The Dementia Screening and Detection Meeting, held on 22 November in New York City, hosted a day of data exchange and debate toward a set of joint recommendations for improving dementia detection in primary care practice. Meeting participants shared, some for the first time, results of dementia-screening programs conducted as part of research studies. The Alzheimer’s Drug Discovery Foundation (ADDF) and the Alzheimer’s Foundation of America (AFA), both based in New York City, organized the meeting.
A common theme running through the presentations, regardless of patient population or setting, was that many of the available cognitive screening tests are effective in detecting dementia, but that a large proportion of patients who screen positive on those tests do not return for further follow-up. Frederick Schmitt, University of Kentucky Medical Center in Lexington, presented preliminary screening results from the PREADVISE (Prevention of AD with Vitamin E and Selenium) study. This trial piggybacked on a larger one (the SELECT trial) being conducted by the National Cancer Institute on the effects of the supplements in reducing prostate cancer. PREADVISE investigators conducted annual cognitive screens on 7,546 men over 65 years of age (or over 60 if they were African-American) who enrolled in the cancer study and were followed for seven to 11 years.
To detect dementia, the investigators used a cut-off score of 5 on the Memory Impairment Screen (MIS), a four-minute, four-item delayed free and cued recall memory test (Kuslansky et al., 2002). The cancer study ended in 2008, seven years after it started, because a futility analysis indicated that the supplements were not beneficial but harmful. When the SELECT trial ended, PREADVISE investigators who had been performing the MIS in person at SELECT trial sites switched to screening participants by phone using a telephone version of the test; the last screen is planned for 2013.
This changing of methodology may complicate the interpretation of results. “We had 9 percent of people failing the MIS by telephone versus 1.6 percent in person,” said Schmitt. He was unsure about the reason for this difference between the two types of test, but suggested the higher failure rate of the telephone MIS could simply reflect the fact that the men being screened by telephone had become older (since this type of screen was done at a later stage of the trial), and thus more of them would have cognitive problems.
People who failed either the in-person or telephone MIS were invited for follow-up and more extensive evaluation for dementia. Two-thirds took up the offer. “Even though this is a research population volunteering for a cancer study, we see the same results as other studies: no return on positive screen, physician reluctance to follow up, noise on comorbidities,” said Schmitt.
The dropout rate in PREADVISE was consistent with findings presented by Malaz Boustani, Indiana University School of Medicine in Indianapolis. He offered cognitive screening to 825 participants enrolled in the Indianapolis Dementia Screening and Diagnosis study. Whereas 97 percent of study participants agreed to be screened for dementia, half of those who screened positive did not return for further follow-up. “We were really surprised about that,” said Boustani.
Switching gears, David Salmon of the University of California, San Diego, presented results of a screening service set up for primary care physicians in the San Diego area. Physicians suspecting that a patient might be cognitively impaired could refer that patient to the screening service at no charge. (The Alzheimer's Research Centers of California and the National Institute on Aging funded the screening.)
Salmon and colleagues conducted a fairly extensive screen which involved taking a history of memory and medical problems, a test for depression, and a memory screening test battery consisting of the Mini-Mental State Examination (MMSE); Wechsler Memory Scale-Revised (WMS-R) Logical Memory Test, which probes immediate and delayed recall; the Trail Making Test parts A and B, in which patients are asked to connect numbers and/or letters in a sequential order; and the Seven Minute Screen for Dementia, which includes the Benton orientation test, enhanced cued recall test, clock drawing test, and semantic verbal fluency test (Solomon et al., 1998). The scores were compared to those of cognitively normal, healthy people. Based on that comparison, the patient was classified as having high, moderate, low, or undetermined probability of dementia.
The screening service saw about 150 people per year. After the first 519 people had been screened, Salmon and colleagues analyzed the findings. This revealed that 196 people had two or more cognitive deficits, indicating possible dementia; 80 had signs of depression; 64 only had memory problems, indicating possible mild cognitive impairment (MCI); 137 had no cognitive deficits; and 42 had other problems, such as stroke, that complicated diagnosis.
Of the people screened, 133 had MMSE scores in the upper twenty-fifth percentile—in other words, normal. Yet 47 percent of those 133 had inappropriate responses on immediate recall, and 61 percent had impaired delayed recall; both deficits were detected using the WMS-R Logical Memory Test. “If the MMSE is all the physician can do, we would not see memory impairment in those people,” said Salmon. The Seven Minute Screen for Dementia is a more detailed test than the MMSE. But with that test, too, 29.5 percent of people with low probability of dementia nonetheless had problems with immediate recall and 39.2 percent with delayed recall.
A year and a half later, Salmon and colleagues retested some of the patients. They found that 12 people who had tested normal in the first screen and five people who were depressed showed no cognitive decline in the second screen. On the other hand, eight people with MCI on the first screen and five people identified as being demented showed a decline in MMSE scores. “This provides a bit of validity for the screen,” said Salmon. “If we call someone impaired, they do actually decline.”
He also looked at whether the primary care physicians who referred their patients to the service were using the information provided. Salmon and colleagues looked at patients’ charts six months after they were screened, and in 75 percent of cases the physician had made a note that he or she had discussed the test with the patient. In 65 percent of cases, the physician had recommended further workup, specialty referral, changes in treatment, or some other action. Salmon did not know, however, whether that action resulted in a better outcome for patients in the long run.
