The Use of Self-Tracking Technology for Health

Uitnodiging Thea Kooiman

Voor het bijwonen van de openbare verdediging van mijn proefschrift

The use of self-tracking technology for health BMI Steps Steps BMI Steps BMI BMI Steps BMI Steps BMI Steps Woensdag 7 november om 14:30 in het Academiegebouw van de Rijksuniversiteit Groningen, Broerstraat 5 te Groningen.

Aansluitend bent u van harte welkom voor een hapje en een drankje in het Goudkantoor,

The use of self-tracking technology for health Waagplein 1 te Groningen.

BMI Steps Graag voor 1 november aanmelden voor de borrel via [email protected]

Thea Kooiman [email protected] [email protected]

Paranimfen

BMI Steps Emmy Wietsma Willemke Nijholt

Contact [email protected]

BMI Steps

BMI Steps BMI

BMI Steps BMI Steps Step

The use of self-tracking s

BMI s Step technology for health BMI Steps

Thea Kooiman s Step BMI BMI Step s

The use of self-tracking technology for health Validity, adoption, and effectiveness

Thea Kooiman

The work presented in this thesis was performed at the Research Group Healthy Ageing, Allied Health Care and Nursing, of the Hanze University of Applied Sciences, Groningen, the Netherlands.

Printing this thesis was financially supported by:

- Research group Healthy Ageing, Allied Health Care and Nursing of the Hanze University of Applied Sciences - University Medical Center Groningen (UMCG) The use of self-tracking technology - University of Groningen - Graduate School for Health Services Research (SHARE) for health - Vereniging van Oefentherapeuten Cesar en Mensendieck (VvOCM) - Nederlandse Obesitas Kliniek Validity, adoption, and effectiveness

Proefschrift

ter verkrijging van de graad van doctor aan de Rijksuniversiteit Groningen op gezag van de rector magnificus prof. dr. E. Sterken en volgens besluit van het College voor Promoties.

De openbare verdediging zal plaatsvinden op Cover design: Nelson Wagenaar Lay-out: Thea Kooiman, Anne Zijlstra and GVO drukkers & vormgevers B.V. woensdag 7 november 2018 om 14.30 uur Printed by: GVO drukkers & vormgevers B.V.

ISBN: 978-94-034-1023-4 (printed version) 978-94-034-1022-7 (electronic version) ISBN: door

geboren op 26 februari 1988 All rights reserved. No parts of this publication may be reproduced, stored in a retrieval system, or transmitted te Breukelen in any form or by any means, without the prior written permission of the copyright owner.

The work presented in this thesis was performed at the Research Group Healthy Ageing, Allied Health Care and Nursing, of the Hanze University of Applied Sciences, Groningen, the Netherlands.

Printing this thesis was financially supported by:

Proefschrift

ter verkrijging van de graad van doctor aan de Rijksuniversiteit Groningen op gezag van de rector magnificus prof. dr. E. Sterken en volgens besluit van het College voor Promoties.

ISBN: 978-94-034-1023-4 (printed version) 978-94-034-1022-7 (electronic version) ISBN: door

Promotor Paranimfen Prof. dr. C.P. van der Schans Emmy Wietsma Willemke Nijholt Copromotores Dr. M. de Groot Dr. A. Kooy

Beoordelingscommissie Prof. dr. R. Sanderman Prof. dr. R.O.B. Gans Prof. dr. J.E.W.C. van Gemert-Pijnen

Promotor Paranimfen Prof. dr. C.P. van der Schans Emmy Wietsma Willemke Nijholt Copromotores Dr. M. de Groot Dr. A. Kooy

Beoordelingscommissie Prof. dr. R. Sanderman Prof. dr. R.O.B. Gans Prof. dr. J.E.W.C. van Gemert-Pijnen

