Title
Step accumulation per minute epoch is not the same as cadence for free-living adults
Authors Names
Philippa Margaret Dall1, Paul Robert Walker McCrorie2, Malcolm Howard Granat1, Benedict
William Stansfield1.
Author Affiliations
1
School of Health and Life Sciences, Glasgow Caledonian University, Glasgow, UK, G4
0BA.
2
Social & Public Health Sciences Unit, Medical Research Council, Glasgow, UK, G12 4RZ.
Corresponding author
Dr Philippa Dall
Senior Research Fellow, School of Health and Life Sciences,
Glasgow Caledonian University,
Cowcaddens Road.
Glasgow, UK
G4 0BA,
(+44) 141 3318003 (phone)
(+44) 141 3318112 (fax)
philippa.dall@gcu.ac.uk
Running Title
step accumulation versus cadence in adults
Disclosure of Funding
This study was funded by the School of Health and Life Sciences, Glasgow Caledonian
University.
Conflict of interest statement
Malcolm Granat is a co-inventor of the activPAL activity monitor, and director of PAL
technologies. The other authors report no conflict of interest.
Abstract
Purpose: The term cadence has been used interchangeably to describe both the rate of
stepping and the number of steps in a minute epoch. This is only strictly true if walking is
continuous within that epoch. This study directly compared these two outcomes in minute
epochs of data from free-living adults to assess the scale of any difference between them.
Methods: A convenience sample of healthy adults wore an activPAL activity monitor for
seven days. The event record output of the activPAL, providing the start time and duration of
each stride to the nearest 0.1s, was used to calculate step accumulation (number of steps),
duration of walking and cadence (number of steps/ duration of walking) for each minute of
measurement.
Results: Data from 117 individuals (78 female, mean age 46±16 years, mean BMI 24.9±3.7
kg·m-2) were analysed. Twenty-one percent of minutes (n=310/day) contained walking. The
distribution (most minutes less than 40 steps/min) and mean (34±9 steps/min) of step
accumulation, was very different from that of cadence (most minutes between 60-100
steps/min, mean 76±6 steps/min). Only 12% of minutes with stepping were walked
continuously, while 69% of minutes with stepping contained less than 30s of walking. This is
key to the difference between step accumulation and cadence, and means that cadence cannot
be reconstructed from step accumulation without also knowing the duration that was walked.
Conclusion: Step accumulation, the number of steps in a fixed period of time, and cadence,
the rate of stepping whilst walking, are not interchangeable outcome measures. It is vitally
important that unambiguous terminology is used to describe the rate of stepping so that the
outcomes of studies can be correctly interpreted.
Key Words step rate; objective measurement; public health; accelerometer; outcome
measures; ambulation
Introduction
Paragraph Number 1 Cadence (rate of stepping) can be measured by body-worn sensors
during habitual physical activity. Cadence, as an outcome measure, is of interest as it can be
used to indicate the intensity of walking activity (18), and act as a proxy of usual walking
speed, which is predictive of mortality in older adults (16). Public health guidelines for
physical activity refer both to the duration and the intensity of activity, for example
recommending that adults spend 30 minutes a day in physical activity that is at least moderate
intensity (9). There is a growing consensus that a cadence of 100 steps/min represents a
reasonable minimum level for moderate physical activity in adults and adolescents
(1,2,8,11,13,19). These studies measured indirect energy expenditure using portable
calorimetry while participants walked on a treadmill or over ground across a range of speeds.
For each measurement point, participants walked for at least 6 minutes, and cadence was
therefore assessed during continuous periods of walking.
Paragraph Number 2 Threshold values for cadence have recently been applied to data
collected from large population based surveys (17). The National Health and Nutrition
Examination Survey (NHANES) from 2005-2006 collected objective accelerometer data
from a sub-sample of 3744 participants, using an Actigraph activity monitor. Data was
provided as the number of steps taken within successive minute-long epochs, and was
presented using the term cadence. Within this population, the vast majority of minutes that
contained any stepping (95%) had less than 60 steps in the minute-epoch. There was a
relatively small number of minutes of those containing any stepping (1%) with more than 100
steps in the minute-epoch. This was perhaps unexpected, given that observational studies of
apparently healthy adults walking, have demonstrated that many people do at times walk
continuously with cadence greater than 100 steps/min in free-living conditions (18).
