|Home | About | Journals | Submit | Contact Us | Français|
Purpose:Using the Activity Card Sort (ACS), we derived a measure of lifestyle-adjusted function and examined the distribution of this measure and its correlates in a community sample of older adults at risk for disability transitions.Design and Methods:Participants in the Sources of Independence in the Elderly project (n = 375) completed the ACS using a Q-sort (successive pile sort procedure), in which respondents sorted 39 nonbasic activities of daily living (non-BADL) and noninstrumental activities of daily living (non-IADL) tasks into four categories: “never performed,” “used to but no longer perform,” “hard to do,” and “easy to do.” Lifestyle-adjusted function was defined as the number reported easy/(number easy + number hard + number no longer performed).Results:Respondents reported that they found it easy to perform about 60% of the activities they ever performed. However, people reporting BADL and IADL disability found it easy to perform only 32% of these activities. Lower extremity performance and depressive symptoms were significant independent correlates of lifestyle-adjusted function in models that controlled for sociodemographic and clinical status. The same 2 correlates were significant in models that excluded people with self-reported BADL or IADL disability.Implications:Because this measure eliminates activities never performed, it can be considered “lifestyle adjusted.” Its ability to identify differences in competency among people who do not report BADL or IADL disability suggests that it may be a useful addition to functional assessment.
Identifying disability in older people usually involves reports of “difficulty” or “need for help” in daily activities. These most often include the instrumental activities of daily living (IADL; such as using the telephone, managing money, or cooking and cleaning) and basic activities of daily living (BADL; such as bathing, dressing, or using the toilet). However, Kovar and Lawton (1994) identified a number of limitations in BADL and IADL measures. These include the following:
BADL and IADL measures are strongly related to the risk for nursing home admission, hospitalization, and mortality in samples of more disabled older adults (as shown, e.g., in findings from the National Long-Term Care Survey: Manton, Corder, & Stallard, 1993; Manton & Gu, 2001; Manton, Gu, & Lamb, 2006). However, in less disabled cohorts, correlations between self-reports and health indicators (number of doctor visits, number of nights hospitalized, number of prescriptions, and risk for nursing home admission) are low (Rodgers & Miller, 1997). For examining risk of disability in high-functioning older adults, it may be necessary to expand the range of tasks sampled and develop other indicators of ability.
A natural move in this direction is to assess performance of tasks usually considered more complex than the BADL or IADL. A number of activity checklists have been developed to explore this approach. These include early (Kautzman, 1984; Morgan & Godbey, 1978; Nystrom, 1974) and newer measures (Ashe, Miller, Eng, & Noreau, 2009; Kruger, Ham, & Sankar, 2008) of leisure activity, such as “pursuing a hobby,” “gardening,” and “going out for social events.” Perhaps the best known is the Frenchay Activities Index (Holbrook & Skilbeck, 1983). Self-reported activity is not ideal for measuring physical movement or energy expenditure (for that, accelerometers or actigraphic devices are better suited, as Chipperfield, 2008, suggests), but for assessing diversity of daily activity, loss of tasks over time, or lifestyle preferences, reports of activity are inescapable.
Assessment of activity profiles is complicated by the discretionary nature of many activities. Although nearly every adult can be expected to take out trash or visit friends (and therefore not doing so may indicate disability), the same cannot be said of certain hobbies, physical activities, or other leisure or intellectual pursuits. It is important therefore to adjust reports of activity for differences in lifestyle or personal environment. One promising approach to these challenges in activity assessment is to allow respondents to sort a representative set of activities on a variety of dimensions. This approach is the basis of the Activity Card Sort (ACS; Baum & Edwards, 2001). In this assessment, respondents sort photographs of elders performing up to 80 different activities on dimensions specified by investigators. Advantages of the ACS over other activity checklist measures include its (a) ability to capture both current and abandoned activity; (b) use of photographs rather than verbal questions, which typically emphasize deficits or difficulty; (c) more diverse set of activities; and (d) clinical utility for identifying client-centered goals (Doney & Packer, 2008). The ACS is administered using a Q-sort methodology, in which respondents sort tasks according to different categories or dimensions (see subsequently). This can also be considered an advantage over other measures in that it allows respondents to consider tasks in light of “importance,” “goals,” or other dimensions immediately meaningful to them. It differs from most measures of activity in not being limited to short time frames, such as the National Institutes of Health (NIH) activity record (Gerber & Furst, 1992). Finally, it has been translated and validated for diverse samples, including Israeli (Katz, Karpin, Lak, Furman, & Hartman-Maeir, 2003), Hong Kong Chinese (Chan, Chung, & Packer, 2006), and Australian (Doney & Packer) older adults.