Table of contents

Chapter 1 General Introduction 9

Chapter 2 Reliability and Validity of ten consumer activity trackers 17 Thea J.M. Kooiman, Manon L. Dontje, Siska R. Sprenger, Wim P. Krijnen, Cees, P. van der Schans, Martijn de Groot BMC Sports Science, Medicine and Rehabilitation (2015) 7:2 Chapter 3 Reliability and validity of consumer activity trackers depend 38 on walking speed Tryntsje Fokkema, Thea J.M. Kooiman, Wim P. Krijnen, Cees P. van der Schans, Martijn de Groot Medicine and Science in Sports and Exercise (2017) 49(4):793-800 Chapter 4 Behavioral Determinants for the Adoption of Self-tracking 55 Devices by Adults – a Longitudinal Study Thea J.M. Kooiman, A. Dijkstra, J. Timmer, Wim P. Krijnen, Adriaan Kooy, Cees P. van der Schans, Martijn de Groot Submitted Chapter 5 Do activity monitors increase physical activity in adults with 77 overweight or obesity? A systematic review and meta- analysis Herman J. de Vries, Thea J.M. Kooiman, Miriam W. van Ittersum, Marco van Brussel, Martijn de Groot. Obesity (2016) 24, 2078–2091 Chapter 6 Self-tracking of physical activity in people with type 2 100 diabetes - a randomized controlled trial Thea J.M. Kooiman, Martijn de Groot, Klaas Hoogenberg, Wim P. Krijnen, Cees P. van der Schans, Adriaan Kooy Computers, Informatics, Nursing (2018) 36(7): 340-349 Chapter 7 The role of self-regulation in the effect of self-tracking of 121 physical activity and weight on BMI Thea J.M. Kooiman, Arie Dijkstra, Adriaan Kooy, Cees P. van der Schans, Martijn de Groot Submitted Chapter 8 General Discussion 137

Samenvatting 154 Dankwoord 160 about the author 161 ן Over de auteur Research Institute SHARE 163 Acknowledgements 167

Table of contents

Chapter 1 General Introduction 9

Chapter 2 Reliability and Validity of ten consumer activity trackers 17 Thea J.M. Kooiman, Manon L. Dontje, Siska R. Sprenger, Wim P. Krijnen, Cees, P. van der Schans, Martijn de Groot BMC Sports Science, Medicine and Rehabilitation (2015) 7:2 Chapter 3 Reliability and validity of consumer activity trackers depend 39 on walking speed Tryntsje Fokkema, Thea J.M. Kooiman, Wim P. Krijnen, Cees P. van der Schans, Martijn de Groot Medicine and Science in Sports and Exercise (2017) 49(4):793-800 Chapter 4 Behavioral Determinants for the Adoption of Self-tracking 57 Devices by Adults – a Longitudinal Study Thea J.M. Kooiman, A. Dijkstra, J. Timmer, Wim P. Krijnen, Adriaan Kooy, Cees P. van der Schans, Martijn de Groot Submitted Chapter 5 Do activity monitors increase physical activity in adults with 79 overweight or obesity? A systematic review and meta- analysis Herman J. de Vries, Thea J.M. Kooiman, Miriam W. van Ittersum, Marco van Brussel, Martijn de Groot. Obesity (2016) 24, 2078–2091 Chapter 6 Self-tracking of physical activity in people with type 2 103 diabetes - a randomized controlled trial Thea J.M. Kooiman, Martijn de Groot, Klaas Hoogenberg, Wim P. Krijnen, Cees P. van der Schans, Adriaan Kooy Computers, Informatics, Nursing (2018) 36(7): 340-349 Chapter 7 The role of self-regulation in the effect of self-tracking of 125 physical activity and weight on BMI Thea J.M. Kooiman, Arie Dijkstra, Adriaan Kooy, Cees P. van der Schans, Martijn de Groot Submitted Chapter 8 General Discussion 141

Samenvatting 159 Dankwoord 166 about the author 167 ן Over de auteur Research Institute SHARE 169 Acknowledgements 173

