Chapter  19:  Recognition and Initial Assessment of Alzheimer's Disease and Related Dementias: Clinical Practice Guideline No. 19.

History of Guideline Focus

The panel was originally convened to develop a guideline entitled Screening for Alzheimer's and Related Dementias. During the first 18 months, the panel devoted considerable time to analyzing the literature on screening tests. The American Psychological Association (APA) conducted the primary literature search under contract to AHCPR; its work was then reviewed and a meta-analysis undertaken. On the basis of this analysis and after considerable deliberation, the panel reached three important conclusions: (a) of the most frequently used screening tests, no evidence existed that some should be recommended over others; (b) none of the tests had a high sensitivity for early or mild dementia; and (c) there was no evidence that a general screen for Alzheimer's disease or related dementias would be efficacious, given the lack of unequivocally effective treatment and the difficulty of recognizing early dementia. These were valuable conclusions in themselves; however, they clearly dictated a redirection of the panel's work.

The panel concluded that the recognition of early dementia remained an important issue, for reasons previously stated, and changed the guideline focus to recognition and initial assessment of Alzheimer's disease and related dementias. The panel agreed to limit its scope to this subject and not to consider differential diagnosis. Work was assigned to panel subgroups whose members specialized in the topics delegated to them, and additional literature searches were conducted. Attachment B lists instruments for which specific literature searches were conducted.

Literature Search and Review

The primary and most extensive literature search, conducted by APA through January 1993, was undertaken to identify empirical studies of assessment of mental status instruments for differentiating between persons with and without dementia. A second search, from January through July 1993, was conducted to identify mental status instruments used in the assessment of persons with Alzheimer's disease. This literature was subjected to meta-analysis, as described below. The Guideline Technical Report for this guideline describes this process in greater detail. The panel conducted additional literature searches related to assessment of functional impairment to inform the development of the assessment flow chart, locate published material on risk factors for dementia, and identify relevant articles published after the date of the original searches. As a result of the peer review process, references unrelated to the meta-analyses were added through August 1995. This guideline includes a total of more than 300 references.

To assist the literature search and analysis on assessment instruments, the panel developed a list of key words or phrases (e.g., screening, assessment, cognitive, cognition, dementia, memory, neuropsychology), as well as criteria for selecting or rejecting abstracts or articles. The search process was guided by three criteria: (a) the tests should be sensitive to early or mild symptoms of dementia; (b) they should lend themselves for use in unselected populations, that is, be useful with a wide spectrum of persons with a number of comorbid conditions; and (c) they should have high clinical utility, that is, not require specialized equipment, extensive or complex procedures of scoring or administration, or extensive specialized training.

Eight inclusion criteria were identified for articles on assessment instruments. They should concern human studies, pertain to adult or elderly subjects, be written in English, assess memory complaints, be screening studies, be editorials or commentaries, be meta-analyses, and, finally, be evaluations of instrument effectiveness or discriminability. In addition, 24 criteria were used to exclude studies or records from the literature search process. If a record dealt with one or more of the following criteria, it was not selected: animal studies, children, non-English languages, etiology or pathology, biological markers, individual case studies, biochemistry, drug models, clinical trials, drug therapy, dexamethasone suppression, toxic encephalopathy, acquired immunodeficiency syndrome (AIDS), syphilis, neurosyphilis, multiple sclerosis, lumbar punctures, cerebrospinal fluid analysis, treatment, management, detection by neuroimaging, pathophysiology, physiological changes, and normal aging changes. The APA and the National Library of Medicine conducted comprehensive computerized literature searches of 20 electronic databases. These searches yielded a total of 8,198 records, which were entered into a Pro-Cite database. An additional 961 abstracts were contributed by panel members and other sources, for a total of 9,159 records.

Panelists then rated the abstracts or records in the database to identify those sufficiently promising to warrant obtaining copies of the full document. The ratings were guided by two broad criteria selected to identify the study as useful for the meta-analysis: (a) relevance to assessment of early detection of mild dementia (i.e., the study used a test, battery, interview, or technology that is at least conceivably useful for the assessment or screening of early or mild dementia) and (b) inclusion of quantitative information (i.e., the study is likely to contain empirical information on persons with dementia and a group of control subjects). On the basis of these criteria, panelists rated abstracts as 1, definitely retrieve; 2, retain citation but do not retrieve now; and 3, not needed, do not retrieve the reprint or retain the citation. The 671 abstracts with an average rating of 1.0 or 1.5 were retrieved and further evaluated.

The 671 articles were read and coded by panel project staff to obtain crucial information about the study design and findings that would be used for the meta-analytic evaluation. Only 178 of these 671 abstracts provided data relevant to the meta-analyses described below; 493 were rejected as unsuitable to provide data necessary for the meta-analysis. The two most frequent reasons for rejection were lack of control or dementia group (40 percent) and no suitable screening instrument (20 percent). Review of new studies yielded an additional 67 studies that were potentially suitable to provide data relevant to the meta-analyses. Of the 67 additional studies, 32 were judged suitable for meta-analysis.

Meta-Analysis

Meta-analysis is the method the panel chose for combining the scientific evidence on dementia assessment instruments. In meta-analysis, the index of an instrument's ability to discriminate between persons who do and do not have dementia is expressed in terms of a mean effect size, a statistic that summarizes the separation of average scores between the two groups in standard deviation units ( Hedges and Okin, 1985). An average effect size is computed for each test. Higher effect sizes correspond to greater differences between the groups in average scores and thus greater discriminative ability (Hasselblad and Hedges, 1995). Further details of the meta-analysis are found in the Guideline Technical Report.

A total of 210 articles (178 from the original search and 32 of the new studies) were included in the meta-analysis (Hedges and Okin, 1985). The initial meta-analyses were conducted on a set of 100 tests or instruments, which included subtests or items from tests or mental status questionnaires. The effect sizes computed in these meta-analyses were calculated for control subjects versus patients with dementia that varied in stage or severity of disease. Studies of persons with severe-only or moderate-to-severe dementia had larger effect sizes than studies of persons with mild-only or mild-to-moderate dementia. The severity of dementia was the only between-study difference considered in the characterization of case patients and control subjects.

Table 1. Five-phase scheme for evaluation of mental (more...)

Table 1. Five-phase scheme for evaluation of mental status tests

Phase	Case patients	Control subjects	Purpose
Uncontrolled
I	Narrow spectrum of disease		Performance of procedure
Case-Control
II	Narrow typical	Healthy	Coarse distinctions
III	Expanded spectrum	Healthy	More subtle distinctions
IV	Inclusion of appropriate comorbidity	Inclusion of appropriate comorbidity	Clinical spectrum of pertinent challenges
Prospective
V	Full specturm	Full spectrum	Cohort clinical trials

Source: Nierenberg AA, Feinstein AR. How to evaluate diagnostic marker tests. Lessons from the rise and fall of dexamethasone suppression test. JAMA 1988;259(11):1699-1702. Copyright 1988, American Medical Association. Used with permission.

Healthy control subjects consisted of community volunteers or sometimes nonaffected spouses. Control subjects with comorbidity could include patients from primary care clinics, psychiatric clinic outpatients, memory or dementia clinic outpatients, or medical or surgical patients. Characteristics of case subjects were as follows: narrow spectrum of disease included moderate to severe Alzheimer's disease; expanded spectrum included mild or mixed Alzheimer's disease; and comorbidity among case subjects was most often depression, stroke, delirium, or other conditions that affected cognitive or functional performance.

Despite the voluminous literature on the subject, very few studies have been conducted on the most informative phases (i.e., IV or V). Thus, for Phase IV studies, only one instrument, the MMSE, had more than four contrasts; the remainder of the instruments had fewer than four (most had only two contrasts). Consequently, two studies, each with a single contrast at Phase IV, represent extremely important if limited information.

Instruments for assessment of functional impairment were also subjected to meta-analysis. The 18 key instruments evaluated in the meta-analyses are listed in Attachment B. The results of the meta-analyses for selecting specific assessment instruments for mental and functional status are described in Chapter 4 (see also Attachment C).

Panel Recommendations and Strength of Evidence

After extensive review and analysis of the available scientific literature, the panel developed guideline recommendations supported by the evidence. Strength of evidence was rated based on the following criteria:

• A

• Strong Evidence. Evidence from studies that compare case patients who have dementia with control subjects who do not have dementia but do have comorbid or interfering conditions (most difficult level of discrimination).

• B

• Suggestive Evidence. The same type of evidence as in category A, but involving a smaller number of studies or a less consistent pattern of findings, or both.

• C

• Expert Opinion. Evidence from clinical experience described in the literature or derived from the consensus of panel members, or both. The panel's recommendations, strength-of-evidence ratings, and discussion of supporting evidence appear in Chapter 4.

Public Comment and Peer Review

A public meeting was announced in the Federal Register and held April 12, 1993, in Washington, DC. The purpose of this meeting was to give interested organizations, individuals, and agencies an opportunity to present written or oral testimony for the panel's consideration. The meeting elicited oral and additional written testimony from eight interested parties.

Later in the process of guideline development, drafts of the Clinical Practice Guideline were distributed for peer review. A second peer review was conducted for a subsequent draft that incorporated major revisions based on earlier review comments. Reviewers were selected from the following groups to represent a broad range of disciplines and clinical practice areas: health care professional and consumer organizations, participants in the public meeting, government agencies, and interested health care providers. A total of 109 reviewers submitted comments, which were collated and reviewed by the panel cochairs.

Cognitive Impairment

Cognitive manifestations may differ by type of dementia. Although memory problems are the most common cognitive deficits of dementia, a decline in other mental functions generally occurs. With Alzheimer's disease, for example, significant, objective, and progressive decline in other mental functions is a required component of the diagnosis.

Alzheimer's disease affects many aspects of cognition. For descriptive purposes, the changes that occur in the early stages of the disease can be divided into three groups: memory, language, and visuospatial dysfunctions (Storandt, Botwinick, Danziger, et al., 1984). It is important to note that these three areas of cognitive function also may be affected by age (as are attention, problem solving and abstraction, and general intelligence). The following discussion therefore contrasts normal aging with symptoms of possible early-stage dementia to help clinicians recognize the difference (see Chapter 4, section on triggers for recognition and initial assessment for the presence of dementia).

Memory. Memory is a complex function. For short-term or primary memory -- the ability to learn and briefly retain small amounts of information (i.e., for a few seconds or minutes, if not distracted) -- normal age-related changes are generally minimal. However, age may bring substantial changes in long-term or secondary memory, particularly the ability to learn large amounts of information and retain them after long delays. Declines in this type of memory become more frequent beginning around age 50 (Crook and West, 1990). Progressive deterioration of memory abilities is a hallmark of Alzheimer's disease. Persons with this disease, even in the earliest stages, have difficulty learning new information and retaining it more than momentarily (Duchek, Cheney, Ferraro, et al., 1991). An example is failure to remember a recent conversation. In the early stage of the disease, a person may be able to remember a phrase or telephone number long enough to repeat it but will lose the information if there is a delay (Welsh, Butters, Hughes, et al., 1991). Repeated conversations are a common result. As the disease progresses, new learning is severely curtailed, and "old" or remote memories also are gradually lost. Persons in the early stages of Alzheimer's disease have problems with the two aspects of information retrieval: prompted recall (e.g., recalling the U.S. President's name) and recognition (e.g., recognizing the President's name among others). Also severely affected, even in early stages, is personal episodic memory time-related information about oneself. An example is forgetting where one left a hairbrush after last using it. Recent research suggests that Alzheimer's disease affects some types of semantic memory or general knowledge (e.g., Mount Everest is the highest mountain in the world; see Nebes, 1989, for a review). An example is difficulty remembering names of common objects. In normal aging, semantic memory is usually well maintained; however, even in patients with Alzheimer's disease, aspects of general semantic knowledge are preserved, while distinctive features are affected (Nebes and Brody, 1988; Nebes, Norton, and Horn, 1984). Procedural memory (remembering how) does not appear to be affected by early-stage Alzheimer's disease to the same extent as declarative memory (remembering what). For example, although persons with Alzheimer's disease may have trouble remembering the name of their third-grade teacher, they are likely to remember the cursive writing skill learned from that teacher (LaBarge, Smith, Dick, et al., 1992). Some evidence suggests that persons with Alzheimer's disease of mild to moderate stage can retain new procedural learning of motor responses to relatively simple stimuli (Grafman, Weingartner, Newhouse, et al., 1990; Knopman, 1991).

