Cognition and Measures of Physical Activity,Mobility,and Gait in Individuals With Multiple Sclerosis: A Systematic Review

Abstract

Introduction

Impairments in cognition are prominent for individuals with multiple sclerosis (MS) and have been linked to low levels of physical activity (PA) and walking impairment. However, this relationship remains inconsistent, necessitating the synthesis of current literature to yield collective knowledge.

Objective

To investigate the relationship between cognition and measures of PA, mobility, and gait quality in individuals with MS.

Methods

Relevant, peer-reviewed research articles were identified through a systematic search of MEDLINE, EMBASE, PsycINFO, SPORTSDiscus, and CINAHL from inception to April 2, 2024. Eligible studies explored the relationship between cognition and measures of PA, mobility, and gait quality. The Standard Quality Assessment Criteria for quantitative studies was employed for quality assessment.

Results

Twenty-six studies with a total of 3248 participants were identified in this review, of which 21 studies indicated strong methodological quality. Our review found that processing speed holds a significant relationship with PA volume, but not PA intensity metrics. Mobility and gait quality outcomes were associated with varying cognitive domains, including processing speed, executive function, verbal memory, and visuospatial memory. The magnitudes of the association between cognition and PA, mobility, and gait quality were mostly weak-to-moderate.

Conclusion

Processing speed appears to be collectively associated with PA volume, mobility, and gait quality. However, the evidence supporting this conclusion is largely based on correlational studies involving individuals with mild-to-moderate ambulation disability, warranting future research.

Keywords

multiple sclerosis cognition mobility physical activity gait quality real-world ambulation

Introduction

Multiple sclerosis (MS) is a chronic neurological disease affecting the central nervous system, characterized by demyelination and axonal loss.¹ Mobility difficulties, one of the most prevalent symptoms in MS,² lead to limitations in physical activity (PA) and gait impairments.³ Moreover, cognitive impairment is common among individuals with MS, with impairments in processing speed, verbal memory, visuospatial memory, and executive function most commonly observed.^4,5 These impairments, in turn, adversely affect social participation and daily life, contributing to reduced overall quality of life.^6-9

While low levels of PA and gait impairment have often been considered primarily linked to mobility impairments,¹⁰ recent research highlights the crucial role of cognition. Previous research has investigated correlations between cognition and PA, including PA volume (eg, steps per day and number of activity bouts) and PA intensity metrics (eg, moderate-to-vigorous PA [MVPA]), indicating significant correlations between processing speed and PA outcomes even after adjusting for covariates.^11-13 However, the findings were inconsistent in previous studies, primarily attributable to different demographic characteristics and PA metrics. For example, Sandroff et al¹⁴ reported non-significant correlations between varying cognitive outcomes with MVPA in persons with MS who had a severe disability (median Expanded Disability Status Scale [EDSS] = 6). Furthermore, a previous study found sedentary behavior time as a significant predictor of processing speed,¹⁵ but other studies did not.^13,16

Consistent with research on the cognitive correlates of PA, studies have highlighted a significant relationship between cognition and mobility in controlled laboratory settings. Specifically, poorer processing speed, verbal memory, and executive function have been associated with decreased Timed-Up-and-Go (TUG) performance and greater step time variability.^17,18 However, a notable gap exists between gait metrics obtained in the lab and those gathered from everyday walking.¹⁹ This gap underscores the importance of wearable sensors enabling the collection of real-world mobility data that offer greater ecological validity compared to lab-based assessments of gait quality.²⁰ Research has particularly identified processing speed as a key contributor to gait speed and stride regularity in real-world settings,^19,21 but not verbal memory or executive function.²¹ Given the observed variations in walking across different settings, the cognitive correlates may also vary, highlighting the need for a comparison of the literature across both environments.

In this regard, synthesizing current literature seems useful to determine collective knowledge given the inconsistent findings between varying cognitive domains and PA, mobility, and gait quality.^11-16,19-21 The synthesis could yield 3 key insights. First, the investigation may help find the existence of a unique association between specific cognitive domains and PA, mobility, and gait outcomes. Second, the holistic exploration may examine whether the associations differ by distinct demographic characteristics or study settings. Third, the collective investigation may help determine the direction of the relationships among cognition, PA, and mobility, as previous studies have suggested a bidirectional relationship in healthy adults.^22,23 That is, decreased cognitive function and brain structure (eg, gray and white matter volume) are associated with reduced mobility and PA, as cognitive decline and poor brain health hinder mobility and PA. Conversely, greater mobility and PA support cognitive function and brain integrity. As such, it is salient to investigate whether similar bidirectional patterns are observed in individuals with MS. Taken together, establishing such relationships through the synthesis may hold clinical implications for the development of tailored interventions and offer recommendations for future studies.

To our knowledge, 2 previous reviews by Morrison and Mayer²⁴ and Lenne et al²⁵ synthesized relationships between PA and cognition in persons with MS. However, the reviews primarily aimed to examine the effects of PA on cognitive functions instead of a thorough investigation of the relationships between the 2. Furthermore, none of the reviews addressed mobility or gait quality outcomes. As such, a new systematic review is needed to synthesize recent literature to further understand the association of cognition with PA, mobility, and gait quality metrics. Therefore, the purpose of this systematic review was to investigate the relationships between cognition and measures of PA, mobility, and gait quality in individuals with MS.

Methods

This systematic review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) recommendations.²⁶ This systematic review was registered on the international database of prospectively registered systematic reviews in health and social care (International Prospective Register of Systematic Reviews [PROSPERO]; Registration number: CRD42024531243).

Information Sources and Study Selection

A systematic search was performed in MEDLINE, EMBASE, PsycINFO, SPORTSDiscus, and CINAHL from inception to April 2nd, 2024. The search query was developed by combining database-specific MeSH terms and keywords. Briefly, the broad keywords included were as followed: (“multiple sclerosis” OR “MS”) AND (“cognition” OR “processing speed”) AND (“physical activity” OR “gait quality”). The full search strategy performed in each database is shown in Appendix 1. The 2 authors (M.V. and M.B.) additionally searched relevant literature through websites (eg, Google Scholar) and checked references in included articles to retrieve further relevant studies (ie, citation search method).

The 2 authors (M.V. and M.B.) independently screened identified literature using the Covidence web-based collaboration software for systematic review (Veritas Health Innovation, Australia). First, the Covidence software automatically removed duplicated articles. Second, 2 authors performed title and abstract screening. Third, full-text screening was performed based on the Population, Intervention, Comparison, Outcome, and Study design (PICOS) study eligibility criteria. The authors discussed and resolved disparities of article inclusion/exclusion after the completion of abstract and full-text screening, respectively.

PICOS Eligibility Criteria

This study determined the PICOS criteria. Adults with MS (age ≥18 years) were included in the current review. We did not limit any MS phenotype (relapsing-remitting, secondary progressive, and primary progressive), EDSS,²⁷ and Patient Determined Disease Steps (PDDS).²⁸ The Intervention and Comparison criteria were not applicable in this study. We included studies if they examined correlational or associative relationships of cognition with PA, gait quality, or clinical mobility assessments. Specifically, studies were eligible in this review if they employed standardized neuropsychological assessments. We excluded studies employing self-reported cognitive assessments (eg, Perceived Deficit Questionnaire), as gaps between neuropsychological assessments and self-reported outcomes were found in this population.²⁹ PA and gait quality outcomes objectively measured by accelerometer, smartphone applications, or electronic walkways in either laboratory or real-world settings were included in this review. Additionally, studies that assessed mobility using clinical assessments (eg, TUG, 25-foot walk test [T25FW] and 6-minute walk test [6MWT]) were included. This review included observational studies published in English, including retrospective design, cross-sectional design, baseline data of clinical trials, secondary data analysis, and prospective cohort design. We excluded conference abstracts and case reports.

