Additional related information may be found at:
Neuropsychopharmacology: The Fifth Generation of Progress

Back to Psychopharmacology - The Fourth Generation of Progress

Alzheimer's Disease :

Treatment of Noncognitive Behavioral Abnormalities

Murray A. Raskind


The original patient described by Dr. Alois Alzheimer in 1907 (3) was remarkable both for her progressive cognitive impairment and for her prominent noncognitive behavioral abnormalities. Clinical interest in these noncognitive abnormalities in Alzheimer's disease (AD) has been substantial because of their high prevalence (70) and because noncognitive behavioral problems complicate patient management and often precipitate institutionalization (64). The real or apparent resemblance of delusions, hallucinations, depressed mood, agitation, hostility, and other noncognitive behavioral abnormalities of AD to the signs and symptoms expressed in such classic psychiatric disorders as depression, schizophrenia, and anxiety disorders has prompted the widespread use of psychotropic drugs in the management of AD (29). However, widespread use does not imply established efficacy. In fact, data establishing efficacy of psychotropic drugs for noncognitive behavioral problems in AD and other dementing disorders are limited. Carefully designed treatment outcome studies incorporating reliable and valid outcome measures, well-defined patient samples, and randomization to an adequate trial of active medication or placebo are few. This chapter will review informative studies of psychopharmacologic management of noncognitive behavioral problems in AD. These include depression (or depressive signs and symptoms), psychotic symptoms (delusions and hallucinations), and nonpsychotic disruptive behaviors (such as physical and verbal aggression, motoric hyperactivity, and uncooperativeness with activities necessary for personal hygiene and safety). Placebo-controlled studies will be emphasized, but results of other studies and reports will be discussed when they suggest directions for future investigation.


The diagnosis and treatment of depression complicating the course of AD has received considerable attention. Because depression per se can impair cognitive function (14), it is reasonable to hypothesize that effective treatment of depression in the AD patient may maximize potential cognitive capacity. Furthermore, there is consensus that reduction of depressive signs and symptoms improves quality of life (16). Unfortunately, the apparently straightforward goal of treating depression complicating AD becomes complex when the problems involved in the diagnosis of depression in the context of AD or other dementing illnesses are considered. A fundamental problem is the substantial overlap of signs and symptoms between depression and AD. Common to both disorders are apathy and loss of interest, impaired ability to think and concentrate, psychomotor changes (both retardation and agitation), and sleep disturbance. The ability to diagnose depression in AD is further compromised by the patient's lack of insight and poor recollection of symptoms.

Even if investigators agreed on uniform diagnostic criteria for syndromal depression in AD and used uniform assessment instruments and interviews, discrepant prevalence rates would likely arise from the differential characteristics of the samples of AD patients studied. Prevalence rates of AD with concurrent depression derived from clinical populations are higher than those derived from research registries that select "pure" AD subjects without a history of major depressive disorder. For example, in an outpatient geriatric clinic, Reifler et al. (48) found that 20% of AD patients met DSM-III criteria for major depressive episode. In contrast, Burke et al. (11) found no incident case of major depression patients in an AD research registry population followed longitudinally through the course of illness. This latter study included a longitudinally followed normal control group matched for age, sex, race, and social position. Signs and symptoms of depression occurred in both AD patients and controls, but major depression could not be diagnosed. A similarly low prevalence of major depression in a sample of AD subjects screened to exclude those with past histories of major psychiatric disorders was reported by Kumar et al. (37). Although depressed mood was more frequent in AD subjects than in age-matched normal controls, depressed mood in AD subjects was unaccompanied by classic vegetative signs and symptoms of depression. These investigators, therefore, interpreted depressed mood as reflecting "demoralization" rather than major depressive disorder. Given these problems, it is not surprising that estimates of the prevalence of depression in AD are widely disparate, ranging from 0% to more than 80% (70).

