The neuropsychiatric symptoms of dementia include signs and symptoms of disturbed perception, thought, mood, or behavior (1). Clinically significant neuropsychiatric symptoms are found in about one-third of dementia patients with mild impairment and in two-thirds with more severe impairment (2, 3) and in an even higher proportion of dementia patients in residential care (4, 5). Neuropsychiatric symptoms contribute significantly to caregiver burden, institutionalization (6), and decreased quality of life for patients with dementia (7).

Psychotropic medications are often prescribed for neuropsychiatric symptoms, but concerns have been raised about the safety and efficacy of these medications (8–10). Psychological approaches may have fewer risks, but little is known about their efficacy. We conducted a systematic review of psychological approaches to neuropsychiatric symptoms in dementia with the aim of making evidence-based recommendations about the use of these interventions. The review included studies examining any therapy derived from a psychological/psychosocial model. We considered the effects of the interventions in terms of neuropsychiatric symptoms and related outcomes and assessed whether the benefit was time limited or sustained.

Search Strategy

We searched electronic databases through July 2003, reference lists from individual and review articles, and the Cochrane Library and sought expert knowledge of additional studies, even those published after July 2003. We also hand-searched the contents of three journals published during the 10-year period up to July 2003.

We used search terms encompassing individual dementias and interventions. We included studies with quantitative outcome measures that were either direct or proxy measures of neuropsychiatric symptoms (e.g., care costs, quality of life, institutionalization, and decreased medication or restraint). Studies of people without dementia, dementia secondary to head injury, or interventions that either involved medication or were not based on a psychological model (e.g., aromatherapy, homeopathy, occupational therapy, light therapy) were excluded.

Data Extraction Strategy

We used a tool adapted from a review of checklists (11). Ratings of the level of evidence were assigned to studies according to the Oxford Centre for Evidence-Based Medicine guidelines (http://www.cebm.net/levels_of_evidence.asp#levels). Levels of evidence grades ranged from 1 to 5, with lower numbers indicating higher quality. Each type of intervention was then given an overall "grade of recommendation" according to the Oxford Centre for Evidence-Based Medicine criteria. The grades ranged from A (consistent level of evidence grade of 1) to D (level of evidence grade of 5 or troublingly inconsistent or inconclusive studies at any level).

We identified 1,632 references; 1,421 were excluded and 162 were included.

Reminiscence Therapy

Reminiscence therapy (t1) uses materials such as old newspapers and household items to stimulate memories and enable people to share and value their experiences. We identified five studies of reminiscence therapy interventions (12–16). Three were small randomized, controlled trials. One had 10 participants and reported behavioral improvements when reminiscence therapy was preceded by reality orientation, but not vice versa (13). The improvement was not clearly significant. The other two studies found no benefit of reminiscence therapy (14, 15). Two level-4 studies had small numbers (12, 16). One reported a significant improvement in mood, although the raters were not masked to participants’ treatment group (16).

• •  We assigned a grade of recommendation of D to reminiscence therapy.

Validation Therapy

Validation therapy (t1), rooted within the Rogerian humanistic psychology premise of individual uniqueness, is intended to give an opportunity to resolve unfinished conflicts by encouraging and validating expression of feelings. We identified three studies of validation therapy. The first, a case series of five individuals, indicated an improvement in irritability after validation therapy (17). The second, which included five patients who served as their own comparison subjects, reported no change in behavior (18). A randomized, controlled trial compared validation therapy to usual care or a social contact group in 88 patients with dementia (19). Although at 1-year follow-up the nursing staff thought the validation therapy group improved, there was no difference in independent outcome ratings, in nursing time needed, or in use of psychotropic medication and restraint.

• •  Because of the absence of conclusive evidence, we assigned a grade of recommendation of D to validation therapy.

Reality Orientation Therapy

Reality orientation therapy (t2) is based on the idea that impairment in orientating information (day, date, weather, time, and use of names) prevents patients with dementia from functioning well and that reminders can improve functioning. Eleven studies assessed reality orientation therapy (13, 20–29). The strongest randomized, controlled trial, which had 57 participants, showed no immediate benefit of reality orientation therapy, compared to active ward orientation (24). In a smaller randomized, controlled trial (N=10), patients who received reality orientation therapy followed by reminiscence therapy had fewer neuropsychiatric symptoms, compared to patients who received the treatments in the reverse order (13). The other smaller nonrandomized, controlled trials mostly found benefits in the reality orientation therapy groups in terms of improved mood, decreased neuropsychiatric symptoms, or delayed institutionalization.

