The JAMA commentary by Carol Brayne et al. carefully considers the issues surrounding routine screening for dementia by primary care practitioners (PCPs). Although enthusiasm has repeatedly been voiced for widespread screening for dementia in the elderly by PCPs, and dementia is under-recognized and not diagnosed adequately in primary practice, there are many obstacles that obstruct screening and detection. Even if brief tools were available, busy PCPs lack the time to screen for dementia and may have difficulty finding the time and resources to complete the dementia workup and institute care and treatment. It is also unclear whether the general public is enthusiastic about screening for dementia, in particular because a positive screen carries with it risks such as loss of driving privileges, and it is unclear what reassurance a negative screen offers. In any case, why screen in the absence of highly effective medical treatment? Systematic studies of screening for dementia and their consequences are rare, and the commentary highlights the findings that older patients and those...  Read more

The JAMA commentary by Carol Brayne et al. carefully considers the issues surrounding routine screening for dementia by primary care practitioners (PCPs). Although enthusiasm has repeatedly been voiced for widespread screening for dementia in the elderly by PCPs, and dementia is under-recognized and not diagnosed adequately in primary practice, there are many obstacles that obstruct screening and detection. Even if brief tools were available, busy PCPs lack the time to screen for dementia and may have difficulty finding the time and resources to complete the dementia workup and institute care and treatment. It is also unclear whether the general public is enthusiastic about screening for dementia, in particular because a positive screen carries with it risks such as loss of driving privileges, and it is unclear what reassurance a negative screen offers. In any case, why screen in the absence of highly effective medical treatment? Systematic studies of screening for dementia and their consequences are rare, and the commentary highlights the findings that older patients and those with mild deficits frequently refused further workup after failing a cognitive screen (e.g., Boustani et al., 2007; Agency for Healthcare Research and Quality; Boustani, 2006; Boustani, 2003). The commentary is sobering, and underscores the difficulty of advocating widespread screening for dementia, particularly in the absence of disease-modifying treatment for Alzheimer disease (AD).

By framing the question as “screening for dementia,” the commentary did not consider the full range of future needs and goals. It is becoming increasingly important to consider how to develop a brief screening test that has adequate sensitivity and specificity to detect early cognitive changes in AD and other dementias, even before the full dementia syndrome is present. Dementia can be viewed as brain failure. Screening for other common medical disorders, for example, cancer or diabetes, also is based on the premise of early detection. Efforts to screen cancer would not be popular if they only targeted metastatic or severe disease. Similarly, efforts aimed at diabetes do not target the disease at the stage when end-organ complications are already apparent. Admittedly, treatment is available, even early in the course, for both diabetes and cancer. However, research into dementia should encourage methods aimed at screening for cognitive difficulties that may constitute very mild cognitive impairment that are likely to progress to future dementia—which would presumably capture more blatant cases of dementia in the process.

Enlarging the question of screening to include early detection of cognitive decline or very mild dementia raises new opportunities, as well as concerns:

If screening assesses cognitive abilities or memory performance sensitively, it could be presented as having positive value, e.g., part of an attempt to maintain healthy cognitive aging, which may circumvent the negative connotations of screening for “dementia.” As Brayne et al. point out, it is extremely unusual to find reversible causes of dementia. However, many potentially reversible factors can contribute to cognitive performance that is worse than expected as a result of aging alone—for example, medical illness, depression, medications, or cardiovascular factors. Many of these may be amenable to intervention. In addition, a negative screen can be used to reassure individuals who are concerned about their self-perceived decline in cognitive performance that they are most likely experiencing age-related changes, rather than the beginnings of Alzheimer disease or another dementia. By comparison, a subject who “passes” a relatively insensitive screening test that targets dementia cannot be reassured; he or she may, in fact, have prodromal features of dementia that the screening cutoff does not detect adequately.

Who should be screened using this type of approach? In the current absence of disease-modifying treatment, it seems most reasonable to screen individuals upon request, either of the person or a concerned spouse or relative. Such screens would most likely be triggered by symptoms or by concerns about risk (the worried well). As progress is made in intervening against AD, it is likely that treatment trials and approved treatments will be targeted at people who show the earliest symptoms, or even at those who are presymptomatic but have markers suggesting underlying pathology. Population-wide screening based on risk factors and age would certainly be appropriate once the field reaches that stage.

