Association Between Late-Life Social Activity and Motor Decline in Older Adults

Background  Loss of motor function is a common consequence of aging, but little is known about the factors that predict idiopathic motor decline. Our objective was to test the hypothesis that late-life social activity is related to the rate of change in motor function in old age.

Methods  Longitudinal cohort study with a mean follow-up of 4.9 years with 906 persons without stroke, Parkinson disease, or dementia participating in the Rush Memory and Aging Project. At baseline, participants rated the frequency of their current participation in common social activities from which a summary measure of social activity was derived. The main outcome measure was annual change in a composite measure of global motor function, based on 9 measures of muscle strength and 9 motor performances.

Results  Mean (SD) social activity score at baseline was 2.6 (0.58), with higher scores indicating more frequent participation in social activities. In a generalized estimating equation model, controlling for age, sex, and education, global motor function declined by approximately 0.05 U/y (estimate, 0.016; 95% confidence interval [CI], −0.057 to 0.041 [P = .02]). Each 1-point decrease in social activity was associated with approximately a 33% more rapid rate of decline in motor function (estimate, 0.016; 95% CI, 0.003 to 0.029 [P = .02]). The effect of each 1-point decrease in the social activity score at baseline on the rate of change in global motor function was the same as being approximately 5 years older at baseline (age estimate, −0.003; 95% CI, −0.004 to −0.002 [P<.001]). Furthermore, this amount of motor decline per year was associated with a more than 40% increased risk of death (hazard ratio, 1.44; 95% CI, 1.30 to 1.60) and a 65% increased risk of incident Katz disability (hazard ratio, 1.65; 95% CI, 1.48 to 1.83). The association of social activity with the rate of global motor decline did not vary along demographic lines and was unchanged (estimate, 0.025; 95% CI, 0.005 to 0.045 [P = .01]) after controlling for potential confounders including late-life physical and cognitive activity, disability, global cognition depressive symptoms, body composition, and chronic medical conditions.

Conclusion  Less frequent participation in social activities is associated with a more rapid rate of motor function decline in old age.

Idiopathic decline in motor function is a familiar consequence of aging, with older persons displaying a wide spectrum of loss of motor abilities ranging from mild decreased muscle strength and bulk and reduced speed and dexterity to overt motor impairment with concomitant disability. The motor deficits observed in older persons have been subsumed under several terms including sarcopenia,¹ physical frailty,² and parkinsonian signs³ and are widely known to be related to adverse health outcomes including death,^4,5 disability,^6,7 and dementia.^8,9 Although risk factors for common diseases known to cause motor dysfunction such as stroke are recognized, few risk factors for idiopathic motor decline have been identified.

Studies by our group and others have identified physical activity as a factor associated with the rate of declining motor function in community-dwelling elders.^10-13 However, accumulating evidence suggests that a much broader range of leisure activities including late-life social activity are associated with health benefits such as longevity,¹⁴ risk of dementia, and rate of cognitive decline.^15,16 In animal studies, a broad array of activities including social, physical, and cognitive activities are associated with a slower rate of functional decline.¹⁷ However, we are unaware of studies that have examined the extent to which late-life social activity is related to the rate of decline of motor performances in old age. We used data from more than 900 older participants in the Rush Memory and Aging Project, who underwent annual detailed examinations for up to 11 years,¹⁸ to test the hypothesis that the frequency of participation in late-life social activity is related to the rate of motor function decline.

