Three distinct diseases, one and the same job. Faced with a patient who has a major neurocognitive disorder, the physiotherapist always does the same four things: adapt the instruction to a memory that no longer holds anything, prevent falls eight times more frequent than at the same age, hold on to independence for as long as possible, and work with the carer rather than alongside them. What changes from one aetiology to the next is the way it is done. This review first sets out that common ground and its exact level of evidence, negative trials included, then what frontotemporal lobar degeneration, Alzheimer's disease and Lewy body disease each require you to do differently. Only one of these three diseases carries an immediate risk to life from an ordinary prescription: it gets a box of its own.

What exactly does a major neurocognitive disorder cover?

The vocabulary has changed, and it is not cosmetic: it has moved the question from “how much memory has been lost” to “which disease, and therefore what course of action”. That is exactly the question a physiotherapist asks before the first session.

The DSM-5, then its 2022 revision, dropped the word “dementia” in favour of major neurocognitive disorder, defined by significant cognitive decline in at least one domain, enough to compromise independence in the activities of daily living.14 The diagnosis is then broken down by aetiology: major neurocognitive disorder due to Alzheimer's disease, to frontotemporal lobar degeneration, with Lewy bodies, of vascular origin, and so on. It is this second part that matters for the session.

The Haute Autorité de Santé, in its care pathway guide of May 2018, states what practice confirms every day: “even though Lewy body disease or vascular and frontotemporal neurocognitive disorders are designated as ‘related diseases’, their warning signs and their management differ appreciably from those of Alzheimer's disease”.12 A patient with a neurocognitive disorder is therefore not interchangeable with another.

The scale of the problem, in orders of magnitude

The Global Burden of Disease estimated 57.4 million people affected worldwide in 2019 (uncertainty interval 50.4 to 65.1), and projects 152.8 million in 2050 (130.8 to 175.9). Age-standardised prevalence, on the other hand, is stable: the growth comes from ageing and demography, not from an epidemic. In Western Europe, the projected rise in the number of cases is 74 % (58 to 90). Women are 1.69 times more numerous than men among those affected.1

In France, Santé publique France puts the figure at about 1,200,000 people affected in 2014, including nearly 24,000 before the age of 65, with a frequency about twice as high in women.13 Those 24,000 young cases are the blind spot of clinical reasoning: that is where frontotemporal degeneration hides, and that is where it is first called depression or burnout.

A caveat that holds for everything that follows: pure pathologies are rare

The autopsy series of the Rush Memory and Aging Project, 141 brains, gives the most useful result in this whole chapter. Among patients with dementia, only 30 % had pure Alzheimer's disease ; 38 % combined Alzheimer's and cerebral infarcts, 12 % Alzheimer's and Lewy body pathology. In all, more than half of those with dementia carried several neuropathological diagnoses, against fewer than 20 % of those without dementia, and the presence of several pathologies multiplied the probability of being demented by 2.8 (CI 1.2 to 6.7).2

Practical consequence: a patient labelled “Alzheimer's” who shows marked fluctuations, visual hallucinations or rigidity is not necessarily mislabelled, they may be both. Reasoning by entity, which is how this article is built, is a tool for deciding what to do, not a promise of nosological purity.

What does the physiotherapist do whatever the aetiology?

Before separating the three diseases, we have to set out what does not separate. Four objectives hold, from diagnosis to the end: they depend neither on the protein involved nor on the lobe affected.

A patient with a major neurocognitive disorder never comes to rehabilitation for their memory. They come because they fall, because they can no longer get out of their armchair, because their wife can no longer walk them to the car. The request is motor and functional. What the disease changes is the channel through which you can act on it.

First objective: getting the instruction across

It is the precondition for everything else, and it is the only point where physiotherapy for neurocognitive disorder is genuinely a discipline apart. An exercise the patient does not understand, does not encode or does not accept has no effect, however sound it is biomechanically. The chapter devoted to adapting the instruction sets out the mechanisms you can use, procedural memory first among them.

Second objective: preventing falls

This is the dominant risk, and it is massive. The only prospective study to have followed patients with dementia for twelve months, split by aetiology, with an age-matched control group, measures 9,118 falls per 1,000 person-years against 1,023 in the controls, an incidence density ratio of 7.58 (CI 3.11 to 18.5).3 The chapter on falls sets out the predictors, three of which can be changed by a care action.

Third objective: holding on to functional independence

The aim is not to make progress, it is to decline more slowly. That is a difference in kind from rehabilitating a sprain, and it must be said to the patient as to the carer: success looks like stability. The Finnish FINALEX trial, 210 Alzheimer's patients living at home with their spouse as carer, is the best illustration available: every group declined over a year, but the loss on the Functional Independence Measure was 7.1 points with an individual home exercise programme, 10.3 with a group programme, and 14.4 with no programme.8 You do not climb back up the slope, you make it half as steep.

Fourth objective: working with the carer, not alongside

The carer is the only person present for the 165 hours of the week when the physiotherapist is not. In FINALEX, the most effective arm was the home arm, with the spouse in the room.8 The Cochrane review also identified, in a single trial of 40 dyads, a fall in carer burden when the carer supervises the programme themselves (mean difference 15.30 points, CI 5.87 to 24.73): fragile data, one trial only, but pointing the same way as everything else.5

The question is not whether the patient will remember the session. They will not. The question is whether their body has kept something, and whether the person who lives with them knows what to do with it tomorrow morning.