Quality Assessment of Included Studies

The 2 authors (M.V. and M.B.) independently evaluated the quality of included studies using the Standard Quality Assessment Criteria for quantitative studies.³⁰ When discrepancies in the assessment scores occurred, the 2 authors discussed their scores until reaching consensus. The quality assessment tool consists of 14 scales that assesses the range of the study’s methodological quality (eg, clarity of study purpose, subject selection, quality of analysis, confounding, and robustness of measurement), with the scales answered with Yes (2 points), Partial (1 point), No (0 point), and N/A (not applicable). The calculation strategies of the assessment tool were as follows: (1) Total sum = (number of “yes” × 2) + (number of “partial” × 1), (2) Total possible sum = 28 − (number of “N/A” × 2), and (3) Final score = Total sum/Total possible sum.³⁰ Higher scores of the final score indicated greater methodological quality, with the scores classified as low quality (<0.7), good quality (0.7-0.79), and strong quality (≥0.8).³¹

Data Extraction

One author (M.B.) initially extracted data from included studies, followed by the other reviewer (M.V.) reviewing the accuracy of the extracted data. The extracted information was as follows: (1) sample size, (2) sex, (3) mean age, (4) MS phenotype, (5) PDDS or EDSS score, (6) cognitive outcome, (7) gait metrics, and (8) brief result of the association between cognition and PA, mobility, or gait quality. We additionally extracted (1) sensor/equipment, (2) wear days, (3) data collection method, (4) data processing method, (5) data analysis, and (6) confounders (see Appendix 2). For the brief result, we extracted correlation metrics, such as Pearson correlation (r) or Spearman rank-order correlation (r_s). Importantly, we prioritized extracting partial correlation values, such as partial Pearson (pr) or Spearman (pr_s) correlations, when both adjusted and unadjusted values were offered. This allows for robustness of interpretation between cognition and PA and gait quality metrics in this review. This study interpreted the magnitude of correlation coefficients based on the cut-off values: negligible (0 to .10, or −.10 to 0), weak (.10 to .39, or −.39 to −.10), moderate (.40 to .69, or −.69 to −.40), strong (.70 to .89, or −.89 to −.70), and very strong correlations (.90 to 1.00, or −1.00 to −.90).³² Similarly, adjusted regression coefficients (B) were prioritized to be extracted over unadjusted models in included studies that examined relationships between cognition and PA and gait quality outcomes.

Data Synthesis

The characteristics of included studies were primarily synthesized into 3 categories: (1) demographic characteristics, (2) study methodology, and (3) relationship between outcomes. The demographic characteristics summarized sample size, age, sex, and disability levels in included studies. As for study methodology, we summarized and compared outcomes (eg, cognition, PA, mobility, and gait quality), analyses, study settings, and data collection methods across included studies. Finally, we determined the relationship between varying cognitive domains and PA (eg, step counts, light-intensity PA [LPA], and MVPA), mobility, and gait outcomes. Specifically, we examined if (1) specific cognitive domains were associated with PA, mobility, and gait outcomes, (2) the magnitude of relationships differed by MS severity, (3) bi-directional relationship was observed between outcomes, and (4) cognitive associations with mobility and gait quality outcomes were similar between different study settings (ie, laboratory vs real-world).

Results

Literature Search and Extracted Studies

The initial literature search yielded 2634 articles from databases. After eliminating duplicated studies (n = 517) and abstract/title screening (n = 2073), 50 studies remained for full-text screening. Full-text screening led to a final 16 studies that met our PICOS criteria. Ten articles retrieved from citation searching (n = 1)¹⁶ and website searching (n = 9)^19,21,33-39 met our eligibility criteria, such that a total of 26 articles were included in this systematic review. The PRISMA search and screening procedure are shown in Figure 1.

Figure 1.

PRISMA flow diagram.

Quality Assessment

About 21 out of 26 studies (80.7%) indicated strong methodological quality (≥0.80), while 3 studies^17,36,40 showed low quality and 2 studies^16,39 indicated good quality. Most of the included studies (n = 12) did not meet the criterion of “Some estimate of variance is reported for the main results?”. On the other hand, 18 out of 26 studies adjusted for covariates in their linear regression or partial correlation, in which age and sex were the most frequent covariates followed by disability status (eg, EDSS and PDDS).^{11-13,15,18,19,21,34,35,38,41-47} Tables 1 and 2 indicates the overall quality score of studies. Specific quality assessment results based on each criterion are shown in Appendix 3.

Table 1.

Methodological Characteristics of Included Studies That Examined Relationship Between Cognition and Physical Activity Metrics.

Study	Demographic characteristics	Cognitive outcome/measures	Gait metrics	Results	Quality score
Bollaert et al., 2019 Cross-sectional United States	N = 40	PS: SDMT VM: CVLT-2 VSM: BVMT-R EF, PS: PASAT	LPA (min/day) MVPA (min/day) # of activity bouts Duration of activity bouts Sedentary behavior # of sedentary bouts Duration of sedentary bouts	PS × LPA, r = .12, P > .05 PS × MVPA, r = .21, P > .05 PS × # of activity bouts, r = .40, P < .01 PS × duration of activity, r = .21, P > .05	0.64
	Sex (F/M) = 25/15
	Mean age = 65.3 ± 4.3
	MS phenotype = N/A			VM, VSM, and EF were not significantly correlated with PA outcomes in older adults with MS. Sedentary outcomes were not significantly correlated with PS, VM, VSM, EF in older adults with MS.
	PDDS = N/A
	EDSS median (IQR) = 4 (2)
Hubbard et al. 2015 Cross-sectional United States	N = 82	PS: SDMT	Sedentary time (min/day)	PS × Sedentary, r_s = −.10, P = .35	0.73
	Sex (F/M) = 62/20
	Mean age = 49 ± 9.1
	MS phenotype = RR
	PDDS = N/A
	EDSS median (IQR) = 3.5 (4.5)
Manglani et al., 2023 Cross-sectional United States	N = 80	PS VM: CVLT-2 VSM: BVMT-R	Total PA Step count LPA (min/day) MVPA (min/day) # of MVPA bouts Sedentary time # of sedentary bouts	MVPA, step count, and sedentary bouts were significantly associated with working memory/processing speed.	0.82
	Sex (F/M) = 63/17
	Mean age = 47.3 ± 8.0
	MS phenotype = N/A
	PDDS = N/A
	EDSS median (range) = 4 (0-5.5)
Motl et al., 2011 Cross-sectional United States	N = 33	PS: SDMT VM: SRT VSM: BVMT-R EF, PS: PASAT	Average steps per day	PS × Step count, pr = .35, P = .047 VM × Step count, pr = .20, P = .18	0.82
	Sex (F/M) = 22/11
	Mean age = 59.0 ± 10.0
	MS phenotype = N/A
	PDDS = N/A
	EDSS median (range) = 6 (3.5-7)
Motl et al., 2022 Cross-sectional United States	N = 60	PS: SDMT VM: CVLT-2 VSM: BVMT-R	Time of sedentary behavior (min/d) LPA (min/d) MVPA (min/d) Average steps per day	PS × MVPA, pr_s = .36, P < .05 PS × step count, pr_s = .31, P < .05 PS × LPA, pr_s = .26, P > .05 PS × Sedentary, pr_s = −.10, P > .05	0.91
	Sex (F/M) = 48/12
	Mean age = 49.5 ± 10.0
	MS phenotype = RR, SP, PP			VM, VSM were not significantly correlated with all PA outcomes after adjusting for covariates.
	PDDS = N/A
	EDSS = 3.5 ± 2.0
Sandroff et al., 2013 Cross-sectional United States	N = 82	PS: SDMT EF, PS: PASAT	Average steps per day	PS × step count, pr_s = .29, P = .01 EF × step count, pr_s = .12, P = .34	0.86
	Sex (F/M) = 62/20
	Mean age = 49 ± 9.1
	MS phenotype = RR, SP, PP
	PDDS = N/A
	EDSS median(range) = 3.5 (0-6.5)
Sandroff et al., 2014 Prospective United States	N = 212	PS: SDMT EF, PS: PASAT	Average steps per day	PS × step count, pr = .26, P < .01	0.86
	Sex (F/M) = 170/42
	Mean age = 50.0 ± 10.3
	MS phenotype = RR, SP, PP
	PDDS = N/A
	EDSS = N/A
Sandroff et al., 2020 Cross-sectional United States	N = 385	PS: SDMT	Average steps per day	PS × step count, pr = .11, P = .04 for MS with no CPS impairment PS × step count, pr = .29, P < .01 for MS with CPS impairment	0.86
	Sex (F/M) = 310/75
	Mean age = 49.9 ± 11.6
	MS phenotype = RR, PP
	PDDS median (IQR) = 2 (4)
	EDSS = N/A
Sandroff et al., 2022 Cross-sectional United States	N = 240	PS: SDMT VM: CVLT-2 VSM: BVMT-R	MVPA (min/d)	PS × MVPA, r = .06, P > .05 VM × MVPA, r = −.05, P > .05 VSM × MVPA, r = .01, P > .05	0.86
	Sex (F/M) = 146/94
	Mean age = 52.4 ± 7.0
	MS phenotype = SP, PP			MVPA were not associated with cognitive performance in persons with progressive MS who had substantial CPS impairment.
	PDDS = N/A
	EDSS median(range) = 6 (1.5-6.5)
Zheng et al., 2024 Cross-sectional United States	N = 147	PS: SDMT VM: CVLT-2 VSM: BVMT-R	LPA (min/d) MVPA (min/d) Daily steps (steps/d) Peak-30 cadence (Steps/min) Time spent in cadence bands (min/d) -Sporadic movement (20-39 step/min) -Purposeful steps (40-59 steps/min) -Slow walking (60-70 steps/min) -Medium walking (80-99 steps/min) -Brisk walking (100-119 steps/min) -Faster locomotion (≥120 steps/min)	PS × LPA, pr_s = .04, P > .05 PS × MVPA, pr_s = .11, P > .05 PS × daily steps, pr_s = .25, P < .01 PS × peak-30 cadence, pr_s = .28, P < .01 PS × Sporadic movement, pr_s = .23, P < .01 PS × Purposeful steps, pr_s = .24, P < .01 PS × Slow walking, pr_s = .32, P < .01 PS × Medium walking, pr_s = .27, P < .01 PS × Brisk walking, pr_s = .25, P < .01 PS × Faster locomotion, pr_s = .16, P < .05	0.95
	Sex (F/M) = 114/33 Mean age = 50.2 ± 13.5 MS phenotype = RR, PP PDDS median (IQR) = 1 (3) EDSS = N/A			VM × peak-30 cadence, pr_s = .18, P < .05 VM × Faster locomotion, pr_s = .19, P < .05
				VSM was not associated with PA outcomes after controlling for age, sex, and education.

