A prospective study of the association of cognitive and motor function with odds of life space constriction in older adults

Abstract

Background:

Many studies indicate that smaller life space is related to worse cognitive and motor function. It is plausible that cognitive and motor function also predict life space constriction, thus long-term, prospective studies are needed of cognitive and motor function as predictors of life space.

Methods:

A total of 1,246 participants of the Rush Memory and Aging Project, who reported initial maximal life space and at least one follow-up assessment were included in this prospective study, with up to 19 years follow-up. The outcome of interest was the Modified version of the Life Space Questionnaire; which we categorized into large (beyond community), medium (neighborhood/community), and small (home/yard) life space. Participants also had detailed composite measures of global cognition and motor function as predictors and available at the first life space assessment. Life space transitions over one-year periods were modeled using multistate Markov modeling, including confounders and both predictors simultaneously.

Results:

Better cognitive and motor function were broadly associated with lower odds of life space constriction [Cognitive: Large → medium: OR = 0.91, 95% CI 0.83–1.00; Large → small: OR = 0.85, 95% CI 0.74–0.97; Medium → small: OR = 1.01, 95% CI 0.82–1.22. Motor: large → medium: OR = 0.76, 95% CI 0.69–0.83; large → small: OR = 0.58, 95% CI 0.51–0.67; medium → small: OR= 0.71, 95% CI = 0.57–0.87].

Conclusions:

Combined with previous literature that life space predicts function, these results support the notion of complex inter-relations of cognitive function, motor function, and life space.

Introduction

Life space is a measure of the spatial environment an individual moves through within a specified time period, ranging from the room where one sleeps to the community at large^{1, 2}. Navigating the environment is a multi-faceted activity integrating a combination of cognitive and motor abilities¹. Indeed, life space is of increasing interest as an important aspect of health in aging.

However, life space is a complex construct, potentially both a determinant of later health status as well as the result of health-related factors. Nonetheless, research examining key variables in aging, such as cognitive and physical function, as predictors of life space, is largely either cross-sectional or prospective with short-term follow-up^2–11. Importantly, directionality is difficult to establish in cross-sectional or even short-term prospective studies. Thus, long-term prospective studies of cognitive and motor function in relation to life space are critical to exploring temporality and directionality^9–17.

In previous work, our group has reported a cross-sectional association between cognitive and motor abilities with life space¹⁸. Further, we found an association between constricted life space and later cognitive decline, as well as risk for incident mild cognitive impairment (MCI) and Alzheimer’s dementia¹⁹. However, there is also potential for life space to constrict as a result of poor cognitive and motor abilities. Thus, here we extend our prior work by now evaluating the long-term relation of cognitive and motor abilities to subsequent life space constriction. By evaluating the multi-directional associations of cognitive as well as motor function with life space, we can better understand these inter-relations, and potentially better evaluate and apply interventions that may improve health in aging. We hypothesize that higher cognitive ability and higher motor function are both independently associated with lower likelihood of developing life space constriction over time.

Methods

Study Participants

Participants were older community-dwelling individuals from the Rush Memory and Aging Project (MAP). The Rush MAP started in 1997, and includes older men and women from across the Chicago metropolitan area, without known dementia at enrollment, and who agreed to annual clinical assessments and brain donation after death²⁰. Over 2,190 completed a baseline evaluation to date, and the follow-up rate among the survivors exceeds 90%. The study was approved by the Institutional Review Board of Rush University Medical Center. All study participants provided written informed consent.

The life space scale was first included in 2001 which was used as the analytic baseline for these analyses. Therefore, to be included in the study, participants were required to have reported their life space and to have at least one follow-up assessment on their life space after the analytic baseline. In order to define our population at risk and reduce opportunities for reverse causation (i.e. lower life space causing lower cognitive ability and poorer motor function rather than the reverse), we included only participants who reported the largest life space (zone 6) on their first assessment. Participants were also required to have complete cognitive and motor assessments at the analytic baseline. Of a total of 2,142 participants who completed the life space scale, 1,928 participants had at least one follow-up life space assessment, and of those, 1,339 reported their first life space zone as 6; and 1,246 had measures of cognitive ability and motor function at analytic baseline. Thus, 1,246 participants were included in the analyses.

Measurement of global cognitive and motor abilities

Global cognitive score.