What does exercise really change, and what does it not change?

This is the most important chapter of this article, and the most uncomfortable. The literature is clear on two opposing points: exercise does very well what exercise does, and it does not do what we would like it to do.

What is established with a strong level of evidence

The systematic review most useful to the physiotherapist is the one in the Journal of Physiotherapy : 43 randomised controlled trials, 3,988 participants with mild cognitive impairment or major neurocognitive disorder, effects graded with GRADE. With a strong level of evidence, supervised exercise improves:4

• the 30-second sit-to-stand, by 2.1 repetitions (CI 0.3 to 3.9);
• step length, by 5 cm (CI 2 to 8);
• the Berg Balance Scale, by 3.6 points (CI 0.3 to 7.0);
• functional reach, by 3.9 cm (CI 2.2 to 5.5);
• the Timed Up and Go, by 1 second (CI 0 to 2);
• gait speed, by 0.13 m/s (CI 0.03 to 0.24);
• the 6-minute walk test, by 50 metres (CI 18 to 81).

The dose associated with these results is precise and reproducible in the clinic: about 60 minutes, 2 to 3 days a week, supervised. And a detail that changes the selection of patients: it is the low motor level at baseline, and not the cognitive level, that predicts the best response to training.4 In other words, a patient who is very diminished physically is not a poor candidate, they are the best one.

On the parameter that is most specifically a physiotherapy matter, walking, a double-blind randomised trial provides the clearest demonstration. Sixty-one patients averaging 81.9 years, mild to moderate neurocognitive disorder, three months of progressive strengthening and functional training two times two hours a week, against a low-intensity placebo motor programme: speed (p < 0.001), cadence (p = 0.002), step length (p = 0.008), cycle duration (p = 0.001) and double support duration (p = 0.001) all improved, with large effect sizes, Cohen's d from 0.80 to 1.27. Adherence reached 91.9 %.9

Two parameters, on the other hand, did not move in that trial: step width and temporal step variability. And it is precisely variability that is the best-established marker of fall risk. Strengthening changes gait; it does not yet change its regularity.

What is established as not being true

The idea that exercise slows cognitive decline is appealing, very widespread, and refuted by the largest trial that has tested it. The DAPA trial randomised 494 patients with mild to moderate neurocognitive disorder, 2 to 1, between four months of moderate to high intensity aerobic and strength training and usual care. At twelve months, the adjusted difference on the ADAS-Cog was 1.4 points against the exercise group (CI 0.2 to 2.6, p = 0.03), that is, slightly poorer cognition in those who had trained. The authors themselves stress that the gap is small and of uncertain clinical relevance. Physical capacity, for its part, had improved: 18.1 metres more on the 6-minute test at six weeks (CI 11.6 to 24.6). Adherence was good, with 214 of the 329 patients attending more than three quarters of the sessions: this is not a trial that failed for want of participation.6

The Cochrane review, 17 trials and 1,067 participants, points the same way on cognition: standardised mean difference of 0.43 (CI 0.05 against to 0.92 in favour), massive heterogeneity at 80 %, quality of evidence judged very low. It does find a signal on activities of daily living, standardised mean difference of 0.68 (CI 0.08 to 1.27), but with the same heterogeneity and the same very low quality. No effect on neuropsychiatric symptoms or on depression.5

Lastly, the British PrAISED trial, published in 2023, is the hardest to take because it tested exactly what this article recommends. Three hundred and sixty-five people with early neurocognitive disorder or mild cognitive impairment, a home rehabilitation programme designed specifically for dementia, covering strength, balance, physical activity and the activities of daily living, up to 50 sessions over twelve months, with a reported average of 121 minutes of exercise a week. Result: no difference on the disability scale for dementia (adjusted difference 1.3 points, CI 5.2 against to 2.6 in favour; Cohen's d 0.06), and a non-significant reduction in falls, 79 falls against 200 in the control group, but an adjusted incidence rate ratio of 0.78 with a confidence interval of 0.5 to 1.3 (p = 0.3).7

The summary by modality and level of evidence

How to read this: the pills give the level of evidence as it was graded by the authors of the reviews cited, and not an appraisal made by this article.

How do you adapt the instruction when memory no longer keeps up?

This is the know-how specific to this population. It rests on a simple neuropsychological dissociation: the memory that lets you recount the session and the memory that lets you perform the movement again are not the same, and they do not deteriorate at the same rate.

What remains available

A patient who cannot say what they did ten minutes ago can nonetheless improve their performance on a repeated motor task. That is procedural memory, and every rehabilitation that is possible in major neurocognitive disorder rests on it. The practical implication is direct: you do not explain an exercise, you get it done, and you get it done again.

