Article Text

Download PDFPDF

Drugs for disruptive features in dementia


It is estimated that around 750,000 people in the UK have dementia and, of these, 60% will have behavioural changes or psychological symptoms.13 These features tend to occur in clusters, of which the five most readily recognised are characterised by aggression, apathy, depression, psychomotor agitation or psychosis.4 Here, we review the place of drug therapy in the management of patients in whom these features are disruptive.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


The commonest types of dementia are Alzheimer's disease (around 60% of those affected), vascular dementia (about 25%) and dementia with Lewy bodies (10-15%). These types differ with respect to the frequency of resulting behavioural and psychological disturbances. For example, delusions are commoner in Alzheimer's disease than in vascular dementia,2 while hallucinations are more frequent in dementia with Lewy bodies than in Alzheimer's disease.5


Assessment of a patient presenting with behavioural or psychological disturbances involves first being sure that the underlying diagnosis of dementia is accurate. There should also be a clinical, functional and social assessment of the nature and causes of the disturbances.6 This requires accurate and detailed descriptions of presenting features, including their severity, duration and frequency. As dementia progresses, this information-gathering will increasingly depend on interviews with carers. Some symptoms may occur only in specific circumstances, for example, aggression in response to particular people. Changes in surroundings may disorientate or upset a patient. Other precipitating factors include physical discomfort or constipation, which are often overlooked. Some problem behaviours are self-limiting,7,8 but those present for 3 months are more likely to persist long term. It is important to remember that if the behaviour does not pose a danger to the patient or others, nor upsetting to the person themselves, little or no intervention may be needed.

Following initial assessment, a specific individualised management plan should be drawn up for the patient and the carers.6 Such plans should be co-ordinated by a designated healthcare professional and reviewed regularly. Delirium and any intercurrent illnesses that may cause or worsen symptoms (e.g. urinary tract infection) should be treated.

Management without drugs

When managing disruptive patients, non-drug interventions should be tried first. If the patient has visual or hearing loss, this should be remedied where possible. Similarly, any environmental features which could precipitate disruption should be tackled. It is worth asking the patient's close friends or relatives about strategies that have helped in the past. A variety of specific non-drug approaches have been assessed in clinical studies, including behavioural and psychological interventions, occupational activities (e.g. music therapy), sensory stimulation (e.g. bright light treatment) and aromatherapy.6,914 However, much of the evidence on these interventions comes from small poorly designed studies that have not provided convincing evidence of benefit.

Should drug therapy be started?

Decisions on whether or when to start drug therapy should be based on the patient's physical and psychological state and apparent level of distress; the severity and duration of disturbances; the risk of harm to the patient or carer of both treating or not treating; any concomitant medication and illness; the patient's needs and preferences (where these can be determined); and, sometimes, the ability of carers to cope with the disturbances.

Informed consent

Many patients with dementia will be unable to give informed consent to drug treatment, and in the UK, relatives and staff have no legal right to give such consent on a patient's behalf.6,15,16 If a person is not capable of giving or refusing consent, and has not validly refused such care in advance, treatment may still be given lawfully if it is deemed to be in the patient's best interests. However, this should happen only after full consideration of its potential benefits and unwanted effects and consultation with the carer(s), relatives and other people close to the patient.

Conventional antipsychotics

Clinical efficacy

There is little convincing evidence from published randomised controlled studies to support the widespread use of antipsychotic drugs for managing difficult behaviour (e.g. agitation, aggression) in patients with dementia. Individual studies have often not included a placebo group or been too small to identify meaningful differences between comparator groups. Also, differences between studies in the types of dementia included, patient selection, and definitions of 'difficult' behaviour make interpretation of the data difficult.6 Moreover, many studies have used outcome measures that have not been validated for the patient groups studied or have been too general to assess particular behaviours adequately.6

