Validity and reliability of a spatiotemporal gait-analysis system
tiotemporal gait parameters for barefoot comfortable
walking with the stationary SGAS setup (Table III).
The CoV decreased gradually when averaging more
footsteps. Four footsteps were required to reach a CoV
below 10% for step length, step time, and stance time.
The Bland-Altman plots showed no heteroscedasticity
(Fig. 4). For double support time, 8 footsteps were re-
quired. For the shod walking condition, similar results
were found. The sample sizes of the toe walking and
slow gait speed conditions were too small, considering
that there were fewer than 10 valid trials of footsteps
per subject (sample sizes of 10 and 15 subjects, re-
spectively).
The minimum number of footsteps needed for ade-
quate reliability during comfortable barefoot walking
by the moving SGAS camera in 30 subjects were 3
footsteps for step length, step time, and stance time
and 7 footsteps for double support time.
Inter- and intra-rater reliability
For assessment of the inter- and intra-rater reliability,
304–316 trials were used. Step length, step time, and
stance time values had excellent agreement between
the 3 observers (ICC ≥ 0.94 and lower limit of the 95%
CIs ≥ 0.86). Inter-rater agreement on double support
time was moderate (ICC 0.68 and 95% CI 0.48–0.79).
Intra-rater reliability was excellent for step length, step
time, and stance time (ICC ≥ 0.98 and lower limit of
the 95% CIs ≥ 0.97), and good for double support time
(ICC 0.84 and 95% CI 0.80–0.87).
DISCUSSION
This study found that the SGAS is a valid and reliable
system to assess step length, step time, stance time,
stride length, and stride time under different walking
conditions. The stationary, as well as the moving, ca-
mera set-up can be used to determine these spatiotem-
poral gait parameters. However, validity could not be
confirmed for double support time and swing time. A
minimum of 4 footsteps was needed to obtain a reliable
assessment of step length, step time, and stance time
with the SGAS. Inter- and intra-rater reliability were
confirmed for step length, step time, and stance time.
There is a need for low-cost and portable gait ana-
lysis technology in clinical settings. Such technology
needs to be assessed for validity and reliability in
assessing spatiotemporal gait parameters, such as in
the current study using a 1-camera method. One other
study using a 1-camera system examined the validity
between this system and a reference 3-dimensional
motion capture system (20). They found differences
in accuracy between the 2 systems that were similar
461
compared with the current study for the temporal
parameters. However, they found larger differences
between the system and the reference for the spa-
tial parameters compared with our study, which may,
among other possible explanations, be due to the
choice of reference system. Results on reliability and
measurement error were not reported in that study.
Furthermore, they examined the validity for different
gait speed conditions with the subjects wearing ankle
socks, but not for the conditions of toe walking or shod
walking. The Microsoft Kinect v2, which is a camera
system extracting data from 3-dimensional skeletal
modelling, has been shown to provide valid results for
temporal parameters. Although results on accuracy for
spatial parameters were inconsistent between studies
(19, 21–24), one study showed an ICC of 0.76 for step
length (95% CI –0.17 to 0.95) and an absolute and
relative error of 10 cm (SD 5 cm) (21). Other low-
cost alternatives using footswitches, accelerometers,
gyroscopes, and inertial measurement units have been
shown to be accurate in measuring temporal gait pa-
rameters, but are currently either unable or inaccurate
to measure spatial parameters (25–30). Moreover, the
advantage of a camera system over these methods is
that video images of the person are obtained, which
can be used for clinical assessment of gait pathology.
The SGAS is a feasible, easy-to-use measurement
instrument for clinical practice and research purposes.
In the current study, position and time assessment to
calculate the spatiotemporal gait parameters from the
SGAS user interface was a manual process. For expe-
rienced observers, position and time assessment for
10 trials in one walking condition took approximately
10–15 min. The observers noted that the video capture
with time resolution of 0.02 s regularly missed the ex-
act moment of initial contact or toe off, complicating
the assessment. However, this did not compromise the
reliability, since the results show that accurate data can
be obtained with a 50 Hz sampling rate. The stationary
camera set-up can be used in all settings, but is restric-
ted by the field of view. In this study, the chosen field
of view was 130 cm, to provide good spatial resolution
for accuracy, but at the expense of being able to assess
full strides. The use of a moving camera set-up solves
this problem and, additionally, requires less effort from
persons, as multiple steps are analysed in a single trial.
A moving SGAS camera does, however, require a rail
placed parallel to the walkway and a steady tripod on
wheels. Recordings from the SGAS can be combined
with other gait recordings (e.g. electromyography and
force plate).
In this study, double support time and swing time
could not be assessed in a valid and reliable way with
the SGAS using the GAITRite® system as a reference.
J Rehabil Med 51, 2019