The cleanest trial on this point compares errorless learning, where the patient is given the cue before they can go wrong, with trial-and-error learning, where the cue only comes after the mistake. Sixty participants, split into three groups of twenty: severe neurocognitive disorder, mild to moderate disorder, and controls with no disorder. Performance was overall better after errorless learning (p = 0.012), with a maximal effect size in the mild to moderate group at delayed testing 1 to 3 days later, Cohen's d of 1.61. And the most interesting point: the gap between the two methods grows with the delay.10

Why does that gap grow? Because a mistake once made is encoded as a trace, and the patient no longer has the explicit memory that would let them recall that the trace was wrong. They retain the error with the same strength as the success. In a healthy person, explicit memory sorts it out; here, it no longer does.

What this gives in a session

And honesty about what does not always work

A case series published in the Journal of Geriatric Physical Therapy tempers the enthusiasm. Three women aged 89 to 95, moderate Alzheimer's disease, in residential care, received twelve physiotherapy sessions over four weeks explicitly incorporating errorless learning, massed repetition and spaced retrieval. All improved on the objective balance measures, but only one exceeded the minimal detectable change, and neither the Timed Up and Go nor comfortable gait speed changed. The authors conclude that the feasibility of the errorless paradigm was limited in practice.25

This is a finding to hold on to: the learning principles that come out of the laboratory are robust, their transfer to a real session with a 93-year-old patient is not yet. We apply them because they are plausible and cost nothing, not because they are demonstrated at that level of detail.

How do you prevent falls in a patient with a neurocognitive disorder?

This is the most frequent reason for referral, and the only area where physiotherapy in this population has an objective that nobody disputes. The difficulty is that the strongest factor is not a motor one.

The size of the risk, measured

The reference study followed 179 people over 65 for twelve months with a prospective falls diary: 38 with Alzheimer's disease, 32 with vascular disorders, 30 with Lewy body disease, 40 with Parkinson's dementia and 39 healthy controls. Patients with dementia fell nearly eight times more often than the controls: 9,118 falls per 1,000 person-years against 1,023, incidence density ratio 7.58 (CI 3.11 to 18.5).3

On univariate analysis, two predictors dominate: belonging to the Lewy body group in the broad sense (Lewy body disease and Parkinson's dementia together) multiplies the risk by 3.33 (CI 2.11 to 5.26); having fallen in the previous twelve months multiplies it by 2.52 (CI 1.52 to 4.17).

But it is the multivariate predictors that matter for care, because they are modifiable:

• symptomatic orthostatic hypotension, HR 2.13 (CI 1.19 to 3.80);
• the autonomic symptom score, HR 1.055 per point on a 0 to 36 scale (CI 1.012 to 1.099);
• the Cornell depression score, HR 1.053 per point on 0 to 40 (CI 1.01 to 1.099);
• and conversely, the level of physical activity, protective, HR 0.827 per point on a 0 to 9 scale (CI 0.716 to 0.956).3

What the physiotherapist does with this

Three direct consequences, and one qualification.

First, the blood pressure cuff enters the assessment. Symptomatic orthostatic hypotension doubles the risk of falling, and it takes three minutes to measure: blood pressure lying down after five minutes of rest, then standing at one and three minutes. This is not an encroachment on the medical role, it is a screening test whose result determines the day's session and the content of the letter to the doctor.

Second, mood is a fall factor. A high depression score predicts falls independently of everything else. An apathetic patient who refuses to stand up is not a non-compliant patient, they may have a symptom that calls for treatment.

Third, physical activity is the only protective lever identified in this cohort. It is the soundest argument in favour of intervening, and it must be quoted for what it is: an observational association, not an effect demonstrated by a trial.

For that is where the qualification lies, and it must be stated plainly. The intervention trials have not confirmed the promise. The GRADE review of 43 trials judges the effect of exercise on falls inconclusive.4 PrAISED, the largest trial designed specifically for this population, finds 79 falls against 200, which looks spectacular, but an adjusted incidence rate ratio of 0.78 with an interval running from 0.5 to 1.3: statistically, no conclusion is possible.7 FINALEX, for its part, reports significantly fewer falls in both of its exercise arms.8

The physiotherapist who promises a family that they will prevent falls promises what the literature does not support. The one who treats orthostatic hypotension, flags depression and maintains walking acts on three factors that we know count.

Frontotemporal degeneration: why does behaviour govern the session?

This is the disease on this list that least resembles the idea we have of a dementia. The patient is often young, remembers their address and their grandchildren's first names perfectly, and does not do what is asked of them. The temptation to conclude a lack of motivation is almost irresistible, and it is wrong.

Who we are talking about

Frontotemporal lobar degenerations cover several clinical syndromes, two of which dominate practice: the behavioural variant, by far the most frequent, and the primary progressive aphasias, themselves split into three variants (non-fluent or agrammatic, semantic, logopenic) by the international classification of 2011.16 The ratio between the two is about four to one in favour of the behavioural variant.18

The most recent meta-analysis, 32 population studies, puts overall incidence at 2.28 per 100,000 person-years (CI 1.55 to 3.36) and prevalence at 9.17 per 100,000 (CI 3.59 to 23.42). For the behavioural variant alone, incidence 1.20 and prevalence 9.74. For the primary progressive aphasias, incidence 0.52 and prevalence 3.67. In the under-65s, incidence is 1.84 per 100,000 person-years and prevalence 7.47. The authors conclude that the frequency of frontotemporal degeneration is comparable to that of Lewy body disease.17

The figure that has to stay in mind is a ratio, not an absolute value. In studies that include people over 65, frontotemporal degeneration accounts on average for 2.7 % of cases of neurocognitive disorder. In those restricted to the under-65s, 10.2 %, with values running up to 15.7 %.18 In a patient under 65 referred for “cognitive problems”, the frontotemporal hypothesis is not exotic: it is about one case in ten.