Perhaps the true prevalence of concurrent depression in AD lies somewhere in between these disparate estimates. As early as 1955, Sir Martin Roth (54) addressed the issue of differentiating the common "affective coloring" seen in dementia patients from the relatively uncommon "sustained depressive symptom complex." He found that the latter syndrome, which can probably be equated with DSM-III-R major depressive episode, occurred in only 3% of dementia patients. In a recent carefully performed study of AD patients admitted to an inpatient geropsychiatry service, Greenwald et al. (25) reported that 8% of patients with AD met DSM-III criteria for major depressive episode. Application of DSM-III-R as opposed to DSM-III criteria may reduce even further the rate of diagnosable major depressive episode in patients with AD (39). Regardless of the true prevalence of major depressive episode complicating AD, when such an episode occurs it produces excessive functional disability (44).

It is disappointing that the extensive interest in defining the prevalence of depression complicating AD has not generated many interpretable studies evaluating the outcome of antidepressant treatment in such patients. The database consists primarily of anecdotal case reports and non-placebo-controlled outcome studies. Although some of these reports appear useful, the only placebo-controlled treatment trial in the literature (50) raises concerns about the interpretability of uncontrolled studies.

Anecdotal reports have suggested that depression complicating AD may respond to tricyclic antidepressants (49), monoamine oxidase (MAO) inhibitor antidepressants (34), or electroconvulsive therapy (ECT) (18, 63). In a carefully performed open trial of nortriptyline, given in doses sufficient to achieve therapeutic plasma concentrations, or ECT in eight inpatients with AD complicated by depression, Reynolds et al. (52) reported significant reduction of mean Hamilton Depression Rating scores from 17 prior to treatment to 9 following treatment. Although the reduction in depressive signs and symptoms was substantial in AD patients with concurrent depression, it was less robust than in a similarly treated group of elderly nondemented depressed patients. Another open trial of "naturalistic" somatic antidepressant treatment of inpatients with dementia and concurrent depression was reported by Greenwald et al. (25). This study carefully documented the presence of major depressive episode in six patients with AD and four with multi-infarct dementia (MID) (see TardiveDyskinesia: Epidemiological and Clinical Presentation). Patients were treated for a mean duration of 11 weeks with a variety of conventional somatic antidepressants (doses not reported) and/or ECT. Hamilton scores significantly and substantially decreased from a mean of 19 on admission to a mean of 5 at discharge. This degree of improvement did not differ significantly from an elderly nondemented depressed inpatient group treated in a similar naturalistic manner. However, the mean length of stay to achieve comparable improvement in the elderly nondemented/depressed group was substantially shorter than in the demented/depressed group. Possible differential treatment responses between AD subjects with major depression and MID subjects with major depression were not reported. Both Reynolds et al. (52) and Greenwald et al. (25) interpreted their results as suggesting that major depression complicating dementia is responsive to standard somatic antidepressant treatments, but both investigators acknowledged that standardized, double-blind placebo-controlled studies of dementia patients with major depression are needed.

In the only placebo-controlled study of treatment of major depressive episode complicating AD reported in the literature, Reifler et al. (50) randomly assigned either the tricyclic imipramine (n = 13) or placebo (n = 12) to subjects who met DSM-III criteria for both primary degenerative dementia of the Alzheimer's type and major depressive episode. AD outpatients (mean age = 72 years) had a mean Mini Mental State Exam (MMSE) score of 17 [very comparable to the mean MMSE scores of the demented/depressed patients studied by Greenwald et al. (25) and Reynolds et al. (52)] and were suffering from a similarly moderate degree of depression (mean Hamilton score = 19). Imipramine (mean dose = 83 mg/day; mean plasma level of imipramine plus desmethylimipramine = 116 ng/ml) or placebo were prescribed for 8 weeks. Substantial, highly significant, and almost identical improvements occurred in mean Hamilton scores in both the imipramine subjects (19.3 prior to treatment compared to 11.5 following treatment) and placebo subjects (18.6 prior to treatment compared to 10.8 following treatment). The improvement of Hamilton scores was similar to those achieved in the open inpatient studies reported by Greenwald et al. (25) and Reynolds et al. (52). This outpatient study demonstrated that depressive signs and symptoms can be reduced in outpatients with AD, but the mechanism of treatment efficacy did not appear to be a specific antidepressant effect of imipramine. It is possible that more aggressive antidepressant medication treatment or a different type of somatic antidepressant treatment may have revealed a difference between an active treatment and placebo in this study. However, the substantial placebo response and the low likelihood that mean Hamilton scores much below 5 are achievable in patients with AD [mean Hamilton scores of 6 or 7 have been reported in samples of AD subjects in whom concurrent depression has been specifically excluded (25, 68)] make this possibility unlikely. At the least, this study (50) makes it mandatory that future antidepressant trials in AD patients with concurrent depression be placebo-controlled.



