Findings
Archived Content
Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.
While cognitive decline is not an inevitable consequence of aging, it is more prevalent at older ages.1 In 2006, one in seven Canadians (13.7% of the total population) was aged 65 or older.2 Among these seniors, the percentage aged 80 or older continues to grow, as does the number of centenarians. These trends suggest that a rise in the prevalence of cognitive impairment can be anticipated.
Mild cognitive decline heightens the risk of further deterioration,3-5 but seniors with relatively low levels of impairment may not be identified in cognition studies, which typically focus on people diagnosed with dementia.6 Nonetheless, a substantial share of the senior population is affected. According to the 1991 Canadian Study of Health and Aging (CSHA), about 17% of Canadians aged 65 or older had mild impairment, often labelled "cognitive impairment—no dementia or CIND."7 Similarly, data from the Health and Retirement Survey indicate that 22% of Americans aged 71 or older had CIND.6 Consequently, examination of the prevalence of various levels of cognitive well-being is warranted.
The last national survey to include measures of cognitive functioning among seniors was the CSHA.7 The present analysis uses data from the Cognition Module of the 2009 Canadian Community Health Survey (CCHS)—Healthy Aging to validate a categorization of levels of cognitive functioning in the Canadian household population aged 45 or older. Five categories of four measures of cognitive functioning are examined for the entire sample, and by age group and language.
Methods
Data source
The 2009 Canadian Community Health Survey (CCHS)—Healthy Aging is a population-based, cross-sectional survey. The sampling frame consisted of people aged 45 or older living in private dwellings in the ten provinces. The survey excluded residents of the three territories, some remote regions, institutions, Indian reserves or Crown lands, and military bases (military and civilian), and full-time members of the Canadian Forces. Data collection took place from December 1, 2008 through November 30, 2009 using Computer-Assisted Personal Interviewing.
The purpose of the Cognition Module of the survey was to examine cognitive functioning (as opposed to cognitive impairment) across the lifespan. The Module was administered in English and French to non-proxy respondents who consented to participate. This differed from the main component of the survey, for which proxy responses were accepted if the mental or physical health of selected participants prevented them from completing the interview (2.2% of the sample). Preliminary analyses suggested that respondents interviewed by proxy were more likely than those who answered on their own behalf to have dementia or to have suffered a stroke. Exclusion of these respondents from the Cognition Module means that the data may slightly overestimate cognitive functioning in the household population.
The overall response rate to the Cognition Module was 62.3% (N = 25,864), compared with 74.4% (N = 30,865) for the entire CCHS—Healthy Aging.
Cognition Module variables
Previous studies have used clinical and non-clinical means to assess cognitive functioning. The Mini-Mental State Examination (MMSE)8 and the Modified Mini-Mental State Examination (3MS)9 are the instruments most commonly employed in clinical settings.10-12 However, when clinical assessment is not possible (in large, survey-based studies), other measures must be used.
Cognition may be defined in terms of domains, including memory and executive functioning (for example, planning, problem-solving, and anticipation of outcomes).13 The 2009 CCHS—Healthy Aging Cognition Module includes four cognitive tasks: two relating to memory (immediate and delayed recall) and two relating to executive functioning (animal-naming and the Mental Alternation Test). These tasks are similar to those used in other population-based surveys,14 as well as in community-based studies.15-18
Recall tasks
A modified version of the Rey Auditory Verbal Learning Test (RAVLT) was administered to CCHS—Healthy Aging respondents. The test involves memorizing 15 common unrelated words (for example, drum, curtain, bell) and performing two recall trials: one immediate and one delayed. The delayed recall trial took place five minutes after the immediate recall trial (the other cognitive tasks were performed between the recalls). Survey-administered tests of immediate and delayed recall have been shown to be related to each other in a consistent way, to have similar consistency across racial groups,19 and to have good construct validity.20
Animal-naming
To test semantic fluency, respondents were given one minute in which to name as many items as possible from a category, in this case, animals. Different types of the same species were counted (for example, robin and parrot counted for two points), but different varieties of the same type (for example, American robin and European robin) received only one point. The animal-naming test has been widely administered, demonstrated to be appropriate for evaluating different populations, and sensitive to different types of brain abnormalities, and it correlates with other tests of verbal fluency. 21
Mental Alternation Test
The Mental Alternation Test (MAT) assesses processing speed.17,22 Respondents are asked recite the alphabet, and then to count from 1 to 26. They then have 30 seconds in which to alternate between numbers and letters in the sequence 1-A, 2-B, 3-C, etc. The maximum possible score is 51.