Participants were recruited from about 40 retirement facilities and subsidized housing facilities, as well as from church groups and social service agencies in northeastern Illinois. All participants signed an informed consent agreeing to annual clinical evaluation. The study was in accordance with the latest version of the Declaration of Helsinki and was approved by the institutional review board at Rush University Medical Center, Chicago, Illinois. The clinical evaluation was uniform and included a medical history, complete neurological examination, and assessment of cognitive and motor function. Follow-up evaluations were performed annually by examiners blinded to previously collected data.¹⁸

At the time of these analyses, 1194 participants had enrolled and completed a baseline evaluation. Eligibility for these analyses required the absence of clinical dementia, stroke, or Parkinson disease at the baseline evaluation and a valid assessment of social, physical, and cognitive activities and motor function assessment at baseline as well as at least 1 follow-up motor function evaluation to assess change. We excluded 71 persons who met criteria for dementia at baseline, 113 persons with stroke alone, 15 persons with Parkinson disease, 1 person with both stroke and Parkinson disease, 41 persons who had completed a baseline evaluation but died before their first follow-up examination or had not been in the study long enough for follow-up evaluation, and 47 persons with incomplete data, leaving 906 participants for these analyses. The project began in 1997, and follow-up data through September of 2008 were analyzed. Because of the rolling admission and mortality, the length of follow-up and number of examinations varied across participants. Of the 906 persons included in these analyses, 195 died (21.5%) during the course of follow-up (mean [SD], 4.5 [2.44] years). There were missing data from 279 of 4747 examinations (5.8%) during the course of follow-up.

Clinical diagnoses were made using a multistep process as previously described.¹⁸ Cognitive function testing including 19 performance tests were summarized into a composite measure of global cognition as previously described.¹⁸ Participants were then evaluated in person by an experienced neurologist or geriatrician who diagnosed dementia, stroke, Parkinson disease, and other common neurologic conditions affecting cognitive or physical function. Criteria for dementia followed the joint working group of the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer Disease and Related Disorders Association.¹⁹ Diagnosis of stroke was made as outlined for the Trial of ORG 10172 in Acute Stroke Treatment (TOAST).²⁰ The diagnosis of Parkinson disease was made according to the clinical criteria recommended by the Core Assessment Program for Intracerebral Transplantation (CAPIT).²¹

Grip and pinch strength were measured bilaterally using the Jamar hydraulic dynamometers (Lafayette Instruments, Lafayette, Indiana). Handheld dynamometry (Lafayette Manual Muscle Test System, Model 01163) was used to assess muscle strength in arm abduction, arm flexion, arm extension, hip flexion, knee extension, plantar flexion, and ankle dorsiflexion bilaterally. Time and number of steps to walk 2.4 meters and turn 360° were measured. Participants were asked to stand on each leg and then on their toes for 10 seconds. We counted the number of steps off line when walking a 2.4-m line in a heel-to-toe manner. We also measured the number of pegs that could be placed in 30 seconds (Purdue Pegboard test) and the rate of index finger tapping for 10 seconds bilaterally (Western Psychological Services, Los Angeles, California). Composite measures have been used effectively in other longitudinal studies of cognitive and motor function.^8,22,23 A composite measure of global motor function was constructed by converting the raw score from each of the 18 motor measures to z scores using the mean (SD) from all participants at baseline (Table 1) and averaging the z scores of all of the motor tests together as previously described.^12,24

We used a previously established composite measure of late-life social activity in these analyses.^25,26 Frequency of participation in social activity was assessed with a previously established scale based on 6 items about activities involving social interaction (1, go to restaurants, sporting events or teletract [off-track betting], or play bingo; 2, go on day trips or overnight trips; 3, do unpaid community or volunteer work; 4, visit relatives or friends houses; 5, participate in groups, such as senior center, Knights of Columbus, Rosary Society, or something similar; and 6, attend church or religious services). Each activity was rated on a 5-point scale, with 1 indicating participation in the activity once a year or less; 2, several times a year; 3, several times a month; 4, several times a week; and 5, every day or almost every day. Responses on each item were averaged to yield the composite measure used in analyses as previously described.²⁶

The patient's sex was recorded at the baseline interview. Age in years was computed from self-reported date of birth, and date of the baseline clinical examination was when the strength measures were first collected. Education (reported highest grade or years of education) was obtained at the time of the baseline cognitive testing. Weight and height were measured and recorded at each visit by a trained technician blinded to previously collected data. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared.

