Measurement of Physical Activity and Energy Expenditure in Wheelchair Users: Methods, Considerations and Future Directions
Nightingale et al. Sports Medicine - Open
Measurement of Physical Activity and Energy Expenditure in Wheelchair Users: Methods, Considerations and Future Directions
Tom E. Nightingale 0
Peter C. Rouse 0
Dylan Thompson 0
James L. J. Bilzon 0
Key Points 0
0 Department for Health, University of Bath , Bath BA2 7AY , UK
Accurately measuring physical activity and energy expenditure in persons with chronic physical disabilities who use wheelchairs is a considerable and ongoing challenge. Quantifying various free-living lifestyle behaviours in this group is at present restricted by our understanding of appropriate measurement tools and analytical techniques. This review provides a detailed evaluation of the currently available measurement tools used to predict physical activity and energy expenditure in persons who use wheelchairs. It also outlines numerous considerations specific to this population and suggests suitable future directions for the field. Of the existing three self-report methods utilised in this population, the 3-day Physical Activity Recall Assessment for People with Spinal Cord Injury (PARA-SCI) telephone interview demonstrates the best reliability and validity. However, the complexity of interview administration and potential for recall bias are notable limitations. Objective measurement tools, which overcome such considerations, have been validated using controlled laboratory protocols. These have consistently demonstrated the arm or wrist as the most suitable anatomical location to wear accelerometers. Yet, more complex data analysis methodologies may be necessary to further improve energy expenditure prediction for more intricate movements or behaviours. Multi-sensor devices that incorporate physiological signals and acceleration have recently been adapted for persons who use wheelchairs. Population specific algorithms offer considerable improvements in energy expenditure prediction accuracy. This review highlights the progress in the field and aims to encourage the wider scientific community to develop innovative solutions to accurately quantify physical activity in this population.
information about the types and purpose of activities
being performed by persons who use wheelchairs.
This has implications to inform the wider public
health agenda by promoting physical activity and
reducing non-communicable diseases in persons with
chronic physical disabilities.
Considerable evidence now exists to support the
beneficial effects of physical activity (PA) for human health
and wellbeing [1–3]. However, the majority of this
evidence is from research in adults without disabilities. Our
understanding of the impact and importance of PA for
populations with chronic physical disabilities,
particularly those who use wheelchairs, is therefore lacking.
© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made.
One of the major barriers to the acquisition and analysis
of PA data in populations who use wheelchairs is the
uncertainty surrounding the validity and reliability of the
existing PA measurement tools. Improved assessment of
habitual PA would permit; appropriate cross-sectional
comparisons to biomarkers of metabolic health allow
researchers to comment on the efficacy of behaviour
change interventions and potentially inform PA
guidelines . This review provides a detailed evaluation of
the available tools within the context of their potential
application in persons who use wheelchairs. Firstly, three
of the most frequently utilised self-reported measures
will be described and evaluated. Then our attention will
turn to the increasingly employed objective methods of
The PA behaviour of persons who use wheelchairs is
inherently difficult to measure due to the heterogeneous
nature of the population, whereby different disability
aetiologies responsible for the use of a wheelchair result in
highly variable movement patterns. Common physical
disabilities that require prolonged use of a wheelchair
include spinal cord injury (SCI), amputation, multiple
sclerosis, cerebral palsy and cerebrovascular disease.
Currently, it is problematic to accurately equate PA into
units of energy expenditure (EE), as EE varies
significantly from person to person depending on body mass,
type of physical disability and efficiency of movement.
Due to movement being primarily restricted to the
upper body, the energy cost of most exercise and
activities of daily living performed by persons who use
wheelchairs result in a considerably lower energy cost (−27%)
than those reported in the general population [5, 6]. The
smaller skeletal muscle mass activated to perform
certain activities does not achieve the same whole-body
metabolic rate. Metabolic equivalents (METS) are often
used to express the energy costs of PA as multiples of
resting metabolic rate (RMR) . However, the
conventional MET value (oxygen uptake of 3.5 mL kg−1 min−1)
is not applicable for persons with a disability, as disuse/
paralysis results in atrophy of leg fat-free mass (FFM)
[8, 9]. RMR is influenced by FFM , which explains
why RMR is reduced in persons with disabilities who use
wheelchairs compared to adults without disabilities .
For example, commonly used equations to predict RMR
in persons with spinal cord injury (SCI) overestimate
measured requirements by 5–32% . Considering RMR is
the largest component of total daily energy expenditure
(TDEE) (up to 80% for sedentary individuals ), error in
the prediction of this component using existing algorithms
for persons without disabilities can have profound
implications for accurately predicting TDEE. Consequently,
approaches that solely measure physical activity energy
expenditure (PAEE) might have greater utility, particularly
as this is the most malleable component of TDEE.
The PA monitoring field is evolving at a rapid pace.
However, the development of validated self-report and
objective tools to quantify PA/EE in persons who use
wheelchairs remains relatively under researched. It is not
always feasible to use criterion methods (i.e. indirect
calorimetry, observation, doubly labelled water) to measure
free-living PA/EE, as these techniques require expensive/
sophisticated equipment or are impractical for use
outside of the laboratory. Therefore, this review provides an
overview of the predominant methods of measuring PA/
EE in persons who use wheelchairs. Specifically, we
describe and review the different self-report and objective
tools currently available whilst also considering their
potential strengths and limitations.
Until recently, the quantification of free-living PA in
persons who use wheelchairs had been reliant on outputs
from self-report measures [14, 15]. Self-report
questionnaires offer researchers an inexpensive and
easy-toadminister method of measuring PA. However, these
methods are reliant upon the accuracy of the
participants’ memory and recall. Furthermore, it has been
suggested that self-report measures fail to adequately
quantify the lower end of the PA continuum [16, 17],
suffer from floor effects (lowest score is too high for
inactive respondents) and participant over-reporting .
Besides these general limitations, specific issues
pertaining to the administration of the three predominant
questionnaires (Table 1) used to predict components of PA
in this population are discussed below. It is noteworthy
that not all were developed and/or validated for persons
who use wheelchairs.
The Physical Activity and Disability Survey (PADS)
 was one of the first questionnaires developed but
was validated for participants with a wide range of
disabilities ranging from stroke to type-2 diabetes, and
subsequently, a revised version (PADS-R) in persons with
neurological conditions [20, 21]. Therefore, it could be
argued that the content of the PADS fails to capture
activities specific to the lifestyle of persons that use
wheelchairs. The Physical Activity Scale for Individuals with
Physical Disabilities (PASIPD)  was adapted from the
Physical Activity Scale for the Elderly (PASE) and follows
a similar format to that of the International Physical
Activity Questionnaire (IPAQ) . Despite being
developed in people with both visual and auditory disabilities,
its implementation in people with locomotor
impairment and SCI means it could be considered sensitive to
persons who use wheelchairs. However, only the Physical
interview or self-administered
questionnaire (20–30 min)
3. General activity
6. Wheelchair use
Score is based on the time
respondents spend doing the
activities multiplied by an intensity
rating of that activity. Each activity
has an assigned weighting
(Aerobic = .3, strength = .2 and
flexibility = .1). Higher scores
represent more activity and negative
scores can be achieved through
questionnaire (~15 min)
1. Home repair/gardening
3. Vigorous sport
4. Moderate sport
Number of days per week and
hours per day of participation
in above dimensions. Intensity
of activity is established by
multiplying the average hours
per day for each item by a
standard MET value (MET-h/day)
whereby a series of flow charts
help the interviewer guide the
participants through 8 periods
of the day (20–45 min)
The mean number of minutes
per day spent in mild, moderate,
and heavy intensity LTPA and ADL.
Scores may be summed to generate
total accumulated PA (min/day)
Activity Recall Assessment for People with Spinal Cord
Injury (PARA-SCI) was specifically developed and
evaluated for people with SCI.
Questionnaire Administration A distinguishing feature
between the three questionnaires is the resource demand
to complete each tool (Table 1). The PARA-SCI was
designed as an interview-based questionnaire that collects
rich behavioural data. Thus, the PARA-SCI is resource
intensive because it was developed, as a research tool, to
be used in epidemiological studies. For example, it can
take between 20–45 min to complete, the cost of the
interviewer needs to be considered and there is
considerable participant demand. Ullrich et al.  also
suggested that the use of the PARA-SCI might have limited
application for other investigators, besides the
developers, due to the exclusion of subjective appraisals and
the technical complexity of interview administration.