Language. Language abilities are largely retained very well up to age 70. Those that decline over age 70 include word retrieval and word list generation. The ability to name an object when shown a picture of it (confrontation naming) declines in many older persons, as does the ability to generate quickly a list of words that all belong to a common category or begin with a specific letter (word fluency) ( Albert, Heller, and Milberg, 1988; Hultsch, Hertzog, Small, et al., 1992). Early in the course of Alzheimer's disease, some patients may have prominent language dysfunction (aphasia). Some studies suggest that aphasia is associated with more rapid deterioration (Kaszniak, Fox, Gandell, et al., 1978; Yesavage, Brooks, Taylor, et al., 1993). Several common language problems also may appear early in Alzheimer's disease (see Hart, 1988,for a review). An example is difficulty finding the desired words.

In early stages of the disease, persons usually maintain sentence structure (syntax) in expressive language, although it may become less complex (Heller, Dobbs, and Rule, 1992; Kemper, LaBarge, Ferraro, et al., 1993). The ability to pronounce single words and to write also remain intact until the disease becomes severe (LaBarge, Smith, Dick, et al., 1992; Nelson and O'Connell, 1978). Persons in the early stages of dementia do not appear to have much trouble with auditory comprehension, at least for short commands.

Visuospatial Abilities. Visuospatial abilities involve both production and recognition of three-dimensional or two-dimensional figures and objects. Compared with the young, many older persons show impairment in producing and recognizing three-dimensional drawings. For example, in a comparison of two groups of adults matched for visual acuity, one group with a mean age of 21 and one with a mean age of 67, the older persons were, on average, less accurate in drawing cubes or in discriminating among distorted and undistorted curves. There was considerable individual variability. The older persons were, however, equally able to copy a cube when given landmarks regarding the size of the lines (Plude, Milberg, and Cerella, 1986).

A possible variant of Alzheimer's disease involves more prominent early visuospatial dysfunction; becoming lost in a familiar neighborhood may be one of the first difficulties reported. Although some aspects of visuospatial difficulties may result from impaired memory ( Henderson, Mack, and Williams, 1989), a subgroup of persons with Alzheimer's disease also may have neuropathologic lesions that affect the visual association cortex (Hof, Bouras, Constantinidis, et al., 1990). These lesions cause problems with complex visual functions such as figure-ground discrimination, visual recognition and synthesis, and spatial localization (Mendez, Mendez, Martin, et al., 1990) and make it difficult for the person to arrange objects at home.

Functional Impairment

In dementia, functional impairment is associated with two major types of abilities -- activities of daily living (ADLs) and instrumental activities of daily living (IADLs). ADLs are self-maintenance skills such as dressing, bathing, toileting, grooming, eating, and ambulating. IADLs are more complex, higher order skills such as managing finances, using the telephone, driving a car, taking medications, planning a meal, shopping, and working in an occupation. In Alzheimer's disease, problems typically appear in the conduct of IADLs and they progress to the inability to manage even less complex ADL functions of personal care (Reisberg, Ferris, and Franssen, 1985). Problems in performing the more complex IADLs may be the first indication of the disease to the patient or family.

Relationship Between Cognitive Impairment and Functional Performance. In the early stages of dementia, when the physician is attempting to make a first diagnosis, the patient may demonstrate different levels of cognitive and functional competence. Examples are driving without getting lost or taking medications without difficulty but then forgetting three-fourths of the items in a delayed-recall task. In more advanced stages of dementia, a clear relationship exists between the extent of cognitive impairment and functional performance ( Bassett and Folstein, 1991; Heaton and Pendleton, 1981), although only a few studies have investigated the precise character of this relationship.

From cross-sectional research, Vitaliano, Breen, Albert, et al. ( 1984) reported that memory and attention deficits largely account for functional impairment in dementia patients. A followup study (Vitaliano, Russo, Breen, et al., 1986) showed that attention deficits predicted decline in functional performance for both ADLs and IADLs, even after the investigators controlled for initial levels of functioning. More recently, Henderson, Mack, and Williams (1989) found that visuoconstructive deficits documented by neuropsychological evaluation predicted spatial disorientation (wandering, getting lost, failing to recognize familiar environments), as reported by the caregivers of dementia patients.

Functional Assessment Instruments. Because dementia symptoms usually appear first in IADL tasks that reflect cognitive deficits (e.g., inappropriate dressing for occasion or weather) rather than physical deficits, measurement of functional performance is warranted. Information on functional performance contributes importantly to the assessment of dementia, particularly the detection of mild forms of disease (Morris, McKeel, Storandt, et al., 1991). Assessment instruments that measure higher order social and instrumental activities are considered especially helpful when used in the context of a comprehensive evaluation for early dementia ( Barberger-Gateau, Commenges, Gagnon, et al., 1992; Wilder, Gurland, Chen, et al., 1994).

Formal assessment of functional health emerged in the 1960s with the Index of Activities of Daily Living (Katz, Ford, Moskowitz, et al., 1963) and the Physical Self-Maintenance Scale (PSMS) and Lawton-Brody Instrumental Activities of Daily Living Scale (Lawton and Brody, 1969). More recently developed instruments include the Functional Activities Questionnaire (FAQ) (Pfeffer, Kurosaki, Harrah, et al., 1982) and the Structured Assessment of Independent Living Skills (SAILS) ( Mahurin, DeBettignies, and Pirozzolo, 1991).

Implications of Functional Decline. Decline in functional abilities is one of the most troubling aspects of dementia for patients (Cotrell and Schulz, 1993) and their caregivers (Green, Mohs, Schmeidler, et al., 1993). Early in the dementing process, the loss of ability to function adequately may cause the patient to feel useless, dependent, and burdensome ( Cotrell and Schulz, 1993). For families of persons with more advanced stages of dementia, the major problems of care include impairments of communication, eating and bathing, and wandering (Rabins, Mace, and Lucas, 1982). Incontinence is also a significant problem that causes stress for the patient and caregiver.

Caregivers have pragmatic concerns about the patient's ability to function independently, which include issues of safety, such as the ability to drive, follow medication directions, and use household appliances (e.g., remember to turn off an oven or stove burner). In addition to evaluating a patient's abilities in these areas, health care providers are often called upon to render judgments about functional competence (Loewenstein, 1990). Detailed information about the functional performance of persons with dementia is important for planning their care needs (Green, Mohs, Schmeidler, et al., 1993; Mahurin, DeBettignies, and Pirozzolo, 1991) and for legal and financial matters, including consent to treatment and competence to manage financial and other affairs. To structure care that is appropriate to each patient's current condition, it is important that functional performance be assessed with some regularity. Scheduled followup visits can help in this process of monitoring patients over time. Attachment D provides a list of national resources for patients, family members, and health care professionals to assist in some of these planning issues.

Behavioral Symptoms

Behavioral disturbances are prevalent in mild Alzheimer's disease (Rubin, Morris, Storandt, et al., 1987; Wild, Kaye, and Oken, 1994); other dementing disorders are also marked by a number of behavioral symptoms. These include personality changes, agitation, and symptoms of psychopathology such as mood disturbances, delusions, hallucinations, and misidentifications. Some of the behavioral disturbances such as agitation and psychotic symptoms usually are not found in early Alzheimer's disease but may be prominent in frontal lobe dementia syndromes. The physical changes in the brain that cause dementia evoke strong emotional responses, both in the patient and in those exposed to the condition. As the dementing disorder becomes evident in behavior changes, the patient and family members often experience considerable distress they may not be able to express. The patient may become frustrated by advancing debility; those who care for and about the patient are likely to be upset by signs of deterioration and behavior at odds with the person they know. The progressive nature of dementia over many years can create situations of unrelenting stress. Encouraging affected patients and families to talk about emotions they may not be able to identify can help them develop coping skills. It also may avert abusive situations that result when people react blindly to feelings they cannot express.

Additionally, when health care professionals are able to learn the personal characteristics that have defined and given meaning to patients' lives -- hobbies, professional accomplishments, family ties, expressions of faith-they can better understand and address patient and family frustrations as functional losses become evident. This knowledge informs decisions about patient care and helps clinicians respect the patient's wishes, even when he or she can no longer express preferences.

Like cognitive and functional deficits, behavioral disturbances vary considerably among patients. There is evidence that agitation, hallucinations, or delusions are associated with increased risk for nursing home placement (Steele, Rovner, Chase, et al., 1990) and may be associated with more rapid progression of cognitive impairment (Stern, Albert, Brandt, et al., 1994; Teri, McCurry, Edland, et al., 1995). Depression may be associated with increased mortality in patients with Alzheimer's disease (Burns, Lewis, Jacoby, et al., 1991).

In early or mild dementia, the most common behavioral symptoms include signs of depression and what can be considered broadly as personality changes (Rubin and Kinscherf, 1989; Rubin, Morris, Storandt, et al., 1987). Psychotic symptoms and clinically significant agitation seem to be fairly infrequent in early or mild dementia and seem to increase in frequency as the disease progresses to severe cognitive impairment (Reisberg, Franssen, Sclan, et al., 1989; Rubin and Kinscherf, 1989; Spector and Jackson, 1994). Reports are less consistent for prevalence of depression as a function of disease severity, and available data vary depending on the assessment methods used and the nature of the depression being considered (Logie, Murphy, Brooks, et al., 1992; Wild, Kaye, and Oken, 1994).

Recent studies comparing levels of psychiatric symptoms in persons with vascular dementia and those with Alzheimer's disease have reported mixed findings (e.g., Cohen, Eisdorfer, Gorelick, et al., 1993; Corey-Bloom, Galasko, Hofstetter, et al., 1993; Sultzer, Levin, Mahler, et al., 1993). Distinguishing Alzheimer's dementia from vascular dementia is often very difficult. Many patients have both dementias, making the clinical diagnosis in such cases difficult if not impossible. In a sample of persons matched for severity of dementia, Sultzer and colleagues (1993) found that those with vascular dementia had more severe symptoms of depression, anxiety, and behavioral retardation (e.g., blunted affect, emotional withdrawal) than those with Alzheimer's disease. [In DSM-IV, nomenclature for "multi-infarct dementia" has been changed to "vascular dementia."] Cohen and colleagues (1993) found similar levels and patterns of psychiatric symptoms in persons with Alzheimer's disease and those with vascular dementia. However, persons with mixed Alzheimer's disease and vascular dementia had higher numbers of behavioral symptoms (e.g., wandering, abusiveness) than those with either Alzheimer's disease or vascular dementia alone.

Personality Changes. Although personality changes are among the most common behavioral problems in persons with dementia, they have been among the least intensively investigated. The available literature (e.g., Strauss, Pasupathi, and Chatterjee, 1993) suggests that assessing current and premorbid personality is feasible in persons with dementia. Data are inadequate, however, to estimate the sensitivity or the specificity of personality changes as a means of identifying persons with dementia. Family members often report that personality changes occur with Alzheimer's disease, but supporting data are retrospective in nature (Chatterjee, Strauss, Smyth, et al., 1992; Siegler, Welsh, Dawson, et al., 1991).

Rubin and colleagues (Rubin and Kinscherf, 1989; Rubin, Morris, and Berg, 1987) studied groups of persons with questionable and mild Alzheimer's disease and controls of similar age. They found that symptoms of passivity at times occurred by themselves, but agitation and self-centered behavior occurred most frequently in combination with symptoms of passivity. Rubin, Morris, and Berg (1987) evaluated these symptoms in this population as the disease progressed from early through moderate to severe dementia. For the respective severity levels, passive behavior increased in prevalence from 66 to 75 to 88 percent; agitated behaviors increased significantly from 30 to 46 to 67 percent; and self-centered behaviors increased from 34 to 58 to 63 percent.