The sex ratio is balanced, unlike neurocognitive disorders as a whole: 373 men to 338 women across the studies reporting the information, a non-significant difference (p = 0.18).18

An early onset, a short course

The international cohort of the genetic forms, 3,403 mutation carriers from 1,492 families, gives age landmarks very different from those of Alzheimer's disease:19

A patient of 52 in full working life, with a mean disease duration of six to nine years, does not raise the same questions as a patient of 84. They often have dependent children, a mortgage running, a spouse still working, and sometimes a disability claim to put together. The HAS explicitly mentions referral to the maisons départementales des personnes handicapées, the French departmental disability offices, “for young people”.12

The clinical picture, and what it does to the session

The international criteria revised in 2011 define a “possible” behavioural variant by the presence of at least three of the following six features : disinhibition, apathy or inertia, loss of sympathy or empathy, perseverative or compulsive behaviours, hyperorality and dietary change, a dysexecutive neuropsychological profile with relative sparing of memory and visuospatial skills. The “probable” form adds functional impact and suggestive imaging. These criteria were validated on 137 autopsy-confirmed cases: sensitivity of 86 % for the possible form and 76 % for the probable, against 53 % for the earlier criteria (p < 0.001).15

One detail of that validation is worth the physiotherapist knowing: the patients the new criteria still missed were older and had an atypical presentation with marked memory impairment.15 In other words, an older patient who forgets a great deal does not rule out frontotemporal degeneration; it only makes it harder to recognise.

The Swedish register of behavioural and psychological symptoms, 10,405 care-home residents assessed with the neuropsychiatric inventory, makes it possible to place the frontotemporal profile precisely against Alzheimer's. Compared with Alzheimer's patients, frontotemporal patients have more apathy, more disinhibition and more abnormalities of appetite and eating behaviour, and fewer delusions, fewer hallucinations and less irritability.42

Each of these items translates directly into the session.

Apathy is not a lack of willingness

This is the most important point of the chapter. Frontotemporal apathy is a deficit of initiation: the patient wants to, understands, and does not start. Reproaching them for it is as useful as reproaching someone with Parkinson's for their tremor. What works is to remove the initiation step: begin the movement with them, place the limbs physically, use a regular external trigger (a rhythm, a count, an object to grasp) rather than a verbal invitation. An instruction that begins with “whenever you are ready” is an instruction that will never be followed.

Disinhibition calls for a framework, not a negotiation

A disinhibited patient may stand up in the middle of the exercise, leave the room, or say inappropriate things. The strategies that work are environmental: short session, room without competing stimuli, same time and same running order every time, equipment brought out at the last moment. Identical repetition also exploits the perseverative tendency rather than fighting it.

Hyperorality and dietary changes concern the physiotherapist

Rapid weight gain, sugar cravings, putting non-food objects in the mouth: this alters the load to be moved, the risk of aspiration, and sometimes the running order of the session itself. In a young patient it is also one of the most frequent causes of carer exhaustion.

The motor signs, and why they are not incidental

Frontotemporal degeneration is often presented as a purely behavioural disease. That is false, and the GENFI study shows it well. Across 322 mutation carriers, a principal component analysis identifies five clusters of motor signs : a profile suggesting progressive supranuclear palsy, a bulbar-onset amyotrophic lateral sclerosis profile, a mixed profile with amyotrophic lateral sclerosis predominance, a parkinsonian profile and a corticobasal syndrome profile. Mixed and amyotrophic lateral sclerosis type signs are the most frequent, followed by parkinsonian signs. In C9orf72 carriers, motor signs are detectable up to 25 years before the expected onset of symptoms.20

In practice, this means that a frontotemporal patient may arrive with a motor picture that resembles Parkinson's disease, a corticobasal syndrome, or motor neurone disease, and that the programme must fit that motor picture, not the label of the diagnosis.

What has been tried, and with what result

The intervention literature is thin. The most useful trial is a feasibility study of the Tailored Activity Program, an occupational therapy programme that prescribes the carer personalised activities calibrated to the preserved abilities, adapted here to frontotemporal degeneration. Twenty dyads randomised, nine in the programme and eleven as controls, assessed at four months: a significant reduction in behavioural symptoms (F = 8.073, p = 0.011) and maintenance of functional performance (p = 0.548), that is, no decline over the period of the study. The qualitative analysis documents acceptability for the carers.21

Twenty dyads is not proof. But the principle transfers and costs little: anchor the exercise in an activity that means something to the person, hand its starting over to the carer, and take not declining over four months as a result.

Faced with a frontotemporal patient, the physiotherapist is not first rehabilitating a motor deficit: they are building the conditions in which a movement can start. The rest of the job comes afterwards.