In this research, we sought to use the ACS to (a) derive a measure of non-BADL and non-IADL competencies that adjusts for differences in lifestyle, (b) examine the distribution of this lifestyle-adjusted function measure in a sample of older adults, and (c) examine correlates of lifestyle-adjusted function. The broad goal was to examine variation beyond standard BADL and IADL competencies.
The Sources of Independence in the Elderly (SITE) project was designed to investigate risk factors for disability, as well as factors associated with recovery from disability, in people older than 70 years. We aimed to enroll high-functioning elderly people likely to experience disability transitions. We excluded people who met criteria for dementia or who were living in assisted housing but used a purposive sampling scheme to ensure that a majority of respondents reported at least some difficulty with upper and lower extremity function and few reported both IADL and BADL disability. To identify dementia, all participants completed a neuropsychological evaluation. This 45-min assessment was conducted by Spanish–English bilingual testers and covered the domains of memory, language, and executive function. Results from cognitive assessments were reviewed in a consensus conference, along with other information, to make the diagnosis, which was based on National Institute for Neurological Disease and Stroke-Alzheimer's Disease and Related Disorders Association criteria.
Sampling and recruitment for SITE are described elsewhere (Albert, Bear-Lehman, Burkhardt, Merete-Roa, & Noboa-Lemonier, 2006). Briefly, SITE participants were selected from the larger pool of New York City elders followed in the Washington Heights-Inwood Columbia Aging Project (WHICAP). WHICAP is a population-based survey of Medicare beneficiaries residing in Northern Manhattan, New York City (Luchsinger, Tang, Stern, Shea, & Mayeux, 2001; Luchsinger et al., 2001). In the parent WHICAP study, 2,165 people were enrolled between 1999 and 2002, with roughly equal numbers of White non-Hispanic, African Americans, and Hispanic Americans. People with and without telephones were targeted, and overall 61% of eligible people agreed to participate. Comparison of the WHICAP sample to the 2000 U.S. Census for the ZIP code catchment area suggests no major biases in the distribution of age, gender, or race–ethnicity between the sample and the source population. Eighty-five percent of eligible WHICAP participants agreed to enroll in SITE. SITE participants were enrolled in 2002–2004 and completed follow-up in 2006.
Of the 375 participants enrolled in SITE, 348 completed a second assessment at 12 months and 289 a third assessment at 24 months. Over follow-up, 26 participants (6.9%) were known to have died. Thus, 82% (289/[375 − 26]) of eligible participants completed all three study assessments. In this research, we examine cross-sectional correlates of lifestyle-adjusted function but use vital status over follow-up as an additional correlate, comparing people at baseline according to whether they survived the 2-year follow-up period.
The ACS (Baum & Edwards, 2001) elicits functional status using photographs of older adults performing activities. Each photograph is labeled with the name of an activity. Thus, the photograph labeled “reading Bible” shows an older woman reading the Bible in her home. The people shown are diverse and include men and women of different races and ethnicity. Activities range from such common tasks as “watch TV” to the more rare “playing a musical instrument.” The original instrument contains 80 photographs of older adults performing activities in four broad domains: instrumental (e.g., grocery shopping and washing dishes), low–physical demand leisure (sewing and using a computer), high–physical demand leisure (swimming and gardening), and social (travel and dancing). As the test developers note, “pictures of people actually performing the activity prompt the person to recall the presence of activity in his or her life” (Baum & Edwards, 2001, p. 12).
The ACS elicits the presence or loss of activities, or their ease or difficulty, through a sorting task rather than a direct rating. For example, in a clinical use of the ACS, respondents recovering from stroke identified activities that have been lost due to illness as well as activities identified as goals for therapy (Hartman-Maeir et al., 2007). A scoring technique for this use would be calculation of the percentage of lost activities that were regained. An alternative use of the ACS is to ask respondents to group activities according to common dimensions or importance (Holberg & Finlayson, 2007; Stineman, Kurz, Kelleher, & Kennedy, 2007).
