Evaluation of Cognitive Impairment in Older Adults

Because the U.S. population of persons 65 years and older is expected to grow, the annual number of cases of Alzheimer’s and other dementias is projected to double by 2050 (1). Current estimates of prevalence are that 10%−20% of persons over 65 have mild cognitive impairment (MCI), while 11% of those over 65 and 32% of those over 85 have dementia due to Alzheimer’s disease. Prevalence of all forms of dementia in the U.S. in those 71 or older is 13.9% (1). Among dementias, Alzheimer’s disease (AD) is by far the most common type, accounting for 50%−60% of cases (1, 2) with vascular, frontotemporal, and Lewy body dementias accounting for another 35% (3, 4). Although aging is the most consistent risk factor, those with fewer years of education are at higher risk for dementia related to having lower cognitive reserve (5).

The evaluation of cognitive disorders in older persons is complicated by the challenge of distinguishing them from depression and delirium, conditions that present with elements of cognitive and noncognitive neuropsychiatric symptoms that overlap with MCI and dementia. Further, they can be comorbid with dementia or its earlier phase, MCI. Another challenge is distinguishing normal aging from a dementing process. Education, cultural background, and primary language differences affect interpretation of symptoms and possible staging of a dementing disorder. The purpose of this article is to provide an overview of the approach to patients, as well as sufficient detail about practical clinical aspects of differential diagnosis of cognitive impairment in older persons, to benefit physicians who do not have additional training in geriatrics and wish to expand their competency, even if they eventually refer patients to a dementia specialist. We focus on the more prevalent conditions, whereas the subspecialist can diagnose atypical and less common disorders.

Depression and delirium are usually quite treatable disorders with high reversibility and therefore always need to be recognized and managed before considering a progressive condition as the cause of cognitive impairment. Delirium can be subsyndromal or even chronic and masquerade as a dementing condition. Delirium is often under-ecognized and misattributed to depression or dementia (6). Delirium is also often comorbid with dementia, ranging from approximately 15% (Alzheimer’s) to 35% (vascular dementia) (7). Major depression can cause memory deficits and be mistaken for MCI. Conversely, dementing illnesses can be accompanied by mood or apathy symptoms that get misattributed to depression. When mild, any of these may be confused as “normal aging.”

MCI, also called mild neurocognitive disorder (NCD) in DSM−5 (8), can have numerous etiologies, including depression, but may also herald the beginning stage of a dementing disorder. Some will revert to normal and many remain stable but approximately 10%−20% per year will convert to dementia with variations in rate depending on the length of follow-up (9, 10). Medical problems such as vitamin deficiencies, thyroid conditions, and adverse effects of drug therapies need to be carefully sought as sole causes or contributing factors to any of the 3 D’s (depression, delirium, dementia). Clinical interview, examination, and laboratory testing will be similar at the beginning of the evaluation irrespective of which of the 3 D’s is being considered. More advanced testing such as neuropsychological testing, neuroimaging, CSF analysis, or genomic testing will depend on the particular disorder(s) being differentially diagnosed for the individual patient; some testing might necessitate referral to a specialist.

The application of neuropathological, genomic, and newer neuroimaging techniques in research has revealed a significant rate of clinical misdiagnosis for types of dementias in both clinical and research settings, even when using strict diagnostic criteria (11, 12). Beach et al. (13) reported sensitivity ranging from 70.9% to 87.3% and specificity from 44.3% to 70.8% for possible or probable Alzheimer’s dementia as compared with autopsy diagnosis using longitudinal data from the National Alzheimer’s Coordination Center’s research database. As compared with neuropathological diagnosis at autopsy when using research criteria, Hogervorst et al. (14) found that identifying patients as a group who were possibly or probably demented had a higher (76%−100%) positive predictive value for Alzheimer’s disease as the cause, but a lower (18%−76%) negative predictive value. This highlights that while clinical evaluation can fairly reliably detect that MCI or dementia is present, advanced adjunctive testing targeting the underlying pathological process may be necessary to determine the actual diagnosis (2). Because prognosis, treatment, and temporal course vary with etiology, obtaining the right diagnosis is important. This article describes the typical clinical features of the major dementing disorders that, when complemented by objective diagnostic testing, may enable greater specificity than was possible prior to the advent of recent neuroimaging and laboratory biomarker tests (15).