• •  The grade of recommendation for reality orientation therapy is D.

Cognitive Stimulation Therapy

Cognitive stimulation therapy (t3), derived from reality orientation therapy, uses information processing rather than factual knowledge to address problems in functioning in patients with dementia. Three of four randomized, controlled trials of cognitive stimulation therapy (31, 32, 34, 35) showed some positive results, although the studies used different follow-up endpoints (immediately after therapy to 9 months after therapy). There were early behavior improvements, relative to waiting list. By 9 months, no significant difference between groups was found. One study showed reduced depression, and another showed improvement in quality of life but not in mood (34, 35). The final study did not report whether the differences in behavior were significant (32).

• •  Given the mostly consistent evidence that cognitive stimulation therapy improves aspects of neuropsychiatric symptoms immediately and for some months afterward, our consensus is that the grade of recommendation is B, although the evidence is not consistent in all respects.

Other Dementia-Specific Therapies

We identified two other dementia-specific therapies (30, 33) (t3). The first, "individualized special instruction," consisted of 30 minutes of focused individual attention and participation in an activity appropriate for each individual (30). The participants in the pilot randomized, controlled trial were their own waiting-list comparison subjects. During the intervention period, their behavior did not deteriorate, compared with deteriorating behavior before the intervention period.

The second dementia-specific therapy was "self-maintenance therapy," which is intended to help the patient maintain a sense of personal identity, continuity, and coherence (33). This intervention incorporates techniques from validation, reminiscence, and psychotherapy. A 3-week admission of patients and caregivers to a specialist unit in which self-maintenance therapy was provided led to a significant decrease in depression and problematic behavior, compared to baseline. This outcome may have been partly attributable to the environment.

• •  These level-4 studies support a grade of recommendation of C for both interventions.

Non-Dementia-Specific Therapies

Twenty-five reports described use of non-dementia-specific psychological therapies for patients with dementia (36–60) (t4). Nearly all of the studies examined behavioral management techniques. In one large randomized, controlled trial, participants received either manual-guided treatment for the patient and caregiver or a problem-solving treatment for the caregiver only (43). The two interventions were equally successful in improving depressive symptoms immediately and at 6-month follow-up (43, 44). Two other small randomized, controlled trials also had positive results (37, 42). In one of those studies, participants had significantly fewer neuropsychiatric symptoms 2 months after being taught progressive muscle relaxation. In the other study, the behavior of patients with the dementia of multiple sclerosis improved with "neuropsychological counseling" (a cognitive behavior intervention). There were two other randomized, controlled trials in which behavioral management techniques were used (36, 40); these techniques were ineffective in one of the studies (36). It used a complex, difficult-to-classify intervention that included a variety of techniques (e.g., life review, sensory stimulation, single-word commands, and problem-oriented strategies) (36). The second used a token economy, which was more effective in reducing "bizarre" behavior in patients with severe dementia, compared to a preintervention condition, but less effective than a milieu treatment (40). Several single-case studies are summarized in t4.

• •  The grade of recommendation for standard behavioral management techniques in dementia is B. The findings of the larger randomized, controlled trials were consistent and positive, and the effects lasted for months.

Psychological Interventions With Caregivers

t5 and t6 summarize 19 reports that describe interventions with family caregivers designed to ameliorate neuropsychiatric symptoms or frequency of institutionalization in dementia (61–79). Seven studies involved training the caregiver to use behavioral management techniques (t5). A randomized, controlled trial (65) found no difference in agitation or global outcome in a comparison of treatment with behavioral management techniques, haloperidol or trazodone alone, or placebo at 16 weeks. Behavioral management techniques taught to caregivers did not reduce psychotropic drug use or symptom frequency at 1-year follow-up (67). Exercise and behavioral management techniques led to significant improvements in depression at 3 months but not at 2 years (66). In a smaller randomized, controlled trial, behavioral management techniques based on the progressive Lowered Stress Threshold Model were taught to caregivers with the aim of reducing stimulation in response to specific stressors identified by caregivers (63). Both study groups received the intervention, one in the form of written materials, and the other in a training program. A positive effect for care recipients was found in the second group. The evidence that behavioral management techniques with caregivers and exercise training with patients helps depression is strong, but it is unclear which component was the active component.