Abbreviations: PS, Processing speed; VSM, Visuospatial memory; EF, Executive function; VM, Verbal Memory; EDSS, Expanded disability status scale; PDDS, Patient determined disease status; PA, Physical activity; LPA, Light intensity physical activity; MVPA, Moderate-to-vigorous intensity physical activity; r, Pearson’s correlation coefficient; pr, Partial Pearson’s correlation coefficient; r_s, Spearman’s correlation coefficient; pr_s, Partial Spearman’s correlation coefficient; B, Unstandardized coefficient.

Table 2.

Methodological Characteristics of Included Studies That Examined Relationship Between Cognition and Mobility and Gait Quality Metrics.

Study	Demographic characteristics	Cognitive outcome/measures	Gait metrics/clinical assessment	Results	Quality score
Cognition and gait quality in real-world setting
Ibrahim et al., 2023 Prospective Germany	N = 505	PS: SDMT	Step count SD of step count	PS × Step count, B = .22, P = .16 PS × SD of step count, B = 0-.17, P = .046	0.91
	Sex (F/M) = 363/142
	Mean age = 45.3 ± 11.6
	MS phenotype = N/A		Turn speed SD of turn speed	PS × Turn speed, B = .005, P < .01 PS × SD of turn speed, B = −.0006, P = .34
	PDDS = N/A
	EDSS = N/A
Shema Shiratzky et al., 2020 Cross-sectional Israel	N = 44	PS: SDMT	Gait speed (m/s) Cadence Stride regularity Sample entropy	PS × Gait speed, pr = .50, P < .01 PS × Cadence, pr = .33, P < .05 PS × Stride regularity, pr = .35, P < .05 PS × Sample entropy, pr = .33, P < .05	0.86
	Sex (F/M) = 32/12
	Mean age = 49.2 ± 10.7
	MS phenotype = N/A		Steps per day Total activity # of walking bouts≥30 s	PS × step count, pr = .45, P < .01 PS × total activity, pr = .32, P < .05 PS × walking bouts, pr = .43, P < .01
	PDDS = N/A
	EDSS median (range) = 3.5 (2-6)
VanNostrand et al., 2024 Cross-sectional United States	N = 20	PS: SDMT VM: SRT IH: Stroop EF, PS: PASAT	Gait speed (m/s) Stride regularity	PS × Gait speed, r = .61, P < .01 PS × Stride regularity, r = .59, P < .01 PS × walking bouts, r = .48, P < .01 PS × Gait speed, B = .44, P = .03 PS × Stride regularity, B = .46, P = .02 PS × walking bouts, B = .27, P = .30	0.91
	Sex (F/M) = 16/4
	Mean age = 56.4 ± 8.6 MS phenotype = RR, SP, PP PDDS median (range) = 1(0-4) EDSS = N/A
			# of walking bout ≥30 sTime spent walking	SDMT was significant predictors of gait speed, stride regularity.EF, VM, IH were not correlated with gait quality and PA outcomes (P > .05).
Cognition and gait quality and mobility in laboratory setting
Allali et al., 2012Cross-sectionalSwitzerland	N = 20	PS: WAIS-III SDMTVM: SRTEF: TMT-B, IH: Stroop	TUG	PS × TUG, r_s = −.62, P < .01VM × TUG, r_s = −.56, P < .01EF × TUG, r_s = .50, P < .05IH × TUG, r = .38, P > .05	0.68
	Sex (F/M) = 14/6
	Mean age = 39.6 ± 8.0
	MS phenotype = N/A
	PDDS = N/A
	EDSS = 2.2 ± 1.14
Benedict et al., 2011Retrospective,Cross-sectionalUnited States	N = 211	PS: SDMTVM: CVLTVSM: BVMT-REF, PS: PASATEF: DKEFS	T25W	PS × T25FW, r = −.48, P < .01VM × T25FW, r = −.47, P < .01VSM × T25FW, r = −.39, P < .01EF, PS × T25FW, r = −.36, P < .01EF × T25FW, r = −.42, P < .01	0.91
	Sex (F/M) = 167/44
	Mean age = 44.9 ± 10.0
	MS phenotype = N/A			The T25W was accounted for by processing speed, working memory, and Executive function.
	PDDS = N/A
	EDSS = 2.8 ± 1.6
Bilgin et al., 2023ProspectiveTurkey	N = 124	PS: SDMT	T25FWTUG	PS × T25FW, r_s = −.32, P < .05PS × TUG, r_s = −.20, P > .05	0.86
	Sex (F/M) = 80/44
	Mean age = 38.2 ± 9.1
	MS phenotype = RR
	PDDS = N/A
	EDSS = 1 ± 1.3
Bollaert et al., 2019Cross-sectionalUnited States	N = 40	PS: SDMTVM: CVLT	T25FWTUG6MWT	PS × T25FW, r = .35, P < .05PS × TUG, r = −.37, P < .05PS × 6MWT, r = .37, P < .05	0.95
	Sex (F/M) = 25/15
	Mean age = 65.3 ± 4.3MS phenotype = RR, SP, PPPDDS = N/AEDSS median (IQR) = 4 (2)			VM × T25FW, r = −.06, P > .05VM × TUG, r = .002, P > .05VM × 6MWT, r = .06, P > .05SDMT was significant predictors of T25W, TUG, and 6MWT.
Cutter et al., 1999Cross-sectionalUnited States	N = N/A	PS: SDMTEF, PS: PASAT	T25W	EF, PS × T25W, r_s = .21	0.45
	Sex (F/M) = N/A
	Mean age = N/A
	MS phenotype = N/A
	PDDS = N/A
	EDSS = N/A
D'Orio et al. 2012Cross-sectionalUnited States	N = 81	PS: SDMTVM: CVLTVSM: BVMT-REF, PS: PASAT	T25FW	PS × T25FW, r = −.51, P < .001VM × T25FW, r = −.26, P = .02VSM × T25FW, r = −.38, P = .001EF, PS × T25FW, r = −.30, P = .015SDMT was a significant predictor of walking speed.	0.86
	Sex (F/M) = 54/27
	Mean age = 47.2 ± 10.2
	MS phenotype = RR, SP, PP
	PDDS = N/A
	EDSS = N/A
Højsgaard Chow et al., 2018RetrospectiveDenmark	N = 52	PS: SDMTEF, PS: PASAT	T25W	PS × T25W, r_s = −.14, P > .05EF, PS × T25W, r_s = −.11, P > .05	0.82
	Sex (F/M) = 27/25
	Median age (IQR) = 51 (47-57)
	MS phenotype = SP, PP
	PDDS = N/A
	EDSS median (IQR) = 4.5 (1.5)
Hsieh et al., 2017Cross-sectionalUnited States	N = 191	PS: SDMT	Step time variabilityStep length variability	PS × Step time variability, B = −.007, P < .05PS × Step length variability, B = −.004, P < .05	0.95
	Sex (F/M) = 129/62
	Mean age = 56.6 ± 11MS phenotype = N/APDDS = N/AEDSS = 4.9 ± 1.5			SDMT was significant predictors of step time variability and step length variability.
Kalron et al., 2018Cross-sectionalIsrael	N = 355	Global cognitive score, Memory, EF, VSM, PS, attention: Mindstreams computerized cognitive battery	Step time variability	Global cognitive score × step time variability, B = −.04, P < .01Memory × step time variability,B = −.03, P < .01EF × step time variability,B = −.024, P < .05Attention × step time variability,B = −.02, P > .05PS × step time variability,B = −.01, P > .05VSM × step time variability,B = −.01, P > .05Global cognitive score, memory, and executive functions were significant predictors of step time variability.	0.95
	Sex (F/M) = 215/137
	Mean age = 41.4 ± 13.5
	MS phenotype = RR, SP, PP
	PDDS = N/A
	EDSS = 2.7 ± 1.8
Motl et al., 2021Cross-sectionalUnited States	N = 62	PS: SDMT	T25W6MWT	PS × T25FW, r = .59, P < .05PS × 6MWT, r = .58, P < .05PS × T25FW, B = .08, P < .05PS × 6MWT, B = .24.7, P < .05	0.91
	Sex (F/M) = 48/14
	Mean age = 50.6 ± 9.3
	MS phenotype = RR, PP			SDMT was significant predictors of T25FW and 6MWT.
	PDDS = N/A
	EDSS median (IQR) = 5 (3)
Sandroff et al., 2015Cross-sectionalUnited States	N = 96	EF, PS: PASAT	SSST	EF, PS × SSST, r_s = −.35, P < .05	0.77
	Sex (F/M) = 77/19
	Mean age = 52.7 ± 11.1
	MS phenotype = RR, PP
	PDDS = N/A
	EDSS median = 4.5
Sebastião et al., 2016Cross-sectionalUnited States	N = 47	PS: SDMT	TUG	PS × TUG, r_s = −.59, P < .05	0.82
	Sex (F/M) = 42/5
	Mean age = 53 ± 11.4
	MS phenotype = RR, SP, PP
	PDDS = N/A
	EDSS median = 4
Takla et al., 2023Cross-sectionalUnited States	N = 39	PS: SDMTVM: SRTVSM: BVMT-R	Forward walk speedBackward walk speed	PS × Forward speed, r = .32, P > .05VM × Forward speed, r = .23, P > .05VSM × Forward speed, r = .15, P > .05	0.86
	Sex (F/M) = 35/4
	Mean age = 50.9 ± 9.5
	MS phenotype = RR			PS × Backward speed, r = .28, P > .05VM × Backward speed, r = .11, P > .05VSM × Backward speed, r = .35, P < .05
	PDDS = 3.3 ± 2.1
	EDSS = N/A

