Received: 12 April 2024 Revised: 25 July 2024 Accepted: 30 July 2024

DOI: 10.1111/jgs.19162
Journal of the
CLINICAL INVESTIGATION American Geriatrics Society

Gait performance in older adults across the cognitive

spectrum: Results from the GAIT cohort

Pauline Ali MD, MSc 1,2,3 | Pauline Renaud MD 3 |

Manuel Montero-Odasso MD, FRCPC, PhD, AGSF, FGSA 4,5,6 |
Jennifer Gautier MS 7,8 | Mickaël Dinomais MD, PhD 2,3 |
Cédric Annweiler MD, PhD 1,7,8
1
Robarts Research Institute, Department of Medical Biophysics, Schulich School of Medicine and Dentistry, Western University, London, Ontario,
Canada
2
Laboratoire Angevin de Recherche en Ingénierie des Systèmes, EA7315, University of Angers, Angers, France
3
Department of Physical and Rehabilitation Medicine, Angers University Hospital, University of Angers, Angers, France
4
Gait and Brain Lab, St. Joseph's Health Care, Parkwood Institute, Lawson Health Research Institute, University of Western Ontario, London,
Ontario, Canada
5
Department of Medicine, Division of Geriatric Medicine, Schulich School of Medicine and Dentistry, Western University, London, Ontario, Canada
6
Department of Epidemiology and Biostatistics, Schulich School of Medicine and Dentistry, Western University, London, Ontario, Canada
7
UNIV ANGERS, UPRES EA 4638, University of Angers, Angers, France
8
Department of Geriatric Medicine and Rehabilitation, Research Center on Autonomy and Longevity, University Hospital, Angers, France

Correspondence
Pauline Ali, CHU Angers, Service de Abstract
Médecine Physique et de Réadaptation, Background: Gait performance can provide valuable insights into cognitive
4 Rue Larrey, 49000 Angers, France
functioning in older adult and may be used to screen for cognitive impairment.
Email: [email protected]
However, the optimal test condition and spatiotemporal parameter for accu-
Funding information racy have not yet been determined. This study aims to determine the gait mea-
French Ministry of Health, Grant/Award
sure with the highest accuracy identifying cognitive decline.
Number: PHRC2009-A00533-54; French
Society of Physical and Rehabilitation Methods: A total of 711 participants were recruited, including 332 cognitively
Medicine (SOFMER) healthy individuals, 264 with mild cognitive impairment (MCI), and 115 with
dementia, with a mean age of 72 years (interquartile range 69–76), and 43%
(n = 307) of women. The participants underwent gait assessment in three dif-
ferent conditions, including a single task and dual tasks of counting backward
by ones and naming animals.
Results: Gait performance was deteriorated as cognitive impairment pro-
gressed. The gait test performed during naming animals condition was the
most accurate in differentiating between cognitive groups. Specifically, the
naming animals gait speed was more accurate in discriminating control partici-
pants from those with cognitive impairment (area under the curve
[AUC] = 76.9% for MCI and 99.7% for people with dementia with control
group as reference). The coefficient of stride length variability while naming

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and reproduction in any
medium, provided the original work is properly cited and is not used for commercial purposes.
© 2024 The Author(s). Journal of the American Geriatrics Society published by Wiley Periodicals LLC on behalf of The American Geriatrics Society.

J Am Geriatr Soc. 2024;72:3437–3447. wileyonlinelibrary.com/journal/jgs 3437

 15325415, 2024, 11, Downloaded from https://agsjournals.onlinelibrary.wiley.com/doi/10.1111/jgs.19162 by Universitat De Valencia, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3438 ALI ET AL.

animals was the most effective parameter in discriminating between MCI and
dementia groups (AUC = 96.7%).
Conclusions: The naming animals dual-task gait test can be a valuable assess-
ment for screening cognitive impairment in older adults, regardless of their
cognitive abilities. The test is useful in clinical settings for subjects with a range
of cognitive profiles.

KEYWORDS
dual-task cognitive impairment, gait parameters, older adults, screening

INTRODUCTION
Key points
With the increasing aging of the population, dementia • Gait impairment is considered as a clinical
and falls have become major public health issues.1 It is marker of cognitive decline in older adults.
important to study gait and cognitive impairments con- However, it is unclear which gait modality to
currently, as they share common neural circuits and evaluate to best identify those individuals.
mechanisms.2 Gait impairment may serve as an early • Among the conditions and parameters investi-
marker of cognitive decline, often preceding cognitive gated on an electronic walkway, the naming
symptoms.3–5 Thus, gait analysis could be a practical animal gait test has the highest accuracy in
method for screening individuals with cognitive impair- identifying people with cognitive impairment.
ment.6 It is less costly, invasive, and time-consuming • The current study shows that not all the gait
than standard diagnostic strategies for improving the assessments are equally effective in identifying
diagnosis of dementia, such as brain imaging and biologi- cognitive decline and dementia. It is important
cal biomarkers. To achieve this, it is necessary to have a to highlight the most efficient assessments for
gait profiling according to the cognitive status of each clinicians and researchers to further investigate
group and evaluate the predictive ability of each gait the identified individuals.
parameter and condition.
Several gait parameters may be taken into account.6,7
Why does this paper matter?
Beyond the well-studied relationship between slower gait
speed and cognitive impairment, gait variability, which Data from the current study illustrate that the
reflects the stability component of gait, also provides naming animal gait test on an electronic walk-
insights into cognitive status.8 For instance, individuals way, which measures parameters of gait speed
with dementia exhibit higher coefficient of stride length and gait variability and is a rapid and inexpensive
variability (CoV SLV) compared with healthy controls, tool for screening cognitive decline and demen-
while those with mild cognitive impairment (MCI) show tia. It shows researchers and clinicians that this
increased coefficient of stride time variability (CoV STV) feature could easily be incorporated into the stan-
when compared with the same reference group.9,10 How- dard battery assessment for cognitive decline and
ever, these associations between gait performance and dementia in older adults.
cognitive impairment have not been always reported for
double support time and width.11
A single usual gait assessment can differentiate
between dementia and control groups. Moreover, adding cognitive resources, and the cognitive domain of the
a cognitive task during gait (i.e., dual-task gait) can refine task.15,16 A complex condition could produce a greater
the diagnosis and unmask latent gait impairment that is change in a cognitively altered group compared with a
only evident under cognitive stress, as seen in individuals control groups.17 Therefore, the challenge is to identify a
with MCI. It can also help identify MCI subtypes and task that is demanding enough to observe a clinical
subjects at higher risk of progressing to dementia.12–14 change but not overly complex to the point of losing
Several conditions (i.e., secondary cognitive tasks) can be information. Some tasks may be too easy and/or not tax
added to gait and impact it differently depending on the the same resources as the systems that control gait,
characteristics of the subjects, such as their age, available whereas others, such as counting backward by sevens,
 15325415, 2024, 11, Downloaded from https://agsjournals.onlinelibrary.wiley.com/doi/10.1111/jgs.19162 by Universitat De Valencia, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
THE SIGNIFICANCE OF DUAL TASK GAIT 3439

may be too difficult and cause subjects with dementia to Clinical assessment
rush through the cognitive task and inadvertently speed
up their gait.15,18,19 Studies that evaluate gait parameters Cognitive assessment
in older adults with different cognitive statuses have been
limited by small sample sizes13,20,21 or have only consid- All participants were evaluated at the Memory Clinic of
ered a single condition, such as single or dual task, or a the University Hospital of Angers (France). Diagnoses
single spatiotemporal characteristic.9,10,22,23 Furthermore, of cognitive status were made during multidisciplinary
it is unclear whether the observed gait characteristics are assessments involving geriatricians, neurologists, and
physiological or associated with a neurodegenerative pro- neuropsychologists, blinded from the gait performance.
cess. It is worth investigating whether the various condi- Participants with normal neuropsychological results
tions have a similar impact on gait parameters in both (MMSE26 score >25 and negative Dubois and Winblad
normal aging and cognitive impairment.20 criteria) were considered as controls. Dementia was diag-
The aim of this study is to determine the most effec- nosed according to the Diagnostic and Statistical Manual
tive gait parameter and condition for distinguishing of Mental Disorder DSM-IV and NINCDS/ADRDA cri-
between several cognitive groups. The main hypothesis is teria.27,28 MCI was diagnosed according to the criteria of
that adding a cognitive task during gait would better Dubois and Winblad et al., which include
characterize the different groups compared with a single the following29: (i) the person is neither normal nor
gait assessment. demented; (ii) there is evidence of cognitive decline, as
evidenced by either objectively measured decline over
time and/or subjective report of decline by self and/or
METHODS informant in conjunction with objective cognitive defi-
cits8; and (iii) activities of daily living are preserved and
Study population complex instrumental functions are either intact or mini-
mally impaired.30
From the “Gait and Alzheimer Interactions Tracking”
(GAIT) prospective cohort conducted at the University
Hospital of Angers from November 2009 to November Gait assessment
2015, participants were recruited from the “Centre
Mémoire Recherche Ressources” (CMRR) or during a Gait assessment was performed using a Gaitrite© system
routine assessment in the cognitive impairment screen- (972-cm long) to evaluate spatiotemporal gait parameters
ing centre of Angers. The study procedure has been according to the European guidelines for spatiotemporal
described in detail previously.24,25 Participants were gait analysis in older adults.7 Principal measures of inter-
required to be 60 years of age or older, community- est were gait speed (comfortable in cm/s), coefficient of
dwelling, and ambulatory and to have an adequate stride time (CoV STV), and coefficient of step length
understanding of French language. Additionally, they variability (CoV SLV). Participants were asked to walk
were required to have a mini-mental state examination at their usual walking speed in their own shoes. For
(MMSE) score of at least >10, visual acuity of at least the dual-task trials, participants walked at their usual
≥2/10, no severe depressive symptoms (15-item Geriat- speed without instructions to prioritize the gait or
ric Depression Scale score ≤ 10), and no acute medical cognitive task while performing the following cognitive
illnesses in the past month. Individuals were excluded tasks aloud: (i) counting backward (CB) from 50 to
from participation if they had preexisting locomotor dis- 0 one at a time and (ii) naming animals (NA). The
orders, a history of stroke or sensorimotor sequelae from GAITRite® software automatically calculated mean values
the central nervous system, any acute medical or surgi- and coefficient of variation (CoV = [SD/mean]  100).14
cal condition less than 3 months old, a score of greater Coefficient of stride time variability (CoV STV) is defined
than 10 on the 15-item Geriatric Depression Scale, no as the measure of the variation of the time elapsed
dual-task gait assessment, or if they were unable to walk between the first contact of two consecutive footfalls of
unaided for less than 15 min (eFigure 1 in Supplemen- the same foot expressed as a percentage. Coefficient
tary material). of stride length variability (CoV SLV) is the variation of
Participants were enrolled after giving written the distance between two consecutive footfalls of the
informed consent to participate in the study. The Angers same foot. Dual-task cost (DTC) of gait speed (%) was cal-
Ethics Committee approved the research protocol (CPP culated as ([simple-task gait value dual-task gait
Ouest II—2009-12). value]/simple-task gait value)  100.16
 15325415, 2024, 11, Downloaded from https://agsjournals.onlinelibrary.wiley.com/doi/10.1111/jgs.19162 by Universitat De Valencia, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3440 ALI ET AL.

Statistics Alzheimer's disease (AD), 4 had non-AD neurodegenera-

tive disease, and 15 had mixed vascular dementia (see
To describe the population, Chi-square tests were used eFigure 1).
for nominal variables and the Kruskal–Wallis test was
utilized for group comparisons, followed by the Mann–
Whitney U test for pairwise comparisons of continuous Participant characteristics
variables, contingent upon where the non-Gaussian
nature of the data. Participant characteristics stratified by cognitive status
A two-way repeated measures analysis of variance are summarized in Table 1.
(ANOVA) or a Kruskal–Wallis test were used depending Individuals with dementia were older, more likely to
on the distribution of the variables to assess the main and be women, and less educated than individuals with MCI
interaction effects between status group (controls, MCI and to the controls ( p < 0.001). In addition, the dementia
and those with dementia) and gait parameters. Student's group had a higher history of falls in the past 6 months,
t-test or Mann–Whitney–Wilcoxon tests were then weaker grip hand strength, and more comorbidities
applied when applicable. including psychotropic prescription (p < 0.001). Individ-
Logistic regression analyses were performed to eval- uals with MCI had a higher body mass index compared
uate the association between gait variables and cogni- with the control group ( p = 0.03). As expected, cognitive
tive groups, including covariates such as age, sex, test scores were lower in the dementia group compared
education level (dichotomized into post-secondary edu- with the others and lower in the MCI group
cation and no post-secondary education), MMSE26 compared with the control group (p < 0.001).
(global cognition), number of comorbidities, and models
with the lowest Akaike information criterion (AIC)
were selected. Association between gait performance and
Receiver operating characteristic (ROC) curves were the cognitive status
used to demonstrate the ability of each parameter to
accurately classify healthy controls and individuals The two-way ANOVA comparing groups by cognitive sta-
with MCI and individuals with dementia. The area tus was significant for total gait performance in all condi-
under the curve (AUC) was calculated as an indicator tions. Post hoc analysis showed that people with
of the overall accuracy of the parameter, with an AUC dementia had slower gait speed and higher variability
>70% considered acceptable (moderate accuracy) and coefficients (CoV STV and CoV SLV) than the other two
>90% considered excellent (high accuracy).31 To groups (Figure 1, Table 2). Similarly, individuals with
enhance the clinical applicability of the findings, we MCI were worse than the control group in all gait tests
established the optimal cut-off values for the selected (Figure 1, Table 2). In addition, the DTC (in all condi-
gait parameters in distinguishing the control group tions) was higher (i.e., worse) in the dementia group
from the cognitive impairment groups. The sensitivity, compared with the others and worse in the MCI
specificity, and Youden Index for each selected gait group compared with the controls.
parameter value were calculated using ROC curve anal- Gait speed remained above the fall risk threshold of
ysis, and the cut-off value that maximized the Youden 100 cm/s in each condition for the control group. People
Index was identified.32 with dementia were below this threshold in all conditions
For each test performed, the significance threshold (Table 2). Individuals with MCI remained above for the
was set at p < 0.05. All analyses were performed using single gait and CB task but dropped below the threshold
SAS software, version 9.4 (SAS Institute Inc.). during the NA gait task (Table 2).
The association between the cognitive status and the
gait parameters, adjusted for covariates is shown in
RESULTS Table 3. During single task, the gait speed and the CoV
SLV were different between the dementia and the control
The GAIT cohort recruited 912 participants. After apply- groups ( p = 0.02 and p = 0.02, respectively) and CoV
ing exclusion criteria, 711 participants were included in STV was different between the MCI and the control
the current study. Of these, 332 were healthy controls, groups (p = 0.046). No other differences in gait parame-
264 had criteria for MCI, and 115 had criteria for demen- ters were found between the groups.
tia. Among the MCI group, 62 had amnestic pure features In the CB condition, the CB gait speed differentiated
and 202 had non amnestic features. In the dementia the cognitively impaired groups (MCI and dementia)
group, 96 individuals were diagnosed with probable from the control group (p < 0.001 for MCI and p = 0.02
 15325415, 2024, 11, Downloaded from https://agsjournals.onlinelibrary.wiley.com/doi/10.1111/jgs.19162 by Universitat De Valencia, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
THE SIGNIFICANCE OF DUAL TASK GAIT 3441

TABLE 1 Participant characteristics according to their cognitive status.

Post hoc analysis

MCI Dementia Dementia

Controls Dementia versus versus versus
Variable Total (n = 711) (n = 332) MCI (n = 264) (n = 115) p-Value controls controls MCI
Age in years, 71.9 [68.8–76.1] 70.2 [68.1–73.3] 72.0 [69.1–75.9] 82.2 [77.5–84.9] <0.001 <0.001 <0.001 <0.001
median [IQR]
Women, n (%) 307 (43.2) 133 (40.1) 99 (37.5) 75 (65.2) <0.001 0.530 <0.001 <0.001
High school 404 (56.8) 227 (68.4) 141 (53.4) 36 (31.3) <0.001 <0.001 <0.001 <0.001
educated, n (%)
BMI (kg/m2), 25.7 [23.5–28.3] 25.3 [23.2–28.1] 26.1 [24.1–28.7] 26.0 [22.9–28.4] 0.028 0.008 0.741 0.157
median [IQR]
Grip hand 30.3 [23.0–39.0] 33.3 [25.7–40.3] 32.0 [25.0–39.0] 20.2 [15.7–26.0] <0.001 0.181 <0.001 <0.001
Strengtha (kg),
median [IQR]
History of fallsb, 88 (12.5) 37 (11.2) 23 (8.7) 28 (24.8) <0.001 0.322 <0.001 <0.001
n (%)
Medications/ 2 [1–5] 2 [1–4] 3 [1–5] 5 [2–8] <0.001 <0.001 <0.001 <0.001
daysc, median
[IQR]
Comorbidities, 2 [1–3] 2 [1–3] 2 [1–3] 3 [2–5] <0.001 <0.001 <0.001 <0.001
median [IQR]
Psychotropic 135 (19.0) 42 (12.7) 48 (18.2) 45 (39.1) <0.001 0.064 <0.001 <0.001
Drugs Use, n (%)
MMSE, median 28 [26–29] 29 [28–30] 27 [26–28.5] 21 [18–23] <0.001 <0.001 <0.001 <0.001
[IQR]
FABd, median 16 [14–17] 17 [16–18] 16 [14–17] 12 [9–14] <0.001 <0.001 <0.001 <0.001
[IQR]

Note: Where appropriate the mean or median is shown with standard deviation in parentheses. Comparison based on Chi-squared or Kruskal–Wallis as
appropriate with Chi-squared or Mann–Whitney post hoc test and Benjamin-Hochberg correction as appropriate. History of falls refers to the past 6 months.
p-Value significant in bold (i.e., <0.05).
Abbreviations: BMI, body mass index; FAB, frontal assessment battery (scores range from 0 to 18, higher scores representing better function); IQR,
interquartile range; MCI, mild cognitive impairment; MMSE, mini-mental state examination (scores range from 0 to 30, higher scores representing better
function); n, number of participants.
a
Missing data: n = 41.
b
Missing data: n = 4.
c
Missing data: n = 6.
d
Missing data: n = 8.

for dementia groups). The variability only differentiated Classifying accuracy of gait parameters to
the MCI from the control individuals. identify each cognitive group (ROC
In the NA condition, the NA gait speed was different analysis)
in all the three groups (significant differences between
the dementia group versus the controls group ( p = 0.01) NA gait speed was the best parameter to discriminate
and versus the MCI group (p = 0.041), and individuals participants with cognitive impairment from those with-
with MCI versus controls, p < 0.001). NA CoV STV dis- out, with a good accuracy to identify individuals
criminated individuals with MCI from the control and with MCI (AUC = 76.9%, specificity = 70.8%, sensitivity =
dementia groups ( p < 0.001 and p = 0.03 respectively). 67.8%, and cut-off score = NA gait speed 90 cm/s) and
NA CoV SLV discriminated individuals with dementia an excellent accuracy to identify people with dementia
from the control and MCI groups (p = 0.046 and (AUC = 99.7%, specificity = 96.1%, sensitivity = 95.7%)
p = 0.03 respectively). (Figure 2). When MCI and dementia were grouped
In both conditions, the DTC of gait speed differenti- together (i.e., cognitive impairment group), the AUC was
ated participants with MCI from controls but not the 84.6%, the specificity was 79.7%, and the sensitivity was
other groups from each other ( p = 0.004 for CB condition 72.8%, and the optimal cut-off score was mean NA gait
and p < 0.001 for the NA condition). speed 88 cm/s.
 15325415, 2024, 11, Downloaded from https://agsjournals.onlinelibrary.wiley.com/doi/10.1111/jgs.19162 by Universitat De Valencia, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3442 ALI ET AL.

F I G U R E 1 Gait performances
stratified by cognitive status. Bar plots
are used to represent the mean and
standard deviation. CB, counting
backward by ones; MCI, mild cognitive
impairment; NA, naming animals; ST,
simple task gait.

The NA CoV SLV was the best parameter for a slower gait speed when exposed to sufficient cognitive
identifying MCI from the dementia participants (AUC = load (i.e., dual-task gait).
96.7%, specificity = 90.2%, sensitivity = 87.8%, cut-off The dual-task gait assessment, specifically when
score = mean NA CoV SLV = 2.36%). NA, has been identified as an effective method for
screening participants with cognitive impairment. This
aligns with existing research highlighting the useful-
DISC USS I ON ness of dual-task gait in screening for cognitive defi-
cits.13,21,33,34 The mechanism involves the central
The aim of this study was to investigate which gait nervous system, which commands both cognitive func-
parameters best distinguish between individuals with tions and gait control. The temporal lobe, responsible
cognitive impairment (either MCI or mild to moderate for memory and spatiotemporal inputs, and the pre-
dementia) and those without. The results showed that frontal cortex, responsible for executive functions and
gait while NA was the most effective test in distinguish- motor planning, are example of neural circuits being
ing between the groups. Finally, participants with cogni- used by both gait and cognitive tasks.35–37 In individ-
tive impairment (MCI and dementia group) experienced uals with cognitive impairment affecting these areas,
 15325415, 2024, 11, Downloaded from https://agsjournals.onlinelibrary.wiley.com/doi/10.1111/jgs.19162 by Universitat De Valencia, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
THE SIGNIFICANCE OF DUAL TASK GAIT 3443

TABLE 2 Comparison between gait parameters and groups according to their cognitive status.

Post hoc analysisb

Controls MCI Dementia MCI Dementia Dementia

Total group group group p-Value versus versus versus
(n = 711) (n = 332) (n = 264) (n = 115) overalla controls Controls MCI
Single-task
Gait speed (s) 104.9 ± 23.7 113.5 ± 18.9 105.4 ± 21.6 79.3 ± 22.7 <0.001 <0.001 <0.001 <0.001
(mean ± SD)
CoV STV (%) 3.2 ± 7.0 2.5 ± 1.5 2.9 ± 1.7 6.0 ± 16.7 <0.001 <0.001 <0.001 <0.001
CoV SLV (%) 3.1 ± 2.3 2.5 ± 1.5 3.0 ± 1.5 5.2 ± 3.9 <0.001 <0.001 <0.001 <0.001
CB task
Gait speed (s) 103.5 ± 28.7 115.8 ± 22.7 102.5 ± 25.4 70.2 ± 24.1 <0.001 <0.001 <0.001 <0.001
CoV STV (%) 4.9 ± 6.5 3.4 ± 3.5 5.4 ± 7.6 8.3 ± 8.5 <0.001 <0.001 <0.001 <0.001
CoV SLV (%) 3.8 ± 3.1 2.8 ± 1.7 3.8 ± 2.8 6.8 ± 4.6 <0.001 <0.001 <0.001 <0.001
NA-task
Gait speed (s) 88.6 ± 29.1 101.3 ± 23.3 87.1 ± 26.4 55.2 ± 22.1 <0.001 <0.001 <0.001 <0.001
CoV STV (%) 10.3 ± 14.2 6.7 ± 8.7 11.1 ± 14.0 19.0 ± 21.8 <0.001 <0.001 <0.001 <0.001
CoV SLV (%) 6.3 ± 4.9 4.9 ± 3.5 6.0 ± 4.1 10.9 ± 7.0 <0.001 <0.001 <0.001 <0.001
DTC
CB 1.6 ± 16.2 2.5 ± 14.8 2.6 ± 15.3 11.1 ± 17.5 <0.001 <0.001 <0.001 <0.001
NA 16.0 ± 19.9 10.2 ± 17.2 17.2 ± 19.8 29.8 ± 20.1 <0.001 <0.001 <0.001 <0.001

Note: p-Value significant in bold (i.e., <0.05).

Abbreviations: CB, counting backward; CoV SLV, stride length variability coefficient; CoV STV, stride time variability coefficient; DTC, dual-task cost; MCI,
mild cognitive impairment; NA, naming animals; SD, standard deviation.
a
Analysis of variance or nonparametric Kruskal–Wallis test as appropriate.
b
Student t-test or nonparametric Mann–Whitney Wilcoxon as appropriate.

the need to manage with the remaining brain resources a new cut score of 88 cm/s for the NA gait speed, which
often leads to a compromise in either the cognitive or effectively distinguishes the “cognitive impairment
the gait task. Consequently, when individuals with MCI group” from the control group. To date, no cut score
or dementia are asked to walk while performing a cog- has been established for dual-task gait speed.12 Our
nitively demanding task, alterations in their gait param- findings provide a new evidence base that can be uti-
eters become manifest and may aid in the diagnostic lized in a clinical setting. Consequently, the NA dual-
process. The NA gait task is particularly revealing of task gait is the most effective method for identifying
such alterations, as it engages executive functions and subtle cognitive impairments and enhancing sensitivity
semantic memory, which are domains frequently to fall risk.41
impaired in MCI subjects.38 Furthermore, this task is The present study analyzed both gait speed and DTC
achievable even for subjects with dementia, making it a of gait speed. DTC assesses the impact of cognitive tasks
valuable tool to supplement the diagnostic for all geriat- on gait.13 Participants with MCI had elevated DTC in
ric populations.21 It presents a greater cognitive chal- both conditions compared with the control group, but
lenge than simple arithmetic tasks, such as CB, which there was no significant difference between the dementia
have a lower cognitive load.15,16 As a result, this condi- and control groups. MCI group may have attempted to
tion causes individuals with MCI to fall below the compensate for their deficits by exerting a high cognitive
100 cm/s threshold,39 even if they were above this effort, leading to a slower gait speed compared with the
threshold in single-task and CB conditions. The control group. In contrast, individuals with dementia
100 cm/s threshold has been identified as a relevant already demonstrate altered baseline gait performance,
cutpoint for the subsequent risk of major health-related even without an additional task. Therefore, the incorpo-
outcomes in older adults, including falls, hospitaliza- ration of a secondary task does not significantly change
tion, and even death.39,40 However, an investigation their overall performance, resulting in no significant high
into the clinical applicability of our parameters revealed DTC relative to the control group. Additionally, the
 15325415, 2024, 11, Downloaded from https://agsjournals.onlinelibrary.wiley.com/doi/10.1111/jgs.19162 by Universitat De Valencia, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3444 ALI ET AL.

TABLE 3 Multivariate analysis showing the association between gait performances and cognitive status.

MCI versus controls (n = 596) Dementia versus controls (n = 447) Dementia versus MCI (n = 379)

OR [95% CI] p-Value OR [95% CI] p-Value OR [95% CI] p-Value

Single-task
Gait speed 0.99 [0.98–1.00] 0.065 0.95 [0.91–0.99] 0.017 0.99 [0.97–1.01] 0.341
CoV STV 1.14 [1.002–1.30] 0.046 1.12 [0.75–1.70] 0.578 0.98 [0.91–1.05] 0.493
CoV SLV 1.10 [0.97–1.24] 0.130 1.34 [1.06–1.70] 0.015 1.20 [0.94–1.55] 0.142
CB task
Gait speed 0.98 [0.97–0.99] <0.001 0.96 [0.93–0.99] 0.018 0.99 [0.97–1.01] 0.153
CoV STV 1.06 [1.01–1.12] 0.028 1.05 [0.92–1.19] 0.478 0.98 [0.91–1.06] 0.597
CoV SLV 1.13 [1.02–1.26] 0.019 1.02 [0.85–1.23] 0.851 1.02 [0.90–1.15] 0.793
Dual-task cost 1.02 [1.01–1.03] 0.004 1.02 [0.98–1.05] 0.419 1.02 [0.99–1.04] 0.186
NA-task
Gait speed 0.98 [0.97–0.99] <0.001 0.95 [0.92–0.99] 0.007 0.98 [0.96–0.999] 0.041
CoV STV 1.04 [1.02–1.06] <0.001 1.04 [0.99–1.08] 0.121 1.02 [1.002–1.05] 0.033
CoV SLV 1.04 [0.99–1.10] 0.104 1.14 [1.002–1.30] 0.046 1.09 [1.01–1.18] 0.029
Dual-task cost 1.02 [1.01–1.03] <0.001 1.03 [0.99–1.07] 0.087 1.02 [1.00–1.04] 0.055

Note: Multivariate logistic regression model adjusted on age, sex, educational level, number of comorbidities, and mini-mental state examination. p-Value
significant in bold (i.e., <0.05).
Abbreviations: CB, counting backward; CI, confidence interval; CoV SLV, stride length variability coefficient; CoV STV, stride time variability coefficient; MCI,
mild cognitive impairment; NA, naming animals; OR, odds ratio.

current analysis did not provide detailed information on the brain, enhancing motor performance and optimizing
cognitive performance during dual-tasking.42 Those with gait coordination.
impaired cognition may score poorly, as indicated by a The analysis of spatiotemporal parameters provides
reduced number of results (number or animal named). In valuable insights. Gait speed indicates propulsion, whereas
fact, individuals with dementia might struggle with mul- variability coefficients reflect stability components.10 Gait
titasking and may be better perform tasks sequentially. speed is more effective in distinguishing between cognitive
Additionally, our study included vascular, mixed, and task and non-task conditions, whereas variability differen-
AD in the dementia group. Subjects with vascular tiates between subjects with MCI and those with demen-
dementia have a tendency to present more severe gait tia. Previous studies have shown that altered variability
impairments.9,43 However, the relationship between neu- distinguishes subjects with AD from other groups.10,45 A
rodegenerative and vascular pathologies in gait alteration trajectory of gait disturbances may begin with an impair-
is not yet fully understood, nor is the extent of vascular ment in the propulsion component. MCI subjects compen-
dementia in patients with AD.44 This uncertainty influ- sate with a partially preserved stability component.
enced our decision to analyze these patients together as a Subsequently, as neurodegeneration progresses and alters
group. motor planning and coordination, stability is affected. CoV
The CB task, which is more routine and automatic, SLV alteration increases the risk of falls and leads to signif-
appears to facilitate gait in control subjects compared icant participation restrictions and activity limitations.8
with the other trials, in line with Gillain et al.33 This CoV STV did not emerge as a parameter with strong diag-
effect can be attributed to several interrelated mecha- nostic discriminatory power. This may be because the con-
nisms, possibly due to an overall increase in arousal or trol of stride time involves lower-order circuits
alertness. The rhythmic nature of counting can provide a (i.e., brainstem and spinal cord) which remain relatively
regular beat that aids in coordinating movement during preserved in participants with high-order system lesions.
gait, potentially faster gait speed. This effect may also be Conversely, it is presumed that gait speed and stride
related to positive distraction, where the cognitive task length are controlled by the cortico-basal ganglia circuit
moderately diverts attention away from potentially inhib- through the thalamus, whereas cadence is governed by
itory aspects of gait, such as self-consciousness, allowing mechanisms in the brainstem and spinal cord.46
for a more fluid and rapid gait speed. Furthermore, the This approach makes the gait assessment practical and
CB task may activate supplementary motor pathways in accessible. The robustness of the study is reinforced by its
 15325415, 2024, 11, Downloaded from https://agsjournals.onlinelibrary.wiley.com/doi/10.1111/jgs.19162 by Universitat De Valencia, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
THE SIGNIFICANCE OF DUAL TASK GAIT 3445

F I G U R E 2 Receiver operating characteristic (ROC) curve. Analysis of the predictive abilities of gait parameters. Naming animals
(NA) gait speed accurately classified cognitive impairment groups from control groups and NA stride length variability (SLV) accurately
classified dementia from mild cognitive impairment (MCI) groups. AUC, area under the curve.

large and well-characterized cohort, and the use of stan- NA condition which is the most effective to differentiate
dardized tests ensures the reliability and validity of the between groups and which can be feasible for many sub-
findings. However, the study has some limitations due to jects even with reduced cognitive resources. This gait
its cross-sectional nature, which limits our ability to draw assessment can be easily integrated into routine examina-
causal inferences. The participants recruited from a mem- tions for the early detection of cognitive decline unmask-
ory clinic may not be representative of the general popula- ing latent cognitive impairment. Future research should
tion. It would also be interesting to analyze the cognitive aim to conduct longitudinal studies to assess the contri-
load by incorporating the number of errors. bution of NA gait performance to predict the cognitive
decline.

C O N C L U S IO N AUTHOR CONTRIBUTIONS
CA was responsible for data acquisition. PA, PR and JG
This study significantly contributes to the understanding were responsible for data analysis; this was overseen by
of how different levels of cognitive impairment are CA, MD, and MM-O. PA and PR participated equally in
reflected in different gait profiles. It underscores the clini- drafting the manuscript. All authors contributed to
cal relevance of analysis dual-task gait, particularly in the reviewing and editing the manuscript.
 15325415, 2024, 11, Downloaded from https://agsjournals.onlinelibrary.wiley.com/doi/10.1111/jgs.19162 by Universitat De Valencia, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
3446 ALI ET AL.

A C K N O WL E D G M E N T S cognitively healthy individuals: results from a cross-sectional

The authors acknowledge the following teams for their pilot study. J Alzheimers Dis. 2016;55(3):965-971. doi:10.
contributions to this article: The staff from Angers Uni- 3233/JAD-160604
9. Allali G, Annweiler C, Blumen HM, et al. Gait phenotype from
versity Memory Clinic, France, for daily assistance. There
mild cognitive impairment to moderate dementia: results
was no compensation for this contribution. from the GOOD initiative. Eur J Neurol. 2016;23(3):527-541.
doi:10.1111/ene.12882
CONFLICT OF INTEREST STATEMENT 10. Beauchet O, Blumen HM, Callisaya ML, et al. Spatiotemporal
The authors declare no conflicts of interest. gait characteristics associated with cognitive impairment: a
multicenter cross-sectional study, the intercontinental “Gait,
SPONSOR'S ROLE cOgnitiOn & Decline” initiative. Curr Alzheimer Res. 2018;
15(3):273-282.
The first author was supported by a grant from the 11. Rosano C, Brach J, Longstreth WT Jr, Newman AB. Quantita-
French Society of Physical and Rehabilitation Medicine tive measures of gait characteristics indicate prevalence of
(SOFMER). The study was financially supported by the underlying subclinical structural brain abnormalities in high-
French Ministry of Health (Projet Hospitalier de functioning older adults. Neuroepidemiology. 2005;26(1):52-60.
Recherche Clinique national no. 2009-A00533-54). The doi:10.1159/000089240
authors thank this institution for its support. This study 12. Montero-Odasso MM, Sarquis-Adamson Y, Speechley M, et al.
has been registered in ClinicalTrials no. NCT01315717. Association of dual-task gait with incident dementia in mild
cognitive impairment: results from the gait and brain study.
JAMA Neurol. 2017;74(7):857-865. doi:10.1001/jamaneurol.
I N F O R M E D C O N S EN T 2017.0643
All participants provided informed consent. 13. Muir SW, Speechley M, Wells J, Borrie M, Gopaul K, Montero-
Odasso M. Gait assessment in mild cognitive impairment and
ORCID Alzheimer's disease: the effect of dual-task challenges across
Pauline Ali https://orcid.org/0000-0002-7652-2548 the cognitive spectrum. Gait Posture. 2012;35(1):96-100. doi:10.
Manuel Montero-Odasso https://orcid.org/0000-0002- 1016/j.gaitpost.2011.08.014
14. Montero-Odasso M, Oteng-Amoako A, Speechley M, et al. The
5796-2361
motor signature of mild cognitive impairment: results from
the gait and brain study. J Gerontol A Biol Sci Med Sci. 2014;
R EF E RE N C E S 69(11):1415-1421. doi:10.1093/gerona/glu155
1. 2023 Alzheimer's disease facts and figures. Alzheimers Dement. 15. Beauchet O, Dubost V, Gonthier R, Kressig RW. Dual-task-
2023;19(4):1598-1695. doi:10.1002/alz.13016 related gait changes in transitionally frail older adults: the type
2. Montero-Odasso M, Verghese J, Beauchet O, Hausdorff JM. of the walking-associated cognitive task matters. Gerontology.
Gait and cognition: a complementary approach to understand- 2005;51(1):48-52. doi:10.1159/000081435
ing brain function and the risk of falling. J Am Geriatr Soc. 16. Montero-Odasso M, Muir SW, Speechley M. Dual-task com-
2012;60(11):2127-2136. doi:10.1111/j.1532-5415.2012.04209.x plexity affects gait in people with mild cognitive impairment:
3. Tian Q, Montero-Odasso M, Buchman AS, et al. Dual cognitive the interplay between gait variability, dual tasking, and risk of
and mobility impairments and future dementia – setting a falls. Arch Phys Med Rehabil. 2012;93(2):293-299. doi:10.1016/j.
research agenda. Alzheimers Dement. 2023;19(4):1579-1586. apmr.2011.08.026
doi:10.1002/alz.12905 17. Hunter SW, Divine A, Frengopoulos C, Montero OM. A frame-
4. Mielke MM, Roberts RO, Savica R, et al. Assessing the temporal work for secondary cognitive and motor tasks in dual-task gait
relationship between cognition and gait: slow gait predicts cog- testing in people with mild cognitive impairment. BMC Geriatr.
nitive decline in the Mayo Clinic Study of Aging. J Gerontol A 2018;18(1):202. doi:10.1186/s12877-018-0894-0
Biol Sci Med Sci. 2013;68(8):929-937. doi:10.1093/gerona/gls256 18. Montero-Odasso M, Camicioli R, Muir-Hunter SW. Dual-task
5. Montero-Odasso M, Speechley M, Muir-Hunter SW, et al. gait and incident dementia-a step forward, but not there yet-
Motor and cognitive trajectories before dementia: results from reply. JAMA Neurol. 2017;74(11):1380-1381. doi:10.1001/
gait and brain study. J Am Geriatr Soc. 2018;66(9):1676-1683. jamaneurol.2017.2880
doi:10.1111/jgs.15341 19. Åhman HB, Cedervall Y, Kilander L, et al. Dual-task tests
6. Montero-Odasso M, Almeida QJ, Bherer L, et al. Consensus on discriminate between dementia, mild cognitive impairment,
shared measures of mobility and cognition: from the Canadian subjective cognitive impairment, and healthy controls – a
Consortium on Neurodegeneration in Aging (CCNA). cross-sectional cohort study. BMC Geriatr. 2020;20(1):258. doi:
J Gerontol A Biol Sci Med Sci. 2019;74(6):897-909. doi:10.1093/ 10.1186/s12877-020-01645-1
gerona/gly148 20. Persad CC, Jones JL, Ashton-Miller JA, Alexander NB,
7. European GAITRite® Network Group, Kressig RW, Giordani B. Executive function and gait in older adults with
Beauchet O. Guidelines for clinical applications of spatio- cognitive impairment. J Gerontol A Biol Sci Med Sci. 2008;
temporal gait analysis in older adults. Aging Clin Exp Res. 2006; 63(12):1350-1355. doi:10.1093/gerona/63.12.1350
18(2):174-176. doi:10.1007/BF03327437 21. Oh C. Single-task or dual-task? Gait assessment as a potential
8. Beauchet O, Launay CP, Chabot J, Levinoff EJ, Allali G. diagnostic tool for Alzheimer's dementia. J Alzheimers Dis.
Subjective memory impairment and gait variability in 2021;84(3):1183-1192. doi:10.3233/JAD-210690
 15325415, 2024, 11, Downloaded from https://agsjournals.onlinelibrary.wiley.com/doi/10.1111/jgs.19162 by Universitat De Valencia, Wiley Online Library on [09/12/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
THE SIGNIFICANCE OF DUAL TASK GAIT 3447

22. Beauchet O, Allali G, Berrut G, Hommet C, Dubost V, Assal F. systematic review and meta-analysis. Neurosci Biobehav Rev.
Gait analysis in demented subjects: interests and perspectives. 2011;35(3):715-728. doi:10.1016/j.neubiorev.2010.08.008
Neuropsychiatr Dis Treat. 2008;4(1):155-160. 36. Yogev-Seligmann G, Hausdorff JM, Giladi N. The role of execu-
23. Park J, Lee HJ, Park JS, et al. Development of a gait feature– tive function and attention in gait. Mov Disord. 2008;23(3):329-
based model for classifying cognitive disorders using a single 342; quiz 472. doi:10.1002/mds.21720
wearable inertial sensor. Neurology. 2023;101(1):e12-e19. doi: 37. Rosso AL, Studenski SA, Chen WG, et al. Aging, the central
10.1212/WNL.0000000000207372 nervous system, and mobility. J Gerontol A Biol Sci Med Sci.
24. Beauchet O, Launay CP, Sekhon H, Montembeault M, Allali G. 2013;68(11):1379-1386. doi:10.1093/gerona/glt089
Association of hippocampal volume with gait variability in pre- 38. Petersen RC, Doody R, Kurz A, et al. Current concepts in mild
dementia and dementia stages of Alzheimer disease: results cognitive impairment. Arch Neurol. 2001;58(12):1985-1992. doi:
from a cross-sectional study. Exp Gerontol. 2019;115:55-61. doi: 10.1001/archneur.58.12.1985
10.1016/j.exger.2018.11.010 39. Cesari M, Kritchevsky SB, Penninx BWHJ, et al. Prognostic
25. Beauchet O, Allali G, Thiery S, Gautier J, Fantino B, value of usual gait speed in well-functioning older people–
Annweiler C. Association between high variability of gait speed results from the Health, Aging and Body Composition Study.
and mild cognitive impairment: a cross-sectional pilot study. J Am Geriatr Soc. 2005;53(10):1675-1680. doi:10.1111/j.1532-
J Am Geriatr Soc. 2011;59(10):1973-1974. doi:10.1111/j.1532- 5415.2005.53501.x
5415.2011.03610_9.x 40. Studenski S, Perera S, Patel K, et al. Gait speed and survival in
26. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”. A older adults. JAMA. 2011;305(1):50-58. doi:10.1001/jama.2010.
practical method for grading the cognitive state of patients for 1923
the clinician. J Psychiatr Res. 1975;12(3):189-198. doi:10.1016/ 41. Verghese J, Robbins M, Holtzer R, et al. Gait dysfunction
0022-3956(75)90026-6 in mild cognitive impairment syndromes. J Am Geriatr
27. Dubois B, Feldman HH, Jacova C, et al. Research criteria for Soc. 2008;56(7):1244-1251. doi:10.1111/j.1532-5415.2008.
the diagnosis of Alzheimer's disease: revising the NINCDS- 01758.x
ADRDA criteria. Lancet Neurol. 2007;6(8):734-746. doi:10.1016/ 42. Goh HT, Pearce M, Vas A. Task matters: an investigation on the
S1474-4422(07)70178-3 effect of different secondary tasks on dual-task gait in older adults.
28. Quinn BP. Diagnostic and statistical manual of mental disor- BMC Geriatr. 2021;21(1):510. doi:10.1186/s12877-021-02464-8
ders, fourth edition, primary care version. Prim Care Compan- 43. Allan LM, Ballard CG, Burn DJ, Kenny RA. Prevalence and
ion J Clin Psychiatry. 1999;1(2):54-55. severity of gait disorders in Alzheimer's and non-Alzheimer's
29. Winblad B, Palmer K, Kivipelto M, et al. Mild cognitive dementias. J Am Geriatr Soc. 2005;53(10):1681-1687. doi:10.
impairment–beyond controversies, towards a consensus: report 1111/j.1532-5415.2005.53552.x
of the International Working Group on Mild Cognitive Impair- 44. Hachinski V. Shifts in thinking about dementia. JAMA. 2008;
ment. J Intern Med. 2004;256(3):240-246. doi:10.1111/j.1365- 300(18):2172-2173. doi:10.1001/jama.2008.525
2796.2004.01380.x 45. Pieruccini-Faria F, Black SE, Masellis M, et al. Gait variability
30. Jutten RJ, Dicks E, Vermaat L, et al. Impairment in complex across neurodegenerative and cognitive disorders: results from
activities of daily living is related to neurodegeneration in Alz- the Canadian Consortium of Neurodegeneration in Aging
heimer's disease–specific regions. Neurobiol Aging. 2019;75:109- (CCNA) and the Gait and Brain Study. Alzheimers Dement.
116. doi:10.1016/j.neurobiolaging.2018.11.018 2021;17(8):1317-1328. doi:10.1002/alz.12298
31. Fischer JE, Bachmann LM, Jaeschke R. A readers' guide to the 46. Takakusaki K. Functional neuroanatomy for posture and gait
interpretation of diagnostic test properties: clinical example of control. J Mov Disord. 2017;10(1):1-17. doi:10.14802/jmd.
sepsis. Intensive Care Med. 2003;29(7):1043-1051. doi:10.1007/ 16062
s00134-003-1761-8
32. Akobeng AK. Understanding diagnostic tests 3: receiver operat-
ing characteristic curves. Acta Paediatr. 2007;96(5):644-647. SU PP O R TI N G I N F O RMA TI O N
doi:10.1111/j.1651-2227.2006.00178.x Additional supporting information can be found online
33. Gillain S, Warzee E, Lekeu F, et al. The value of instru- in the Supporting Information section at the end of this
mental gait analysis in elderly healthy, MCI or Alzhei- article.
mer's disease subjects and a comparison with other
eFigure 1. Flowchart of the study.
clinical tests used in single and dual-task conditions. Ann
Phys Rehabil Med. 2009;52(6):453-474. doi:10.1016/j.rehab.
2008.10.004 How to cite this article: Ali P, Renaud P,
34. Maquet D, Lekeu F, Warzee E, et al. Gait analysis in elderly
Montero-Odasso M, Gautier J, Dinomais M,
adult patients with mild cognitive impairment and patients
with mild Alzheimer's disease: simple versus dual task: a pre-
Annweiler C. Gait performance in older adults
liminary report. Clin Physiol Funct Imaging. 2010;30(1):51-56. across the cognitive spectrum: Results from the
doi:10.1111/j.1475-097X.2009.00903.x GAIT cohort. J Am Geriatr Soc. 2024;72(11):
35. Al-Yahya E, Dawes H, Smith L, Dennis A, Howells K, 3437‐3447. doi:10.1111/jgs.19162
Cockburn J. Cognitive motor interference while walking: a