• •  Because the findings of other studies are inconsistent, the grade of recommendation for teaching caregivers behavioral management techniques to manage psychological symptoms is D.

t6 summarizes the results of nine studies (seven randomized, controlled trials) involving psychoeducation to teach caregivers how to change their interactions with patients with dementia. In one large trial, improvement in neuropsychiatric symptoms at 16 weeks was found, but the difference only approached significance (73). In a second trial, primarily powered to improve mental health in caregivers rather than in patients, improvement in neuropsychiatric symptoms occurred immediately after 12 weeks of training in stress management, dementia education, and coping skills but was not maintained at 3-month follow-up (69). A third, smaller trial examining the effects of an intervention with individual families found significant improvements at 6 months in mood and ideational disturbance (74). In a randomized, controlled trial of an educational program for family carers that included supportive counseling, psychoeducation and training in management strategies, and home visits, the rate of institutionalization of patients was decreased (70). The effect continued for 3 months but not 2 years. A fifth randomized, controlled trial involved psychoeducation, instruction to caregivers in how to change their interactions with the patient, or both (68). Patients’ behavior improved at 6 months, but the difference only approached significance. The researchers attributed the nonsignificant result to the fact that the trial was a pilot study that had limited power. Another study examined the effects of caregiver psychoeducation in working with nursing home residents to enhance social activities and self-care; the intervention resulted in a decrease in agitation after 6 months (77). Finally, a level-1 study investigated a comprehensive support and counseling intervention for spouse caregivers that included problem solving, management of troublesome behavior, education, and increased practical support, followed by long-term support groups (78). Patients’ neuropsychiatric symptoms were not directly measured, but the intervention was found to delay time to institutionalization by nearly a year. The other studies were noncontrolled and showed either improvement that approached significance or significant improvement (71, 72).

• •  The grade of recommendation for behavioral management techniques in the form of psychoeducation and teaching caregivers how to change their interactions with patients is A, because evidence from level-1, level-2, and level-4 studies consistently supports these interventions, and the effects have been shown to last months.

An uncontrolled study suggested that family counseling is helpful in reducing institutionalization of patients (76). In a nonrandomized, controlled trial, a family support group resulted in a decrease in problem behavior but not in depression (75).

• •  The grade of recommendation for family counseling is C, because the intervention is supported by two level-4 studies.

A single controlled study compared the effects of "admiral" nurses—specialists in treatment of dementia who worked in the community with persons caring for patients with dementia—to those of usual treatment and showed no effect on institutionalization of patients (79).

• •  The grade of recommendation for caregiver support by specialist nurses in the community is D.

Psychosocial Interventions

Sensory enhancement

Music/music therapy

Music/music therapy interventions (t7) included playing music from specific eras or particular genres, such as Big Band music, as part of activity sessions or at certain times of day, including mealtimes or bath times. Participants also played musical instruments, moved to music, or participated in composition and improvisation sessions. Of 24 music/music therapy interventions (15, 80–102), six were investigated in randomized, controlled trials (15, 84, 89, 92, 95, 98). All were small trials and showed improvements in disruptive behavior. In two, behavior was observed during the music sessions, but there was no evidence that benefit carried over after the sessions (84, 92). In three studies, behavioral change was observed outside of the music/music therapy session. In the first study, patients were significantly less agitated, both during and immediately after music/music therapy in which the music was chosen to fit the individuals’ preference (89). The results of the second study were similar (95). In the third study, which assessed music, hand massage, or a combination of both for 10 minutes, decreased agitation was observed 1 hour after the intervention (98). All but one of the other studies (100) were controlled. Most of them found a benefit, although some did not (83).

• •  The grade of recommendation for music therapy for immediate amelioration of disruptive behavior is B, because consistent level-2 evidence suggests that music therapy decreases agitation during sessions and immediately after. There is, however, no evidence that music therapy is useful for treatment of neuropsychiatric symptoms in the longer term.