Prevalence of Psychotic Symptoms

Prevalence rates of psychotic symptoms (delusions and hallucinations) complicating AD cluster between 30% and 40% in carefully performed cross-sectional studies (17, 70). Because psychotic symptoms can emerge at any time during the course of AD and probably are more prevalent in the later stages of the illness, longitudinal studies reveal even higher prevalence rates. Drevets and Rubin (22) longitudinally followed subjects with AD from the early through the later stages of illness and documented the occurrence of psychotic symptoms both cross-sectionally and cumulatively. This study was strengthened by the inclusion of a longitudinally followed age-matched normal control population (none of whom developed psychotic symptoms during the course of the study). Slightly more than 50% of AD subjects manifested psychotic symptoms at some point during the course of their illness. Subjects were not considered positive if psychotic symptoms occurred only rarely or if they occurred only in the context of a possible delirium. As in other studies of psychotic symptoms in AD (51, 70), delusions were usually simple and persecutory, most commonly involving theft. Systematized delusions were uncommon. Hallucinations were most frequently visual, although auditory hallucinations were also common.

Nonpsychotic Disruptive Behaviors

Verbal and physical aggression, motor agitation, uncooperativeness with essential care, persistent irritability, and other disruptive or potentially disruptive behaviors are common in moderately to markedly demented patients with AD (15). In one well-conducted nursing home study, Rovner et al. (55) documented the occurrence of problematic disruptive behaviors in the majority of a random sample of 50 demented residents. Most frequently reported were restlessness, noisiness, and aggressive behaviors. Less demented AD patients still able to reside in the community also manifest disruptive behaviors. Ryden (56) surveyed caregivers of outpatients with AD and found a prevalence of aggressive behavior occurring at least once a week in 31% of subjects and daily in 16% of subjects.

Although psychotic symptoms and "nonpsychotic" disruptive behaviors may not always reflect the same underlying pathophysiologic process or processes, these two classes of noncognitive behavioral problems appear associated. Lopez et al. (38) evaluated the presence of belligerence, uncooperativeness, and physical and verbal aggression in AD patients with (n = 17) and without (n = 17) delusions and hallucinations. A greater proportion of psychotic AD patients (11 of 17) than of nonpsychotic AD patients (1 of 17) manifested these behavioral disturbances. A study addressing the relationship between psychotic symptoms and physical aggression in AD patients was reported by Deutsch et al. (19). Delusions (most commonly persecutory) and misidentifications (such as the patient believing his or her house was not their home or that strangers were living in the house) frequently preceded and were significantly associated with episodes of physical aggression. However, the presence of delusions could account for episodes of physical aggression in only a minority of cases. Therefore, other factors not clearly apparent were involved in the precipitation of physically aggressive behavior.



