone you know has Lewy body dementia, or might have it, the single most important thing on this page is that antipsychotic drugs can cause severe and occasionally fatal reactions in this condition. That is not a general caution about side effects. It is specific to this diagnosis, it is well recognised, and the people most exposed to it are the ones who have not been diagnosed yet.
This is the fourth post in our series on the dementia diagnoses, after mild cognitive impairment, Alzheimer's disease and vascular dementia. We are not doctors, and nothing here is treatment advice.
What it is, and why it gets missed
Lewy body dementia is caused by deposits of misfolded alpha-synuclein protein — Lewy bodies — inside neurons. It is widely cited as the third commonest cause of dementia, and it is routinely mistaken for Alzheimer's, for Parkinson's, or for a psychiatric illness.
The four core features are not the ones people expect from dementia:
- Fluctuating cognition. Not a steady decline — marked variation, sometimes within a single day. Alert and conversational in the morning, vacant by afternoon. Families are often told they must be imagining it.
- Visual hallucinations. Typically well-formed and detailed — people, animals, children — and often not frightening, at least at first. These appear early, which is unusual.
- REM sleep behaviour disorder. Acting out dreams: shouting, punching, kicking, falling out of bed. This can precede any cognitive symptom by years or decades.
- Parkinsonism. Slowness, stiffness, reduced facial expression, shuffling gait, sometimes tremor.
Memory, notably, is often not the first thing to go. Attention, visuospatial ability and executive function usually suffer earlier. So a person who is seeing things, moving stiffly and varying hour to hour, but who can still remember yesterday's conversation, does not look like what most people — including many clinicians — picture when they think dementia.
The sleep disorder that arrives first
REM sleep behaviour disorder is the most powerful early warning sign of this family of diseases, and the evidence base for that is unusually strong.
The International RBD Study Group pooled prospective follow-up from 24 centres and 1,280 patients with polysomnography-confirmed isolated RBD — no parkinsonism, no dementia at baseline — followed for an average of 4.6 years, with some out to 19.1 The conversion rate to an overt neurodegenerative syndrome was 6.3% per year, and 73.5% had converted by twelve years.
That is a very high rate, and it needs stating carefully: it does not mean everyone with dream-enactment behaviour is destined for dementia. These were people with polysomnography-confirmed isolated RBD recruited into specialist sleep centres, which is a selected group. But within that group, the trajectory is clear.
The study also ranked predictors of conversion by hazard ratio: abnormal quantitative motor testing (HR 3.16), objective motor examination (3.03), loss of smell (2.62), mild cognitive impairment (1.91–2.37), erectile dysfunction (2.13), abnormal DAT scan (1.98), colour vision abnormalities (1.69), constipation (1.67), loss of REM muscle atonia (1.54) and age (1.54).
Equally useful is what did not predict anything: sex, daytime sleepiness, insomnia, restless legs, sleep apnoea, urinary problems, orthostatic symptoms, depression, anxiety, or substantia nigra ultrasound findings. A list of non-predictors is as clinically valuable as the list of predictors, and it rarely gets printed.
And the finding most relevant to this page: among all those markers, only the cognitive variables differed at baseline between people who went on to develop primary dementia versus those who developed parkinsonism. In other words, RBD plus early cognitive change points towards the Lewy body dementia path; RBD plus motor findings points towards Parkinson's. That is the fork in the road, and cognition is the signpost.
Antipsychotics: the specific danger
People with Lewy body dementia hallucinate. Hallucinations in an older person reliably prompt a prescription for an antipsychotic. And people with Lewy body dementia are extremely sensitive to antipsychotics — reactions include severe worsening of parkinsonism, profound sedation, confusion, and neuroleptic malignant syndrome, which can be fatal.
The mechanism follows directly from what antipsychotics actually do: they block dopamine D2 receptors, and in a disease that is already destroying dopaminergic neurons, that blockade lands on a system with no reserve left. Acetylcholinesterase inhibitors are the safer option for the neuropsychiatric symptoms, and review literature puts them first-line.