These limitations were acknowledged by the authors
who subsequently developed a new questionnaire to
address these limitations. The Leisure Time Physical
Activity Questionnaire for People with Spinal Cord Injury
(LTPAQ-SCI)  is a brief (5 min) self-report
questionnaire specifically designed for persons with SCI that
measures minutes of mild, moderate and heavy-intensity
leisure time physical activity (LTPA) performed over the
previous 7 days but is not capable of measuring other
activities of daily living.
Reliability and Validity The test-retest reliability of the
three questionnaires has been examined; however, the
PADS has had no reliability studies conducted in
persons who use wheelchairs. Therefore, it remains unclear
whether the PADS can be reliably used as a measure of
physical activity behaviour in this population. A
testretest reliability correlation of .77 was established for the
PASIPD in a study of 45 adult patients with a range of
disabilities, but these patients did not use wheelchairs
. The PARA-SCI is the only instrument tested for
reliability in a sample solely consisting of persons who
use wheelchairs. To establish the test-retest reliability of
the PARA-SCI, 102 people with SCI completed the
instrument on two separate occasions a week apart
. Intra-class correlations revealed good test-retest
reliability for total cumulative activity (.79). However,
moderate-intensity (LTPA) and heavy-intensity (lifestyle
activity) demonstrated poor levels of reliability (ICC = .45
and .56, respectively).
Establishing the validity of questionnaires is important
to ensure that the tool effectively measures what it
intends to (i.e. the activity of persons that use a
wheelchair). Manns and colleagues  revealed a significant
moderate relationship between scores on the PADS and
V̇ O2 max (r = 0.45). Likewise, comparison of scores from
the PASIPD with indicators of physical capacity revealed
weak to moderate relationships (V̇ O2 max; r = 0.25,
manual muscle test; r = 0.35) . However, we contend
that equating self-reported PA to physical capacity, rather
than a criterion measure of PA, may not be the most
appropriate way to ascertain concurrent validity. Measures
of physical capacity can be related to numerous variables
beyond the users’ PA level.
Results from validity studies indicate that of the three
questionnaires, the PARA-SCI has the strongest
relationships with criterion measures. During the development
and evaluation of the PARA-SCI , criterion (V̇ O2
reserve) values displayed a very large correlation with
cumulative (LTPA plus lifestyle) activity data (r = 0.79). When
data was coded for intensity of activity, large to very
large positive correlations were seen for
moderateintensity (r = 0.63) and heavy-intensity (r = 0.88) activity.
However, this relationship was weak and non-significant
for low-intensity activities (r = 0.27) and consequently, the
PARA-SCI under-reported time spent doing activities of
low intensity by 10%. Therefore, although these findings
indicate some evidence of convergent validity, the results
also highlight limitations of self-report measures.
Measuring Intensity A distinguishing feature between
the three disability questionnaires is how they gather
information pertaining to the intensity of activity conducted.
Evidence from adults without disabilities would suggest
superior reductions in mortality risk with
vigorousintensity PA in comparison to light-to-moderate intensity
PA [30–32]. Therefore, failure to consider individual
differences in PA intensity makes it difficult to detect
relationships between lifestyle activities and health outcomes
. The PADS employs a single item to examine the
overall intensity of structured activity but does not assess
the intensity of leisure time activities. A fundamental
limitation of the PASIPD is the use of standard MET values as
a measure of activity intensity regardless of the
participant’s type of disability. If MET values are to be used, it
will be necessary to develop a new empirically based
supplement to the compendium of physical activity
appropriate for persons that use wheelchairs . The inability of
the PASIPD and PADS to effectively measure activity
intensity prompted the development of the PARA-SCI.
Subsequently, the authors of the PARA-SCI conducted a
systematic process to develop definitions of three different
exercise intensities (i.e. mild, moderate and heavy)
specifically for people with SCI . The empirical development
of intensity-based definitions suggests the PARA-SCI may
be the most effective self-report questionnaire at
measuring the intensity of PA in persons with SCI. However, it
should be noted that even with such a rigorous
development of intensity definitions, the PARA-SCI is still
dependent upon the accurate recall of behaviour. Research
has also challenged the use of psychophysiological indexes
as a measure of perceived exertion in persons with SCI
. This could have implications for the prediction of
activity intensity using self-report measures in persons
with disabilities, which could be influenced by secondary
conditions such as chronic pain and discomfort, coupled
with the inability to engage large muscle groups in
constant rhythmic activities.
Objective sensors overcome many of the shortcomings
of self-report methods, predominantly by removing the
subjective recall element. The next section will discuss
the use of these objective sensors in wheelchair users.
Accelerometers or movement sensors report their
outcomes in ‘activity counts’ per unit time or epoch, which
are the product of the frequency and intensity of
movement. Accelerometers are therefore capable of providing
temporal information about specific variables such as the
total amount, frequency and duration of PA . They
can also monitor the accumulation of
moderate-tovigorous intensity PA (MVPA) and/or sedentary behaviour
thanks to the development of population-specific
cutpoints for activity counts per minute. Despite enormous
differences in signal processing and internal components,
all accelerometers have similar fundamental properties
defined by accuracy, precision, range and sensitivity and
should be compared against criterion measurements to
demonstrate validity . Monitors have been compared
to a selection of criterion laboratory measurements in
persons that use wheelchairs: oxygen uptake (V̇ O2) [37–39],
EE [40, 41] and PAEE [42, 43] measured by indirect
calorimetry (Table 2). Studies have utilised different
commercial monitors, worn at various locations, validated using
diverse activity protocols including propulsion on a
wheelchair-adapted treadmill, wheelchair ergometer or
over ground, arm-crank ergometry (ACE) and various
activities of daily living. Two fundamentally different
varieties of accelerometers are widely used in PA research,
uniaxial and, increasingly, tri-axial. Uniaxial
accelerometers register movement in the vertical axis only, whereas
tri-axial accelerometers register movement in the
anteroposterior (X), mediolateral (Y) and vertical (Z) axes. In
keeping with pooled data from a systematic review of
laboratory and free-living validation studies in adults
without disabilities , it appears that the greater sensitivity
of the tri-axial accelerometer leads to a better prediction
of EE than uniaxial accelerometers in persons who use
wheelchairs (Table 2).
Monitors Attached to the Wheelchair Researchers
have explored attaching a custom data logger  or
biaxial  and tri-axial  accelerometers onto the
wheels of a wheelchair. Other preliminary research has
simply attached a smartphone (containing a gyroscope
and accelerometer) onto the armrest of a wheelchair [48,
49]. Considering the exponential growth of smartphone
ownership , this later approach in particular can
widely be used to capture certain mobility characteristics
such as average speed and distance travelled, functioning
in a similar manner to pedometers in persons who do not
use wheelchairs. However, despite these approaches being
o E E E E
ir S S S S
e M M M M
it , , , ,
rc lts .86 .86 .68 .67 .95 .93 .94
o su 0 0 0 0 0 0 0
t e =1 =1 =1 =1 = = =
n R r − r − r − r − r r r
m no its
o i r
.ltscooC illtccaao itannwm irttsnw irts ts ed
rryop taonm -onod ianom tseh itsa ithgw ifrtew inobm
A N D C W R L C
d e 2
k rso 4% k rs ( E 1 k rs
8 2 o 3 E n 0 o
.3 rr . .0 rr .6 .) e .0 rr
1 e 38 1 e % 0 n e 1 e %
(a n s
io a m
rm rm ta W rm ra
a a u ( ) a r
ft ) t q ts r e
lre ino lfre ice iend ram irts irw epp upp r-am
e ta e if b re w t u t S
pp uq pp cep om pp fte ihg fte ihg AM
U e U s C u L R L R P
−1in −1in 1. −1. from )to on .381% 1. from
−..1kgml −..1lkgm −.inml endg ·ilnm endg rskkeow iittsam .a2nd −.ilnm endg
ca ra ca ra ca ra
rm rm e e ed
A A en en itc
r r d
irts epp irts epp ,sEE iittyv rep
tW tU tW tU iton lca red
h h h h a a n
iRg iRg iRg iRg ilcp iscy ruo
A E ,e b
C E rd n fn
A A a ap co
7 C B P M t
relatively unobtrusive, they are unable to quantify the
intensity of activities being performed and are limited
in deriving accurate EE estimates. Conger et al. 
tried to address this limitation by using a PowerTap
Hub attached to the wheel of a wheelchair. The
measured hand rim propulsion power explained 48% of the
variance in predicting criterion EE. The authors
revealed three significant prediction models from this
laboratory protocol, with model 3 (incorporating power,
speed and heart rate) explaining the greatest variance
(87%). Seemingly, the incorporation of a physiological
signal significantly improved the prediction of EE.