As done in previous studies, physical activity was assessed using questions adapted from the 1985 National Health Interview Survey. Participants were asked if they had engaged in any of the activities within the past 2 weeks (eg, walking for exercise, gardening, or yard work) and, if so, the number of occasions and average minutes per occasion. Minutes spent engaged in each activity were summed and expressed as hours of activity per week.¹²

Frequency of participation in cognitively stimulating activities was quantified with a previously established scale.²⁷ People rated how often they had participated in each of 7 cognitive activities (eg, reading a newspaper) in the past year on a 5-point scale, and the mean score for the 7 activities was used in these analyses.

Disability was assessed at baseline with the 6-item Katz disability scale,²⁸ the 3-item Rosow-Breslau disability scale,²⁹ and 8 items that assessed instrumental activities of daily living (IADL) as adapted from the Duke Older Americans Resources and Services project.³⁰ Depressive symptoms were assessed with a 10-item version of the Center for Epidemiologic Studies Depression scale.³¹ Persons were asked whether they had experienced each of 10 symptoms in the past week, and the score was the number of symptoms reported.³²

As in previous studies, the sum of the number of vascular risk factors (ie, the sum of hypertension, diabetes mellitus, and smoking), and vascular diseases (ie, myocardial infarction, congestive heart failure, and claudication) were used in these analyses.³³ Joint pain was based on participant report.

We examined the bivariate associations of late-life social activity and global motor function with age, education, and other covariates. Then we divided the participants into 2 groups according to high and low frequency of participation in social activity at baseline based on the median value and compared their demographic and covariate measures at baseline. We used generalized estimating equation models³⁴ to assess the relation of social activity with baseline level of global motor function and its annual rate of change. The core model included terms for time in years since baseline as well as terms for social activity at baseline, which was centered at its mean, and a term for its interaction with time since baseline. The term for time indicates the average rate of change in global motor function for a typical participant with a social activity score of 2.6; the term for social activity indicates the average difference in motor function at baseline associated with a 1-point change in social activity score; and the interaction of social activity with time indicates the effect of a 1-point change in social activity score on the annual rate of change in global motor function. To control for the effect of demographic variables, these and all subsequent models included terms for age, sex, and education and their interaction with time. In subsequent models, we added terms for the interactions of age, sex, and education with social activity.

Next, we examined several potential confounders of the association of social activity with motor function. Because of sex differences in level and rate of decline in motor function, we also examined 3-way interactions of sex × social activity × motor function. To determine the clinical significance of the amount of change in global motor function, we constructed Cox proportional hazards models examining adverse health consequences of change in motor function and estimated the hazard ratios associated with a given unit of change. These models controlled for age, sex, education, and baseline global motor function. For these analyses we used ordinary least squares regression to estimate the annual rate of change in global motor function for each person.

Finally, in exploratory analyses, we examined whether individual social activities were associated with rate of global motor decline. Models were examined graphically and analytically, and assumptions were judged to be adequately met. An a priori level of statistical significance was P < .05. Using a mixed-model crude estimate of power, we estimate that a sample size of 900 persons with a follow-up pattern and distribution of social activity similar to that seen would have 80% power to detect a coefficient of 0.0082 for the coefficient measuring the effect of social activity on motor function.³⁵ Programming was performed with SAS version 9.1.3 (SAS Institute Inc, Cary, North Carolina) statistical software.

There were 906 persons in these analyses, with a mean (SD) follow-up of 4.9 (2.21) years (range, 2-11 years). Baseline motor function ranged from −2.1 to 2.1 (mean [SD], −0.02 [0.57]). Global motor function was inversely related to age (r = −0.45; P < .001) and positively associated with education (r = 0.19; P < .001), and men had higher levels of global motor function (mean [SD], 0.25 [0.58]) than women (mean [SD], −0.11 [0.54]) (t₉₀₄= −8.69; P < .001). As expected, global motor function was associated with other activity measures, disability, cognition, depressive symptoms, vascular diseases, and joint pain (Table 2).