In a meta-analysis of seven randomised placebo-controlled studies, involving a total of 252 patients with various types of dementia (mean age 71-83 years), conventional antipsychotic drugs (at doses equivalent to 66-267mg daily of chlorpromazine given for 3-8 weeks) improved behavioural symptoms in 18% more patients treated than did placebo.17 No individual study showed a difference between active and placebo groups, and high placebo responses were seen in some studies (up to 67%). Another meta-analysis of 16 randomised controlled (either placebo or active drug) trials, involving a total of 734 patients with dementia, found that conventional antipsychotics given for 3-12 weeks improved behaviour in 61% (95% CI 47-75%) of patients treated (vs. 34% [95% CI 18-50%] with placebo).18 This was assessed in the majority of patients by Clinical Global Impression scores, which measured changes in medical status, cognitive dysfunction, daily functioning and behaviour. There were no significant differences between the antipsychotic drugs tested. In a third, larger meta-analysis of five randomised placebo-controlled trials, involving a total of 856 patients with dementia, haloperidol (0.5-6mg daily for 3-16 weeks) had no effect on agitation or Clinical Global Impression scores, but appeared to reduce aggression.19

Unwanted effects

The long-term unwanted effects of antipsychotics in older people, particularly those with coexisting medical conditions, are largely unknown. Unwanted effects such as akathisia may resemble an exacerbation of the disruptive problem being treated. Older people are particularly susceptible to tardive dyskinesia, which may develop within 1 year and is dose-dependent. For example, in a cohort study, involving 261 patients aged over 55 years and previously untreated with conventional antipsychotic drugs, the cumulative rates of tardive dyskinesia were 25% after 1 year and 53% after 3 years of cumulative exposure to conventional antipsychotics.20 Such effects may be irreversible despite stopping therapy and unresponsive to treatment. Another cohort study, involving 71 patients with dementia and lasting 2 years, suggested that treatment with conventional antipsychotics may double the decline in cognitive score compared with no antipsychotic treatment.21 Other unwanted effects that may occur in older people include hypotension, hypothermia, confusion, blurred vision and constipation.

Atypical antipsychotics


Two randomised placebo-controlled trials have assessed the efficacy of risperidone (taken for 12 weeks) in controlling behavioural problems in patients aged 55 years or over with DSM-IV dementia (Alzheimer type, vascular dementia, or mixed dementia). In one study, involving 625 institutionalised patients, more of those on risperidone 1-2mg daily had a 50% or greater reduction in the Behavioural Pathology in Alzheimer's disease (BEHAVE-AD) total score, the primary endpoint (45% with 1mg/day vs. 50% with 2mg daily vs. 33% with placebo).22 Of the features assessed, risperidone reduced psychosis and aggressiveness, but not activity disturbances, diurnal rhythm disturbances, affective disturbances, anxieties or phobias. More patients on the 2mg dose developed somnolence (28% vs. 17% with 1mg dose vs. 8% with placebo), peripheral oedema (18% vs. 13% vs. 6%) and extrapyramidal unwanted effects (21% vs. 13% vs. 7%). In the second study, involving 344 patients, the proportion reaching the primary endpoint (at least 30% reduction from baseline in BEHAVE-AD total score) were similar for risperidone (72%, mean final dose 1.1mg daily), haloperidol (69%, mean dose 1.2mg daily) and placebo (61%).23 The therapeutic effect of risperidone was clearest for aggressive symptoms. The incidence of extrapyramidal unwanted effects was similar with risperidone (15%) and placebo (11%), but higher with haloperidol (22%).23


In a 6-week randomised placebo-controlled trial involving 206 nursing home patients with Alzheimer's disease, olanzapine 5-10mg daily reduced the sum of core symptoms (agitation/aggression, hallucinations and delusions) of the Neuropsychiatric Inventory-Nursing Home scale (primary outcome measure) from baseline by 43-53% (vs. 25% with placebo; p<.05). This rating scale is designed to assess psychopathology in patients with Alzheimer's disease and other dementias.24 Somnolence and abnormal gait, but not extrapyramidal unwanted effects, were reported more frequently with olanzapine.