Cognitive ability was assessed with a battery of 19 neuropsychological tests, as previously described^{21, 22}. Individual test scores were transformed into z-scores using baseline means and standard deviations of the entire cohort. A composite measure of global cognitive function was constructed by averaging the z-scores across the 19 tests, and a higher score represents better cognitive function. In the composite, there were 7 measures of episodic memory; 3 measures of semantic memory: 3 measures of working memory: 4 measures of perceptual speed; and 2 measures of visuospatial ability.

Global motor score.

A composite measure of 10 motor tasks was constructed by converting the performance score for each motor measure to a score using the mean from all participants at baseline and averaging all the motor tests together, as previously described^{23, 24}. To make the results of these analyses comparable to cognitive function, we converted the performance score for each motor measure to a z-score using the mean and SD from all participants at baseline, and averaging the motor tests together. A higher score represents better motor function. The Jamar hydraulic hand and pinch dynamometers (Lafayette Instruments, Lafayette) were employed to test bilateral grip and pinch strength (tests 1 – 3). Dexterity of the arms was based on the number of pegs placed in the Purdue Pegboard in thirty seconds. Two trials for each hand were averaged to provide a Purdue Pegboard score (test 4). An electronic tapper (Western Psychological Services, Los Angeles, CA) was employed to determine how quickly participants were able to tap with their index fingers for ten seconds. Two trials for each hand were averaged to yield a tapping score (tests 5 – 8). The time and number of steps taken to walk eight feet and turn 360° measured gait speed (test 9). Lastly, participants stood on each leg for ten seconds to assess lower extremity strength and balance. They were then requested to stand on their toes for ten seconds; standing time was recorded (test 10).

The use of composite measures of cognition and motor function in our analyses helped in minimizing floor and ceiling artifacts and other forms of measurement error.

Measurement of life space

Life space was assessed beginning in 2001, and annually thereafter, using a modified version of the Life Space Questionnaire²⁵, which is a self-reported measure of spatial movement through the environment. The original scale²⁵ consisted of 9 questions that address specific life-space zones, each zone further removed from the participant’s room. Participants are asked whether they have been to each concentric zone within the past three days beginning with the room where one sleeps, and then areas immediately outside the home, neighborhood, and the country. Our modified scale has 6 questions that ask participants to indicate whether or not they have been to a particular zone within the environment in the past week. Each zone is a concentric enlargement of life space, beginning with: (i) the room where one sleeps (life space zone 1); to immediately outside the home or apartment building such as the porch, patio, deck, hallway, or garage (life space zone 2); to outside the home or apartment building, such as the driveway, parking lot, yard, courtyard (life space zone 3); to the immediate neighborhood (life space zone 4); to the town or community (life space zone 5); and outside of the town or community (life space zone 6). The response options for each zone were yes (scored as a 1) or no (scored as a 0). The life space score for each person was the sum of responses with possible scores ranging from 0 (not leaving the room where one sleeps) to 6 (outside of town). Higher scores indicate larger life space. Since there were minimal missing data, we carried forward the last valid observation for any missing data on life space.

Other clinical covariates

We considered primary potential confounding factors, including age, sex, and the number of years of formal education²⁶, which were all assessed at cohort enrollment. We also considered other potential confounders, which have previously been reported to be associated with cognitive ability, motor function, and/or life space²⁷, including depressive symptomatology, medical comorbidities, social factors, and purpose in life. Depressive symptoms were assessed with a 10-item version of the Center for Epidemiologic Studies Depression (CES-D) scale²⁸. Medical comorbidities included a count of seven self-reported medical conditions including diabetes, hypertension, heart disease, cancer, thyroid disease, head injury, and stroke²⁹. Perceived social support was assessed using the mean of 4 questions from the Multidimensional Scale of Perceived Social Support³⁰. Perceived social isolation was assessed using five questions that rated the level of agreement with each item (e.g. I feel like I don’t have enough friends) on a five-point scale, with higher scores indicating more loneliness³¹. Purpose in life was assessed using a modified ten-item measure derived from Ryff’s and Keyes’s scales of Psychological Well-Being. On a five-point scale, participants were asked to rate their level of agreement with each item (e.g. I feel good when I think of what I’ve done in the past and what I hope to do in the future). All covariates were identified from the same visit as the initial question on life space.