However, we propose that a device attached to the
wheelchair cannot distinguish between self or assisted
propulsion, certainly not without complex analyses of
the raw acceleration outputs , and are unable to
quantify activity out of the wheelchair. Further, it is
common for persons who use wheelchairs to have
different chairs to participate in various sports or undertake
ACE as a mode of exercise. Therefore, a single device
attached to a wheelchair will fail to capture
moderate-tovigorous-intensity activity in structured exercise, likely
to contribute a large proportion towards TDEE.
Moreover, if a person uses a power-assisted wheelchair, signals
from devices attached to the chair will provide
erroneous measurements regarding upper body PAEE. These
limitations need to be considered when using this
approach to predict free-living PA/EE in persons that use
Body-Borne Accelerometers Waist-mounted
singlesensor devices, positioned within close proximity to an
individual’s centre of mass, have been the mainstay of
activity monitoring in cohorts without physical disabilities.
Single units worn on the waist can be limited for certain
types of upright behaviours that have a low ambulatory
component and may involve upper-body work . The
measurement error of waist-mounted devices is generally
related to the inability to detect arm movements as well as
static work (e.g. lifting, pushing, carrying loads). With
movement of persons that use wheelchairs predominantly
restricted to the upper-body, it is unsurprising that
stronger correlations between accelerometer outputs and
criterion measurements were reported for devices worn on
the upper arm and wrist, r = 0.83–0.93 and r = 0.52–0.93,
respectively (Table 2). While two studies [37, 41] have
found differences in the strength of correlations between
the left and right wrist, these discrepancies could be due
to hand dominance or the specific asymmetry of the
activities performed in these studies. The predominance
of research, however, suggests little to no difference
between dominant and non-dominant wrists [38, 39],
suggesting freedom/flexibility in selecting either wrist to
predict PA/EE in this population.
Combining data from two anatomical locations
seemingly does not yield substantial improvements in the
strength of correlations or EE estimation error [39, 54]. In
some research and development laboratories,
accelerometers have been arranged in parallel arrays and positioned
at various anatomical locations to monitor the types of
activity being performed by postural identification. Such
prototype PA monitors were developed to primarily target
specific population groups during rehabilitation, including
amputees  or inpatients with SCI . Devices with
multiple arrays have shown good specificity (92%),
agreement (92%) and sensitivity (87%) for the detection of
wheelchair propulsion in observational studies . Yet,
even when worn for a relatively short period of time,
participants self-reported moderate burden . These
monitors are relatively obtrusive and, due to reduced memory
capacity and battery life, are restricted to short monitoring
durations (<48 h). This is not in keeping with current end
user requirements of PA monitors. Multi-site prototype
arrays are also not typically available outside of the
developing laboratory, making validation by other researchers
A simpler set-up, the Physical Activity Monitoring
System (PAMS) , which incorporates a gyroscope-based
wheel rotation monitor (G-WRM) and one tri-axial
accelerometer attached to the arm or wrist, overcomes the
shortcomings of accelerometers attached to the
wheelchair alone. When this approach was recently evaluated
using a robust laboratory protocol and home-based
follow-up session, both the PAMS-arm and PAMS-wrist
estimated EE with small biases (M ± E < 10%) . Yet,
MAE for predicting EE in persons that use a wheelchair
remained elevated (>25%). Kooijmans et al.  also
assessed the utility of a tri-axial accelerometer (GT3X+)
attached to the wrist and spokes of a wheelchair.
However, rater observations reported less agreement (85%)
and specificity (83%) for wheelchair propulsion than
using multiple-arrays . Whilst less burdensome,
disagreement between GT3X+ (Actigraph, Pensacola, FL)
outputs and observers was greatest for propulsion on a
slope and being pushed whilst making excessive arm
movements. Therefore, it is likely that physiological
signals, such as heart rate, should be incorporated into the
prediction of EE to improve accuracy.
Heart rate (HR) is useful as a physiological variable as it
increases linearly and proportionately with exercise
intensity and thus oxygen uptake , at least in
individuals without disabilities. Keytel et al.  concluded that
EE can be accurately predicted from HR after adjusting
for age, sex, body mass and fitness. However, during
lower intensity PA, there is a weak relationship between
HR and EE . This is most likely due to small postural
changes causing alterations in stroke volume, or that HR
during low intensity PA is affected by external factors such
as psychological stress, stimulants, ambient temperature,
dehydration and illness . There are a number of ways
to use HR data to predict EE, one of the most promising
being the FLEX-HR method , which has previously
been used in persons with SCI [66, 67]. Despite recent
research into the use of various HR indices  and artificial
neural networks  in the prediction of V̇ O2 in
individuals with SCI, it is clear that the accurate prediction of EE
using HR is heavily reliant on individual calibration. Hayes
et al.  found that the variance in measured EE was
considerably improved using an individual calibration
(55%) compared to HR alone (8.5%) during five activities
of daily living in thirteen individuals with SCI. Considering
the type of activities performed, the large variations in
cardiovascular fitness and cardiovascular responses to
exercise stress persons who use wheelchairs, individual
specific HR-EE relationships are necessary for the accurate
prediction of EE using HR. This consideration is perhaps
even more important for persons with considerable
functional impairment or various disability aetiologies that
may disrupt the autonomic nervous system, such as
highlevel SCI (>T6).
New multi-sensor technologies, which include the
combination of physiological parameters and accelerometry,
have great potential for increased accuracy in assessing
EE as they incorporate and minimise the strengths and
weaknesses of physiological signals and accelerometry
alone. The use of multi-sensor devices has mostly been
limited to laboratory based validation of the SenseWear®
Armband (SWA) (BodyMedia Inc., Pittsburgh, PA),
which is worn on the upper arm, a preferential
anatomical location for the prediction of EE in persons that use
wheelchairs (Table 2). More detailed components and
specifications of this activity monitor have been
described elsewhere . It is clear that the proprietary
manufacturer’s algorithms intrinsic to the SWA device
are not appropriate to predict EE in persons that use
wheelchairs, with ICCs < 0.64 [40, 70, 71]. The
overestimation of EE by the SWA manufacturer’s model is likely
due to the movements typically performed by persons
that use wheelchairs (e.g. wheelchair propulsion and
ACE) not being included in predefined activity
categories. Hence, such activities are misclassified into more
strenuous types of PA.
Researchers have developed new EE prediction models
(SCI general and activity specific) for the SWA device
that have been cross-validated [40, 71]. Where MAE
statistics are available [40, 70, 71] weighted means were
calculated, with the SCI general (22.7%) and activity
specific (18.2%) models performing significantly better
than the manufacturer’s model (54.4%). Whilst these
findings provide encouragement for the use of the
SWA in persons that use wheelchairs with new
prediction models, Conger et al.  noticed that even when
using the SCI general model, the SWA tended to
overestimate EE (27 to 43%), whereas a wrist-mounted
accelerometer more accurately predicted EE (9 to 25%)
during wheelchair prolusion (Table 3). It is noteworthy
that the SWA utilizes upwards of twenty possible
output parameters, including heat flux, galvanic skin
response and temperature to predict EE. Individuals with
high level SCI (>T6) experience impaired
thermoregulatory function (reduced sweating response and
inability to dilate superficial vasculature ), which
might intrinsically effect the error when using SWA in
this population. Unfortunately, the acquisition of the
company BodyMedia by Jawbone in 2013 resulted in
the device being taken off the market and cessation of
all BodyMedia web applications. Despite considerable
improvements in EE prediction error it seems the
future use of this technology is limited.