The clinical rule of thumb is to first identify any component attributable to a major depression and/or delirium prior to considering a dementing disorder, acknowledging the complexity when these disorders are comorbid (see Figure 1). Some symptoms overlap across conditions, such as dysphoric mood, apathy, or forgetfulness, necessitating a careful characterization of the type of memory loss (secondary memory deficits due to inattention from depression or delirium; forgetfulness due to a retrieval memory disturbance related to white matter disease; amnestic memory pattern with the inability to learn and remember new information due to AD). Serial evaluations would include laboratory or imaging tests to identify reversible/treatable causes of cognitive impairment. In a meta-analysis (with 39 studies and 5620 dementia patients), potentially reversible causes were reported in 9% of dementia cases, with 0.6% cases actually resolved (0.29% partially, 0.31% fully) (16). These included metabolic (hypothyroidism, vitamin B12 deficiency), normal pressure hydrocephalus, brain tumor, depression, alcohol, infection, subdural hematoma, trauma, anoxia, and medications. This means that almost 1 patient in every 10 with dementia has a potentially reversible cause that should not be missed.

No matter the ultimate diagnosis determined, many patients and their caregivers/families are keenly aware of and fearful about the specter of having Alzheimer’s disease, which will often be top of mind and needs to be compassionately managed during the evaluation process (17, 18). Further, there is value in knowing the particular cause of MCI or dementia so appropriate medical management and psychosocial, financial, and occupational planning can occur more proactively (19–21). Caregivers note delays of about 3 years after symptom onset in obtaining a firm diagnosis of dementia type (usually by referral to a specialist) when general practitioners provide medical care (22–25). Delays and missed diagnoses—as many as 50% (26)—may be attributed to physicians’ discomfort with making a diagnosis or perceived harm (27).

Identifying cognitive impairment sooner may lead to earlier treatment of the cognitive condition, allow increased supervision of individuals to more adequately perform their activities of daily living (ADLs), improve treatment adherence rates for their other chronic conditions, reduce medication use errors, decrease ER and hospital visits, improve quality of life for the patient, and reduce financial and caregiver burden. For those patients with AD, we know that the current cognitive drugs work better the earlier they are started (28) and that when disease-modifying agents are discovered, earlier identification of cognitive impairment will be critical. Delaying the onset of or preventing AD would save billions of health care dollars and improve the quality of life for millions (1).

Depression and delirium need to be identified and managed prior to embarking on a workup for a dementing process (Figure 1). After treatment occurs for depression and/or delirium, if symptoms remain, the patient should be evaluated for a possible dementing process. A substantial stroke can result in an initial delirium that clears to reveal a significant depression and a more permanent cognitive problem. Isolated depressive symptoms that are subthreshold to a major depression may be secondary to a delirium, a dementing process or another medical/neurological condition.

Subtle cognitive deficits that are stable over a period of months and temporally related to mood changes suggest possible depression. Acute onset of cognitive impairment suggests delirium or a major neurological insult such as stroke. Fluctuating cognitive impairment suggests delirium or DLB. Chronic mild cognitive deficits suggest MCI or mild NCD or subsyndromal delirium. A rapid trajectory of decline but without all the features of delirium suggests a prion neurodegenerative process. Gradual persistent cognitive decline with increasing functional impairment suggests a neurodegenerative dementing process.

In addition to a careful history with an informant, the next step is formal mental status testing and a physical examination with an emphasis on measuring mood, behaviors, cognition, function (ADLs), and neurological signs. Administration of standardized scales for depression or delirium and cognitive tests objectively measure severity and patterns of symptoms. Tests can be repeated over time to monitor the patient’s trajectory. Findings on physical, laboratory, and neuroimaging examinations can help identify underlying medical conditions that may be causative.

Once depression and delirium are ruled out or managed, the diagnosis of dementia commences and involves a multistep approach. Additional medical history and physical examination may need to be performed, including assessment of risk factors and early warning signs (2, 57). Results from office-based cognitive and standard laboratory tests may need to be supplemented by formal neuropsychological testing, neuroimaging, and cerebrospinal fluid (CSF) examination to enable a more specific diagnosis. Some degenerative disorders are not well detected with current methods (e.g., tauopathies). The presentation of dementia symptoms may differ based on the nature of affected brain regions and stage of progression of the underlying neuropathology both inter- and intraindividually. Nonetheless, characteristic phenotypes (see disease state sections) can provide some direction to the order of tests performed in an evaluation. It is important to be aware that more than one etiology for a dementing process may be present so that mixed dementias need to be considered. AD mixed with either vascular or DLB etiologies is not uncommon and produces atypical clinical presentations.