Of those who completed any part of the CCHS—Healthy Aging Cognition Module, 85.9% responded to the immediate recall, 75.5% to the delayed recall, 92.6% to animal-naming, and 90.9% to the Mental Alternation Test. Existing French versions of the recall and animal-naming instruments were used for interviews conducted in French; the English version of the MAT was translated into French.
The various measures reflect independent markers of cognitive functioning, and may have different associations with health outcomes. For example, memory impairment may be important for the early detection of dementia,23 and declines in executive functioning, as well as memory, may influence activities of daily living.24 It is also possible that subgroups respond differently to the various measures of cognitive functioning. For instance, people aged 45 to 64 may not demonstrate the same patterns of cognitive functioning as seniors, and patterns may vary by language group.
Socio-demographic characteristics
Cognitive functioning is typically evaluated in terms of age, sex and education, factors known to be related to cognitive performance.6,7,10-12,25-28 For instance, results from the English Longitudinal Study of Aging revealed better cognitive performance among younger people, women and individuals with higher education.14
Respondents to the CCHS—Healthy Aging reported their sex, age in years and highest level of education. Ten education levels were specified. The language of the interview (English or French) was recorded by the interviewer; respondents who did not complete the CCHS—Healthy Aging in either English or French were excluded from the Cognition Module.
Analysis variables
Numerous physical and psychological correlates of impaired cognitive functioning have been identified (Table 1). Cognitive difficulties have been associated with lower self-rated health,29,30 depression,31-33 loneliness,34,35 decreased life satisfaction,36,37 and reduced ability to perform instrumental activities of daily living.24,38-40 People with cerebrovascular disease,29 diabetes,33,41-44 hypertension,45 or stroke12,46 are more likely to be cognitively impaired than are individuals without these conditions. Psychiatric disorders have also been associated with poor cognitive functioning.47,48
Table 1 Selected characteristics of 2009 Canadian Community Health Survey— Healthy Aging Cognition Module respondents, household population aged 45 or older, Canada excluding territoriesSelf-perceived health
CCHS―Healthy Aging respondents were asked about their general and mental health: "In general, would you say your [mental] health is. . . ." The response options―"excellent," "very good," "good," "fair," and "poor"―were dichotomized to reflect good (excellent/very good/good) versus poor (fair/poor) health.
Activities of daily living
Questions about respondents' ability to perform activities of daily living (ADL) were based on the OARS Multidimential Assessment Questionnaire.49 An overall summary measure of ratings on the ADL capacity-instrumental and physical dimensions was derived. A score of 0 indicates no functional impairment; 1 = mild impairment; 2 = moderate impairment; 3 = severe impairment; and 4 = total impairment. Responses were dichotomized to identify respondents with no impairment versus mild/moderate/severe/total impairment.
Life satisfaction
On a scale from 0 to 10, with 0 representing very dissatisfied and 10, very satisfied, respondents were asked: "How do you feel about your life as a whole right now?" Scores were dichotomized to identify those whose life satisfaction was low (at least 1 standard deviation below the mean) versus not low.
D epression
The CCHS—Healthy Aging measure of depression is a shortened version of the World Health Organization Composite International Diagnostic Interview (CIDI) Scale, which is based on the DSM-III-R and the Diagnostic Criteria for the Research of the ICD-10. The depression subscale pertains to people who felt depressed or lost interest in things for two or more weeks in the past year. For the CCHS—Healthy Aging, a derived variable was created based on the depression score, indicating the probability that respondents would have been diagnosed as having experienced a major depressive episode in the past 12 months if they had completed the Long-Form CIDI. A probability of 0 was assigned to respondents who replied negatively to the stem question (did not have depression for two or more weeks in the past year); a cut-off value of 0.9 was used to distinguish those with a high probability of depression (above 0.9) from those with a lower probability.
Loneliness
The 3-Item Loneliness Scale50 measures an individual's reported loneliness. On a 3-point Likert scale ("hardly ever," "some of the time," and "often"), CCHS―Healthy Aging respondents answered the questions: "How often do you feel:
- that you lack of companionship?"
- left out?"
- isolated from others?"