Using an instrument designed for persons with head injury, Petry, Cummings, Hill, et al. (1988) compared spouses' ratings of personality changes in patients with Alzheimer's disease with changes related to retirement in control subjects. They reported that changes occurred regularly in the persons with Alzheimer's disease. These subjects were rated as having become more out of touch, more unreasonable, more lifeless, more unhappy, more irritable, less self-reliant, less mature, less enthusiastic, less stable, less affectionate, less kind, and less generous than before the onset of the disease. In contrast, there were no significant changes in control subjects. In the Alzheimer's disease group, only an increase in excitability was correlated with decreasing score on the Mini-Mental State Examination (MMSE). When the sample of persons with Alzheimer's disease was followed longitudinally, certain traits appeared to increase over time, such as listlessness, insensitivity, and coldness (Petry, Cummings, Hill, et al., 1989). Other traits increased and then plateaued, such as being out of touch, childish, reliant on others, unreasonable, lifeless, and uninterested in company.

Burns, Folstein, Brandt, et al. (1990) developed a scale to assess irritability, aggression, and apathy in persons with dementia. They reported that 58 percent of 31 persons with Alzheimer's disease were irritable, 32 percent were aggressive, and 48 percent were apathetic, with no relationship between symptomatology and score on the MMSE.

Both Siegler, Welsh, Dawson, et al. (1991) and Chatterjee, Strauss, Smyth, et al. (1992) investigated personality changes using relatives' ratings of patients on the NEO Personality Inventory ( Costa and McCrae, 1989). Siegler and colleagues (1991), evaluating a series of persons with heterogeneous dementias or amnestic disorders, reported that they had become less conscientious, less extroverted, more neurotic, slightly less open, and more antagonistic. There was no correlation between these changes and MMSE scores. In a group of persons with Alzheimer's disease, Chatterjee and colleagues (1992) found a major decrease in conscientiousness followed by lesser increases in neuroticism and decreases in extroversion; changes in openness and agreeableness were smaller. Of these personality changes, only introversion was inversely associated with changes in MMSE scores; the other changes did not show a statistically significant association with the MMSE score.

Agitation. The most prevalent and the most intensively investigated behavioral disturbances in persons with dementia are states of agitation. Agitation poses a major clinical challenge in patients and in care settings such as nursing homes. Cohen-Mansfield, Marx, and Rosenthal (1990) investigated the phenomenology of dementia in nursing home residents; factor analysis suggested that agitation includes physically aggressive, physically nonaggressive (i.e., agitated behaviors, wandering), and verbally agitated behaviors. Although agitation occurred in both cognitively intact and impaired nursing home residents, the symptoms differed. Cognitively intact residents exhibited primarily verbal agitation through symptoms that resembled exaggerations of coping mechanisms (e.g., complaining); those with dementia were more likely to show both physically nonaggressive and physically aggressive symptoms.

As reviewed by Deutch and Rovner (1991), reports on the frequency of agitation in persons with Alzheimer's disease from the community attending a dementia clinic varied from 24 to 48 percent; the frequency of physically nonaggressive behaviors was estimated at between 33 and 50 percent; physically aggressive behaviors, at approximately 30 percent; and verbally agitated behaviors, at 24 percent. The frequency of agitated behavior increases with the progression of Alzheimer's disease, with the highest rates in those with moderate or severe disease and a possible decrease in those with end-stage, extremely severe dementia ( Reisberg, Franssen, Sclan, et al., 1989).

For caregivers at home, coping on a daily basis with agitation and other behavioral disturbances can engender mixed feelings, such as helplessness, distress, and anger toward the patient. Health care professionals can give family caregivers a valuable outlet for such emotions and acknowledge that they are natural responses to a highly stressful situation.

Symptoms of Psychopathology. Some behavioral symptoms have been more widely investigated, possibly because they are more obvious sources of disability, result in distress and danger to both affected persons and their caregivers, and are potentially treatable components of the dementia syndrome. These include mood disturbances and psychotic symptoms.

Mood Disturbances.Mood disturbances in dementia include both major depression and less severe but significant depressive states, such as depressive symptoms or "minor" depression, affective lability, and manic or hypomanic states (see Chapter 4 for a more extensive discussion of dementia and depression).

Wragg and Jeste (1989) summarized findings from 18 studies on depression in Alzheimer's disease. They reported that the frequencies of depressive symptoms ranged from 0 to 87 percent with a median of 41 percent and modal frequency between 40 and 50 percent. One study cited, Lazarus, Newton, Cohler, et al. ( 1987), comparing persons with Alzheimer's disease and control subjects, found a depression frequency of 40 percent in the Alzheimer's disease group and 12 percent in controls. Five studies reported a 19-percent median frequency of specific depressive disorders, primarily major depression, with most studies clustered between 10 and 20 percent. Additional studies reviewed by Burns (1991) do not alter these estimates.

Psychotic Symptoms. Wragg and Jeste (1989) also reviewed 21 studies of the prevalence of behavioral problems such as hallucinations and delusions in persons with Alzheimer's disease. The reported frequency of delusions at some time during the course of this illness ranges from 10 to 73 percent, with a median of 33.5 percent. Most reports show a frequency clustered between 30 and 38 percent. Reports on hallucinations range from 21 to 49 percent, with a median of 28 percent. Other symptoms such as misidentification of objects or people also are relatively common but have not been as intensively investigated. Although there is no clear consensus on the extent to which the prevalence of psychotic symptoms varies as a function of the stage of dementia or the degree of cognitive impairment, available data suggest that the frequency of psychotic symptoms at the earliest stages of the disease is low. Rubin and Kinscherf (1989), for example, report that delusions occurred in none of a group of control subjects without dementia, in 2 percent of those with questionable dementia, and in 21 percent of those with mild dementia. They report that hallucinations occur in 0, 0, and 13 percent, respectively; that misidentifications occur in 0, 0, and 12 percent; and that any psychotic symptoms occur in 0, 2, and 29 percent.

Estimates of Direct Costs

To identify the direct costs of diagnosing, managing, and living with a health condition, one must know how many persons have the condition and the types, number, and costs of services provided. The number of persons with Alzheimer's disease is very difficult to determine. Estimates range from 1.35 million (Ernst and Hay, 1994) to 3.75 million (Evans, Scherr, Cook, et al., 1990). Other studies have yielded different estimates ( Arenberg, 1990; Sapir, 1988). Estimates of prevalence of dementia vary among studies principally because of differences in diagnostic criteria and how they are implemented. Other reasons include whether they (a) are retrospective or prospective, (b) screen subjects or perform chart reviews, and (c) use test results alone or perform clinical assessments.

For a nationally representative study, the most efficient means of identifying cases is the diagnosis code of health care claims data; however, Alzheimer's disease, even when present, is rarely coded in available databases. In a random sample of 5 percent of 1992 Medicare claims submitted for inpatient services (acute care, long-term care, skilled nursing facilities, rehabilitation hospitals, and psychiatric hospitals), only six cases among approximately 700,000 claims were identified with an ICD-9-CM (International Classification of Diseases, 9th revision, Clinical Modification) code of 331.0 (Alzheimer's disease) (Weis, 1992).

Alzheimer's disease is not coded for many reasons (Torian, Davidson, Fulop, et al., 1992), including the coding of other acute conditions as reasons for hospitalization ( Congress of the United States, 1989). Without the means to identify cases reliably, one cannot use secondary databases to analyze the cost of diagnosing Alzheimer's disease. A prospective study is needed, and the cost of conducting a study that would provide reliable national estimates is great.

Costs of services are also difficult to determine. In addition to diagnosis, services needed by dementia patients and their families may include management, acute care, home- and community-based services, information and support for caregivers, respite care, legal and financial services, and nursing home care. The specific medical and long-term services that patients require are determined by factors such as stage and manifestations of disease and availability of caregivers in the home (Davis and Neuman, 1986).

Estimates of Total Costs

Either prevalence-based or incidence-based estimates are used to estimate the costs of illness. Prevalence-based estimates look at the direct and indirect economic burden incurred during a specific period of time (base period, often a year) as a result of the prevalence of a disease or illness. This includes the value of resources used or lost, regardless of time of onset of the disease. Incidence-based estimates, in aggregate, refer to the total lifetime costs of all cases of the disease or illness with onset in a given base year. Of the several studies that have attempted to estimate the indirect costs of informal care of Alzheimer's disease and other dementias (Hay and Ernst, 1987; Huang, Cartwright, and Hu, 1988; Rice, Fox, Max, et al., 1993), all but the Hay and Ernst study used the prevalence approach.

Rice and Max (1993) reviewed 19 studies of costs of aging-related diseases, converting the estimated costs of Alzheimer's disease and related dementias in those studies to 1991 dollars. When they converted data from the Hay and Ernst ( 1987) study (incidence data), total costs were $54 billion; the converted 1991 figure for costs from the Huang et al. (1988) study (prevalence data) totaled $120.8 billion. The substantial variation in estimated costs reflects the different approaches to cost estimation, methodologies, and sources of data.

The cost studies cited here attempt to capture the economic burden of dementia. Less quantifiable, but staggering, are the emotional and psychological costs dementia exacts from patients and their families. Along with the costs to society of Alzheimer's disease and other dementing conditions, it is important to acknowledge the profound nonfinancial toll, which may escalate over a period of many years.

Family History of Dementia

Data from family studies vary by type of dementia and are complicated by confounding variables. Evidence suggests that by age 90, about one-half of the first-degree relatives of dementia patients develop dementia (Breitner, Silverman, Mohs, et al., 1988; Mohs, Breitner, Silverman, et al., 1987). In their reanalysis of 11 case-control studies, van Duijn, Clayton, Chandra, et al. (1991) reported an adjusted pooled odds ratio of 3.5 for family history. The risk is highest if a sibling was affected with Alzheimer's disease and increases with the number of first-degree relatives affected (Graves, White, Koepsell, et al., 1990).

Genetic studies based on the evidence from family histories show that genetic factors contribute to the development of Alzheimer's disease. Autosomal-dominant forms of the disease typically appear at young ages and are usually associated with early death. At least three autosomal-dominant forms of the disease, all with early-onset symptoms, have been identified: a common one on chromosome 14 (Schellenberg, Bird, Wijsman, et al., 1992; Sherrington, Rogaev, Liang, et al., 1995) and rare ones on chromosome 1 (Levy-Lahad, Wijsman, Nemens, et al., 1995) and chromosome 21.

Recent studies suggest an association between alleles of the apolipoprotein E (apoE) gene locus on chromosome 19 and either increased risk for late-onset Alzheimer's disease (apoE 4) or protection against this risk (apoE 2 or 3) (Corder, Saunders, Risch, et al., 1994; Nathan, Bellosta, Sanan, et al., 1994; Poirier, Davignon, Bouthillier, et al., 1993; Roses, Strittmatter, Pericak-Vance, et al., 1994; Saunders, Schmader, Breitner, et al., 1993). Different studies have shown varying levels of association between the different apoE genotypes: 30 to 40 percent prevalence of Alzheimer's disease in persons with the homozygous apoE 4 genotype and as low as 15 to 20 percent prevalence in older persons who have either the apoE 2 or apoE 3 genotype.

It seems clear that other factors, such as other genes, life experiences, and environmental factors, must be involved to result in late-onset Alzheimer's disease. As Roses and colleagues ( 1994) suggest, it is not yet possible to depend on apoE genotyping for definitive guidance about diagnosis or treatment of Alzheimer's disease.