Chapter 1 | General introduction

Chapter 1

The current health care system has begun to change. Rising costs, an ageing population, and work or social activities, and lack of exercise facilities.15 Therefore, more knowledge is an increase in the number of people with lifestyle related diseases have exposed the need needed on how to overcome these barriers to physical activity and how to increase physical for a transformation from a centralized health care model towards one that is user-centered activity in the general population, individuals with overweight/obesity, and those with type 2 and preventive.1,2 In the centralized model, health care is delivered from centralized places diabetes. such as health care institutions and hospitals. Patients have a relatively inactive role in their A possible, relatively new approach for stimulation of physical activity is the disease management in this model. Evaluations and patient-related measurements are deployment of eHealth technology 1,16 which refers to the use of internet and primarily conducted within the walls of the health care facility. In contrast, in the user- communication technologies to improve health, well-being, and healthcare.16 The centered model, self-management is an important concept. Self-management is defined as manifestation of eHealth technology is very varied and can include digital health platforms, the tasks that a person must do to monitor their own health and to make adjustments difference types of self-monitoring devices, smartphone applications, and patient monitoring towards a satisfying health.3 The patient is more informed, has more responsibility, and is systems. eHealth might have many benefits for health care delivery. For example, health even a producer of individual knowledge. This focus on self-management is also reflected in data that is electronically uploaded by patients can add valuable information for the health a renewed conceptualization of health: ‘the ability to adapt and self-manage in the face of care provider for both diagnosis and treatment purposes. In addition, the design and social, physical, and emotional challenges’.4 There is a general consensus that shifting the manifestation of eHealth technology can be adapted for a specific goal or target group, and focus from ‘care’ to ‘self-management’ is crucial for many patient groups to be able to eHealth can increase access to care (e.g., it is flexible with regard to time and place, or achieve sufficient disease management and subsequently a satisfying quality of life.1,2,5,6 individuals who have a rare disease or live in rural areas might receive better access to An important reason for the need of a renewed health care model is the large group care).1,16,17 Digital self-monitoring devices are a form of eHealth. Self-monitoring devices are of people with overweight/obesity and the related chronic diseases. This group has been often wearable devices that enable the user to monitor, for example, physical activity, diet, growing rapidly over the last few decades, with now one in every two adults in Europe sleep, respiration, heart rate, blood pressure, or blood glucose.18 These self-monitoring having overweight (BMI 25-30) and one out of six having obesity (BMI>30).7–9 devices are also known as ‘self-tracking devices’, ‘health self-quantification devices’, or Overweight/obesity increases the risk of diabetes, cardiovascular diseases, cancer, and simply ‘wearables’. Self-tracking devices are increasingly acknowledged as possible (neuro-)degenerative deterioration.10 It is well known that lifestyle factors such as physical facilitators for self-management abilities.18–21 This is because of their ability to empower inactivity strongly contribute to the onset of overweight/obesity and type 2 diabetes9,10 people with insight into their own health data and associated possibilities to stimulate the whereas engagement in sufficient physical activity such as 7500-10.000 steps per day is adoption of healthier behavior based on this data. For example, an individual who wants to associated with major health benefits such as a healthy weight and glucose metabolism, a increase his or her exercise activities is now able to gain insight into their current physical better functional performance, a lower risk for various chronic diseases, and a better quality activity pattern through the use of an activity tracker. Whit this data visualized on an of life.11,12 internet account or mobile application the user can see the course of their own physical activity pattern over time and how this is related to health recommendations such as taking However, despite of the widespread evidence of the benefits of physical activity, a certain number of steps per day. This prevents overestimation of individual physical most people do not comply with physical activity recommendations. In addition, attempts to activity behavior and might stimulate making behavioral adjustments in physical activity increase physical activity within intervention programs, e.g., by education or counseling, are habits, for example, by means of goals, reminders, prompts, and rewards. often disappointing. For instance, adherence to exercise recommendations of health care professionals has been found to be low in people with type 2 diabetes, but also in other From a theoretical perspective on behavioral change, self-monitoring of behavior is patient groups.13,14 This low adherence is caused by different reasons, e.g., lack of known as one of the self-regulation skills that are crucial to motivate and guide the desired motivation, lack of an adequate exercise plan, and an inadequate building up which causes behavior.22 Another important self-regulation skill is goal-setting. Interventions targeting injuries.14 Also, intrapersonal factors (perceived health and beliefs towards physical activity), lifestyle behaviors have been shown to be more effective when these self-regulation social factors (lack of support from friends or family), organizational factors (lack of components were included.23,24 Therefore, goal-setting and self-monitoring of behavior are accessible exercise facilities and costs), and environmental factors (friendly physical activity used as important Behavioral Change Techniques (BCTs).25,26 Other important BCTs are environment) have been determined as being relevant barriers for people with mobility providing information (tips and suggestions on how to increase physical activity), prompting problems to engage in physical activity.13 Barriers for exercise in a general middle aged and review of behavior, providing feedback on behavior, and rewards. All of these BCTs are being elderly population are partly overlapping: lack of time, tiredness, lack of knowledge how to increasingly incorporated within modern consumer activity trackers.27 What is conceptually be active, inconvenience of being active, lack of an exercise companion, interference with new about these devices is that they provide the user with objective knowledge about