Alzheimer's disease: why does the instruction matter more than the programme?

This is the most frequent aetiology, and therefore the one we believe we know. Its particularity for physiotherapy comes down to one sentence: the patient keeps the ability to execute for a long time, and loses very early the ability to retain what they have been asked to execute.

What the diagnosis covers today

The reference clinical criteria remain those of the working group of the National Institute on Aging and the Alzheimer's Association, published in 2011, which define dementia due to Alzheimer's disease and recognise, alongside the classic amnestic presentation, non-amnestic presentations: language, visuospatial, or dysexecutive.22

In 2024, an Alzheimer's Association workgroup published a revision that changes the very nature of the diagnosis: Alzheimer's disease is defined there biologically, by its biomarkers, with a clinical staging system distinct from the biological definition.23 A patient can therefore be “biologically Alzheimer's” with no symptom, or symptomatic at a given stage. For the physiotherapist, the practical consequence is not to confuse the biological label with the functional stage: it is the second that decides the session, and it cannot be read off a PET scan report.

Citing the version in force matters: between the 2011 criteria and the 2024 revision, what moved was not a detail of wording, it was the very object of the definition.

The clinical sequence, and where it tips over

The most frequent presentation begins with a disorder of anterograde episodic memory : the patient no longer encodes new events. They can recount their military service and not what they had for lunch. At that stage their motor function is normal, and they perform perfectly any movement shown to them. The problem is never execution, it is autonomous repetition between two sessions.

Then, later, the instrumental signs appear: aphasia, apraxia, agnosia. Apraxia changes everything, because it attacks directly the object of physiotherapy work: the patient can no longer organise the motor sequence, get dressed, use a walking frame, turn over in bed in the right order. This is no longer a memory problem, it is a movement problem.

This shift has prognostic value. Two cohorts followed with a repeated Mini-Mental State, 57 and 70 patients, show that patients who develop aphasia or apraxia early decline faster than those who do not develop them, and that this difference is not explained by the baseline score.24 Seeing apraxia appear in a session is therefore not only a technical nuisance on the day: it is a signal of acceleration, to be recorded and to be passed on.

What exercise brings, specifically in this disease

The Finnish FINALEX trial is the most convincing, because it involves Alzheimer's patients living at home with their spouse as carer, which is exactly the situation in community practice. Two hundred and ten patients, three arms: group exercise (four-hour sessions including about one hour of training), individual home exercise (one hour of training), both twice a week for a year, and a usual care arm.8

Two lessons for the clinic. The first: the most effective arm is the home one, with the spouse present, and not the group run by professionals. The second: the intervention did not increase health costs, it even reduced them in the group arm (p = 0.03), a useful argument when the funding has to be justified.8

On walking itself, the double-blind trial already cited elsewhere in this article involved 61 patients averaging 81.9 years, with a confirmed mild to moderate neurocognitive disorder, with no distinction of aetiology: progressive strengthening and functional work, two times two hours a week for three months, with large gains in speed, cadence, step length and double support, Cohen's d from 0.80 to 1.27.9

The cognitive-motor interface, and a drug avenue worth knowing

The gait of an Alzheimer's patient deteriorates above all under dual task, when walking demands part of the attentional resources. A randomised trial in patients at the mild cognitive impairment stage tested donepezil on this precise parameter: the dual-task cost fell significantly in the treated group, whether the second task was counting backwards by 1 (10.25 % against 1.75 %, p = 0.048) or by 7 (21.38 % against 14.64 %, p = 0.037). Dual-task speed improved by 4 to 11 cm/s but without reaching significance, and the reduction in the fall rate was not significant.27

This result is not a prescription, it is a framework for reasoning: in these patients, part of the instability is attentional and not muscular. It justifies testing dual-task walking in the assessment, and not concluding too quickly that there is a balance deficit when the patient walks well as long as nobody speaks to them.

The advanced stage

When walking disappears, the role does not disappear. The HAS guide lists explicitly, for the stages of major loss of independence: maintaining walking and balance and preventing falls for as long as the patient is mobile; measures against the effects of immobility once they are not; prevention of the complications linked to immobility and incontinence; prevention of aspiration ; adapted stimulation in the non-communicating person, relational touch being given as an example.12 Dysphagia in Alzheimer's disease has been the subject of a dedicated systematic review, which documents its reality and what is at stake.28

In an Alzheimer's patient, the best-chosen exercise is worth nothing if it is not done again between two sessions. The useful programme is not the best programme: it is the one the carer knows how to restart.

Lewy body disease: which emergencies must the physiotherapist know?

This is the only one of the three where information passed on in time by a physiotherapist can prevent a death. This chapter therefore begins there, before even the clinical description.