A number of studies consistently suggest that the presence of psychotic symptoms in AD is associated with more rapid deterioration of cognitive function. Stern et al. (66) were the first to report this phenomenon, and the association between psychotic symptoms and more rapid decline has since been confirmed by other groups. Lopez et al. (38) reported that AD patients with delusions and hallucinations had a more rapid decline in MMSE scores over a 1-year follow-up than did nonpsychotic AD patients and appeared to manifest a specific defect in receptive language. Drevets and Rubin (22) reported that the presence of psychotic symptoms predicted increased rate of cognitive deterioration. Recently, Jeste et al. (35) compared the performance of delusional and nondelusional AD patients on a neuropsychological test battery. Patients with delusions had a more rapid rate of dementia progression than did nondelusional AD patients. The explanation for this apparent association between psychotic symptoms and more rapid cognitive deterioration in AD is unclear.



Psychotropic drugs are widely prescribed to AD patients in long-term care facilities (29). In fact, the repeated documentation of widespread and often chronic prescription of antipsychotic and other psychotropic drugs in the long-term care setting has prompted Federal regulations designed to limit this practice to short-term treatment regimens with clear indications (30). Although the goals of such regulations are laudable, their application in the absence of empirically and soundly derived treatment guidelines continues to be problematic. Typical of studies documenting a high use rate for psychotropic medications in elderly long-term care recipients is a study of intermediate care facility residents in the state of Massachusetts (8). To ensure that the institutions surveyed represented typical geriatric facilities rather than those caring for deinstitutionalized patients with such long-term illnesses as chronic schizophrenia, facilities were not studied if they had greater than 20% of their residents admitted from inpatient psychiatric hospitals. More than half of all elderly residents surveyed were prescribed a psychotropic medication. Twenty-six percent were receiving an antipsychotic medication, and 28% were receiving a sedative-hypnotic medication (primarily benzodiazepines) or a sedating antihistamine. In a survey of nursing homes in Illinois (10), 60% of residents received at least one psychotropic medication during a 1-year period. The antipsychotics haloperidol and thioridizine and the benzodiazepine flurazepam were most frequently prescribed. Similarly, a random sample of 55 long-term care facilities in Massachusetts (5) found that over half of the residents were prescribed at least one psychoactive medication. In this study, antipsychotic medications were being administered to 39% of patients.


The rationale for the use of antipsychotic drugs in AD is partially attributable to the phenomenologic similarities of delusions and hallucinations and other disruptive behaviors occurring in AD to the symptoms occurring in schizophrenia. Unfortunately, this analogy to psychotic and other behavioral signs and symptoms in schizophrenia is imperfect. For example, delusions in AD are usually unelaborated persecutory beliefs often involving theft. Systematized, complex, and grandiose delusions are uncommon (70). In addition, the memory deficits basic to AD often appear to play a role in the development of delusional beliefs. For example, an AD patient who has forgotten where an item has been placed and who lacks insight as to his or her cognitive deficits might assume that it has been stolen. Or, an AD patient can stubbornly insist in a delusional manner that long-deceased persons who still remain alive in available memory traces are, in fact, alive. Although such beliefs of theft, of decreased persons continuing to exist, or of a spouse being an impostor may meet formal criteria for delusions, they are in some ways phenomenologically dissimilar to the delusions of schizophrenia for which the antipsychotic drugs have been demonstrated to be so effective. In fact, a meta-analysis of studies evaluating antipsychotic drugs in behaviorally disturbed dementia patients (60) concluded that although these drugs are more effective than placebo in patients with AD, the effect size is relatively small. The phenomenologic differences between psychotic features of AD and psychotic features of schizophrenia as well as the inclusion of nonpsychotic disruptive behaviors as target symptoms in outcome studies of antipsychotic drugs in AD may explain why the overall efficacy of antipsychotic drugs in AD is modest.