10 General introduction

The current health care system has begun to change. Rising costs, an ageing population, and work or social activities, and lack of exercise facilities.15 Therefore, more knowledge is an increase in the number of people with lifestyle related diseases have exposed the need needed on how to overcome these barriers to physical activity and how to increase physical for a transformation from a centralized health care model towards one that is user-centered activity in the general population, individuals with overweight/obesity, and those with type 2 1 and preventive.1,2 In the centralized model, health care is delivered from centralized places diabetes. such as health care institutions and hospitals. Patients have a relatively inactive role in their A possible, relatively new approach for stimulation of physical activity is the disease management in this model. Evaluations and patient-related measurements are deployment of eHealth technology 1,16 which refers to the use of internet and primarily conducted within the walls of the health care facility. In contrast, in the user- communication technologies to improve health, well-being, and healthcare.16 The centered model, self-management is an important concept. Self-management is defined as manifestation of eHealth technology is very varied and can include digital health platforms, the tasks that a person must do to monitor their own health and to make adjustments difference types of self-monitoring devices, smartphone applications, and patient monitoring towards a satisfying health.3 The patient is more informed, has more responsibility, and is systems. eHealth might have many benefits for health care delivery. For example, health even a producer of individual knowledge. This focus on self-management is also reflected in data that is electronically uploaded by patients can add valuable information for the health a renewed conceptualization of health: ‘the ability to adapt and self-manage in the face of care provider for both diagnosis and treatment purposes. In addition, the design and social, physical, and emotional challenges’.4 There is a general consensus that shifting the manifestation of eHealth technology can be adapted for a specific goal or target group, and focus from ‘care’ to ‘self-management’ is crucial for many patient groups to be able to eHealth can increase access to care (e.g., it is flexible with regard to time and place, or achieve sufficient disease management and subsequently a satisfying quality of life.1,2,5,6 individuals who have a rare disease or live in rural areas might receive better access to An important reason for the need of a renewed health care model is the large group care).1,16,17 Digital self-monitoring devices are a form of eHealth. Self-monitoring devices are of people with overweight/obesity and the related chronic diseases. This group has been often wearable devices that enable the user to monitor, for example, physical activity, diet, growing rapidly over the last few decades, with now one in every two adults in Europe sleep, respiration, heart rate, blood pressure, or blood glucose.18 These self-monitoring having overweight (BMI 25-30) and one out of six having obesity (BMI>30).7–9 devices are also known as ‘self-tracking devices’, ‘health self-quantification devices’, or Overweight/obesity increases the risk of diabetes, cardiovascular diseases, cancer, and simply ‘wearables’. Self-tracking devices are increasingly acknowledged as possible (neuro-)degenerative deterioration.10 It is well known that lifestyle factors such as physical facilitators for self-management abilities.18–21 This is because of their ability to empower inactivity strongly contribute to the onset of overweight/obesity and type 2 diabetes9,10 people with insight into their own health data and associated possibilities to stimulate the whereas engagement in sufficient physical activity such as 7500-10.000 steps per day is adoption of healthier behavior based on this data. For example, an individual who wants to associated with major health benefits such as a healthy weight and glucose metabolism, a increase his or her exercise activities is now able to gain insight into their current physical better functional performance, a lower risk for various chronic diseases, and a better quality activity pattern through the use of an activity tracker. Whit this data visualized on an of life.11,12 internet account or mobile application the user can see the course of their own physical activity pattern over time and how this is related to health recommendations such as taking However, despite of the widespread evidence of the benefits of physical activity, a certain number of steps per day. This prevents overestimation of individual physical most people do not comply with physical activity recommendations. In addition, attempts to activity behavior and might stimulate making behavioral adjustments in physical activity increase physical activity within intervention programs, e.g., by education or counseling, are habits, for example, by means of goals, reminders, prompts, and rewards. often disappointing. For instance, adherence to exercise recommendations of health care professionals has been found to be low in people with type 2 diabetes, but also in other From a theoretical perspective on behavioral change, self-monitoring of behavior is patient groups.13,14 This low adherence is caused by different reasons, e.g., lack of known as one of the self-regulation skills that are crucial to motivate and guide the desired motivation, lack of an adequate exercise plan, and an inadequate building up which causes behavior.22 Another important self-regulation skill is goal-setting. Interventions targeting injuries.14 Also, intrapersonal factors (perceived health and beliefs towards physical activity), lifestyle behaviors have been shown to be more effective when these self-regulation social factors (lack of support from friends or family), organizational factors (lack of components were included.23,24 Therefore, goal-setting and self-monitoring of behavior are accessible exercise facilities and costs), and environmental factors (friendly physical activity used as important Behavioral Change Techniques (BCTs).25,26 Other important BCTs are environment) have been determined as being relevant barriers for people with mobility providing information (tips and suggestions on how to increase physical activity), prompting problems to engage in physical activity.13 Barriers for exercise in a general middle aged and review of behavior, providing feedback on behavior, and rewards. All of these BCTs are being elderly population are partly overlapping: lack of time, tiredness, lack of knowledge how to increasingly incorporated within modern consumer activity trackers.27 What is conceptually be active, inconvenience of being active, lack of an exercise companion, interference with new about these devices is that they provide the user with objective knowledge about