Paragraph Number 3 In using the term cadence for the number of steps recorded within a
minute-long epoch, there is an underlying assumption that walking was continuous
throughout the minute. Non-walking breaks within that minute, however, would lead to a
situation where the number of steps recorded was not identical to the actual cadence of
walking. In that case, the data presented by the Actigraph represented the accumulation of the
number of steps taken within the minute (step accumulation), but not the actual rate at which
those steps were taken (cadence). An analogy would be driving a car at a speed of 30 miles
per hour. If driven at that speed for a whole hour, the distance travelled (analogous to steps)
would be 30 miles, and the speed (analogous to cadence) would also be 30 miles/hour.
However if the car was only driven for half an hour, the speed (i.e. cadence) would still be 30
miles/hour, but the distance travelled (i.e. steps) would only be 15 miles.
Paragraph Number 4 The validity with which cadence and step accumulation can be used as
interchangeable terms, depends on the relative proportion of minutes with continuous
walking to those with non-continuous walking, and how close the duration of non-continuous
walking is to the full minute. When using an event-based approach, where data is analysed as
a set of events representing continuous periods of walking rather than artificially imposed
epochs (5), studies have shown that walking is often comprised of short events (3,12). For
example, it has been found that the mean length of a walking event was of the order of a
minute (3), indicating that there are many walking events of less than 1 minute duration. It is
therefore possible that these short walking events would have a large effect on the
relationship between cadence and step accumulation.
Paragraph Number 5 The purpose of this analysis was to quantify the scale of any
discrepancy between step accumulation and cadence, by directly comparing the two outcome
measures of using data from an activity monitor which records the duration of walking as
well as step count. We hypothesise that step accumulation will be lower than cadence in some
epochs, but are unclear as to how many epochs this will effect or how different the two
outcome measures will be.
Method
Paragraph Number 6 This study is an analysis of data collected from healthy adults during
their usual activities. Throughout the article we define step accumulation as the number of
steps taken within a minute-long epoch, regardless of the actual time spent walking within
that minute. We define cadence as the number of steps within that minute divided by the
duration of walking in that minute, representing the actual stepping rate within the minute.
Both step accumulation and cadence have the same units (steps/min).
Paragraph Number 7 A convenience sample of 121 healthy adults were recruited via staff
working in a Scottish University, and data was collected between October 2007 and October
2009. Participants were included if they were aged over 18, lived in the community, and had
no mobility problems. Ethical approval was granted by the School of Health & Social Care
Research Ethics Committee, Glasgow Caledonian University, and all participants provided
written informed consent.
Paragraph Number 8 Physical activity data was collected using the activPAL activity
monitor (Pal Technologies Ltd, Glasgow). This small, light, monitor is worn on the front of
the thigh, and attached using double sided adhesive sticky pads. The device records
acceleration at a 10Hz sampling rate, and proprietary software classifies the data into
categories of sit/lie, stand and walk. The data is time stamped to the nearest 0.1s, and the start
time and duration of each step can be provided in the output. The monitor is valid for posture
allocation, step count and cadence in adults and older adults (4,6,7,14).
Paragraph Number 9 On day one, participants met with a researcher and were provided with
the monitor and instructed in its use. Basic demographic data (age, sex, BMI, employment
status) were collected from each participant. Participants were asked to start wearing the
monitor that day before midnight, and to then wear the monitor for seven consecutive days
whilst going about their usual activities. Monitors were worn overnight but not during waterbased activities. Participants were provided with additional sticky pads, and advised to
change them each day. On day nine, participants removed the monitor and returned it to the
researcher, data was analysed from midnight to midnight for days two to eight, inclusive.