The Actiheart (Cambridge Neurotechnology Ltd,
Papworth, UK) integrates an accelerometer and HR monitor
into a single-piece movement monitor. The Actiheart
(AHR) unit has been described in detail previously ,
along with the detailed branched modelling technique it
utilises to estimate PAEE through the combination of
HR and accelerometer counts . Previous work from
our research group  has assessed the performance of
this device in a controlled-laboratory environment with
a heterogeneous sample of persons who use wheelchairs.
Across all activities considerable mean absolute error
(MAE) was reported (51.4%) using the manufacturer’s
proprietary algorithms to predict PAEE. By using an
incremental ACE test, which permitted an individual
HR-EE relationship similar to that performed by Hayes
et al , individual calibration was incorporated and
MAE was considerably reduced to 16.8% across all
activities. Individual calibration has also been shown to
improve the prediction of EE estimations using this device
in free-living [76, 77] and laboratory settings  in
adults without disabilities during walking and running.
The sizeable improvement in EE prediction error in
persons who use wheelchairs with individual calibration
may be due to a larger degree of individual variance in
cardiovascular function and responses to exercise in this
population. Consequently, individual calibration of this
monitor is of upmost importance for the accurate
prediction of PAEE in persons that use wheelchairs.
Furthermore, incorporating individually calibrated HR and
acceleration data better captures the differing energy
costs of bespoke activities, despite similar acceleration
profiles, such as wheelchair propulsion up a gradient or
with additional load (e.g. shopping) .
m m su lag
EE (a EE (m E r E r
E a E a
Prediction Accuracy of Methodologies in Free Living
The majority of PA/EE validation research in this
population has been performed in a controlled-laboratory
environment but there is a paucity of free-living studies
(Table 4) primarily due to the practical difficulties or
expense associated with ‘gold standard’ EE measurement
(DLW). This method is not without limitations; for
example minimal information regarding frequency, duration
or intensity of activity can be obtained . Furthermore,
the estimation of EE is based on the assumption of a mean
respiratory exchange ratio (RER) of 0.85, indicative of a
standard western diet . Yet, carbohydrate and fat
oxidation has been shown to be altered with arm
compared to leg exercise  and in paraplegics compared to
non-disabled controls . These factors may lead to an
increased RER in persons that use wheelchairs, which
could violate the assumptions used in the prediction of EE
via the DLW technique.
Irrespective of this, Tanhoffer et al.  compared four
aforementioned prediction methods (SWA, FLEX-HR,
PARA-SCI, PASIPD) to DLW during habitual routines
over an extended 14-day period. The authors
demonstrated that the two best prediction methods were
PARA-SCI and FLEX-HR for both TDEE and PAEE
(Table 4). The SWA and PASIPD both performed
particularly poorly in the prediction of PAEE, displaying
considerable random error as demonstrated by the large
95% limits of agreement. It is noteworthy that the SWA
used the aforementioned error-prone manufacturer’s
model [40, 70, 71] but could be improved with the SCI
general EE prediction model developed by Hiremath et
al. . One limitation of the Tanhoffer et al  study
is that the length of PA monitoring period for each
prediction method varied compared to the criterion
method. Total EE collected over a two-week period for
the criterion DLW technique was divided by 14 to
estimate mean TDEE. However, the objective measures
(SWA and FLEX-HR) were only worn ≥12 h on two
separate days and subjective measures, the PARA-SCI
and PASIPD ask participants to recall the previous 3 and
7 days, respectively. This weakness in the experimental
design means it is difficult to identify whether the error is
intrinsic to each prediction method or simply an artefact
of the comparison between different days or time-periods.
In the absence of other suitable criterion free-living
methods, researchers have encouraged the simple
evaluation of the agreement and disagreement between
measures . Previous studies have compared prediction
methods to daily PA record scores over 7 days  and
PAEE estimated from the adapted PA compendium over
24 h . Again the PASIPD was poorly correlated with
the reference standard, whereas outputs from a tri-axial
Actiwatch demonstrated a stronger correlation, r = 0.51
. Nightingale and colleagues  supported their
earlier laboratory findings, demonstrating that the
Actiheart™ with individual HR calibration explained more of
the variance in free-living PAEE than using the Actiheart™
with proprietary algorithms. However, these analyses were
only performed on a subsample of participants (n = 8)
who had provided enough detailed information in PA logs
to allow accurate estimation of PAEE using the adapted
PA compendium for manual wheelchair users . This
compendium only describes the energy cost of 63
wheelchair activities compared to the 821 specific activities
included in the updated version of the compendium of
physical activities for adults without disabilities .
Consequently, coding of activities is less specific and
accuracy of data is reliant on the quality of the self-report
PA log. There are clear discrepancies between validating
objective tools in a controlled-laboratory and free-living
environments. Therefore, renewed efforts are required to
validate measurement tools in both settings to determine
Statistical Approaches, Analytical Considerations and
The majority of studies found strong associations between
criterion measurements and outputs from wearable
devices. However, in some instances where results from
Bland Altman methods are also available, considerable
random error has been reported [42, 66]. Where devices
have been validated over a wide range of activities, of
various intensities (in keeping with best practice guidelines
) a stronger correlation coefficient is likely.
Consequently we encourage researchers to conduct multi-trait
multi-method approaches , such as Bland Altman
methods to assess agreement  or report measurement
error . It is also important that authors are very clear
about what error calculations have been performed and
what error statistics are reported. Furthermore, it would
be advisable for the wider academic community to
produce a consensus statement addressing the clinical
limits for PA/EE assessment error for devices used in this
It is possible that predicting EE/PAEE from linear
regression equations may be too simple an approach to use
in examining complex movements or behaviours .
The activity protocols adopted by laboratory validation
studies cited here mostly focus around wheelchair
propulsion of various velocities. It is important to characterise
this behaviour, as it will likely make a significant
contribution to TDEE in free-living conditions (similar to
ambulation in adults without disabilities). But as push frequency
increases to match higher velocities, so too will
accelerometer outputs. Therefore, whilst it might be appropriate
to use linear regression methods to quantify PAEE
associated with wheelchair propulsion, this approach
might misclassify other types of physical activity. This
is highlighted by considerable increases in
measurement error for sedentary or atypical movements such
as folding clothes [38, 43]. Greater error in more
frequently performed low-intensity or sedentary
behaviours has potentially considerable implications for the
accurate determination of free-living EE in persons
who use wheelchairs. A more ecologically valid approach
would be to develop regression models based on a
smorgasbord of activities common in the everyday lives of
persons who use wheelchairs. It is possible that, by giving
more weight to everyday activities (i.e. household chores
or work-based tasks), such regression models may reduce
An alternative solution to regression models would be
to use new data analysis methodologies , including
hidden Markov models , artificial neural networks
[92, 93] and classification trees , which use the rich
information to classify certain activities and derive a more
accurate estimate of EE . To obtain such rich
information, the shortest possible epoch (1 s) should be selected
for activity monitor data collection , primarily to
maximise the original PA related bio-signal being retained.
Garcia-Masso et al.  recently developed and tested
classification algorithms based on machine learning using
accelerometers to identify specific activities performed by
persons who use wheelchairs. This is encouraging since
activity-specific EE algorithms developed for resting,
wheelchair propulsion, arm-ergometry and deskwork can
improve overall EE estimation [70, 97]. One important
consideration that remains to be addressed is, how well
objective measurement tools and associated algorithms
capture elevated energy expenditure during recovery from
MVPA (i.e. excess post-exercise oxygen consumption). It is
conceivable that a physiological signal is required to
accurately capture this information when acceleration signals
post-exercise might be similar to resting values. Future
research should consider; (i) applying and further developing
new data analysis techniques, (ii) using more ecologically
valid assessments that better resemble free-living
conditions for persons that use a wheelchair and, (iii) evaluating
the performance of EE prediction models during recovery
after exercise (which contributes to TDEE).