11 Chapter 1 individual daily routines and provide different forms of feedback which stimulates learning.28 Aims and outline of this dissertation This may enhance sustained behavior change.29,30

Modern consumer level technology may thus have potential for broad applications This dissertation aims to increase knowledge about the use and effectiveness of eHealth and for both general public health purposes and within health care for specific patient groups. self-monitoring techniques, especially activity trackers, in the current healthcare system. The However, before activity trackers can be deployed within health care, they must comply with focus will be on the general population as well on people with overweight/obesity and those certain conditions such as a satisfying reliability and validity. Not much is known yet about with type 2 diabetes. Three domains will be distinguished. the reliability and validity of the large number of activity trackers that are currently on the market. This information is very important for users, health care providers, and researchers The first domain is the reliability and validity of new consumer self-tracking devices. Before in order to be able to rely on information from these devices. new technology can be integrated into health care, it must be known whether these devices are reliable and valid. Therefore, the purpose of Chapter 2 and 3 is to examine the reliability Another important point of consideration when using self-tracking technology, is and validity of 20 activity trackers, apps, and smartwatches. knowledge about the adoption of these devices. Before self-tracking technology can impact behavior, they must be adopted by the user, and there has to be an certain engagement The second domain focuses on the adoption of self-monitoring devices in the general with the device.31 Although the development of self-monitoring technology has led to an population. For this, the purpose in Chapter 4 is to examine the adoption and factors increased number of people who actively engage in self-measurements,18 the sustained use associated with the adoption of self-tracking devices that quantify physical activity, sleep, of wearable devices by consumers is not yet that high. Several studies on the adoption of and weight. consumer level self-tracking devices have found that the percentages of people who stopped The third domain is about the effect of using consumer level activity-tracking devices and using their device within relative short-term follow up periods may vary between 33 and eHealth applications in healthy people, people with overweight/obesity, and those with type 32–35 75%. Factors that influence the adoption of self-monitoring technology are not yet 2 diabetes. First, the aim of Chapter 5 is to systematically review all studies done so far to completely clear. Factors determined thus far include different types, including personal the impact of self-monitoring of physical activity on activity levels in people with overweight factors, devices factors, and behavioral factors.32–34,36 Not much is known yet about theory and obesity. In Chapter 6 a randomized controlled trial will be conducted to the effect of a based behavioral factors that explain adoption of technology. Knowledge about these factors modern consumer activity tracker connected to an online lifestyle program on physical is important in order to be able to tailor interventions based on this information, direct activity, glycemic control, and other health outcome measures in people with type 2 future developments, or possibly distinguish between people who may or may not be diabetes. The purpose of Chapter 7 is to investigate short-term and long-term effects of self- suitable for treatment with use of modern consumer level technology. tracking of physical activity and weight on BMI change in the general population and to what extent a change in self-regulation capabilities can explain weight loss. When activity trackers are employed as (part of) an intervention to increase physical activity, an important condition is knowledge about the potential effects this technology may have. At the moment, eHealth including consumer activity trackers is not yet widely used in healthcare.16,37,38 In accordance with this, not much is known yet about the impact of these devices on lifestyle behaviors and health outcomes both in the general population as well as for people who have overweight/obesity, or type 2 diabetes. As described earlier, the use of consumer-level self-monitoring devices that can measure physical activity might be an effective approach for the incremental increase of physical activity, including people with overweight/obesity or individuals who have type 2 diabetes. However, thus far, mostly simple pedometers without additional BCTs have been deployed in interventions targeting this group, and the evidence for physical activity and health outcome measures such as weight/BMI and glycemic control is not yet conclusive.39,40 Therefore, more knowledge is needed about the effectiveness of eHealth technology, including the use of activity trackers.