Carbamazepine and valproic acid

Carbamazepine and valproic acid (as the semisodium salt) have been studied in two, 6-week, randomised placebo-controlled studies involving nursing home patients aged 60 years or over with a variety of dementia types and exhibiting agitation, tension, hostility, unco-operativeness or excitement.25,26 In the first study, involving 51 patients, carbamazepine (mean final dose 304mg daily) reduced the mean total Brief Psychiatric Rating Scale score, the primary outcome measure (by 7.7 vs. 0.9 with placebo), primarily by decreasing agitation and aggression.26 This scale assesses withdrawal/depression, agitation, cognitive dysfunction, hostility/suspiciousness and pyschotic distortion. The treatment benefits were apparent 3 weeks after therapy was started. Overall, unwanted effects (e.g. ataxia, drowsiness, disorientation, rash) were more common with carbamazepine (59% vs. 29% for placebo). In the second study, involving 56 patients, semisodium valproate (mean final dose 826mg daily) had no effect on the primaryoutcome measure of agitation on the Brief Psychiatric Rating Scale score.25


Around 40-50% of patients with dementia have depressive symptoms at some stage,27 although depression may be difficult to diagnose in such people. Randomised controlled trials lasting 4-8 weeks suggest that antidepressants can help in relieving depression in these patients.2730 There is no clear evidence that any one antidepressant is more effective than any other, but unwanted effects may differ between them.6 In general, tricyclic-related drugs (e.g. trazadone) and selective serotonin re-uptake inhibitors (SSRIs; e.g. paroxetine) produce fewer antimuscarinic and cardiovascular unwanted effects than do older tricyclic antidepressants. Drugs with pronounced anticholinergic effects should be avoided in patients with dementia. SSRIs should be used with caution in patients on concomitant medications that increase the risk of bleeding (e.g. low-dose aspirin).

Anxiolytics and hypnotics

Anxiety and insomnia are common in dementia. However, anxiolytics and hypnotics should only be used if symptoms are severe and persistent and only for the shortest time possible (preferably no more than 1 week for hypnotics and 2-4 weeks for anxiolytics).6,31

Cholinesterase inhibitors

Few trials of cholinesterase inhibitors have reported the effects of these drugs on behavioural and psychological disturbances. These drugs are only licensed for the treatment of mild to moderate dementia in Alzheimer's disase. Systematic reviews on donepezil,32 ▼galantamine33 and rivastigmine34 in Alzheimer's disease, involving 2,664-3,370 patients, have suggested that such treatment improves cognitive scores, clinicians' global rating and activities of daily living scores. However, it is difficult to know whether any of these improvements are accompanied by beneficial changes in behavioural and psychological features. Results specifically from clinical trials published since the systematic review on donepezil32 are contradictory. In a 24-week randomised placebo-controlled study, involving 208 patients with Alzheimer's disease in a nursing home, donepezil 5-10mg daily had no effect on neuropsychiatric symptoms (the primary outcome measure).35 However, another 6-month randomised placebo-controlled trial involving 290 patients with Alzheimer's disease found that donepezil 5-10mg daily improved neuropsychiatric symptoms (a secondary outcome measure).36 Also, two placebo-controlled trials of galantamine have provided conflicting evidence on whether the drug helps relieve neuropsychiatric symptoms in Alzheimer's disease.33 In a randomised placebo-controlled 20-week study involving 120 patients with dementia with Lewy bodies, rivastigmine 3-12mg daily improved neuropsychiatric symptoms by at least 30% in twice as many patients as did placebo.37 All three cholinesterase inhibitors cause dose-related unwanted gastrointestinal cholinergic effects (e.g. diarrhoea). Hallucinations, agitation and aggressive behaviour have also been reported with donepezil; these resolve when the drug is stopped or the dose reduced.

Practical considerations

Most of the drugs discussed in this article are not licensed for the management of behavioural disturbances or psychological symptoms in elderly people with dementia. Many, moreover, should be used with particular caution in this population.