Some of the principal limitations of previous validation
studies are the relatively small sample sizes recruited,
the mixed aetiologies for wheelchair use and, use of EE
prediction algorithms without cross-validation. This is
likely due to difficulties associated with recruiting from
various disabled populations , and we encourage
research groups to work collaboratively to recruit larger
sample sizes. Using a diverse sample of participants and
aetiologies for wheelchair use has been widely adopted
[37, 39, 72, 83] and provides a robust model for the
assessment of EE in the wider population of individuals
who use wheelchairs, rather than a subgroup of that
population. When the development of regression
equations to predict EE and subsequent evaluation was
conducted on the same sample of participants [37, 42],
there is a tendency for the evaluation statistics to be
biased and overly optimistic . Cross-validation is
necessary, whereby the validity of developed algorithms are
assessed using an independent sample of participants
[54, 70]. We advocate employing a leave-one-out cross
validation analysis  which has been employed
previously [38, 43]. This permits an ‘independent’ assessment
of EE prediction algorithms, and is an optimal approach
when participant recruitment is particularly challenging.
Wearable Technology and Physical Activity Guidelines for
Persons who use Wheelchairs
The American College of Sports Medicine (ACSM) have
highlighted wearable technology as the top fitness trend
for 2016 . Available consumer devices (Apple Watch,
Microsoft Band, Fitbit Charge HR) are becoming
increasingly sophisticated, incorporating multi-sensor
technologies and are worn on the most appropriate anatomical
location (wrist) to predict EE in persons who use
wheelchairs. Apple recently announced at its Annual
Worldwide Developer’s Conference that they have developed
fitness tracking algorithms specifically for persons who
use wheelchairs. Such wearable devices have the potential
to provide wheelchair users with physical activity feedback
which is informative and motivating . The feasibility
of combining estimation methods should also be explored.
Greater context regarding the location, type and purpose
of physical activity behaviours are of huge importance in
public health research. More detailed information may be
achieved by combination of GPS and accelerometer
outputs, especially when also incorporating self-report
measures. This approach could help to understand specific
personal and environmental barriers to exercise, which
are numerous for persons who use wheelchairs .
It has been suggested that individuals with
disabilities should strive to meet PA guidelines of 150 min
of MVPA per week . These general population
guidelines were informed by epidemiological evidence,
using questionnaires, which capture the amount of activity
required above normal lifestyle activities. While minutes
per week represent an easy target for people to understand
and attain, only the PARA-SCI and multi-sensor devices
can currently be used in persons who use wheelchairs to
generate total accumulated MVPA per day/week.
Discrepancies have been shown between self-reported and
objectively measured PA [104, 105]. Consequently, a recent
review of data collected with accurate multi-sensor
devices in adults without disabilities has suggested
that ~1000 min per week of MVPA is a more
appropriate target . To our knowledge, only one paper
has attempted to establish MVPA cut-points for wrist
worn accelerometer outputs in persons who use
wheelchairs . However, accelerometer outputs alone
(without complex data processing techniques) cannot easily
detect the resistance of various movements that have
similar acceleration profiles i.e. arm-crank exercise at 70
revolutions per minute; with no resistance (light-intensity
activity) vs. 40 W (likely MVPA). Therefore, deriving
MVPA cut-points for single unit wrist/arm accelerometers
might have limited applicability, as direct outputs are
unable to differentiate the resistance of certain arm
movements (thus activity intensity) common in the everyday
lives of persons who use wheelchairs. As such, measuring
activity intensity is of utmost importance to accurately
estimate MVPA, above and beyond daily PAEE/EE.
Improvements in measurement techniques that capture
this specific variable would significantly help to inform
specific PA guidelines for persons with chronic disabilities
who use wheelchairs.
There is now a renewed impetus to translate progress in
measuring PA in adults without disabilities to persons
who use wheelchairs, with the techniques reviewed here
(i.e. self-report, physiological signals, accelerometry and
multi-sensor devices), displaying varying degrees of
success. Currently, selecting a PA assessment tool to use
in this population presents a challenging proposition for
clinicians and researchers alike due to differing outcome
variables of interest, practicality/usability of the tool and
population specific considerations. To help guide
decision-making, Fig. 1 was developed to provide a
systematic evaluation of the strengths and limitations of the
different measurement tools reported herein. The
PARA-SCI has been extensively developed and is the
most suitable self-report measure to predict time spent
performing various intensity activities. This methodology
also captures the type of activities being performed,
categorised as either LTPA or activities of daily living, which
provide useful behavioural information. Tri-axial
accelerometers worn on the wrist or arm are well tolerated and
relatively unobtrusive . They offer a promising
alternative to self-report methods for predicting PA/EE,
particularly when combined with devices attached to the
wheelchair or by incorporating complex data analysis
methodologies. Multi-sensor devices, with algorithms
developed specifically for the individual or generally for
persons who use wheelchairs, demonstrate considerably
improved error in the prediction of PA/EE during
controlled laboratory protocols. It is possible that due to
altered movement patterns and variations in
metabolically active mass, predicting PA/EE in persons that use
wheelchairs might be intrinsically more challenging.
However, building on the current progress outlined in
this review, we encourage the scientific community to
rise to the challenge and provide innovative solutions to
accurately predict free-living PA behaviours in this
population. This is particularly important given the
greater risk of non-communicable diseases, which are
often associated with reduced activity, in persons with
chronic physical disabilities who use wheelchairs.
The authors would like to thank Roger and Susan Whorrod, and the Medlock
Charitable Trust for their kind donations to the DisAbility Sport and Health
(DASH) research group at the University of Bath. The authors would like to
acknowledge Dr Andrew Siddall for his review.
No financial support was received for the conduct of this study or
preparation of this manuscript.
TEN conceived the review, which was revised with input from PCR, DT and
JLJB. TEN conducted the literature review and drafted the manuscript with
PCR also contributing to the manuscript preparation. The manuscript was
critically revised for important intellectual content by DT and JLJB. All
authors approved the final manuscript.
Tom Nightingale (TEN), Peter Rouse (PCR), Dylan Thompson (DT) and James
Bilzon (JLJB) declare that they have no conflict of interest.
1. Booth FW , Gordon SE , Carlson CJ , Hamilton MT . Waging war on modern chronic diseases: primary prevention through exercise biology . J Appl Physiol . 2000 ; 88 ( 2 ): 774 - 87 .
2. Haskell WL , Lee IM , Pate RR , Powell KE , Blair SN , Franklin BA , et al. Physical activity and public health: updated recommendation for adults from the American College of Sports Medicine and the American Heart Association . Med Sci Sports Exerc . 2007 ; 39 ( 8 ): 1423 - 34 .
3. Kesaniemi YA , Danforth E , Jensen MD , Kopelman PG , Lefebvre P , Reeder BA . Dose-response issues concerning physical activity and health: an evidencebased symposium . Med Sci Sports Exerc . 2001 ; 33 ( 6 ): S351 - 8 .
4. Brage S , Brage N , Franks PW , Ekelund U , Wareham NJ . Reliability and validity of the combined heart rate and movement sensor Actiheart . Eur J Clin Nutr . 2005 ; 59 ( 4 ): 561 - 70 .
5. Collins EG , Gater D , Kiratli J , Butler J , Hanson K , Langbein WE . Energy cost of physical activities in persons with spinal cord injury . Med Sci Sports Exerc . 2010 ; 42 ( 4 ): 691 - 700 .
6. Conger SA , Bassett Jr DR . A compendium of energy costs of physical activities for individuals Who use manual wheelchairs . Adapt Phys Act Q . 2011 ; 28 ( 4 ): 310 - 25 .
7. Butte NF , Ekelund U , Westerterp KR . Assessing physical activity using wearable monitors: measures of physical activity . Med Sci Sports Exerc . 2012 ; 44 : S5 - S12 .
8. Dionyssiotis Y , Petropoulou K , Rapidi CA , Papagelopoulos P , Papaioannou N , Galanos A , et al. Body composition in paraplegic men . J Clin Densitom . 2008 ; 11 ( 3 ): 437 - 43 .
9. Biering-Sorensen B , Kristensen IB , Kjaer M , Biering-Sorensen F. Muscle after spinal cord injury . Muscle Nerve . 2009 ; 40 ( 4 ): 499 - 519 .
10. Schofield WN . Predicting basal metabolic rate, new standards and review of previous work . Hum Nutr Clin Nutr . 1985 ; 39 Suppl 1 : 5 - 41 .