The commentary does not discuss in detail who should carry out screening and how the screening should be performed. Several brief screening tests already exist, some of which have reasonable sensitivity and specificity for dementia. Interviews about functional status, such as the AD8 featured on Alzforum previously, are one example. These have higher sensitivity when the information is obtained from an informant than from the patients themselves, adding a little more complexity to gathering the information. Cognitive screening is more difficult to introduce into primary practice. The shorter the test, the more likely it is to have unsatisfactory metric properties, compromising sensitivity and specificity. Longer tests require greater expertise and time on the part of the PCP.

It may be useful to think outside the paradigm of the PCP carrying out the screening. A number of cognitive assessments have been developed that can be accessed on a computer or via the Internet. These could be carried out in the physician’s waiting room or in the comfort of the patient’s home, and the patient could bring the results of such testing, together with available normative data, to the PCP’s office for interpretation. Another approach would be to use a brief, standard neuropsychological battery as an intermediate level screen, administered by a trained psychometrist. This second approach lends itself more easily to screens triggered by requests or complaints from patients and their families rather than to broad-based, population-based screening.

In the absence of disease-modifying treatment, the merits of early detection of dementia are thought to include social interventions and planning the future. The studies discussed by Brayne et al. suggest that these are not necessarily perceived as positives that outweigh potential loss of independence or of driving privileges. Further studies that attempt to implement screening will need to measure these factors. However, further study of attitudes and perceptions, together with public education efforts about potential benefits of early detection, could change this landscape.

Finally, the lack of convenient biomarkers as screening tools was mentioned and merits further discussion. A biomarker based on a blood test or imaging may be able to detect features related to amyloid or other brain pathology. However, the biomarker cannot read out the clinical consequences of such pathology. It requires a clinical assessment to decide whether somebody likely to harbor pathology is cognitively intact or has deficits. Because the prevalence of asymptomatic AD pathology rises with age and may precede clinical symptoms by decades, biomarkers will provide a positive readout far more frequently than a clinical screening method. Sensitive biomarkers of pathology may find a place as a secondary step after clinical screening. Thinking ahead to a critical future need, screening biomarkers that signal the presence of pathology could be linked to select people for preventive treatment aimed at targets based on pathology.

As efforts to diagnose AD and other dementias tilt toward earlier interventions and disease-modifying strategies, it behooves us to consider a number of approaches to screening. Brayne et al. raise many critical issues that will need to be assessed rigorously when field tests of novel screening methods are conducted in the future.

View all comments by Douglas Galasko

Comment by Paul Solomon and Cynthia Murphy
Carol Brayne, Chris Fox, and Malaz Boustani provide further fuel to the growing debate of whether widespread screening for Alzheimer disease should be implemented. In this paper, they specifically address screening in primary care settings and conclude that it is not appropriate.

The authors’ position in their paper is consistent with previous reports of the United States Public Service Task Force (USPSTF) in 1996 and 2003, of which Boustani was lead author (Boustani et al., 2003). Their position is also consistent with the UK National Screening Committee. Nevertheless, early in the JAMA paper the authors recognize a growing trend supporting screening for Alzheimer disease. Specifically, a number of groups have begun to advocate for routine screening in elderly Americans. For example, 16 November 2007 was “National Alzheimer’s Screening Day,” and including memory screening as part of the “Welcome to Medicare physical” is now under consideration by the Alzheimer’s Disease...  Read more

Comment by Paul Solomon and Cynthia Murphy
Carol Brayne, Chris Fox, and Malaz Boustani provide further fuel to the growing debate of whether widespread screening for Alzheimer disease should be implemented. In this paper, they specifically address screening in primary care settings and conclude that it is not appropriate.