12 General introduction individual daily routines and provide different forms of feedback which stimulates learning.28 Aims and outline of this dissertation This may enhance sustained behavior change.29,30 1 Modern consumer level technology may thus have potential for broad applications This dissertation aims to increase knowledge about the use and effectiveness of eHealth and for both general public health purposes and within health care for specific patient groups. self-monitoring techniques, especially activity trackers, in the current healthcare system. The However, before activity trackers can be deployed within health care, they must comply with focus will be on the general population as well on people with overweight/obesity and those certain conditions such as a satisfying reliability and validity. Not much is known yet about with type 2 diabetes. Three domains will be distinguished. the reliability and validity of the large number of activity trackers that are currently on the market. This information is very important for users, health care providers, and researchers The first domain is the reliability and validity of new consumer self-tracking devices. Before in order to be able to rely on information from these devices. new technology can be integrated into health care, it must be known whether these devices are reliable and valid. Therefore, the purpose of Chapter 2 and 3 is to examine the reliability Another important point of consideration when using self-tracking technology, is and validity of 20 activity trackers, apps, and smartwatches. knowledge about the adoption of these devices. Before self-tracking technology can impact behavior, they must be adopted by the user, and there has to be an certain engagement The second domain focuses on the adoption of self-monitoring devices in the general with the device.31 Although the development of self-monitoring technology has led to an population. For this, the purpose in Chapter 4 is to examine the adoption and factors increased number of people who actively engage in self-measurements,18 the sustained use associated with the adoption of self-tracking devices that quantify physical activity, sleep, of wearable devices by consumers is not yet that high. Several studies on the adoption of and weight. consumer level self-tracking devices have found that the percentages of people who stopped The third domain is about the effect of using consumer level activity-tracking devices and using their device within relative short-term follow up periods may vary between 33 and eHealth applications in healthy people, people with overweight/obesity, and those with type 32–35 75%. Factors that influence the adoption of self-monitoring technology are not yet 2 diabetes. First, the aim of Chapter 5 is to systematically review all studies done so far to completely clear. Factors determined thus far include different types, including personal the impact of self-monitoring of physical activity on activity levels in people with overweight factors, devices factors, and behavioral factors.32–34,36 Not much is known yet about theory and obesity. In Chapter 6 a randomized controlled trial will be conducted to the effect of a based behavioral factors that explain adoption of technology. Knowledge about these factors modern consumer activity tracker connected to an online lifestyle program on physical is important in order to be able to tailor interventions based on this information, direct activity, glycemic control, and other health outcome measures in people with type 2 future developments, or possibly distinguish between people who may or may not be diabetes. The purpose of Chapter 7 is to investigate short-term and long-term effects of self- suitable for treatment with use of modern consumer level technology. tracking of physical activity and weight on BMI change in the general population and to what extent a change in self-regulation capabilities can explain weight loss. When activity trackers are employed as (part of) an intervention to increase physical activity, an important condition is knowledge about the potential effects this technology may have. At the moment, eHealth including consumer activity trackers is not yet widely used in healthcare.16,37,38 In accordance with this, not much is known yet about the impact of these devices on lifestyle behaviors and health outcomes both in the general population as well as for people who have overweight/obesity, or type 2 diabetes. As described earlier, the use of consumer-level self-monitoring devices that can measure physical activity might be an effective approach for the incremental increase of physical activity, including people with overweight/obesity or individuals who have type 2 diabetes. However, thus far, mostly simple pedometers without additional BCTs have been deployed in interventions targeting this group, and the evidence for physical activity and health outcome measures such as weight/BMI and glycemic control is not yet conclusive.39,40 Therefore, more knowledge is needed about the effectiveness of eHealth technology, including the use of activity trackers.