11. Gomes AID , Vigario PD , Mainenti MRM , Ferreira MD , Ribeiro BG , Soares ED. Basal and resting metabolic rates of physically disabled adult subjects: a systematic review of controlled cross-sectional studies . Ann Nutr Metab . 2014 ; 16 ( 4 ): 243 - 52 .
12. Buchholz AC , Pencharz PB . Energy expenditure in chronic spinal cord injury . Curr Opin Clin Nutr Metab Care . 2004 ; 7 ( 6 ): 635 - 9 .
13. Landsberg L , Young JB , Leonard WR , Linsenmeier RA , Turek FW . Is obesity associated with lower body temperatures? Core temperature: a forgotten variable in energy balance . Metab-Clin Exp . 2009 ; 58 ( 6 ): 871 - 6 .
14. Martin Ginis KA , Latimer AE , Buchholz AC , Bray SR , Craven BC , Hayes KC , et al. Establishing evidence-based physical activity guidelines: methods for the Study of Health and Activity in People with Spinal Cord Injury (SHAPE SCI) . Spinal Cord . 2007 ; 46 ( 3 ): 216 - 21 .
15. Buchholz AC , Martin Ginis KA , Bray SR , Craven BC , Hicks AL , Hayes KC , et al. Greater daily leisure time physical activity is associated with lower chronic disease risk in adults with spinal cord injury . Appl Physiol Nutr Metab . 2009 ; 34 ( 4 ): 640 - 7 .
16. Tudor-Locke CE , Myers AM . Challenges and opportunities for measuring physical activity in sedentary adults . Sports Med . 2001 ; 31 ( 2 ): 91 - 100 .
17. Shephard RJ . Limits to the measurement of habitual physical activity by questionnaires . Br J Sports Med . 2003 ; 37 ( 3 ): 197 - 206 .
18. Sallis JF , Saelens BE . Assessment of physical activity by self-report: status, limitations, and future directions . Res Q Exerc Sport . 2000 ; 71 ( 2 ): S1 - S14 .
19. Rimmer JH , Riley BB , Rubin SS. A new measure for assessing the physical activity behaviors of persons with disabilities and chronic health conditions: the Physical Activity and Disability Survey . Am J Health Promot . 2001 ; 16 ( 1 ): 34 - 45 .
20. Kayes NM , McPherson KM , Taylor D , Schluter PJ , Wilson BJK , Kolt GS . The Physical Activity and Disability Survey (PADS): reliability, validity and acceptability in people with multiple sclerosis . Clin Rehabil . 2007 ; 21 ( 7 ): 628 - 39 .
21. Kayes NM , Schluter PJ , McPherson KM , Taylor D , Kolt GS . The Physical Activity and Disability Survey-Revised (PADS-R): an evaluation of a measure of physical activity in people with chronic neurological conditions . Clin Rehabil . 2009 ; 23 ( 6 ): 534 - 43 .
22. Washburn RA , Zhu WM , McAuley E , Frogley M , Figoni SF . The physical activity scale for individuals with physical disabilities: development and evaluation . Arch Phys Med Rehabil . 2002 ; 83 ( 2 ): 193 - 200 .
23. Craig CL , Marshall AL , Sjostrom M , Bauman AE , Booth ML , Ainsworth BE , et al. International physical activity questionnaire: 12-country reliability and validity . Med Sci Sports Exerc . 2003 ; 35 ( 8 ): 1381 - 95 .
24. Ullrich PM , Spungen AM , Atkinson D , Bombardier CH , Chen Y , Erosa NA , et al. Activity and participation after spinal cord injury: state-of-the-art report . J Rehabil Res Dev . 2012 ; 49 ( 1 ): 155 - 74 .
25. Martin Ginis KA , Phang SH , Latimer AE , Arbour-Nicitopoulos KP . Reliability and Validity Tests of the Leisure Time Physical Activity Questionnaire for People With Spinal Cord Injury . Arch Phys Med Rehabil . 2012 ; 93 ( 4 ): 677 - 82 .
26. van der Ploeg HP , Streppel KRM , van der Beek AJ , van der Woude LHV , Vollenbroek-Hutten M , van Mechelen W. The physical activity scale for individuals with physical disabilities: test-retest reliability and comparison with an accelerometer . J Phys Act Health . 2007 ; 4 ( 1 ): 96 - 100 .
27. Martin Ginis KA , Latimer AE , Hicks AL , Craven BC . Development and evaluation of an activity measure for people with spinal cord injury . Med Sci Sports Exerc . 2005 ; 37 ( 7 ): 1099 - 111 .
28. Manns PJ , McCubbin JA , Williams DP . Fitness, inflammation, and the metabolic syndrome in men with paraplegia . Arch Phys Med Rehabil . 2005 ; 86 ( 6 ): 1176 - 81 .
29. de Groot S , van der Woude LHV , Niezen A , Smit CAJ , Post MWM . Evaluation of the physical activity scale for individuals with physical disabilities in people with spinal cord injury . Spinal Cord . 2010 ; 48 ( 7 ): 542 - 7 .
30. Wen CP , Wai JP , Tsai MK , Yang YC , Cheng TY , Lee MC , et al. Minimum amount of physical activity for reduced mortality and extended life expectancy: a prospective cohort study . Lancet . 2011 ; 378 ( 9798 ): 1244 - 53 .
31. Gebel K , Ding D , Chey T , Stamatakis E , Brown WJ , Bauman AE . Effect of moderate to vigorous physical activity on All-cause mortality in middleaged and older Australians . JAMA Intern Med . 2015 ; 175 ( 6 ): 970 - 7 .
32. Samitz G , Egger M , Zwahlen M. Domains of physical activity and all-cause mortality: systematic review and dose-response meta-analysis of cohort studies . Int J Epidemiol . 2011 ; 40 ( 5 ): 1382 - 400 .
33. Lee M , Zhu WM , Hedrick B , Fernhall B. Determining metabolic equivalent values of physical activities for persons with paraplegia . Disabil Rehabil . 2010 ; 32 ( 4 ): 336 - 43 .
34. Lewis JE , Nash MS , Hamm LF , Martins SC , Groah SL . The relationship between perceived exertion and physiologic indicators of stress during graded arm exercise in persons with spinal cord injuries . Arch Phys Med Rehabil . 2007 ; 88 ( 9 ): 1205 - 11 .
35. Westerterp KR . Assessment of physical activity: a critical appraisal . Eur J Appl Physiol . 2009 ; 105 ( 6 ): 823 - 8 .
36. Chen KY , Janz KF , Zhu W , Brychta RJ . Redefining the roles of sensors in objective physical activity monitoring . Med Sci Sports Exerc . 2012 ; 44 : S13 - 23 .
37. Washburn RA , Copay AG . Assessing physical activity during wheelchair pushing: validity of a portable accelerometer . Adapt Phys Act Q . 1999 ; 16 ( 3 ): 290 - 9 .
38. Garcia-Masso X , Serra-Ano P , Garcia-Raffi LM , Sanchez-Perez EA , LopezPascual J , Gonzalez LM . Validation of the use of Actigraph GT3X accelerometers to estimate energy expenditure in full time manual wheelchair users with spinal cord injury . Spinal Cord . 2013 ; 51 ( 12 ): 898 - 903 .
39. Learmonth YC , Kinnett-Hopkins D , Rice IM , Dysterheft JL , Motl RW . Accelerometer output and its association with energy expenditure during manual wheelchair propulsion . Spinal Cord . 2016 ; 54 ( 2 ): 110 - 114 .
40. Hiremath SV , Ding D. Evaluation of activity monitors in manual wheelchair users with paraplegia . J Spinal Cord Med . 2011 ; 34 ( 1 ): 110 - 7 .
41. Kiuchi K , Inayama T , Muraoka Y , Ikemoto S , Uemura O , Mizuno K. Preliminary study for the assessment of physical activity using a triaxial accelerometer with a gyro sensor on the upper limbs of subjects with paraplegia driving a wheelchair on a treadmill . Spinal Cord . 2014 ; 52 ( 7 ): 556 - 63 .
42. Nightingale TE , Walhim JP , Thompson D , Bilzon JL . Predicting physical activity energy expenditure in manual wheelchair users . Med Sci Sports Exerc . 2014 ; 46 ( 9 ): 1849 - 58 .