Snoezelen therapy/multisensory stimulation

Snoezelen therapy/multisensory stimulation (t8), which combines relaxation and exploration of sensory stimuli, such as lights, sounds, and tactile sensations, is based on the idea that neuropsychiatric symptoms may result from periods of sensory deprivation. Interventions occurred in specially designed rooms and lasted 30–60 minutes. Of six trials of Snoezelen therapy/multisensory stimulation, three were randomized, controlled trials. The first was a very small trial with no clear results (103). The other two found that disruptive behavior briefly improved outside the treatment setting but that there was no effect after the treatment had stopped (104, 105). The other reports described studies of individual cases (106, 107) and an uncontrolled trial in which improvements were found but no statistics were provided (108).

• •  The grade of recommendation for Snoezelen for amelioration of disruptive behavior immediately after the intervention is B, on the basis of consistent evidence from level-2 studies. The effects are apparent only for a very short time after the session.

Other sensory stimulation

Of seven trials of other forms of sensory stimulation (t8), three were randomized, controlled trials. The first trial compared massage with a comparison condition, music, or a combination of massage and music (98). Decreased agitation was observed 1 hour after the intervention. The second trial examined a sensory integration program that emphasized bodily responses, sensory stimulation, and cognitive stimulation; this intervention had no effect on behavior (112). Similarly, a small randomized, controlled trial found that white noise had no effect on sleep disturbance and nocturnal wandering (114). An "expressive physical touch" intervention (5.5 minutes/day of touching, including 2.5 minutes/day of gentle massage and 3 minutes/day of intermittent touching with some talking) over a 10-day period decreased disturbed behavior from baseline immediately and for 5 days after the intervention (111). White noise tapes led to immediate decrease in agitation (109). A controlled trial of stimulation with "natural elements" while bathing (sounds of birds, brooks, and small animals were played and large bright pictures were displayed) found that agitation decreased significantly only during bathing (115). The other study of single cases found no difference in agitation before and after therapeutic touch or massage (113). In the final two studies, the effects of several forms of sensory stimulation involving touch, smell, and taste were examined. A small randomized, controlled trial reported no change associated with the intervention (110), and the other study found that the intervention was helpful (116).

• •  The grade of recommendation for short-term benefits of sensory stimulation is C, but there is no evidence for sustained usefulness.

Simulated presence therapy

Six studies investigated the effects of simulated presence therapy, in which positive autobiographical memories are presented to the patient in the form of a telephone conversation usually involving a continuous-play audiotape made by a family member or surrogate (t9). One randomized, controlled trial found no change in agitated or withdrawn behaviors (117). Staff observations suggested reduced agitation in patients who received the intervention, compared to a placebo group but not compared to patients receiving usual care (117). A small study found improved social interaction and attention (118). Simulated presence therapy used to address agitation led to significant decreases in agitation and improved social interaction but no change in aggressive behaviors (119). When simulated presence therapy was used regularly, problem behaviors were reduced by 91% (119). Finally, in a series of single case studies, Peak and Cheston (120) reported mixed results, with increased ill-being in one participant and reduced anxiety and increased social interaction in other participants. Use of video to provide simulated presence was not associated with significant changes in agitated behavior (121).

• •  The grade of recommendation for simulated presence therapy is D.

Structured activity

Therapeutic activity programs

There were five randomized, controlled trials of therapeutic activities (t9). In a small-scale randomized, controlled trial, therapeutic activities at home were associated with significant decreases in agitation (123). Another study found that small group discussion and being carried on a bicycle pedaled by volunteers alleviated patients’ depression but not agitation at 10 weeks (122). The third found no effects of puzzle play on social interaction and mood (95). Similarly, a comparison of games and puzzle play with Snoezelen and another study comparing structured activity with a control condition found no improvements in mood and behavior (104, 129).