Test–retest reliability of the ACS was 0.90 over a 1-week interval in a community-dwelling sample of older adults (Baum & Edwards, 2001), and the ACS (in checklist form) was validated relative to Medical Outcomes Study: Short-Form 12 measures of physical and mental health (Everard, Lach, Fisher, & Baum, 2000). More recent validation efforts have been reported for an Australian sample. Currently performed activity elicited in the ACS was correlated with a variant of the Frenchay Activities Index (r = .43) as well as a measure of well-being (r = .35; Doney & Packer, 2008).
In our use of the ACS, respondents were asked to sort 39 photographs of activities. From the larger set of photographs in the ACS, we excluded outdoor activities unlikely to be performed in an urban cohort, such as boating and hunting, as well as BADL and IADL tasks. We retained a few ACS tasks that appear in some IADL measures but not in the IADL correlate analyzed subsequently. These include “do laundry,” “managing investments,” and “fixing things around the house.” The 39 activities are listed subsequently.
We administered the ACS using a Q-sort procedure. Respondents were given the full stack of labeled photographs and told to view each picture. They were then asked to divide the stack of photographs into two piles, a pile of activities they currently do not perform and a pile of activities they currently perform. Next, they were told to break up each pile into two separate piles. Respondents broke the pile of activities they reported they do not currently perform into a pile of “never-performed” and a pile of “used-to-but-no-longer-perform” activities. Respondents broke the pile of activities currently performed into a pile of “hard-to-do” and a pile of “easy-to-do” activities. The time frame was “today or the past 30 days.” Respondents were allowed to move photos between piles before their ordering was recorded. The ACS was the first of the tasks in the study research battery, an excellent icebreaker, and all respondents were able to complete the task.
To adjust for lifestyle differences, we computed an index that removes any activities that respondents reported they never performed. Adjusted daily function can then be defined as the proportion of activities respondents report they easily perform relative to activities they ever performed. Thus, lifestyle-adjusted function = number reported easy/(number easy + number hard + number no longer performed). Alternatively, lifestyle-adjusted function = number easy/(39 − number never performed). For example, a respondent who reported that she never performed 9 of the tasks and currently considers 15 of the remaining 30 tasks easy to perform would receive a lifestyle-adjusted function score of 0.50, 15/(39 − 9).
The measure offers a number of advantages. It eliminates “never-did” activities and so adjusts for lifestyle differences; it includes activities no longer performed and so captures loss of activity; and it allows assessment of more complex activity profiles beyond the household and personal self-maintenance activities covered in IADL and BADL measures.
We have not been able to identify prior use of the ACS that adjusts individual activity scores for the never-performed category of activity. Prior studies have typically computed separate indexes for categories of activity, such as the percentage of activities given up or currently performed. But this approach presumes the relevance of the full set of activities when, in fact, most respondents perform a varying subset according to opportunity, ability, or choice. The ACS allows for this adjustment for lifestyle.
We investigated a series of factors likely to be associated with lifestyle-adjusted function. These included sociodemographic indicators (age, gender, education, and race–ethnicity), medical status (whether respondents died over the follow-up period and sum of 13 chronic conditions reported by respondents), self-reported BADL and IADL status, and a series of physical, cognitive, sensory, and mood measures.
Reported medical conditions included myocardial infarction, angina, congestive heart failure, hypertension, diabetes, arthritis, stroke, cancer, hip fracture, chronic obstructive pulmonary disease, Parkinson’s or other neurological conditions, and problems with hearing and vision.
Respondents reporting difficulty in any BADL (bathing, dressing, grooming, using the toilet, and handling utensils) or IADL (using telephone, light shopping, light cleaning, managing medications, and handling money) were considered disabled in the domain. We used the question formats from the Women’s Health and Aging Study I (Guralnik, Fried, Simonsick, Kasper, & Lafferty, 1995).
