Published online: 18 Oct 2012.

Transcription

1 This article was downloaded by: [b-on: Biblioteca do conhecimento online UTL] On: 14 July 2015, At: 07:50 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: Registered office: 5 Howick Place, London, SW1P 1WG Click for updates European Journal of Sport Science Publication details, including instructions for authors and subscription information: Calibration of ActiGraph GT3X, Actical and RT3 accelerometers in adolescents Marcelo Romanzini a, Edio Luiz Petroski b, David Ohara a, Antonio Carlos Dourado c & Felipe Fossat Reichert d a Department of Physical Education, State University of Londrina, Londrina, Brazil b Department of Physical Education, Federal University of Santa Catarina, Florianópolis, Brazil c Department of Sport of Science, State University of Londrina, Londrina, Brazil d School of Physical Education, Federal University of Pelotas, Pelotas, Brazil Published online: 18 Oct To cite this article: Marcelo Romanzini, Edio Luiz Petroski, David Ohara, Antonio Carlos Dourado & Felipe Fossat Reichert (2014) Calibration of ActiGraph GT3X, Actical and RT3 accelerometers in adolescents, European Journal of Sport Science, 14:1, 91-99, DOI: / To link to this article: PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the Content ) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at

2 European Journal of Sport Science, 2014 Vol. 14, No. 1, 9199, ORIGINAL ARTICLE Calibration of ActiGraph GT3X, Actical and RT3 accelerometers in adolescents MARCELO ROMANZINI 1, EDIO LUIZ PETROSKI 2, DAVID OHARA 1, ANTONIO CARLOS DOURADO 3, & FELIPE FOSSAT REICHERT 4 Downloaded by [b-on: Biblioteca do conhecimento online UTL] at 07:50 14 July Department of Physical Education, State University of Londrina, Londrina, Brazil, 2 Department of Physical Education, Federal University of Santa Catarina, Florianópolis, Brazil, 3 Department of Sport of Science, State University of Londrina, Londrina, Brazil, and 4 School of Physical Education, Federal University of Pelotas, Pelotas, Brazil Abstract The objective of this study was to develop count cut-points for three different accelerometer models: ActiGraph GT3X, RT3 and Actical to accurately classify physical activity intensity levels in adolescents. Seventy-nine adolescents (1015 years) participated in this study. Accelerometers and oxygen consumption ( _V O 2 ) data were collected at rest and during 11 physical activities of different intensities. Accelerometers were worn on the waist and _V O 2 was measured by a portable metabolic system: Cosmed K4b2. Receiver operating characteristic (ROC) curves were used to determine cut-points. Cutpoints for sedentary (SED), moderate-to-vigorous (MVPA) and vigorous-intensity physical activity (VPA) were 46, 607 and 818 counts 15s 1 to the vertical axis of ActiGraph; 180, 757 and 1112 counts 15s 1 to the vector magnitude of ActiGraph; 17, 441 and 873 counts 15s 1 for Actical; and 5.6, 20.4 and 32.2 counts s 1 for RT3, respectively. For all three accelerometer models, there was an almost perfect discrimination of SED and MVPA (ROC 0.97) and an excellent discrimination of VPA (ROC 0.90) observed. Areas under the ROC curves indicated better discrimination of MVPA by ActiGraph (AUC0.994) and Actical (AUC0.993) when compared to RT3 (AUC0.983). The cut-points developed in this study for the ActiGraph (vector magnitude), RT3 and Actical accelerometer models can be used to monitor physical activity level of adolescents. Keywords: Motion sensors, cut-points, physical activity, sedentary behaviour, youth Introduction Valid measurements of physical activity from moderate-to-vigorous intensity (MVPA) and sedentary behaviour in young people are of fundamental importance, particularly because of the associations between physical activity behaviours and health (Janssen & Leblanc, 2010; Strong et al., 2005). Since accelerometers allow for the identification of associations that might not be noticeable with subjective measures, minimise information bias and may improve knowledge related to the relationship between physical activity, sedentary behaviour and health (Reilly et al., 2008), the use of accelerometers in studies involving children and adolescents has increased (Rowlands, 2007). Various accelerometer models are used to detect the acceleration of body movements and to transform them into a measure called counts. Counts vary according to the accelerometer characteristics (Chen & Bassett, 2005) and do not have a behavioural or biological significance (Freedson, Pober, & Janz, 2005), but may be influenced by the subject s physical characteristics (Freedson et al., 2005). Thus, the interpretation of counts of accelerometers can be accomplished through the use of cut-points derived from calibration studies specific to the population (children, adolescents, adults) and manufacturer. ActiGraph is the most widely used accelerometer in physical activity studies (De Vries et al., 2009) and Correspondence: Marcelo Romanzini, Department of Physical Education, State University of Londrina, Rod. Celso Garcia Cid, km 380, Campus Universitário, Londrina, Brazil. mromanzini@uel.br # 2012 European College of Sport Science