Higher scores indicate greater loneliness. Scores were dichotomized to identify those with high loneliness (at least 1 standard deviation above the mean) versus not high loneliness.
Self-reported cognition
(Health Utilities Index)
The Health Utilities Index (HUI) Mark III assesses functional health status in eight domains: vision, hearing, speech, ambulation, dexterity, emotion, cognition, and pain.51,52 The HUI has been shown to have strong reliability and validity in general,53 as well as for patients with lower cognitive functioning.54
Only the cognition subscale of the HUI was pertinent to the current study. The items of interest were: "How would you describe your usual ability to:
- remember things (able to remember most things; somewhat forgetful; very forgetful; unable to remember anything at all)."
- think and solve day-to-day problems (able to think clearly and solve problems; having a little difficulty; having some difficulty; having a great deal of difficulty; unable to think or solve problems)."
Items were dichotomized as "able to remember most things" versus at least "somewhat forgetful," and "able to think clearly and solve problems" versus having "at least some difficulty."
Chronic conditions
Respondents were asked if they had been diagnosed with specific long-term health conditions. Conditions relevant to the current analysis were grouped into neurological (Alzheimer's Disease or other dementia, Parkinson's Disease, effects of stroke), vascular (high blood pressure, diabetes, heart attack, heart disease), and psychiatric (mood disorder or anxiety) disorders.
Analytical techniques
T-score creation
Selecting cut-points to identify impairment implies that definitive lines demarcate "normal" from "dysfunctional" scores. It is more likely that cognitive functioning operates on a continuum and that several categories are more appropriate as indicators of impairment.4 Consequently, for this analysis, multiple categories of cognitive functioning were identified.55 T-scores that control for age, sex and education can be calculated for each cognitive measure. Using the sample data for the current study, five categories of cognitive functioning were created, representing t-scores of 0 to 34, 35 to 44, 45 to 54, 55 to 64, and 65 or more.
To generate t-scores from the results of each of the four cognitive tasks, raw scores were converted to scaled scores (mean = 10, standard deviation = 3); higher scaled scores indicate better performance.55 Scaled scores were regressed separately for each task on age, sex and education. In this manner, equations were created for each dependent variable (cognitive outcome) in the form:
DV = intercept + b(age) + b(education) + b(gender)
Each respondent's predicted scaled score was generated from this equation (that is, independent of age, sex and education). The respondent's predicted score was subtracted from the actual scaled score to determine the residual, indicating how well the individual performed, compared with what would be expected based on his/her age, sex and education. Finally, residual scores were converted to t-scores with the following equation:
T = {[(residual/standard deviation of the residual) x 10] + 50}
Thus, the t-scores are independent of age, sex and education; are normally distributed; have a mean of 50 and a standard deviation of 10; and are independent of a unit of measurement.55
Validation
Once t-scores were created and individuals were assigned to one of the five cognitive function categories, the first step in empirically validating the categories was to examine cross-tabulations of the categories by health outcome.
Stratum-specific likelihood ratios (SSLRs) were calculated to determine the accuracy of assigning individuals to levels of cognitive functioning based on the health outcomes.56-58 SSLRs are generalizable and independent of actual probabilities in the population.59 The likelihood that people in each cognitive functioning category (stratum) will experience a certain outcome (for example, fair/poor self-rated health) is given relative to their likelihood of experiencing a positive outcome (in this example, excellent/very good/good self-rated health), according to the formula:
SSLR = (x1g/n1)/(x0g/n0)
where x1g is the number of people with the health outcome (fair/poor health) in the gth stratum; n1 is the total number of people with the health outcome; x0g is the number of people without the health outcome (in good health) in the gth stratum; and n0 is the total number of people without the health outcome.
An SSLR of 10 or more indicates that the health outcome is highly likely; an SSLR below 0.1 indicates that it is highly unlikely.56,57 It is anticipated that SSLRs would be high when a poor health outcome is more likely, and low when a poor health outcome is unlikely.
The final step was to examine all relevant health variables as predictors of cognitive functioning, comparing lower levels of functioning to the highest category (t-scores of 65 or more) for each cognitive outcome. Because the dependent variable (cognitive functioning category) comprised five levels, a multinomial logit regression analysis was used. The odds of reporting a health problem (for example, fair/poor health) should be greatest for those in the lowest (versus the highest) cognitive functioning category, with odds decreasing for those in progressively higher categories of functioning.