Statistical analysis

We combined the six life space zones into three categories that captured movement in and out of the house (small: zones 1 – 3), in and out of the neighborhood (medium: zones 4 – 5), and out of town (large: zone 6). A three-state discrete time Markov model was used to assess the transitioning of older adults through different life space zones. A multi-state discrete-time Markov chain is a statistical approach that captures progression of events over multiple states (eg, large to medium to small life space), and estimates associations of risk factors with transitions of interest. The model is based on a one-step transition probability matrix that determines the rate of transitions across the defined states and estimates the probability of moving from one state to the next given the current state. We used a three-state model with State 1 being defined as large life space (zone 6), State 2 as medium life space (zone 4 – 5), and State 3 as small life space (zones 1 – 3). Thus, our models provide odds ratios for the associations of risk factors with the likelihood of transitions between states (eg, from large to medium life space, and large to small life space). Forward transitions represent movement from a larger to a smaller life space. Since life space constriction is not inherently permanent, we also assessed back-transitions; that is, transitions reflecting recovery of life space from a previous constriction (i.e. from a medium to a large, or from a small to a medium or large life space).

We used person-years of follow-up as our metric for tracking participants; this allowed us to account for the number of years that each participant was under observation while allowing participants to contribute follow-up to more than one life space state if they transitioned between states over time. For example, if a participant reported maximal life space at two annual follow-up visits and then reported lowest life space at their third visit, this individual would contribute two person-years to maximal life space and then another person-year to the lowest life space. Incidence rates for constriction and recovery were then calculated by counting the number of incident cases for a given transition between states divided by the number of person-years in the state of origin.

We used two sets of models to investigate the odds of life space transition for every unit increase of baseline cognitive function and motor ability: i) life space constriction, and ii) life space recovery. We implemented models with cognitive ability and motor function included separately, and then we created models with both together. Primary models adjusted for primary confounders of age, sex, and education. We also constructed models adding medical comorbidities, depressive symptomatology, perceived social support, perceived social isolation, and purpose in life as potential confounding factors, based on the literature (Supplementary Tables S1 and S2); however, results were nearly identical to models where we did not include those covariates, thus we focus here on the parsimonious models with primary confounders.

Results

Characteristics of the study sample

A total of 1,246 participants (mean age = 79.1, SD = 7.4; females = 72.7%, non-Hispanic White = 94.2%, mean years of education = 15.3 years, SD = 3.1) who reported the largest life space at study baseline (i.e. outside of the town or community) were included in these analyses (Table 1). The mean for the composite measure of motor function was 1.05 standard units (SD= 0.21; range= 0.52 – 1.66);and the mean for the composite measure of cognitive function was 0.16 (SD = 0.56; range= −2.78 – 1.47).

Table 1.

Baseline descriptive characteristics of participants (N = 1,246).

Baseline Characteristics	n (%)
Female, n (%)	974 (72.7)
White	1,173 (94.2%)
	Mean (SD)
Age at baseline, years	79.1 (7.4)
Years of education	15.3 (3.1)
Global cognitive composite	0.2 (0.6)
Motor function composite	0.1 (0.2)
Number of depressive symptoms	1.0 (1.6)
Number of comorbidities	1.3 (1.1)
Social support	4.4 (0.7)
Social isolation	2.2 (0.7)
Purpose in life	3.8 (0.4)

We had 8,490 person-years of follow-up across all three life space states (Figure 1). Rates of life space constriction were 303.6 cases per 1000 person-years for a large to a medium life space, 57.8 cases per 1000 person-years for a large to a small life space, and 164.4 cases per 1000 person-years for a medium to small life space. The rate of life space recovery from a medium to a large life space was 462.2 cases per 1000 person-years of observation, and from small to medium and small to large the rates were 210.2 cases per 1000 person-years, each. In absolute terms, the number of observations with life space recovery was modest; for example, there were 119 “cases” for recovery from a small to medium life space, in contrast to 1,085 for constriction from large to medium life space.

Figure 1. The multistate transition model of life space constriction and life space recovery.

