Contributors to Poor Mobility in Older Adults: Integrating White Matter Hyperintensities and Conditions Affecting Other Systems

Abstract

Background

Age-related mobility limitations are debilitating and common. Cerebral white matter hyperintensities (WMH) and conditions affecting other systems are known contributors, but have been studied in isolation.

Methods

In 2,703 adults aged 65 years or older, we assessed cross-sectional and longitudinal gait speed and mobility disability (self-reported difficulty walking half mile) in those with and without high burden of MRI-defined WMH along with six other conditions (OCs) affecting mobility: gender-specific weak grip; poor self-reported vision; gender-specific lowest quartile of forced vital capacity; self-reported joint pain; ankle-arm index less than 0.9; and body mass index (BMI) greater than 30 kg/m². Separate regression models adjusted for age, gender, and race were repeated for each OC and based on a 4-level predictor: −WMH/−OC; −WMH/+OC; +WMH/−OC; and +WMH/+OC.

Results

Gait speed was fastest in those with −WMH/−OC and slowest for those with +WMH/+OC. Gait speed was similar for either WMH or one of the OC (p range: .07–.9), except for BMI. Those with a high BMI had slower gait speed than those with WMH (p = .01). Declines in gait speed over 6 years were similar for all groups. Results for both prevalent and incident mobility disability showed that associations for WMH and OC were similar for weak grip, poor vision, and low forced vital capacity (p range: .1–.7). Having joint pain, low ankle-arm index, or high BMI was associated with higher prevalent and incident mobility disability compared with having WMH (p range: <.001–.02).

Conclusions

Cerebral WMH should be considered along with conditions affecting mobility from other systems when considering risk and treatment for mobility limitations.

Mobility limitations affect between one quarter and one half of community-dwelling older adults (1,2) and are important public health targets (1). Mobility limitations are associated with poor health, reduced quality of life, and earlier mortality (2,3). However, the success of clinical interventions to maintain or improve mobility has been modest, most likely because the specific causes of age-related mobility limitations are often unclear.

Although the central nervous system is an important contributor to coordination, balance, and mobility control, research and practice have traditionally focused on neurological diseases rather than on subclinical age-related abnormalities. White matter hyperintensities (WMH) observed on magnetic resonance imaging (MRI) are associated with mobility impairment in older adults (4), even without overt neurologic disease. WMH in older adults comprise a mix of inflammatory and vascular pathology and are a marker of cerebral small vessel disease (5). Although progression of WMH may be modifiable (5–7), their contribution is rarely considered along with other treatable contributors to mobility limitations (4). Other contributors may be confounders of the WMH–mobility association. Alternatively, a synergistic interaction may exist such that the effects of WMH are worse in those who also have other contributors.

Our understanding of the underlying mechanisms is complicated by the fact that age-related mobility limitations typically have contributors from multiple domains (8,9). These contributors include weak muscles (10), poor vision (11), low lung function (12), joint pain (13), poor peripheral vasculature (14), and obesity (15), as well as WMH (16). Further, aging affects multiple domains simultaneously; older adults rarely have isolated impairments. However, our methods often do not account for multiple contributors but focus on a single domain, only considering others as confounders. Here, we aim to more carefully describe the contribution of WMH to mobility function and disability by quantifying the role of WMH in combination with other conditions (OCs) on gait speed and mobility disability in a cohort of community-dwelling older adults.

Methods

Study Participants

Participants were from the Cardiovascular Health Study, a community-based study of 5,888 adults aged 65 years or older in four regions of the United States (17). Recruitment from Medicare eligibility lists and age-eligible household members was conducted in 1989/1990 with minority supplement in 1992/1993. Participants were eligible if they had no active cancer, were not wheelchair- or bedbound, and did not plan to move within 3 years (17). Participants received MRI between 1991 and 1994 if they had no exclusions (eg, metal implants and pacemakers) and agreed. The analytic sample included 2,703 participants with complete data for MRI, all OCs, and mobility. Those included in the analytic sample, compared with those excluded, were younger, had faster gait speed, and were less likely to have mobility disability, stroke, obesity, low ankle-arm index (AAI), and weak grip (all p < .001). Gender, race, joint pain, and poor vision were similar for included and excluded participants (all p > .05). Among those with MRI data but missing other data, included had higher WMH compared with excluded participants (p < .001).

Institutional review board approval was obtained from all sites, and all participants provided written informed consent.

Mobility

Mobility was defined as both an objective functional limitation and a subjectively reported disability, following the disablement model (18). Gait speed was assessed over a 15-foot (4.57 m) course at usual speed and reported in meters per second (m/s). Mobility disability was self-reported difficulty in walking half mile (0.80 km). Both were collected at six annual follow-up visits.