Abbreviations: PS, Processing speed; VSM, Visuospatial memory; EF, Executive function; VM, Verbal Memory; EDSS, Expanded Disability Status Scale; PDDS, Patient determined disease status; PA, Physical activity; LPA, Light intensity physical activity; MVPA, Moderate-to-vigorous intensity physical activity. T25W, Timed 25-Foot Walk Test; TUG, Timed-Up-and-Go; 6MWT, 6-Minute Walk Test; SSST, Six-Spot Step Test. SDMT, CVLT, BVMT-R, SRT; PASAT, Paced auditory serial addition test; DKEFS, Delis–Kapan Executive Function System Sorting Test; r, Pearson’s correlation coefficient; pr, Partial Pearson’s correlation coefficient; r_s, Spearman’s correlation coefficient; pr_s, Partial Spearman’s correlation coefficient; B, Unstandardized coefficient; SD, Standard deviation.

Participant Characteristics

A total number of 3248 persons with MS participated in the included studies, with a median sample size of 81 (range = 20-505). As expected, most participants were female (n = 2346, 72.2%). About 23 out of 26 studies recruited participants with a mean age ranging from 38.2 to 59.0 years.^42,48 Only 2 studies primarily recruited older adults with MS (mean age = 65.3 ± 4.3 years),^40,45 and 1 study did not offer age information.³⁶ All included studies did not limit MS phenotypes in their eligibility criteria, except for 2 studies that only recruited those with progressive MS diagnosis (ie, primary progressive and secondary progressive).^14,37 Based on the PDDS and EDSS cut-off scores,^49,50 all studies primarily recruited those having mild-to-moderate MS ambulatory disability, except for 2 studies whose participants had a severe ambulatory disability (median EDSS score = 6.0).^14,42 Notably, 2 studies examined the relationship between cognition and PA outcomes in those with cognitive processing speed impairment.^13,43 Detailed methodological characteristics of the included studies are shown in Table 1.

Characteristics of Included Studies

All studies examined the association between cognition and PA, mobility, or gait quality outcomes conducted cross-sectional (n = 21),^{11,13-19,21,33,34,36,38-40,42-47} retrospective (n = 2),^35,37 or prospective (n = 3)^12,41,48 designs. Specifically, all studies analyzed correlational relationships between 2 outcomes. Nine studies^{15,18,21,35,38,41,45-47} further employed regression analyses to examine relationships of cognition with PA, gait quality, or mobility outcomes, and vice versa. Out of 26 studies, 13 studies measured PA only (n = 10)^{11-16,40,42-44} and both PA and gait quality outcomes (n = 3)^19,21,41 in real-world settings. Thirteen studies assessed mobility (n = 10)^{17,33,35-39,45,46,48} and gait quality outcomes (n = 3)^18,34,47 in laboratory settings. Most studies (19 out of 26 studies, 73.1%) were conducted in the United States. Notably, all 10 studies that measured PA were conducted in the United States.

All but 2 studies^18,39 assessed cognitive processing speed using the Symbol Digit Modality Test (SDMT) assessed by written or oral version and in a laboratory or remote setting. About 12 out of 26 (46.1%) studies assessed verbal memory using either the California Verbal Learning Test-2 (CVLT-2)^{13-15,35,40,44-46} or the Selective Reminding Test (SRT).^17,21,34,42 Brief Visuospatial Memory Test-Revised (BVMT-R) and Paced Auditory Serial Addition Test (PASAT) were used to assess visuospatial memory (n = 9)^{13-15,34,35,40,42,44,46} and executive function/processing speed (n = 11),^{11,12,21,35-40,42,46} respectively. Only 2 studies assessed cognitive inhibition using the color-word Stroop test.^17,21 Manglani et al¹⁵ utilized exploratory factor analysis to develop composite measures of processing speed/executive function (SDMT, PASAT, Processing Speed Index and Working Memory Index from the Wechsler Adult Intelligence Scale, and the List Sorting Test from the NIH Toolbox), and verbal memory (CVLT and NIH Toolbox Flanker).

The studies assessing PA outcomes used ActiGraph GT3x+ (n = 9)^{11-16,40,43,44} or Step Activity Monitor (n = 1)⁴² accelerometers, all of which instructed participants to wear the sensor for 7 days, except for Bollaert and Motl⁴⁰ who did not articulate wear time. It was consistent across the studies that at least 10 hours of wear time was eligible for a valid day, whereas number of valid days included in the analyses varied from either +1 valid days,^13,14,44 +2 valid days,^15,43 or +3 valid days.^16,42 The outcomes extracted from the accelerometer included varying PA metrics (eg, light-intensity PA, moderate-to-vigorous PA, and step counts) and sedentary metrics (eg, sedentary time and number of sedentary bouts). Interestingly, Zheng et al⁴⁴ calculated a peak 30-minute cadence (ie, the average of the highest 30-minute cadences in a day) and time spent cadence bands in addition to typical PA outcomes. Detailed descriptions regarding sensor instruction, processing, and analysis are shown in Appendix 2.