The Haute Autorité de Santé has built this constraint into its care pathway guide: low-dose risperidone comes in only as a third-line option, and the text states “after ruling out a MCL”, that is, a maladie à corps de Lewy, a Lewy body disease. Where hallucinations or delusions are associated with this disease, the guide points towards rivastigmine or clozapine, after a specialist opinion. Any neuroleptic prescription must also be preceded by an electrocardiogram.12

Recognising the disease: the criteria in force

The reference criteria are those of the fourth consensus report of the DLB Consortium, published in 2017, and not the third report of 2005 that is still often cited. This revision separates clinical features clearly from diagnostic biomarkers, and gives greater diagnostic weight to REM sleep behaviour disorder and to myocardial metaiodobenzylguanidine scintigraphy. It also recalls that its management recommendations rest largely on expert opinion, randomised trials in this disease being few.29

The four core clinical features are:

• the cognitive fluctuations, with marked variation in attention and alertness;
• the recurrent visual hallucinations, typically well formed and detailed;
• the REM sleep behaviour disorder, which may precede cognitive decline by several years;
• one or more spontaneous parkinsonian signs.29

REM sleep behaviour disorder deserves particular attention, because it is both highly informative and greatly under-recorded. In a cohort of patients given a diagnosis of Lewy body disease in secondary care, it was found in only 8.4 % of the records, which the authors themselves judge well below what was expected.35 A multicentre study of idiopathic REM sleep behaviour disorder has also quantified the risk of conversion to a dementia or a parkinsonism.36 The question “does your husband wave his arms, talk or hit out during his sleep?” put to the spouse takes thirty seconds at the end of the assessment, and it points the way.

A frequent and under-diagnosed disease

The reference systematic review puts the figure at 4.2 % of all the dementias diagnosed in the population, and 7.5 % in secondary care, with an incidence representing 3.8 % of new cases. That is about one case in 25 in the community and one in 13 in hospital. The authors add that the real prevalence is probably far higher, diagnoses being frequently missed.31

A British study of memory services confirms the order of magnitude, 4.6 % of cases (CI 4.0 to 5.2), and documents the extent of the under-diagnosis: the rate varies from 2.4 to 5.9 % across services, and Parkinson's dementia accounts for only 9.7 % of Parkinson's patients (CI 8.3 to 11.1), far fewer than expected. The authors conclude that both forms are considerably under-diagnosed.32

What this means for the clinic: across a geriatric caseload, a physiotherapist sees Lewy body disease. Some cases carry the right diagnosis, many carry the label “Alzheimer's” or “Parkinson's”.

What makes this disease dangerous in a session

Falls, far above the other aetiologies

In the prospective falls cohort already cited, belonging to the Lewy body group multiplied the risk of falling by 3.33 (CI 2.11 to 5.26) compared with the other causes of neurocognitive disorder, at equal age and sex.3

And these falls precede the diagnosis. A comparison of 4,682 Lewy body patients, 20,214 with Alzheimer's, 28,222 with Parkinson's and 20,214 controls, drawn from general practice data, shows that falls, psychiatric symptoms and dysautonomia are more strongly associated with Lewy body disease than with Alzheimer's disease or Parkinson's disease in the five years preceding the diagnosis.34 The physiotherapist who takes on a 72-year-old for repeated unexplained falls, with a little constipation and a sad mood, is sometimes sitting opposite a Lewy body disease that nobody has yet named.

Orthostatic hypotension, in one patient in two

A meta-analysis of 18 studies finds orthostatic hypotension in 50.8 % of 662 patients with Lewy body disease, with an odds ratio of 7.71 (CI 4.42 to 13.44) compared with controls.33 One patient in two: measuring blood pressure on standing is not a precaution of principle, it is a high-yield examination, and it determines the way the patient is moved from the treatment couch to standing.

The fluctuations, which make an isolated assessment misleading

A Lewy body patient may pass a balance test on Tuesday and fail it on Thursday with no worsening of the disease at all. This has two operational consequences. First, an assessment done on a good day overestimates the real capacities, and the home adaptations set on that assessment will fall short. Second, the variability seen in a session is not a failure of motivation or a problem of measurement: it is a clinical sign, and it is worth recording and dating.

What rehabilitation can do, and the real state of the evidence

It has to be said plainly: this is the area where we know least, and for a structural reason. A systematic review sifted 111,485 articles, read 288 in full text, and retained only five studies totalling sixteen participants. The cause of that collapse is explicit: 89.6 % of the articles set aside were set aside because they excluded participants who had both a dementia and a parkinsonism. Exercise trials in Parkinson's disease exclude people with dementia; exercise trials in dementia exclude people with parkinsonism. Lewy body patients fall between the two.39

Across the thirteen participants whose habitual gait speed was measured, the gain was 0.18 m/s (CI 0.02 against to 0.38 in favour), which exceeds the moderate important change threshold used in Parkinson's disease (0.14 m/s). The authors ask that this result be treated with caution, and they are right: thirteen people.39

The PRIDE trial, published in 2022, is the first conducted specifically in this population. A non-randomised crossover design, nine participants included, six having received the intervention (three Parkinson's dementias, three Lewy body diseases), with an eight-week waiting phase followed by eight weeks of high-intensity progressive resistance training, demanding balance work and functional exercises, three days a week. During the waiting list, the participants declined to a clinically significant degree in functional independence, cognition, physical function, muscle mass and weight. After the intervention, they improved to a clinically significant degree in functional independence, cognition, physical function and strength. Adherence exceeded 80 % and only one minor exercise-related adverse event was reported.40

Six participants do not make a proof. But the tolerance message is useful: high-intensity resistance training was carried out in patients with dementia and parkinsonism, from the mild to the severe stage, without notable incident.