Note. Large life space (life space = 6) denotes traveling outside of town, medium life space (life space = 5/4) denotes traveling outside and within the neighborhood; small life space (life space ≤ 3) denotes remaining in areas immediately outside or within the home. The arrows depict the number of years for which persons contribute data to a new life space category over the duration of follow-up. Each participant can contribute person-years to multiple life space categories over time, as appropriate. The 1,246 participants included in the study contributed to a total of 8,490 person-years. Blue arrows in panel A show transitions from a large life space to smaller life spaces, i.e., life space constriction (arrows going into orange and red boxes); orange arrows in panel B depict transitions to either larger, i.e., life space recovery (arrows going into the blue box) or smaller (arrows going into the red box) transitions; red arrows in Panel C depict transitions to larger life spaces (blue and orange boxes).

Cognitive ability, motor function, and life space constriction

Overall, both higher cognitive ability and higher motor function were associated with lower odds of transitioning from a large to a smaller life space. As expected, since life space was conceived as a measure of mobility, there was a stronger relation of motor function than cognitive ability to life space constriction (for large to medium life space: cognitive ability, OR = 0.87, 95%CI = 0.79, 0.95; motor function, 0.74, 95%CI = 0.68, 0.81; and to a small life space, cognitive ability, OR = 0.73, 95%CI = 0.64, 0.83; motor function, 0.56, 95%CI = 0.49, 0.64). Higher baseline motor function was also associated with lower odds of further transitioning to even smaller life space (from a medium to a small life space; motor function: OR = 0.71, 95%CI = 0.57, 0.87; cognitive ability: 0.95, 95%CI = 0.78, 1.15) (Supplementary Table S3).

When both cognitive ability and motor function were included in the same model, effect estimates were attenuated for both predictors; however, motor function continued to be more strongly associated with life space than cognitive ability. Specifically, higher motor function was associated with lower odds of transitioning from a larger to a smaller life space (large to medium: OR = 0.76, 95%CI = 0.69, 0.83, large to small: OR = 0.58, 95%CI = 0.51, 0.67; medium to small: 0.71, 95%CI = 0.57, 0.87). Higher cognitive ability was moderately associated with lower odds of transitioning from a large to a medium life space, and from a large to a small life space (large to medium: OR = 0.91, 95%CI =0.83, 1.00; large to small: OR = 0.85, 95%CI = 0.74, 0.97), although the former did not reach the statistical significance. There was no association of cognition with transitioning from a medium to a small life space (OR = 1.01, 95%CI = 0.82, 1.22) (Figure 2).

Figure 2.

Three-state models of the odds of life space constriction over time, according to simultaneous baseline cognitive and motor functions.

* OR=odds ratio; CI=confidence intervals. Both cognitive and motor functions were modeled as continuous variables, in standard units. Models adjusted for baseline age, sex, and education.

Cognitive ability, motor ability, and life space recovery

We also considered life space recovery, although the number of observations for recovery was modest, and odds ratios may not be stable. Higher cognitive function at baseline was associated with higher odds of incremental recovery (from a medium to a large life space: OR = 1.21, 95%CI = 1.04, 1.41 and from a small to a medium life space: OR = 1.30, 95%CI = 1.01, 1.66). Odds of recovery were also slightly higher with higher motor function (from a medium to a large life space: OR = 1.07, 95%CI = 0.92, 1.24; from a small to a large life space: OR = 1.26, 95%CI = 0.99, 1.59; from a small to a medium life space: OR = 1.17, 95%CI = 0.90, 1.52), although results did not reach statistical significance. The associations of higher baseline cognitive function and higher motor function with recovery from a small to a large life space were not statistically significant (cognitive function: OR = 1.23, 95%CI = 0.99, 1.54; motor function: OR = 1.26, 95%CI = 0.99, 1.59) (Supplementary Table S4).

When considering both cognitive ability and motor function in the same model, higher cognitive ability at baseline remained associated with higher odds of life space recovery (from a medium to a large life space: OR = 1.22, 95%CI = 1.04, 1.43 and from a small to a medium life space: OR = 1.32, 95%CI = 1.01, 1.73 as opposed to remaining in the smaller life space), but it was not associated with recovery from a small to a large life space (OR = 1.13, 95% CI = 0.89, 1.44). Higher motor function at baseline was not significantly associated with recovery, although all odds ratios were greater than one (Figure 3).

Figure 3.

Three-state model of the odds of life space recovery over time, according to simultaneous baseline cognitive and motor functions.

