https://orcid.org/0000-0002-5102-5940
Abstract
Social support predicts functional and cognitive decline in aging. Yet, the associations between social support and gait speed decline—a functional vital sign—are not well understood. This study examined associations between social support and gait speed decline in aging.
Social support and gait data from 542 older adults without dementia were examined (mean age 76.1 ± 6.5 years). Baseline emotional support, tangible support, affectionate support, positive social interactions, and overall support from the Medical Outcomes Study Social Support Survey were the predictors of interest. Annual change in simple (normal pace walking) and complex (walking while reciting alternate letters of the alphabet) gait speed (cm/s) were the outcomes of interest. Linear mixed effects models examined associations between social support and gait speed decline, after adjusting for gender, race, depressive symptoms, overall cognition, and comorbidities.
The mean annual change in gait speed was 1.8 cm/s during simple walking and 1.13 cm/s during complex walking. Tangible support was the only category of social support that predicted decline in simple and complex gait speed over a median follow-up of 3 years. The annual decline in gait speed was 0.51 cm/s (p = .008, 95% confidence intervals [CI] 0.13, 0.89) and 0.58 cm/s (p = .007, CI 0.16, 1.0) greater among those with low tangible support than in those with high tangible support during simple and complex walking, respectively.
Tangible support is a potentially modifiable risk factor for gait speed decline. Further study is needed to examine mechanisms behind the observed associations and the potential for intervention.
Robust Social Support Is Associated With Positive Health Outcomes
There is a large body of research on social support and different aspects of physical and mental health including quality of life, well-being, cognition, physical function, and mortality (1,2). Higher levels of perceived support are associated with reduced risk of institutionalization or nursing home stay following hospitalization and less disability and functional impairment (3,4). Poor social support is associated with frailty, functional impairment, disability, cognitive decline, and a gait-based predementia syndrome known as Motoric Cognitive Risk (MCR) (3–7). Additionally, gait speed and MCR share neural substrates with social support in aging (8–10). Physical and social functions are typically studied individually; however, their shared neural substrates and previously observed associations between social support and physical functions highlight the need for studying the relationship between social support and gait in aging.
Perceived Social Support
Social support can be defined as the perceived availability of support, actual support received, the perception of being cared for, and social integration (11). Support can be structural (eg, marital status, number of social ties) or functional (eg, emotional or informational support, someone to listen) (11). Perceived social support can be further subdivided into 4 different categories: (1) emotional and/or informational support (eg, someone to confide in), (2) tangible support (eg, availability of help when needed), (3) affectionate support (eg, someone to love), and (4) positive social interaction (eg, social companionship) (11).
The impact of social support on health may differ by social support category but few studies have examined this issue—and what category (or categories) of support is important for different health outcomes are not well understood. Better emotional support is associated with better immune and neuroendocrine function, whereas poor tangible support is associated with functional impairment and cortical atrophy (2,3,10). The current study focuses on perceived support rather than actual support received because received support is confounded by need and may not accurately reflect the amount of support received (11).
Poor Gait Performance Is Associated With Adverse Outcomes
Gait speed is a vital sign for functional status and overall health and a robust predictor of prognosis (12,13). Gait speed alone is a sensitive predictor of poor health outcomes compared to physical performance batteries (14–16). Gait speed is also predictive of a range of negative health outcomes including falls, frailty, disability, institutionalization, hospitalization, and mortality (14,17). Gait speed is now known to be an early feature of dementia, especially non-Alzheimer disease dementias (18,19). Gait speed declines faster than cognition and can precede mild cognitive impairment (MCI) and dementia diagnosis by many years (20,21). Thus, gait speed identifies individuals at increased risk for cognitive decline and dementia. Potential modifiable risk factors for gait speed decline, such as social support, present an opportunity for early intervention to prevent functional decline, cognitive decline, and dementia in older adults.
Complex Gait Performance Is Associated With Adverse Outcomes
Walking in the real world can be more complex than laboratory walking because individuals often engage in other activities while walking such as talking to a companion, crossing a street, or navigating busy environments—all of which require divided attention. Limited attentional resources associated with normal aging or cognitive impairment can affect the ability of older adults to perform competing tasks at the same time (22). Dual-task walking tests involve walking while performing a cognitive task (eg, reciting alternate letters of the alphabet, counting backwards from 100, subtracting serial sevens from 100), and are considered mobility stress tests because they necessitate the ability to divide and switch attention between 2 tasks (23). When cognitive demands are introduced during walking, attentional resources must be shared between the cognitive and motor tasks and, as evidenced by decrements in gait performance while performing a Walking While Talking (WWT) test, individuals with fewer cognitive resources have difficulty compensating for the increased cognitive load (24). Dual task cost, or the decrement in gait performance with the addition of a cognitive task during walking while talking, is evident in healthy older adults, young and middle-aged adults, and even children, reflecting the reliance of dual task gait on cognitive function, particularly executive function, and attention (25). In aging, WWT performance is associated with falls, frailty, disability, dementia, and mortality (22,23,26).