The other studies of therapeutic activities were nonrandomized, controlled trials. Ishizaki et al. (124) found no beneficial effects of weekly therapeutic activities on depression. In another study, a combination of group and individualized activity sessions in day care significantly increased agitation over 10 weeks (125). A controlled, nonrandomized clinical trial of weekly activity groups led by nursing assistants found no behavioral changes (126). There was, however, less use of physical restraint generally, and psychotropic medication use was reduced in seven of 20 participants. A specialist day-care program providing structured daily activities for patients with dementia was associated with decreased institutionalization and was more cost-effective than nursing home care (29). Patients who were rocked on a swing did not show a decrease in aggression (128). Three case studies of diverse group activities (games, music, exercise, socializing) found equivocal effects on behavior (127). In two studies that used reading sessions as an intervention, some improvement was seen in wandering (86) and disruptive behaviors were decreased in both patients in the study both during and 1 week after the reading intervention (87).

• •  Not all therapeutic activity programs used the same interventions, but overall, the study findings are inconsistent and inconclusive. The grade of recommendation is D.

Montessori activities

Montessori activities use rehabilitation principles and make extensive use of external cues and progression in activities from simple to complex (t10). Three nonrandomized, controlled trials utilized Montessori-based activities and found no change in depression and agitation (132, 135, 138).

• •  The grade of recommendation for Montessori activities is D.

Exercise

Three studies examined the use of exercise/movement/walking as an intervention for neuropsychiatric symptoms (t10). A well-conducted randomized, controlled trial found no effects on behavior in a "walk-talk" program in which one caregiver walked with two residents or walked and talked with two residents (131). A randomized, controlled trial of a psychomotor activation program found no behavioral effect (133). The other two studies were nonrandomized, controlled trials. One study, in which 11 patients were their own comparison subjects, found a significant reduction in aggressive behaviors on days when a walking group was held (134). The other study, a small matched, controlled trial of exercise groups, found no significant reduction in agitated behaviors (136).

• •  The grade of recommendation for exercise is D.

Social interaction

A small report of single cases studies showed decreased neuropsychiatric symptoms in one-third of patients who had enforced social interaction with nurses for 3 hours/day for 1–2 months (137).

• •  The grade of recommendation for enforced social interaction is D.

Decreased sensory stimulation

Two small studies investigated decreased sensory stimulation (t10). A "quiet week" intervention (turning off the television, lowering voices. and reducing fast movement by staff at a day center) led to an immediate significant reduction in agitation as measured by a nonstandardized scale, compared to the period before the intervention (135). In another study, patients on a specially designed reduced stimulation unit—without television, radio, telephones; with scheduled rest periods and limited access to visitors—had no reduction in neuropsychiatric symptoms as measured by a standardized scale, compared with the period before the intervention, but use of restraint decreased (130).

• •  The grade of recommendation for decreased sensory stimulation is D.

Environmental manipulation

Visually complex environments

Eight studies (no randomized, controlled trials) investigated the effects of changing the visual environment (t10). The presence of two-dimensional grids on the floor near doors did not reduce exiting behaviors (150). However, two studies in which a horizontal grid pattern was used reported significant decreases in patients’ attempts to open doors and in patients’ ambulation (142, 144). Similar results were found in a study of the effects of murals on the walls above doorways (145). Blinds and cloth barriers placed over doors/door handles and signs installed to provide a focus of patients’ attention were also effective in reducing time spent attempting to exit the ward (140, 143, 148). Enhancement of the visual environment in a selected area of a residential home was associated with a decrease in agitated behaviors, although the finding was not statistically significant (139).

• •  Consistent evidence from level-4 studies for changing the environment to obscure the exit indicates a grade of recommendation of C.

Mirrors

Two small nonrandomized, controlled trials investigated the effects of mirrors in the patient’s environment (t10). In a study with a single case design, one of two patients was less agitated after removal of mirrors from the ward environment (146). Placing a full-length mirror over a doorway led to a significant decrease in exiting during the intervention for nine patients (147).

• •  The grade of recommendation for use of mirrors is D.

Signposting

Three nonrandomized, controlled trials investigated the effects of signposting on neuropsychiatric symptoms (t10). Two single case studies found that signposting alone was ineffective, but signposting in combination with reality orientation therapy led to improvements in ward orientation in two of four and five of five patients, respectively (141, 149). In the third study, signposts were placed alongside prompts that served to draw attention to the signs; this arrangement led to a reduction in neuropsychiatric symptoms in all five study participants (143).

