Population Health Expert Panel

Larry W. Chambers, Christina Bancej and Ian McDowell, Editors

Measurement of Cognitive Impairment and Dementia

Dementia refers to progressive impairments in memory and other cognitive functions.

A diagnosis of dementia is based on a pattern of signs symptoms such as the ten warning signs of dementia:

  1. Memory loss that affects day-to-day abilities,
  2. Difficulty performing familiar tasks,
  3. Problems with language such as forgetting or using wrong words,
  4. Disorientation in time and space,
  5. Impaired judgment,
  6. Frequent problems with complicated tasks,
  7. Misplacing things,
  8. Sudden changes in mood and behavior,
  9. Changes in personality, and
  10. Loss of interest in doing things (Alzheimer Society of Canada 2015).

Dementia lies at the severe end of a spectrum of cognitive disorders and the level of disability and need for care rises across the spectrum.
It is difficult to set a precise threshold between cognitive impairment and dementia.
The US Preventive Services Task Force (Lin at al 2013) identified the following cognitive impairment screening tests as accurate as their psychometric properties are reported to be greater than 80 to 90 percent sensitivity and specificity:

  • Mini-Mental State Exam
  • Clock drawing test
  • Mini-cog test
  • Memory impairment screen
  • Abbreviated mental test
  • Short Portable Mental Status Questionnaire
  • Free and cued selective reminding test
  • 7-minute screen
  • Telephone Interview for Cognitive Status
  • Information Questionnaire on Cognitive Decline in the Elderly

International population surveys of prevalence of mild cognitive impairment estimates vary from 5% to 37% (Sachdev et al 2015).
Sometimes “mild cognitive impairment”, typically identified using one or more of these screening tests, is counted as “dementia” in population prevalence estimates.
However, only about 60% of people with mild cognitive impairment progress to dementia and experience disability (Savva et al, 2009).
Changing the threshold for ‘dementia’ to Include mild cognitive impairment may create new patients (over diagnosis) and, in clinical settings, result in overtreatment (Figure 1) (Welch et al 2011, Lin et al 2013, Institute of Medicine 2015).
The Gerontology Society of America Workgroup on Cognitive Impairment Detection and Earlier Diagnosis (2015) concluded there is lack of evidence that specific medical conditions or functional limitations are inevitably linked to development of any type of dementia. And, this Workgroup went further, choosing not to endorse any specific risk factors for dementia that would automatically trigger the need for cognitive assessment (The Gerontology Society of America Workgroup on Cognitive Impairment Detection and Earlier Diagnosis 2015).3.0 Measurement of Cognitive Impairment and Dementia.

What is Dementia Prevalence?

Prevalence refers to the number or percentage of people who have dementia.
Dementia incidence is the number or frequency of people who are newly diagnosed with dementia in a given period.
Incident cases are new cases while prevalence counts the total number of existing cases, old and new, for the time period studied.
Prevalence will exceed incidence, reflecting the duration of survival of people with dementia, which has been increasing.
In a steady state, prevalence is equal to incidence multiplied by length of survival. Because the prevalence of dementia rises with age it is desirable to calculate age-standardized prevalence when comparing between populations if their age-structures differ.
Prevalence, incidence and survival figures serve different purposes.
Prevalence summarizes the overall magnitude of the disorder in the population.
Incidence is used to indicate the possible impact of preventive measures.
Survival is used as a marker of success in treatment.

What are the Sources of Information on Dementia in Canada?

Dementia prevalence and monetary costs in the population can be estimated using two sources of information: community health surveys and health administrative databases (or combinations of these) (Figure 2).

Community surveys involve trained interviewers who contact people in the community and ask them and/or their caregivers about their health, and administer simple cognitive tests.
Some respondents may then undergo further medical testing at a study clinic; the additional information collected can supplement the cognitive testing to allow neuropsychologist interpretation of test results.
Such studies provide estimates for dementia prevalence in a population. However, results vary: self-reported survey data relevant to dementia are not likely to be accurate and Canadians living in the community may not be reliable sources of information on neurological conditions, even when proxy responses are included (Loney et al 1998 and Erkinjunnti et al 1997).
Reflecting this, the prevalence estimate for dementia derived from the Canadian Community Health Survey (2010–2011) was much lower than an estimate obtained from a meta-analysis (Grignon et al 2015).
Health administrative data can be a second source.

When people access the health care system, administrative data are created, for example, recording each encounter with a physician and each hospital visit, Clinical point of care assessment data, for example using the InterRAI data collection system in long term care facilities and homecare programs, are collected for administrative purposes but can also be used in research studies (Danila et al 2014).
Computer files within hospitals and for physicians who have electronic medical records are used to store basic information such as reason for the visit, diagnosis and medications prescribed.
These data are used to reimburse providers but also can be used in estimating prevalence, incidence and monetary costs of dementia.

Several service delivery locations generate administrative data (Figure 2) but many such databases do not cover all these locations and the records are often incomplete.
For example, interRAI data are not collected in every Canadian province, limiting comparisons across long term care facilities or home care programs.
In the absence of a unitary system such greater established validity of data generated using interRAI, information on dementia prevalence can be assembled from various sources using clinical algorithms.
For example, persons with a diagnosis of dementia admitted to a hospital, persons for whom dementia was the reason for three visits to a family physician, and/or persons who were prescribed a drug used in the treatment of dementia could each be counted by the algorithm as a “dementia case”.