Social Support As a Potentially Modifiable Risk Factor for Age-Related Gait Decline
The relationship between social support and different aspects of function including functional impairment, disability, and frailty in older adults is well established (3,4,7). However, the relationship between overall and different categories of social support and gait speed is less clear. Slow gait is part of the definition of frailty and MCR (17,27). Furthermore, increased tangible and overall support decreased the risk of MCR by 30% and 33%, respectively (5). Overall social support and tangible support are also associated with distributed networks of brain areas previously linked to memory, executive function, aging, and dementia (10). Simple and complex walking share partially overlapping brain networks (28). Simple walking is also associated with brain regions associated with executive function and processing speed, whereas dual task walking cost, or the decrement in performance associated with the addition of a cognitive task to simple walking, is associated with brain regions associated with memory (8). Gait speed also shares common demographic, functional, and cognitive risk factors with social support (1,18,29). Given these shared neural substrates, risk factors, and cognitive outcomes of gait speed and social support—along with evidence that social support is associated with reduced risk of functional and cognitive decline—we hypothesized that increased social support would decrease the annual rate of decline in gait speed.
The purpose of this study was to examine the longitudinal relationship between simple and complex gait speed, perceived social support, and social support categories as a function of aging using linear mixed effects models to accommodate repeated observations and permit random effects where gait speed trajectories are allowed to vary by individual participants (30). We hypothesized that gait speed and social support will be associated with age. We further hypothesized that social support is a potentially modifiable risk factor for gait speed decline because less social support and gait speed are both associated with disability and functional decline.
Method
Participants
This prospective cohort study examined gait speed and social support in a sample of 542 adults aged 65 and older enrolled in the Central Control of Mobility in Aging (CCMA) study. The CCMA study is a longitudinal cohort study at the Albert Einstein College of Medicine. The primary goal of CCMA is to determine cognitive and brain predictors of mobility in aging. The study procedure has been previously described (31). Briefly, adults aged 65 and older were identified from population lists in lower Westchester County, NY and invited to participate via mail and telephone. Participants were screened for eligibility over the telephone. Exclusion criteria were inability to speak English, inability to ambulate independently, dementia, significant vision or hearing loss, major psychiatric disorders, recent or anticipated procedures that may affect mobility, and hemodialysis. Eligible participants received comprehensive neuropsychological, cognitive, psychological, and mobility assessments as well as annual follow-up testing. Social support was also assessed annually; however, only baseline scores were utilized for longitudinal analyses in this study because we were examining social support at baseline as a predictor of gait speed decline. The Rush University Medical Center and Albert Einstein College of Medicine Institutional Review Boards approved the study. Informed consent was obtained prior to enrollment.
Social Support
Perceived social support was our predictor variable and was assessed with the Medical Outcomes Study Social Support Survey (MOS-SSS) (11). This measure was used because it measures overall social support as well as 4 different categories of support: (1) emotional/informational support (eg, someone to give you good advice), (2) tangible support (eg, someone to help you if you were confined to a bed), (3) affectionate support (eg, someone who shows you love and affection), and (4) positive social interactions (eg, someone to have a good time with; detailed in Supplementary Table 1). Scores were based on a Likert scale. Response options were none of the time, a little of the time, some of the time, most of the time, or all of the time. A score for overall social support was calculated by adding these 4 subcategories of social support into one composite score. Higher scores indicated higher levels of social support. The MOS-SSS, as well as the 4 support categories generated from this measure, have been previously validated (11,32).
Gait Speed
The primary outcome was gait speed (or gait velocity) in centimeters per second. The validity and reliability of gait speed as a sensitive functional measure have been previously established (12,33). For this study, gait speed (cm/s) was measured during simple walking and complex walking using an 8.5-m-long instrumented walkway with embedded pressure sensors (GAITRite, CIR Systems, Havertown, PA). The GAITRite system is a valid and reliable gait analysis tool and is commonly used for research and clinical purposes (34). Gait speed was assessed using the distance covered in 2 ambulation trials. Assessments were conducted in a quiet, well-lit hallway, and participants wore comfortable footwear. During simple walking, participants completed 1 straight walk, without turning, at their usual pace. During complex walking, participants completed 1 walk while reciting alternate letters of the alphabet starting with the letter “A” or “B,” a test referred to as WWT. Participants were asked to pay equal attention to both the walking and the cognitive task. The WWT test has high test–retest reliability and is easily administered in clinical settings (35). For this study, slow gait speed during simple walking was defined as 1.5 standard deviations below the sample mean by age and gender to describe the sample. For men under age 75, slow gait was defined as 75.3 cm/s or less. For men aged 75 and older, slow gait was a velocity of 63.4 cm/s or less. For women under age 75, slow gait was defined as 70.5 cm/s or less. For women aged 75 and older, slow gait was defined as 67 cm/s or less. Simple and complex gait were assessed as continuous variables in all analyses.
Other Covariates
Global cognitive function was measured using the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS), which measures attention, language, visual-spatial skills, immediate memory, and delayed memory. The RBANS is a validated cognitive assessment tool with established test–retest reliability (36). Lower scores indicated worse cognitive function. Comorbidities were assessed using a Global Health Score (GHS; range 0–10) generated from the self-reported presence or absence of the following conditions: diabetes, heart failure, hypertension, arthritis, depression, stroke, Parkinson’s disease, chronic obstructive pulmonary disease, angina, and myocardial infarction (31). Higher scores indicated the presence of more comorbidities. The Geriatric Depression scale was used to assess for depression (37). Higher scores indicated increased levels of depressive symptoms. The Social Network Index was used to determine the number of high-contact relationships quantified by biweekly interactions with 12 types of social relationship domains (38). Close contacts included spousal relationships, parents, parents-in-law, children, other close family members, neighbors, members of religious or nonreligious groups, work colleagues, school peers, and fellow volunteers.