White Matter Hyperintensities

Cranial MRI was obtained on 1.5-T (three centers) or 0.35-T (one center) scanners by standard protocol with interpretation at a central reading center. Details are previously published (19). WMH were visually graded from 0 (least) to 9 (most). White matter grade of 3 or more defined presence of WMH (16).

Other Conditions

Six OCs were included that are known mobility contributors and for which data were available. Clinical cutpoints were used where available. Weak grip was defined by gender-specific cutoffs of less than 16 kg for women and less than 26 kg for men (20). Vision problems were self-reported being unable to see sufficiently to drive, watch TV, or recognize someone across a room with or without glasses. Lung function was measured by forced vital capacity (FVC) with low FVC defined as being in the worst gender-specific quartile: women less than 2.01 L and men less than 3.00 L (8). Joint pain was self-reported as any pain in the feet, knees, hips, or back. Peripheral vascular impairments were defined by an AAI less than 0.9 (21). Obesity was a body mass index (BMI) greater than 30 kg/m².

Covariates

Age, gender, and race were self-reported. Height was measured using standard procedures. Smoking status was never, former, or current. Diabetes, coronary heart disease (CHD), hypertension, and stroke were defined or centrally adjudicated as described elsewhere (22). Use of Parkinson’s medications was from a complete medication inventory (23). The Modified Mini-Mental State (3MS) scores less than 80 indicated likely cognitive impairment (24). The Digit Symbol Substitution Test (DSST) measures processing speed with higher scores indicating faster speed (25). A modified version of the Centers for Epidemiologic Studies Depression (CES-D) scale was administered with scores greater than 10 indicating likely depression (26). Recurrent fallers were those with two or more self-reported falls in the past year.

Statistical Analysis

Descriptive comparisons used chi-square statistics and t tests. For each of the six OCs, a four-level variable was created: no WMH or OC, only the OC, only WMH, or both WMH and the OC. Separate regression models (six total, one per condition) assessed the association of this four-level variable with gait speed or mobility disability, both cross-sectionally and longitudinally. All analyses were adjusted for age, gender, and race. Supplemental analyses adjusted for additional covariates described earlier.

Mixed effects models with random intercepts and slopes estimated baseline gait speed and change over 6 years for WMH and OCs after adjustment. Interaction terms for time with gender and age and a time squared term were included. Estimates and 95% confidence intervals (CIs) are presented for white women of mean age of 74 years; results were similar for other demographic groups.

Logistic regression models estimated the proportion with mobility disability at baseline after adjustment. Estimates and 95% CI are presented for white women of mean age of 74 years; results were similar for other demographic groups. Cox proportional hazard models with discrete times assessed hazard ratios of newly reported mobility disability by WMH and OCs after adjustment in those free of mobility disability at baseline (n = 2,234). Death and loss-to-followup were treated as censored non-events.

Interactions were tested between WMH and each of the OCs for both outcomes; no synergistic interactions were present and were not included in final models. Sensitivity analyses excluded those with stroke or taking Parkinson’s medication at baseline. SAS version 9.4 was used.

Results

The sample had a mean age of 74 years (SD = 4.8) and was 56.3% female and 84.7% white (Table 1). WMH were present in 880 (32.6%) participants. At baseline, mean gait speed was 0.93 m/s (SD = 0.22) and 469 (17.4%) participants reported mobility disability. Prevalence of conditions and co-occurrence with WMH are shown in Table 2. Among those with any of the OCs, most had only one (n = 979; 53.2%) or two (n = 601; 32.7%). The mean number of OCs was slightly higher for those with WMH (1.3, SD = 1.1) than for those without (1.1, SD = 1.0; p < .001). All OCs were more common in those with WMH than in those without, except for joint pain and obesity (Table 2). Of those with mobility disability, only 25 (5.3%) had no OCs or WMH.

Table 1.