Lower extremity function was assessed with the Short Physical Performance Battery (SPPB; Guralnik, Ferrucci, Simonsick, Salive, & Wallace, 1995; Guralnik et al., 1994). The SPPB consists of a 4-m walk to assess gait speed, progressively more challenging static balance tests (side-by-side, semitandem, and full-tandem stand), and a chair stand test. Performance is categorized by quartile of performance using population-based norms, and scores across the three domains are summed to yield a score of 0–12, with scores of 10–12 representing best performance. Upper extremity tests included grip strength of the dominant hand using a dynamometer (Jamar; Sammons-Preston, Bolingbroke, IL).
We assessed vibratory sensation using a biothesiometer (Bio-Medical Instrument Co., Newbury, OH). Respondents were told to look away and report when they first felt a “moving or fluttering” sensation as vibration amplitude was gradually increased. We examined sensory thresholds for vibration perception for two anatomical locations on each foot (great toe and instep). Because correlations between the four sites were high, tests were averaged to yield a single measure of foot vibratory threshold (Bear-Lehman, Albert, & Burkhardt, 2006).
To define mild cognitive impairment (MCI), subtests of the memory, language, and executive function assessments were converted to z scores and summed to create separate composites. Assessment of memory included total and delayed recall from the Selective Reminding Test (Bushke & Fuld, 1974) and recognition from the Benton Visual Retention Test (Benton, 1955). Assessment of language included performance on the Boston Naming Test (Kaplan, Goodglass, & Weintraub, 1983) and repetition and comprehension from the Boston Diagnostic Aphasia Examination (Goodglass & Kaplan, 1983). Assessment of executive function included letter and category fluency (Benton & Hamsher, 1976). Composites were then converted to T scores using a regression model to adjust for differences in age, education, and race. Respondents scoring lower than 1.5 SD on any of the composites using these normative corrections were considered to meet criteria for MCI (Manly et al., 2005).
Depressive symptoms were assessed with the nine-item Patient Health Questionnaire (PHQ), a self-report elicitation consistent with Diagnostic and Statistical Manual of Mental Disorders (fourth edition) criteria (Spitzer, Kroenke, & Williams, 1999). Symptom frequency over the prior 2 weeks is scored not at all, several days, more than half the days, and nearly everyday. Responses were summed to create an index of depressive symptoms.
These measures were selected as correlates because all have been used extensively in gerontological research and tap domains likely to be important for the lifestyle-adjusted function measure. As a measure of validity for the correlates, we determined that all were significantly correlated with physical performance, as represented by SPPB, and that physical measures (such as grip strength, r = .38, p < .001 or vibratory sensibility, r = .28, p < .001) were, as expected, more highly correlated than measures of cognitive function (r = .14, p < .02 for executive function). Reliability for self-report scales was also acceptable in the sample. For example, the internal consistency alpha for the PHQ was .74.
We examined descriptive statistics for each photograph in the ACS and computed lifestyle-adjusted function using the formula described earlier. In univariate analyses, we compared subgroups of respondents according to this outcome using an independent samples t test or analysis of variance. We examined lifestyle-adjusted function relative to traditional BADL and IADL measures in similar fashion, nesting people within categories defined by reports of any or no disability in the two measures. We then developed two regression models using lifestyle-adjusted function as an outcome to examine correlates that were independently associated with the measure. One model used the full sample and included IADL and BADL as predictors along with other measures. The second was limited to people reporting no IADL or BADL disability. Regression models used forced entry of all predictors rather than stepwise procedures. Finally, in the sample with no reported BADL or IADL disability, we used an analysis of covariance (ANCOVA) model to examine interactions between measures of physical health and depressive symptoms as correlates of self-reported lifestyle-adjusted function.
Of the 375 respondents, 316 completed cognitive assessments. Accordingly, regression models that included MCI as a correlate were limited to these participants. However, participants with and without cognitive assessments did not significantly differ in sociodemographic status, disability, or activity status. Also, results from regression models that excluded MCI as a correlate were similar to those presented subsequently (results available upon request).
Table 1 shows features of the sample as well as mean lifestyle-adjusted function scores according to categories of each correlate. Respondents were primarily women (69.3%), Hispanic (52.3%), and older than 75 years (71.7%). A minority (29.1%) had education beyond high school. A majority of respondents scored less than 10 (58.1%) on the SPPB. About 14.1% of respondents met criteria for MCI, 26.1% reported an IADL disability, and 10.9% reported a BADL disability.