Paragraph Number 10 Data from the activPAL activity monitor was downloaded using
proprietary software (PAL Professional v5.9.1.1). Data was visually inspected to ensure that
the monitor had been worn, and a participant was included if there was data for at least four
days. Data output from the event record provided by the activPAL software was used for this
analysis. This output provided a series of records representing a sitting event, a standing
event, or a stride. For each record the start time and duration were reported to the nearest
0.1s. Custom software (written in Matlab, Mathworks) used the start time and duration of
each stride to divide the data into minute-long epochs starting at midnight on day two. Each
stride has a duration, and we assumed the stride count to be equally spread across that
duration. For each stride occurring entirely within an epoch, stride count for the epoch was
increased by 1, and the duration of that stride was added to the total duration of walking in
that epoch. For strides that occurred across epochs, both the stride count and the duration of
the stride were split between the two epochs in proportion to the duration in each. For
example, if 30% of a 2 second long stride occurred in the first epoch, the stride count of that
epoch would be increased by 0.3 (30% of the stride count of ‘1’), and the duration of walking
in the epoch would be increased by 0.6s (30% of the duration of 2s). The number of strides
was then doubled to provide the number of steps in each minute epoch.
Paragraph Number 11 Three primary outcome measures were derived for each minute
epoch, step accumulation (the total number of steps in the epoch), duration of walking, and
cadence (total number of steps in the epoch/duration of walking). The difference between
cadence and step accumulation (cadence – step accumulation) was also calculated for each
minute epoch. Additionally, for minute epochs where the step accumulation and cadence
agreed (i.e minutes where walking was continuous) the cadence was reported as “agreement”
cadence.
Paragraph Number 12 For the whole sample, the number of minutes analysed, number of
minutes with walking, and proportion of those minutes where walking was continuous were
calculated. For minute-epochs containing walking, the frequency distribution of minuteepochs was reported for the outcome measures, step accumulation, cadence, and duration of
walking. Although not normally distributed, mean and standard deviation were used to
illustrate the central tendency of the outcome measures. To assess whether there was a
relationship between step accumulation and cadence, a scatter graph was created. As there
were a large number of points to be graphed, a random selection of 1 in 64 data points
(~4,000) were used. The frequency distribution of cadence was also reported for those
minute-epochs with continuous walking (i.e. when step accumulation was equal to cadence).
Finally, to assess the impact of using step accumulation or cadence on the public health
message, the number of minutes for each participant which had a value of 100 steps/minute
or more were calculated both for step accumulation and for cadence. The actual duration of
walking in minutes with a cadence of over 100 steps/minute was also calculated for each
participant. Data were not normally distributed, and were presented as median and interquartile range for the population, and were compared using related-samples Wilcoxon signed
rank tests.
Results
Paragraph Number 13 Four participants were excluded as they had fewer than 4 days of
data. Therefore, the multiday physical activity records of 117 participants (78 female) were
included within the analysis. The participants had a mean age of 46 ± 16 years with a wide
range (23-82 years old). Mean BMI (24.9 ± 3.7 kg·m-2) was within the normal category,
ranging from underweight to obese (17.7-38.4 kg·m-2). Ninety-six participants were in
employment, and 83 of them worked full time.
Paragraph Number 14 A mean of 6.8 days of physical activity was recorded from each
participant giving a total of 797 days of data. The mean daily number of minutes that
included walking was 310, representing 21.6% of all minutes within each 24 hour period. In
total, there were 247,453 minutes with stepping and an associated 8,424,866 steps.
Participants walked an average of 10,627 ± 4,157 steps/day ranging from 3,389 to 32,244
steps/day, and just over half the participants (n=59) took an average of more than 10,000
steps/day.
Paragraph Number 15 The distribution of step accumulation and cadence across the range of
values were very different (figure 1). Step accumulation was distributed across the range 0180 steps/min, with the largest contribution to the number of minutes walking at the lower
end of the step accumulation range, below 40 steps/min (table 1). However, there was a small
peak in the distribution of step accumulation at about 100-120 steps/min. The mean (±
standard deviation) step accumulation across all participants was 34 ± 9 steps/min. Cadence
was distributed very differently, across the range 20-180 steps/min, with stepping
concentrated around 50-100 steps/min. The mean cadence across all participants was 76 ± 6
steps/min.