Note. OR=odds ratio; CI=confidence intervals. Both cognitive and motor functions were modeled as continuous variables, in standard units. Models adjusted for baseline age, sex, and education.

Discussion

We examined the association of cognitive and motor abilities with subsequent changes in life space in 1,246 well-characterized community-dwelling older adults, considering both life space constriction and recovery. During 8,500 person-years of follow-up, higher cognitive ability and higher motor function at baseline were associated with lower odds of subsequent life space constriction. Our findings here, combined with our previous research¹⁹ are supportive of associations both of: (1) life space constriction as a predictor of subsequent cognitive and motor decline,¹⁹ as well as (2) cognitive and motor abilities as predictors of subsequent life space constriction. That is, associations between life space and cognitive/motor function appeared “bi-directional” since we previously found that life space is associated with cognitive and motor ability, and report here that cognitive and motor ability are associated with life space constriction. This bi-directionality does not appear due to reverse causality since in both studies we only included healthy individuals (i.e., without cognitive impairment in our previous publication, and with maximal life space in the current report). These results have important implications for understanding the complexity of health in aging, including how to test and implement potential interventions involving life space.

Our results extend previous research in our cohort and others by providing evidence that life space is associated with earlier health as well as being a risk factor for subsequent health outcomes^{19, 32}. While there have been some previous studies of cognition and mobility as predictors of life space, these frequently examined basic measures of cognition (such as the MMSE and single cognitive tests⁹) and of motor function (such as step count¹⁰, or walking difficulty¹¹). Thus, our findings here with rigorous measures of cognition and mobility, as well as long-term follow-up, substantially strengthen evidence for the apparent contribution of these factors to changes in life space.

It is plausible that there is a complex interplay between cognitive and motor abilities and life space, with cognitive and motor abilities contributing to subsequent life space status, as well as life space status contributing to subsequent trajectories of cognitive and motor function. Cognitive and motor abilities are health factors that represent multiple mechanisms that can lead to changes in life space (i.e., the ability to navigate one’s environment). These mechanisms may span biologic factors, behavioral factors, as well as environmental factors. In parallel, pathways that are a consequence of constricted life space, such as pathology or engagement in less activities, may then lead to further changes in health factors, including cognitive and motor function.

The significance of life space as both a risk factor for^{18 19}, and a consequence of these same processes, helps inform on the practical consideration of life space in aging adults in several ways. First, if worse cognitive and motor processes are related to subsequent life space constriction, then interventions on life space for the purpose of maintaining health may be less effective if these functions have already declined. Intervention efforts that aim to maintain or increase life space³⁶, may perform best when targeted earlier in the aging process. Intervention might be more successful if factors that also influence life space, such as cognition and motor function, are still intact. Indeed, it is more challenging to reverse impairment (or constriction) once decline has already occurred. Second, our results point to the potential clinical utility of life space – our findings that lower cognitive and motor function are predictors of life space constriction suggest that simple information on life space could potentially serve as part of a simple screen for early cognitive and motor deficits in older persons given the relative ease in identifying changes in life space.

This study has a number of strengths. Participants had up to 19 years of annual follow-up on their assessments of life space, which enabled us to capture long-term relations of cognition and mobility with life space transitions. Follow-up rates were also high, thus the possibility of differential bias was reduced. This study also has some limitations. Results are based on a brief assessment of life space, whereas the “gold standard” Life Space Assessment is substantially more detailed³⁷. Therefore, our outcome may incorporate some misclassification. Similarly, assessment of life space was done annually. In a large, long-term study such as ours, it is extremely challenging to implement more frequent assessments of life space. We might thus have some random misclassification of life space. In a prospective study, random misclassification would lead to underestimating associations of cognition or motor function to life space, and true associations may actually be somewhat stronger than those we observe here. Confounding is also an issue in observational studies. However, when we controlled for an array of potential confounding factors, including health and psychosocial variables these did not meaningfully change the association of motor and cognitive function with life space. Thus, although we cannot rule out unmeasured or residual confounding, it seems unlikely that this could completely explain our results. Lastly, the vast majority of the sample was non-Latino White older adults; since motor and cognitive function both appear to differ across racial and ethnic groups, it will be important to explore associations in diverse samples of older adults.