Mean lifestyle-adjusted function for the sample was 57.7 (SD 15.9), with a range of 7.1–91.4. The median score was 58. In other words, respondents reported that they found it easy to perform about 60% of the activities they ever performed. Of the 39 activities, respondents on average reported that 18 were easy to perform, 2 hard, 11 no longer performed, and 8 never performed.
Reliability of respondent reports was assessed by correlating baseline counts with counts elicited 12 months later within each of the four categories of activity. This approach yields an estimate of test–retest reliability. Because the number of never-performed activities should not change, the correlation between the count of activities in this category across the two time points is a good indication of reliability. The correlation was .74, suggesting adequate reliability.
Table 1 presents measures of validity for the lifestyle-adjusted function measure. Concurrent validity can be seen in the lower mean lifestyle-adjusted function scores for people reporting BADL and IADL disability as well as people with poorer lower extremity performance. For example, people with optimal SPPB performance (scores 10–12, 42% of the sample) had a mean lifestyle-adjusted function of 64.9. People with poorer SPPB performance scored significantly lower, 49.8 (scores 4–6) and 42.4 (scores 0–3). Convergent validity is evident in lower lifestyle-adjusted function for people reporting depressive symptoms, a greater number of medical conditions, older age, and fewer years of school. Discriminant validity is evident in the lack of significant differences between men and women and between race and ethnicity groups. Finally, predictive validity is evident in the association between lower lifestyle-adjusted function and mortality risk for a period of 2 years. People dying over follow-up had lower lifestyle-adjusted function at baseline relative to people who survived the 2-year follow-up period.
Table 2 shows the result of the card sort. Each of the 39 activities is grouped according to the proportion of respondents who reported that they could easily perform the activity. For example, more than 80% reported that they could easily watch TV, take out trash, listen to music, attend family gatherings, watch movies, read magazines or books, or visit with friends. Less than 2% of the sample reported that they never performed these common activities, but about 10% of respondents reported that they no longer watched movies, read magazines or books, or visited friends.
Within the set of 39 activities, the least frequently performed were playing yard games, running, and playing a musical instrument. Less than 10% reported that they performed these activities easily, and most respondents reported that they no longer or never engaged in such activity.
The distribution of activities indicates the primarily urban, immigrant, and pre-digital history of the cohort. Half reported never learning to drive, three quarters never used a computer, and half never managed investments. A quarter or more of the sample also reported that they never swam, ran, took care of a pet, did volunteer work, did puzzles, or collected as a hobby.
An important benefit of the card sort procedure is identifying activities that have dropped out of people’s repertories. As shown in Table 2, by the time of our interviews, half or more of the sample reported they no longer helped with child care, danced, took care of pets, studied, or ran. If we reduce this threshold to 25% or more of the sample, activities no longer performed also include going to children’s activities, attending concerts, playing table games, exercising, collecting as a hobby, going to museums, doing puzzles, performing volunteer work, driving, swimming, making handicrafts, or playing yard games. Thus, a quarter of the sample or more had given up over half the activities surveyed.
As shown in Table 3, mean lifestyle-adjusted daily function was highest in people with no BADL or IADL disability (61.9) and lowest in people reporting both types of disability (31.6). BADL disability and IADL disability were each independent correlates (p = .025 for BADL and p < .001 for IADL), and the interaction between the two was also significant (p = .047). A small proportion (20/375; 5.3%) reported BADL disability in the absence of IADL disability and had relatively high lifestyle-adjusted function scores (56.2). These people mostly reported trouble handling utensils, which evidently did not affect their IADL or activity profiles.
The regression analyses shown in Table 4 indicate that lower extremity performance and depressive symptoms are significant independent correlates of lifestyle-adjusted function in models that control for sociodemographic factors and indicators of medical, cognitive, physical, and self-report disability status. The same two correlates were significant in models that excluded people with self-reported BADL or IADL disability. Women were significantly more likely to report higher function than men but only in the model that included people with BADL or IADL disability. In both models, African American and Hispanic elders were likely to report lower function, but these differences did not achieve significance. MCI, vibratory sensibility, and grip strength were not significant correlates in either model.