3 Downloaded by [b-on: Biblioteca do conhecimento online UTL] at 07:50 14 July M. Romanzini et al. calibration studies with children and adolescents have already been performed (Evenson, Catellier, Gill, Ondrak, & McMurray, 2008; Mattocks et al., 2007; Puyau, Adolph, Vohra, & Butte, 2002; Treuth et al., 2004; Vanhelst, Beghin, Turck, & Gottrand, 2011). Unlike previous versions, the newer model of ActiGraph GT3X accelerometer is a solid-state triaxial accelerometer that can measure motion data on three axes. Since the collection and storage process of uniaxial and triaxial versions of ActiGraph accelerometer are different, cut-points developed for models 7164 and GT1M might not be applicable to GT3X. In addition, to date, only one study calibrated the vector magnitude of the GT3X for children and adolescents (Hänggi, Phillips, & Rowlands, in press). Therefore, the development of cut-points for the new version of ActiGraph is crucial. RT3 and Actical are two other accelerometer models commonly used in physical activity measurement (Rowlands, 2007). Both the RT3 and Actical are triaxial accelerometers suggesting high validity, with Actical being particularly attractive in studies with children and adolescents due to its small size. Some studies have calibrated these accelerometers in children and adolescents (Chu, McManus, & Yu, 2007; Colley & Tremblay, 2011; Evenson et al., 2008; Kavouras, Sarras, Tsekouras, & Sidossis, 2008; Puyau, Adolph, Vohra, Zakeri, & Butte, 2004; Rowlands, Thomas, Eston, & Topping, 2004; Vanhelst et al., 2010), but cut-points have been divergent. Virtually all calibration studies so far have been carried out in high-income settings. Whether the thresholds developed in such studies are valid for other settings remains to be determined. In addition to verifying the consistency of the thresholds, the current study will calibrate the ActiGraph GT3X. In contrast with earlier versions of the ActiGraph, which were uniaxial, this later model is triaxial and thus new thresholds are necessary to be developed. Furthermore, the current study adds to the literature by calibrating the RT3, Actical and ActiGraph GT3X using a single sample of participants who are engaging in the same activities; thus, appropriate cut-points can be determined for the different devices. Further, this allows for comparisons among studies that used different devices to measure physical activity. Therefore, the objective of this study was to establish cut-points for ActiGraph (GT3X), RT3 and Actical accelerometers to identify sedentary behaviour and light- (LPA), MVPA and vigorousintensity physical activity (VPA) intensity in adolescents from 10 to 15 years of age. Methods Sample From September 2010 to March 2011, adolescents aged 1015 years enrolled from 5th to 8th grade in a public school in the city of Londrina, Southern Brazil, were invited to participate in this study. The sample size calculation took into account the area under the receiver operating characteristic (ROC) curve of 0.79, which is the lowest area observed in accelerometer calibration studies in adolescents (De Bock et al., 2010), a type I error of 0.05, and power of 90%, resulting in a minimum of 78 subjects needed. Among those who agreed to participate in the study, a total of 79 adolescents were selected, with approximately half male. The selection process involved stratifying adolescents by sex (males and females) and age group ( years and years). The adolescents were then randomly selected into each stratum in order to ensure the proportion of 50% for each sex and age group. The study protocol was approved by the Ethics Committee and Human Research of the Federal University of Santa Catarina (protocol 445/2010); parental consent was obtained for all child participants in the study. Anthropometry Anthropometric measures of body mass and height were collected using electronic scales (Urano, model PS 180) with 0.1 kg resolution and stadiometers with 0.5 cm scales. During measurements, the subjects were barefoot and wore light clothing. Nutritional status was determined according to body mass index cut-points values suggested by Cole, Bellizzi, Flegal, and Dietz (2000). Accelerometry The accelerometers studied were ActiGraph GT3X, RT3 and Actical. ActiGraph (ActiGraph LLC, Pensacola, FL, USA), model GT3X, weighs 27 grams and has small dimensions ( cm). It has a triaxial accelerometer that collects information in three axes (vertical, mediolateral and anteroposterior) and can combine this information into a vector magnitude. ActiGraph GT3X records accelerations in a magnitude range from 0.05 to 2.5 G s. The signal is digitised and passes through a filter that limits the frequency to a variation from 0.25 to 2.5 Hz. The GT3X epoch can vary from 1 to 360 seconds. The RT3 accelerometer (StayHealthy, Inc., Monrovia, CA) has dimensions of cm and weighs 65.2 grams. Similar to ActiGraph GT3X, RT3 detects acceleration in three axes to generate a triaxial measure (vector magnitude) (Chu et al., 2007; Kavouras et al., 2008). RT3 records