Paragraph Number 16 The difference between cadence and step accumulation for each
minute ranged between 0 and 170 steps/min, similar to the ranges of the individual variables.
As illustrated by a random sample of points (figure 2), there was an upper limit on the value
of step accumulation when walking occurred during the whole minute, and at higher cadences
(> 140 steps/min) step accumulation tended to agree with cadence. However, within these
limits, there was no apparent relationship between the step accumulation and cadence of
individual minutes.
Paragraph Number 17 Walking was continuous in only 12% (38 ± 28 min/day) of minutes
with walking (figure 3). A large proportion of the minutes with stepping contained only a
short duration of walking within that minute, for example 32% of minutes with stepping
contained less than 10s of stepping, whilst 69% of minutes contained less than 30s of
stepping. Where a minute contained continuous walking there was agreement between step
accumulation and cadence. The distribution of the cadence for these minutes demonstrated a
concentration of outcomes between 90-130 steps/min (Figure 4). The mean cadence across all
participants for those minutes with continuous stepping was 109 ± 9 steps/min, over three
times that of the average step accumulation (34 steps/min) and approximately fifty percent
higher than the mean value of cadence (76 steps/min).
Paragraph Number 18 There were significantly fewer minutes a day with a step
accumulation of over 100 steps/min (median (interquartile range), 29 (26) min, p<0.01)
compared with minutes with a cadence greater than 100 steps/min (65 (41) min). As indicated
above, many of those minutes will not have contained a full minute of walking. However, the
duration of walking within minutes with a cadence greater than 100 steps/min (39 (34) min)
was also significantly higher than the number of minutes with a step accumulation greater
than 100 steps/min (p<0.001).
Discussion
Paragraph Number 19 The term cadence has widespread use in the literature when
describing stepping activity. The rate of stepping per time interval is considered important as
it is indicative of how intensively someone is engaging in physical activity. However, the
term cadence has been used to describe both an output that is the total number of steps within
a minute epoch (step accumulation), and an output that is the rate of steps taken within a
period of walking (cadence). It is clear from this direct comparison of the two concepts on the
same habitual physical activity data, that step accumulation and cadence are very different
outcome measures. They are significantly different both in terms of their distribution, and the
summative information that can be derived from the data. For example, mean cadence in this
population (76 steps/min) was twice that of mean step accumulation (34 steps/min).
Paragraph Number 20 The key factor in the difference between these distributions appears
to be the amount of time spent stepping within each minute. Over 30% of minutes with
walking activity contained less than 10s of stepping. In these cases, the step accumulation
will be low compared to the cadence. This is particularly important when applying a
threshold based on continuous walking, to judge the intensity of walking. A suggested
minimum threshold for walking at moderate intensity is 100 steps/min. However, with so few
minutes with close to continuous walking, step accumulation does not adequately judge the
time people spend walking at a cadence above this threshold. The number of minutes with a
step accumulation above 100 steps/min was significantly fewer than both the number of
minutes and duration of walking with a cadence above 100 steps/min.
Paragraph Number 21 In an individual minute, the difference between step accumulation
and cadence varied considerably, between 0 and 170 steps/min. There was also no evident
relationship between step accumulation and cadence in any particular minute, meaning that
imputing cadence from step accumulation was not feasible. As this difference was related to
duration of walking, there is no direct way of reconstructing cadence from step accumulation
without also knowing the duration of stepping, an outcome which is not available to those
instruments which have previously been used to report step accumulation (e.g. Actigraph
(18)).
Paragraph Number 22 From the definitions of step accumulation and cadence, it follows that
these two outcome measures were identical when the entire minute epoch was continuous
walking. The distribution of agreement cadence for those minutes with continuous walking
was different from both that of step accumulation and cadence across all minutes. The mean
value of agreement cadence (109 steps/min) was considerably higher than both cadence (76
steps/min) and step accumulation (34 steps/min). The difference in mean values of cadence
and agreement cadence might be explained by a difference in the purpose of walking bouts.