Up to 30% of patients in UK nursing homes receive anti-psychotic drugs.38 In the USA, such overuse has led to legislation restricting provision (Omnibus Reconciliation Act of 1987).39 In one UK study, involving 909 nursing home residents, only 27 of 217 of those on antipsychotic therapy would have been allowed such treatment according to US guidelines.38

UK guidelines suggest that antipsychotic drugs should not be regarded as a treatment option unless a patient has serious problems, in particular, psychotic symptoms, or is seriously distressed or exhibits behaviour that poses a danger.6 It is important to recognise that even this restricted use is largely unsupported by convincing evidence. Antipsychotic drugs should not be given to patients who may have dementia with Lewy bodies, which is characterised by visual hallucinations, fluctuating cognitive impairment and parkinsonism. In these patients, antipychotics may precipitate irreversible parkinsonism, further impair consciousness levels and induce an autonomic disturbance similar to neuroleptic malignant syndromes, and increase mortality rates 2-3 fold.5

Treatment with an antipsychotic should start with a low dose, and be increased slowly (not more often than once weekly) if symptoms persist. Once the patient is stabilised, treatment should be reviewed every 4-6 weeks. Doses should be reduced as soon as possible and treatment stopped if no longer essential or helpful. Multiple-drug therapy should be avoided, particularly when regular therapy is given concomitantly with 'as required' therapy. If patients develop tardive dyskinesia, early withdrawal of the drug at the earliest sign (fine involuntary movements of the tongue) may prevent full development of the condition. The routine use of anticholinergic medication to prevent extrapyramidal unwanted effects is contraindicated, as these drugs can worsen cognitive function or precipitate delirium. Concomitant use of other drugs with pronounced anticholinergic effects (e.g. some antidepressants) should also be avoided.


Antidepressants should only be used to treat patients with severe and persistent depression. The choice of drug will depend on various factors including presence of concomitant disease or other therapy, suicide risk and previous response to antidepressant therapy. In general, sedating preparations are more likely to be suitable for agitated and anxious patients, while less sedating treatments are more appropriate for withdrawn and apathetic patients. As with antipsychotic therapy, initial doses should be low, with close monitoring for unwanted effects, and increased only gradually as needed.


Management of behavioural disturbance or psychological symptoms in patients with dementia is often difficult, not least because those affected are often unable to give informed consent to treatment. First-line steps include treating intercurrent illnesses and addressing aggravating physical and environmental factors, and employing appropriate non-pharmacological interventions. Antipsychotic drugs are widely used in such patients, but there is little convincing evidence to support this practice, which in most instances is unlicensed. Such therapy should only be used as a last resort in patients with disturbances that are causing severe problems. Atypical antipsychotic drugs cause fewer extrapyramidal unwanted effects than conventional antipsychotics. Antipsychotic drugs should be avoided in patients suspected of having dementia with Lewy bodies. Antidepressants, anxiolytics and hypnotics should only be used in patients with relevant marked and persistent symptoms. The routine use of drugs with pronounced anticholinergic effects may worsen cognitive function or precipitate delirium. Whether carbamazepine, valproic acid and cholinesterase inhibitors help in the management of problem behaviours in dementia is not known.


[M=meta-analysis; R=randomised controlled trial]

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. R 9.
  10. R 10.
  11. R 11.
  12. R 12.
  13. 13.
  14. R 14.
  15. 15.
  16. 16.
  17. M 17.
  18. M 18.
  19. M 19.
  20. 20.
  21. 21.
  22. R 22.
  23. R 23.
  24. R 24.
  25. R 25.
  26. R 26.
  27. R 27.
  28. R 28.
  29. R 29.
  30. R 30.
  31. 31.
  32. M 32.
  33. M 33.
  34. M 34.
  35. R 35.
  36. R 36.
  37. R 37.
  38. 38.
  39. 39.
View Abstract