• •  The grade of recommendation for signposting is D.

Other environmental manipulations

Group living

Group living is the name given to specially designed nursing homes that encourage a homelike atmosphere (t11). In a randomized, controlled trial, no change in neuropsychiatric symptoms was found in those in a group living setting, compared to community-dwelling waiting-list comparisons (155). Two other randomized, controlled trials showed decreased aggression, anxiety, and depression and less use of neuroleptic medication for 1 year in residents in group living settings (151, 152). No differences between group living and comparison subjects were observed 3 years later. Both studies were limited, because residents were selected for admission and were ineligible if they had frontal lobe symptoms, severe dementia, or a severe physical morbidity. A smaller uncontrolled trial of group living reported beneficial effects on neuropsychiatric symptoms at 6 months and reduced use of physical restraints (153). However, in another study, neuropsychiatric symptoms significantly increased in group living subjects, relative to comparison subjects, at 6 months and 1 year (156). In summary, group living may have beneficial or deleterious effects—or no effect—on neuropsychiatric symptoms.

• •  The grade of recommendation for group living is D.

Unlocking doors

One small uncontrolled study examined the effect of unlocking ward doors for 3-hour periods (154) (t11). Patients showed fewer neuropsychiatric symptoms and decreased wandering when the door was open (154).

• •  The grade of recommendation for unlocking doors is D.

Staff education in managing behavioral problems

Nine studies investigated the effects of staff education in treatment of neuropsychiatric symptoms. Three of the studies were randomized, controlled trials (163–165) (t11). A randomized, controlled trial of communication skills training for nursing and auxiliary staff showed significant reductions in patients’ aggression at 3 months and in patients’ depression at 6 months (163). Education of staff to implement an emotion-focused care program (validation, reminiscence, sensory stimulation) did not result in any change in neuropsychiatric symptoms (164). Staff education programs focused on knowledge of dementia and potential management strategies reduced use of physical restraint use (165) and, in a nonrandomized, controlled trial, decreased aggressive behavior toward staff (161). Specialized care programs for individuals in a residential home plus staff education improved emotional status and quality of life for residents 12 months later (157). A similar approach in a controlled trial with only 11 people in each arm led to nonsignificant differences favoring the intervention group (160). The result of a client-centered approach to agitation and sleep disturbance for 33 residents of a nursing home was equivocal. Verbal aggression decreased significantly, but the (less frequent) episodes of nonverbal agitation increased (162). Training staff in integrity-promoting care (staff gave more time, made the environment more homelike, encouraged patients to do more and to wear their own clothes) improved patients’ anxiety and depressed mood in a small controlled trial (158). In a large uncontrolled trial, training for nursing staff in using unstandardized observational outcomes led to an increase in restraint use but had no effect on agitated behavior (159).

• •  The grade of recommendation for specific staff education programs in managing neuropsychiatric symptoms is B, on the basis of consistent evidence from level-1 and level-2 studies, as well as supportive evidence from level-4 studies.

Environmental interventions combined with staff education

Eight nonrandomized, controlled trials investigated the effects of environmental interventions such as special care units designed for patients with dementia and staffed by specially trained workers who received ongoing training (t12). In a controlled trial, admission to a "low-density" special care dementia unit, which had fewer residents and larger living areas than standard units, was associated with a decrease in disruptive behavior (171). Similarly, in a controlled trial, a combination of group living and staff training was found to improve patients’ emotional and physical outcomes and was less costly than standard care (166, 167). In other studies, special care dementia units were associated with a reduction in neuropsychiatric symptoms, especially agitation and depression, and with a reduction in use of neuroleptic medication (167, 169). Aggression and activity disturbances were reduced in a small controlled trial of a special care dementia unit care (170). However, three other studies found no effect (168, 172, 173).

• •  The grade of recommendation for special care units combined with staff education is D.