Interpretation of antipsychotic drug outcome studies in dementia patients performed prior to the introduction of DSM-III are often hampered by the use of unclear diagnostic nomenclature. For example, the term "senile psychosis" connoted severe dementia rather than the presence of delusions and hallucinations. In addition, early studies were usually performed in state hospital populations which included patients with degenerative neurologic dementing disorders (the majority of whom presumably were suffering from AD) and patients with chronic schizophrenia who had grown old in the institutional setting. In one of these early studies (61), 29 patients with the diagnosis of dementia complicated by disturbed behaviors were treated with either chlorpromazine or placebo in a double-blind crossover trial. Global ratings favored chlorpromazine, but serious adverse effects such as excessive sedation and falls also were more common in the active drug group. In a small study of dementia patients who manifested "hyperactive behaviors" such as assaultiveness, irritability, and hyperexcitability, acetophenazine was more effective than a placebo (26). Again, excessive sedation was a major adverse effect of active medication. In another small placebo-controlled study (67), haloperidol was more effective than placebo for the control of agitation, overactivity, and hostility in a group of elderly demented patients. The haloperidol group had more subjects who developed unsteady gait and/or extrapyramidal symptoms than did the placebo group. In this study, the best responders to haloperidol were those patients with the most severe agitated behaviors prior to study entry. All three of the above small placebo-controlled positive studies are notable for having included only dementia patients with disruptive behaviors.

Other early studies of antipsychotic drugs in dementia patients which did not include psychotic or nonpsychotic disruptive behaviors as target symptoms and signs found antipsychotic drugs no more effective than placebo. A comparison of trifluoperazine and placebo on target symptoms of apathy, withdrawal, and cognitive and behavioral deterioration (loss of ambulation, disorientation, and incontinence) demonstrated no therapeutic effect of the active medication and was most remarkable for a high prevalence of trifluoperazine-induced sedation and extrapyramidal signs and symptoms (27). In a comparison of thiothixene and placebo for the amelioration of cognitive deficits in a group of demented patients, therapeutic response to active drug again did not differ from placebo (47). Thirteen of 22 patients receiving thiothixene were rated as globally improved, and 11 of 20 patients in the placebo group were rated as globally improved. The presence of an apparent placebo response in this study confirmed an earlier report (1) which documented a positive placebo response in elderly demented patients.



Since the introduction of DSM-III, a small number of placebo-controlled studies have evaluated the efficacy of antipsychotic drugs in dementia patients. Diagnostic criteria for AD and MID increase confidence that elderly patients with chronic psychiatric disorders beginning in early life, such as schizophrenia, have been excluded. In addition, these latter studies have targeted psychotic and disruptive nonpsychotic behaviors as outcome measures. Perhaps the study which best supports the efficacy of an antipsychotic drug in the treatment of psychotic signs and symptoms complicating AD is a small but carefully performed outpatient study by Devanand et al. (20). Nine outpatients with predominantly early-onset AD (mean age = 67) were treated with either haloperidol (mean dose = 3 mg/day) or placebo in a crossover design. All nine AD subjects had clear delusions and/or hallucinations, and eight of the nine had other nonpsychotic disruptive behaviors. Although haloperidol was superior to placebo for psychotic symptoms, cognitive function worsened during active drug treatment and extrapyramidal adverse effects limited the improvement in quality of life achieved by subjects in the haloperidol condition.