Presence of Down Syndrome

Most persons with Down syndrome develop the characteristic neuropathological brain lesions of Alzheimer's disease (i.e., neuritic plaques, granulovacuolar changes, cerebral vascular amyloidosis, Hirano bodies, and neurofibrillary tangles) by the age of 40 years ( Wisniewski, Wisniewski, and Wen, 1985), although many do not demonstrate clinical features of dementia. According to Mortimer and Hutton (1985), almost all autopsies of persons with Down syndrome who have survived to at least 40 years of age reveal brain lesions typical of Alzheimer's disease. A recent study of apoE genotypes in persons with Down syndrome and in their first-degree relatives suggests that having the apoE 2 or 3 genotype may protect against what appeared to be an inevitable development of Alzheimer's disease, rather than the apoE 4 genotype being a causative factor ( Hyman, West, Rebeck, et al., 1995; Royston, Mann, Pickering-Brown, et al., 1994). If this observation is confirmed, it could offer clues to the type of compound, analogous to apoE 2 or 3 gene products, that might prevent or delay the onset of Alzheimer's disease in susceptible persons.

Other Possible Risk Factors

There is evidence for a modest but significant association between Alzheimer's disease and earlier head trauma with loss of consciousness (Mortimer, van Duijn, Chandra, et al., 1991). This reported association may be explained by recall bias -- that is, overreporting by affected persons and underreporting by control subjects.

A low level of education is another suggested risk factor for dementia, but the evidence is mixed and suggests complex explanations. Although some studies find a strong association (Fratiglioni, Grut, Forsell, et al., 1991; Zhang, Katzman, Salmon, et al., 1990), other studies find none (Folstein, Bassett, Anthony, et al., 1991; Heyman, Fillenbaum, Prosnitz, et al., 1991). However, persons with limited formal education or occupational experience may be at higher risk for early detection ( Amaducci, Falcini, and Lippi, 1992). They also may be at higher risk for false-positive diagnosis if a baseline evaluation is not available to demonstrate cognitive decline.

In summary, the only reasonably well-established risk factors for Alzheimer's disease are age, family history of dementia, and the presence of Down syndrome. Even though the incidence of Alzheimer's disease increases exponentially with age, it does not seem warranted to conclude that age alone is a risk factor that justifies steps to detect or rule out dementia.

Reversible Dementias

Despite the relatively modest percentages represented by reversible causes of dementia, as well as treatable aspects of nonreversible dementias, it is important to recognize them. The potential to reverse the condition or halt or delay deterioration is among the most compelling justifications for early recognition of a dementing disorder. Knowledge of the patient and alertness to changes suggesting dementia can trigger timely actions or referrals that enhance quality of life and, in the case of delirium, even save lives.

As noted, among the most frequent reversible causes of dementia are depression (depressive pseudodementia), alcohol abuse, and drug toxicity. Katzman, Lasker, and Bernstein (1988) reviewed diagnoses in nine published series of clinical cases and found that drug toxicity was the most common cause of reversible dementia (2.7 percent of all cases). Other causes of dementia, such as normal-pressure hydrocephalus, neoplasms, metabolic disorders, trauma, and infection, occurred less often.

Few studies of potentially reversible dementia have provided followup of identified cases to determine if the patients improved with treatment. Of the 32 studies analyzed by Clarfield (1988), 11 had followup. In the approximately 13 percent of patients with potentially reversible causes of dementia, only 11 percent actually improved with treatment, and only 3 percent had complete reversal.

Alcohol. Of the several toxins that can produce dementia, alcohol is associated with the highest frequency of dementia. The increasing incidence of alcoholism in older persons makes this an important consideration. Prolonged, heavy ingestion of alcohol may result in an amnestic disorder with cognitive deficits limited to memory impairment that is not a classic dementia by DSM-IV criteria (American Psychiatric Association, 1994; Brust, 1993; O'Brien, 1994; Victor, Adams, and Collins, 1989). This disorder, Wernicke-Korsakoff syndrome, is generally attributed to associated thiamine deficiency. Alcohol amnestic syndrome often follows an acute episode of Wernicke's encephalopathy; if the Wernicke's encephalopathy is treated early, alcohol amnestic syndrome could be prevented. Once this syndrome is established, there is generally only slight improvement over time.

A dementia that does meet DSM-IV criteria, alcohol-induced persisting dementia, has also been associated with prolonged, heavy ingestion of alcohol (American Psychiatric Association, 1994; Brust, 1993; O'Brien, 1994). This dementia persists after alcohol intake has stopped, and the etiologic role of alcohol in this disorder is controversial. Other nutritional disorders, such as multiple vitamin, protein, and calorie deficiencies, occur commonly in chronic alcoholics and have been linked to cognitive impairment. Correcting these deficiencies may improve cognitive function to some extent.

Table 5. Some medications that may cause cognitive (more...)

Table 5. Some medications that may cause cognitive impairment[a]

Type of medication	Generic name	Common trade name(s)
Anticholinergic agents	scopolamine	Transderm Scop, Isopto-Hyoscine
	orphenadrine	Norflex, Norgesic[b], Norgesic Forte[b]
	atropine	various, Lomotil[c]
	trihexyphenidyl	Artane
	benztropine	Cogentin
	meclizine	Antivert, Bonine
	homatropine	Isopto-Homatropine, Hycodan[c]
Antidepressants	amitriptyline	Elavil, Endep, Etrafon[c], Triavil[c], Limbitrol[c]
	imipramine	Tofranil
	desipramine	Norpramin
	doxepin	Sinequan
	trazodone	Desyrel
	fluoxetine	Prozac
Antimanic agents	lithium	Eskalith, Lithobid, Lithotabs
Antipsychotic (neuroleptic) agents	thioridazine	Mellaril
	chlorpromazine	Thorazine
	fluphenazine	Prolixin
	prochlorperazine	Compazine
	trifluperazine	Stelazine
	perphenazine	Trilafon, Etrafon[c], Triavil[c]
	haloperidol	Haldol
Antiarrhythmic agents (oral)	quinidine	Quinidex, Quinaglute
	disopyramide	Norpace
	tocainide	Tonocard
Antifungal agents	amphotericin B	Fungizone
	ketoconazole	Nizoral
Sedative/hypnotic agents
Benzodiazepine derivatives	diazepam	Valium, Valrelease
	chlordiazepoxide	Librium, Libritabs, Librax[c]
	lorazepam	Ativan
	oxazepam	Serax
	flurazepam	Dalmane
	triazolam	Halcion
	alprazolam	Xanax
Barbiturate acid derivatives	phenobarbital	various, Donnatal[c]
	butabarbital	Butisol
	butalbital	Fiorinal[b] [c], Fioricet[c], Esgic[c]
	pentobarbital	Nembutal
Chloral & carbamate derivatives	chloral hydrate	Noctec, Aquachloral
	meprobamate	Miltown, Equanil, Equagesic[b]
Antihypertensive agents
Beta adrenergic antagonists	propranolol	Inderal, Inderide[c]
	metoprolol	Lopressor
	atenolol	Tenormin
	timolol	Timoptic
Alpha-2 agonists	methyldopa	Aldomet, Aldoril[c]
	clonidine	Catapres, Catapres-TTS, Combipres[c]
Alpha-1 antagonists	prazosin	Minipress
Calcium channel blockers	verapamil	Calan, Isoptin
	nifedipine	Procardia, Adalat
	diltiazem	Cardizem, Cardizem CD
Inotropic (cardiotonic) agents	digoxin	Lanoxin, Lanoxicaps
Corticosteroids	hydrocortisone	Cortef, Cortisporin[c], Neo-Cortef[c], Cortaid
prednisone	Deltasone, Prednisone Intensol
methylprednisone	Medrol, Solu-Medrol
dexamethasone	Decadron, Neo-decadron[c]
Nonsteroidal anti-inflammatory agents	ibuprofen	Motrin, Rufen, Advil, Nuprin, Medipren
	naproxen	Naprosyn, Anaprox, Aleve
	indomethacin	Indocin
	sulindac	Clinoril
	diflunisal	Dolobid
	choline magnesium trisalicylate	Trilisate, Tricosal
	aspirin	various
Narcotic analgesics	codeine	Tylenol with Codeine[c], Robitussin AC[c], Brontex[c], other codeine cough preparations
	hydrocodone	Lortab[c], Lorcet[c], Vicodin[c], Hycodan[c], Hycomine[c], Tussionex[c]
	oxycodone	Percodan[b], Percocet[c], Tylox[c], Roxicet[c]
	meperidine	Demerol, Mepergan[c]
	propoxyphene	Darvon, Darvon-N, Darvocet-N[c], Wygesic[c], Darvon Compound[b]
Antibiotics	metronidazole	Flagyl, Metrogel
	ciprofloxacin	Cipro
	norfloxacin	Noroxin
	ofloxacin	Floxin
	cefuroxime	Zinacef, Ceftin
	cephalexin	Keflex
	cephalothin	Keflin
Radiocontrast media	metrizamide	Amipaque
	iothalamate	Conray
	diatrizoate	Hypaque, Renovist
	iohexol	Omnipaque
H[2] receptor antagonists	cimetidine	Tagamet, Tagamet HD
	ranitidine	Zantac
	famotidine	Pepcid
	nizatidine	Axid
Immunosuppressive agents	cyclosporine	Sandimmune
	interferon	Intron A, Roferon A, Actimmune
Antineoplastic agents	chlorambucil	Leukeran
	cytarabine	Cytosar-U
	interleukin-2
	spirohydantoin mustard	Spiromustine
Anticonvulsants	phenytoin	Dilantin
	valproic acid	Depakene, Depakote
	carbamazepine	Tegretol
Anti-Parkinsonian agents (see also anticholinergic agents)	levodopa levodopa/carbidopa bromocryptine pergolide	Larodopa Sinemet Parlodel Permax
Antiemetics	prochlorperazine	Compazine
	metoclopramide	Reglan
	hydroxyzine	Atarax, Vistaril
	promethazine	Phenergan
	trimethobenzamide	Tigan
	diphenhydramine	Benadryl, Dramamine
	meclazine	Antivert
Skeletal muscle relaxants	cyclobenzaprine	Flexaril
	methocarbimol	Robaxin
	carisoprodol	Soma, Soma Compound[b]
	baclofen	Lioresal
	chlorzoxazone	Parafon Forte, Paraflex
Antihistamines/decongestants	diphenhydramine	Benadryl, Tylenol PM[c], Sominex, other OTC cough/cold preparations
	chlorpheniramine	Chlor-Trimeton, Deconamine[c], Contac[c], Tylenol Cold[c], Hycomine[c], other OTC cough/cold preparations[c]
	brompheniramine	Dimetane, Dimetapp[c], Drixoral[c], other OTC cough/cold preparations[c]
	pseudoephedrine	Sudafed[c], Actifed[c], Robitussin PE[c], Dimetapp[c], Entex[c], Drixoral[c], Tylenol Cold[c], Claritin- D[c], other OTC cough/cold preparations[c]
	phenylpropanolamine	Ornade[c], Triaminic[c], Poly- Histine[c], Hycomine[c], other OTC suppressant preparations[c]

[a] These are examples only; new medications appear regularly.

[b] These compounds contain aspirin.

[c] These compounds may contain other active ingredients.

Source: Guideline panel.

Psychiatric Disorders. Clinically significant numbers of patients who present with apparent dementia have a major psychiatric disorder that accounts for their cognitive problems. Estimates range from 1 to 31 percent of persons diagnosed with progressive dementing illness (Katzman, Lasker, and Bernstein, 1988). Depression is often misdiagnosed as dementia and vice versa (see the section in Chapter 4, Assessing for Delirium and Depression). The AHCPR-sponsored guideline, Depression in Primary Care: Volume 1. Detection and Diagnosis, offers a detailed discussion of diagnosis and treatment of depression (Rush, Golden, Hall, et al., 1993). Other psychiatric conditions that sometimes present with or mimic manifestations of dementia include schizophrenia (especially the paranoid form), bipolar disorders, and severe personality disorders. Given the frequency of a dementia misdiagnosis involving a psychiatric problem, followup monitoring and continuity of care are especially valuable in ensuring that the patient has been correctly diagnosed and is receiving suitable treatment. Indeed, the fact that psychiatric disorders can mimic dementia is one reason for undertaking the steps involved in initial assessment of an apparent dementing disorder.