13 Chapter 1

References associated with increased effectiveness in dietary and physical activity interventions. BMC Public Health. 2011;11(1):1. 24. Teixeira PJ, Carraa E V, Marques MM, et al. Successful behavior change in obesity interventions in adults: a systematic review of self-regulation mediators. BMC Med. 2015;13(1):1. 1. Arnrich B, Mayora O, Bardram J, Trster G. Pervasive healthcare. Methods Inf Med. 2010;49(1):67-73. 25. Michie S, Ashford S, Sniehotta FF, Dombrowski SU, Bishop A, French DP. A refined taxonomy of 2. Swan M. Emerging patient-driven health care models: an examination of health social networks, behaviour change techniques to help people change their physical activity and healthy eating consumer personalized medicine and quantified self-tracking. Int J Environ Res Public Health. behaviours: The CALO-RE taxonomy. Psychol Health. 2011;26(11):1479-1498. 2009;6(2):492-525. doi:10.1080/08870446.2010.540664. 3. El-Gayar O, Timsina P, Nawar N, Eid W. A systematic review of IT for diabetes self-management: Are 26. Michie S, Richardson M, Johnston M, et al. The behavior change technique taxonomy (v1) of 93 we there yet? Int J Med Inform. 2013;82(8):637-652. doi:10.1016/j.ijmedinf.2013.05.006. hierarchically clustered techniques: building an international consensus for the reporting of behavior 4. Huber M, André Knottnerus J, Green L, et al. How should we define health? BMJ. 2011;343(7817). change interventions. Ann Behav Med. 2013;46(1):81-95. doi:10.1136/bmj.d4163. 27. Lyons EJ, Lewis ZH, Mayrsohn BG, Rowland JL. Behavior change techniques implemented in electronic 5. Sharon T. Self-Tracking for Health and the Quantified Self: Re-Articulating Autonomy, Solidarity, and lifestyle activity monitors: A systematic content analysis. J Med Internet Res. 2014;16(8). Authenticity in an Age of Personalized Healthcare. Philos Technol. 2016:1-29. doi:10.2196/jmir.3469. 6. Panagioti M, Richardson G, Small N, et al. Self-management support interventions to reduce health 28. Menninga K. Learning abstinence theory - PhD thesis under supervision of A. Dijkstra. Univ Groningen. care utilisation without compromising outcomes: a systematic review and meta-analysis. BMC Health 2012. Serv Res. 2014;14(1):356. doi:10.1186/1472-6963-14-356. 29. Carver CS, Scheier MF. Control theory: A useful conceptual framework for personality-social, clinical, 7. Tamayo T, Rosenbauer J, Wild SH, et al. Diabetes in Europe: An update. Diabetes Res Clin Pract. and health psychology. Psychol Bull. 1982;92(1):111-135. doi:10.1037/0033-2909.92.1.111. 2014;103(2):206-217. doi:10.1016/j.diabres.2013.11.007. 