Here is how often that goes wrong in practice. A 2025 Swedish registry study identified 362 people registered with Lewy body dementia who filled at least one antipsychotic prescription in 2019.2 Of those, 114 (31.5%) had been given an antipsychotic as first-line treatment instead of a cholinesterase inhibitor. And within that 114, 60 people did not receive their Lewy body dementia diagnosis until after the antipsychotic had already been prescribed.
Read that last number again. For more than half of the people who got an antipsychotic first, nobody yet knew they had the condition that makes antipsychotics dangerous. The hallucinations were the presenting sign of the disease, and they were treated with the drug class the disease is most sensitive to.
The study's own conclusion is the practical one: emerging psychotic symptoms in an older person should be considered a possible manifestation of Lewy body dementia. If an older relative starts seeing things and an antipsychotic is proposed, "could this be Lewy body dementia?" is a reasonable and potentially important question to ask the prescriber. It is not a reason to stop a prescribed medication on your own — abrupt antipsychotic discontinuation carries its own risks.
The genuinely good news, which this series has not had much of
Across these four posts the treatment picture has been bleak: anti-amyloid antibodies with effects at or below what a patient would notice, and a vascular dementia drug literature that stopped in 2012. Lewy body dementia is the exception.
A meta-analysis of 17 randomised trials and 1,798 patients with Lewy body disorders found cholinesterase inhibitors significantly improved cognitive function (standardised mean difference −0.53), global function (−0.52), behavioural disturbances (−0.28) and activities of daily living (−0.28).3 Crucially, motor function did not worsen — which was the standing fear about using cholinergic drugs in a parkinsonian condition.
An SMD of −0.53 for cognition is a moderate effect. Set it beside the anti-amyloid antibodies' sub-threshold CDR-SB differences and the two-point ADAS-Cog shifts in vascular dementia, and this is the strongest drug signal anywhere in the dementias.
Donepezil in Lewy body dementia specifically: three randomised placebo-controlled trials, 312 patients on 10 mg, pooled at 12 weeks.4 MMSE improved by 1.50 points (95% CI 0.67–2.34) and the odds of a clinician-rated global improvement more than doubled (OR 2.20, 1.13–4.26) — both with acceptable heterogeneity (I² = 32.4% and 19.8%). But the neuropsychiatric outcomes in the same analysis had heterogeneity of I² = 87.2% and 67.7%, which means the trials disagreed badly on whether donepezil helps the hallucinations and behavioural symptoms. We are flagging that rather than quoting the convenient half.
The costs are real too. The 17-trial analysis found more discontinuation on cholinesterase inhibitors — all-cause RR 1.48 (number needed to harm 14), discontinuation for adverse events RR 1.59 (NNH 20), nausea RR 2.50 (NNH 13) and tremor RR 2.30 (NNH 20).3 Roughly one person in thirteen gets nausea who wouldn't have. That is a tolerable trade for a moderate cognitive benefit, and it is a trade rather than a free win.
Where we come in — and a loop this series needs to close
We sell nothing that treats or prevents Lewy body dementia. No ingredient in our range has been tested for it, and the one drug class with a real effect here is prescription-only.
But there is a connection to something we do sell, and it closes a loop we opened earlier in the medication series.
In what benzodiazepines actually do, we printed a finding that counted against us: a large observational study in which people prescribed melatonin showed a higher rate of subsequent dementia than people prescribed benzodiazepines — hazard ratio 2.09. We said at the time that this almost certainly reflected who gets prescribed melatonin rather than what melatonin does, and we named REM sleep behaviour disorder as the clearest example.
This post is where that explanation becomes concrete. Melatonin is one of the things used for RBD. RBD converts to an overt neurodegenerative syndrome at 6.3% per year, reaching 73.5% at twelve years. So a cohort of older people newly prescribed melatonin contains a meaningful number of people with a prodromal synucleinopathy — people who were already on the road to Lewy body dementia or Parkinson's before the first capsule. A study comparing them to people prescribed something else will find exactly the signal that study found, and it will have nothing to do with the drug.