Normal-Pressure Hydrocephalus. This is a disorder in which an abnormal accumulation of fluid in the cerebral ventricles compresses the brain. It is important as a diagnostic consideration because it is one of the causes of potentially reversible dementia (Arnold and Kumar, 1993). Although normal-pressure hydrocephalus is generally reported to account for less than 2 percent of all cases of dementia, some series indicate that it accounts for up to 6 percent of all cases (Katzman and Rowe, 1992; Wells, 1977). It occurs in middle-aged and older persons and is associated with dementia consisting of memory loss, confusion, slowness to respond, and paucity of thought that occurs over the course of weeks to years. Gait apraxia and urinary incontinence are also usually present. Although the condition may result from subarachnoid hemorrhage, trauma, or meningeal infection, most patients have no known precipitating illness or event.

Nonreversible Dementias

Vascular Dementia. Vascular dementia, generally cited as the second most common of the nonreversible dementias, results from loss of neuronal tissue through death of brain tissue caused by interruption of blood supply, for example, by atherosclerotic lesions or from emboli. Hypertension with resultant cerebral infarctions is the most important, recognized, and preventable risk factor for vascular dementia; emboli from the heart or elsewhere within the vascular system are other causes. Vascular dementia frequently presents with a stepwise progression of symptoms, each with an abrupt onset, often in association with a neurologic incident. Focal neurologic findings, changes in tone and reflexes, and pseudobulbar palsy are common features of the illness. Vascular dementia is generally classified as a nonreversible dementia; however, appropriate treatment of hypertension and other risk factors for cerebrovascular disease is critical for prevention of further damage.

Pick's Disease. Pick's disease is a neurodegenerative disease characterized by severe cortical atrophy, most prominently of the frontal and temporal lobes. Persons with Pick's disease commonly show early signs of severe frontal lobe or temporal lobe dysfunction, characterized by general decline in mental function, changes in behavior patterns, and lack of insight. Later, loss of retentive memory, loss of language functions, and prominent grasp and sucking reflexes may occur. The disease is relentlessly progressive, generally over the course of 2 to 7 years, with the later stages characterized by features attributable to damage of the basal ganglia. The condition is usually sporadic but can occur in families.

Parkinson's Disease. Parkinson's disease is a progressive neurodegenerative disorder characterized by rigidity, bradykinesia, tremor, and postural instability resulting in disturbances of speech, gait, and coordination. Dementia associated with Parkinson's disease is insidious in onset and heralded by disorientation at night; its etiology has not been determined (Cedarbaum and Gancher, 1992; Yahr and Bergmann, 1986). Some cognitive impairment occurs in a large percentage of persons with Parkinson's disease; estimates range from 22 to 40 percent (Celesia and Wanamaker, 1972; Sroka, Elizan, Yahr, et al., 1981). Overall, Parkinson's disease accounted for about 1 percent of all instances of dementia in the studies analyzed by Clarfield (1988) and in the neuropathologic series of Katzman and Rowe (1992).

AIDS-Related Dementia. Although AIDS-related dementia (also referred to as "AIDS-related complex") is uncommon in older persons, it warrants diagnostic consideration for patients with a suspected history of risk factors or behaviors, such as pre-1985 transfusions, history of unprotected sex, and injection drug use. As described in the DSM-IV (American Psychiatric Association, 1994), where it is called "dementia due to HIV disease," AIDS-related dementia is characterized by forgetfulness, slowness, poor concentration, and difficulties with problem solving. Prominent behavioral symptoms include apathy and social withdrawal. Physical examination may reveal tremor, impaired rapid repetitive movements, imbalance, ataxia, hypertonia, generalized hyperreflexia, positive frontal release signs, and impaired pursuit and saccadic eye movements (Arendt, Hefter, Neuen-Jacob, et al., 1993; Kaeming and Kaszniak, 1989). Although not designed to address AIDS-related dementia, the algorithm for initial assessment presented in the Figurein Chapter 4 might be considered during diagnosis, particularly for patients at high risk.

Knowledge of the risk factors and causes of dementing disorders, particularly the most common forms, is a valuable tool in the conduct of an initial assessment for suspected dementia, as described in the next chapter.

Focused History

The medical history should include inquiries about relevant systemic diseases, psychiatric disorders, and known neurological disorders, including history of head trauma. It also should include any history of alcohol or substance abuse and exposure to environmental toxins. Because a variety of medical conditions can cause or contribute to cognitive impairment, a careful review is indicated of any intercurrent, infectious, or metabolic illness, such as pneumonia, urinary tract infection, diabetes, or acute or chronic renal failure.

A thorough family history can contribute importantly to the initial evaluation for dementia. A family history of early-onset Alzheimer's disease or other rare genetic conditions leading to dementia such as Huntington's disease should be sought.

The social and cultural history should include information about education, literacy, and socioeconomic, ethnic, and cultural background, as well as information on recent life events and social support networks. As noted, these factors can affect the risk for dementia as well as current performance on mental status tests.

The informant should be a person familiar with the patient, for example, a family member or close friend. In addition to having memory loss, which limits the validity of subject reports, many patients with a dementing syndrome lack insight into the severity of their decline. This is especially true in patients with more severe impairment ( Grut, Jorm, Fratiglioni, et al., 1993; McGlynn and Kaszniak, 1991). Informant reports can both corroborate and supplement reports of the affected person.

The importance of informant reports for establishing decline cannot be overstated (Berg, 1990; Henderson and Huppert, 1984). Informant reports also can provide important evidence for impairment in multiple cognitive domains. Notably, research concerning the retrospective accounts of dementia symptoms has shown that relatives' reports are reasonably reliable (La Rue, Watson, and Plotkin, 1992) and valid (Bayles and Tomoeda, 1991). Relatives' reports, however, can be influenced by the nature of the relationship to the patient. For example, La Rue and colleagues ( 1992) found that spouses of memory-impaired persons reported lower levels of impairment than did younger relatives. Thus, even though it may be difficult, it appears prudent to include more than one family informant or a family consensus approach to increase the accuracy of conclusions on the presence and range of cognitive impairments in those with suspected dementia.

Concern should be given to preserving a good doctor-patient relationship and for safeguarding the dignity of the patient. Patients should be informed that others may be interviewed. It is best to interview family members and other informants separately from the patient; family members are often very aware of deficits but may not be willing to discuss them in the presence of the patient for fear of upsetting him or her. Butler, Finkel, Lewis, et al. (1992) describe practical approaches to the initial office visit for a person suspected of having dementia who is accompanied by friends or family members. They emphasize the need to examine the patient alone and before the informant is interviewed to maintain the patient's dignity. Examining the patient alone also has implications for the development of a trusting physician-patient relationship (Greene, Majerovitz, Adelman, et al., 1994).

In evaluating informant reports, clinicians should be aware of the possibility of questionable motives, as well as the family's level of stress and fatigue, which frequently attend the process of diagnosing disturbing symptoms. For example, reports of behavioral and cognitive symptoms might be minimized if the family member is concerned that the patient will be denied admission to a nursing home. Conversely, symptoms could be exaggerated or even fabricated if the informant is motivated by financial gain or other issues.

Focused Physical Examination

A focused physical examination, including a brief neurological evaluation, is an essential part of the initial clinical assessment. It must be directed by the same principles that guide initial assessment of any medical condition; that is, life-threatening or rapidly progressing etiologies must be identified first. Life-threatening conditions include mass lesions, vascular lesions, and infections. Special attention should be paid to assessing for those conditions that cause delirium because delirium represents a medical emergency (see the section on Assessing for Delirium and Depression later in this chapter). The examination should include measurement of blood pressure and pulse, both lying and standing; assessment of vision and hearing; and evaluation for any evidence of cardiac failure, poor respiratory function, or problems in mobility or balance. A complete physical examination should be scheduled if not done at this visit.

All health care providers also must be alert to signs of a caregiver's abuse and neglect of patients with dementia. One recent study reported that severe physical abuse of the affected person occurs in approximately 20 percent of families (Paveza, Cohen, Eisdorfer, et al., 1992). Suspected abuse should be reported to the proper authorities.

Functional Status Assessment

A functional status assessment should be part of the initial clinical assessment (see Figure). The panel recommends the use of an explicit and standardized evaluation of functional status (see Chapter 2, Functional Impairment section). Structured measures of functional performance, especially those that measure higher order social and instrumental activities, are particularly helpful when used in the context of a comprehensive evaluation for dementia (see Wilder, Gurland, Chen, et al., 1994). Performance testing of function can be helpful, particularly when informants are not available (Gurland, Cross, Chen, et al., 1994). Informant-based functional assessment scales, in particular, are important for evaluating complex or difficult cases (O'Connor, Pollitt, Hyde, et al., 1991) or persons who fall into "borderline" (Hershey, Jaffe, Greenough, et al., 1987) or "questionable" (Morris, McKeel, Storandt, et al., 1991) or "borderzone" (Wilder, Cross, Chen, et al., 1995) categories.

The Functional Activities Questionnaire (FAQ) (Pfeffer, Kurosaki, Harrah, et al., 1982), an informant-based measure, can provide critical information on an individual's functional abilities. This determination was made on the basis of meta-analyses conducted with data from studies that met the panel's criteria for inclusion. An informant provides performance ratings of the person being assessed (see Attachment E for information on administration and scoring of the FAQ). The FAQ is a good discriminator, with an average effect size of 2.46 for patients with mild-to-severe dementia and comorbid conditions. Effect size is a summary statistic that provides an index of an instrument's ability to discriminate between, in this case, subjects with and without dementia. An effect size of 2.46 for the FAQ corresponds to sensitivity and specificity values in the range of about 85 to 90 percent ( Hasselblad and Hedges, 1995).

On the FAQ, an informant provides performance ratings of the target person on 10 complex, higher order functional activities (see Attachment E). Little evidence supports the use of the FAQ directly with the patient rather than (or in addition to) an informant. A recent study by Weinberger, Samsa, Schmader, et al. (1992) suggests that persons with very mild dementia may be able to provide valid reports of functional performance. This is confirmed by Wilder, Cross, Chen, et al. (1995). As a rule, however, every effort should be made to find a reliable informant.

Two measures of functional status were included in the meta-analysis -- the Katz ADL and the FAQ -- along with the other 50 psychological, neuropsychological, and mental status tests listed in Attachment C. Although the panel's recommendation of the FAQ is based on only a minimum number of studies, these studies yielded contrasts at Phase IV, the highest and thus most important level for the purpose of recommendations. Only 3 other tests of the total of 18 tests involved in Phase IV analyses had more than two contrasts: the MMSE (12), Wechsler Adult Intelligence Scale (WAIS) Digit Symbol (4), and WAIS Information (3). The remaining 15 tests contributed only two contrasts, as did the FAQ. However, the FAQ emerged as the best discriminator in these Phase IV analyses, and only its effect size of 2.46 was significantly different from (better than) the MMSE effect size of 1.78. Findings for the MMSE, although based on more studies, showed poorer specificity (ranging from .44 to .63) compared with a specificity of .80 for the FAQ.

It should be noted that neither the Katz ADL nor any other ADL or IADL scale had data that met Phase IV criteria. Furthermore, the FAQ was superior not only to the MMSE but also to all other 17 tests and measures in the Phase IV analyses. Together, these results support and lead to the panel's recommendation of the FAQ in the initial assessment of dementia. Recently, Pfeffer (1995) reported a larger, more representative study that gives further confirmation and validation.

Mental Status Assessment

The panel determined that universal or broad-band screening for Alzheimer's disease or related dementias among older adults was not warranted (see Chapter 1). In lieu of routine screening, the practitioner plays a pivotal role in being alert to the changes from baseline status that suggest the development of a dementing disorder in a patient and in recognizing the difference between such symptoms and the normal processes of aging.