Petrie et al. (45) compared low-dose haloperidol and loxapine to placebo in 64 inpatients of a state psychiatric hospital (mean age = 73 years). The sample included subjects who met diagnostic criteria for either AD or MID. Both antipsychotic medications were more effective than placebo for improvement of hallucinations, suspiciousness, hostility, excitement, and uncooperativeness. Global improvement ratings, however, only modestly favored the active medications. Thirty-five percent of haloperidol subjects and 32% of loxapine subjects as compared to 9% of placebo subjects were rated as moderately or markedly improved. Not only were therapeutic responses to active drugs in these elderly demented patients lower than would have been predicted from outcome studies in younger subjects with schizophrenia, adverse drug effects including sedation and extrapyramidal signs and symptoms were frequent. In the only placebo-controlled study of antipsychotic drugs in a typical community nursing home sample of elderly demented patients, Barnes et al. (7) randomized 60 behaviorally disturbed dementia patients (mean age = 83 years) to thioridazine, loxapine, or placebo. All subjects met diagnostic criteria for either AD or MID. Both active antipsychotic drugs were more effective than placebo for ratings of excitement and uncooperativeness. Although suspiciousness and hostility tended to improve more with active drugs than with placebo, substantial improvements in these two factors also were documented in subjects receiving placebo. Therefore, differences between active drugs and placebo for improvement in suspiciousness and hostility were not statistically significant. As in the Petrie study (45), Barnes et al. (7) found that only approximately one-third of patients in the active medication conditions achieved global ratings of either moderate or marked improvement. In both of these studies, classic delusions and hallucinations were present only in a minority of patients, and the heterogeneous target symptoms may have limited the ability to detect antipsychotic drug efficacy in specific subgroups of subjects with more schizophreniform delusions and hallucinations. Taken together, these three studies suggest that antipsychotic drugs are modestly effective for psychotic and nonpsychotic disruptive behaviors in AD. They also suggest that adverse effects limit their utility in these elderly subjects.

The high frequency of adverse effects in elderly dementia patients receiving low doses of antipsychotic drugs also raises the possibility that pharmacokinetic factors may be clinically important in elderly patients. Several recent studies of haloperidol pharmacokinetics in human aging suggest that pharmacodynamic factors may be more important than pharmacokinetic factors in the differential responsivity with aging to low-dose antipsychotics. Aoba et al. (4) failed to demonstrate an age effect on plasma haloperidol neuroleptic levels. Dysken et al. (23) measured haloperidol and reduced haloperidol concentrations in plasma and red blood cells in 29 inpatients with AD assigned to fixed doses of haloperidol (1, 2, or 4 mg/day for 3 weeks) to treat noncognitive behavioral problems. Although plasma concentrations achieved at these low doses were substantially below the haloperidol therapeutic window (5-12 ng/ml) postulated by Van Putten et al. (69), substantial behavioral improvement was noted in the majority of subjects. Similarly low plasma haloperidol concentrations measured by radioimmunoassay were reported by Devanand et al. (21) in 19 patients with AD treated with low dose haloperidol. These subjects included the smaller group previously reported (20) who showed substantial therapeutic (compared to placebo) as well as substantial adverse effects at these doses of haloperidol and the plasma concentrations achieved. In a recent study more directly addressing pharmacokinetic and pharmacodynamic effects of aging on response to haloperidol, Kelly et al. (36) administered intravenous haloperidol (0.014 mg/kg) in a double-blind, placebo-controlled crossover study to 13 young subjects (mean age = 27 years) and 10 older subjects (mean age = 72 years). There were no statistically significant age differences in plasma haloperidol pharmacokinetics. However, older subjects showed significantly greater decreases in cognitive function following drug administration. These data suggest important pharmacodynamic mechanisms for the increased sensitivity to haloperidol with aging. In contrast, pharmacokinetic factors with aging may be important for some other antipsychotic drugs. Movin et al. (42) found that both total and unbound remoxipride following fixed oral doses increased with age. They demonstrated a twofold increase in area under the curve of both total and unbound remoxipride concentrations in elderly subjects compared to young subjects.


When a satisfactory response to antipsychotic drugs is achieved during an acute treatment trial, the clinician must next decide how long to maintain patients on these drugs. This question recently was addressed by Risse et al. (53) in a small but informative antipsychotic drug discontinuation study in behaviorally disturbed elderly dementia patients who appeared to have benefited from an acute course of antipsychotic medications and who had then been maintained on these medications chronically. Placebo was substituted for maintenance antipsychotic medication in nine male dementia patients (mean age = 65 years) who had shown clear improvement in noncognitive behavioral problems following treatment with antipsychotic medication and who subsequently had been maintained on antipsychotic medication for at least 90 days. At the end of the 6-week placebo substitution period, only one patient had developed disruptive behaviors severe enough to warrant reinstitution of antipsychotic medication. Of the remaining eight patients, five actually were less agitated, two were unchanged, and only one was determined to be more agitated than when receiving maintenance antipsychotic medication. This small but important study supports the wisdom of periodic discontinuation of chronic antipsychotic medication to evaluate the need for long-term maintenance.