The authors’ position in their paper is consistent with previous reports of the United States Public Service Task Force (USPSTF) in 1996 and 2003, of which Boustani was lead author (Boustani et al., 2003). Their position is also consistent with the UK National Screening Committee. Nevertheless, early in the JAMA paper the authors recognize a growing trend supporting screening for Alzheimer disease. Specifically, a number of groups have begun to advocate for routine screening in elderly Americans. For example, 16 November 2007 was “National Alzheimer’s Screening Day,” and including memory screening as part of the “Welcome to Medicare physical” is now under consideration by the Alzheimer’s Disease Screening Discussion Group, November 2007, Washington D.C.

Brayne et al. distinguish between the current method of detecting AD patients in clinical settings, in which the clinician suspects cognitive problems for the patient and initiates an evaluation, and screening asymptomatic patients over the age of 65 in a primary care practice. The authors aptly point out that clinical recognition of symptoms is different from screening a population with no reported or observed symptoms.

The authors indicate that before screening can be advocated, the benefits must outweigh the risks. They conclude that this is not currently the case and present arguments in four general areas to support their view (see Table below). While each of these arguments has merit, we might suggest that there are equally meritorious arguments in the affirmative for broad-based screening:

Brayne et al. Criticisms	Alternative Viewpoints
Screening Tests: Although existing instruments have good sensitivity and specificity, the positive predictive values are low.	Positive predictive values can be improved by enriching the screening sample. Recent guidelines (Solomon and Murphy, 2005) address this by recommending frequency of screening as a function of age, risk factors, and cognitive complaints. For example, one set of guidelines suggests discretionary screening from age 65-74 (prevalence = 3%) based on family history and cognitive complaints from the family or patient; from age 75-84 (prevalence = 19%) every 2 years or if there are complaints from the family or patient; and for age 85 and up (prevalence = 47%) annually.
Treatments: Disease-modifying drugs may only be beneficial if the disease can be modified before the onset of symptoms.	There are multiple potentially disease-modifying treatments in clinical trials. For example, Flurizan has positive Phase 2 data; Phase 3 results will be available in 2008. Bapineuzumab (passive vaccination) will present Phase 2 results in early 2008, and Elan/Wyeth have already announced a Phase 3 trial. Several companies are pursuing trials with secretase inhibitors. It is self-apparent that in the treatment of a neurodegenerative disease, intervention at the earliest possible stage will be most beneficial.
Screening Programs: Patients may refuse follow-up care after a positive screen because of fear of loss of driving privileges or denial of long-term care insurance.	Unfortunately, denial of long-term care insurance is already taking place because some insurance companies routinely screen applicants for cognitive deficits. Similarly, several states are now considering screening elderly drivers. The question then becomes not whether broad-based screening will come about, but who will conduct the screening.
Limited Evidence: There have been no large-scale RCTs demonstrating the benefits of screening.	Absence of proof is not proof of absence. Interestingly, although the most recent USPSTF report did not recommend broad-based screening, it did recognize that screening to detect dementia at an early stage is desirable because only intervention at an early stage can modify an otherwise certain decline.

All of the above notwithstanding, perhaps the best argument for the implementation of broad-based screening in primary care and/or community settings is that although there are currently an estimated 5.5 million people in the United States with AD, by some estimates only half of these patients are diagnosed and one quarter are treated (Solomon and Murphy, 2005). Undiagnosed patients experience difficulty medically, socially, and financially. They have difficulty managing medications and do not accurately report the symptoms of comorbid illnesses; they are at greater risk for automobile accidents; they make poor financial judgments and can be victimized by scams. They do not have access to symptomatic medications that have been shown to be most beneficial when introduced early in the disease, nor will they have access to future disease-modifying medications, which will also be most beneficial early in the disease. While Brayne et al. summarize the consequences of broad-based screening for AD, I might suggest that they do not adequately consider the consequences of not screening. When fewer than half of patients with the disease are diagnosed, and only half of those diagnosed are treated, what we are doing is clearly inadequate. As the number of AD patients is set to grow four- to fivefold in the next 50 years, these inadequacies will be greatly magnified. With this in mind, we might argue that broad-based screening is a program whose time has come.

View all comments by Paul Solomon