Assessing the mental status of a person suspected of having dementia is the most critical component of the physical examination; a quantitative mental status examination should be part of the initial assessment. Ideally, this evaluation will provide a detailed cognitive profile of the patient. This approach, unfortunately, is impractical for most health care providers. However, in lieu of a comprehensive mental status examination, several brief, quantitative tests are available to assess mental status. These tests were developed to standardize the variety of individual approaches to the clinical evaluation of mental functioning.

Table 7. Components of commonly used brief mental status (more...)

Table 7. Components of commonly used brief mental status tests and requirements for testing

	MMSE	BIMC	BOMC	STMS
Components of cognition measured (listed in the DSM-IV criteria for dementia):
-- Immediate memory	X			X
-- Short-term recall	X	X	X	X
-- Abstract thinking				X
-- Judgment
-- Aphasia	X
-- Apraxia	X
-- Agnosia	X
-- Constructional ability	X			X
Additional components of cognition measured:
-- Concentration	X	X	X	X
-- Spatial ability, orientation	X	X	X	X
Requirements for testing:
-- Verbal responses	X	X	X	X
-- Reading ability	X
-- Writing ability	X			X
-- Mathematical ability	X			X
-- Vision	X			X
-- Motor control skills	X

Note: MMSE = Mini-Mental State Examination; BIMC = Blessed Information-Memory-Concentration Test; BOMC = Blessed Orientation-Memory-Concentration Test; STMS = Short Test of Mental Status.

Source: Adapted from White H, Davis PB. Cognitive screening tests: an aid in the care of elderly outpatients. J Gen Intern Med 1990;5;438-45. Copyright 1990, Society for Research and Education in Primary Care Internal Medicine. Used with permission.

The literature is growing for many of the brief mental status tests developed in the last 10 to 15 years. This evidence was submitted to meta-analysis as discussed in Chapter 1. To identify mental status tests that would be clinically useful across a wide spectrum of persons ( Ransohoff and Feinstein, 1978), the panel adopted an analytic scheme of phases of evaluation for diagnostic marker tests (Nierenberg and Feinstein, 1988). Lower phases in the scheme involve relatively easy discriminations between case patients with severe dementia and healthy, high-functioning control subjects. Higher phases involve a wider range of severity of illness (i.e., including patients with mild or early dementia and control subjects with comorbid conditions that mimic the performance of case patients). Attachment C presents the effect sizes results (in "D" columns), along with the number of contrasts and the Z values for 52 neuropsychological tests or subtests and functional status measures. Tests with Z scores in Attachment C of 1.96 or greater are significantly better than the MMSE (Folstein, Folstein, and McHugh, 1975) in distinguishing case patients from control subjects.

As seen in Attachment C, the great majority of studies and contrasts come from Phases II and III, where comorbidity among control subjects is absent. Only three tests have data at Phase III. In addition to the MMSE, the results in Attachment C indicate that the FAQ emerges as the best discriminator, especially at the highest level of discrimination difficulty, and its effect size is the only one significantly better than the MMSE. Several other mental status tests evaluated at the highest phase (IV) had effect sizes that were equivalent to the MMSE: the 26-item Blessed Information-Memory-Concentration (BIMC) test, the 6-item Blessed Orientation-Memory-Concentration (BOMC) test, and the Short Test of Mental Status (STMS) (the Selective Reminding-Total Words and the Symbol Digit Modalities are also equivalent to the MMSE).

Taken as a whole, the results of the meta-analyses from the most relevant contrasts -- those involving an expanded spectrum of case patients and control subjects with interfering or comorbid conditions -- indicate that four tests are largely equivalent in terms of their ability to differentiate between persons with and without dementia who have comorbid conditions of a general medical, psychiatric, and/or neurologic nature. Thus, any of these four mental status questionnaires -- the MMSE, the BIMC, the BOMC, or the STMS -- can be used to gather mental status information. The panel's meta-analyses from the most relevant contrasts available -- those involving an expanded spectrum of case patients and control subjects -- show that the various mental status tests are largely equivalent in their ability to differentiate between subjects with and without dementia. The MMSE (Folstein, Folstein, and McHugh, 1975) is the most widely used brief mental status examination in the United States, and the clinical and research literature on the MMSE is extensive and broad. However, clinicians who have experience using the BIMC, the BOMC, or the STMS could choose to continue using these tests.

Because structured mental status tests can provide a quantitative baseline for monitoring the course of cognitive impairment, they can be useful in establishing a decline from previous cognitive levels. Repeated administration of the same mental status tests would be particularly valuable in reassessing mental status in persons found to have treatable delirium or depression on initial evaluation. Subsequent examination of these individuals may reveal improvement of cognition or, in the absence of significant improvement, may assist in confirming the presence of dementia. Currently, there is no evidence to suggest a timeframe for reassessment.

Whenever possible, clinicians should document the presence of decline from reports obtained from a close relative or another informant. In some instances, however, no reliable informant is available. In other cases, particularly in mild or early dementia, relatives may deny that the person has demonstrated cognitive deterioration, or they may minimize the extent of decline (Pollitt, O'Connor, and Anderson, 1989). Evidence for decline can be documented in these cases by comparing performance on a mental status test over time.

Because most brief mental status tests assess a range of cognitive functions, they also can be used to document multiple cognitive impairments, as required for a diagnosis of dementia. The MMSE is the most comprehensive of these tests. It examines orientation, attention, immediate and delayed memory, visuospatial ability, aphasia, agnosia, and apraxia. However, its approach is limited by differential sensitivity for various cognitive domains (Mitrushina and Satz, 1994). For instance, several studies show that the MMSE language items are too simple to detect mild impairment (Tombaugh and McIntyre, 1992), even though persons with mild dementia of Alzheimer's disease may have clear language impairments on more difficult language tasks (Bayles and Kaszniak, 1987). In one study of 76 subjects (about half of whom were diagnosed with Alzheimer's disease), four of the five MMSE language items showed very low (<15 percent) sensitivity, and three of these language items did not correlate with language measures of neuropsychological testing (Feher, Larrabee, and Crook, 1992). In this study, however, the MMSE language items did show reasonably high specificity.

On the basis of these studies, an MMSE finding of impairments in memory and at least one other cognitive area is consistent with the multiple cognitive deficits associated with dementia; however, a finding of impairment in memory only does not automatically rule out dementia.

Regardless of the test selected, its results should be interpreted in conjunction with the patient's signs and symptoms and with the functional status assessment. Brief mental status tests are important to the initial assessment of dementia, but they are not diagnostic and thus should not be used as stand-alone diagnostic tests.

This approach rests on the concept of incremental validity -- the notion that information from multiple, reliable sources enhances the validity of the assessment. Several studies provide support for this notion within the context of assessing dementia (Chaves and Izquierdo, 1992; Wilder, Gurland, Chen, et al., 1994). Wilder and associates (1994) showed, for example, that combining an informant's report of function with the results of a cognitive rating scale can yield sizable gains in specificity (reduction in the number of false-positives) without appreciable loss of sensitivity. These investigators also showed that, for the intermediate range of cognitive scores (i.e., neither very high nor very low), the additional information from a functional scale improves the prediction of dementia. This combination of an informant-based measure of functional performance with a brief cognitive rating scale has been widely endorsed in the dementia literature ( Jorm and Korten, 1988; van der Cammen, van Harskamp, Stronks, et al., 1992; see also Gurland, Wilder, Chen, et al., 1995).

Confounding and Comorbid Conditions

Visual and auditory impairments, as well as various physical disabilities, may affect performance on mental status and neuropsychological tests (O'Connor, Pollitt, Hyde, et al. 1988; Peters, Potter, and Scholer, 1988). Individuals suspected of having dementia should be screened for visual and auditory impairment and for physical disability (e.g., neuromuscular disorder, severe arthritis) before mental status or other cognitive tests are administered. Visual and auditory deficits should be corrected, when possible, before testing. When visual or auditory deficits cannot be corrected or physical disability is present, the patient may need to be referred for neuropsychological evaluation employing tests that use his or her unimpaired sensory modalities and motor capacities. The MMSE in particular should not be used when the subject has impaired motor control because of the drawing and praxis portions of these tests; other tests (e.g., the BIMC) should be used. Care also should be taken when testing any older adult to ensure adequate illumination and contrast of visual stimuli and adequate volume and distinctiveness of auditory verbal stimuli.

Delirium

Recognition and treatment of delirium are important because the underlying physical disorder, as well as the associated cognitive deficits and behavioral symptoms, can be sources of danger, distress, and mortality. The presence of delirium can indicate that medical illnesses or physiological abnormalities are affecting cerebral activity. Early recognition of delirium with mild symptoms may offer significant opportunities to prevent disability and irreversible deterioration. Levkoff, Evans, Liptzin, et al. (1992) suggest that acute onset of disturbances of consciousness, disorientation, and perceptual disturbances is associated with significant morbidity, even in the absence of the full-blown syndrome.

Delirium is a medical emergency requiring immediate further evaluation and treatment; some of the underlying causes (such as bacterial meningitis or hypoglycemia) can be fatal. Delirium is usually reversible if treated in a timely fashion.

Delirium is common among elderly persons with acute illnesses. Prevalence rates are estimated in the range of 10 to 30 percent among medical inpatients and 7 to 52 percent among postsurgical patients, depending on clinical settings and the specific diagnostic criteria used (Levkoff, Cleary, Liptzin, et al., 1991). Delirium is also a common complication of chronic disease; prevalence rates in nursing homes have been estimated to be 6 to 7 percent ( Rovner, German, Broadhead, et al., 1990).

Medical record reviews indicate that delirium is underrecognized in clinical settings (Johnson, Kerse, Gottlieb, et al., 1992). One potential problem in diagnosing delirium is that many clinicians may not know how to evaluate attention and level of consciousness. States of lethargy and stupor in which patients have difficulty in maintaining wakefulness should be readily recognized as symptoms of delirium, but more subtle disturbances in consciousness and attention that could identify persons needing urgent evaluations and treatment may not be recognized as such.

Recognition of dementia can be improved by using mental status tests to identify patients with cognitive impairment and by establishing the history of the onset and the degree of fluctuation in symptoms. Recent research has defined symptoms of delirium in operational terms, establishing the feasibility of using systematic methods for assessment at bedside or in the clinic. Anthony, LeResche, Von Korff, et al. (1985), for example, developed a method by which clinicians can rate "how well the patient kept his mind on interaction with the observer" and determined that these "global ratings of accessibility" are reliable and that they have high levels of both sensitivity (90 percent) and specificity (95 percent) against independent clinical diagnoses of delirium. More recently, Gottlieb, Johnson, Wanich, et al. (1991) assessed these symptoms through a combination of ratings of accessibility and measures of performance on brief tasks (e.g., spelling a word and stating the days of the week forward and backward). Albert, Levkoff, Reilly, et al. (1992) developed the Delirium Symptom Interview, which evaluated the disturbances of consciousness by administration of brief tests to patients (asking for days of the week and months of the year backwards) and by observations of behavior (e.g., staring into space unaware of the environment, inappropriate distraction by environmental stimuli). The Confusion Assessment Method and the Delirium Rating Scale are validated methods that distinguish between delirium and dementia (Inouye, van Dyck, Alessi, et al., 1990). Inouye, Viscoli, Horwitz, et al. (1993) developed a predictive model for delirium in hospitalized elderly patients.

Depression

Depression is often mistaken for dementia (Koskinen, 1992) and vice versa (Siegel and Gershon, 1986). Differentiation between depression and dementia is complicated by the fact that dementia can initially present as depression, sometimes in the absence of marked cognitive deficit (Reding, Haycox, and Blass, 1985). Conversely, depression may present as dementia. This has been referred to as "pseudodementia" (Kiloh, 1961) or, more recently, "depression-related cognitive dysfunction" (Stoudemire, Hill, Gulley, et al., 1989). Depression and dementia also may coexist. It has been estimated that 1 to 31 percent of persons diagnosed with a progressive dementing illness might actually be depressed and have associated memory deficits (Katzman, Lasker, and Bernstein, 1988). Depression can develop secondary to a dementing illness, such as Alzheimer's disease or vascular dementia, particularly in persons with less severe cognitive impairment.