43. Nightingale TE , Walhin JP , Thompson D , Bilzon JLJ . Influence of accelerometer type and placement on physical activity energy expenditure prediction in manual wheelchair users . PLoS One . 2015 ; 10 ( 5 ): e0126086 .
44. Van Remoortel H , Giavedoni S , Raste Y , Burtin C , Louvaris Z , Gimeno-Santos E , et al. Validity of activity monitors in health and chronic disease: a systematic review . Int J Behav Nutr Phys Act . 2012 ; 9 : 84 .
45. Tolerico ML , Ding D , Cooper RA , Spaeth DM , Fitzgerald SG , Cooper R , et al. Assessing mobility characteristics and activity levels of manual wheelchair users . J Rehabil Res Dev . 2007 ; 44 ( 4 ): 561 .
46. Sonenblum SE , Sprigle S , Caspall J , Lopez R. Validation of an accelerometerbased method to measure the use of manual wheelchairs . Med Eng Phys . 2012 ; 34 ( 6 ): 781 - 6 .
47. Coulter EH , Dall PM , Rochester L , Hasler JP , Granat MH . Development and validation of a physical activity monitor for use on a wheelchair . Spinal Cord . 2011 ; 49 ( 3 ): 445 - 50 .
48. Fu JC , Liu T , Jones M , Qian G , Jan YK , Ieee . Characterization of wheelchair maneuvers based on noisy inertial sensor data: a preliminary study , 2014 36th Annual International Conference of the Ieee Engineering in Medicine and Biology Society (Embc). 2014 . p. 1731 - 4 .
49. Fu JC , Jones M , Liu T , Hao W , Yan YQ , Qian G , et al. A novel mobile-cloud system for capturing and analyzing wheelchair maneuvering data: a pilot study . Assist Technol . 2016 ; 28 ( 2 ): 105 - 14 .
50. Topol EJ , Steinhubl SR , Torkamani A. Digital medical tools and sensors . JAMA . 2015 ; 313 ( 4 ): 353 - 4 .
51. Conger SA , Scott SN , Bassett Jr DR . Predicting energy expenditure through hand rim propulsion power output in individuals who use wheelchairs . Br J Sports Med . 2014 ; 48 ( 13 ): 1048 - 53 .
52. Popp WL , Brogioli M , Leuenberger K , Albisser U , Frotzler A , Curt A , et al. A novel algorithm for detecting active propulsion in wheelchair users following spinal cord injury . Med Eng Phys . 2016 ; 38 ( 3 ): 267 - 74 .
53. Matthews CE , HagstrÖMer M , Pober DM , Bowles HR . Best practices for using physical activity monitors in population-based research . Med Sci Sports Exerc . 2012 ; 44 : S68 - 76 .
54. Hiremath SV , Ding D. Regression equations for RT3 activity monitors to estimate energy expenditure in manual wheelchair users . 33rd Conf Proc IEEE Eng Med Biol Soc . 2011 ; 7348 - 7351 .
55. Bussmann HBJ , Reuvekamp PJ , Veltink PH , Martens WLJ , Stam HJ . Validity and reliability of measurements obtained with an “activity monitor” in people with and without a transtibial amputation . Phys Ther . 1998 ; 78 ( 9 ): 989 - 98 .
56. van den Berg-Emons RJ , Bussmann JB , Haisma JA , Sluis TA , van der Woude LH , Bergen MP , et al. A prospective study on physical activity levels after spinal cord injury during inpatient rehabilitation and the year after discharge . Arch Phys Med Rehabil . 2008 ; 89 ( 11 ): 2094 - 101 .
57. Postma K , van den Berg-Emons HJG , Bussmann JBJ , Sluis TAR , Bergen MP , Stam HJ . Validity of the detection of wheelchair propulsion as measured with an activity monitor in patients with spinal cord injury . Spinal Cord . 2005 ; 43 ( 9 ): 550 - 7 .
58. Bussmann JBJ , Kikkert MA , Sluis TAR , Bergen MP , Stam HJ , Vaccaro AR . Effect of wearing an activity monitor on the amount of daily manual wheelchair propulsion in persons with spinal cord injury . Spinal Cord . 2009 ; 48 ( 2 ): 128 - 33 .
59. Hiremath SV , Intille SS , Kelleher A , Cooper RA , Ding D. Detection of physical activities using a physical activity monitor system for wheelchair users . Med Eng Phys . 2015 ; 37 ( 1 ): 68 - 76 .
60. Hiremath SV , Intille SS , Kelleher A , Cooper RA , Ding D. Estimation of energy expenditure for wheelchair users using a physical activity monitoring system . Arch Phys Med Rehabil . 2016 ; 97 ( 7 ): 1146 - 53 . e1.
61. Kooijmans H , Horemans HLD , Stam HJ , Bussmann JBJ . Valid detection of self-propelled wheelchair driving with two accelerometers . Physiol Meas . 2014 ; 35 ( 11 ): 2297 - 306 .
62. Keytel LR , Goedecke JH , Noakes TD , Hiiloskorpi H , Laukkanen R , Van der Merwe L , et al. Prediction of energy expenditure from heart rate monitoring during submaximal exercise . J Sports Sci . 2005 ; 23 ( 3 ): 289 - 97 .
63. Luke A , Maki KC , Barkey N , Cooper R , McGee D. Simultaneous monitoring of heart rate and motion to assess energy expenditure . Med Sci Sports Exerc . 1997 ; 29 ( 1 ): 144 - 8 .
64. Achten J , Jeukendrup AE . Heart rate monitoring-applications and limitations. Sports Med . 2003 ; 33 ( 7 ): 517 - 38 .
65. Ainslie PN , Reilly T , Westerterp KR . Estimating human energy expenditure-a review of techniques with particular reference to doubly labelled water . Sports Med . 2003 ; 33 ( 9 ): 683 - 98 .
66. Tanhoffer RA , Tanhoffer AIP , Raymond J , Hills AP , Davis GM . Comparison of methods to assess energy expenditure and physical activity in people with spinal cord injury . J Spinal Cord Med . 2012 ; 35 ( 1 ): 35 - 45 .
67. Hayes AM , Myers JN , Ho M , Lee MY , Perkash I , Kiratli BJ . Heart rate as a predictor of energy expenditure in people with spinal cord injury . J Rehabil Res Dev . 2005 ; 42 ( 5 ): 617 - 23 .
68. Coutinho ACB , Neto FR , Beraldo PSS . Validity of heart rate indexes to assess wheeling efficiency in patients with spinal cord injuries . Spinal Cord . 2014 ; 52 ( 9 ): 677 - 82 .
69. Garcia-Masso X , Serra-Ano P , Garcia-Raffi L , Sanchez-Perez E , Giner-Pascual M , Gonzalez LM . Neural network for estimating energy expenditure in paraplegics from heart rate . Int J Sports Med . 2014 ; 35 ( 12 ): 1037 - 43 .
70. Hiremath SV , Ding D , Farringdon J , Cooper RA . Predicting energy expenditure of manual wheelchair users with spinal cord injury using a multisensor-based activity monitor . Arch Phys Med Rehabil . 2012 ; 93 ( 11 ): 1937 - 43 .
71. Tsang K , Hiremath SV , Cooper RA , Ding D. Evaluation of custom energy expenditure models for SenseWear armband in manual wheelchair users . J Rehabil Res Dev . 2015 ; 52 ( 7 ): 793 - 804 .
72. Conger SA , Scott SN , Flynn JI , Tyo BM , Bassett DR . Validity and accuracy of physical activity monitors for estimating energy expenditure during wheelchair locomotion . San Francisco: American College of Sports Medicine Annual Meeting ; 2012 .
73. Petrofsky JS . Thermoregulatory stress during rest and exercise in heat in patients with a spinal-cord injury . Eur J Appl Physiol Occup Physiol . 1992 ; 64 ( 6 ): 503 - 7 .
74. Brage S , Brage N , Franks PW , Ekelund U , Wong MY , Andersen LB , et al. Branched equation modeling of simultaneous accelerometry and heart rate monitoring improves estimate of directly measured physical activity energy expenditure . J Appl Physiol . 2004 ; 96 ( 1 ): 343 - 51 .