Finally, a point worth knowing even though it is not a physiotherapy matter: rivastigmine showed, in an international randomised double-blind placebo-controlled trial, an improvement in behavioural symptoms, with 37 patients (63 %) improved by at least 30 % against 18 (30 %) on placebo, and gains bearing particularly on tasks with a strong attentional component.37 A patient whose attentional fluctuations improve on treatment becomes a patient with whom the session is possible.

In the other two diseases, the physiotherapist adapts their practice. In this one, they also act as a sensor: they watch the patient walk every week, which nobody else in the care chain does.

Lewy bodies or Parkinson's dementia: where does this article stop?

The question necessarily arises, because the two diseases share their lesion, their motor signs and a good part of their management. The boundary is a convention, and you have to know which one you are using.

Both entities belong to the same Lewy body pathology. The diagnostic convention that separates them is chronological: when the neurocognitive disorder appears before the parkinsonian signs or in the year that follows them, we speak of Lewy body disease ; when it arises in a patient known to have Parkinson's for more than a year, we speak of Parkinson's dementia. The fourth consensus explicitly maintains this rule, while calling it an arbitrary threshold based on expert opinion, useful in clinical practice and open to revision once the mechanisms of these diseases are better known.29 The exercise studies, for their part, treat the two together under the term Lewy body dementia in the broad sense.3940

Parkinson's dementia is frequent: it concerns 9.7 % of Parkinson's patients in the British survey already cited, a figure the authors themselves judge far below reality.32

How do you tell the three entities apart in a session?

The physiotherapist does not make the diagnosis. They do, however, need to know which disease they probably have in front of them, because the running of the session, the precautions and the information to pass on all depend on it.

Wide table: scroll it horizontally to see the three columns.

What do real clinical cases teach us?

Four publications, one for each typical situation. They are cited with their identifier: these are real patients, not vignettes made up to illustrate a point.

An avoidable fall in a care home, in an Alzheimer's patient

An 85-year-old woman with Alzheimer's disease, living in a long-term care facility. She develops balance problems and weakness leading to gait deviations and falls; the whole set of clinical measures and her history place her at high risk of falling.

The programme put in place is ordinary in content and unusual in form: lower limb and trunk strengthening, work on balance, on walking and on walking aids, and above all training the care staff in a functional maintenance programme, centred on activities that engage implicit memory and on the aspects of communication usually preserved in dementia.

Result after four weeks and twelve 30-minute sessions: the Tinetti score goes from 8/28–16/28, the Berg Balance Scale from 7/56–19/56, and the number of incident reports linked to a fall goes from 2 to 0 over a four-week period. The authors stress that these results were obtained by modifying the usual fall prevention interventions, taking account of the fall in cognitive capacity, the communication deficits and the behavioural problems.26

Three Alzheimer's patients, and the limit of errorless learning

Three women aged 89 to 95, moderate Alzheimer's disease, in assisted living. Twelve physiotherapy sessions over four weeks, with explicit integration of high-frequency repetition, errorless learning and spaced retrieval. All attended at least ten of the twelve sessions.

Result: improvement in the objective balance measures in all three patients, but only one exceeded the minimal detectable change, and none changed their Timed Up and Go or their comfortable gait speed. The authors conclude that the feasibility of the errorless learning paradigm was limited in practice and call for further research.25

This case is cited here precisely because it is disappointing. One positive case series and one mixed case series in comparable populations: that is the real state of the literature, signals rather than certainties.

An 87-year-old Lewy body patient whose trajectory reverses

An 87-year-old man with Lewy body disease, living in a care home for dependent older people. Over nine months he shows a rapid functional decline and weight loss, together with injurious falls. He is assessed by a geriatrician and an exercise physiologist, independently of one another, during a prolonged waiting period before entry into an exercise pilot trial.

The analysis identifies three treatable contributors to his frailty and his functional collapse: polypharmacy, sarcopenia and undernutrition. The intervention combines deprescribing of hazardous medications, intensive exercise and increased calorie intake. The authors report an improvement in his physical and neuropsychological state, and conclude with a sentence worth quoting: frailty is not a contraindication to robust exercise, it is on the contrary one of the most important reasons to prescribe it.41

Three lessons for the clinic. The first: faced with a rapid decline, look for what is reversible before concluding that the disease is progressing. The second: the list of medications is part of the assessment, and a physiotherapist who reads it can trigger a review. The third: very advanced age and an advanced stage do not rule out progressive resistance work.

Twenty frontotemporal dyads, and a definition of success

The personalised activity programme already mentioned was tested in twenty patient and carer dyads, nine receiving the intervention and eleven as controls, with assessment at baseline and at four months. The intervention consists of working with the carer to prescribe activities calibrated to the patient's preserved abilities, with a view to managing behaviour.

Results: a significant reduction in behavioural symptoms (F = 8.073, p = 0.011) and maintenance of functional performance (p = 0.548). The qualitative analysis brings out five themes, including the benefits perceived by the carer, their readiness to change, the strategies they use to engage the patient, and the obstacles to implementation.21

In a disease whose mean duration is six to nine years,19 an unchanged functional performance at four months is a result, provided it is announced as such at the first appointment.