Results are presented only for the immediate recall outcome; results were similar for delayed recall, animal-naming, and the MAT (Appendix Tables A to I). Correlations between the four outcome variables were moderate (immediate recall with delayed recall, r = .69; immediate recall with animal-naming, r = .36, immediate recall with MAT, r = .34; delayed recall with animal-naming, r = .33, delayed recall with MAT, r = .30; animal-naming with MAT, r = .45; all p's ≤ .001). Survey sampling weights were applied to all point-estimates to account for the complex survey design of the CCHS—Healthy Aging.
Because the response rate for the Cognition Module was lower than that for the full CCHS—Healthy Aging, separate sampling weights were created for use with the cognitive outcome variables sample. All analyses were performed with SAS 9.1 and SAS-callable SUDAAN. Standard errors in modelling were computed using a bootstrapping technique.60
Results
Descriptive statistics
As expected,55 the distribution of immediate recall scores across cognitive functioning categories was normal, with the most common category (39% of respondents) being t-scores in the 45-to-54 range (Table 2). Approximately 6% of respondents scored in the lowest category, and 8%, in the highest.
People who reported fair/poor general health were more likely to have relatively low immediate recall scores. About 9% of them had scores in the lowest category, compared with 6% overall. Conversely, 5% of those with fair/poor health had immediate recall scores in the in the highest category, compared with 8% overall. This pattern was even more pronounced for self-reported mental health. Similarly, relatively high percentages of people who reported difficulties with activities of daily living, lower life satisfaction and loneliness had low immediate recall scores. By contrast, no pattern emerged for depression.
The HUI cognitive functioning variables were associated with immediate recall scores. Respondents who reported that they were at least somewhat forgetful and who had at least some difficulties thinking clearly and solving problems were more likely than others to have immediate recall scores in the lowest category and less likely to have scores in the highest category.
The presence of a neurological or psychiatric disorder was related to cognitive functioning. Relatively high percentages of people who reported such conditions had low immediate recall scores. However, this was not the case for people with vascular disorders.
Stratum-specific likelihood ratios
In general, the stratum-specific likelihood ratios (SSLR) supported the cognitive functioning categories: the higher their immediate recall score, the less likely were respondents to have negative health characteristics. (Although the SSLR patterns were generally as anticipated, some differences emerged for delayed recall, animal-naming and MAT– Appendix Tables D to F).
SSLRs for fair/poor self-rated general and mental health, difficulties with activities of daily living, low life satisfaction and loneliness decreased as immediate recall scores rose (Table 3). In general, depression also followed the expected pattern, with the highest SSLR for the lowest immediate recall score category. The two HUI cognition variables also demonstrated the anticipated pattern.
Similarly, the SSLRs for neurological and psychiatric disorders followed the expected pattern in that the likelihood of the conditions was associated with low immediate recall scores; no association was shown for vascular disorders.
Multinomial logistic regression
The final step was to examine the odds of being in a low immediate recall score category given the presence of a negative health characteristic. The highest t-score category was set as the reference group (Table 4). As expected, scoring in the lowest immediate recall category was associated with the highest odds of poor health. For instance, compared with people whose immediate recall scores were in the highest category, those with scores in the lowest category had more than twice the odds of being in fair/poor general health, almost six times the odds of being in fair/poor mental health, and more than twice the odds of having difficulties with activities of daily living. Results were similar for low life satisfaction and loneliness. Not surprisingly, people with the lowest immediate recall scores had almost twice the odds of reporting that they were at least somewhat forgetful, and almost four times the odds of reporting that they had some difficulty thinking clearly and solving problems, compared with people with the highest immediate recall scores. However, no association was shown between depression and immediate recall scores.
People with immediate recall scores in the lowest category had more than three times the odds of reporting a neurological condition and twice the odds of reporting a psychiatric disorder. (Odds ratios were generally similar for the other three measures of cognitive functioning, with the exception of psychiatric disorders and the MAT—Appendix Tables G, H and I).
Subgroup analyses
Validation conducted for people aged 45 to 64 and for those aged 65 or older, as well as for English and French respondents, yielded results similar to those obtained for the entire sample (results available upon request). Whether they were in the younger or older age group, English or French, respondents demonstrated similar patterns between health outcomes and the five categories of cognitive functioning (in both cross-tabulations and SSLR comparisons).