75. Nightingale TE , Walhin JP , Thompson D , Bilzon JLJ . Predicting physical activity energy expenditure in wheelchair users with a multisensor device . BMJ Open Sport ExercMed . 2015 ; 1 ( 1 ) ;0:e000008 . doi:10.1136/bmjsem- 2015 - 000008 .
76. Villars C , Bergouignan A , Dugas J , Antoun E , Schoeller DA , Roth H , et al. Validity of combining heart rate and uniaxial acceleration to measure free-living physical activity energy expenditure in young men . J Appl Physiol ( 1985 ). 2012 ; 113 ( 11 ): 1763 - 71 .
77. Brage S , Westgate K , Franks PW , Stegle O , Wright A , Ekelund U , et al. Estimation of free-living energy expenditure by heart rate and movement sensing: a doubly-labelled water study . PLoS One . 2015 ; 10 ( 9 ): 19 .
78. Brage S , Ekelund U , Brage N , Hennings MA , Froberg K , Franks PW , et al. Hierarchy of individual calibration levels for heart rate and accelerometry to measure physical activity . J Appl Physiol . 2007 ; 103 ( 2 ): 682 - 92 .
79. Plasqui G , Westerterp KR . Physical activity assessment with accelerometers: an evaluation against doubly labeled water . Obesity . 2007 ; 15 ( 10 ): 2371 - 9 .
80. Bray GA . Energy expenditure using doubly labeled water: the unveiling of objective truth . Obes Res . 1997 ; 5 ( 1 ): 71 - 7 .
81. Tremblay JH , Peronnet F , Lavoie C , Massicotte D. Fuel selection during prolonged Arm and Leg exercise with C-13-glucose ingestion . Med Sci Sports Exerc . 2009 ; 41 ( 12 ): 2151 - 7 .
82. Jacobs KA , Burns P , Kressler J , Nash MS . Heavy reliance on carbohydrate across a wide range of exercise intensities during voluntary arm ergometry in persons with paraplegia . J Spinal Cord Med . 2013 ; 36 ( 5 ): 427 - 35 .
83. Warms CA , Whitney JD , Belza B. Measurement and description of physical activity in adult manual wheelchair users . Disabil Health J . 2008 ; 1 ( 4 ): 236 - 44 .
84. Ainsworth BE , Haskell WL , Herrmann SD , Meckes N , Bassett DR , Tudor-Locke C , et al. 2011 compendium of physical activities: a second update of codes and MET values . Med Sci Sports Exerc . 2011 ; 43 ( 8 ): 1575 - 81 .
85. Bassett Jr DR , Rowlands A , Trost SG . Calibration and validation of wearable monitors . Med Sci Sports Exerc . 2012 ; 44 ( 1 Suppl 1 ): S32 - 8 .
86. Post MW . What to Do with “moderate” reliability and validity coefficients? Arch Phys Med Rehabil . 2016 ; 97 ( 7 ): 1051 - 2 .
87. Bland JM , Altman DG . Statistical methods for assessing agreement between two methods of clinical measurement . Int J Nurs Stud . 2010 ; 47 ( 8 ): 931 - 6 .
88. Staudenmayer J , Zhu W , Catellier DJ . Statistical considerations in the analysis of accelerometry-based activity monitor data . Med Sci Sports Exerc . 2012 ; 44 ( 1 Suppl 1 ): S61 - 7 .
89. Strath SJ , Pfeiffer KA , Whitt-Glover MC . Accelerometer use with children, older adults, and adults with functional limitations . Med Sci Sports Exerc . 2012 ; 44 : S77 - 85 .
90. Clark CCT , Barnes CM , Stratton G , McNarry MA , Mackintosh KA , Summers HD. A review of emerging analytical techniques for objective physical activity measurement in humans . Sports Med . 2017 ; 47 ( 3 ): 439 - 447 .
91. Pober DM , Staudenmayer J , Raphael C , Freedson PS . Development of novel techniques to classify physical activity mode using accelerometers . Med Sci Sports Exerc . 2006 ; 38 ( 9 ): 1626 - 34 .
92. Staudenmayer J , Pober D , Crouter S , Bassett D , Freedson P. An artificial neural network to estimate physical activity energy expenditure and identify physical activity type from an accelerometer . J Appl Physiol . 2009 ; 107 ( 4 ): 1300 - 7 .
93. Trost SG , Wong WK , Pfeiffer KA , Zheng YL . Artificial neural networks to predict activity type and energy expenditure in youth . Med Sci Sports Exerc . 2012 ; 44 ( 9 ): 1801 - 9 .
94. Bonomi AG , Plasqui G , Goris AHC , Westerterp KR . Improving assessment of daily energy expenditure by identifying types of physical activity with a single accelerometer . J Appl Physiol . 2009 ; 107 ( 3 ): 655 - 61 .
95. Heil DP , Brage S , Rothney MP. Modeling physical activity outcomes from wearable monitors . Med Sci Sports Exerc . 2012 ; 44 : S50 - 60 .
96. Garcia-Masso X , Serra-Ano P , Gonzalez LM , Ye-Lin Y , Prats-Boluda G , Garcia-Casado J. Identifying physical activity type in manual wheelchair users with spinal cord injury by means of accelerometers . Spinal Cord . 2015 ; 53 ( 10 ): 772 - 7 .
97. Hiremath SV , Ding D , Farringdon J , Vyas N , Cooper RA . Physical activity classification utilizing SenseWear activity monitor in manual wheelchair users with spinal cord injury . Spinal Cord . 2013 ; 51 ( 9 ): 705 - 9 .
98. Yilmaz DDCB . Recruitment of spinal cord injury patients to clinical trials: challenges and solutions . Top Spinal Cord Inj Rehabil . 2006 ; 11 ( 3 ): 12 - 23 .
99. Hastie Y , Tibshirani R , Friedman JH . The elements of statistical learning; data mining , inference, and prediction. 2nd ed . New York : Springer; 2009 .
100. Thompson WR . Worldwide survey of fitness trends for 2016. 10th Anniversary Ed ACSM's Health Fit J . 2015 ; 19 ( 6 ): 9 - 18 .
101. Western MJ , Peacock OJ , Stathi A , Thompson D. The understanding and interpretation of innovative technology-enabled multidimensional physical activity feedback in patients at risk of future chronic disease . PLoS One . 2015 ; 10 ( 5 ): 13 .
102. Rimmer JH , Schiller W , Chen MD . Effects of disability-associated Low energy expenditure deconditioning syndrome . Exerc Sport Sci Rev . 2012 ; 40 ( 1 ): 22 - 9 .
103. Nash MS , Cowan RE , Kressler J. Evidence-based and heuristic approaches for customization of care in cardiometabolic syndrome after spinal cord injury . J Spinal Cord Med . 2012 ; 35 ( 5 ): 278 - 92 .
104. Prince SA , Adamo KB , Hamel ME , Hardt J , Gorber SC , Tremblay M. A comparison of direct versus self-report measures for assessing physical activity in adults: a systematic review . Int J Behav Nutr Phys Act . 2008 ; 5 : 56 .
105. Steene-Johannessen J , Anderssen SA , Van der Ploeg HP , Hendriksen IJM , Donnelly AE , Brage S , et al. Are self-report measures able to define individuals as physically active or inactive ? Med Sci Sports Exerc . 2016 ; 48 ( 2 ): 235 - 44 .
106. Thompson D , Batterham AM , Peacock OJ , Western MJ , Booso R. Feedback from physical activity monitors is not compatible with current recommendations: A recalibration study . Prev Med . 2016 ; 91 : 389 - 94 .
107. Learmonth YC , Kinnett-Hopkins D , Rice IM , Dysterheft JL , Motl RW . Accelerometer output and its association with energy expenditure during manual wheelchair propulsion . Spinal Cord . 2016 ; 54 ( 2 ): 110 - 4 .
108. Warms CA , Belza BL . Actigraphy as a measure of physical activity for wheelchair users with spinal cord injury . Nurs Res . 2004 ; 53 ( 2 ): 136 - 43 .
109. Conger SA , Scott SN , Fitzhugh EC , Thompson DL , Bassett DR . Validity of physical activity monitors for estimating energy expenditure during wheelchair propulsion . J Phys Act Health . 2015 ; 12 ( 11 ): 1520 - 6 .