Two studies employed an AX3 sensor for assessing both PA and gait quality outcomes, with the sensor wear time of 3²¹ and 7 days.¹⁹ Both studies extracted metrics of gait speed, stride regularity, Time spent walking, and total number of walking bouts, while Shema-Shiratzky et al¹⁹ additionally extracted cadence and sample entropy. Both studies only included walking bouts ≥30 seconds in data analysis. A smartphone application was employed to assess gait quality in real-world settings.⁴¹ Notably, the data collection methods were unique in that participants were instructed to perform clinical assessments (ie, 2-minute walking test and Five-U-Turn Test) in their community settings. The application assessed step count, step power, turn speed, and standard deviation of step count and turn speed.

Thirteen studies assessed mobility and gait quality metrics in laboratory settings.^{17,18,33-39,45-48} Specifically, 10 studies employed clinical assessments to measure mobility, where Timed 25-foot walk test (T25FW; n = 7)^{35-38,45,46,48} was most frequently assessed, followed by TUG (n = 4),^17,33,45,48 6MWT (n = 2),^38,45 and Six-Spot Step Test (SSST; n = 1).³⁹ Three studies employed an electronic walkway to measure gait variability metrics (eg, step time variability and step length variability)^18,47 or gait speed.³⁴

Cognitive Function and PA Outcomes

We observed relatively consistent findings that higher cognitive processing speed was significantly correlated with increased PA outcomes. All 7 studies assessing step counts indicated a significantly positive correlation with processing speed, regardless of disability level or cognitive impairment, while the magnitudes of correlation were mostly weak in 6 of 7 studies (correlation range = .11-.35).^{11-13,19,42-44} Similarly, number of walking/activity bouts demonstrated significantly positive and moderate correlations with processing speed in all studies (correlation range = .40-.48).^19,21,40 However, the correlation between processing speed and MVPA remained insufficient given 3 of 4 studies indicating non-significant and negligible-to-weak magnitude (correlation range = .06-.36).^40,43,44 The correlation of LPA with cognitive processing speeds was negligible-to-weak and non-significant (correlation range = .04-.26) in all studies,^13,40,44 as did sedentary time (pr_s = −0.10).^13,16

The findings of the correlational relationship were limited and inconsistent in advanced MS. Of the 2 studies recruiting advanced MS with severe disability levels (median EDSS = 6.0), they assessed distinct PA metrics, including step counts⁴² or MVPA.¹⁴ While Motl et al⁴² found a significant but weak correlation between processing speed and step counts (pr_s = .36, P < .05), processing speed was not correlated with MVPA in the other study (r = .06, P > .05).¹⁴

We found bidirectional relationships between PA metrics and cognition. Specifically, Manglani et al¹⁵ indicated that MVPA, step count, and number of sedentary bouts were significant predictors of working memory/processing speed. On the other hand, Ibrahim et al⁴¹ indicated that higher processing speed was significantly associated with greater step counts. VanNostrand et al²¹ found that processing speed did not significantly predict total walking bouts after adjusting covariates.

Other cognitive outcomes, including visuospatial memory, verbal memory, and executive function, were not significantly correlated with both PA and sedentary outcomes, regardless of the severity of MS or cognitive impairment. Only Zheng et al⁴⁴ found a significant correlation between verbal memory and higher PA intensity. Specifically, verbal memory showed a significant but weak correlation with peak 30-minute cadence (pr_s = .18, P < .05) and faster locomotion (≥120 steps/min, pr_s = .19, P < .05).

Cognitive Function and Gait Quality Outcomes

As with PA outcomes, cognitive processing speed was significantly correlated with gait quality and mobility outcomes, regardless of walking environment.^19,21 Specifically, processing speed showed moderate correlations with gait speed (correlation range = .50-.61) and weak-to-moderate relationship with stride regularity (correlation range = .35-.59) in a real-world setting.^19,21 Similarly, all studies conducted in a laboratory setting showed a significant weak-to-moderate correlation between processing speed and T25FW (correlation range = .20-.59),^{35,36,38,45,46,48} except for Højsgaard Chow et al³⁷ that only recruited advanced MS (r_s = −.14, P > .05). We also observed significant weak-to-moderate correlation between processing speed and TUG (correlation range = −.62 to −.37) in 3 of 4 studies^17,33,45 and 6MWT (correlation range = .37-.58) in all studies.^38,45

Out of 8 studies examining the association in adjusted models,^{18,21,35,38,41,45-47} 6 studies indicated impaired processing speed was significantly associated with slower gait and turn speed, decreased stride regularity, greater variability in step time and step length, and decreased T25FW and 6MWT performance. However, 2 studies found a non-significant association of processing speed with turn speed variability and step time variability.^18,41 No studies examined the impact of mobility and gait quality outcomes on cognitive domains, making the bidirectional relationship unclear.

Only VanNostrand et al²¹ examined the relationship between gait quality metrics with cognitive outcomes other than SDMT in a real-world setting, reporting that none of visuospatial memory, verbal memory, executive function, and cognitive inhibition was significantly associated with gait speed and stride regularity. Contrary to the study, we observed significant relationships between varying cognitive domains and mobility and gait quality outcomes in laboratory settings. All studies measuring PASAT indicated a significant but weak correlation with T25FW (correlation range = .21-.36) and SSST (r_s = −.35, P < .05), except for Højsgaard Chow et al³⁷ that only recruited advanced MS (r_s = −.11, P > .05). Two studies measured executive function other than PASAT showed a significant relationship with step time variability (B = −.02, P < .05) and TUG (r_s = .50, P < .05).^17,18 About 4 out of 6 studies found significant relationship of verbal memory with TUG (r_s = −.56, P < .01), T25FW (correlation range = −.47 to −.26, all P < .05), and step time variability (B = −.03, P < .01).^17,18,35,46 Visuospatial memory was weakly but significantly correlated with T25FW and backward gait speed in 2 out of 3 studies (correlation range = .11-.38).^34,46

Discussion

Our review highlights that cognition, particularly processing speed, holds significant and positive relationship with varying PA volume metrics (eg, step counts and activity bouts), while the magnitudes were mainly negligible-to-moderate. Impairments in processing speed were weakly-to-moderately associated with reduced varying gait quality metrics (eg, gait speed, turn speed, and stride regularity) and mobility performance (eg, TUG, T25FW, and 6MWT). Furthermore, we observed that other cognitive domains, including executive function, verbal memory, and visuospatial memory, were associated with mobility and gait quality metrics. Notably, these findings were consistent across individuals with MS with varying levels of cognitive impairment, as some studies included samples exhibiting cognitive impairment while others did not.⁵¹

Our study observed significant relationships between processing speed and PA metrics, indicating that persons with MS who have greater processing speed exhibit greater PA levels, and vice versa. Previous studies supported the bidirectional relationship. First, research has suggested that regular PA or exercise may enhance cerebral perfusion, synaptic neuroplasticity, brain volume, brain structure, brain connectivity, and brain-derived neurotrophic factor, which could ultimately support improvements in cognitive functions.⁵² The unique association between processing speed and PA outcomes indicates that PA or exercise mode involves distinct mechanisms leading to cognitive adaptations of interest.^53,54 Second, cognitive dysfunctions are correlated with decreased activities of daily living and increased depressive symptoms,⁴ all of which are significant predictors of physical inactivity.⁵⁵ Furthermore, sufficient cognitive performance is necessary to maintain healthy behavior, including PA.⁵² In this regard, cognitive dysfunction may deteriorate the ability to refrain from unhealthy behavior, such as sedentary behavior. Thus, it is not surprising that cognitive impairment may lead to declines in PA in persons with MS. Nevertheless, given the cross-sectional and correlational nature of the findings, the causal relationships should be interpreted with caution.

We found that PA volume showed negligible-to-moderate but significant association with processing speed, indicating that PA volume may be a salient component relevant to cognitive performance. This holds clinical implication that targeting higher PA levels may induce effective neural adaptations that lead to improvement in cognitive function.⁵⁶ However, contrary to PA volume metrics, the correlation between processing speed and PA intensity metrics (eg, LPA and MVPA) were negligible-to-weak and non-significant in most studies.^13,40,43,44 Only Zheng et al⁴⁴ investigated PA intensity metrics beyond MVPA and observed that faster locomotion (≥120 steps/min) and peak 30-minute cadence (ie, the average of the highest 30-minute cadences in a day) were significantly correlated with both processing speed and verbal memory, independent of age, sex, and education. As such, a further investigation regarding PA intensity metrics is recommended in future studies to corroborate the relationship between cognition and PA intensity.

Only 2 studies recruited advanced MS with severe disability levels (median EDSS = 6.0), and the evidence remains inconclusive. Although Motl et al⁴² showed a weak but significant correlation (pr_s = .36) between processing speed and step counts, it should be interpreted carefully as the study only recruited 33 participants. On the other hand, Sandroff et al¹⁴ involving 385 individuals with advanced MS found that MVPA showed a negligible and non-significant correlation with processing speed. These findings may be more representative of the advanced MS population as a whole. It is plausible that the severe MS disease process deteriorates the capacity to induce neuroplastic changes in response to PA.⁵⁶ This aligns with a previous study, wherein the same PA program showed clinically meaningful changes in SDMT in those with mild disability, but not in those with greater disability.⁵⁴ Thus, further investigation targeting advanced MS is needed to develop tailored PA interventions for this population.

Extending the association between processing speed and PA, our review further highlighted the importance of this cognitive domain for mobility and gait quality metrics. That is, studies found that lower processing speed was weakly to moderately but significantly correlated with decreased gait speed, gait variability, and mobility performance in both real-world and laboratory-based settings.^{17,19,21,33,35,36,38,39,45,46,48} Given the existing literature identifying the shared cortical regions associated with mobility and cognition,^57,58 it is not surprising that declines in processing speed are significantly correlated with reduced mobility and gait quality. However, this finding may not be generalizable to advanced MS, as Højsgaard Chow et al,³⁷ who specifically recruited individuals with advanced MS, reported a weak and non-significant relationship between processing speed and mobility outcomes.

It is notable that other cognitive functions, including executive function, verbal memory, and visuospatial memory, were also correlated with varying mobility and gait quality outcomes, particularly in laboratory settings. These findings indicate that varying cognitive domains are involved in mobility and gait, while the evidence remains to be further expanded in ecologically valid settings. Only VanNostrand et al²¹ measured processing speed, verbal memory, executive function, and cognitive inhibition indicating that no cognitive domains other than processing speed correlated with real-world measures of gait quality metrics. It is contrary to a previous finding that real-world measures of mobility were more closely associated with cognitive function than those measured in laboratory in different populations.⁵⁹ However, given that the study was conducted with a small sample size (n = 20), further research with a larger sample size and varying gait and mobility metrics should be conducted to corroborate the relationship in real-world settings.

The methods of processing PA, sedentary behavior, and gait quality outcomes through accelerometers were relatively heterogeneous. Particularly, while all studies collecting PA and sedentary outcomes had 7 days of wear time, the number of valid days analyzed ranged from 1 to 3 days or more. Considering Klaren et al⁶⁰ suggested 3 to 5 and 4 to 6 consecutive or non-consecutive days of valid data for acceptable reliability of measures of PA and sedentary behavior, studies involving a low number of valid days may lack ecological validity of the data and limit the generalizability of findings.^13-15,43,44 Of studies measuring real-world gait quality data, VanNostrand et al collected for 3 days whereas Shema-Shiratzky et al¹⁹ did so for 7 days. While differences in collection period length were seen, previous studies have identified 2 to 3 days of gait monitoring as sufficient to capture most variability in gait metrics.⁶¹ Nonetheless, our current review builds upon previous literature demonstrating the usefulness of wearable sensors in understanding real-world PA and gait quality in individuals with MS.²⁰ Furthermore, while the studies included in this review characterized gait speed as a gait quality metric, this is not universal across the literature regarding real-world data collection.^62,63 Therefore, establishing standardized terminology is crucial for greater consistency. We suggest future studies assess PA and gait quality using efficient but reliable data processing methods to ensure ecological validity of the data.

Only 3 studies in this review examined the relationship between cognition and real-world gait quality metrics, and only a limited number of cognitive outcomes were assessed.^19,21,41 Thus, future studies should be conducted with specific demographic populations and varying cognitive outcomes to offer further detailed insights regarding cognitive correlates with gait quality in real-world settings. While the inclusion of wearables in the acquisition of real-world mobility data has been justified by the variability seen in mobility resulting from an MS diagnosis,⁶⁴ cognition is still examined cross-sectionally, under controlled conditions. Given the variability of cognition within and across days in individuals with MS,^65,66 and evidence that environmental distractions can influence cognitive performance,⁶⁷ future research should incorporate ecological momentary assessment. This approach would enhance the ecological validity of cognitive assessments while offering a deeper understanding of the dynamic interplay between mobility and cognition over time. It also provides valuable insights into the directionality of their relationship and the potential presence of a lead-lag effect.

Limitations

Despite noteworthy findings and strengths in this review, there are several limitations. Most studies were conducted in the selected countries (ie, 19 out of 26 studies conducted in the US) and the same research groups (ie, 13 out of 26 studies), which may restrict generalizability to our study findings. This study did not perform a meta-analysis due to the heterogeneous PA and gait quality metrics in the included studies. Future meta-analysis is warranted to further extend our findings. Finally, there may be a potential bias in this review as the literature search and screening were performed by only 2 authors and the data extraction procedures were not performed independently.

Conclusion

This systematic review indicates that cognition, particularly cognitive processing speed, holds significant correlations with varying PA volume, mobility, and gait quality metrics in persons with mild-to-moderate disability, whereas the association remains inconclusive in advanced MS. In this review, however, cognitive processing speed was not associated with PA intensity metrics (eg, LPA and MVPA) in most studies. We also found that executive function, verbal memory, and visuospatial memory showed an association with mobility and gait quality outcomes, however, the evidence remains to be further investigated in real-world settings.

Footnotes

Appendix

Appendix 3.

Quality Assessment of Studies.

Study	C1	C2	C3	C4	C5	C6	C7	C8	C9	C10	C11	C12	C13	C14	Final	Quality
Studies that examined relationship between cognition and physical activity
Bollaert et al. 2019	1	2	1	1	N/A	N/A	N/A	1	2	2	1	0	1	2	0.64	Low
Hubbard et al. 2015	2	2	1	1	N/A	N/A	N/A	2	2	2	0	0	2	2	0.73	Good
Manglani et al. 2023	2	2	1	2	N/A	N/A	N/A	1	2	2	0	2	2	2	0.82	Strong
Motl et al. 2011	2	2	0	2	N/A	N/A	N/A	2	2	2	1	2	1	2	0.82	Strong
Motl et al. 2022	2	2	2	2	N/A	N/A	N/A	2	2	2	0	2	2	2	0.91	Strong
Sandroff et al. 2013	2	2	1	2	N/A	N/A	N/A	2	2	2	0	2	2	2	0.86	Strong
Sandroff et al. 2014	2	2	1	2	N/A	N/A	N/A	2	2	2	0	2	2	2	0.86	Strong
Sandroff et al. 2020	2	2	1	2	N/A	N/A	N/A	2	2	2	0	2	2	2	0.86	Strong
Sandroff et al. 2022	2	2	1	2	N/A	N/A	N/A	2	2	2	0	2	2	2	0.86	Strong
Zheng et al. 2024	2	2	2	2	N/A	N/A	N/A	2	2	2	1	2	2	2	0.95	Strong
Studies that examined relationship between cognition and mobility and gait
Allali et al. 2012	2	2	2	2	N/A	N/A	N/A	1	1	0	1	0	2	2	0.68	Low
Benedict et al. 2011	2	2	2	2	N/A	N/A	N/A	2	2	2	0	2	2	2	0.91	Strong
Cutter et al. 1999	2	2	0	0	N/A	N/A	N/A	1	2	0	0	0	1	2	0.45	Low
Bilgin et al. 2023	2	2	2	2	N/A	N/A	N/A	2	2	2	1	0	2	2	0.86	Strong
Bollaert et al. 2019	2	2	2	2	N/A	N/A	N/A	2	2	2	1	2	2	2	0.95	Strong
D’Orio et al. 2021	2	2	2	2	N/A	N/A	N/A	1	2	1	1	2	2	2	0.86	Strong
Højsgaard Chow et al. 2018	2	2	2	2	N/A	N/A	N/A	2	2	2	0	0	2	2	0.82	Strong
Hsieh et al. 2017	2	2	2	2	N/A	N/A	N/A	1	2	2	2	2	2	2	0.95	Strong
Ibrahim et al. 2023	2	2	0	2	N/A	N/A	N/A	2	2	2	2	2	2	2	0.91	Strong
Kalron et al. 2018	2	2	2	2	N/A	N/A	N/A	1	2	2	2	2	2	2	0.95	Strong
Motl et al. 2021	2	2	2	2	N/A	N/A	N/A	2	2	2	1	1	2	2	0.91	Strong
Sandroff et al. 2015	2	2	2	2	N/A	N/A	N/A	2	2	1	0	0	2	2	0.77	Good
Sebastião et al. 2016	2	2	1	2	N/A	N/A	N/A	1	2	2	2	0	2	2	0.82	Strong
Shema-Shiratzky et al. 2020	2	2	1	2	N/A	N/A	N/A	2	2	2	0	2	2	2	0.86	Strong
Takla et al. 2023	2	2	2	2	N/A	N/A	N/A	2	1	2	0	2	2	2	0.86	Strong
VanNostrand et al. 2024	2	2	2	2	N/A	N/A	N/A	2	1	2	1	2	2	2	0.91	Strong
# of “0” scores	0	0	3	1	N/A	N/A	N/A	0	0	2	12	8	0	0

Abbreviations: C, Criteria; Yes = 2, Partial = 1, No = 0, N/A, not applicable; Final score is indicative of low quality (<0.7), good quality (0.7-0.79), and strong quality (≥0.8).

C1: Question/objective sufficiently described?

C2: Study design evident and appropriate?

C3: Method of subject/comparison group selection or source of information/input variables described and appropriate?

C4: Subject (and comparison group, if applicable) characteristics sufficiently described?

C5: If interventional and random allocation was possible, was it described?

C6: If interventional and blinding of investigators was possible, was it reported?

C7: If interventional and blinding of subjects was possible, was it reported?

C8: Outcome and (if applicable) exposure measure(s) well defined and robust to measurement/misclassification bias? Means of assessment reported?

C9: Sample size appropriate?

C10: Analytic methods described/justified and appropriate?

C11: Some estimate of variance is reported for the main results?

C12: Controlled for confounding?

C13: Results reported in sufficient detail?

C14: Conclusions supported by the results?

Author Contributions

Myeongjin Bae; Conceptualization; Data curation; Formal analysis; Methodology; Writing—original draft; and Writing—review & editing. Michael VanNostrand; Conceptualization; Data curation; Formal analysis; Methodology; Supervision; Writing—original draft; and Writing—review & editing.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Myeongjin Bae

Michael VanNostrand

References

Calabresi

PA.

Diagnosis and management of multiple sclerosis. Am Fam Physician. 2004;70(10):1935-1944.

LaRocca

NG.

Impact of walking impairment in multiple sclerosis. Patient. 2011;4(3):189-201.

Socie

Motl

Pula

Sandroff

Sosnoff

JJ.

Gait variability and disability in multiple sclerosis. Gait Posture. 2013;38(1):51-55.

Chiaravalloti

DeLuca

Cognitive impairment in multiple sclerosis. Lancet Neurol. 2008;7(12):1139-1151.

van Dam

Krijnen

Nauta

, et al. Identifying and understanding cognitive profiles in multiple sclerosis: a role for visuospatial memory functioning. J Neurol. 2024;271(5):2195-2206.

Coote

Comber

Quinn

Santoyo-Medina

Kalron

Gunn

Falls in people with multiple sclerosis: risk identification, intervention, and future directions. Int J MS Care. 2020;22(6):247-255.

Kasser

Zia

Mediating role of resilience on quality of life in individuals with multiple sclerosis: a structural equation modeling approach. Arch Phys Med Rehabil. 2020;101(7):1152-1161.

Rooney

Riemenschneider

Dalgas

, et al. Physical activity is associated with neuromuscular and physical function in patients with multiple sclerosis independent of disease severity. Disabil Rehabil. 2021;43(5):632-639.

Bae

VanNostrand

Burt

Kasser

SL.

Contribution of dual-tasking to social participation in persons with multiple sclerosis: a structural equation modeling approach. Mult Scler Relat Disord. 2024;90:105835.

10.

Kohn

Coleman

Michael White

Sidovar

Sobieraj

DM.

Mobility, walking and physical activity in persons with multiple sclerosis. Curr Med Res Opin. 2014;30(9):1857-1862.

11.

Sandroff

Pilutti

Dlugonski

Motl

RW.

Physical activity and information processing speed in persons with multiple sclerosis: a prospective study. Mental Health Phys Act. 2013;6(3):205-211.

12.

Sandroff

Dlugonski

Pilutti

Pula

Benedict

Motl

RW.

Physical activity is associated with cognitive processing speed in persons with multiple sclerosis. Mult Scler Relat Disord. 2014;3(1):123-128.

13.

Motl

Sandroff

Benedict

RHB

. Moderate-to-vigorous physical activity is associated with processing speed, but not learning and memory, in cognitively impaired persons with multiple sclerosis. Mult Scler Relat Disord. 2022;63:103833.

14.

Sandroff

Motl

Amato

, et al. Cardiorespiratory fitness and free-living physical activity are not associated with cognition in persons with progressive multiple sclerosis: baseline analyses from the CogEx study. Mult Scler. 2022;28(7):1091-1100.

15.

Manglani

Phansikar

Duraney

, et al. Accelerometry measures of physical activity and sedentary behavior: associations with cognitive functioning in MS. Mult Scler Relat Disord. 2023;79:104963.

16.

Hubbard

Motl

RW.

Sedentary behavior is associated with disability status and walking performance, but not cognitive function, in multiple sclerosis. Appl Physiol Nutr Metab. 2015;40(2):203-206.

17.

Allali

Laidet

Assal

, et al. Adapted timed up and go: a rapid clinical test to assess gait and cognition in multiple sclerosis. Eur Neurol. 2012;67(2):116-120.

18.

Kalron

Aloni

Dolev

Frid

Givon

Menascu

The relationship between gait variability and cognitive functions differs between fallers and non-fallers in MS. J Neural Transm (Vienna). 2018;125(6):945-952.

19.

Shema-Shiratzky

Hillel

Mirelman

, et al. A wearable sensor identifies alterations in community ambulation in multiple sclerosis: contributors to real-world gait quality and physical activity. J Neurol. 2020;267(7):1912-1921.

20.

Frechette

Meyer

Tulipani

Gurchiek

McGinnis

Sosnoff

JJ.

Next steps in wearable technology and community ambulation in multiple sclerosis. Curr Neurol Neurosci Rep. 2019;19(10):80.

21.

VanNostrand

Bae

Ramsdell

Kasser

SL.

Information processing speed and disease severity predict real-world ambulation in persons with multiple sclerosis. Gait Posture. 2024;111:99-104.

22.

Hartley

Monaghan

Donoghue

Kenny

Romero-Ortuno

Exploring bi-directional temporal associations between timed-up-and-go and cognitive domains in the Irish longitudinal study on ageing (TILDA). Arch Gerontol Geriatr. 2022;99:104611.

23.

Rodriguez-Ayllon

Neumann

Hofman

Vernooij

Neitzel

The bidirectional relationship between brain structure and physical activity: a longitudinal analysis in the UK Biobank. Neurobiol Aging. 2024;138:1-9.

24.

Morrison

Mayer

Physical activity and cognitive function in adults with multiple sclerosis: an integrative review. Disabil Rehabil. 2017;39(19):1909-1920.

25.

Lenne

Donze

Massot

Degraeve

. Impact of physical activity, physical fitness and exercises on cognitive impairment in patients with multiple sclerosis: A review of evidence and underlying mechanisms. Rev Neurol (Paris). 2024;180(7):583-598.

26.

Page

McKenzie

Bossuyt

, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71.

27.

Meyer-Moock

Feng

Maeurer

Dippel

Kohlmann

Systematic literature review and validity evaluation of the Expanded Disability Status Scale (EDSS) and the Multiple Sclerosis Functional Composite (MSFC) in patients with multiple sclerosis. BMC Neurol. 2014;14:58.

28.

Learmonth

Motl

Sandroff

Pula

Cadavid

Validation of patient determined disease steps (PDDS) scale scores in persons with multiple sclerosis. BMC Neurol. 2013;13:37.

29.

Rosti-Otajärvi

Ruutiainen

Huhtala

Hämäläinen

Relationship between subjective and objective cognitive performance in multiple sclerosis. Acta Neurol Scand. 2014;130(5):319-327.

30.

Kmet

Cook

Lee

RC.

Standard quality assessment criteria for evaluating primary research papers from a variety of fields. Alberta Heritage Found Med Res. 2004;13:31.

31.

Abou

Peters

Freire

Sosnoff

JJ.

Fear of falling and common symptoms of multiple sclerosis: physical function, cognition, fatigue, depression, and sleep: a systematic review. Mult Scler Relat Disord. 2024;84:105506.

32.

Schober

Boer

Schwarte

LA.

Correlation coefficients: appropriate use and interpretation. Anesth Analg. 2018;126(5):1763-1768.

33.

Sebastião

Sandroff

Learmonth

Motl

RW.

Validity of the timed up and go test as a measure of functional mobility in persons with multiple sclerosis. Arch Phys Med Rehabil. 2016;97(7):1072-1077.

34.

Takla

Chargo

Daugherty

Fritz

NE.

Cognitive Contributors of Backward Walking in Persons with Multiple Sclerosis. Mult Scler Int. 2023;2023:5582242.

35.

Benedict

Holtzer

Motl

, et al. Upper and lower extremity motor function and cognitive impairment in multiple sclerosis. J Int Neuropsychol Soc. 2011;17(4):643-653.

36.

Cutter

Baier

Rudick

, et al. Development of a multiple sclerosis functional composite as a clinical trial outcome measure. Brain. 1999;122 ( Pt 5):871-882.

37.

Højsgaard Chow

Schreiber

Magyari

, et al. Progressive multiple sclerosis, cognitive function, and quality of life. Brain Behav. 2018;8(2):e00875.

38.

Motl

Sandroff

Benedict

Hubbard

Pilutti

Sutton

BP.

Do subcortical gray matter volumes and aerobic capacity account for cognitive-motor coupling in multiple sclerosis?

Mult Scler. 2021;27(3):401-409.

39.

Sandroff

Motl

Sosnoff

Pula

JH.

Further validation of the Six-Spot Step Test as a measure of ambulation in multiple sclerosis. Gait Posture. 2015;41(1):222-227.

40.

Bollaert

Motl

RW.

Physical and cognitive functions, physical activity, and sedentary behavior in older adults with multiple sclerosis. J Geriatr Phys Ther. 2019;42(4):304-312.

41.

Ibrahim

Adler

Gaßner

Rothhammer

Kluge

Eskofier

BM.

Association between cognition and gait in multiple sclerosis: a smartphone-based longitudinal analysis. Int J Med Inform. 2023;177:105145.

42.

Motl

Gappmaier

Nelson

Benedict

RH.

Physical activity and cognitive function in multiple sclerosis. J Sport Exerc Psychol. 2011;33(5):734-741.

43.

Sandroff

Motl

RW.

Device-measured physical activity and cognitive processing speed impairment in a large sample of persons with multiple sclerosis. J Int Neuropsychol Soc. 2020;26(8):798-805.

44.

Zheng

Sandroff

Motl

. Free-living ambulatory physical activity and cognitive function in multiple sclerosis: the significance of step rate vs. step volume. J Neurol. 2024;271(4):1638-1648.

45.

Bollaert

Sandroff

Stine-Morrow

EAL

Sutton

Motl

RW.

The intersection of physical function, cognitive performance, aging, and multiple sclerosis: a cross-sectional comparative study. Cogn Behav Neurol. 2019;32(1):1-10.

46.

D’Orio

Foley

Armentano

Picone

Kim

Holtzer

Cognitive and motor functioning in patients with multiple sclerosis: neuropsychological predictors of walking speed and falls. J Neurol Sci. 2012;316(1-2):42-46.

47.

Hsieh

Sun

Sosnoff

JJ.

Cognition is associated with gait variability in individuals with multiple sclerosis. J Neural Transm. 2017;124:1503-1508.

48.

Bilgin

YOU

Koskderelioglu

Gedizlioglu

. Fall risk is related to cognitive functioning in ambulatory multiple sclerosis patients. Neurol Sci. 2023;44(9):3233-3242.

49.

Sandroff

Pilutti

Benedict

Motl

RW.

Association between physical fitness and cognitive function in multiple sclerosis: does disability status matter?

Neurorehabil Neural Repair. 2015;29(3):214-223.

50.

Giroux

Sogoloff

Bergmans

Bae

Vannostrand

Kasser

SL.

An exploratory study of community mobility in adults with multiple sclerosis across different ambulation levels. Int J MS Care. 2023;25(4):145-151.

51.

Benedict

Cookfair

Gavett

, et al. Validity of the minimal assessment of cognitive function in multiple sclerosis (MACFIMS). J Int Neuropsychol Soc. 2006;12(4):549-558.

52.

Audiffren

André

The exercise-cognition relationship: a virtuous circle. J Sport Health Sci. 2019;8(4):339-347.

53.

Netz

Is there a preferred mode of exercise for cognition enhancement in older age? A narrative review. Front Med (Lausanne). 2019;6:57.

54.

Sandroff

Klaren

Pilutti

Dlugonski

Benedict

Motl

RW.

Randomized controlled trial of physical activity, cognition, and walking in multiple sclerosis. J Neurol. 2014;261(2):363-372.

55.

Motl

McAuley

Symptom cluster as a predictor of physical activity in multiple sclerosis: preliminary evidence. J Pain Symptom Manage. 2009;38(2):270-280.

56.

Sandroff

Motl

Reed

Barbey

Benedict

DeLuca

Integrative CNS plasticity with exercise in MS: the PRIMERS (PRocessing, Integration of Multisensory Exercise-Related Stimuli) conceptual framework. Neurorehabil Neural Repair. 2018;32(10):847-862.

57.

Hernandez

Holtzer

Chaparro

, et al. Brain activation changes during locomotion in middle-aged to older adults with multiple sclerosis. J Neurol Sci. 2016;370:277-283.

58.

VanNostrand

Belanger

Purin

Kasser

Cannizzaro

Cognitive-motor interference and cortical activation while walking in individuals with multiple sclerosis. Motor Control. 2022;26(4):677-693.

59.

Giannouli

Bock

Zijlstra

Cognitive functioning is more closely related to real-life mobility than to laboratory-based mobility parameters. Eur J Ageing. 2018;15(1):57-65.

60.

Klaren

Hubbard

Zhu

Motl

RW.

Reliability of accelerometer scores for measuring sedentary and physical activity behaviors in persons with multiple sclerosis. Adapt Phys Activ Q. 2016;33(2):195-204.

61.

Meyer

Depetrillo

Franco

, et al. How much data is enough? A reliable methodology to examine long-term wearable data acquisition in gait and postural sway. Sensors. 2022;22(18):6982.

62.

Weiss

Herman

Giladi

Hausdorff

JM.

Objective assessment of fall risk in Parkinson’s disease using a body-fixed sensor worn for 3 days. PLoS ONE. 2014;9(5):e96675.

63.

Weiss

Sharifi

Plotnik

van Vugt

Giladi

Hausdorff

JM.

Toward automated, at-home assessment of mobility among patients with Parkinson disease, using a body-worn accelerometer. Neurorehabil neural Repair. 2011;25(9):810-818.

64.

Ibrahim

Küderle

Gaßner

Klucken

Eskofier

Kluge

Inertial sensor-based gait parameters reflect patient-reported fatigue in multiple sclerosis. J Neuroeng Rehabil. 2020;17(1):165.

65.

Kratz

Murphy

Braley

TJ.

Pain, fatigue, and cognitive symptoms are temporally associated within but not across days in multiple sclerosis. Arch Phys Med Rehabil. 2017;98(11):2151-2159.

66.

Kratz

Murphy

Braley

TJ.

Ecological momentary assessment of pain, fatigue, depressive, and cognitive symptoms reveals significant daily variability in multiple sclerosis. Arch Phys Med Rehabil. 2017;98(11):2142-2150.

67.

Kratz

Whibley

Kim

Williams

Clauw

Sliwinski

The role of environmental distractions in the experience of fibrofog in real-world settings. ACR Open Rheumatol. 2020;2(4):214-221.