Baseline social activity scores were approximately normally distributed (mean [SD], 2.6 [0.58]; skewness, −0.18). Scores ranged from 1.00 to 4.17, with higher values indicating more frequent participation in social activity. Social activity was inversely related to age (r = −0.17; P < .001) and positively associated with education (r = 0.14; P < .001), and women had higher levels of social activity (mean [SD], 2.6 [0.57]) than men (mean [SD], 2.5 [0.61]) (t₉₀₄ = 2.23; P = .03). Social activity was associated with global motor function, activity measures, disability, cognition, and depressive symptoms (Table 2).

Compared with participants who reported high social activity at baseline, those who reported low social activity were older, more likely to be male, and less educated; reported less frequent participation in physical and cognitive activities; reported more disability; had lower cognitive function; and were more likely to have lower BMI and diabetes (Table 3).

We used a generalized estimating equation model to test the hypothesis that more frequent participation in social activity is associated with a slower rate of decline in global motor function. On average, global motor function declined at a rate of approximately 0.05 U/y (Table 4, “time”). Baseline frequency of participation in social activity was associated with both baseline level of global motor function (Table 4, “social activity”) and the rate of change in global motor function (Table 4, “social activity × time”). That is, for each point below the mean social activity score at baseline, the average rate of decline in global motor function was 33% more rapid (Table 4, “time”). Since age was also related to the rate of global motor decline, we can compare the amount of global motor decline associated with increased age with the amount of motor decline associated with social activity. For each additional year of age, global motor function declined an additional 0.003 standard unit (Table 4, “age × time”). In contrast, for each 1-point decrease in social activity, global motor function declined an additional 0.016 standard unit (Table 4, “social activity × time”). Thus, in terms of declining motor function, a 1-point decrease on the social activity scale was equivalent to being approximately 5 years older at baseline.

The association of social activity with decline in motor function did not vary along demographic lines (results not shown). In a sensitivity analysis, we excluded participants who were unable to ambulate at baseline, and the association was unchanged (estimate, 0.017; 95% CI, −0.004 to 0.030 [P < .01]).

To illustrate the findings with a common measure, we used a similar model to examine the relationship between social activity and the rate of change in walking speed. In the average participant, walking speed at baseline was approximately 65 cm/s and declined by approximately 2 cm/s/y. In contrast, gait speed in a person with high social activity (score = 3.3, 90th percentile) declined by approximately 1.5 cm/s/y vs 2.6 cm/s/y for a participant with low social activity (score = 1.8, 10th percentile).

We examined a number of covariates that might affect the association of social activity with change in motor function. None of these additional analyses altered the estimate of the association (Table 5). First, we adjusted for cognitive and physical activity (Table 5, model 1). Next, we added terms for baseline disability using the Katz disability, Rosow-Breslau disability, and IADL scales (Table 5, model 2). We then adjusted for baseline global cognition and depressive symptoms (Table 5, models 3 and 4). Next, we examined a number of health-related covariates including body composition, vascular risk factors, vascular disease burden, and joint pain (Table 5, model 5). Finally, all of these covariates were included in a single model, and social activity remained associated with the rate of motor decline (Table 5, model 6).

To determine the clinical significance of the amount of change in global motor function associated with social activity identified in the analyses in the previous subsection, we constructed Cox proportional hazards models examining the association of change in motor function with death and disability and subsequently estimated the hazard ratios associated with a change of 0.16 U/y (ie, the amount of change in global motor function associated with a 1-point decrease on the social activity scale). From these models (data not shown), we calculated that a mean annual change in motor function of 0.16 U/y (Table 4, “social activity × time”) is associated with a more than 40% increased risk of death (hazard ratio, 1.44; 95% confidence interval [CI], 1.30 to 1.60); a 65% increased risk of incident Katz disability (hazard ratio, 1.65; 95% CI, 1.48 to 1.83); and a 34% increased risk of incident Rosow-Breslau disability (hazard ratio, 1.34; 95% CI, 1.18 to 1.52).