Characteristics of the Sample (≥65 years) by WMH Status

	Mean (SD) or n (%) in Whole Sample	Mean (SD) or n (%) in Those Without WMH	Mean (SD) or n (%) in Those With WMH	p Value
	n = 2,703	n = 1,823	n = 880
Mean (SD) Age, y	74.4 (4.8)	73.7 (4.5)	75.9 (5.1)	<.001
Female gender	1,521 (56.3)	1,011 (55.5)	510 (58.0)	.2
Black race	413 (15.3)	274 (15.0)	139 (15.8)	.6
Mean (SD) Height, cm	165.5 (9.3)	166.0 (9.2)	164.6 (9.4)	<.001
Recurrent faller	140 (5.2)	84 (4.6)	56 (6.4)	.05
Low 3MS score	589 (11.9)	91 (5.0)	76 (8.6)	<.001
Mean (SD) DSST score	40.5 (13.0)	41.9 (12.7)	37.6 (13.2)	<.001
High depressive symptoms	736 (14.5)	175 (9.6)	99 (11.3)	.2
Smoking status
Never	1,276 (47.3)	866 (47.6)	410 (46.6)	.9
Former	1,136 (42.1)	763 (41.9)	373 (42.4)
Current	288 (10.6)	191 (10.5)	97 (11.0)
Diabetes	248 (9.2)	159 (8.7)	89 (10.1)	.2
CHD	602 (22.3)	374 (20.5)	228 (25.9)	.002
Stroke	114 (4.2)	56 (3.1)	58 (6.6)	<.001
Hypertension	1,055 (39.1)	643 (35.3)	412 (46.9)	<.001
Mean (SD) Gait speed, m/s	0.93 (0.22)	0.95 (0.20)	0.89 (0.20)	<.001
Mean (SD) Per year change gait speed	−0.015 (0.040)	−0.014 (0.047)	−0.019 (0.073)	.003
Mobility disability	469 (17.4)	263 (14.4)	206 (23.4)	<.001
6-Year incident mobility disabilitya	779/2,234 (34.9)	512/1,560 (32.8)	267/674 (39.6)	.002

Note: 3MS = Modified Mini-Mental State Examination; CHD = coronary heart disease; DSST = Digit Symbol Substitution Test; SD = standard deviation; WMH = white matter hyperintensities.

^aOf those free of mobility disability at baseline

Table 2.

Prevalence of Other Conditions at Baseline by WMH Status in Adults Aged 65 Years and Older (n = 2,703)

	Prevalence n (%) in Whole Sample	Prevalence n (%) in Those Without WMH	Prevalence n (%) in Those With WMH	p Value
		n = 1,823	n = 880
Other conditiona:
Weak grip	318 (11.8)	181 (9.9)	137 (15.6)	<.001
Poor vision	404 (15.0)	251 (13.8)	153 (17.4)	.01
Low FVC	607 (22.5)	356 (19.5)	251 (28.5)	<.001
Joint pain	919 (34.0)	620 (34.0)	299 (34.0)	.9
Low AAI	276 (10.2)	163 (8.9)	113 (12.8)	.002
High BMI	502 (18.6)	353 (19.4)	149 (16.9)	.1
No other conditions	864 (32.0)	619 (34.0)	245 (27.8)	.001

Note: AAI = ankle-arm index; BMI = body mass index; FVC = forced vital capacity; WMH = white matter hyperintensities.

^aDefined as—Weak grip: gender-specific low grip strength; Poor vision: poor self-reported vision; Low FVC: gender-specific lowest quartile of FVC; Joint pain: self-reported joint pain; Low AAI: AAI < 0.9; and High BMI: BMI > 30.

At baseline and after adjustment for age, gender, race, and all six OCs, those with WMH compared with those without had significantly slower gait speed (mean difference (95% CI): −0.03 m/s (−0.05, −0.01)) and greater proportion with mobility disability (mean difference (95% CI): 0.06 (0.03, 0.09)). Among those with none of the OCs (n = 864) and after adjustment for age, gender, and race, those with WMH compared with those without had significantly slower gait speed (mean difference (95% CI): −0.06 m/s (−0.08, −0.03)) and greater proportion with mobility disability (mean difference (95% CI): 0.05 (0.01, 0.08)).

Gait Speed

Estimated gait speed by WMH and each of the OCs adjusted for age, gender, and race are shown in Figure 1. For each condition, those without WMH or the OC walked fastest, while those with both walked slowest (Figure 1, Supplementary Table 1). Participants with only WMH or only the OC walked more slowly than those with neither WMH nor the OC (all p ≤ .0001). Except for high BMI, those with only WMH had similar gait speed to those with only the OC (p range: .07–.9; Supplementary Table 1). Those with only a high BMI walked more slowly than those with only WMH (mean difference = 0.03 m/s, p = .01).

Figure 1.

Estimated gait speed (m/s) by white matter hyperintensities (WMH) or other condition only (two middle lines (red and green lines online)), both (bottom line), or neither (top line) over 6 years of follow-up (n = 2,703). Results are for white women of mean age (74 years); relative differences were similar for other groups.