That is the fourth time this pattern has turned up in our medication writing, and it is the clearest instance of it: the reason for the prescription masquerades as the effect of the prescription. We print it here because it is the honest explanation of a number that looks bad for a product we sell, and because learning to spot the pattern is worth more than any individual hazard ratio.
On what actually helps: the risk-factor evidence in modifiable dementia risk is about dementia broadly and is not specific to this subtype. If dream-enactment behaviour is what brought you here — shouting, hitting out, falling out of bed — that belongs in front of a doctor, and is worth raising specifically rather than as "trouble sleeping."
The short version
- Lewy body dementia is widely cited as the third commonest dementia and is routinely misdiagnosed.
- Core features: fluctuating cognition, early well-formed visual hallucinations, REM sleep behaviour disorder, parkinsonism. Memory is often not first.
- Across 1,280 patients at 24 centres, isolated RBD converted to an overt neurodegenerative syndrome at 6.3% per year — 73.5% by twelve years.
- Loss of smell, abnormal motor testing and early cognitive impairment raise that risk. Sex, insomnia, sleep apnoea, depression and anxiety do not predict it.
- Only cognitive markers distinguished those heading for dementia from those heading for parkinsonism.
- Antipsychotics can cause severe, occasionally fatal reactions in this condition. In a Swedish registry, 31.5% of antipsychotic users with Lewy body dementia got the antipsychotic before a cholinesterase inhibitor — and 60 of those 114 had no diagnosis at the time.
- Cholinesterase inhibitors work better here than in any other dementia: cognition SMD −0.53, global function −0.52, no motor worsening. Nausea NNH 13.
- Donepezil improved MMSE by 1.50 points, but the trials disagreed badly (I² up to 87.2%) on neuropsychiatric symptoms.
- RBD is why melatonin looks bad in observational dementia data. The indication, not the drug.
The rest of this series
Six diagnoses, each written the same way: what it is, what the evidence actually shows, and what we do and do not sell for it.
- Mild cognitive impairment: what it is and what follows
- Alzheimer's disease: what actually happens in the brain
- Vascular dementia: the one most tied to risk factors
- Frontotemporal dementia: when it isn't memory first
- Young-onset dementia: why the hard part comes after the diagnosis
This article is for informational purposes only and has not been evaluated by the FDA. It is not intended to diagnose, treat, cure, or prevent any disease. We are not doctors. Lewy body dementia requires specialist diagnosis. Nothing here is a reason to start, stop or change any prescribed medication — including an antipsychotic, which should never be stopped abruptly without medical supervision. If an older person is newly seeing things, or acting out dreams in their sleep, those are specific symptoms worth describing to a doctor in those words.
References
- Postuma RB, Iranzo A, Hu M, Högl B, Boeve BF, Manni R, Oertel WH, Arnulf I, Ferini-Strambi L, et al. "Risk and predictors of dementia and parkinsonism in idiopathic REM sleep behaviour disorder: a multicentre study." Brain, 2019, 142(3):744–759. doi:10.1093/brain/awz030
- Kindstedt J, Lövheim H, Gustafsson M. "Pharmacy Dispensing Records to Describe and Evaluate the Use of Acetylcholinesterase Inhibitors Among Swedish Antipsychotic Drug Users with Symptoms of Lewy Body Dementia." Drugs - Real World Outcomes, 2025, 12(3):383–390. doi:10.1007/s40801-025-00501-1
- Matsunaga S, Kishi T, Yasue I, Iwata N. "Cholinesterase Inhibitors for Lewy Body Disorders: A Meta-Analysis." International Journal of Neuropsychopharmacology, 2015. doi:10.1093/ijnp/pyv086
- Mori E, Ikeda M, Ohdake M. "Donepezil for dementia with Lewy bodies: meta-analysis of multicentre, randomised, double-blind, placebo-controlled phase II, III, and IV studies." Psychogeriatrics, 2024. doi:10.1111/psyg.13101