Accurate diagnosis of older adults who show concomitant signs of memory impairment and depression presents a continuing problem for physicians and other health care professionals. Depression in older persons, particularly those with physical illness, is frequently underdiagnosed. Depressed older adults are at high risk of suicide and mortality from other causes (National Institutes of Health Consensus Development Panel on Depression in Late Life, 1992). However, depression in older adults often responds to treatment with antidepressant medication, psychotherapy, electroconvulsive therapy, or all of the above (Benedict and Nacoste, 1990; Koenig and Blazer, 1992; National Institutes of Health Consensus Development Panel on Depression in Late Life, 1992).

Problems are also associated with misdiagnosing a progressive dementing illness as depression. In reviewing the literature, Siegel and Gershon (1986) concluded that 25 to 30 percent of older persons referred for evaluation and treatment of depression are ultimately diagnosed as having a progressive dementing illness. Inappropriate treatment for nonexistent depression in a person with a progressive dementia not only entails unnecessary expense but also can exacerbate the dementia because treatment with antidepressant medications that have anticholinergic properties may contribute to further confusion or memory impairment (Benedict and Nacoste, 1990; Harris, Gierz, and Lohr, 1989).

Assessing for depression is therefore an important part of the initial evaluation of any older adult suspected of cognitive impairment. This is particularly true for the person who presents with complaints of memory difficulty (as contrasted with the person who is brought in for evaluation by a spouse or other relative who suspects memory impairment) (Kahn, Zarit, Hilbert, et al., 1975; Larrabee and Levin, 1986). In addition to complaints of memory problems, some depressed older persons may show evidence of functional disability and impairment on tests of memory. Although changes in memory, attention, and executive function (i.e., goal formulation, planning, and execution of plans) are commonly associated with depression, marked visuospatial or language impairment would suggest a dementing process.

Instruments to Assess Depression. Brief self-report questionnaires can facilitate initial screening for depression-related symptoms. Several reviews comparing older-adult normative, reliability, and validity data from self-report depression screening instruments are available (Fry, 1986; Gallagher, 1986; Kaszniak and Allender, 1985; Yesavage, 1986).

However, for persons suspected of having a dementing illness, standard screening instruments for assessing depression should be used with caution. Considerable difficulties are encountered in assessing for depression in older adults with known or suspected memory impairment (Kaszniak, Sadeh, and Stern, 1985; Raskind and Peskind, 1992; Rubin, Zorumski, and Burke, 1988). For example, memory difficulty, agitation, disrupted sleep-wake cycle, and personality changes such as apathy and increased dependence are classic symptoms of Alzheimer's disease that may be mistaken for depressive signs of poor concentration, decreased interest, changes in psychomotor activity, sleep disturbance, and fatigue. Depression-related symptoms may be common in the early stages of Alzheimer's disease and may occur in the absence of the full syndrome of clinical depression (Burke, Rubin, Morris, et al., 1988; Burns, 1991; Merriam, Aronson, Gaston, et al., 1988).

One self-report depression screening instrument, the Geriatric Depression Scale (GDS) (Brink, Yesavage, Lum, et al., 1982; Yesavage, Brink, Rose, et al., 1983), was developed specifically for use with older adults. It is reliable and valid for the assessment of depression in elderly outpatients, inpatients, and nursing home residents (Lesher, 1986; Parmelee, Katz, and Lawton, 1989; Scogin, 1987; Yesavage, Brink, Rose, et al., 1983). The GDS is a 30-item questionnaire, with clearly and simply phrased items and a "yes/no" response format that takes only 8 to 10 minutes to administer. The content focuses on mood and mood-related symptoms and excludes somatic items that may reflect physical illness in older persons. A score of 11 or greater is taken as an indication of possible depression. A 15-item short form of the GDS also may be useful for many purposes (Yesavage, 1988).

Although the GDS has merit as a screening instrument for depression in older adults with suspected memory impairment, the possibility of false-negative results must be considered. Several studies indicate that, in comparison with self-report instruments, collateral sources (such as spouses and other relatives) and trained clinical observers consistently report significantly higher frequencies of depression-related symptoms in Alzheimer's disease patients who have mild dementia (Burke, Rubin, Morris, et al., 1988; Mackenzie, Robiner, and Knopman, 1989; Miller, 1980; Rubin, 1990; Teri and Wagner, 1992). Thus, persons with Alzheimer's disease may report fewer depression-related symptoms than their clinicians or caregivers report. Conversely, Burke et al. ( 1988) suggest that collateral sources may have a "heightened awareness" of depression-related symptoms that occur in Alzheimer's disease and thus may be more likely to overreport them.

Considering the possibilities of observer bias in reporting depression-related symptoms, optimal depression screening in patients suspected of dementia should include information from both individual self-report and caregiver report, as well as direct clinician observation of the patient's behavior. Among those instruments appropriate for obtaining caregiver report of depression-related symptoms, the Hamilton Depression Rating Scale (HAM-D) has been used successfully in studies of Alzheimer's disease, and the Center for Epidemiological Studies-Depression Scale (CES-D) has been modified successfully as a caregiver report of patient depression (Teri and Wagner, 1992). The recently developed Columbia University Scale for Psychopathology in Alzheimer's Disease (CUSPAD) (Devanand, Miller, Richards, et al., 1992) is a broad measure of behavioral problems and appears promising. The CUSPAD is a brief semistructured instrument that can be administered by a trained lay interviewer to an informant in 10 to 25 minutes. The PRIME-MD procedure, developed by Spitzer, Williams, Kroenke, et al., 1994, was designed specifically to assist primary care practitioners in identifying psychiatric symptoms in their patients. Another promising new instrument is the Cornell Scale for Depression in Dementia (CSDD) (Alexopoulos, Abrams, Young, et al., 1988), a 19-item scale that requires a trained clinician to obtain information from both the patient and an informant. The AHCPR Clinical Practice Guideline, Depression in Primary Care: Volume I. Detection and Diagnosis(Rush, Golden, Hall, et al., 1993), also should be consulted.

Age and Educational Influences

For most mental status and neuropsychological tests, performance may be affected by both age and education (Tombaugh and McIntyre, 1992). In studies employing regression analyses ( Brayne and Calloway, 1990; Uhlmann and Larson, 1991), education has accounted for more MMSE performance variance than other demographic variables including race, sex, and social class. In the largest normative study of the MMSE in the United States (Crum, Anthony, Bassett, et al., 1993), significant correlations were found between MMSE scores and both age and years of schooling. Crum and associates (1993) provide MMSE mean and percentile data, broken down by age and education. Age-specific MMSE norms have also been provided by Bleecker, Bolla-Wilson, Kawas, et al. (1988).

Many clinicians and investigators have assumed that the effects of education represent a psychometric bias, leading to misclassification of persons having different educational histories. Indeed, there is evidence that low education increases the probability of false-positive errors (misclassifying normal individuals as cognitively impaired), particularly when the person has fewer than 9 years of education ( Anthony, LeResche, Niaz, et al., 1982; Murden, McRae, Kaner, et al., 1991).

There is also evidence that higher educational levels may result in increased false-negative errors (identifying a cognitively impaired person as unimpaired). For example, O'Connor, Pollitt, Hyde, et al. (1989) found that persons with false-negative MMSE identification of dementia had relatively high levels of education. The rate of decline in MMSE scores among persons with Alzheimer's disease has not been found to be associated with education (Burns, Jacoby, and Levy, 1991).

Despite the evidence that differing educational levels can be related to MMSE misclassification, education also may partially reflect the contribution of etiologic factors in the development of dementia (e.g., lower education could be associated with risk factors for vascular dementia such as hypertension) (Mortimer and Graves, 1993). Overall, however, the available data suggest that age- and education-stratified normative data, such as those provided by Crum, Anthony, Bassett, et al. (1993), are valuable in determining whether a particular individual's MMSE score is consistent with significant cognitive impairment.

Cultural Influences

Not only educational level, but also primary language, race, ethnicity, and cultural bias can affect performance on both mental status tests and particular neuropsychological tests. The Consortium to Establish a Registry for Alzheimer's Disease (CERAD -- see Attachment D) has created a Spanish-language version of both its clinical and neuropsychological assessments (copies are available from CERAD), which have been administered to various Hispanic populations in the United States (Welsh, Butters, Mohs, et al., 1994). Escobar, Burnam, Karno, et al. (1986) studied more than 3,000 English- and Spanish-speaking residents of Los Angeles. Spanish-speaking persons were administered a Spanish version of the MMSE; English-speaking persons were administered the English version. Six of the MMSE items showed differential performance on the basis of language and ethnic status of the respondents. When these "ethnically biased" items were removed from the analysis, similar rates of cognitive impairment were found for Hispanics and non-Hispanics.

Neuropsychological tests have been found to be equivalently valid in differentiating persons with and without dementia in Hispanic and non-Hispanic populations. The possibility exists, however, that normative cutpoints for defining impairment or dementia differ. Lopez and Taussig (1991) found that unimpaired elderly Hispanic subjects scored lower on four Wechsler Adult Intelligence Scale Revised (WAIS-R) subtests (i.e., Similarities, Vocabulary, Digit Span, and Block Design) (Wechsler, 1981) than unimpaired elderly non-Hispanic subjects. Comparison of scores for Hispanic and non-Hispanic patients with Alzheimer's disease on these same subtests found no significant differences, suggesting a "floor effect"; that is, performance of both groups was so low that differences were not detectable. When the two verbal WAIS-R subtests (Similarities and Vocabulary) were scored on the basis of culturally specific norms (scale scores of a Spanish version of the WAIS-R), the scores of the unimpaired Hispanic group were raised, eliminating the previous difference. The authors caution that the use of nonculturally specific norms may lead to underestimation of cognitive ability or, conversely, to overestimation of the level of cognitive impairment.

Loewenstein, Argüelles, Barker, et al. (1993) compared the neuropsychological test performance of 38 Hispanic and 38 non-Hispanic women who met criteria for probable or possible Alzheimer's disease. When their test results were compared, the Hispanic women tested in Spanish performed worse on the WAIS-R Digit Span and Comprehension tests as well as on the FAS Controlled Oral Word Association Test. The authors then divided their sample of patients based on a median split of the Fuld Object Memory Evaluation scores (Fuld, 1981), with below the median defining a moderately impaired group and above the median a mildly impaired group. Moderately impaired Hispanic patients scored worse on 5 of their 11 tests (WAIS-R Digit Span, Block Design, Object Assembly, Comprehension, and the FAS Controlled Oral Word Association Test) than their non-Hispanic counterparts. Furthermore, the mildly impaired Hispanic patients scored significantly worse on WAIS-R Digit Span than their non-Hispanic counterparts.

These studies suggest that several neuropsychological tests may place Hispanic persons and possibly members of other racial or ethnic groups at a disadvantage, particularly in the use of culturally inappropriate norms or cutpoints.

Tests of intellectual function used to assess cognitive performance or impairment must be interpreted according to norms established for levels of education and for the culture in which the test is administered. Because of the difficulties in comparing test results across cultures, norms for each test must be established in the culture in which it is applied.

Not only primary language, but also race and ethnicity (e.g., black versus white subjects of comparable educational level), social class, and socioeconomic status have been found to be related to MMSE scores in most studies examining these relationships (Tombaugh and McIntyre, 1992;for exception, see Murden, McRae, Kaner, et al., 1991). No information exists on whether identification of clinically significant cognitive impairment in minority racial and ethnic groups or in lower socioeconomic groups would be improved by the addition of IADL or neuropsychological test data.

Objectives of Neuropsychological Testing

Formal neuropsychological assessment can address any or all of the following general aims (Albert and Moss, 1988; Bayles and Kaszniak, 1987; La Rue, 1992; Zec, 1993):

• Give information to patients, family members, and health care providers about the specific nature of strengths and deficits in cognitive functions.

• Assist in diagnosis, particularly in cases where the impairment is mild, the patient is of high premorbid intellectual ability, or there is an unusual combination of cognitive impairments.

• Contribute to recommendations for treatment and management of behavior problems.

• Provide a baseline measurement against which the effects of treatment or disease progression can be judged.

Available research suggests that neuropsychological assessment makes an important contribution to the identification of mild dementia, particularly when delayed recall is measured (Welsh, Butters, Hughes, et al., 1992). Because mental status tests such as the MMSE are less sensitive in detecting mild compared with moderate to severe cognitive impairment (for a review, see Tombaugh and McIntyre, 1992), neuropsychological testing may be particularly helpful when there is a history of apparent cognitive decline, but the results of a brief mental status test are normal.

In addition, the pattern of performance across neuropsychological tests has been found useful in identifying the presence of dementia among persons with high premorbid intellectual functioning, in discriminating patients with dementing illness from those with focal cerebral disease, and in differentiating among certain etiologies of dementia (La Rue, 1992; La Rue, Yang, and Osato, 1992; Parks, Zec, and Wilson, 1993). However, neuropsychological evaluation must be interpreted within the context of other clinical data, such as informant-based history of cognitive decline, evidence of impairment in IADLs, educational background, assessment for depression, sensory impairment, or factors other than dementia that may account for impaired performance.

Role of Neuropsychological Testing

Brief mental status tests usually have substantial false-negative rates in detecting cognitive impairment, especially if the impairment is caused by focal lesions of the right cerebral hemisphere (Nelson, Fogel, and Faust, 1986). The standardized tests used in neuropsychological assessment, however, have established reliability and sensitivity in the detection of cognitive deficits (see Grant and Adams, 1986; Kolb and Whishaw, 1990; Lezak, 1995; Spreen and Strauss, 1991). Appropriate neuropsychological tests may reveal subtle and circumscribed cognitive impairments in persons who show no evidence of deficits on brief mental status tests (see, e.g., Bondi, Kaszniak, Bayles, et al., 1993). The results of neuropsychological assessment instruments for the detection of mild dementia are correlated with the subject's level of education (Ganguli, Ratcliff, Huff, et al., 1991). They also can be affected by any of the following factors: physical disability, visual or auditory impairment, psychiatric illness (including depression), and limited facility with the English language (O'Connor, Pollitt, Hyde, et al., 1991).

Assessment in Various Cognitive Domains. Neuropsychological tests can examine performance across different domains of cognitive functioning, including orientation and attention, language functions, visual motor constructional ability (praxis), memory functions, abstract and conceptional reasoning, and executive functions. For this reason, they may be valuable in assessing for possible dementia in persons with high premorbid intellectual functioning when the clinician has a basis to determine that a loss of function has occurred (Naugle, Cullum, and Bigler, 1990). Furthermore, some evidence suggests that persons with mild dementia can improve their use of coping mechanisms and strategies to compensate for memory loss when they have information about their cognitive strengths and deficits derived from neuropsychological tests and given in the context of brief supportive counseling (LaBarge, Rosenman, Leavitt, et al., 1988).

Differentiation of Normal Aging and Causes of Dementia. Recent research has focused on the diagnostic utility of neuropsychological testing in differentiating dementia from normal aging and in distinguishing among different causes of dementia (Heaton, Grant, and Matthews, 1986; La Rue, 1992; La Rue, Yang, and Osato, 1992; Parks, Zec, and Wilson, 1993). For example, detailed neuropsychological assessment of memory function that distinguishes features such as learning and forgetting rates, recall versus recognition, and vulnerability to interference has been found useful in differentiating etiologies of dementia, such as Alzheimer's disease from Huntington's disease (Delis, Massman, Butters, et al., 1991).

Differentiation of Mild Dementia. In studies that address differentiation of healthy older adults from persons with very mild dementia caused by probable Alzheimer's disease, the most discriminating measures are those of recent memory, particularly delayed recall of newly learned material (Knopman and Ryberg, 1989; Morris, McKeel, Storandt, et al., 1991; Welsh, Butters, Hughes, et al., 1991). In one study (Morris, McKeel, Storandt, et al., 1991), postmortem evidence of Alzheimer's disease correlated completely with the results of psychometric differentiation of persons with mild cognitive impairment from normal older adults. Tests of recent memory also have been reported to show high sensitivity and specificity in discriminating mildly impaired persons who decline cognitively at 2-year reexamination from those who do not (Flicker, Ferris, and Reisberg, 1991).

Although these data appear encouraging for the application of neuropsychological testing (particularly of memory function) in detection of very mild dementia, high sensitivity may be achieved at the cost of a decrease in (or loss of) specificity. Not all persons with Alzheimer's disease have recent memory deficits as the most prominent initial deficits. In some patients with mild Alzheimer's disease, language or visuospatial abilities, including psychomotor performance, are the prominent features affected. Another early problem in some patients may be executive function (e.g., Becker, Bajulaiye, and Smith, 1992) or access to semantic memory ( LaBarge, Balota, Storandt, et al., 1992; Storandt, Stone, and LaBarge, 1995).

The severity of recent memory deficits early in the course of Alzheimer's disease renders memory tests less useful than others for staging the severity of dementia across individuals (Welsh, Butters, Hughes, et al., 1992). Neuropsychological measures of other cognitive functions, such as recognition memory, verbal fluency, confrontation naming, and praxis, appear better for staging the severity of dementia or tracking its progression (Kaszniak, Wilson, Fox, et al., 1986; Welsh, Butters, Hughes, et al., 1992). When neuropsychological tests are carefully selected, they can play a valuable role in the evaluation of trials of medication or other interventions for persons with dementing illness (see Flicker, 1988).

Clinical Management. Identification of relatively intact cognitive function can be of assistance in developing plans for sharing daily responsibilities between the patient and others (see La Rue, Yang, and Osato, 1992). Conversely, neuropsychological documentation of specific deficits can direct caregivers to areas where the patient will require additional supervision or assistance. In a study of persons with probable Alzheimer's disease, Henderson, Mack, and Williams (1989) found that visuoconstructive deficits documented by neuropsychological tests significantly predicted "real world" spatial disorientation as reflected in caregiver reports of the patient's becoming lost, failing to recognize familiar environments, and wandering. Although research results to date have been mixed, neuropsychological evaluation also may be able to make contributions to decisions about whether a person can safely continue in potentially risky activities such as driving (for review, see Kaszniak, Keyl, and Albert, 1991).

Administration on Aging

The Administration on Aging (AoA) coordinates delivery of services specified by the Older Americans Act. Services are coordinated and provided through 57 State agencies and 657 areas.

Agencies on Aging (AAAs). The range of services provided by each AAA varies, but all include nutrition services, access services, in-home services, and community services.

Addresses and phone numbers of State and area AAAs may be obtained from the national office:

• Administration on Aging

• Department of Health and Human Services

• 330 Independence Avenue, SW

• Washington, DC 20201

• (202) 619-1006, (202) 619-7586 fax

• Internet address: gopher://gopher.os.dhhs.gov/11/dhhs/aoa/aoa

The Elder Care Locator provides a toll-free access number to locate State agency networks: 800-667-1116.

Alzheimer's Disease Centers

The National Institute on Aging, a component of the National Institutes of Health, supports 28 Alzheimer's Disease Centers across the country. This program provides clinical services, conducts basic and clinical research, disseminates professional and public information, and sponsors educational activities. A growing number of satellite clinics associated with this program are helping to expand diagnostic and treatment services in rural and minority communities and collect research data from a more diverse population. For information, contact the ADEAR Center (see below).

Alzheimer's Association

The Alzheimer's Association is a national voluntary organization with 220 local chapters and more than 2,000 support groups. The Alzheimer's Association funds research, promotes public awareness, advocates legislation for patients and families, and provides support services, including support groups, adult day care programs, respite care programs, and telephone helplines through its national, local chapter, and volunteer network.

• Alzheimer's Association,

• 919 North Michigan Avenue, Suite 100

• Chicago, IL 60611-1676

• (312) 335-8700, 800-272-3900 for information and local chapter referrals nationwide (24-hour telephone line)

• Internet address: http://www.alz.org

Alzheimer's Disease Education and Referral Center

The Alzheimer's Disease Education and Referral (ADEAR) Center, a service of the National Institute on Aging, provides information and publications on Alzheimer's disease for health professionals, people with Alzheimer's disease and their families, and the public. The ADEAR Center serves as a national resource for information on diagnosis, treatment issues, patient care, caregiver needs, long-term care, education, research, and ongoing programs. In addition, the Center provides referrals to national and State resources. For information, contact:

• Alzheimer's Disease Education and Referral Center

• P.O. Box 8250

• Silver Spring, MD 20907-8250

• 800-438-4380

• Internet address: adear@alzheimers.org

The Corporation for National Service

The Corporation for National and Community Service is a public corporation that administers Federal service programs, including AmeriCorps, the Foster Grandparent Program, and the Senior Companion Program (SCP), which provides supportive services to adults with physical, emotional, and health limitations. A major emphasis of the SCP is preventing or delaying institutionalization. Foster Grandparent volunteers work with children, including those with disabilities. AmeriCorps members address a range of local health issues.

The address and phone number for regional offices may be obtained through the main office in Washington, DC.

• Corporation for National Service

• Office of Public Liaison

• 1201 New York Avenue, NW

• Washington, DC 20525

• (202) 606-5000

Other Resources

• American Association of Retired Persons (AARP), Washington, DC, (202) 434-2277; 800-424-3410.

• AARP Pharmacy Price Quote Center, 800-456-2226

• American Bar Association Commission on Legal Problems of the Elderly, Washington, DC, (202) 662-8690

• Children of Aging Parents, Levittown, PA, (215) 945-6900

• Consortium to Establish a Registry for Alzheimer's Disease (CERAD), Durham, NC, (919) 286-6406 or 6405 National Association for Continence, Spartanburg, SC, 800-BLADDER (800-252-3337)

• Insurance Consumer Helpline, Washington, DC, 800-942-4242 Medicare Hotline, Baltimore, MD, 800-638-6833

• Medicare Beneficiaries Defense Fund, New York, NY, (212) 869-3850; 800-333-4114

• National Citizens' Coalition for Nursing Home Reform, Washington, DC, (202) 332-2275 National Hospice Organization, Arlington, VA, (703) 243-5900; 800-658-8898

• National Parkinson's Foundation, 800-327-4545, Miami, FL (East Coast); and 800-522-8855, Encino, CA (West Coast) National Stroke Association, Englewood, CO, (303) 771-1700; 800-STROKES.

• Social Security Information, 800-772-1213 (open 7 am-7 pm in all time zones) U.S. Department of Veterans Affairs, Regional Office, Veterans Assistance, Washington, DC, (202) 418-4343; 800-827-1000.

Instructions for Administration of Mini-Mental State Examination

Median Mini-Mental State Examination Score by Age and Educational Level

Overview

The six-item Blessed Orientation-Memory-Concentration (BOMC) test was developed by Katzman, Brown, Fuld, et al. (1983) from the longer (29-item) Blessed Information-Memory-Concentration (BIMC) test (Blessed, Tomlinson, and Roth, 1968). Katzman and colleagues (1983) selected 6 of the original 29 BIMC items based on a series of statistical analyses. The scores from each of the six items are multiplied as detailed below to yield a weighted score. Possible total scores on the BOMC range from 0 (all items answered correctly) to 28 (all items answered incorrectly). Weighted error scores greater than 10 are consistent with dementia, according to Katzman and colleagues.

Instructions for Scoring of the Blessed Information-Memory-Concentration (BIMC) Test

Instructions for Administration and Scoring of the Short Test of Mental Status (STMS)