4 Downloaded by [b-on: Biblioteca do conhecimento online UTL] at 07:50 14 July 2015 information only in epochs of 1 or 60 seconds. Actical (Mini Mitter Co., Inc., Bend, OR) is the smallest ( cm) and lightest (17 grams) of the accelerometers studied. It is classified as omnidirectional, i.e., captures the movement in all directions, although it is oriented to primarily detect vertical acceleration (Heil, 2006). When positioned in the hip, the device becomes more sensitive to vertical movements of the trunk. Actical is sensitive to movements in a frequency range from 0.5 to 3 Hz. This accelerometer is capable of storing data in epochs of 15, 30 or 60 seconds. Two Actical accelerometers and five ActiGraph GT3X and RT3 accelerometers were used in the study. Each adolescent was simultaneously monitored by an accelerometer from each manufacturer. The accelerometers were fixed at the waist and positioned on the anterior axillary line at the iliac crest level of the right or left hip. This procedure was to ensure that accelerometers were positioned in the same location on all participants. Although Treuth et al. (2004) observed similar results for accelerometers data on either side of the hips, we chose to systematically change the side on which they were worn for every individual. Counts were recorded in epochs of 15 seconds for ActiGraph GT3X and Actical and in epochs of one second for RT3. Indirect calorimetry Oxygen uptake ( _V O 2 ) and heart rate (HR) measurements were obtained by portable metabolic system: Cosmed (Model K4b2, Rome, Italy). Cosmed K4b2 is a lightweight device (925 grams) consisting of a mask connected to a miniaturised analysis system, a battery and a POLAR Pacer HR transmitter. About an hour before each measurement session, the unit was calibrated with standard gases according to manufacturer s instructions. This equipment has been validated for adolescents (Harrell et al., 2005) and has been used in calibration protocols for accelerometers involving this population group. Procedures All data were collected at the Center for Physical Education and Sports of the State University of Londrina (Brazil) on a single occasion. Adolescents were asked to fast for at least two hours prior to the beginning of the tests to minimise the impact of the thermic effect of food on the _V O 2 measures. This procedure is similar to those adopted in other studies (Pate, Almeida, McIver, Pfeiffer, & Dowda, 2006; Pfeiffer, McIver, Dowda, Almeida, & Pate, 2006). Initially, anthropometric measures of body mass and height were obtained. Then, Cosmed K4b2 and accelerometers were fixed in individuals to collect Calibration of motion sensors 93 _V O 2 measures and acceleration at rest and during 11 types of physical activities. The description and order of physical activities are described in Table I. The intensities of activities ranged from sedentary to vigorous intensity. Activities were selected to represent typical daily physical activities for adolescents. Each activity was performed for 5 minutes with the exception of the rest period, which lasted 20 minutes. The duration of each activity is similar to other studies (Heil, 2006; Mattocks et al., 2007). Between each activity, a fiveminute recovery period was given. Accelerometers and the Cosmed K4b2 system were synchronised and initialised in the first minute of the rest period. Each activity was started in a new minute to facilitate the processing of information. The rest period and sedentary activities were conducted in an air-conditioned laboratory (average temperature and humidity of 22.48C and 58.5%, respectively), while the other physical activities were held in a gymnasium (average temperature and humidity values of 26.58C and 52.7%, respectively). For the _V O 2 measurement at rest, the adolescents remained lying on a stretcher in a quiet and partially illuminated environment. They were instructed to remain lying in supine position without sleeping or performing sudden body movements. During the walking and jogging activities, an appraiser set the Table I. Description of activities to be performed in the calibration protocol. Activity Description Intensity Rest Lying in supine position, arms Sedentary at