The study Salmon presented involved patients who were referred to be screened by their physicians. Another presentation instead focused on screening patients in a primary care setting in the absence of a physician’s recommendation. The project, presented by J. Riley McCarten of the University of Minnesota Medical School, Minneapolis, was conducted at seven Veterans Affairs (VA) medical centers within the VA Midwest Healthcare Network, in the states of Minnesota, North Dakota, South Dakota, Iowa, and Nebraska. At these centers, McCarten’s group put together teams including a coordinator familiar with treating dementia, a primary care physician, and a nurse trained to conduct cognitive screens. During routine primary care visits, the nurses screened veterans over 70 years old using the Mini-Cog test, which took 90 to180 seconds to administer. The test combines three-item recall with the clock-drawing test. When someone failed the test, which had a cut-off score of 3, the nurse had the following scripted response: “Based on the screening test, my recommendation is that we get further evaluation to help sort things out.”
Out of 43,023 patients seen at the clinic, nurses offered the screening to 8,342 people, primarily white men, over 70 years of age who had no previous diagnosis of dementia. Almost all (8,063) accepted. Of those screened, about a quarter failed the Mini-Cog test and 28 percent of those who failed said they would do a thorough diagnostic workup. Of those 580 people, 432 proved on follow-up to have dementia, 108 had cognitive impairment, and 40 had no cognitive impairment, i.e., were false positives.
On the flip side, of the 5,982 who passed the test, 118 requested a follow-up evaluation, and of those, 82 had dementia, 21 had cognitive impairment, and 15 had no cognitive impairment. This means that some people with dementia or cognitive impairment had escaped detection in the original screen (i.e., false negatives); the people who requested the follow-up knew subjectively that something was wrong, and the 15 who proved fine on the test and follow-up fall into a group researchers call the ‘worried well.’ McCarten also highlighted the fact that 132 of the patients screened had Mini-Cog scores of 0, and 256 patients had a score of 1, which indicates severe cognitive impairment, “yet there was no record they had anything wrong with them, again confirming that dementia goes undetected in the primary care system,” he said at the meeting.
The last presentation of the AFA-ADDF meeting dealt with a different type of screening, that is, one carried out in the community rather than a doctor’s office. Peter Bayley of the VA Palo Alto Healthcare System in California presented data from the AFA’s 8th annual National Memory Screening Day held 16 November 2010. The event engaged more than 2,300 sites that provide confidential memory screenings, including senior centers, libraries, long-term care residences, hospitals, doctors’ offices, and pharmacies. Volunteers ranging from medical students and nurses to physicians and retired physicians conducted the screenings. More than 60,000 people were screened on that day, but Bayley presented preliminary analyses on subsets.
The AFA allows sites to use the screening tool of their choice, and the seven that were most widely used by participating sites were: Mini-Cog, MIS, General Practitioner Assessment of Cognition (GPCOG), the Kokmen Short Test of Mental Status (Kokmen et al., 1991), MMSE, Montreal Cognitive Assessment (MoCA), and Saint Louis University Mental Status Exam (SLUMS). “All tests had very similar failure rates,” said Bayley in his presentation. Taking the results from 3,398 people tested on National Screening Day, Bayley found that, on average, 11.7 percent of participants failed their test. “This result was in agreement with published data that one in eight people aged 65 and over have AD,” he added.
In another analysis, Bayley looked at the proportion of people recommended for follow-up with their primary care physicians. Out of 2,420 people screened, 294 scored below normal on their test. The volunteer test administrators recommended that 94 percent of these 294 people (i.e., 276 people) follow up with their doctors for further evaluation. However, of the 2,126 people who passed their test, 47 percent (i.e., 999 people) were also recommended for follow-up. “Whether an individual was recommended for follow-up might depend on who is conducting the screening. We had a variety of healthcare providers involved at different sites, and they were likely to err on the side of caution,” Bayley told ARF. He is continuing to analyze the data and will present a fuller interpretation at the annual meeting of the American Association for Geriatric Psychiatry to be held in Washington, DC, in March 2012.
For Bayley, the central conclusion is that many people in the community want to be screened. “Regardless of whether you think screening is good or not, there is a level of concern in the community that needs to be addressed,” he told ARF.—Laura Bonetta.
This is Part 2 of a two-part series. See also Part 1.
- Kuslansky G, Buschke H, Katz M, Sliwinski M, Lipton RB. Screening for Alzheimer's disease: the memory impairment screen versus the conventional three-word memory test. J Am Geriatr Soc. 2002 Jun;50(6):1086-91. PubMed.
- Solomon PR, Hirschoff A, Kelly B, Relin M, Brush M, DeVeaux RD, Pendlebury WW. A 7 minute neurocognitive screening battery highly sensitive to Alzheimer's disease. Arch Neurol. 1998 Mar;55(3):349-55. PubMed.
- Kokmen E, Smith GE, Petersen RC, Tangalos E, Ivnik RC. The short test of mental status. Correlations with standardized psychometric testing. Arch Neurol. 1991 Jul;48(7):725-8. PubMed.