Objective assessment of cognitive abilities is an initial step in differential diagnosis. The cognitive loss in those developing MCI and degenerative dementia usually starts insidiously. These patients may have impaired insight and typically do not complain of significant memory loss, and casual conversation with their physician does not uncover deficits. Consequently, these patients present to a physician an average of 3 to 4 years after cognitive symptoms began (88). Screening for cognitive impairment in physician offices may help with detection in older at-risk patients (89). Each person has different natural talents and will have different baseline scores on their cognitive testing. Therefore, it would be ideal for every senior to have screening cognitive assessments prior to any decline in their cognitive abilities. This may practically be performed in the U.S. during the Medicare Annual Wellness Visit (90).

It is important to distinguish among cognitive screening (not in response to a cognitive complaint), office-based multidomain cognitive assessments to aid diagnosis, and formal neuropsychological testing also to aid diagnosis (performed by computer, neuropsychologist or practitioner). Patients with subtle or atypical patterns of cognitive impairment may require neuropsychological testing, whereas for those with mild to moderate cognitive symptoms, an office-based multidomain test may suffice.

When evaluating cognitive decline, the first step is objective assessment appropriate to the suspected condition. For example, delirium may be tested using simple tests of attention such as having the individual repeat a series of numbers or recite the months of the year backward, whereas depression has subtler deficits requiring more challenging tests. In MCI or dementia, patterns of deficits can be helpful in the total patient evaluation. For example, deficits in executive function suggest prefrontal dysfunction, and deficits on word list learning tests suggests hippocampal dysfunction. However, cognitive tests alone cannot determine the etiology of a delirium or dementia process.

A multidomain brief cognitive test (see Table 2) such as the well-known Mini-Mental State Exam (MMSE) (91), the Montreal Cognitive Assessment (MoCA) (92), Saint Louis University Mental Status (SLUMS) (93), or the Self-Administered Gerocognitive Examination (SAGE) (94) is an easy way to obtain a standardized score across a range of cognitive domains and takes about 10 to 15 minutes to complete. While they alone cannot distinguish among the 3 D’s, they offer a quick measure of severity of overall cognitive impairment. Delirium and dementia patients would be expected to score lower than those with MCI or depression.

Depressed patients will have subtle cognitive deficits where their complaints may be out of proportion to the actual deficits on objective testing. An office-based screening test may be suitable for initial testing. Follow-up for unresolved deficits after the depression has remitted is recommended. Delirium is defined by attentional deficits that, if severe, make it difficult to test other cognitive domains.

Office-based cognitive tests for MCI need to be sensitive enough to detect early deficits. For example, the MoCA can measure executive function and the impairments in language and memory often found in early stages of AD better than the MMSE (95). The SLUMS has some similarities to the MoCA but has more items related to memory and fewer to language, executive, and visuospatial tasks.

The SAGE can identify both MCI and early dementia (94). The SAGE is unique in that it is self-administered in pen and paper format and does not require staff time to administer. It has four equivalent interchangeable forms. It can be used for cognitive screening or as an office-based multidomain test.

The AD-8 is an 8-item informant-based dementia screening questionnaire inquiring about the presence or absence of cognitive and functional symptoms but does not directly measure cognition (96). Scores of 2 points or more suggest cognitive impairment. Using amyloid biomarkers, the AD-8 was found superior in detecting dementia of the Alzheimer’s type than the MMSE (97). Simple word list learning tests like the Memory Impairment Screen (MIS) and animal fluency have been advocated in primary care settings as a cognitive screen followed by more sensitive tests (98).