Given the limited efficacy and the frequent adverse effects of antipsychotic drugs in the elderly patient with AD, attempts to demonstrate efficacy for other types of psychotropic drugs with more benign adverse effect profiles are both reasonable and important. Unfortunately, the database derived from well-designed clinical trials of non-antipsychotic drugs for the management of disruptive behaviors in AD is even less robust than for the antipsychotic drugs. The following review, therefore, will rely heavily on anecdotal reports and non-placebo-controlled studies.



The use of benzodiazepines in patients with AD and other dementing disorders recently has been reviewed (65). In a group of "emotionally disturbed" elderly patients (mean age = 81), Sanders (59) evaluated the efficacy of oxazepam compared to placebo in an 8-week treatment trial. Oxazepam was superior to placebo, particularly for reduction of agitation and anxiety. Interpretation of this study is hampered by the vagueness of the diagnoses and the likelihood that many subjects were not demented. Coccaro et al. (13) compared oxazepam, haloperidol, and the sedating antihistamine diphenhydramine in elderly institutionalized patients. The mean age of these subjects was 75 years, most met criteria for AD, and target signs and symptoms included tension, excitement, aggressiveness, pacing, and increased motor activity. Ratings of target signs and symptoms improved over an 8-week period in all treatment groups. Although statistically significant differences among groups did not emerge, there was a trend for greater improvement with diphenhydramine and haloperidol than with oxazepam. The lack of a placebo group in this study complicates interpretation of the modest improvements in objective ratings of disruptive behaviors. A recent study (32) which used an "n of 1" design methodology (6) compared oxazepam, diphenhydramine, and the antipsychotic thiothixene in resistive dementia patients. Results were interpreted as supporting efficacy for the antipsychotic, and benzodiazepines had little to offer to behaviorally disruptive dementia patients. Salzman et al. (58) tapered and then discontinued benzodiazepines in 13 elderly nursing home residents and compared memory function and ratings of depression, anxiety, irritability, and sleep to 12 nursing home residents who continued their benzodiazepine therapy. The drug discontinuation group showed greater improvements in memory than did the drug maintenance group, and there were no differences between groups in measures of depression, anxiety, irritability, or sleep. This study suggests that at least a subgroup of patients maintained chronically on short-acting benzodiazepines may benefit from a trial of drug discontinuation. In addition to adverse effects on cognitive function, benzodiazepines have been associated with falls in geropsychiatric inpatients (2). Taken together, these studies of benzodiazepine use in behaviorally disturbed dementia patients suggest that the widespread prescription of these drugs noted in several epidemiologic studies (8, 10) needs stronger justification.



Buspirone is a partial 5-HT1A agonist which has antianxiety activity and a benign adverse effect profile. Two recent uncontrolled studies of buspirone in dementia patients with agitated behavior have been reported. Sakauye et al. (57!popup(ch134) prescribed buspirone to 10 patients with AD complicated by agitated behaviors in an open-label study. A modest but statistically significant overall reduction of agitated behaviors was noted. There was substantial variability in response, with 4 of the 10 patients demonstrating marked declines in disruptive behaviors. In a similar study, Hermann et al. (31!popup(ch134) prescribed buspirone to a group of elderly nursing home residents who exhibited heterogeneous types of dementia (AD, MID, and alcoholic dementia). All subjects had demonstrated severe behavioral disturbances including agitation and aggression, and all had failed to improve with previous trials of other types of psychotropic medications, principally antipsychotic drugs. Six of 16 patients were rated as much or very much improved in terms of agitation and aggressive behavior. In both of these studies, adverse effects were unusual. Although neither of these studies was placebo-controlled, they suggest that a subgroup of dementia patients with disruptive behaviors may benefit from buspirone. Given the low toxicity of this drug, further placebo-controlled investigations appear warranted.