Data Analysis
Baseline characteristics of participants overall and as a function of slow versus normal gait were summarized in Table 1. We used t tests or Mann–Whitney nonparametric tests for comparisons of continuous variables and chi-square or Fisher’s exact tests for comparisons of categorical variables to examine bivariate associations of normal and slow gait speed and covariates. We applied linear mixed-effects models to examine the longitudinal associations of social support and gait speed. We used separate models for each social support category as they are highly correlated with each other. Age was included as a continuous time variable after graphically assessing the linearity of gait speed decline. Individual participants were included as random effects in the model to account for nonindependence in measures within participants. Because we wanted to measure the influence of social support on rate of change in gait speed over time, we restricted the analyses to baseline social support. Because social support scores were not normally distributed, social support was dichotomized at the median and low social support was used as the reference group. The core models included terms for gait speed, baseline social support and its interaction with time to model the effect of social support on rate of change in gait speed. We also included covariates that could account for or confound the association of gait and social support based on bivariate associations and previous observations that they are associated with social support and gait speed including gender, ethnicity, comorbidities, cognition, and depression (7,39,40). A GHS excluding depression history was computed for use in all regression analyses because depression was assessed by the Geriatric Depression scale and considered as a separate covariate. Depressive symptoms did not significantly contribute to our models but were included based on prior studies and biological relevance. Although marital status was noted as an important predictor of social support in other studies (41), we did not include marital status in our final models because based on likelihood ratio tests, it did not improve the goodness of fit of the models. We used baseline covariate data to be concurrent with social support and to ensure any observed change in gait speed over time was a change in gait speed and not a change in the covariates. We assessed for confounding and effect modification of all covariates, as is our standard practice. Results are reported as parameter estimates with 95% confidence intervals (CI). We conducted multiple sensitivity analyses (summarized in Supplementary Table 2). We ran additional regression models to test whether tangible support was a function of social network size, or the number of high-contact social roles. This was also suggested by prior studies that found larger social networks were associated with less functional decline and activities of daily living and instrumental activity of daily living (ADL/IADL) disability (42). We added ADL and IADL disability as covariates to assess whether results were driven by baseline disability. We also reran the mixed effects models excluding 58 participants who had slow gait at baseline to assess whether results were driven by individuals with accelerated gait decline at baseline. All statistical tests were 2-tailed and p < .05 was considered statistically significant. A false discovery rate (FDR) correction was used as an adjustment for multiple comparisons. Data were inspected descriptively and graphically, and model assumptions were found to be adequately met. Stata version 17.0 (StataCorp LLC, College Station, TX) was used for all analyses.
Baseline Characteristics of Participants Overall and by Gait Speed With Bivariate Associations
| Overall Sample | Slow Gait | Normal Gait | p Value* | |
|---|---|---|---|---|
| n = 543 | n = 58 | n = 485 | ||
| Age (y), mean, SD | 76.1 ± 6.5 | 77.9 ± 6.6 | 75.9 ± 6.5 | .03 |
| Female, n (%) | 326 (55.3) | 39 (67.2) | 262 (54.0) | .05 |
| Ethnicity n (%) | ||||
| Caucasian | 430 (79.2) | 37 (63.8) | 393 (81.0) | .002 |
| Black | 93 (17.1) | 20 (34.5) | 73 (15.1) | |
| Other | 20 (3.7) | 1 (1.7) | 19 (3.9) | |
| Education (y), mean, SD | 14.6 ± 2.9 | 14.1 ± 3.2 | 14.6 ± 2.9 | .2 |
| Marital status n (%) | ||||
| Married | 217 (41.8) | 18 (32.7) | 199 (42.9) | .15 |
| Comorbidities | ||||
| Global Health Score (range 0–5), median, IQR | 1 (1, 2) | 2 (1, 2) | 2 (1, 2) | .04 |
| Cognitive function | ||||
| Total RBANS Score (range 62–137), mean, SD | 91.2 ± 12.1 | 84.8 ± 11.9 | 92 ± 11.9 | <.001 |
| Social factors | ||||
| Social support overall | ||||