Table 3 lists types of laboratory and neuroimaging tests that are performed during differential diagnosis for depression, delirium, and dementing disorders. Blood and urine laboratory tests are similar when evaluating among the 3D’s and serve to help determine underlying medical problems causing a delirium or reversible causes of cognitive impairment such as vitamin deficiency, urinary tract infections, and hypothyroidism. These are performed to identify and manage possibly reversible conditions prior to obtaining CSF examination and neuroimaging, the latter being reserved for when there is suspicion of a CNS abnormality that is causative for the symptoms. It is not uncommon to obtain structural neuroimaging (CT or MRI) in delirium and dementia but less common for depression. Depending on the dementia etiology suspected, an MRI may often be the wiser choice because it detects more subtle abnormalities such as white matter hyperintensities that are associated with slow information processing speed. MRI is very good at detecting abnormalities such as strokes, tumors, hematomas, infectious cysts, inflammation, hemorrhages, and subtle lesions in the brainstem. Although MRI may detect cortical or hippocampal atrophy patterns and ventricular dilation associated with some dementias (e.g., AD, FTD), it is unable to reliably determine the cause for degenerative dementias such as AD, DLB, PDD, and FTD which require specific molecular detection methods (e.g., of beta amyloid, tau, dopamine, and so on). Because a structural brain scan (e.g., MRI) can provide information about a wide variety of lesions that might be attributed to the cognitive impairment, it is always ordered prior to PET scans or CSF examination. However, because conditions can be comorbid—for example, AD and small strokes—both MRI and PET may be ordered to more fully evaluate the underlying etiologies to differentially diagnose the individual patient.

The question often arises as to whether, when, and in what order to obtain CSF and other neuroimaging tests on an individual patient (Figure 1). Unclear, mixed and atypical findings may precipitate consideration of these additional evaluations. According to Jack et al. (99), amyloid is the earliest biomarker that can be detected as abnormal in those with AD. There is a progression of biomarkers that appear later that can be detected after amyloid in those with AD pathology: markers of neurodegeneration using MRI looking for cortical and hippocampal atrophy and metabolic defects that may reflect reduced synaptic activity by using FDG PET.

PET neuroimaging available clinically includes amyloid plaque detection, an important marker of AD neuropathology, and FDG PET to measure brain metabolism. Florbetapir-F18 PET is currently the only FDA-approved amyloid PET radiotracer. Florbetapir specifically labels and estimates the burden of beta amyloid plaques (positive scan indicates moderate to frequent plaques and negative scan indicates none to sparse plaques) that are found in patients with AD and some other amyloidogenic conditions (100). (See also Appropriate Use Criteria for amyloid PET [101]). Pittsburgh Imaging Compound-C11 (PIB) also labels amyloid plaques but is used in research settings. Amyloid PET should be considered only in those with documented cognitive deficits suspected of being due to AD and interpreted in the context of an individual patient’s other clinical information. A percentage of normal older persons can have positive amyloid PET scans where the prognostic value is unclear at this time, although those with amyloid have poorer neuropsychological performance than those without it (102, 103). Emerging research also finds faster trajectories of decline in those who have amyloid pathology than those who do not (104–107).

Those with atypical presentations often have AD pathology that can be detected using amyloid PET while their metabolic patterns on FDG PET reflect the clinical phenotype (110). Mixed VD is most likely to be comorbid with AD than with other neurodegenerative diseases (69), where both MRI and amyloid PET can be applied. FDG PET detects frontal hypometabolism that underlies FTD but also in frontal variant AD that presents with prominent executive dysfunction and can be mistaken for FTD (111), in which case amyloid PET may also be needed.

Figure 2 displays results of MRI, FDG PET, and florbetapir PET scans from two patients diagnosed as either FTD or AD to show how they contribute different types of information about the brain. In AD the MRI shows more generalized cortical atrophy often more prominent in hippocampal regions, a temporoparietal pattern of metabolic defect on FDG PET, and abnormal levels of amyloid plaque on florbetapir PET. In FTD the MRI shows frontotemporal cortical atrophy, FDG PET shows metabolic defects in anterior regions, and florbetapir PET does not show abnormal levels of amyloid plaque.

When extrapyramidal motor symptoms are present and PDD or DLB are in the differential diagnosis, a SPECT scan that labels the dopamine transporter can detect reduced dopamine integrity in the striatum. Because DLB and PD can have coexisting Alzheimer’s pathology, an amyloid PET scan may also be useful for amyloid detection (76). Rates of AD are much higher in DLB than PD patients (75) and amyloid presence may impact disability (77).