To explore the combined effect of lower extremity performance and depressive symptoms on lifestyle-adjusted function, we plotted lifestyle-adjusted function scores by SPPB category and presence or absence of depressive symptoms. We excluded people reporting BADL or IADL disability and used ANCOVA to adjust for age, education, and gender. Separate plots were generated for men and women.
Figure 1 shows that people reporting depressive symptoms scored lower in lifestyle-adjusted daily function at every level of SPPB performance. This pattern held true for men and women.
In this community sample of people older than 70 years, respondents found it easy to perform about 60% of activities they had once performed. This proportion dropped to 32% among respondents who reported BADL and IADL disability, showing that the lifestyle-adjusted activity indicator and traditional measures of function are associated. However, lifestyle-adjusted function was also associated with physical performance and depressive symptoms in people not reporting BADL or IADL disability. Thus, the lifestyle-adjusted function measure may be useful for expanding the range of competencies assessed in gerontological research. In fact, a recent report from the Office of Disability, Aging, and Long-Term Care Policy suggested this kind of expansion in indicators of IADL competency (Waidmann & Freedman, 2006).
Because this measure eliminates activities never performed, it can be considered “lifestyle adjusted.” Because it includes activities no longer performed, it also captures the longer term effect of declining physical or cognitive abilities on activity patterns. The acceptability of the measure to participants, its reliability and validity, and its ability to identify differences among people who do not report BADL or IADL disability all suggest that it may be a useful addition to functional assessment.
We recognize that personal environment, motivation, opportunity, and the perceived value or meaning of activity are all relevant in the choice to perform or withdraw from an activity. Thus, it is impressive to see that indicators of physical performance and mood were significant correlates of lifestyle-adjusted function. Because respondents in the sample on the whole reported very low levels of depressive symptoms, the effect of these symptoms on activity patterns is especially important. In this sample, even mild levels of distress were associated with poorer adjusted function after controlling for physical performance status. Both lower extremity weakness (Espeland et al., 2007) and depressive symptoms (Sriwattanakomen et al., 2008) in older adults can be remediated, and improvements in these areas may lead to more extensive and diverse involvement in activity, one of the key indicators of the “good life” in old age (Lawton, 1999).
The lifestyle-adjusted function measure may be valuable clinically. The ACS is currently used by occupational therapists to identify lost activities, set goals, and assess recovery, as in the setting of stroke rehabilitation (Hartman-Maeir et al., 2007). The index developed here gives clinicians and patients a personalized metric to measure progress. This index may be useful for outcomes assessment in clinical settings as well.
Results from longitudinal cohorts also suggest that participation in diverse activities may be an independent predictor of survival overall and dementia-free survival in particular. In the Established Populations for the Epidemiologic Studies of the Elderly, greater participation in social (going to church, taking trips, and playing cards), physical (sports and walking), and productive (working and volunteering) activities each reduced the likelihood of death for a follow-up period of 13 years in models that adjusted for age, disease, disability, and socioeconomic status (Glass, Mendes de Leon, Marottoli, & Berkman, 1999). In the Bronx Aging Study, greater participation in cognitively stimulating tasks, such as reading, playing board games, and dancing, reduced the risk of dementia for a follow-up period of 5 years (Verghese et al., 2003). These add epidemiological support to social gerontological research showing that loss of activities and roles is associated with poorer well-being and the greater likelihood of physical limitation and depressive symptoms reported here.
The generalizability of results reported here is limited by the primarily urban sample assessed. Mean levels of activity may differ in other communities that offer more outdoor activities or more occasion to drive, greater opportunity to interact with grandchildren or family, or travel. However, it should be noted as well that the race–ethnic groups in the sample did not differ in lifestyle-adjusted function, suggesting that these estimates are stable across diverse groups in a single setting. It may be useful in future research to expand the realm of sampled activities and consider comparing the pile sort results to results obtained using experience-sampling methodologies (Shiffman, Stone, & Hufford, 2008) or time-use diaries (McKenna, Broome, & Liddle, 2007).
Research supported by National Institutes of Health, AG18234.
The authors thank Belkis Merete-Roa and Rafael Noboa-Lemonier and participants in the SITE project.