Full minute walking bouts might have been associated with purposeful unhindered travel,
whereas incomplete minute walking bouts might have been associated with movement about
the home or work place with a need to move around obstacles, or to stop to perform tasks, all
of which may be performed at a lower cadence. Continuous walking in an environment free
of barriers (a park) has been shown to allow participants to walk at a higher cadence and with
fewer breaks in walking, than in an urban environment (15), which may also affect the
distribution of cadence for minutes of continuous walking. Future studies simultaneously
reporting cadence with either walking duration or step accumulation, could provide insight
into the role of light, broken-up or household activity on health and physical function.
Paragraph Number 23 The distribution of cadence per minute epoch has not been previously
reported, but the distribution of step accumulation has been previously reported for the
NHANES national survey of the US population (17). Although the shape of distribution was
similar to that found in this study, the proportion of walking minutes with step accumulation
less than 60 steps/min was higher in the NHANES data (95%) compared with this study
(80%). Interestingly we found 10% of minutes with step accumulation greater than 100
steps/min, in contrast to just 1% in the NHANES data. This could have been due to the nature
of our sample, as the group contained a number of people who went running recreationally,
increasing the number of minutes with a high step accumulation. However, it should also be
noted that the proportion of minutes during the day with any walking in was much lower in
our study (310 min), compared to the NHANES data (554 min). Different instruments were
used to collect the data, and while both have similar percentage accuracy for detecting steps
(10,14), the proportional difference in low levels of step accumulation might suggest that
there were more minutes with a small number of steps reported by the Actigraph record than
the activPAL.
Paragraph Number 24 Whilst the study had a relatively small sample size (n=117), drawn
from a single source, this analysis represents new insight into the implications of methods
used to describe physical activity. The strengths of this analysis include the use of an
objective and validated monitoring tool capable of measuring both step count and duration of
stepping, to allow a direct comparison of the outcome measures of step accumulation and
cadence within this population. Participants were all adults without any notable mobility
impairment or disease state that restricted walking ability, and the physical activity performed
was all self-selected without any restriction or encouragement imposed by the study.
Paragraph Number 25 The distribution of step accumulation and cadence were very
different, and summative output, such as mean values, were also different (step accumulation,
34 steps/min; cadence 76 steps/min). Only 12% of minutes with stepping were walked
continuously, while 69% of minutes with stepping had less than 30s of stepping time,
providing further evidence that stepping in free-living is mostly composed of broken up
‘household’ type ambulation, rather than purposeful directed walking. Differences between
step accumulation and cadence were due to the duration of walking within that minute, and
without reference to such duration information, were apparently unrelated. This means that
data cannot be reliably scaled between the two measurements, and it is not possible to
directly compare reported outcomes of step accumulation and cadence when reported in the
literature.
Paragraph Number 26 Step accumulation, the number of steps in a fixed period of time, and
cadence, the rate of stepping while walking, are not interchangeable outcome measures. It is
vitally important that unambiguous terminology is used to describe the rate of stepping so
that the outcomes of studies can be correctly interpreted.
Acknowledgements
This study was funded by the School of Health & Life Sciences, Glasgow Caledonian
University. Data was collected by Sebastien Chastin, Elaine Coulter, Philippa Dall, Margaret
Grant, Gillian MacLellan, and Cormac Ryan. The custom software for this analysis was
written by Dr Benedict Stansfield. Malcolm Granat is a co-inventor of the activPAL activity
monitor, and director of PAL technologies. The other authors report no conflict of interest.
The results of the current study do not constitute endorsement by ACSM.
References
(1) Abel M, Hannon J, Mullineaux D, Beighle A. Determination of step related thresholds
corresponding to physical activity intensity classifications in adults. J Phys Activ Health
2011;8:45-51.
(2) Beets MW, Agiovlastitis S, Fahs CA, Ranadive SM, Fernhall B. Adjusting step count
recommendations for anthropometric variations in leg length. J Sci Med Sport 2010;13:50912.
(3) Chastin SFM, Dall PM, Tigbe WW, Grant PM, Ryan CG, Rafferty D, Granat MH.
Compliance with physical activity guidelines in a group of UK based postal workers using an
objective monitoring technique. Eur J App Physiol 2009;106(6):893-9.
(4) Godfrey A, Culhane KM, Lyons GM. Comparison of the performance of the activPAL™
Professional physical activity logger to a discrete accelerometer-based activity monitor. Med
Eng Phys. 2009;29:930-4.
(5) Granat MH. Event-based analysis of free-living behaviour. Physiol Meas 2012;33:1785800.
(6) Grant PM, Dall PM, Mitchell SL, Granat MH Activity monitor accuracy in measuring
step number and cadence in community-dwelling older adults. J Aging Phys Act 2008;
16:201-14.
(7) Grant PM, Ryan CG, Tigbe WW, Granat MH. The validation of a novel activity monitor
in the measurement of posture and motion during everyday activities. Brit J Sports Med
2006;40(12):992-7.
(8) Harrington DM, Dowd KP, Tudor-Locke C, Donnelly AE. A steps/minute value for
moderate intensity physical activity in adolescent females. Pediatr Exerc Sci 2012;24:399408.
(9) Haskell WL, Lee I-M, Pate RR, et al. Physical activity and public health: updated
recommendations for adults from the American College of Sports Medicine and the
American Heart Association. Circulation 2007;116(9);1081-93.
(10) Le Masurier GC, Lee SM, Tudor-Locke C. Motion sensor accuracy under controlled and
free-living circumstances. Med Sci Sports Exerc 2004;36:905-10.
(11) Marshall SJ, Levy SS, Tudor-Locke CE, et al. Translating physical activity
recommendations into a pedometer-based step goal. Am J Prev Med 2009;36(5):410-5.
(12) Orendurff MS, Schoen JA, Bernatz GC, Segal AD, Klute GK. How humans walk: bout
duration, steps per bout, and rest duration. J Rehabil Res Dev 2008;45(7):1077-90.
(13) Rowe DA, Welk GJ, Heil DP, Mahar MT, Kemble CD, Calabro MA, Camenisch K.
Stride rate recommendations for moderate intensity walking. Med Sci Sports Exerc
2011;43(2):312-8.
(14) Ryan CG, Grant PM, Tigbe WW, Granat MH. The validity and reliability of a novel
activity monitor as a measure of walking. Brit J Sports Med 2006;40:779-84.
(15) Sellers CE, Grant PM, Ryan CG, O’Kane C, Raw K, Conn D. Take a walk in the park?
A cross-over pilot trial comparing brisk walking in two different environments: park and
urban. Prev Med 2012;55(5):438-43.
(16) Studenski S, Perera S, Patel K, et al. Gait speed and survival in older adults. JAMA
2011;305(1):50-8.
(17) Tudor-Locke C, Camhi SM, Leonardi C, Johnson WD, Katzmarzyk PT, Earnest CP,
Church TS. Patterns of adult stepping cadence in the 2005-2006 NHANES. Prev Med
2011;53:178-81.
(18) Tudor-Locke C, Rowe DA. Using cadence to study free-living ambulatory behaviour.
Sports Med 2012; 42(5):381-98.
(19) Tudor-Locke C, Sisson SB, Collova T, Lee SM, Swan PD. Pedometer-determined step
count guidelines for classifying walking intensity in a young ostensibly healthy population.
Can J Appl Physiol 2005;30:666-76.
Figure Captions
Figure 1
Mean minutes per day for step accumulation (open bars) and cadence (solid bars) across
step rate bands.
Figure 2
Relationship of cadence and step accumulation for a random sample of 1 in every 64
minutes (~4,000 data points) from all minutes with walking.
Figure 3
Mean minutes per day across duration of walking in that minute.
Figure 4
Mean minutes per day across cadence bands for those minutes with complete walking.