Because of the clear serotonergic deficit demonstrated in AD postmortem brain tissue (28!popup(ch134) and cerebrospinal fluid (9!popup(ch134), studies addressing the behavioral efficacy of drugs which purport to enhance central serotonergic activity in AD would be of interest. Simpson and Foster (62!popup(ch134) treated four dementia patients who manifested disruptive behaviors with trazodone after antipsychotic drug treatment had proved ineffective. In this anecdotal report, trazodone in doses of 200-500 mg daily was associated with decreased agitation and aggressive behavior. Pinner and Rich (46!popup(ch134) treated seven dementia patients with trazodone for symptomatic aggressive behavior. Again, all subjects had failed to improve with antipsychotic drug therapy. Three of the seven patients demonstrated an apparent marked decrease in aggressive behavior following 4-6 weeks of trazodone at doses ranging from 200 to 350 mg/day. Olafsson (43!popup(ch134) reported a placebo-controlled trial of fluvoxamine in a mixed group of nondepressed demented patients with MID and AD. There was a trend (p = 0.08) for greater improvement with fluvoxamine than with placebo for noncognitive symptoms (irritability, anxiety, mood level, and restlessness).



Because the hyperactive and aggressive behaviors encountered in the manic phase of bipolar disorder at least superficially can resemble agitated behaviors in AD and other dementias, drugs effective in the treatment of mania have been used in open trials in behaviorally disturbed dementia patients. Marin and Greenwald (40!popup(ch134) treated two AD patients and one MID patient with carbamazepine in an attempt to reduce combative agitated behaviors. Within 2 weeks of carbamazepine treatment at doses ranging from 100 to 300 mg/day, behavioral improvement was noted in all subjects. In a larger open study in AD patients who had failed to respond to antipsychotic drugs (24!popup(ch134), reduction in hostility, agitation, and uncooperativeness was noted in five of nine patients. In this study, two patients whose agitated behaviors decreased manifested ataxia and confusion which resolved with carbamazepine dose reduction. The mean dose of carbamazepine in this study was 480 mg/day. In contrast to these enthusiastic open studies, Chambers et al. (12!popup(ch134) noted no overall benefit from carbamazepine in 19 elderly dementia patients who were prescribed carbamazepine at 100-300 mg/day. Target symptoms in this study were wandering, overactivity, and restlessness.

Sodium valproate is another anticonvulsant drug with demonstrated antimanic activity. Mellow et al. (41!popup(ch134) treated four AD patients who manifested disruptive and agitated behaviors for 1-3 months at doses of valproate ranging from 500 mg/b.i.d. to 500 mg/t.i.d. Substantial behavioral improvement was noted in two of the four patients, and adverse effects did not appear. The classic antimanic drug lithium was evaluated by Holton and George (33!popup(ch134) in an open study of 10 dementia patients who manifested disruptive behaviors. Although no improvement in disruptive behaviors was observed, the low dose of lithium (250 mg/day) administered may have been inadequate.



It is remarkable how little is really known about specific pharmacologic management of noncognitive behavioral abnormalities in AD despite the prevalence of these problems and their impact on patient management. Extrapolating from psychotherapeutic drug outcome studies in younger patients with such diseases as depression and schizophrenia is not a satisfactory way to develop a meaningful database for the pharmacologic management of noncognitive behavioral disturbances in elderly patients with dementia. The clear placebo responses noted in several carefully performed pharmacologic trials in AD patients with noncognitive behavioral problems emphasize the necessity for inclusion of a placebo condition if a study is to be interpretable. These placebo responses also suggest that nonpharmacologic behavioral and environmental strategies deserve careful evaluation in the management of behaviorally disturbed dementia patients. Critical needs in all further studies are careful phenomenologic descriptions of target behavioral abnormalities and blind randomization to active treatment and placebo conditions.

published 2000