Regardless of their age group, people with low immediate recall scores were more likely to have fair/poor self-rated general and mental health, difficulties with activities of daily living, low life satisfaction, loneliness, less ability to think and solve problems, and neurological disorders, compared with people whose scores placed them in higher immediate recall categories. The only differences between the younger and older cohort were in memory and psychiatric disorders—lower immediate recall scores were not strongly associated with ability to remember things and psychiatric disorders among 45- to 64-year-olds, but they were for seniors.
For English and French respondents, lower immediate recall scores were associated with fair/poor self-rated general and mental health, difficulties with activities of daily living, low life satisfaction, loneliness, lower self-rated cognition, neurological disorders, and psychiatric disorders. Depression and vascular disorders were not associated with immediate recall scores for either language group
Discussion
The results of the current study confirm that categories of cognitive functioning can be described from the CCHS—Healthy Aging Cognition Module. Four tests of cognitive functioning—immediate recall, delayed recall, animal-naming and the Mental Alternation Test—were validated based on literature-supported correlates of cognitive functioning. Lower cognitive functioning (notably, t-scores less than 34) was associated with poorer self-rated general and mental health, difficulties with activities of daily living, lower life satisfaction, and loneliness. As might be expected, self-reported cognitive difficulties (forgetfulness and difficulty thinking clearly and solving problems) were associated with low immediate recall scores. The fact that the strongest correlates of the cognitive functioning categories were self-rated mental health and difficulties thinking clearly and solving problems lends the greatest support to the use of the categories presented in this analysis.
Cognitive functioning was not associated with the probability of depression. However, the literature on this subject is inconsistent. Some studies have found no association between depression and cognition,61 while others have shown a relationship, even accounting for socio-economic factors.33 Beirman et al.31 suggested a non-linear relationship between depression and cognitive decline, with elevated levels of depression (and anxiety) in the early stages of decline, but diminished levels as deterioration progresses. Further research on the association between depression and cognitive functioning is obviously necessary.
While neurological and psychiatric disorders were associated with lower cognitive functioning, no patterns emerged for vascular disorders. Previous work, too, has suggested that heart disease, hypertension and diabetes are not necessarily associated with cognitive decline, especially over and above other risk factors such as low educational attainment.33,62,63
Limitations and future directions
A major strength of the current study is the large, nationally representative sample. However, several limitations should be acknowledged.
Proxy responses were not accepted for the Cognition Module. Other research has shown that individuals for whom proxy responses are provided tended to perform poorly on cognitive measures and were more likely to have dementia.20 Thus, the CCHS―Healthy Aging data may underestimate the prevalence of lower cognitive functioning in the Canadian household population.
The CCHS―Healthy Aging Cognition Module used non-clinical measures of cognitive functioning. A clinical assessment would have allowed a test of sensitivity and specificity of the measures in identifying cognitive decline or dementia. This may explain why relationships were not found between vascular disorders (and/or depression) and the cognition categories. Muller et al. 64 found a significant relation between cardiovascular disease and MMSE scores, but not administered tests.
The longitudinal assessment of cognitive functioning among the population is warranted. Such studies would allow researchers to focus on associations between specific risk factors (or correlates) and cognitive functioning over time. For instance, Wilson et al. 36 found that loneliness was associated with a more rapid cognitive decline in elderly people.
Conclusions
Based on the results of tests of immediate and delayed recall, animal-naming, and the MAT in the CCHS—Healthy Aging Cognition Module, five categories describing low to high cognitive functioning were created. These categories were validated for the household population aged 45 or older overall, and by age group and language.
The aging of Canada's population will likely be accompanied by a growing number of people experiencing cognitive decline. CCHS―Healthy Aging data can contribute an understanding of the prevalence of this condition in the household population. This validation study enhances the analytic value of the information in the Cognition Module.
Acknowledgements
Statistics Canada thanks all participants for their input and advice during the development of the Canadian Community Health Survey―Healthy Aging. The survey content was developed by the Health Statistics Division at Statistics Canada in consultation with Health Canada, the Public Health Agency of Canada, and experts conducting the Canadian Longitudinal Study on Aging (CLSA), a major strategic initiative of the Canadian Institute of Health Research. Consultations included stakeholders from Human Resources and Social Development Canada and provincial and territorial health ministries. The addition of 5,000 respondents aged 45 to 54 was funded by the CLSA.
- Date modified: