Klotzbier et al.

European Review of Aging and Physical Activity (2021) 18:17

https://doi.org/10.1186/s11556-021-00271-z

RESEARCH ARTICLE Open Access

An interrater reliability study of gait

analysis systems with the dual task
paradigm in healthy young and older
adults
Thomas Jürgen Klotzbier1* , Bettina Wollesen2,3, Oliver Vogel2, Julian Rudisch4, Thomas Cordes2,
Thomas Jöllenbeck5,6 and Lutz Vogt7

Abstract
Background and aims: One reason for the controversial discussion of whether the dual task (DT) walking
paradigm has an added value for diagnosis in clinical conditions might be the use of different gait measurement
systems. Therefore, the purpose was 1) to detect DT effects of central gait parameters obtained from five different
gait analysis devices in young and old adults, 2) to assess the consistency of the measurement systems, and 3) to
determine if the absolut and proportional DT costs (DTC) are greater than the system-measurement error under ST.
Methods: Twelve old (72.2 ± 7.9y) and 14 young adults (28.3 ± 6.2y) walked a 14.7-m distance under ST and DT at a
self-selected gait velocity. Interrater reliability, precision of the measurement and sensitivity to change were
calculated under ST and DT.
Results: An age effect was observed in almost all gait parameters for the ST condition. For DT only differences for
stride length (p < .029, ɳ2p = .239) as well as single and double limb support (p = .036, ɳ2p = .227; p = .034,
ɳ2p = .218) remained. The measurement systems showed a lower absolute agreement compared to consistency
across all systems.
Conclusions: When reporting DT effects, the real changes in performance and random measurement errors should
always be accounted for. These findings have strong implications for interpreting DT effects.
Keywords: Cognitive-motor interference, Dual task walking, Older adults, Verbal fluency, Minimal detectable
change, Gait analysis

Introduction [2] proposed that there are two reasons why older adults
It is well accepted that walking outside of clinical set- (OA) show decreased performance in multiple-task con-
tings requires dual tasking (DT) or multiple-task per- dition compared to young adults (YA). First, usual
formance, where walking is combined with cognitive or physiological changes associated with ageing (decreased
motor tasks, for example crossing the street while read- muscle mass, visual acuity, changes in proprioception,
ing signs or observing traffic [1]. Paul, Ada and Canning the vestibular- and somatosensory system) as well as ac-
companying alterations (postural adjustments, atten-
* Correspondence: thomas.klotzbier@inspo.uni-stuttgart.de tional capacity, increased reaction time, etc.) could
1
Department of Sport and Exercise Science, University of Stuttgart, interfere with DT performance. Second, decrements in
Allmandring 28, 70569 Stuttgart, Germany physical activity at this age means that multiple-task
Full list of author information is available at the end of the article

© The Author(s). 2021, corrected publication 2022. Open Access This article is licensed under a Creative Commons Attribution
4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as
long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence,
and indicate if changes were made. The images or other third party material in this article are included in the article's Creative
Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative
Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need
to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/
licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/)
applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Klotzbier et al. European Review of Aging and Physical Activity (2021) 18:17 Page 2 of 12

performance may no longer be a prominent feature of width, single and double limb support phase, step length,
everyday activities [3, 4] and therefore, with the lack of gait cycle length, step duration and gait cycle duration;
practice, performance declines. Thus, for example, cf. [18]). These parameters can be classified into parame-
changes in the gait pattern while dual tasking can lead ters of rhythm (e.g., cadence, single and double support)
to injourious falls [5] or serious traffic accidents [6], if and pace (e.g., gait velocity and step length) [19]. How-
the attentional resources are not sufficient to process ever, different gait analysis systems are used in the vari-
environmental conditions (e.g., a car coming up to cross ous studies to measure these gait parameters, which
the road [7];). Dual tasking describes the simultaneous might limit direct comparability. Only few studies deal
processing of two tasks. In research and clinical settings, with the measurement accuracy of these systems in
the aim of DT paradigms is to calculate proportional comparison to established reference systems [20]. Also,
dual task costs (DTC) as evidence for the limitation of regarding the DT gait paradigm, recommendations on
the information processing system. When calculating methodical procedures regarding walking distance, walk-
DTC, performance in each task under DT condition is ing condition (self-selected gait velocity in a slow or fast
related to the respective performance under single task gait conditions), etc. are rarely provided [16]. Klotzbier
(ST) condition [8]. These proportional DTCs are and Schott [21] were able to show that especially walk-
expressed as a percentage decrease in performance com- ing with directional changes is sensitive to the produc-
pared to performance in the ST. The term DTC implies tion of DTC. Straight walking does not sufficiently
that under DT conditions there is an interfering inter- address real-life gait [16]. However, in most studies,
action and a deterioration in the processing of the indi- walking straight ahead is used as a motor task, as most
vidual tasks, i.e. ST. However, DTs do not always and in gait analysis systems are constrained to a straight walk-
all situations lead to performance declines compared to way due to the design of the system (e.g., pressure plates
STs, therefore the term “dual task effect” (DTE) or “cog- [GAITrite; Zebris] or LED photoelectric switches [Opto-
nitive-motor interference (CMI)” is more commonly Gait]). In addition, the algorithms for calculating the gait
used. We use the term DTC to emphasize the perform- parameters from the acceleration data of the inertial sen-
ance decline and DTE to compare the accuracy of differ- sors (GaitUp; MobilityLab) are explicitly and exclusively
ent measurement systems. designed for conditions with straight walking. The differ-
There are several models that try to explain age related ent studies use a range of walking conditions, and the
drecrements in DT performance [9, 10]. With the most measurement range is not always identical [22–24].
common resource-theoretical conception of the atten- With Zebris, for example, it is only possible to cover a
tion construct [11, 12], it can be assumed that especially range of two meters (OptoGait and GAITrite also have
in OA with reduced resources for cognitive and motor limits). Moreover, one must reflect the algorithms in in-
control, the challenges of dealing with DT are greater ertial sensor systems like the MobilityLab that only allow
[13]. Lindenberger et al. [14] were able to show that the the detection of so-called “steady-state” walking. The
DTC become larger with increasing age. Thus, the The question remains unanswered to what extent the
cognitive-motor DT gait paradigm can be used to detect common gait parameters of the different gait analysis
gait deficits that would otherwise remain hidden during systems agree despite different measuring principles
normal walking without additional tasks. Based on the or walking conditions (ST vs. DT conditions) [24].
assumed interrelation of motor- and cognitive function Overall, mobile gait analysis systems show excellent
for gait, the DT paradigm is used for diagnosis, preven- agreement for spatio-temporal variables (gait velocity,
tion and treatment of falls or cognitive impairment (e.g., cadence, gait cycle time, double step time) compared to
intervention measures) and there is a controversial dis- more elaborate “gold-standard” systems [25 – 27]. Less
cussion of whether such paradigms have an added value agreement has usually been observed for stance-, swing-
[15, 16]. The heterogeneity of the study results can be and double stance phase [25, 27– 29]. Although the
explained primarily by the choice of cognitive tasks. To first direct comparisons of GAITrite and OptoGait [30, 31]
address this problem, Al-Yahya and colleagues [17] have as well as GAITrite and MobilityLab [20, 32] resulted
published a task classification and were able to show in good agreement between systems, no data is available;
that mental tracking tasks in particular (internal disturb- neither comparisons of others systems, nor for a simul-
ing factors such as counting backwards or verbal fluency taneous data collection on all systems
tasks) cause significant DTCs on gait. This effect is em- It is crucial that the change of gait parameters from
phasized in old age and already impaired cognitive ST to DT is higher than the random measurement error
abilities. between the systems, especially when retrospectively
Regarding walking performance and gait kinematics, viewing the effect of DT on gait parameters in meta-
most studies currently focus on about eight gait parame- analyses or intervention studies. One way to differentiate
ters and their variability (gait velocity, cadence, step between real change and random measurement error is
Klotzbier et al. European Review of Aging and Physical Activity (2021) 18:17 Page 3 of 12

through the utilization of the standard error of measure- Table 1 Sampling characteristics of older adults (OA) and
ment (SEM) and the minimal detectable change (MDC). young adults (YA), including mean values (standard deviation)
Hence, in this study we compared five different gait and statistic analyses of the mean value differences
measurement systems regarding their reliability – in OA YA stat. Analyses
terms of agreement – in a DT paradigm thereby provid- (n = 12) (n = 14)
ing an indication of the minimum amount of the DT ef- Age (years) 72.2 ± 7.9 28.3 ± 6.2 F(1,24) = 252***,
fect that is necessary to be sure not to consider this as a ɳ2p = .913
measurement error. Sex (n female) 6 8 CHI2(1) = .133ns
Therefore, the purpose of the present study was 1) to Weight (kg) 74.0 ± .08 72.9 ± 14.2 F(1,24) = .045ns,
investigate the average DT effects in a cohort of YA and ɳ2p = .002
OA, (i.e., the change in gait parameters from ST to DT), Height (cm) 1.69 ± .08 1.79 ± .09 F(1,24) = 6.16*,
2) to compare the obtained gait parameters between the ɳ2p = .204
measurement systems in YA and OA under ST condi- BMI (kg/m2) 25.1 ± 3.11 22.6 ± 2.75 F(1,24) = 6.68*,
tion, and 3) to investigate if the DT effects are greater d = .218
than the measurement error of the systems measured Leg length left 83.9 ± 4.79 86.9 ± 6.18 F(1,24) = 1.08ns,
under ST condition. We assumed that DTCs influencing ɳ2p = .043
gait parameters would be particularly evident in OA Leg length right −.916 ± 4.67 86.7 ± 6.03 F(1,24) = 1.11ns,
ɳ2p = .044
and, that the comparison of the different gait measuring
systems indicate no systematic or random differences. Shoe size 41.8 ± 3.08 42.3 ± 3.09 F(1,24) = .161ns,
ɳ2p = .007
Furthermore, we predicted that the DTCs are greater
ns Not significant
than the average difference of the measurement systems, ***p < .001; *p < .05
otherwise the DT effect may be due to measurement
error. Experimental Setup and Procedure). In the PROCARE
multicenter study our group collected data with five dif-
ferent mobile gait measurement systems, according to
Methods different institutional resources (see study protocol [35]).
Participants The PROCARE project was conducted to develop a
A total of 26 participants were recruited. Community training intervention to increase mobility and psycho-
dwelling OA (n = 12) who participated in regular fall logical well-being of nursing home residents. To show
prevention programs and sports activities for senior citi- effects of the intervention on DT performance (as the
cens at the University of Hamburg and YA (sport stu- DT paradigm is also used in the PROCARE study), it is
dents, n = 14) were recruited for this study (see Table 1 necessary to secure that training effects exceed the
for group characteristics). All participants had normal or measurement error.
corrected-to-normal vision and no known neurologic or
orthopaedic disorder affecting their gait. The study, ap- Overground walking systems
proved by the local ethics of the University of Hamburg The OptoGait (Microtgate, Bolzano, Italy) system is an
(registration number 2020_2077; 12.2.2020), followed optoelectronic measurement system using parallel bars
the Declaration of Helsinki [33]. that are positioned on the ground with 0.6 m distance
(adjusted to the width of the Zebris plate) and has a
spatial and temporal resolution of 1.041 cm (distance be-
Measurement systems tween diodes) and 1 kHz respectively. Ground contacts
Five different commercially available systems for per- are measured when the photoelectronic bridge between
forming gait analysis in clinical and research settings an LED and photodiode is interrupted. We used an
were compared. We included systems that used external OptoGait system of 6 m length. The system showed a
hardware to measure ground contact either through high level of correlation with all spatio-temporal param-
pressure sensors or via optoelectronic devices (body-de- eters (ICCs: 0.79–0.95) [31]. The GAITrite (CIR Systems,
tached), as well as systems composed of inertial meas- New Jersey, USA) system is an electronic walkway con-
urement units (body-attached sensors). taining a matrix of pressure sensors. We used an 8.7 m
In analogy to the study procedure of Rudisch and col- GAITrite walkway with an active measuring range of
leagues [34] we used the same systems. Rudisch et al. 7.93 m × 0,75 m and with a spatial and temporal reso-
compared different outcomes for the overlapping gait lution of 1.27 cm (length/width of sensors) and 120 Hz
phases, during which all systems measured the same respectively. It is well accepted that the GAITrite mat
steps. The present study focused on the mean values exhibits excellent reliability for most temporal-spatial
and standard deviation for all walking conditions (see gait parameters in both YA (ICCs: 0.83–0.94) and OA
Klotzbier et al. European Review of Aging and Physical Activity (2021) 18:17 Page 4 of 12

(ICCs: 0.82–0.91) [36]. The Zebris (zebris Medical

GmbH, Isny, Germany) plantar pressure system is, like
GAITrite, an electronic walkway containing a matrix of
pressure sensors. We used a 2 m × 0.6 m Zebris walkway
with a spatial and temporal resolution of 0.85 cm
(length/width of sensors) and 100 Hz respectively. Reli-
ability was excellent for gait velocity, cadence, gait cycle
time, step and double step length (ICC: 0.93–0.99) and
poor for relative stance, swing and double stance phases
(ICC: 0.24–0.47) [27].

Body-attached inertial sensors Fig. 2 Measurement setup of the overground walking systems
The GaitUp is a six-channel inertial sensor system (Phy- Zebris, OptoGait and GAITrite
siolog. Lausanne, Switzerland), which is worn on each
foot. It is attached to the MobilityLab straps and posi- Zebris (2 m), as well as two mats (each of 2 m length)
tioned lateral to them on each foot (see Fig. 1). Data was positioned on both ends of the walkway with the same
recorded over 14.7 m operating at a sampling frequency height as GAITrite, to consider gait initiation and gait
of 128 Hz. Moderate to excellent agreement was shown termination on an even surface. The walkway (between
for temporal parameters (ICCs: 0.72–0.97) [29]. The OptoGait bars) was limited to a width of 0.6 m, corre-
MobilityLab (Opal from APDM Inc., Portland, USA) for sponding to the width of the Zebris system. Upon arrival
gait analysis consists of three inertial sensors that are bi- in the gym of the Institute of Sports Sciences at the Uni-
laterally attached to both feet with straps and to the fifth versity of Hamburg, where the study was conducted, the
lumbar vertebrae. The data recording lasted 14.7 m and participants were informed about the content of the
sampling at a frequency of 128 Hz. Compared to a tread- study and signed a declaration of consent. Afterwards
mill integrated force measuring plate, MobilityLab re- the height, weight, leg length (left and right) and shoe
vealed excellent (ICC: 0.99; range or CI were not size were measured. Then the acceleration sensors were
reported) agreement for gait velocity, cadence, gait cycle attached (see Fig. 1). Waiting at the starting position the
time and double stride length, but only moderate to test person was instructed and the walking conditions
weak (ICC: 0.50; range or CI were not reported) correla- were described.
tions for stance and swing phase [20].
Single task and dual task walking conditions
Experimental setup and procedure The subjects had to walk the 14.7 m distance (see Fig. 2)
Figure 2 shows the measurement setup of the systems, two times without a cognitive task in order to become
illustrating that the gait paths of the different systems familiar with the body-attached inertial sensors and an-
overlap only partially. The overall length of the walkway other two times (one trial in each condition) for data
setup was 14.7 m. This is the sum of the distances of the collection in ST (walking only) and DT condition (walk-
overground walking systems GAITrite (8.7 m) and ing plus verbal fluency task) in randomized order. In the
ST condition the participants were instructed to walk
through the walkway at a comfortable, self-selected gait
velocity, where in the DT condition the participants
should additionally name as many words with a pre-
defined letter (B, D, S or A in random order given just
before the start signal [35]) as they could think of. The
DT condition with an additional verbal fluency task was
performed after a short explanation. Participants were
allowed to name any word except for proper nouns
(such as Bernd or Berlin), numbers, or words that start
with the same sound but have a different ending, e.g.,
love, lover, lovers. These instructions were given and,
using a letter that was not assigned by randomization,
some examples (3–5 words) were given to ensure that
the task was understood. Gait parameters were recorded,
Fig. 1 Image showing the attachment of the inertial sensors of
and the number of words was counted while walking
GaitUp and MobilityLab
straight forward. After each trial, the participants were
Klotzbier et al. European Review of Aging and Physical Activity (2021) 18:17 Page 5 of 12

asked to stand still behind the 2 m mat (intended for the instruments the SEM/SEM% and MDC/MDC95% should
deceleration phase) for 5 s so that the accelerometers be as small as possible.
could transmit their data without interference. Partici- Power analysis (using G*Power3; a statistical power
pants were then asked to return to the starting position analysis program [44];) was conducted to estimate the
for the next trial. Data collection for the ST and DT necessary sample size. With a sample size of 16 in one
walking conditions was about 10 min. group, an ANOVA with repeated measures would have
80% power to detect the interaction effect size of 0.403
Statistical analysis at the 0.05 level of significance. To detect significant
The mean values and standard deviations of the gait pa- group differences (MANOVA between factor: OA vs.
rameters and the respective condition (ST and DT) were YA) of 0.5 or larger (p < 0.05) and a power of 80%,
considered, using individual data acquisition and analysis twelve participants in each group are necessary. With
software of the respective systems. Six outcome variables twelve OA and 14 YA we achieved a total number of 26.
were analyzed as they were recorded by every system: Data were analyzed using SPSS, version 25.0 (SPSS
velocity (m/s); cadence (steps/min); stride length ([m], Inc., Chicago, Illinois). To compare the different systems
distance either foot moves forward); single limb support an ANOVA with measurement repetition with the sys-
([%], time of only 1 foot supporting the body weight); tems as measurement repetition factor was calculated
double limb support phase ([%]; ground contact time for for each gait parameter. To calculate the differences be-
both feet); stance phase ([%], duration of ground contact tween YA and OA a 6 (gait parameter) × 2 (group)
[heel-strike to toe-off]). MANOVA was calculated for ST and DT. To calculate
Absolute motor DTCs were calculated as follows: the differences between ST and DT, post hoc analyses
(−(STperformance - DTperformance)), negative values were calculated for each individual parameter. The mean
indicate decreases from ST to DT. Proportional motor values of the five systems were used as the basis for the
DTCs were calculated as follows: [((DTperformance - calculation. There were no missing values. If the result
STperformence)/STperformance) *100] expressed in % of the ANOVAs was significant, post-hoc tests (Bonfer-
[8, 37]. Since we did not perform cognitive performance oni) were used to analyze which factor levels signifi-
under ST condition, it was not possible to calculate cog- cantly differed from each other (p values set to .05 [45].
nitive DTCs. Effect sizes for all ANOVAs were reported using the
The interrater reliability (ICC) for the comparison partial Eta2 (η2p).
within and between the various systems was calculated
with the ICC (two-way random model for absolute agree- Results
ment and for consistency), if an ICC of < 0.5 is bad, 0.5– Participants
0.75 is moderate, between 0.75 and 0.90 is good, and Table 1 shows the characteristics of the sample. The sex
greater than 0.90 is excellent [38]. Using the ICC, the distribution did not differ between the groups. With
standard error of measurement (SEM) and the minimal 1.79 m YA were significantly taller than OA (1.69 m) and
detectable change (MDC) can be calculated (as preferred had a significantly lower BMI (22.6 ± 2.75) compared to
statistics according to the COSMIN standards [39]. SEM OA (25.1 ± 3.11).
is an indicator of absolute reliability and precision of the
measurement in the same units as the original measure- Age-related differences in gait parameters under single
ment (SEM = SD × √1 – ICC). Measurement error was and dual task condition
expressed as a percentage of the mean, which was defined Overall, an average of 7.39 (SD = 5.75) steps were de-
as SEM% = (SEM / mean) × 100. A SEM% smaller than tected in YA under ST condition and across all systems
10% indicates excellent agreement or reliability [40]. The (GaitUp = 18.1; OptoGait = 5.14; GAITrite = 7.43; Zebris
MDC (MDC95 = SEM × 1.96 × √(2)) can be calculated, 1.07; MobilityLab = 5.17). Under DT, an average of 10.1
where 1.96 derives from the 95% confidence interval of no (SD = 4.97) steps was detected (GaitUp = 19.21; Opto-
change and √2 is included because two measurements are Gait = 5.64; GAITrite = 8.29; Zebris = 1.29; MobilityLab =
involved in measuring change (ST and DT). The MDC is 6.08). In OA, an average of 8.89 (SD = 6.10) steps could
interpreted as the smallest amount of change required to be detected in the ST (GaitUp = 20.3; OptoGait = 6.33;
designate a change as real and beyond the bounds of GAITrite = 8.83; Zebris = 2.25; MobilityLab = 6.71) and
measurement error [41–43], also referred to as the sensi- an average of 9 (SD = 6.04) steps was detected in the DT
tivity to change. Also, the MDC95 was expressed as a per- condition (GaitUp = 20.1; OptoGait = 6.92; GAITrite =
centage, which was defined as MDC95% = (MDC95 / 9.5; Zebris = 2.00; MobilityLab = 6.50).
mean) × 100. The mean is the average for all the param- The 6 × 2 MANOVA showed that YA and OA differ
eter values in ST [43]. While the ICC ranges from 1 to 0, under ST conditions in gait velocity, F(1,20) = 9.11, p =
with 1 being perfect and 0 being no correlation, for good .007, ɳ2p = .313, stride length, F(1,20) = 11.4, p = .003,
Klotzbier et al. European Review of Aging and Physical Activity (2021) 18:17 Page 6 of 12

ɳ2p = .364, single limb support, F(1,20) = 9.99, p = .019, Reliability and minimal detectable changes
ɳ2p = .268, double limb support, F(1,20) = 10.3, p = .004, The mean values of the different measurement systems
ɳ2p = .340, and in the stance phase, F(1,20) = 4.82, p = under ST condition were compared (cf. Table 2). The
.040, ɳ2p = 192. Most gait parameters deteriorated under relative and absolute reliability measures (ICCa; c, SEM,
DT conditions, while differences between YA and OA MDC95) are shown in Table 2. The absolut agreement
were found in stride length, F(1,20) = 5.66, p < .029, (ICCa) between the systems was poor to excellent for all
ɳ2p = .239, single limb support, F(1,20) = 4.12, p = .036, groups and parameters, with values between .255 and
ɳ2p = .22, double limb support, F(1,18) = 5.29, p = .034, .992 [47]. The phase parameters single limb support
ɳ2p = .227, and no significant differences in gait velocity, (.255–.310), double limb support (.272–.309) and stance
F(1,29) = 3.36, p = .082, ɳ2p = .144 (see Fig. 3). phase (−.448–.475) in particular showed poor absolute
Multiple comparisons revealed that differences in agreement between the systems. The consistency of
gait parameters between ST and DT in YA could only measurement across all systems (ICCc) was moderate to
be observed in stride length, p = .016. In OA, differ- excellent, with values between 0.708 and 0.993. The
ences were observed for single limb support, p = .014, SEM% was low in all conditions and groups (0.771–
and double limb support, p = .017. In all other gait 4.52%). In 100% of the observations a SEM% ≤ 10% was
parameters analyzed, no difference between ST and found. The SEM% varied between 1.09–4.52% for YA
DT were observed. and between 0.77–46.4% for OA. The MDC95% was

Fig. 3 Differences in gait parameters between young adults (YA) and older adults (OA) under ST and DT conditions for all gait
measurement systems
Table 2 Mean values and standard deviation for gait parameters under single task condition (Mean ± SD), and intra-class correlation (ICC), inter-trial reliability (SEM; SEM%) and
sensitivity to change (MDC95, MDC95%) for these gait parameters across measurement systems
Parameter GaitUp Opto-Gait GAIT-rite Zebris Mean Mobility Lab ICCc (95% CI) ICCa (95% CI) SEM MDC95 ANOVA
Mean (SD) Mean (SD) Mean (SD) (SD) Mean (SD) (SEM%) (MDC95%)
velocity (m/s) YA 1.47 (0.19) 1.45 (0.19) 1.48 (0.19) 1.44 (0.24) 1.36 (0.19) 0.99 (0.98–0.99) 0.98 (0.93–0.99) 0.02 (1.26) 0.05 (3.49) F(4,9) = 21.9***
velocity (m/s) OA 1.23 (0.19) 1.23 (0.20) 1.24 (0.19) 1.19 (0.22) 1.10 (0.19) 0.87 (0.69–0.96) 0.86 (0.68–0.95) 0.08 (6.43) 0.21 (17.8) F(4,11) = 1.39ns
cadence (steps/s) YA 111.8 (9.68) 111.5 (9.97) 112.1 (9.35) 117.1 (19.66) 111.6 (9.73) 0.96 (0.91–0.99) 0.96 (0.89–0.98) 2.24 (1.99) 6.22 (5.51) F(4,9) = 2.29T
Klotzbier et al. European Review of Aging and Physical Activity

cadence (steps/s) OA 106.9 (9.21) 106.6 (9.86) 107.5 (8.75) 106.3 (12.18) 105.6 (9.42) 0.99 (0.98–0.99) 0.99 (0.98–0.99) 0.82 (0.77) 2.28 (2.14) F(4,11) = 1.69ns
stride length (m) YA 1.56 (0.09) 1.56 (0.09) 1.58 (0.10) 1.49 (0.04) 1.44 (0.13) 0.95 (0.88–0.97) 0.88 (0.62–0.97) 0.02 (1.27) 0.05 (3.51) F(4,9) = 18.1***
stride length (m) OA 1.37 (0.16) 1.37 (0.16) 1.38 (0.16) 1.35 (0.15) 1.25 (0.14) 0.99 (0.98–0.99) 0.96 (0.84–0.99) 0.02 (1.12) 0.04 (3.11) F(4,11) = 34.4***
single limb support (%GCT) YA 82.9 (3.01) 75.1 (2.45) 75.9 (2.59) 86.7 (2.07) 82.1 (1.73) 0.71 (0.27–0.92) 0.26 (−0.01–0.64 0.89 (1.09) 2.45 (3.05) F(4,9) = 61.9***
(2021) 18:17

single limb support (%GCT) OA 79.9 (2.33) 73.3 (2.57) 73.5 (2.19) 85.4 (1.32) 79.1 (2.93) 0.81 (0.57–0.94) 0.31 (0.02–0.66) 0.76 (0.98) 2.11 (2.69) F(4,11) = 105.6***
double limb support (%GCT) YA 17.0 (3.01) 24.9 (2.45) 24.0 (2.59) 13.3 (2.07) 17.7 (1.88) 0.73 (0.33–0.92) 0.27 (−0.01–0.66) 0.88 (4.52) 2.43 (12.5) F(4,9) = 63.4***
double limb support (%GCT) OA 20.1 (2.33) 26.7 (2.57) 26.5 (2.19) 14.6 (1.32) 20.9 (2.93) 0.81 (0.56–0.94) 0.31 (0.02–0.66) 0.76 (3.51) 2.12 (9.72) F(4,11) = 104.9***
stance (%GCT) YA 58.4 (1.39) 62.5 (1.27) 61.9 (1.29) 66.7 (4.19) 58.8 (1.06) −7.49 (−20.2–1.41)§ −0.45 (−0.54–0.16)§ 1.08 (1.75) 2.99 (4.85) F(4,9) = 19.5***
stance (%GCT) OA 60.2 (1.23) 63.5 (1.36) 63.2 (1.09) 65.1 (2.08) 60.4 (1.35) 0.82 (0.59–0.94) 0.46 (0.07–0.79) 0.47 (0.76) 1.31 (2.09) F(4,11) = 49.3***
YA Young adults, OA Older adults, GTC Gait cycle time, ICC Intra-class-correlation, A absolute agreement, C consistency, 95% CI 95% confidence interval; §The Tukey Additivity Test shows a significant interaction effect
between the systems and the persons being assessed, which contradicts the requirements for the ICC analysis. An interpretation is not possible due to a possible under- or overestimation and since reliability measures
are by definition limited to a value range from 0 to 1, negative ICCs indicate a reliability of 0 [46]; SEM Standard error of measurement, mdc minimal detectable change. In order to be able to compare both measures,
they were additionally expressed as percentages (SEM% and MDC95%). ANOVA to calculate the differences between the measurement systems: T Tendency, ns Not significant, *p < .05, **p < .01, ***p < .001
Page 7 of 12
Klotzbier et al. European Review of Aging and Physical Activity (2021) 18:17 Page 8 of 12

between 2.09–17.8% for all goups and parameters. The DT, YA only demonstrated reductions in stride length
MDC95% fluctuated around 17.1% for the total sample. and OA demonstrated longer single and double limb
In a variance-analytical comparison of the systems, dif- support times.
ferences can be reported for almost all parameters. Overall, our results show the expected differences in
walking performance between OA and YA as ageing is
Comparison between real modification and random associated with many changes in the locomotor system
measurement error [48, 49]. It is well described that neural reflexes, visual
Table 3 shows the motor DTC of the six gait parameters and vestibular feedback decrease with age [50] and in
for the two groups separately as well as the smallest combination or interaction, these age-related changes
amount of change required to designate a change as real lead to decrements of the locomotor coordination and
and beyond the bounds of measurement error. It can be have an impact on walking performance because of de-
observed that the percentage DTC was lower than the creasing gait stability [51]. We found consistent results
MDC in percentage for most of the gait parameters in with previous studies for reduced gait velocity [52], re-
both YA and OA. Especially the low sensitivity of change duced step length [49] as well as increased double limb
detection for gait velocity in OA with MDC95% of 17.8% support [53] for OA in comparison to YA. Also, most of
is noticeable. the studies focusing on falls prevention reported higher
decrements of gait parameters for fallers in comparison
Discussion to non fallers including gait velocity, step length, step
The aim of the study was to detect the amount for DT width and double limb support time over several years
decrements for OA and YA across five gait analysis sys- [53] or under DT conditions [54, 55]. Consistent with
tems and to determine wheter the DT effect is greater the literature and according to our results only single
than the measurement error between these systems. This and double limb support phase show significant changes.
is important in order to interpret the DTC for example Interestingly, following the classification by Beauchet
in age comparisons or future training studies. and colleagues [19], YA only showed DT decrements for
The main findings for the age comparison of this study parameters of pace whereas OA showed DTC for rele-
were that OA and YA differ under ST conditions in gait vant parameters of pace and rhythm (e.g., velocity and
velocity, stride length, single and double limb support as step length; rhythm: double support time). The DTC for
well as the stance phase. OA walk slower with shorter both aspects of gait quality might be one explanation for
stride lengths and lower times in the single limb support greater gait instabilities of OA.
phase, but with higher times in the double limb support On the other hand, some studies showed that there is
phase and stance phase. Regarding changes from ST to not always a deterioration in performance under DT

Table 3 The comparison between real modification in performance and the contribution of random measurement error
Parameter ST DT DT effect DT increase/ SEM MDC95 Real
Mean (SD) Mean (SD) Δ (DTE%) decline* (SEM%) (MDC95%) modification
N (%)
velocity (m/s) YA 1.44 (.198) 1.31 (.237) −0.13 (− 9.03) 22.2 0.02 (1.26) 0.05 (3.49) yes
velocity (m/s) OA 1.18 (.207) 1.11 (.237) −0.07 (− 5.93) 36.4 0.08 (6.43) 0.21 (17.8) no
cadence (steps/s) YA 112.8 (11.5) 109.9 (14.2) −2.9 (− 2.57) 22.2 2.24 (1.99) 6.22 (5.51) no
cadence (steps/s) OA 106.5 (9.81) 102.1 (14.9) −4.4 (−4.13) 36.4 0.82 (0.77) 2.28 (2.14) yes
stride length (m) YA 1.53 (.088) 1.44 (.094) −0.09 (−5.88) 66.7 0.02 (1.27) 0.05 (3.51) yes
stride length (m) OA 1.34 (.151) 1.29 (.151) −0.05 (−3.73) 75.0 0.02 (1.12) 0.04 (3.11) yes
Single limb support (%GCT) YA 80.6 (1.63) 78.9 (2.11) −1.7 (−2.11) 11.1 0.89 (1.09) 2.45 (3.05) no
single limb support (%GCT) OA 78.3 (1.77) 77.1 (2.17) −1.2 (− 1.53) 27.3 0.76 (0.98) 2.11 (2.69) no
double limb support (%GCT) YA 19.4 (1.69) 21.0 (2.13) 1.6 (8.25) 88.9 0.88 (4.52) 2.43 (12.5) no
double limb support (%GCT) OA 21.8 (1.76) 22.9 (2.18) 1.1 (5.05) 81.8 0.76 (3.51) 2.12 (9.72) no
stance (%GCT) YA 61.7 (.369) 61.1 (4.23) −0.6 (−.972) 22.2 1.08 (1.75) 2.99 (4.85) no
stance (%GCT) OA 62.5 (1.11) 61.7 (4.16) −0.8 (−1.28) 45.5 0.47 (0.76) 1.31 (2.09) no
The mean values of the measurement systems for the gait parameters divided into YA and OA in ST and DT condition are shown. YA Young adults, OA Older
adults, GTC Gait cycle time, ST single task, DT dual task. Δ = difference between ST and DT [were calculated as follows: (−(STmean – DTmean)). negative values
indicate decreases from ST to DT]. DTC% = proportional dual task costs [were calculated as follows: ((DT - ST) / ST) *100]. *DT increase/decline = % of individuals
showing higher values under DT compared to ST condition (whether this is an improvement or a deterioration in performance depends on the gait parameter; for
example, for velocity we observe that the value increases for 22.2% of the YA in the DT. Accordingly. the velocity decreases for 77.8%, what is to be expected for
the gait velocity. SEM = standard error of measurement of the gait systems. MDC Minimal detectable change. Real modification = MDC95% < DTC%
Klotzbier et al. European Review of Aging and Physical Activity (2021) 18:17 Page 9 of 12

conditions [54, 56]. According to the “Constrained-Ac- Thus, when interpreting a change in DT studies with
tion” hypothesis [57], focusing attention on a highly walking and under consideration of the different meas-
automated movement (internal focus) leads to perform- uring systems, a change in velocity of 0.21 m/s in OA
ance limitations. In contrast, an external focus of atten- may be considered a real change and indicates that
tion towards the cognitive task leads to a self-organized change is not the result of measurement error. It is pos-
and automated motion sequence and improved perform- sible to state with 95% certainty that the change is reli-
ance. As cognitive demands increase, the negative effect able rather than measurement error, if the absolute and
of competition for limited attention resources and the propotional motor DTC for gait velocity are at least
beneficial effect of an external attention focus overlap. A 17.8%. These findings have particularly strong implica-
comparison between different age groups shows a de- tions for the interpretation of study results or to describe
creasing positive effect of an additional cognitive task in training effects on DT performance. Therefore, in line
older persons [58, 59]. Since we neither manipulated the with recommendations by Wollesen and colleagues [16]
difficulty level of the cognitive task nor calculate propor- meta-analysis that deals with gait parameters, the results
tional cognitive DTC, we cannot confirm the predictions must consider that the different systems measure with
of the “Constrained-Action” hypothesis [57]. different accuracy.
The second aim of our study was to analyze the effect
of different measurement conditions. Within this study,
the variables of rhythm that described the main signifi- Limitations
cant differences between OA and YA for ST conditions A limitation in the context of this study is that the setup
(single and double limb support as well as stance phase) and positioning of the five measuring systems prevent
showed the poorest absolute agreements between the the overlapping areas being maintained over the entire
systems. Therefore, a direct comparison of these param- walking distance. This is especially the case with the
eters of different studies with different systems is only overground walking systems, as they have different
possible to a limited extent. The SEM% was low (0.76– measuring ranges. We consider the Zebris with a length
6.43%) in all conditions and both age groups. In 100% of of two meter as comparatively short, which limits its
the observations a SEM% ≤ 10% was found. Overall, the suitability for overground gait analysis (on a treadmill,
MDC95% values ranged from 2.09 to 17.8%. In line with the two-meter system can be quite useful). OptoGait
previous results the SEM values of basic spatiotemporal and GAITrite are limited by constraints including the
parameters (step length and gait velocity) were lower length of the walkway, and the suitability for only flat
than values of relative phase parameters (i.e., double surfaces, however these devices deliver very accurate re-
support time) [30, 34, 60]. The accuracy of the measure- sults. MobilityLab and GaitUp are ecologically valid as
ment of spatiotemporal gait parameters of rhythm de- they do not constrain the gait of participants and the en-
pends on the precision of the heel strike and toe off tire walking distance could be recorded, even though a
detection [61]. A greater variance in the section of these participant has to get used to the sensors. They do pro-
two parameters might also lead to a greater variance of vide good accuracy for basic spatiotemporal parameters
the calculations with the implemented algorhythms. This but are limited with respect to parameters of relative
might be an explanation for the higher SEM values in phase. When comparing MobilityLab and GaitUp, the
the gait parameters of rhythm. former filters all steps that are not representative of a
Thirdly, we wanted to investigate the minimum re- “steady-state” walking, which makes accurate step detec-
quired magnitude of change between ST and DT per- tion impossible.
formance to ensure that the gait systems detect a real The study meets the standards for excellent quality ac-
modification and to be 95% certain that it is not a meas- cording to COSMIN (COnsensus-based Standards for
urement error. The main results showed that for the the selection of health Measurement Instruments [39];).
relative phase parameters single limb support, double The general requirements for studies that use item re-
limb support and stance, the DTE is lower than the sponse theory (IRT) models, the general design issues,
minimum required magnitude of change according to
and questions regarding reliability (with its measurement
the MDC95 (calculated based on the SEM). The rela-
properties: reliability and measurement error) are ful-
tively low agreement in the values for the phase parame-
ters was already observed by Rudisch and colleagues filled. However, the COSMIN recommendations that ad-
[34], who were able to show that the basic spatiotempo- vise a sample size of 50 ([39], see also [62]) were not
ral parameters (i.e., stride length, cadence, and gait vel- met in this study. Thus, the results on the consistency of
ocity) showed better agreement than measures of the measurement systems and the results regarding the
relative phase parameters (i.e., single support phase, comparison between DTE and system-measurement
double support phase, stance; see also [20, 27, 30, 31]). error under ST condition must be interpreted with
Klotzbier et al. European Review of Aging and Physical Activity (2021) 18:17 Page 10 of 12

caution. No firm conclusion can be made for these aims draft together with Thomas Klotzbier, Review & Editing, Project
of the study, due to the small sample size. administration.

In addition to the two trials under ST condition to get

Funding
used to the attached sensors, further DT familiarization This research did not receive any specific grant from funding agencies in the
trials would have been useful to get used to this condi- public, commercial, or not-for-profit sectors. Open Access funding enabled
tion (specifically related to the additional verbal fluency and organized by Projekt DEAL.
task). Multiple trails in both conditions could quite pos-
sibly increase the validity of the study. Since the per- Availability of data and materials
Data can be obtained from the corresponding author upon reasonable
formance of the cognitive task under ST condition was request.
not recorded, a conclusion about the cognitive DTC is
not possible. It can be assumed that differences in the Declarations
DTCs are more pronounced in persons with impair-
ments [62]. In this respect, it is uncertain whether a reli- Ethics approval and consent to participate
All assessments were conducted in accordance with ethical rules for research
able detection of the steps is possible for persons with in human subjects following the Declaration of Helsinki (Fortalenza 2013).
severe locomotion problems. Especially with overground The study protocol (AZ 2020_2077) was approved by the ethics committee
walking systems it can be difficult to detect a gait char- of the Hamburg University.
All participants received written and verbal information about the study and
acteristic with small shuffling as we see for example in signed informed consent prior to their participation.
people with Parkinson’s disease or fragile OA.
Competing interests
The authors have no financial or personal relationships with any other
Conclusions
person or organization that could improperly influence or otherwise
It seems important that studies on gait parameters in influence their work in this study. On behalf of all authors, the corresponding
motor-cognitive DTs provide information not only on author states that there is no conflict of interest.
the significance and statistical DTC, but also on the
Author details
probable cause of all reported changes in performance, 1
Department of Sport and Exercise Science, University of Stuttgart,
i.e., on the contribution of both real changes in perform- Allmandring 28, 70569 Stuttgart, Germany. 2Department of Human
Movement Science, University of Hamburg, Mollerstraße 10, 20148 Hamburg,
ance and random measurement errors to the reported
Germany. 3Biological Psychology and Neuroergonomics, TU Berlin,
changes. Fasanenstr. 1, 10623 Berlin, Germany. 4Department of Neuromotor Behavior
For studies which are to be compared directly with and Exercise, Institute of Sport and Exercise Sciences, University of Münster,
Horstmarer Landweg 62B, 48149 Münster, Germany. 5Institute for
each other and in those where different systems are
Biomechanics, Clinic Lindenplatz, Weslarner Str. 29, 59505, Bad Sassendorf,
used, the comparability of the gait parameters must be Germany. 6Department of Exercise & Health, University of Paderborn,
queried due to the low absolute agreement (absolute re- Warburger Straße 100, 33098 Paderborn, Germany. 7Department of Sports
Medicine, Goethe University Frankfurt am Main, Ginnheimer Landstr. 39,
liability). If, on the other hand, intervention effects on
60487 Frankfurt, Germany.
gait parameters from different studies are compared with
each other, the “acceptable” consistency across the Received: 13 April 2021 Accepted: 20 July 2021
measurement systems ensures comparability and is less
problematic. When it comes to the choice of an appro-
References
priate measurement system, this must always be seen in 1. Faulkner KA, Redfern MS, Cauley JA, Landsittel DP, Studenski SA, Rosano C.
relation to the research question, the walking task to be Multitasking: association between poorer performance and a history of
performed and the respective setting in which the sys- recurrent falls. J Am Geriatr Soc. 2007;55(4):570–6. https://doi.org/10.1111/j.1
532-5415.2007.01147.x.
tems are used, clinical routine or scientific interest. 2. Paul SS, Ada L, Canning CG. Automaticity of walking–implications for
physiotherapy practice. Phys Ther Rev. 2005;10(1):15–23. https://doi.org/1
Acknowledgements 0.1179/108331905X43463.
The authors thank the volunteers who participated in the study. 3. Takagi D, Nishida Y, Fujita D. Age-associated changes in the level of physical
activity in elderly adults. J Phys Ther Sci. 2013;27(12):3685–7. https://doi.
Authors’ contributions org/10.1589/jpts.27.3685.
We confirm that all authors were fully involved in the study and preparation 4. Sun F, Norman IJ, While AE. Physical activity in older people: a systematic
of the manuscript and all authors have approved the manuscript and agree review. BMC Public Health. 2013;13(1):449. https://doi.org/10.1186/1471-24
with its submission to European Journal of Aging. All authors fulfill the ICMJE 58-13-449.
(International Comitee of Medical Journal Editors) recommendations on 5. Tomas-Carus P, Biehl-Printes C, Pereira C, Veiga G, Costa A, Collado-
authorship. Credit taxonomy: T.J.K.: Conceptualization, Methodology, Mateo D. Dual task performance and history of falls in community-
Validation, Formal analysis, Investigation, Writing - Original Draft, Writing - dwelling older adults. Exp Geronto. 2019;120:35–9. https://doi.org/10.101
Review & Editing, Visualization, Project administration; J.R.: Conceptualization, 6/j.exger.2019.02.015.
Methodology, Validation, Formal analysis, Investigation, Review & Editing; 6. Nasar JL, Troyer D. Pedestrian injuries due to mobile phone use in
Thomas Cordes: Conceptualization, Methodology, Investigation, Review & public places. Accid Anal Prev. 2013;57:91–5. https://doi.org/10.1016/j.aa
Editing; O.V.: Formal analysis, Investigation, Resources, Review & Editing; L.V.: p.2013.03.021.
Conceptualization, Methodology, Supervision, Writing - Review & Editing; T.J.: 7. Palmiero M, Piccardi L, Boccia M, Baralla F, Cordellieri P, Sgalla R. Neural
Conceptualization, Methodology, Validation, Formal analysis, Investigation, correlates of simulated driving while performing a secondary task: a review.
Review & Editing; B.W.: Conceptualization, Methodology, Writing – Original Front Psychol. 2019;10:1045. https://doi.org/10.3389/fpsyg.2019.01045.
Klotzbier et al. European Review of Aging and Physical Activity (2021) 18:17 Page 11 of 12

8. Doumas M, Smolders C, Krampe RT. Task prioritization in aging: effects of 28. Mariani B, Hoskovec C, Rochat S, Büla C, Penders J, Aminian K. 3D gait
sensory information on concurrent posture and memory performance. Exp assessment in young and elderly subjects using foot-worn inertial
Brain Res. 2008;187(2):275–81. https://doi.org/10.1007/s00221-008-1302-3. sensors. J Biomech. 2010;43(15):2999–3006. https://doi.org/10.1016/j.
9. Lacour M, Bernard-Demanze L, Dumitrescu MM. Posture control, aging, and jbiomech.2010.07.003.
attention resources: models and posture-analysis methods. Neurophysiol 29. Bourgeois AB, Mariani B, Aminian K, Zambelli PY, Newman CJ. Spatio-
Clin. 2008;38(6):411–21. https://doi.org/10.1016/j.neucli.2008.09.005. temporal gait analysis in children with cerebral palsy using, foot-worn
10. Wollesen B, Voelcker-Rehage C, Regenbrecht T, Mattes K. Influence of a inertial sensors. Gait Posture. 2014;39:436–42. https://doi.org/10.1016/j.ga
visual–verbal Stroop test on standing and walking performance of older itpost.2013.08.029.
adults. Neuroscience. 2016;318:166–77. https://doi.org/10.1016/j. 30. Lienhard K, Schneider D, Maffiuletti NA. Validity of the Optogait photoelectric
neuroscience.2016.01.031. system for the assessment of spatiotemporal gait parameters. Med Eng Phys.
11. Kahneman D. Attention and effort. Englewood Cliffs: Prentice-Hall; 1973. 2013;35(4):500–4. https://doi.org/10.1016/j.medengphy.2012.06.015.
12. Wickens CD. Processing resources and attention. Multiple-task performance; 31. Lee MM, Song CH, Lee KJ, Jung SW, Shin DC, Shin SH. Concurrent validity
1991. p. 3–34. and test-retest reliability of the OPTOGait photoelectric cell system for the
13. Schaefer S, Schumacher V. The interplay between cognitive and motor assessment of spatio-temporal parameters of the gait of young adults. J
functioning in healthy older adults: findings from dual-task studies and Phys Ther Sci. 2014;26(1):81–5. https://doi.org/10.1589/jpts.26.81.
suggestions for intervention. Gerontology. 2011;57(3):239–46. https://doi. 32. Schmitz-Hübsch T, Brandt AU, Pfueller C, Zange L, Seidel A, Kühn AA.
org/10.1159/000322197. Accuracy and repeatability of two methods of gait analysis–GaitRite™ und
14. Lindenberger U, Marsiske M, Baltes PB. Memorizing while walking: increase mobility lab™–in subjects with cerebellar ataxia. Gait Posture. 2016;48:194–
in dual-task costs from young adulthood to old age. Psychol Aging. 2000; 201. https://doi.org/10.1016/j.gaitpost.2016.05.014.
15(3):417–36. https://doi.org/10.1037/0882-7974.15.3.417. 33. World Medical Association. Declaration of Helsinki, ethical principles for
15. Menant JC, Schoene D, Sarofim M, Lord SR. Single and dual task tests of medical research involving human subjects. 64 nd WMA General Assembly,
gait speed are equivalent in the prediction of falls in older people: a Fortaleza, Brazil. 2013.. https://www.wma.net/policies-post/wma-declaration-
systematic review and meta-analysis. Ageing Res Rev. 2014;16:83–104. of-helsinki-ethical-principles-for-medical-research-involving-human-subjects.
https://doi.org/10.1016/j.arr.2014.06.001. Accessed 1 Apr 2021.
16. Wollesen B, Wanstrath M, Van Schooten KS, Delbaere K. A taxonomy of 34. Rudisch J, Jöllenbeck T, Vogt L, Cordes T, Klotzbier TJ, Vogel O. Agreement
cognitive tasks to evaluate cognitive-motor interference on and consistency of five different clinical gait analysis systems in the
spatiotemoporal gait parameters in older people: a systematic review assessment of spatiotemporal gait parameters. Gait Posture. 2021;85:55–64.
and meta-analysis. Eur Rev Aging Phys Act. 2019b;16(1):12. https://doi. https://doi.org/10.1016/j.gaitpost.2021.01.013.
org/10.1186/s11556-019-0218-1. 35. Cordes T, Bischoff LL, Schoene D, Schott N, Voelcker-Rehage C, Meixner C. A
17. Al-Yahya E, Dawes H, Smith L, Dennis A, Howells K, Cockburn J. Cognitive multicomponent exercise intervention to improve physical functioning,
motor interference while walking: a systematic review and meta-analysis. cognition and psychosocial well-being in elderly nursing home residents: a
Neurosci Biobehav Rev. 2011;35(3):715–28. https://doi.org/10.1016/j. study protocol of a randomized controlled trial in the PROCARE (prevention
neubiorev.2010.08.008. and occupational health in long-term care) project. BMC Geriatr. 2019;19(1):
18. Gschwind Y, Bridenbaugh S. The role of gait analysis. Early detection of 369. https://doi.org/10.1186/s12877-019-1386-6.
dementia and risk of falling. Der Informierte Arzt. 2011;6:39–41 Available 36. Menz HB, Latt MD, Tiedemann A, San Kwan MM, Lord SR. Reliability of the
from: https://www.tellmed.ch/include_php/previewdoc.php?file_id=8223. GAITRite® walkway system for the quantification of temporo-spatial
19. Beauchet O, Allali G, Sekhon H, Verghese J, Guilain S, Steinmetz JP, et al. parameters of gait in young and older people. Gait Posture. 2004;20(1):20–5.
Guidelines for assessment of gait and reference values for spatiotemporal https://doi.org/10.1016/S0966-6362(03)00068-7.
gait parameters in older adults: the biomathics and Canadian gait 37. Plummer P, Eskes G. Measuring treatment effects on dual-task performance:
consortiums initiative. Front Hum Neurosci. 2017 Aug;11(353):1–14. https:// a framework for research and clinical practice. Front Hum Neurosci. 2015;9:
doi.org/10.3389/fnhum.2017.00353. 225. https://doi.org/10.3389/fnhum.2015.00225.
20. Washabaugh EP, Kalyanaraman T, Adamczyk PG, Claflin ES, Krishnan C. 38. Koo TK, Li MY. A guideline of selecting and reporting intraclass correlation
Validity and repeatability of inertial measurement units for measuring gait coefficients for reliability research. J Chiropr Med. 2016 Jun;15(2):155–63.
parameters. Gait Posture. 2017;55:87–93. https://doi.org/10.1016/j.gaitpost.2 https://doi.org/10.1016/j.jcm.2016.02.012.
017.04.013. 39. Mokkink LB, Terwee CB, Gibbons E, Stratford PW, Alonso J, Patrick DL, et al.
21. Klotzbier TJ, Schott N. Cognitive-motor interference during walking in older Inter-rater agreement and reliability of the COSMIN (COnsensus-based
adults with probable mild cognitive impairment. Front Aging Neurosci. standards for the selection of health status measurement instruments)
2017;9:350. https://doi.org/10.3389/fnagi.2017.00350. checklist. BMC Med Res Methodol. 2010 Sep;10(82):1–11. https://doi.org/1
22. Gomes GDC, Teixeira-Salmela LF, Freitas FASD, Fonseca MLM, Pinheiro MDB, 0.1186/1471-2288-10-82.
Morais VADC. Gait performance of the elderly under dual-task conditions: 40. Atkinson G, Nevill AM. Statistical methods for assessing measurement error
review of instruments employed and kinematic parameters. Rev Bras Geriatr (reliability) in variables relevant to sports medicine. Sports Med. 1998;26(4):
Gerontol. 2016;19(1):165–82. https://doi.org/10.1590/1809-9823.2016.14159. 217–38. https://doi.org/10.2165/00007256-199826040-00002.
23. Smith E, Cusack T, Cunningham C, Blake C. The influence of a cognitive dual 41. Hollman JH, Beckman BA, Brandt RA, Merriwether EN, Williams RT, Nordrum
task on the gait parameters of healthy older adults: a systematic review and JT. Minimum detectable change in gait velocity during acute rehabilitation
meta-analysis. J Aging Phys Act. 2017;25(4):671–86. https://doi.org/10.1123/ following hip fracture. J Geriatr Phys Ther. 2008;31(2):53–6. https://doi.org/1
japa.2016-0265. 0.1519/00139143-200831020-00003.
24. Wollesen B, Mattes K, Rönnfeldt J. Influence of age, gender and test 42. Hollman JH, Childs KB, McNeil ML, Mueller AC, Quilter CM, Youdas JW.
conditions on the reproducibility of dual-task walking performance. Number of strides required for reliable measurements of pace, rhythm and
Aging Clin Exp Res. 2017;29(4):761–9. https://doi.org/10.1007/s40520-016- variability parameters of gait during normal and dual task walking in older
0664-9. individuals. Gait Posture. 2010;32(1):23–8. https://doi.org/10.1016/j.gaitpost.2
25. Webster KE, Wittwer JE, Feller JA. Validity of the GAITRite® walkway system for 010.02.017.
the measurement of averaged and individual step parameters of gait. Gait 43. Schwenk M, Gogulla S, Englert S, Czempik A, Hauer K. Test–retest reliability
Posture. 2005;22(4):317–21. https://doi.org/10.1016/j.gaitpost.2004.10.005. and minimal detectable change of repeated sit-to-stand analysis using one
26. Cutlip RG, Mancinelli C, Huber F, DiPasquale J. Evaluation of an body fixed sensor in geriatric patients. Physio Meas. 2012;33(11):1931–46.
instrumented walkway for measurement of the kinematic parameters of https://doi.org/10.1088/0967-3334/33/11/1931.
gait. Gait Posture. 2000;12(2):134–8. https://doi.org/10.1016/S0966-6362 44. Faul F, Erdfelder E, Lang AG, Buchner A. G* power 3: a flexible
(00)00062-X. statistical power analysis program for the social, behavioral, and
27. Lee M, Song C, Lee K, Shin D, Shin S. Agreement between the spatio- biomedical sciences. Behav Res Methods. 2007;39(2):175–91. https://doi.
temporal gait parameters from treadmill-based photoelectric cell and the org/10.3758/BF03193146.
instrumented treadmill system in healthy young adults and stroke patients. 45. Tabachnick BG, Fidell LS. Using multivariate statistics 6th edn. Pearson
Med Sci Monit. 2014;20:1210. https://doi.org/10.12659/MSM.890658. Education Limited: New International Edition; 2013.
Klotzbier et al. European Review of Aging and Physical Activity (2021) 18:17 Page 12 of 12

46. Weir JP. Quantifying test-retest reliability using the intraclass correlation
coefficient and the SEM. J Strength Cond Res. 2005;19(1):231–40. https://doi.
org/10.1519/15184.1.
47. Shrout PE, Fleiss JL. Intraclass correlations: uses in assessing rater reliability.
Psychol Bull. 1979;86(2):420–8. https://doi.org/10.1037/0033-2909.86.2.420.
48. Kerber KA, Ishiyama GP, Baloh RW. A longitudinal study of oculomotor
function in normal older people. Neurobiol Aging. 2006;27(9):1346–53.
https://doi.org/10.1016/j.neurobiolaging.2005.07.009.
49. Seidler RD, Bernard JA, Burutolu TB, Fling BW, Gordon MT, Gwin J. Motor
control and aging: links to age-related brain structural, functional, and
biochemical effects. Neurosci Biobehav Rev. 2010;34(5):721–33. https://doi.
org/10.1016/j.neubiorev.2009.10.005.
50. Verdú E, Ceballos D, Vilches JJ, Navarro X. Influence of aging on peripheral
nerve function and regeneration. J Peripher Nerv Syst. 2000;5:91–208.
https://doi.org/10.1111/j.1529-8027.2000.00026.x.
51. Hausdorff JM, Rios DA, Edelberg HK. Gait variability and fall risk in
community-living older adults: a 1-year prospective study. Arch Phys Med
Rehabil. 2001;82(8):1050–6. https://doi.org/10.1053/apmr.2001.24893.
52. Morrison S, Colberg SR, Parson HK, Neumann S, Handel R, Vinik EJ. Walking-
induced fatigue leads to increased falls risk in older adults. J Am Med Dir
Assoc. 2016;17(5):402–9. https://doi.org/10.1016/j.jamda.2015.12.013.
53. Scott D, McLaughlin P, Nicholson GC, Ebeling PR, Stuart AL, Kay D. Changes
in gait performance over several years are associated with recurrent falls
status in community-dwelling older women at high risk of fracture. Age
Ageing. 2015;44(2):287–93. https://doi.org/10.1093/ageing/afu169.
54. Wollesen B, Voelcker-Rehage C. Differences in cognitive-motor interference
in older adults while walking and performing a visual-verbal Stroop task.
Front Aging Neurosci. 2019;10:426. https://doi.org/10.3389/fnagi.2018.00426.
55. Muhaidat J, Kerr A, Evans JJ, Skelton DA. The test–retest reliability of gait-
related dual task performance in community-dwelling fallers and non-fallers.
Gait Posture. 2013;38(1):43–50. https://doi.org/10.1016/j.gaitpost.2012.10.011.
56. Stoffregen TA, Hove P, Bardy BG, Riley M, Bonnet CT. Postural stabilization of
perceptual but not cognitive performance. J Mot Behav. 2007;39(2):126–38.
https://doi.org/10.3200/JMBR.39.2.126-138.
57. Wulf G, McNevin N, Shea CH. The automaticity of complex motor skill
learning as a function of attentional focus. Q J Exp Psychol [A]. 2001;54:
1143–54. https://doi.org/10.1080/713756012.
58. Huxhold O, Li SC, Schmiedek F, Lindenberger U. Dual-tasking postural
control: aging and the effects of cognitive demand in conjunction with
focus of attention. Brain Res Bull. 2006;69(3):294–305. https://doi.org/10.101
6/j.brainresbull.2006.01.002.
59. Verrel J, Lövdén M, Schellenbach M, Schaefer S, Lindenberger U. Interacting
effects of cognitive load and adult age on the regularity of whole-body
motion during treadmill walking. Psychol Aging. 2009;24(1):75–81. https://
doi.org/10.1037/a0014272.
60. Bilney B, Morris M, Webster K. Concurrent related validity of the
GAITRite® walkway system for quantification of the spatial and temporal
parameters of gait. Gait Posture. 2003;17(1):68–74. https://doi.org/10.101
6/S0966-6362(02)00053-X.
61. Kobsar D, Charlton JM, Tse CT, Esculier JF, Graffos A, Krowchuk NM, et al.
Validity and reliability of wearable inertial sensors in healthy adult walking: a
systematic review and meta-analysis. J Neuroeng Rehabilitation. 2020;17(62):
1–21. https://doi.org/10.1186/s12984-020-00685-3.
62. Terwee CB, Bot SD, de Boer MR, van der Windt DA, Knol DL, Dekker J, et al.
Quality criteria were proposed for measurement properties of health status
questionnaires. J Clin Epidemiol. 2007;60(1):34–42. https://doi.org/10.1016/j.
jclinepi.2006.03.012.

Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.