We found numerous studies reporting psychological approaches to neuropsychiatric symptoms. We have tried to summarize and classify these studies using evidence-based guidelines in order to help clinicians understand which interventions are efficacious and over what time period. We also tried to distinguish interventions that are ineffective from interventions for which too little evidence is available to judge their effectiveness. Because some interventions are made up of several elements, we could have classified them in different ways. We tried to use the best fit and, by describing the interventions, to make our judgments transparent. Some therapies may require a huge amount of work for very little benefit, and we did not measure this aspect. In addition, some therapies may provide pleasure (either for the patients with dementia or for staff members) and thus may be worthwhile even if the intervention does not alter the patients’ neuropsychiatric symptoms. We did not attempt to judge these differential effects. Similarly, we did not study cognition as an endpoint, although some therapies are intended to have an effect on cognition.

Effective Psychological Therapies

Behavioral management techniques centered on individual patients’ behavior are generally successful for reduction of neuropsychiatric symptoms, and the effects of these interventions last for months, despite qualitative disparity. Psychoeducation intended to change caregivers’ behavior is effective, especially if it is provided in individual rather than group settings, and improvements in neuropsychiatric symptoms associated with these interventions are sustained for months. We therefore recommend these types of interventions.

Music therapy and Snoezelen, and possibly some types of sensory stimulation, are useful treatments for neuropsychiatric symptoms during the session but have no longer-term effects. The cost or complexity of Snoezelen for such small benefit may be a barrier to its use.

Specific types of staff education lead to reductions in behavioral symptoms and use of restraints and to improved affective states. Staff education is, however, heterogeneous, although instruction for staff in communication skills and enhancement of staff members’ knowledge about dementia may improve many outcomes related to neuropsychiatric symptoms. Teaching staff to use dementia-specific psychological therapies for which there is limited evidence of efficacy may not improve these outcomes.

What Interventions Need More Evidence?

Little evidence is available on the effectiveness of reminiscence therapy, but more positive evidence exists for cognitive stimulation therapy. Training for caregivers in behavioral management techniques had inconsistent outcomes but merits further study. The evidence for therapeutic activities is very mixed, and the study findings for these interventions are contradictory and inconclusive. Specialized dementia units were not consistently beneficial, but changing the environment visually and unlocking doors successfully reduced wandering in institutions. These promising interventions merit more study. There is no convincing evidence that simulated presence interventions or reduced stimulation units are efficacious for neuropsychiatric symptoms.

Which Interventions Were Ineffective?

Reality orientation therapy, validation therapy, "admiral" nurses, and Montessori activities had no effect on neuropsychiatric symptoms. In addition, convincing evidence suggests that simple repetitive exercise does not work for neuropsychiatric symptoms.

Overall our conclusions are limited because of the paucity of high-quality research. We found only nine studies with level-1 evidence. However, lack of evidence of efficacy does not mean lack of efficacy. Because the system of rating research assigns the highest ratings to randomized, controlled trials, most published studies of psychological interventions will not be rated as having the highest quality. The literature on behavioral interventions places greater weight on experimental single case studies, particularly in describing individualized interventions. The purpose of publication, however, is to provide evidence that can be generalized for future use. We have, therefore, used the Oxford Centre for Evidence-Based Medicine’s system for assessing evidence. We encourage the use of standardized interventions (which can be individualized within a context of adherence to basic principles) in future research so that interventions found to be effective can be used in other populations.

Received Nov. 15, 2004; revisions received Jan. 9 and Jan. 24, 2005; accepted Feb. 22, 2005. From the Department of Mental Health Sciences, University College London. Address correspondence and reprint requests to Dr. Livingston, Department of Mental Health Sciences, University College London, Holborn Union Building, Archway Campus, Highgate Hill, London, UK, N19 5LW; g.livingston@ucl.ac.uk (e-mail). The Old Age Task Force of the World Federation of Biological Psychiatry includes John Copeland, M.D., F.R.C.P., F.R.C.Psych., Bob Woods, M.Sc., Linda Teri, Ph.D., Henry Brodaty, A.O., M.B.B.S., M.D., F.R.A.C.P., F.R.A.N.Z.C.P., Pedro Ridruejo, Yong Ku Kim, M.D., Ph.D., Masatoshi Takeda, M.D., Ph.D., Manabu Ikeda, M.D., Ph.D., Dan Blazer, M.D., Ph.D., Carlos Augusto de Mendonca Lima, M.D., D.Sci., and Sirkka-Liisa Kivela, M.D.