New strategies may help PCPs diagnose cognitive impairment, dementia
Click Here to Manage Email Alerts
The CDC has previously said that diagnoses of Alzheimer’s disease and related dementia conditions are expected to increase 178% by 2060.
The Alzheimer’s Association issued a report last year to “sound an alarm regarding the future of dementia care in America.” According to the report, 39% of 1,000 surveyed PCPs reported “never or only sometimes” or “never being comfortable” making a diagnosis of Alzheimer’s disease or other dementias. The same survey showed that 50% of PCPs felt the medical profession is either “not very prepared” or “not at all prepared” to care for the growing number of people living with Alzheimer’s or other dementias.
Current methods used to detect cognitive decline and dementia in primary care are often underutilized or used too late, Pierre N. Tariot, MD, the director of the Banner Alzheimer’s Institute, told Healio Primary Care.
“Research, including that conducted and published by the Alzheimer’s Association, shows that up to half of persons with dementia in the U.S. will not have their dementia diagnosed or discussed with them before they die, and recognition occurs often when the dementia is already moderately severe,” he said.
During the virtual Alzheimer’s Association International Conference, experts presented data on three strategies that could potentially improve cognitive impairment and dementia diagnoses in primary care.
Natural language processing
Laura Maclagan, MSc, a quality improvement epidemiologist at the Institute for Clinical Evaluative Sciences, and colleagues used natural language processing to create a list of 910 words and features associated with dementia. The researchers then ran five different models that used patient progress notes, physician consultation notes and a combination of both notes to see how well those words and features predicted dementia among a cohort of primary care patients in Canada aged older than 66 years. The cohort included 526 patients with dementia and 44,148 patients without the condition.
The researchers reported that the risk prediction algorithms “achieved relatively good performance.” A model that combined consult and progress notes performed better than models using only one of these sources, they added. The best-performing model that used both notes had a sensitivity of 71.5% and a positive predictive value of 85.2%.
Maclagan told Healio Primary Care that with a little tweaking, PCPs may eventually be able to incorporate an algorithm or algorithms within electronic medical records “to help improve the recognition of cognitive issues and dementia in primary care.”
“It is most useful for capacity planning across health systems, as we will be able to more accurately pinpoint the time between early signs/symptoms and use of physician, hospital and long-term care services as well as medications related to dementia,” she said.
Mobile app
The second strategy involved a mobile app called Digital Mobile Assessment Cognition (DMAC) that Suresh Kumar, MD, a headache neurologist at the Headache TBI & Memory Institute in Frisco, Texas, and colleagues are developing as a “quick, comprehensive cognitive test.” The app would help physicians prescreen patients for memory loss and score various physical and cognitive functions. In general, lower DMAC scores indicate a greater likelihood that the patient may have dementia, according to the researchers.
Kumar and colleagues conducted a study to test the accuracy of DMAC in 318 patients with known memory loss and a control group of 33 patients without memory loss. The researchers reported that, on average, the memory loss group had DMAC visual test scores that were 15.18% to 27.47% lower than the control group (P < .001); auditory scores that were between 20.06% and 35.97% lower (P < .001); delayed recall scores that were between 32.95% and 53.16% lower (P < .001); disinhibition scores that were between 27.65% and 48.26% lower (P <. 001); attention scores that were between 18.64% and 35.73% lower (P < .001), executive function scores that were between 18.11% and 32.46% lower (P < .001), numerical recall scores that were between 19.83% and 35.67% lower (P < .001) and working memory scores that were between 16.21% and 35.64% lower (P < .001).
The average sensitivity of the app across these domains was 84.75% and the average specificity was 59.56%, according to the researchers. They said there was no clinically significant difference between the groups regarding language skills. On average, processing times were between 154.68 minutes and 445.15 minutes slower among patients with memory loss compared with the control group (P < .001).
“Most primary care physicians do not know how to interpret and what to do next with [standardized] cognitive testing reports,” Kumar told Healio Primary Care.
DMAC, when available, can help physicians “plan and provide” therapy based on patients’ scores, he added.
Telehealth
The third strategy uses a pre-existing tool known as Telephone Interview for Cognitive Status (TICS) to detect dementia. New data on this tool were presented by Hanna Abdulrahman, a PhD student at Amsterdam University Medical Center/Radboud University Medical Center in the Netherlands, and colleagues.
A previously published study in Neuroepidemiology described TICS as a telephone assessment of cognition. Patients receive a certain number of points based on their ability to recite their name, date, age and phone number; count backwards; complete word-list exercises; know the current president’s and vice president’s names; know the opposites of common words; and perform subtraction and finger tapping.
For the current study, the researchers set the optimal TICS cut-off score at 30 points. In general, the higher one’s score, the less likely he or she is to have dementia.
Abdulrahman told Healio Primary Care that although TICS is the “most widely used and researched telephone-administered diagnostic screening instrument for cognition,” few studies have assessed its value in the real world.
The researchers administered TICS to 810 community-dwelling older adults living in the Netherlands who had no dementia at baseline. Afterwards, 155 adults who had TICS scores of 30 or lower and a random sample of the 655 adults with a TICS score of 30 or higher were screened for dementia. EMRs showed 8.4% of the adults with TICS scores of 30 or lower were ultimately diagnosed with dementia compared with none of the adults with TICS scores of 30 or higher.
Overall, TICS performed with a sensitivity of 86.9%, specificity of 94.9%, positive predictive value of 77.4%, negative predictive value of 97.2%, and an area under the curve of 90.9% (95% CI, 81.2-90.9), according to the researchers.
“Our study results show the TICS to have an almost optimal negative predictive value compared to other known telephone-administered diagnostic screening instruments, such as the TELE and the Short Portable Mental Status Questionnaire,” Abdulrahman said.
‘Keep it simple’
Tariot, who is unaffiliated with the studies, said a decade of research and consultation with PCPs has convinced him that the best primary care tools for detecting cognitive decline and dementia are those that are straightforward and easy to understand.
“What we have learned is that we need to keep it simple in order to succeed,” he told Healio Primary Care.
“Standard treatments can be implemented sooner and have a better chance of preserving independence and quality of life,” he said. “Proactive avoidance of mismanagement of comorbid medical conditions and medications can improve health outcomes, reduce risk of delirium and avoid avoidable utilization of the acute care system.”
Other benefits include allowing patients to plan for the future, addressing safety issues such as driving and fall risks, and providing education and support to patients and their families, according to Tariot.
References:
Abdulrahman AH, et al. Diagnostic accuracy of the telephone interview for cognitive status (TICS) for the detection of dementia in primary care in the Netherlands. Presented at: Alzheimer’s Association International Conference; July 26-30, 2021 (virtual meeting).
Alzheimer's Association. Alzheims Dement. 2020;doi.org/10.1002/alz.12068..
Knopman DS, et al. Neuroepidemology. 2010;doi:10.1159/000255464.
Kumar S, et al. The digital mobile assessment of cognition as a quick comprehensive cognitive test for primary care. Presented at: Alzheimer’s Association International Conference; July 26-30, 2021 (virtual meeting).
Maclagan L, et al. Using natural language processing to identify signs and symptoms of dementia and cognitive impairment in primary care electronic medical records. Presented at: Alzheimer’s Association International Conference; July 26-30, 2021 (virtual meeting).
Matthews KA, et al. Alzheims Dement. 2018;doi:10.1016/j.jalz.2018.06.3063.
PAR Products. Telephone Interview for Cognitive Status. https://www.parinc.com/Products/Pkey/445. Accessed Aug. 16, 2021.