CSF assays of biomarkers for AD (Amyloid-beta 42 and tau levels) as well as more routinely performed CSF assays for blood cells and immunoglobulins, proteins, bacteria, and viruses can be adjunctive in the diagnostic process depending on the suspected etiologies. Low levels of amyloid-beta 42 (Aβ₄₂) and high levels of tau and phosphorylated tau in the CSF are consistent with AD. CSF assays for AD biomarkers still need more standardization and have not been approved by the FDA (112). CSF Aβ₄₂ has high correlations with amyloid PET (107). Referral to a dementia clinic for CSF examination when appropriate is recommended.

Further, genomic testing is useful for certain conditions such as Huntington’s disease. Apolipoprotein E epsilon 4 allele has been associated with greater risk and earlier onset of sporadic AD but itself is insufficient for a diagnosis of AD. Autosomal dominant genes implicated in familial AD include mutations of APP, PS1 and PS2 genes. Certain genetic biomarkers are being established for FTD (see Table 1).

When delirium is suspected, in addition to cognitive testing, a more complete assessment of the range of delirium symptoms can be accomplished using validated tools like the 16-item Delirium Rating Scale-Revised−98 (DRS-R98) (113) for diagnosis and severity ratings or the Memorial Delirium Assessment Scale (MDAS) (114) for symptom severity rating, both designed for use by psychiatrists. Patients who are not rousable for interview may instead be stuporous or sedated by medications, requiring the evaluation be delayed until sedation is reduced or level of arousal improved to allow consciousness to be evaluated for presence of delirium symptoms.

Routine use of a symptom checklist or depression rating scale such as the Geriatric Depression Scale (GDS) (115) and maintaining a high index of suspicion for atypical depression is important, rather than simply accepting the patient or care partner’s report of “depression.”

The Clinical Dementia Rating (CDR) scale is a global measure where the clinician rates the patient’s orientation, executive skills, memory, and functional abilities after a careful evaluation (116). The score ranges from 0 to 3 with 0.5 suggesting mild cognitive impairment or very early dementia and 1 or more suggesting a dementing condition.

The Neuropsychiatric Inventory-Community version (NPI-Q) (117) can be completed by an observer or caregiver who knows the patient. It rates severity of 12 items on a 3-point Likert scale: dysphoria/depression, irritability/lability, anxiety, euphoria/elation, apathy/indifference, aberrant motor behavior, agitation/aggression, disinhibition, delusions, hallucinations, nighttime disturbances, and appetite/eating disturbances. These noncognitive neuropsychiatric symptoms are commonly associated with dementing conditions, where symptom patterns may vary depending on the brain regions affected by the disease.

The Functional Activities Questionnaire (FAQ) (118) is a validated tool that is easy to use in office settings to rate 10 Independent Activities of Daily Living and some basic ADL items on a severity scale from 0 to 3 that can be completed by the patient or the informant. Scores range from 0 to 30 and those greater than 6 may indicate functional impairment in Independent Activities of Daily Living (119). The FAQ can be used across the range of cognitive impairment from milder to full dementia stages.

The Mayo Fluctuations Scale can assist with the quantification of fluctuations in symptom severity when DLB is in the differential diagnosis, where fluctuation is a core defining feature (120).

The modified Hachinski index (121) is a short scale capturing the essence of stepwise deterioration of vascular disease that can lead to cognitive impairment. Scores ≤4 are consistent with AD and scores ≥7 points are consistent with a likely ischemic origin of symptoms and can be adjunctive in evaluation of possible vascular dementia.

The information provided in this article discusses common disorders impairing cognition that may present in patients coming to the general psychiatrist’s office and practical ways to identify and diagnose these conditions in an office-based practice. The demographic shift toward a growing elder population means many more physicians beyond dementia specialists will be needed to perform thorough assessments to identify and diagnose cognitive impairments, especially at their earlier stages when management and interventions targeted at the disease, the patient’s welfare and functioning, and the family’s needs can make an enormous difference in the lives of patients. Many patients and care partners want to know the specific diagnosis of their cognitive impairment, even if it is a dementia diagnosis, and psychiatric training is well-suited to conduct both the neuropsychiatric evaluations and medical discussions that are sensitive.