Having WMH was significantly or nearly significantly related to faster slowing of gait speed over 6 years compared with those with neither WMH nor the OC (Supplementary Table 1; p range: .03–.1). Low FVC (p = .04) or low AAI (p = .05) were significantly related to faster slowing while high BMI was marginally related (p = .07). Weak grip, poor vision, and joint pain were not significantly associated with gait slowing (p range: .2–.7).

Analyses adjusted for diabetes, CHD, hypertension, stroke, smoking status, and height were attenuated but qualitatively similar (Supplementary Table 2). Results were unchanged when removing participants with stroke or using Parkinson’s medications at baseline or when adjusting for cognitive status, fall history, and depressive symptoms (data not shown).

Mobility Disability

At baseline, the proportion with mobility disability was lowest in those without WMH or OCs and was highest in those with both (Figure 2). Compared with those with neither WMH nor OCs, the proportion with mobility disability was higher for those with either WMH (all p ≤ .005) or the OC (p ≤ .001), except for poor vision (p = .2) and weak grip (p = 0.3). For low FVC, weak grip, and poor vision, the proportion with mobility disability did not differ between those having only WMH or only the OC (all p > .1; Figure 2). For high BMI, low AAI, and joint pain, those with only the OC were more likely to have mobility disability than those with only WMH (all p < .006; Figure 2).

Figure 2.

Estimated cross-sectional proportion with mobility disability by white matter hyperintensities (WMH), other conditions (see x-axis), both, or neither (n = 2,703). Results are for white women of mean age (74 years); relative differences were similar for other groups.

In those free of mobility disability at baseline, longitudinal results from Cox models were similar to the cross-sectional results (Table 3). Having only WMH or only the OC, except vision, was associated with a significantly increased hazard of mobility disability compared with having neither. A similar hazard ratio was observed for WMH as for weak grip, poor vision, and low FVC (p range: .1–.5). Joint pain, low AAI, and high BMI were each associated with a higher hazard ratio of developing mobility disability compared with WMH (p range: .0006–.02; Table 3).

Table 3.

HRs and 95% CIs for Incident Mobility Disability Over 6 Years by WMH, Other Conditions, Both, or Neither in Participants Free of Mobility Disability at Baseline (n = 2,234), Adjusted for Age, Gender, and Race

	Other Condition Absent				Other Condition Present
	WMH Absent		WMH Present		WMH Absent		WMH Present
Other Conditionsa	n	HR (CI)	n	HR (CI)	n	HR (CI)	n	HR (CI)	p Valueb
Weak grip	1,415	REF	583	1.27 (1.07–1.50)	145	1.56 (1.18–2.05)	91	1.46 (1.03–2.07)	.2
Poor vision	1,353	REF	564	1.23 (1.04–1.47)	207	1.13 (0.87–1.47)	110	1.29 (0.92–1.81)	.5
Low FVC	1,286	REF	507	1.28 (1.07–1.54)	274	1.54 (1.23–1.93)	167	1.55 (1.17–2.04)	.1
Joint pain	1,099	REF	486	1.2 (0.98–1.46)	461	1.76 (1.46–2.12)	188	2.37 (1.85–3.03)	.0006
Low AAI	1,453	REF	606	1.25 (1.06–1.48)	107	1.9 (1.39–2.60)	68	1.78 (1.20–2.64)	.02
High BMI	1,299	REF	570	1.19 (1.00–1.42)	261	1.59 (1.27–2.00)	104	2.26 (1.66–3.07)	.02

Note: AAI = ankle-arm index; BMI = body mass index; CI = confidence interval; FVC = forced vital capacity; HR = hazard ratio; WMH = white matter hyperintensities.

^aDefined as—Weak grip: gender-specific low grip strength; Poor vision: poor self-reported vision; Low FVC: gender-specific lowest quartile of FVC; Joint pain: self-reported joint pain; Low AAI: AAI < 0.9; and High BMI: BMI > 30.

^bCompares those with WMH and without the other condition to those with the other condition but without WMH.

Both cross-sectional and longitudinal results were attenuated but qualitatively similar after adjustment for diabetes, CHD, hypertension, stroke, smoking status, and height (Supplementary Tables 3 and 4). Results were unchanged when removing participants with stroke or using Parkinson’s medications at baseline or when adjusting for cognitive status, fall history, and depressive symptoms (data not shown).