| Overall MOS-SSS (range 1.16–5), median, IQR | 4.2 (3.5, 4.8) | 4.0 (3.2, 4.3) | 4.3 (3.6, 4.8) | .008 |
| Social support categories, median, IQR | ||||
| Emotional/informational | 4.1 (3.5, 5) | 3.9 (3.1, 4.6) | 4.1 (3.5, 5) | .06 |
| Tangible | 4.3 (3.3, 5) | 4.1 (3, 5) | 4.3 (3.3, 5) | .16 |
| Affectionate | 4.7 (4, 5) | 4.3 (3.7, 5) | 5 (4, 5) | .03 |
| Positive social interaction | 4.3 (3.3, 5) | 3.7 (3, 4.7) | 4.3 (3.7, 5) | <.001 |
| Social network, mean, SD | 5.2 ± 1.6 | 4.8 ± 1.7 | 5.2 ± 1.6 | .06 |
| Function | ||||
| ADL (0–8), median, IQR | 0 (0, 1) | 2 (0.5, 3) | 0 (0, 1) | <.001 |
| IADL (0–23), median, IQR | 1 (0, 4) | 3 (1, 7) | 1 (0, 4) | <.001 |
| Psychological factors | ||||
| Geriatric Depression Score (range 0–21), median, IQR | 4 (2,7) | 5 (2, 9) | 4 (2, 6) | .03 |
| Overall Sample | Slow Gait | Normal Gait | p Value* | |
|---|---|---|---|---|
| n = 543 | n = 58 | n = 485 | ||
| Age (y), mean, SD | 76.1 ± 6.5 | 77.9 ± 6.6 | 75.9 ± 6.5 | .03 |
| Female, n (%) | 326 (55.3) | 39 (67.2) | 262 (54.0) | .05 |
| Ethnicity n (%) | ||||
| Caucasian | 430 (79.2) | 37 (63.8) | 393 (81.0) | .002 |
| Black | 93 (17.1) | 20 (34.5) | 73 (15.1) | |
| Other | 20 (3.7) | 1 (1.7) | 19 (3.9) | |
| Education (y), mean, SD | 14.6 ± 2.9 | 14.1 ± 3.2 | 14.6 ± 2.9 | .2 |
| Marital status n (%) | ||||
| Married | 217 (41.8) | 18 (32.7) | 199 (42.9) | .15 |
| Comorbidities | ||||
| Global Health Score (range 0–5), median, IQR | 1 (1, 2) | 2 (1, 2) | 2 (1, 2) | .04 |
| Cognitive function | ||||
| Total RBANS Score (range 62–137), mean, SD | 91.2 ± 12.1 | 84.8 ± 11.9 | 92 ± 11.9 | <.001 |
| Social factors | ||||
| Social support overall | ||||
| Overall MOS-SSS (range 1.16–5), median, IQR | 4.2 (3.5, 4.8) | 4.0 (3.2, 4.3) | 4.3 (3.6, 4.8) | .008 |
| Social support categories, median, IQR | ||||
| Emotional/informational | 4.1 (3.5, 5) | 3.9 (3.1, 4.6) | 4.1 (3.5, 5) | .06 |
| Tangible | 4.3 (3.3, 5) | 4.1 (3, 5) | 4.3 (3.3, 5) | .16 |
| Affectionate | 4.7 (4, 5) | 4.3 (3.7, 5) | 5 (4, 5) | .03 |
| Positive social interaction | 4.3 (3.3, 5) | 3.7 (3, 4.7) | 4.3 (3.7, 5) | <.001 |
| Social network, mean, SD | 5.2 ± 1.6 | 4.8 ± 1.7 | 5.2 ± 1.6 | .06 |
| Function | ||||
| ADL (0–8), median, IQR | 0 (0, 1) | 2 (0.5, 3) | 0 (0, 1) | <.001 |
| IADL (0–23), median, IQR | 1 (0, 4) | 3 (1, 7) | 1 (0, 4) | <.001 |
| Psychological factors | ||||
| Geriatric Depression Score (range 0–21), median, IQR | 4 (2,7) | 5 (2, 9) | 4 (2, 6) | .03 |
Notes: ADL = activities of daily living; IADL = instrumental activities of daily living; MOS-SSS = Medical Outcomes Study Social Support Survey; RBANS = Repeatable Battery for the Assessment of Neuropsychological Status; SD = standard deviation; IQR = interquartile range.
*Represents bivariate associations of covariates and slow versus normal gait speed.
Baseline Characteristics of Participants Overall and by Gait Speed With Bivariate Associations
| Overall Sample | Slow Gait | Normal Gait | p Value* | |
|---|---|---|---|---|
| n = 543 | n = 58 | n = 485 | ||
| Age (y), mean, SD | 76.1 ± 6.5 | 77.9 ± 6.6 | 75.9 ± 6.5 | .03 |
| Female, n (%) | 326 (55.3) | 39 (67.2) | 262 (54.0) | .05 |
| Ethnicity n (%) | ||||
| Caucasian | 430 (79.2) | 37 (63.8) | 393 (81.0) | .002 |
| Black | 93 (17.1) | 20 (34.5) | 73 (15.1) | |
| Other | 20 (3.7) | 1 (1.7) | 19 (3.9) | |
| Education (y), mean, SD | 14.6 ± 2.9 | 14.1 ± 3.2 | 14.6 ± 2.9 | .2 |
| Marital status n (%) | ||||
| Married | 217 (41.8) | 18 (32.7) | 199 (42.9) | .15 |
| Comorbidities | ||||
| Global Health Score (range 0–5), median, IQR | 1 (1, 2) | 2 (1, 2) | 2 (1, 2) | .04 |
| Cognitive function | ||||
| Total RBANS Score (range 62–137), mean, SD | 91.2 ± 12.1 | 84.8 ± 11.9 | 92 ± 11.9 | <.001 |
| Social factors | ||||
| Social support overall | ||||
| Overall MOS-SSS (range 1.16–5), median, IQR | 4.2 (3.5, 4.8) | 4.0 (3.2, 4.3) | 4.3 (3.6, 4.8) | .008 |
| Social support categories, median, IQR | ||||
| Emotional/informational | 4.1 (3.5, 5) | 3.9 (3.1, 4.6) | 4.1 (3.5, 5) | .06 |
| Tangible | 4.3 (3.3, 5) | 4.1 (3, 5) | 4.3 (3.3, 5) | .16 |
| Affectionate | 4.7 (4, 5) | 4.3 (3.7, 5) | 5 (4, 5) | .03 |
| Positive social interaction | 4.3 (3.3, 5) | 3.7 (3, 4.7) | 4.3 (3.7, 5) | <.001 |
| Social network, mean, SD | 5.2 ± 1.6 | 4.8 ± 1.7 | 5.2 ± 1.6 | .06 |
| Function | ||||
| ADL (0–8), median, IQR | 0 (0, 1) | 2 (0.5, 3) | 0 (0, 1) | <.001 |
| IADL (0–23), median, IQR | 1 (0, 4) | 3 (1, 7) | 1 (0, 4) | <.001 |
| Psychological factors | ||||
| Geriatric Depression Score (range 0–21), median, IQR | 4 (2,7) | 5 (2, 9) | 4 (2, 6) | .03 |
| Overall Sample | Slow Gait | Normal Gait | p Value* | |
|---|---|---|---|---|
| n = 543 | n = 58 | n = 485 | ||