the body s sides, with instructions to minimise body movements DVD Sitting on a chair and watching Sedentary children s films Writing Sitting on a chair and writing Sedentary a standard text with arms on a table Videogame Sitting on a chair and using Sedentary hand controls to play video games (Playstation 3) Standing Standing, with the possibility Sedentary of light movement of limbs Walking 2 km h -1 Walking at a speed of 2 km h -1 Light Walking 4 km h -1 Walking at a speed of 4 km h -1 Moderate Volleyball Double playing volleyball on Moderate a court with reduced size Running 7.2 km h -1 Running at a speed of Vigorous 7.2 km h -1 Soccer Making dribbles in movement Vigorous with a soccer ball and then kicking it to the goal Basketball Performing dribbling between Vigorous obstacles and throwing the ball to the basket Jumping rope Jumping rope rhythmically Vigorous

5 94 M. Romanzini et al. Downloaded by [b-on: Biblioteca do conhecimento online UTL] at 07:50 14 July 2015 pace of adolescents using a speed sensor (Polar S1 Foot Pod, Polar RS 300X TM ). Data reduction Information recorded by indirect calorimetry and accelerometers were transferred to a Microsoft Office Excel 2007 spreadsheet. For purposes of analysis, minutes of the rest period and 45 minutes of each physical activity were considered. This approach is consistent with previous studies. _V O 2 measures recorded by calorimetry and counts recorded by accelerometers were computed as mean values for each activity. The steady state _V O 2 in 1620 minutes of the rest period and 45 minutes of each activity were confirmed by visual inspection. Metabolic equivalent (MET) scores were individually computed by dividing _V O 2 values (ml kg min 1 ) recorded during each activity by the V _ O 2 values (ml kg min 1 )recorded at rest. Then, MET scores were categorised as sedentary (SED) (B1.5 METs), LPA (]1.5 and B3 METs), MVPA (]3 METs), or VPA (]6 METs). Despite some controversy over the use of three or four METs to define moderate-intensity activities in children and adolescents, this classification is consistent with previous studies (Chu et al., 2007; Puyau et al., 2004; Rowlands et al., 2004). Statistical analysis Mean values and standard deviations were determined for all variables. Cut-points for SED, LPA, MVPA and VPA were determined by ROC (Jago, Zakeri, Baranowski, & Watson, 2007). The criterion for the determination of each cut-point was the point at which sensitivity and specificity were maximised. Sensitivity is maximised by correctly identifying activities at or above the cut-point for intensity. Specificity is maximised by correctly excluding activities below the cut-point for intensity. Independent variables were the mean counts recorded for each type of accelerometer in each activity. Dependent variables were the MET values recorded in each activity, which were transformed into dummy variables (0 and 1). Thus, to determine the cut-point for SED, activities were categorised as sedentary 1 (B1.5 METs) or not sedentary 0 (]1.5 METs). For the MVPA cut-point, activities were classified as MVPA 1 (]3 METs) or lower than moderate intensities 0 (B3 METs). Finally, to determine the cut-point for VPA, the activities were classified as vigorous 1 (]6 METs) or lower than vigorous intensity 0 (B6 METs). The cut-points corresponding to sedentary and moderate intensities were adopted as the lower and upper cut-points for LPA. The Hanley and McNeil (1983) test was used to compare eventual differences in the area under the ROC curve among the accelerometer models. The leave-one-out cross-validation analysis was carried out to test the validity of the thresholds derived to each accelerometer model. Analyses were conducted using statistical software SPSS version 20.0 (descriptive analyses, drawing graphs of ROC curves and leave-one-out cross-validation) and MedCalc version (identification of thresholds, areas under the ROC curve and sensitivity and specificity values). Results The study sample was balanced in sex (40 boys and 39 girls) and in age (from 10 to 12.5 years48.1%; 12.5 to 15 years51.9%). Body mass and height measures ranged from 26.4 to 73.3 kg ( kg) and from to cm ( cm), respectively. Approximately 23% of the adolescents studied were overweight. Mean oxygen consumption, METs, HR and accelerometer counts for each task are presented in Table II. Of a total of 948 Table II. Descriptive statistics (mean and standard deviation) for _V O 2, MET, heart rate and counts of accelerometers for each activity. _V O 2 (ml kg min 1 ) MET ActiGraph (VT) (counts 15s 1 ) ActiGraph (VM) (counts 15s 1 ) RT3 (counts s 1 ) Actical (counts 15s 1 ) Rest DVD Writing Videogame Standing Walking 2 km h Walking 4 km h Volleyball Running 7.2 km h Soccer Basketball Jumping rope Note: VT, vertical axis; VM, vector magnitude.