Discussion

In this sample of community-dwelling older adults, WMH were related to slow gait speed and mobility disability, both cross-sectionally and longitudinally, to a similar extent as other well-established contributors to mobility. Moreover, coexistence of WMH with OCs was common. Gait speed and mobility disability were worst for participants with both WMH and another condition, though no synergistic interactions were observed. Cross-sectional and longitudinal findings for mobility disability were similar to one another. However, most of the effect for gait speed was observed at baseline and not longitudinally. This suggests that a trajectory of gait speed decline was already in place when we assessed mobility contributors. Possibly if we had assessed incident mobility contributors, we could have observed onset of this trajectory. These results indicate that WMH, either in isolation or in the presence of OCs, should not be overlooked in the understanding and treatment of mobility problems. The assessment of WMH is often absent from mobility research and practice and is rarely considered concurrently with other contributors (4).

Previous studies have reported associations between WMH and mobility limitations (4). However, this association has not been carefully assessed in relation to other age-related contributors. WMH are indicative of damage to white matter tracts that are essential for coordination of sensory processing and motor control during walking (27). WMH also disrupts tracts involved in executive function (28), a known cognitive correlate of mobility (29). Despite the plausibility of a causal association between WMH and mobility limitations, existing studies have been limited by not accounting for other age-related co-occurring contributors that could be confounders. We show here that the association of WMH with mobility is independent of many of these other contributors. However, having both WMH and another condition was no worse than expected from the sum of their individual associations. Age-related mobility limitations are typically of multifactorial origin (9). Our results highlight the need for an integrated approach that considers specific conditions in the context of other co-occurring conditions to fully understand and effectively treat mobility impairments.

These results have important implications for the management of mobility problems. First, evaluations should include indicators of WMH along with more traditionally assessed contributors and overt neurologic diseases. For a complex problem like mobility limitations, independent impairments in separate systems can have a relatively similar impact. By ignoring some systems, we may be underestimating the risk for mobility limitations. Second, common contributors to age-associated mobility limitations often co-occur. Therefore, clinical interventions will likely have greater impact if they account for impairments in multiple systems, including WMH (30,31). In our findings, high BMI and low AAI had the strongest associations with mobility. However, having WMH in the absence of high BMI or low AAI still resulted in poorer mobility, suggesting that addressing BMI or AAI alone would have limited effectiveness.

We observed similar results for the two mobility outcomes, measured gait speed and self-reported mobility disability. However, weak grip and poor vision may contribute more to gait speed and less to mobility disability whereas, low AAI, joint pain, and high BMI may be particularly important for mobility disability. Two important differences may explain these findings. First, mobility disability is self-reported and may reflect perception as much as ability (32). Second, gait speed is measured over a short distance (15 feet) whereas mobility disability is for a longer distance (half mile). Likely, some conditions have greater impacts on walking over longer distances than shorter ones.

These analyses utilized a well-characterized, representative cohort of older adults with many relevant measures. This allowed us to calculate precise estimates of mobility both cross-sectionally and longitudinally, which are likely generalizable to the community-dwelling older population. However, not all relevant measures were available, most notably peripheral nerve function. For each analysis, the group without WMH or the OC may have had one of the other contributors. However, only 14.1% of participants had more than two conditions. To simplify results, we considered all mobility contributors as dichotomous, which did not allow for assessment of severity. We used clinical cutpoints where available, but for many conditions, including WMH, these are not established. We also utilized a self-report measure of mobility disability which could introduce misclassification. Given the descriptive nature of these analyses, we did not account for multiple comparisons; however, our primary findings were that there were not significant differences between having WMH or another condition. Finally, MRI imaging in this study used older methodology, possibly reducing precision of WMH assessment. However, reduced precision in our predictor should result in reduced ability to detect associations. Thus, we may be underestimating the associations of WMH with mobility limitations.

Although treatments to reduce WMH do not yet exist, progression of WMH may be delayed with control of cardiovascular risk factors (5,6). Further research is needed to develop interventions that treat WMH. Effects of WMH on mobility may also be alleviated through specific rehabilitation strategies, including those that incorporate motor skill learning and cognitive training (31,33,34). Ultimately, addressing WMH along with other contributors may lead to more effective interventions to promote and preserve mobility in older adults.

Supplementary Material

Please visit the article online at http://gerontologist.oxfordjournals.org/ to view supplementary material.

Funding

This research was supported by contracts HHSN268201200036C, HHSN268200800007C, N01 HC55222, N01HC85079, N01HC85080, N01HC85081, N01HC85082, N01HC85083, N01HC85086, and grant HL080295 from the National Heart, Lung, and Blood Institute, with additional contribution from the National Institute of Neurological Disorders and Stroke. Additional support was provided by AG023629 from the National Institute on Aging. A full list of principal CHS investigators and institutions can be found at CHS-NHLBI.org. Additional funding for this analysis came from the National Institutes of Health through grant number KL2 TR000146 and the National Institute on Aging through grant number K01AG053431.