| Age (y), mean, SD | 76.1 ± 6.5 | 77.9 ± 6.6 | 75.9 ± 6.5 | .03 |
| Female, n (%) | 326 (55.3) | 39 (67.2) | 262 (54.0) | .05 |
| Ethnicity n (%) | ||||
| Caucasian | 430 (79.2) | 37 (63.8) | 393 (81.0) | .002 |
| Black | 93 (17.1) | 20 (34.5) | 73 (15.1) | |
| Other | 20 (3.7) | 1 (1.7) | 19 (3.9) | |
| Education (y), mean, SD | 14.6 ± 2.9 | 14.1 ± 3.2 | 14.6 ± 2.9 | .2 |
| Marital status n (%) | ||||
| Married | 217 (41.8) | 18 (32.7) | 199 (42.9) | .15 |
| Comorbidities | ||||
| Global Health Score (range 0–5), median, IQR | 1 (1, 2) | 2 (1, 2) | 2 (1, 2) | .04 |
| Cognitive function | ||||
| Total RBANS Score (range 62–137), mean, SD | 91.2 ± 12.1 | 84.8 ± 11.9 | 92 ± 11.9 | <.001 |
| Social factors | ||||
| Social support overall | ||||
| Overall MOS-SSS (range 1.16–5), median, IQR | 4.2 (3.5, 4.8) | 4.0 (3.2, 4.3) | 4.3 (3.6, 4.8) | .008 |
| Social support categories, median, IQR | ||||
| Emotional/informational | 4.1 (3.5, 5) | 3.9 (3.1, 4.6) | 4.1 (3.5, 5) | .06 |
| Tangible | 4.3 (3.3, 5) | 4.1 (3, 5) | 4.3 (3.3, 5) | .16 |
| Affectionate | 4.7 (4, 5) | 4.3 (3.7, 5) | 5 (4, 5) | .03 |
| Positive social interaction | 4.3 (3.3, 5) | 3.7 (3, 4.7) | 4.3 (3.7, 5) | <.001 |
| Social network, mean, SD | 5.2 ± 1.6 | 4.8 ± 1.7 | 5.2 ± 1.6 | .06 |
| Function | ||||
| ADL (0–8), median, IQR | 0 (0, 1) | 2 (0.5, 3) | 0 (0, 1) | <.001 |
| IADL (0–23), median, IQR | 1 (0, 4) | 3 (1, 7) | 1 (0, 4) | <.001 |
| Psychological factors | ||||
| Geriatric Depression Score (range 0–21), median, IQR | 4 (2,7) | 5 (2, 9) | 4 (2, 6) | .03 |
Notes: ADL = activities of daily living; IADL = instrumental activities of daily living; MOS-SSS = Medical Outcomes Study Social Support Survey; RBANS = Repeatable Battery for the Assessment of Neuropsychological Status; SD = standard deviation; IQR = interquartile range.
*Represents bivariate associations of covariates and slow versus normal gait speed.
Results
Baseline Characteristics
Data from 542 CCMA participants with up to 7 years of follow-up (median 3 years; interquartile range (IQR) 2, 4) were analyzed. From the initial sample of 591 participants, 49 participants were excluded for missing baseline gait speed or social support measures. Baseline characteristics of the study cohort overall, and as a function of slow gait are displayed in Table 1. The mean age of the overall cohort at baseline was 76.2 ± 6.5; 55.3% were female; 79.2% were Caucasian. Participants were quite educated (mean 14.6 ± 2.96 education years) and healthy (median global health score of 1 [IQR 1, 2]) with intact cognitive function (91.2 ± 12.1 Total RBANS score). The median overall score on the MOS-SSS was 4.2 (IQR 3.5, 4.8). Overall, the cohort was functionally intact, with no ADL/IADL disability. Slow walkers were older, more depressed, had more comorbidities, lower cognitive function, and greater ADL/IADL impairment. Slow walkers also had significantly less overall social support, less affectionate support, and less positive social interaction. Out of 9 possible ADL difficulties, only 16 (3%) individuals reported difficulty with 1 or more ADLs at baseline; 8 reported high tangible support and 8 reported low tangible support.
Longitudinal Association of Baseline Social Support With Gait Speed Over Time
The longitudinal associations between social support and gait speed are displayed in Table 2. As expected, simple and complex gait speeds were significantly associated with age. On average, gait speed declined by 1.8 cm/s per year in the simple walking condition and 1.13 cm/s per year in the complex walking condition, after adjustment for gender, race, depressive symptoms, cognition, and comorbidities. Tangible support predicted both simple and complex gait speed declines over a median follow-up of 3 years. In individuals with high tangible support (characterized by the perceived availability of actual assistance such as someone to help you if you were confined to bed, someone to take you to the doctor, someone to prepare your meals, and someone to help with daily chores), expected gait speed decline was reduced by 0.51 cm/s per year on simple gait (p = .008, CI 0.13, 0.89) and 0.58 cm/s per year on complex gait (p = .007, CI 0.16, 1.0) compared to individuals with low tangible support, after adjustment for gender, race, depressive symptoms, cognition, and comorbidities (Figure 1). In the complex walking condition, overall support and emotional support also reduced expected gait speed decline however, the association between emotional support and complex gait speed decline did not remain significant after adjustment for multiple comparisons. Covariate details are provided in Supplementary Table 3).