How do you apply this in the clinic?

This chapter is deliberately operational: what you measure, what you do during the hour, what you write. It does not replace clinical reasoning, it stops you forgetting what counts.

The minimum assessment, and what is added to it according to the aetiology

The running order of a session that holds

1. Always the same beginning. Same time, same room, same first task. Routine does the work that explicit memory no longer does, and it exploits the perseverative tendency instead of suffering it.
2. One action, one sentence. No verbal chaining, no open question, no “do you remember what we did last time”. Guide physically before the mistake happens.10
3. A slow transition to standing in every parkinsonian patient or patient suspected of Lewy body disease: sitting on the edge, pause, then standing, with something to hold on to.33
4. The body of the session: strength and balance, weight-bearing, with progression. That is where the demonstrated effects lie.49 Aim for about 60 minutes, 2 to 3 times a week, on exercises the patient can succeed at.
5. Finish with what will be done again at home. Two exercises, not ten. Shown to the carer, performed in front of them, written on a sheet that is handed to them.8
6. Record the variability. A performance that differs greatly from one session to the next is clinical data, especially in Lewy body disease.

What we tell the family, and what we do not promise

The systematic review devoted to communication strategies in supporting people with a neurocognitive disorder is a reminder that the way of saying things is part of the care, in the same way as what is said.11 Three sentences are enough to frame a course of treatment honestly:

• “The aim is for the loss of independence to be slower, not for it to be reversed.” That is what FINALEX shows, and it can be checked.8
• “Exercise will not slow the memory disease.” That is what DAPA shows, and saying it avoids a disillusion that stops the sessions in the sixth month.6
• “What we can improve is strength, balance, gait speed and endurance.” That is the highest level of evidence we have.4

The letter to the doctor, in five lines

Frequently asked questions

Should a patient with a major neurocognitive disorder be given exercise?

Yes, for what exercise demonstrably does: strength, balance, gait speed, endurance, with a strong level of evidence and an established dose of 60 minutes, 2 to 3 times a week, supervised.4 No, if the stated objective is to slow cognitive decline or to improve quality of life: both promises are contradicted by the trials.64

Can a very frail or very old patient do intensive strengthening?

The GRADE review indicates that a low motor level, and not the cognitive level, predicts the best response to training.4 A pilot trial ran high-intensity progressive strengthening in Lewy body patients from the mild to the severe stage, with more than 80 % adherence and a single minor adverse event.40 And an 87-year-old man in a care home, in a full frailty spiral, saw his trajectory reverse under anabolic exercise, deprescribing and refeeding.41 Frailty is an indication, not a contraindication.

How do you know whether the patient has understood the instruction?

Not by asking them. You see it by watching them do it, and you prevent the mistake instead of correcting it: that is the principle of errorless learning, whose benefit grows with the recall delay.10 A successful instruction is one whose execution starts within two seconds.

What do you do with a patient who systematically refuses to stand up?

Before talking about refusal, consider two hypotheses. An apathy of frontal origin, which is a deficit of initiation and not of willingness: the answer is to start the movement with them rather than inviting them to. Or a depression, which independently predicts falls in this population (HR 1.053 per point of the Cornell scale) and which calls for treatment.3

Can a patient with dementia learn a new exercise?

Yes, by the procedural route. It is the basis of everything described in this article: a patient who will never be able to recount the session can nonetheless improve their performance on a repeated task.10 Honesty requires adding that the transfer of this principle into measurable functional gains, in patients over 89, produced only an inconsistent result in the one published series.25

Why should a Lewy body patient not be given a neuroleptic?

Because 54 % of the patients who received one, in the reference autopsy series, had a severe reaction, with excess mortality within the year (HR 2.70), against 7 % in the Alzheimer's patients.30 The HAS allows risperidone only as a third-line option and only after ruling out this disease.12 The physiotherapist does not prescribe, but is often the first to see the motor consequence of the prescription.

Does the distinction between Lewy body disease and Parkinson's dementia change my management?

Little, in practice. Both belong to the same pathology and the few exercise studies treat them together.3940 The one-year rule that separates them is explicitly called arbitrary by the 2017 consensus.29 What really changes what you do is the fluctuations, orthostatic hypotension and the contraindication to neuroleptics, which hold for both. For the motor content of the sessions, our article Parkinson's disease and physiotherapy remains the reference.

A young patient with behavioural problems, is that really a dementia?

It is one of the hypotheses to consider. In studies restricted to the under-65s, frontotemporal degeneration accounts for 10.2 % of neurocognitive disorders, against 2.7 % across all age groups.18 The genetic forms begin at 49.5 years on average for MAPT.19 And the initial picture is behavioural, with memory normal for a long time: that is exactly what leads to a wrong diagnosis of depression or burnout.

Should balance tests be repeated several times?

In Lewy body disease, yes, systematically. Cognitive fluctuations are a core feature of the disease,29 and an assessment carried out on a good day overestimates the real capacities, and so under-sizes the adaptations. In the other two diseases, a single measure is generally enough, provided it is repeated under the same conditions of time of day.