6 Calibration of motion sensors 95 Downloaded by [b-on: Biblioteca do conhecimento online UTL] at 07:50 14 July 2015 possible cases for each variable (79 subjects 12 activities), complete information on counts and _V O 2 from ActiGraph GT3X, RT3 and Actical accelerometers were obtained in 896 (94.5%), 740 (78.0%) and 918 (96.8%) cases, respectively. Exclusion criteria included: (1) voluntary fatigue (2.5%), (2) Cosmed failure (0.5%), (3) initialisation or download of accelerometer data failure (2.5, 21.9 and 0.0% for ActiGraph GT3X, RT3 and Actical, respectively). The intensity of activities ranged from (watching a DVD) to (playing soccer) METs scores. For the three accelerometers investigated, despite a 40% increase in V _ O 2, counts for volleyball were lower than those observed during walking at 4 km h -1. In this activity, there were several moments of non-involvement assessed in some individuals, resulting in scattered counts at each epoch recorded. Thus, volleyball data were excluded of subsequent analyses. Counts cut-points for SED, LPA, MVPA and VPA are shown in Table III. For the three accelerometers analysed, areas under the ROC curve (ROC-AUC) showed almost perfect discrimination of SED and MVPA (ROC ]0.97) and excellent discrimination of VPA (ROC-AUC ]0.91). In general, cut-points were more sensitive than specific. By examining the ROC-AUC by accelerometer, better discrimination of MVPA by ActiGraph GT3X and Actical accelerometers when compared to RT3 was observed (P B0.05) (Figure 1). The identification of SED and VPA was similar among the three accelerometers. The prediction of SED, MVPA and VPA were similar for the vertical axis and vector magnitude of the GT3X. The leave-one-out cross-validation analysis for the magnitude vector of the ActiGraph showed that 88.5% of the MET SED, 90.2% of the MET MVPA and 88.4% of the MET VPA were accurately classified. To the Actical accelerometers, these values were 82.5, 86.9 and 87.0%, respectively, while to the magnitude vector of the RT3 the values were 85.5, 85.2 and 87.4%, respectively. Discussion This study established cut-points for counts for the new version of the ActiGraph (GT3X) accelerometer and the RT3 and Actical accelerometers for the identification of different intensities of physical activity in adolescents between 10 and 15 years of age. According to the cut-points developed for the three accelerometers, the discrimination of SED, MVPA and VPA was excellent (ROC-AUC 0.90). In addition, the ActiGraph GT3X and Actical accelerometers showed significantly better ability to identify MVPA compared to the RT3 accelerometer. This is one of the first calibration studies of the triaxial version of ActiGraph (GT3X) in adolescents. In a protocol with 11 physical activities, the cutpoints derived from the vertical axis and the vector magnitude of ActiGraph GT3X were able to discriminate physical activity intensities similarly. Several of the physical activities included in the current study have a prominent component of vertical acceleration, which might have contributed to the Table III. Sensitivity, specificity, area under the ROC curve and cut-points for counts of ActiGraph (GT3X), Actical and RT3 accelerometers for adolescents. Sensitivity (%) Specificity (%) Area under the ROC curve (95% CI) Cut-points ActiGraph (VT) Sedentary ( ) 046 counts 15s 1 Light counts 15s 1 Moderate ( ) counts 15s 1 Vigorous ( ) ]818 counts 15s 1 ActiGraph (VM) Sedentary ( ) 0180 counts 15s 1 Light counts 15s 1 Moderate ( ) counts 15s 1 Vigorous ( ) ]1112 counts 15s 1 RT3 Sedentary ( ) 05.6 counts s 1 Light counts s 1 Moderate ( ) counts s 1 Vigorous ( ) ]32.2 counts s 1 Actical Sedentary ( ) 017 counts 15s 1 Light counts 15s 1 Moderate ( ) counts 15s 1 Vigorous ( ) ]873 counts 15s 1 Note: VT, vertical axis; VM, vector magnitude.