Longitudinal Associations of High Social Support and Rate of Change in Gait Speed (N = 542)*
| Simple Gait Speed (cm/s) | Complex Gait Speed (cm/s) | |||||||
|---|---|---|---|---|---|---|---|---|
| b (SE) | 95% CI | p Value | FDR-Adjusted Critical Value | b (SE) | 95% CI | p Value | FDR-Adjusted Critical Value | |
| Gait speed over time | −1.8 (0.1) | −2.0, −1.6 | <.001 | −1.1 (0.1) | −1.3, −0.9 | <.001 | ||
| Overall support × age | 0.3 (0.2) | −0.09, 0.69 | .13 | 0.02 | 0.59 (0.22) | 0.16, 1.02 | .008** | 0.02 |
| Emotional support × age | 0.06 (0.2) | −0.32, 0.45 | .74 | 0.05 | 0.46 (0.22) | 0.03, 0.9 | .034 | 0.03 |
| Tangible support × age | 0.51 (0.2) | 0.13, 0.89 | .008** | 0.01 | 0.58 (0.22) | 0.16, 1.0 | .007** | 0.01 |
| Affectionate support × age | 0.3 (0.2) | −0.1, 0.69 | .14 | 0.03 | 0.15 (0.22) | −0.29, 0.59 | .51 | 0.04 |
| Positive social interaction × age | 0.12 (0.13) | −0.15, 0.38 | .38 | 0.04 | 0.08 (0.16) | −0.23, 0.4 | .62 | 0.05 |
| Simple Gait Speed (cm/s) | Complex Gait Speed (cm/s) | |||||||
|---|---|---|---|---|---|---|---|---|
| b (SE) | 95% CI | p Value | FDR-Adjusted Critical Value | b (SE) | 95% CI | p Value | FDR-Adjusted Critical Value | |
| Gait speed over time | −1.8 (0.1) | −2.0, −1.6 | <.001 | −1.1 (0.1) | −1.3, −0.9 | <.001 | ||
| Overall support × age | 0.3 (0.2) | −0.09, 0.69 | .13 | 0.02 | 0.59 (0.22) | 0.16, 1.02 | .008** | 0.02 |
| Emotional support × age | 0.06 (0.2) | −0.32, 0.45 | .74 | 0.05 | 0.46 (0.22) | 0.03, 0.9 | .034 | 0.03 |
| Tangible support × age | 0.51 (0.2) | 0.13, 0.89 | .008** | 0.01 | 0.58 (0.22) | 0.16, 1.0 | .007** | 0.01 |
| Affectionate support × age | 0.3 (0.2) | −0.1, 0.69 | .14 | 0.03 | 0.15 (0.22) | −0.29, 0.59 | .51 | 0.04 |
| Positive social interaction × age | 0.12 (0.13) | −0.15, 0.38 | .38 | 0.04 | 0.08 (0.16) | −0.23, 0.4 | .62 | 0.05 |
Notes: CI = confidence interval; SE = standard error.
*Adjusted for gender, race, depression, cognition, and comorbidities. Social support dichotomized at the median as high versus low social support with high social support coded as 1.
**Indicates statistically significant value. Simple and complex gait speed were assessed as continuous variables.
Longitudinal Associations of High Social Support and Rate of Change in Gait Speed (N = 542)*
| Simple Gait Speed (cm/s) | Complex Gait Speed (cm/s) | |||||||
|---|---|---|---|---|---|---|---|---|
| b (SE) | 95% CI | p Value | FDR-Adjusted Critical Value | b (SE) | 95% CI | p Value | FDR-Adjusted Critical Value | |
| Gait speed over time | −1.8 (0.1) | −2.0, −1.6 | <.001 | −1.1 (0.1) | −1.3, −0.9 | <.001 | ||
| Overall support × age | 0.3 (0.2) | −0.09, 0.69 | .13 | 0.02 | 0.59 (0.22) | 0.16, 1.02 | .008** | 0.02 |
| Emotional support × age | 0.06 (0.2) | −0.32, 0.45 | .74 | 0.05 | 0.46 (0.22) | 0.03, 0.9 | .034 | 0.03 |
| Tangible support × age | 0.51 (0.2) | 0.13, 0.89 | .008** | 0.01 | 0.58 (0.22) | 0.16, 1.0 | .007** | 0.01 |
| Affectionate support × age | 0.3 (0.2) | −0.1, 0.69 | .14 | 0.03 | 0.15 (0.22) | −0.29, 0.59 | .51 | 0.04 |
| Positive social interaction × age | 0.12 (0.13) | −0.15, 0.38 | .38 | 0.04 | 0.08 (0.16) | −0.23, 0.4 | .62 | 0.05 |
| Simple Gait Speed (cm/s) | Complex Gait Speed (cm/s) | |||||||
|---|---|---|---|---|---|---|---|---|
| b (SE) | 95% CI | p Value | FDR-Adjusted Critical Value | b (SE) | 95% CI | p Value | FDR-Adjusted Critical Value | |
| Gait speed over time | −1.8 (0.1) | −2.0, −1.6 | <.001 | −1.1 (0.1) | −1.3, −0.9 | <.001 | ||
| Overall support × age | 0.3 (0.2) | −0.09, 0.69 | .13 | 0.02 | 0.59 (0.22) | 0.16, 1.02 | .008** | 0.02 |
| Emotional support × age | 0.06 (0.2) | −0.32, 0.45 | .74 | 0.05 | 0.46 (0.22) | 0.03, 0.9 | .034 | 0.03 |
| Tangible support × age | 0.51 (0.2) | 0.13, 0.89 | .008** | 0.01 | 0.58 (0.22) | 0.16, 1.0 | .007** | 0.01 |
| Affectionate support × age | 0.3 (0.2) | −0.1, 0.69 | .14 | 0.03 | 0.15 (0.22) | −0.29, 0.59 | .51 | 0.04 |
| Positive social interaction × age | 0.12 (0.13) | −0.15, 0.38 | .38 | 0.04 | 0.08 (0.16) | −0.23, 0.4 | .62 | 0.05 |
Notes: CI = confidence interval; SE = standard error.
*Adjusted for gender, race, depression, cognition, and comorbidities. Social support dichotomized at the median as high versus low social support with high social support coded as 1.
**Indicates statistically significant value. Simple and complex gait speed were assessed as continuous variables.
Simple and complex gait speed decline as a function of tangible support.
We assessed the interaction between social support and covariates. Using an a priori cutoff of 0.1 to determine if interaction was present, we found depressive symptoms were an effect modifier with overall social support (p = .03) in the simple walking model but we found no significant effect modification based on depression scores in stratified models. See Supplementary Appendix A and Supplementary Table 4 for a detailed discussion of the stratified models. Although prior studies found differential associations of social support by gender (3,6), we assessed for interaction between social support and gender in this sample and found none.
Discussion
The main findings of this study were that (a) both simple and complex gait speed declined with increasing age; (b) high levels of tangible support were associated with a reduced rate of simple and complex gait speed decline; and (c) high levels of overall support were associated with a slower rate of complex gait speed decline. In this study, tangible support was characterized by the availability of actual assistance in terms of someone to help you if you were confined to bed, someone to take you to the doctor, someone to prepare your meals, and someone to help with daily chores. The relationship between tangible support and gait speed decline was not a function of social network size, or the number of high-contact social roles. There were no differences in tangible support between individuals with ADL/IADL disability compared to those without ADL/IADL disability and accounting for ADL/IADL disability in the analyses did not change our results, which suggests that our results were not driven by individuals with ADL/IADL disability who may have required more tangible support.
In this study, high tangible support reduced gait speed decline by 0.5% per year. Slow gait consistently predicts a range of adverse outcomes in community-dwelling older adults, including falls, frailty, disability, institutionalization, hospitalization, and mortality (12,14,17,43,44). Gait is required for independent living in the community and is a marker of overall health, functional status, and healthcare utilization (15). Thus, a reduction of 0.51 cm/s in gait speed decline per year as a function of tangible support is clinically meaningful in terms of improved physical function, reducing disability, increased community ambulation, quality of life, and mortality (15,16). The 0.51 cm/s decline in gait speed predicted by tangible support is over and above the 1.8 cm/s decline in gait speed predicted by age. Thus, tangible support represents one of several risk factors for accelerated gait speed decline in aging and an opportunity for intervention to promote healthy aging.
Like simple gait, complex gait speed decline was associated with baseline tangible support. This finding is consistent with prior work that showed overlapping neural networks and neural substrates between simple and complex gait speed and associations between overall and tangible support with brain regions associated with both, simple and complex gait speed (eg, memory and executive function) (8,10,28). Overall and tangible support decreased the risk for MCR, which is a prodrome of dementia characterized by slow gait and subjective cognitive complaints and clinically detectable prior to the onset of cognitive decline (5,27). In our sample, the relationship between tangible support and complex gait persisted even when adjusting for baseline cognitive performance, which indicates our results are not attributable to slow gait associated with cognitive status and supports the independent association of tangible support and gait speed.
Prior research on social support and physical function and performance is mixed. Few studies used gait speed as their functional outcome. One study found no association between the 6-minute walk test and any of the social support categories, possibly related to a younger study sample (41). Low positive social interaction was associated with poorer physical function, which suggests less physical activity related to less social interaction might play a role in this relationship (41). Less social support was associated with frailty, functional impairment, and disability, and increased tangible support was associated with increased functional decline (3,7,40). In our sample, only tangible support predicted both simple and complex gait speed declines. These mixed results might be related to heterogeneity in functional and social support measures as well as statistical methods. This highlights an important limitation in the current research body on social support. The use of a standardized definition and a standardized assessment tool for social support is crucial for establishing the body of evidence on social support and health outcomes. The current study extends the research base in several important ways. First, our results show that tangible support is associated with the rate of simple and complex gait speed decline and overall support is associated with complex gait speed decline derived from objective gait assessments tested annually for up to 7 years. Second, we addressed a gap in the current research base by using a social support assessment tool that includes 4 distinct and empirically validated dimensions of perceived social support (11) to distinguish between associations of gait speed and different types of social support. Third, the association between gait speed and tangible support was not a function of social network size and fourth, the association between gait speed and tangible support persisted even after controlling for a range of possible confounding variables including gender, ethnicity, comorbidities, cognition, and depression. Taken together, these results suggest that public health interventions to maintain gait speed in older adults must consider the potential role of tangible support as a modifiable risk factor for accelerated gait speed decline, which might increase efficacy of interventions for age-related gait speed decline.
The mechanisms behind the longitudinal association of tangible support and gait speed decline are unclear. A commonly cited theory to explain the health-promoting or health-damaging effects of social support is Cohen’s Stress Buffering Hypothesis (1). This model was borne out of the observation that individuals with greater levels of material and psychological support have better health compared to individuals with fewer resources (1). The stress buffering hypothesis is useful for understanding how social support can intervene between the stressor and the stress reaction by attenuating the stress appraisal as well as the stress experience (1). It is possible that tangible support represents a bare minimum of social support that, when absent, has unique effects on function. It is also possible that individuals with slower gait or poorer physical function are less able to manage taxing or complex tasks in the absence of tangible assistance, which leads to physical inactivity, sedentariness, and further decline in gait and function. This is supported by findings in this sample, where a significantly greater number of individuals with low versus high support (38% vs 33%) reported decreased physical activity. Additionally, social support and gait rely on overlapping neural substrates that support both social, cognitive, and motor processes, both of which might be affected by common pathophysiology. More research is needed to define the mechanisms behind the association of tangible social support and gait speed.
Strengths and Limitations
Strengths of this study include the use of longitudinal data to examine social support as a predictor of gait speed decline. We also used comprehensive and validated assessments of gait and social support, including separate analyses by category of support. We included comprehensive analyses of gait speed and social support with consideration of confounders based on prior research. Our study is not without limitations. As an observational study, we cannot draw conclusions regarding causal inference. Additionally, reverse causality cannot be excluded, despite the persistence of our findings to potential confounding variables and sensitivity analyses, and residual confounding is possible. Our cohort is also healthier, less frail, and well supported, and composed of individuals who were able to come to our center and participate in the study. Additionally, our sample was majority Caucasian and well-educated; thus, results may not be generalizable to the general U.S. population. Future studies are needed to examine the relationship between gait speed and social support in ethnically and socioeconomically diverse samples. We analyzed gait speed as our functional measure and did not evaluate associations with other gait variables (eg, variability or rhythm). We chose velocity as our measure for ease of comparison with existing studies because this is a commonly used quantitative gait variable in the literature. Additionally, dual-task costs affect gait speed, not variability (25). Given the association of complex gait speed with executive function and IADLs, and the association of IADLs with tangible support (3,25), we chose to focus on gait speed only. Different gait variables reflect partially overlapping but also different brain areas, cognitive processes, and health outcomes in both simple and complex gait speeds (26,45–47). Thus, the potentially differential associations of other gait variables and social support are an area for future study. Additionally, although we did not find differences between those with ADL/IADL disability and those without, only 16 individuals reported ADL/IADL disability in this sample and should be further investigated in larger samples including individuals with more varied functional abilities. Further, actual support received, which may influence perceived tangible support, was not measured in this study. Finally, changes in social support over time may influence effects on gait speed and remains an area for future research.
Implications
The identification of low tangible support as a risk factor for gait speed decline is important due to the host of poor health outcomes associated with gait speed decline. Gait speed is known to be an early feature of dementia, especially non-Alzheimer’s disease dementias, and can precede dementia diagnosis by many years (18–20). Targeting modifiable risk factors for cognitive and functional decline is key to prevent morbidity and mortality as well as to support older adults in maintaining independence. Our study suggests tangible support as a potentially modifiable risk factor for functional and cognitive decline in older adults. This information has important translational implications in the clinical and policy arenas in terms of resource investment and allocation and in the design of public health interventions to prevent age-related gait speed decline. Directing resources toward tangible support infrastructure might include investment in home health services to assist with ADLs and IADLs, which are particularly important for maintenance of independent living in older adults (48). Other services that provide tangible support might include meal delivery, transportation to appointments or recreational activities, medication support, and in-home provider services. Taken together with our results, these findings support a role for programs that provide older adults living in the community with tangible support that would allow for continued maintenance of function and independence and prevent accelerated cognitive and gait decline. Although we report associations between perceived tangible support versus actual support received, we hypothesize that perceived support is reflective of actual support received and boosting received support would in turn influence perceptions of support received. Still, actual and perceived support are weakly correlated and increased actual support results in increased perceived support when the support is actually needed (49). An alternative approach to improve perceived tangible support is cognitive reframing or positive mental health interventions to affect individual perceptions of actual support received (50).
Conclusion
Tangible support is a potentially modifiable risk factor for gait speed decline and suggests the need for intervention and policy investment in the provision of tangible support to prevent cognitive and functional decline in older adults.
Funding
This work was supported by NIH/National Center for Advancing Translational Science (NCATS) Einstein-Montefiore CTSA [KL2 TR002558] and NIA/National Institute on Aging [R01AG062659-01A1, R01AG044007-01A1, and R01AG036921].
Conflict of interest
None.
Author Contributions
C.P. conducted all data analyses, wrote the manuscript, and contributed to the critical revisions of the manuscript. H.M.B. conceived of the presented idea, designed the analysis, contributed to the analysis and interpretation of data, writing of the manuscript, and the critical revision of the manuscript. J.V. contributed to the development and design of the article, analysis and interpretation of data, and critical revision of the manuscript. All authors reviewed the final manuscript.