In a series of exploratory analyses, we examined the relation of each social activity index to the rate of global motor decline. Of the 6 activities, the following 3 were related to motor decline: unpaid volunteer or community work (estimate, 0.006; 95% CI, 0.001 to 0.021 [P = .03]); visiting friends or relatives (estimate, 0.012; 95% CI, 0.005 to 0.019 [P < .001]), and attending church or religious services (estimate, 0.011; 95% CI, 0.001 to 0.002 [P = .03]).

In a cohort of more than 900 older persons free of dementia, stroke, or Parkinson disease at baseline, we found that a lower frequency of participation in social activity was associated with a more rapid rate of motor function decline. The effect size was equivalent to being approximately 5 years older at baseline, an amount of change associated with more than a 40% increased risk of death and more than a 65% increased risk of developing disability. Moreover, the association of social activity was robust to a wide range of potential confounding variables and remained unchanged after controlling for disability and excluding persons unable to ambulate at baseline, reducing the potential for reverse causality. These findings expand on the accumulating literature showing that participation in a broad spectrum of late-life activities is associated with positive health outcomes in old age and suggest that more frequent participation in social activity may be protective against motor function decline in older persons.

It is widely recognized that increased levels of physical activity are associated with a slower rate of motor function decline and a reduced risk of other adverse health outcomes.^10-13 However, emerging data suggest that physical activity is only 1 component of an active and healthy lifestyle.¹⁶ For example, increased cognitive and social activities in elderly persons are associated with increased survival and a decreased risk of dementia.^36-40 In addition, a number of studies have reported a link between social activity and disability or functional status.^25,41 The present study extends these previous studies by showing that late-life participation in social activity is related to the rate of change in motor function based on objective quantitative measures. Furthermore, the association persisted even after controlling for the frequency of participation in physical and cognitive activities. These findings may be particularly relevant for intervention strategies designed for older adults, for whom participation in physical activities may be constrained because of underlying health problems. These results have important translational implications because they suggest that public health interventions using a broader range of leisure activities might increase the efficacy of efforts to decrease the burden of age-related motor function decline.

The basis for the association between social activity and decline in motor function is uncertain. Emerging evidence suggests that efficient goal-directed movement requires the orchestration and integration of a wide range of sensory, motor, and cognitive functions,^42,43 Human social interaction is complex, and social behavior is generated in the brain through interconnected brain structures that process different elements of sociocognitive and socioaffective information, which are eventually integrated and translated into action.⁴⁴ Thus, both successful social and motor behavior depend on the structural and functional integrity of neural systems that integrate the varied inputs needed for planning and execution of behavior. For example, mirror neurons are thought to play important roles not only for generating movement but also for a wide range of activities essential for social interaction including self-awareness, empathy, and language.^45,46 Recent work with mirror neurons suggest that social and motor behavior may be linked not only at the neural system levels but also at the level of single neurons.^45,46 Moreover, mirror neurons discharge not only when a particular motor act is being performed but also when we observe the same movement being done by others. Although the functional and structural links between social and motor behavior do not explain how higher levels of social activity are related to decline in motor function, it is noteworthy that physical activity in humans is thought to contribute to improved motor function by increasing neuronal plasticity and protecting against ischemic or neurotoxic damage.^47-49 Animal studies suggest that physical activity may be associated with improved function through changes in brain plasticity.¹⁷

Our study has some limitations. Most importantly, inferences regarding causality must be drawn with great caution from observational studies. While the findings were robust to potential confounding variables and sensitivity analyses, the potential for reverse causality cannot be excluded. Furthermore, it is possible that residual confounding from an unmeasured latent variable is related to both social activity and decline in motor function. Other limitations include the selected nature of the cohort and the self-reporting of chronic diseases, in addition to the self-reporting social, physical, and cognitive activities. The combination of diaries and devices that provide quantitative measures of activity such as actigraphy would provide more accurate information about the duration of activity and energy expenditure. Death as informative censoring is also problematic in studies of aging.