7 Downloaded by [b-on: Biblioteca do conhecimento online UTL] at 07:50 14 July M. Romanzini et al. Figure 1. ROC curves for (a) sedentary, (b) moderate- and (c) vigorous-intensity physical activities. similar performance between the vertical axis and the vector magnitude of the GT3X. However, the literature is not consistent in showing that triaxial accelerometers perform better than uniaxial ones (Loprinzi & Cardinal, 2011). The cut-points derived for SED and MVPA from the vertical axis of ActiGraph GT3X were quite similar to those validated by Evenson et al. (2008) for children and adolescents (100 and 2300 counts min 1 to SED and MVPA, respectively). This supports the recently recommended cut-points for SED and MVPA by Trost, Loprinzi, Moore, and Pfeiffer (2011), which provide good confidence for the use of the thresholds derived in this study for SED and MVPA using data from the magnitude vector of GT3X and the counts from RT3 and Actical. Regarding the VPA cut-point, a smaller threshold was found for the vertical axis of GT3X (3300 counts min 1 ) when compared to those shown in literature (Table IV). However, Trost et al. (2011) reported that high thresholds ( counts min 1 ) related to VPA tend to have high specificity, low sensitivity and significantly lower capacity to predict VPA when compared to a lower threshold of 4012 counts min 1. Therefore, it is likely that the cut-point for the vector magnitude of ActiGraph GT3X related to VPA is slightly higher than that found in this study. Another triaxial accelerometer calibrated in this study was RT3. Currently, the cut-points available for this accelerometer are inconsistent, especially for SED and VPA (Table IV). In this study, the cutpoints established may provide insight on the determination of different intensity physical activities, particularly of SED and MVPA. In this study, SED and MVPA were measured by both the ActiGraph GT3X and RT3 accelerometers for same sample using the same physical activity protocols. The resulting thresholds of the vertical axis of ActiGraph GT3X resembled those previously validated by Trost et al. (2011) providing insight on the validity of the cut-points for the RT3 accelerometer compared to the ActiGraph GT3X. Additionally, a cut-point of 5.6 counts s 1 for SED was identified in this study and is consistent with the findings in the Chu et al. (2007) study that suggest that cut-points near 6 counts s 1 are appropriate to determine SED by RT3. For MVPA, the cut-point of 20.4 counts s 1 (1224 counts min 1 ) established in this study is similar to those proposed in other studies (Kavouras et al., 2008; Rowlands et al., 2004; Vanhelst et al., 2010). Thus, based on an excellent discrimination of MVPA (ROC-AUC 0.98), the use of the cut-point established in this study can be recommended. With respect to VPA, the cut-point established in this study (1932 counts min 1 ) is lower than other cut-points available in literature (Table IV).

8 Calibration of motion sensors 97 Table IV. Cut-points developed in the current study and previously published cut-points for ActiGraph, RT3 and Actical accelerometers in studies with children and adolescents. Cut-points (counts min 1 ) Study Age (n) Sedentary Light Moderate Vigorous Downloaded by [b-on: Biblioteca do conhecimento online UTL] at 07:50 14 July 2015 ActiGraph (vertical axis) Evenson et al. (2008) 58 yr (33) ]4012 Mattocks et al. (2007) 12 yr (163) ]6130 Puyau et al. (2002) 616 yr (26) B ]8200 Treuth et al. (2004) 1314 yr (74) B Vanhelst et al. (2011) 1016 yr (40) Current study 1015 yr (79) ]3272 ActiGraph (vector magnitude) Hänggi et al. (in press) 1015 yr (32) ]3366 Current study 1015 yr (79) ]4448 RT3 Chu et al. (2007) 812 yr (35) B ]4110 Kavouras et al. (2008) 1014 yr (42) ]2610 Rowlands et al. (2004) 991 yr (19) ]2333 Vanhelst et al. (2010) 1016 yr (40) Current study 1015 yr (79) ]1932 Actical Colley and Tremblay (2011) 915 yr (12) ]4760 Puyau et al. (2004) 718 yr (32) B ]6500 Current study 1015 yr (79) ]3492 Note: Cut-points reported for comparison as counts per minute and rounded where appropriate. Higher cut-points have been observed in studies using only ambulatory activities (running) in the calibration protocol (Chu et al., 2007; Kavouras et al., 2008; Vanhelst et al., 2010). Rowlands et al. (2004) observed a cut-point of 3022 counts min 1 for RT3 in activities performed by boys on treadmill. When other activities were considered, the cut-point was considerably lower (2333 counts min 1 ). Thus, it is plausible that the lower cut-point observed in this study results from the use of varied vigorousintensity activities. Since vigorous-intensity activities performed by adolescents are not typically restricted to running, the use of high cut-points to identify VPA through RT3 seems inappropriate and may significantly underestimate participation in VPA. Actical accelerometer has been calibrated in samples of adolescents. In