Accelerometer Data Collection and Processing Criteria to
Assess Physical Activity and Other Outcomes: A Systematic
Review and Practical Considerations
Jairo H. Migueles
1,*
, Cristina Cadenas-Sanchez
1
, Ulf Ekelund
2,3
, Christine Delisle
Nyström
4
, Jose Mora-Gonzalez
1
, Marie Löf
4,5
, Idoia Labayen
6
, Jonatan R. Ruiz
1,4
, and
Francisco B. Ortega
1,4
1
PROFITH “PROmoting FITness and Health through physical activity” Research Group,
Department of Physical Education and Sports, Faculty of Sport Sciences, University of Granada,
Ctra. Alfacar s/n, 18011 Granada, Spain
2
Department of Sport Medicine, Norwegian School of Sport Sciences, Oslo, Norway
3
MRC Epidemiology Unit, Institute of Metabolic Science, University of Cambridge, Addenbrooke’s
Hospital Hills Road, Cambridge, UK
4
Department of Biosciences and Nutrition, Karolinska Institutet, Huddinge, Sweden
5
Department of Clinical and Experimental Medicine, Faculty of the Health Sciences, Linköping
University, Linköping, Sweden
6
Department of Nutrition and Food Science, University of the Basque Country, UPV-EHU, Vitoria-
Gasteiz, Spain
Abstract
Accelerometers are widely used to measure sedentary time, physical activity, physical activity
energy expenditure (PAEE), and sleep-related behaviors, with the ActiGraph being the most
frequently used brand by researchers. However, data collection and processing criteria have
evolved in a myriad of ways out of the need to answer unique research questions; as a result there
is no consensus.
Objectives—The purpose of this review was to: (1) compile and classify existing studies
assessing sedentary time, physical activity, energy expenditure, or sleep using the ActiGraph
GT3X/+ through data collection and processing criteria to improve data comparability and (2)
review data collection and processing criteria when using GT3X/+ and provide age-specific
practical considerations based on the validation/calibration studies identified.
Methods—Two independent researchers conducted the search in PubMed and Web of Science.
We included all original studies in which the GT3X/+ was used in laboratory, controlled, or free-
living conditions published from 1 January 2010 to the 31 December 2015.
*
Phone +34 958 24 43 53, jairohm@ugr.es.
Conflict of interest
Jairo H. Migueles, Cristina Cadenas-Sanchez, Ulf Ekelund, Christine Delisle Nyström, Jose Mora-Gonzalez, Marie
Löf, Idoia Labayen, Jonatan R. Ruiz, and Francisco B. Ortega declare that they have no conflicts of interest relevant to the content of
this review.
Europe PMC Funders Group
Author Manuscript
Sports Med. Author manuscript; available in PMC 2018 November 12.
Published in final edited form as:
Sports Med
. 2017 September ; 47(9): 1821–1845. doi:10.1007/s40279-017-0716-0.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
Results—The present systematic review provides key information about the following data
collection and processing criteria: placement, sampling frequency, filter, epoch length, non-wear-
time, what constitutes a valid day and a valid week, cut-points for sedentary time and physical
activity intensity classification, and algorithms to estimate PAEE and sleep-related behaviors. The
information is organized by age group, since criteria are usually age-specific.
Conclusion—This review will help researchers and practitioners to make better decisions before
(i.e., device placement and sampling frequency) and after (i.e., data processing criteria) data
collection using the GT3X/+ accelerometer, in order to obtain more valid and comparable data.
PROSPERO registration number—CRD42016039991.
1 Introduction
Health benefits of physical activity (PA) across a person’s lifespan have been widely
reported [1, 2, 3]. The use of accelerometers to assess sedentary time (SED) and PA [4, 5, 6,
7] has become an objective and feasible alternative to self-report methods such as
questionnaires, which are characterized by their poor reliability and validity, especially in
younger populations [8, 9, 10]. Accelerometers are wearable devices that measure
accelerations of the body segment to which the monitor is attached. The signal is usually
filtered and pre-processed by the monitor to obtain activity counts, i.e., accelerations due to
body movement. The amount and intensity of daily SED and PA may be obtained by
classifying activity counts accumulated in a specific time interval (epoch length) with a set
of cut-points, i.e., intensity thresholds for PA intensity classification [11, 12, 13, 14, 15].
Physical activity energy expenditure (PAEE) or sleep-related behaviors may also be
estimated by applying algorithms to objectively-determined activity counts [16, 17, 18, 19,
20, 21]. New methods to estimate these variables from raw acceleration signals (gravity
units) instead of activity counts have been developed recently [22, 23, 24].