However, several factors increase confidence in our findings. Perhaps most importantly, the study had a high follow-up participation, thus reducing bias due to attrition. In addition, social activity was assessed among persons without dementia based on a detailed clinical evaluation, and motor function was evaluated as part of a uniform clinical evaluation and incorporated many widely accepted and reliable strength and motor performance measures. Strength testing was performed in all 4 extremities, and motor performances were tested in both the arms and legs. The aggregation of multiple measures of motor function into a composite measure yielded a more stable measure of motor function and increased the statistical power to identify associations. In addition, a relatively large number of older persons representative of the general population were studied so that there was adequate statistical power to identify the associations of interest, while controlling for several potentially confounding demographic variables.

Decline in motor function is a common condition with adverse health outcomes including death, disability, and the development of other conditions. Thus, it is increasingly being recognized as a major public health problem. Yet little is known about risk factors for motor function decline that could translate into potential public health or clinical interventions. These data raise the possibility that social engagement can slow motor function decline and possibly delay adverse health outcomes from such decline. Further work is needed to ensure that this is a causal relationship. First, the findings will need replication in other cohorts. Second, intervention studies may be needed. In fact, demonstration projects are already under way that may inform on the potential value of interventions. For example, in a novel translational study, a randomized trial of participation in Experience Corps is under way in Baltimore, Maryland.⁵⁰ Participants are randomized to volunteering in elementary schools, which serves as a rich source of cognitive, physical, and social engagement vs being on a wait list. Second, preclinical animal studies potentially could be used to determine whether different types of activities work through a common biological mechanism. Finally, in this study, we excluded clinical stroke and Parkinson disease; however, subclinical manifestations of these or other conditions, in addition to nonneurologic conditions, are responsible for motor function decline. Additional knowledge of the biological, in particular the neurobiological, mechanisms of motor function decline is needed. Such information would allow for much more refined hypotheses regarding the mechanisms underlying the association that will be important for the design and execution of potential interventions.

Correspondence: Aron S. Buchman, MD, Rush Alzheimer's Disease Center, Rush University Medical Center, 600 S Paulina, Ste 1038, Armour Academic Facility, Chicago, IL 60612 (Aron_S_Buchman@rush.edu).

Accepted for Publication: March 7, 2009.

Author Contributions: Dr Buchman had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Buchman, Boyle, Wilson, Fleischman, and Bennett. Acquisition of data: Buchman, Fleischman, and Bennett. Analysis and interpretation of data: Buchman, Boyle, Wilson, Fleischman, Leurgans, and Bennett. Drafting of the manuscript: Buchman and Fleischman. Critical revision of the manuscript for important intellectual content: Buchman, Boyle, Wilson, Fleischman, Leurgans, and Bennett. Statistical analysis: Boyle and Leurgans. Obtained funding: Buchman, Fleischman, and Bennett. Administrative, technical, and material support: Buchman, Wilson, and Bennett. Study supervision: Buchman and Bennett.

Financial Disclosure: None reported.

Funding/Support: This work was supported by National Institute on Aging grants R01AG17917 (Drs Buchman, Wilson, Leurgans, and Bennett), R01AG24480 (Drs Buchman, Leurgans, and Bennett), K23 AG23040 (Dr Boyle), the Illinois Department of Public Health, and the Robert C. Borwell Endowment Fund.

Role of the Sponsors: The organizations funding this study had no role in the design or conduct of the study; in the collection, management, analysis, or interpretation of the data; or in the preparation, review, or approval of the manuscript.

Additional Contributions: We thank all the participants in the Rush Memory and Aging Project. We also thank the staff employed at the Rush Alzheimer's Disease Center, including Traci Colvin, RN, and Tracey Nowakowski, BA, for project coordination; Barbara Eubeler, Mary Futrell, Karen Lowe Graham, MA, and Pamela Smith, MA, for participant recruitment; Wenqing Fan, MS, and Liping Gu, MS, for statistical programming; and John Gibbons, MS, and Greg Klein, BS, for data management.