this study, cut-points for SED (17 counts 15s 1 or 68 counts min 1 ) and MVPA (441 counts 15s 1 or 1764 counts min 1 ) were similar to those observed in other calibration studies for Actical (Colley & Tremblay, 2011; Puyau et al., 2004), while cut-point for VPA was lower (873 counts 15s 1 or 3492 counts min 1 ). Once again, it is important to consider that the VPA of other studies only involved ambulatory activities (Colley & Tremblay, 2011; Puyau et al., 2004). Considering only ambulatory activities, data from this study indicate a cut-point for VPA (1671 counts 15s 1 or 6684 counts min 1 ) very similar to that found by Puyau et al. (2004) (data not shown). Analysing the areas under the ROC curve of the three accelerometers studied decreased the ability to discriminate MVPA activities by RT3 in relation to ActiGraph GT3X and Actical. Nevertheless, discrimination of these activities by RT3 was excellent (ROC-AUC0.983) and there were no differences between accelerometers for the prediction of SED and VPA. It should be considered, however, that the main deficiency of RT3 in physical activity studies is its low data storage capacity. For epochs of onesecond vector magnitude, it is only able to store data for nine hours. Some limitations of this study should be considered. The main limitation is the lack of the crossvalidation of cut-points using an independent sample. Such a validation would be important to increase the confidence in the cut-points developed. Other studies that submitted their cut-points for cross-validation with independent samples did not identify any relevant differences in the results (Chu et al., 2007; Vanhelst et al., 2010Vanhelst et al., 2011). However, it should be highlighted that the counts are more dependent of the activities tested than an individual s characteristics (Corder et al., 2007). In this context, cut-points for the vertical axis of the ActiGraph GT3X in this study were similar to those validated by Trost et al. (2011), particularly for SED and MVPA, even though a different protocol of activities different than those used in the present study were used. Thus, the cut-points developed in this study are valid. Another limitation is regarding the V _ O 2 measurements. In order to obtain steady state _V O 2 measurements, activities were standardised and, therefore, do not necessarily reflect the

9 Downloaded by [b-on: Biblioteca do conhecimento online UTL] at 07:50 14 July M. Romanzini et al. patterns of intermittent activities commonly observed in adolescents. Some strengths points of this study should also be noted. First, the simultaneous derivation of thresholds for three models of accelerometers commonly used in literature can provide comparability between studies that used these accelerometers to determine the different physical activity intensities of adolescents. Second, the thresholds derived for the vertical axis of the ActiGraph GT3X were very similar to those previously validated by Trost et al. (2011) for SED and MVPA. This study suggests that since data were recorded simultaneously by the three models of accelerometers for the same subjects and using the same physical activity protocols, the thresholds derived for the magnitude vector of GT3X, RT3 and Actical counts, especially for SED and MVPA, are reliable. Other strengths of the study include adequate sample size, the use of physical activity protocols representative of the daily life of adolescents and the use of multiple units of each type of accelerometer to consider the variability between accelerometers from the same manufacturer. Conclusions The similarity between cut-points found in this study and those in the Evenson et al. (2008) study for the vertical axis of the ActiGraph provide support for the continued use of Evenson s cut-points. For the vector magnitude of the ActiGraph as well as for the RT3 and the Actical, the cut-points developed in this study should be used in Brazilian children aged 10 to 15 years old. References Chen, K. Y., & Bassett, D. R. Jr. (2005). The technology of accelerometry-based activity monitors: Current and future. Medicine and Science in Sports and Exercise, 37, S490S500. Chu, E. Y., McManus, A. M., & Yu, C. C. (2007). Calibration of the RT3 accelerometer for ambulation and nonambulation in children. Medicine and Science in Sports and Exercise, 39, Cole, T. J., Bellizzi, M. C., Flegal, K. M., & Dietz, W. H. (2000). Establishing a standard definition for child overweight and obesity worldwide: International survey. British Medical Journal, 320, Colley, R. C., & Tremblay, M. S. (2011). Moderate and vigorous physical activity intensity cut-points for the Actical accelerometer. Journal of Sports Sciences, 29, Corder, K., Brage, S., Mattocks, C., Ness, A., Riddoch, C., Wareham, N. J., et al. (2007). Comparison of two methods to assess PAEE during six activities in children. Medicine and Science in Sports and