Among the commercially available brands, the ActiGraph (Pensacola, FL, USA)
accelerometers are the most frequently used by researchers, accounting for >50% of
published studies [25]. This review only considered the latest generation of ActiGraph
devices, i.e., GT3X, GT3X+, and wGT3X-BT (hereinafter referred to as GT3X/+). The
continuous change in the features of these devices makes it difficult to compare data
between studies.
The first ActiGraph accelerometers available were uniaxial (i.e., they could only detect
vertical axis accelerations) and consequently cut-points and algorithms were developed to
assess SED, PA intensity, PAEE, and sleep-related behaviors from vertical axis accelerations
[11, 17, 21]. In mid-2009, ActiGraph released the triaxial GT3X, which detected
accelerations in three axes (i.e., vertical, medio-lateral and antero-posterior axes). The
transition from uniaxial to triaxial devices implied new calibration processes, and the
algorithms developed for the vertical axis were not applicable to vector magnitude (i.e., the
square root of the sum of squared activity counts from the three axes) [7, 13, 18, 20, 26, 27,
28].
Migueles et al.
Page 2
Sports Med
. Author manuscript; available in PMC 2018 November 12.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
Due to the extremely rapid development in this field, there is an overwhelming amount of
data collection and processing criteria decisions, and there is no consensus about which
approaches to use. Consequently, it is difficult for researchers and practitioners to make the
right decisions about which criteria should be used in a given situation. This is important as
the chosen criteria have a huge impact on the outcome. In order to address this problem,
some studies have compared certain GT3X/+ outcomes estimated by different cut-points and
algorithms [4, 29, 30, 31] in an attempt to recommend which decisions are the most
accurate; however, this information is still scarce.
It is important to note that algorithms validated in a specific age group might not be valid for
other age groups due to different PA patterns, so whenever possible, data collection and
processing criteria should be age-specific. Accelerometer methods can be grouped into two
categories: (1) data collection protocols, which are decisions that need to be made a priori
such as device placement or sampling frequency; and (2) data processing criteria, which
involve decisions that can be made a posteriori such as filters, epoch length, non-wear-time
definition, cut-points, and algorithms. The present review will address all of these criteria
separately and specifically by age group. In this review we aimed to: (1) compile and
classify existing studies assessing sedentary time, physical activity, energy expenditure, or
sleep using the ActiGraph GT3X/+ by data collection and processing criteria to improve
data comparability, and (2) review data collection and processing criteria when using
GT3X/+ and provide age-specific practical considerations based on the validation/calibration
studies identified. Both objectives were approached separately for the following age groups:
preschoolers, children/adolescents, adults, and older adults. Although there is a large amount
of information included in this review, we believe that it is useful for readers to have a single
article that summarizes the most important accelerometer methods for each age group
separately. This will allow readers to go directly to a specific criteria for the age group they
are interested in (e.g., PAEE in preschoolers). In this review, we provide a section with
examples of how the information presented can be used in practical terms, as well as a table
with practical considerations.
2 Methods
2.1 Study Design
The present review focuses on 11 key methodological issues related to GT3X/+ data
collection and processing criteria: (1) device placement, (2) sampling frequency, (3) filter,
(4) epoch length, (5) non-wear-time definition, (6) what constitutes a valid day and a valid
week, (7) registration period protocol, (8) SED and PA intensity classification, (9) PAEE
algorithms, (10) sleep algorithms, and (11) step counting. Available information was
classified into two different types of studies: (1) any cross-sectional, longitudinal, or
intervention study which used the GT3X/+ device and met the inclusion criteria indicated in
Sect. 2.3 (objective 1); and (2) studies focused on validation, calibration, or comparison of
functions related to data collection or processing criteria (objective 2). Therefore, the
practical considerations provided for each age group are based on the results from the
validation/calibration studies (see Table 1). Furthermore, we provide a summary of all data
extracted from the validation/calibration papers included in this review by age group in the
Migueles et al.
Page 3
Sports Med
. Author manuscript; available in PMC 2018 November 12.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
Electronic Supplementary Material Appendix S1. Inclusion/exclusion criteria and analytical
methods were specified in advance and registered in the PROSPERO (http://
www.crd.york.ac.uk/PROSPERO/) international database of systematic reviews
(CRD42016039991) [32]. The study is conducted according to the Preferred Reporting
Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [33].
2.2 Search Strategy
We searched PubMed and Web of Science for studies using the ActiGraph GT3X/+ model
and classified the studies into the following age groups: preschoolers (2–5 years), children
(6–11 years), adolescents (12–18 years), adults (19–59 years), and older adults (≥60 years).
We combined (using the Boolean operator “OR”) the following search terms: GT3X, GT3X
+, and ActiGraph. Although we wanted to limit the search to GT3X/+, the word ActiGraph
was entered in the search because we found that some studies specified the brand (i.e.,
ActiGraph) instead of the model (i.e., GT3X/+) in the title/abstract/keywords. Since the
GT3X/+ models were launched in mid-2009, we limited the dates of the search to 1 January
2010 to the 31 December 2015 and conducted the final search on 3 January 2016. We
contacted authors of those studies where the data processing and collection information was
unavailable in the published article. In a final step, we extended the search to the IEEE
(Institute of Electrical and Electronics Engineers) Xplore database, in case we had missed
any relevant studies.
2.3 Inclusion/Exclusion Criteria
We included all original studies (cross-sectional, longitudinal, or intervention studies) in
which the GT3X/+ was used in a laboratory, or under controlled or free-living conditions.
Protocol studies, reviews, editorials, and abstract or congress communications were
excluded, as well as studies conducted in people with mobility problems or in periods of life
in which their mobility could have been markedly altered (e.g., pregnancy).
Two authors working independently (JHM and CCS) read the articles and checked whether
they met the inclusion/exclusion criteria. They obtained 76% agreement on the papers
selected for the review before consensus and 100% agreement after discrepancies were
resolved in a consensus meeting. Risk of bias assessment was also conducted independently
by JHM and CCS in order to assess the quality of studies (see Electronic Supplementary
Material Appendix S2).
3 Results
A total of 940 articles were identified (Fig. 1), of which 444 were excluded after reading the
title and abstract and 261 articles were additionally excluded after reading the full text and
did not meet the inclusion/exclusion criteria stated above. Finally, a total of 235 studies were
considered eligible for the current systematic review. Of them, 78 were validation/calibration
studies. Methods and results of these validation/calibration studies are summarized in the
Electronic Supplementary Material Appendix S1. Detailed information about the methods
and results for the rest of studies (i.e., those using GT3X/+ that were not validation/
calibration studies) included in this review is available upon request.
Migueles et al.
Page 4
Sports Med
. Author manuscript; available in PMC 2018 November 12.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts
Forty-four percent (
N
= 103) of the included studies were conducted in adults (46%
validation/calibration studies), 34% (
N
= 81) in youth (30% validation/calibration studies),
22% (
N
= 51) in older adults (11% validation/calibration studies), and 10% (
N
= 24) in
preschoolers (13% validation/calibration studies).
Studies including two or more age groups are summarized in both age group sections in this
review. Table 2 presents the criteria used for data collection and processing by age group. A
list of references for each of the criteria is found in Electronic Supplementary Material
Appendix S3. The information provided in Table 2 and Electronic Supplementary Material
Appendix S3 allows researchers to make comparisons between studies that have used the
same data collection and processing criteria.
Figure 2 shows the percentage of studies that did not report key methodological issues by
age group. Of the studies reviewed, 15–20% did not report criteria such as sampling
frequency, epoch length, and a non-wear-time definition, and 60–80% of studies did not
report information on the filter used.
Table 3 presents the studies that compared the differences in several outcomes when the
GT3X/+ device was simultaneously worn on the wrist and hip. The optimal place to attach
the GT3X/+ should be chosen based on reliability, validity, and compliance. Table 4 shows
the references for the studies sorted by age group and placement site that have developed
SED and PA cut-points, PAEE prediction equations, and sleep algorithms. Table 5 shows the
intensity cut-points used in the included studies together with the pre-processing criteria
used in the study which developed each set of cut-points. Therefore, the practical
considerations provided for each age group are based on the results from the validation/
calibration studies (see Table 1).
In the following sub-sections, we will focus only on information from validation/calibration
studies presented in Electronic Supplementary Material Appendix S1. Sections 3.1, 3.2, and
3.3 correspond to data collection protocols (i.e., pre-processing stage) and Sects. 3.4 – 3.10
correspond to processing criteria (i.e., processing stage).
3.1 Device Placement
3.1.1 Preschoolers—In young preschoolers Johansson et al. [34] reported receiver
operating characteristic area under the curve (ROC-AUC) data for intensity thresholds
between 0.88 to 0.98 using a left wrist-mounted GT3X+. Similarly, a ROC-AUC of 0.90–
0.94 was reported by Costa et al. [35] using a hip placement, suggesting high potential for
both placements to correctly classify PA intensity in preschoolers.
3.1.2 Children and Adolescents—A higher compliance for wrist-worn versus hip-
worn devices has been reported in children/adolescents [23]. However, similar wear-time
was achieved in protocols using 24-h waist-worn compared to 24-h wrist-worn
accelerometers [36].
With regard to cut-points to classify SED and PA intensity, non-dominant wrist placement
achieved a lower ROC-AUC (0.64–0.89) [15] compared to the dominant wrist (0.83–0.94)
Migueles et al.
Page 5
Sports Med
. Author manuscript; available in PMC 2018 November 12.
Europe PMC Funders Author Manuscripts Europe PMC Funders Author Manuscripts