Exercise, 39, De Bock, F., Menze, J., Becker, S., Litaker, D., Fischer, J., & Seidel, I. (2010). Combining accelerometry and HR for assessing preschoolers physical activity. Medicine and Science in Sports and Exercise, 42, De Vries, S. I., Van Hirtum, H. W., Bakker, I., Hopman-Rock, M., Hirasing, R. A., & Van Mechelen, W. (2009). Validity and reproducibility of motion sensors in youth: A systematic update. Medicine and Science in Sports and Exercise, 41, Evenson, K. R., Catellier, D. J., Gill, K., Ondrak, K. S., & McMurray, R. G. (2008). Calibration of two objective measures of physical activity for children. Journal of Sports Sciences, 26, Freedson, P., Pober, D., & Janz, K. F. (2005). Calibration of accelerometer output for children. Medicine and Science in Sports and Exercise, 37, S523S530. Hänggi, J. M., Phillips, L. R., & Rowlands, A. V. (in press). Validation of the GT3X ActiGraph in children and comparison with the GT1M ActiGraph. Journal of Science and Medicine in Sport. Hanley, J. A., & McNeil, B. J. (1983). A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology, 148, Harrell, J. S., McMurray, R. G., Baggett, C. D., Pennell, M. L., Pearce, P. F., & Bangdiwala, S. I. (2005). Energy costs of physical activities in children and adolescents. Medicine and Science in Sports and Exercise, 37, Heil, D. P. (2006). Predicting activity energy expenditure using the Actical activity monitor. Research Quarterly for Exercise and Sport, 77, Jago, R., Zakeri, I., Baranowski, T., & Watson, K. (2007). Decision boundaries and receiver operating characteristic curves: New methods for determining accelerometer cutpoints. Journal of Sports Sciences, 25, Janssen, I., & Leblanc, A. G. (2010). Systematic review of the health benefits of physical activity and fitness in school-aged children and youth. International Journal of Behavioral Nutrition and Physical Activity, 7, 40. Kavouras, S. A., Sarras, S. E., Tsekouras, Y. E., & Sidossis, L. S. (2008). Assessment of energy expenditure in children using the RT3 accelerometer. Journal of Sports Sciences, 26, Loprinzi, P. D., & Cardinal, B. J. (2011). Measuring children s physical activity and sedentary behaviors. Journal of Exercise Science and Fitness, 9, Mattocks, C., Leary, S., Ness, A., Deere, K., Saunders, J., Tilling, K., et al. (2007). Calibration of an accelerometer during freeliving activities in children. International Journal of Pediatric Obesity, 2, Pate, R. R., Almeida, M. J., McIver, K. L., Pfeiffer, K. A., & Dowda, M. (2006). Validation and calibration of an accelerometer in preschool children. Obesity (Silver Spring), 14, Pfeiffer, K. A., McIver, K. L., Dowda, M., Almeida, M. J., & Pate, R. R. (2006). Validation and calibration of the Actical accelerometer in preschool children. Medicine and Science in Sports and Exercise, 38, Puyau, M. R., Adolph, A. L., Vohra, F. A., & Butte, N. F. (2002). Validation and calibration of physical activity monitors in children. Obesity Research, 10, Puyau, M. R., Adolph, A. L., Vohra, F. A., Zakeri, I., & Butte, N. F. (2004). Prediction of activity energy expenditure using accelerometers in children. Medicine and Science in Sports and Exercise, 36, Reilly, J. J., Penpraze, V., Hislop, J., Davies, G., Grant, S., & Paton, J. Y. (2008). Objective measurement of physical activity and sedentary behaviour: Review with new data. Archives of Disease in Childhood, 93, Rowlands, A. V. (2007). Accelerometer assessment of physical activity in children: An update. Pediatric Exercise Science, 19, Rowlands, A. V., Thomas, P. W., Eston, R. G., & Topping, R. (2004). Validation of the RT3 triaxial accelerometer for the assessment of physical activity. Medicine and Science in Sports and Exercise, 36,

10 Downloaded by [b-on: Biblioteca do conhecimento online UTL] at 07:50 14 July 2015 Calibration of motion sensors 99 Strong, W. B., Malina, R. M., Blimkie, C. J., Daniels, S. R., Dishman, R. K., Gutin, B., et al. (2005). Evidence based physical activity for school-age youth. Journal of Pediatrics, 146, Treuth, M. S., Schmitz, K., Catellier, D. J., McMurray, R. G., Murray, D. M., Almeida, M. J., et al. (2004). Defining accelerometer thresholds for activity intensities in adolescent girls. Medicine and Science in Sports and Exercise, 36, Trost, S. G., Loprinzi, P. D., Moore, R., & Pfeiffer, K. A. (2011). Comparison of accelerometer cut-points for predicting activity intensity in youth. Medicine and Science in Sports and Exercise, 43, Vanhelst, J., Beghin, L., Rasoamanana, P., Theunynck, D., Meskini, T., Iliescu, C., et al. (2010). Calibration of the RT3 accelerometer for various patterns of physical activity in children and adolescents. Journal of Sports Sciences, 28, Vanhelst, J., Beghin, L., Turck, D., & Gottrand, F. (2011). New validated thresholds for various intensities of physical activity in adolescents using the Actigraph accelerometer. International Journal of Rehabilitation Research, 34,