30. Kluger AN, DeNisi A. The effects of feedback interventions on performance: A historical review, a 8. Eurostat. European Health Interview Survey: Almost 1 adult in 6 in the EU is considered obese. Prem meta-analysis, and a preliminary feedback intervention theory. Psychol Bull. 1996;119(2):254-284. Off News. 2016;2014(October):7-11. http://ec.europa.eu/eurostat/documents/2995521/7700898/3- doi:10.1037/0033-2909.119.2.254. 20102016-BP-EN.pdf/c26b037b-d5f3-4c05-89c1-00bf0b98d646. 31. Perski O, Blandford A, West R, Michie S. Conceptualising engagement with digital behaviour change 9. Lee IM, Shiroma EJ, Lobelo F, et al. Effect of physical inactivity on major non-communicable diseases interventions: a systematic review using principles from critical interpretive synthesis. Transl Behav worldwide: An analysis of burden of disease and life expectancy. Lancet. 2012;380(9838):219-229. Med. 2016:1-14. doi:10.1016/S0140-6736(12)61031-9. 32. Fritz T, Huang EM, Murphy GC, Zimmermann T. Persuasive technology in the real world: a study of 10. Musto AA. The effects of an incremental pedometer program on metabolic syndrome components in long-term use of activity sensing devices for fitness. In: Proceedings of the SIGCHI Conference on sedentary overweight women. Diss Abstr Int Sect B Sci Eng. 2008;69(3-B):1598. Human Factors in Computing Systems. ACM; 2014:487-496. http://search.ebscohost.com/login.aspx?direct=true&AuthType=ip,shib&db=psyh&AN=2008-99180- 33. Gouveia R, Karapanos E, Hassenzahl M. How do we engage with activity trackers?: a longitudinal study 343&site=ehost-live&custid=s4121186. of Habito. In: Proceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous 11. Haskell WL, Lee IM, Pate RR, et al. Physical activity and public health: Updated recommendation for Computing. ACM; 2015:1305-1316. adults from the American College of Sports Medicine and the American Heart Association. Med Sci 34. Lazar A, Koehler C, Tanenbaum J, Nguyen DH. Why we use and abandon smart devices. In: Proceedings Sports Exerc. 2007;39(8):1423-1434. doi:10.1249/mss.0b013e3180616b27. of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing. ACM; 12. Tudor-Locke C, Craig CL, Brown WJ, et al. How many steps/day are enough? for adults. Int J Behav Nutr 2015:635-646. Phys Act. 2011;8(1):79. doi:10.1186/1479-5868-8-79. 35. Clawson J, Pater JA, Miller AD, Mynatt ED, Mamykina L. No longer wearing: investigating the 13. Vasudevan V, Rimmer JH, Kviz F. Development of the Barriers to Physical Activity Questionnaire for abandonment of personal health-tracking technologies on craigslist. In: Proceedings of the 2015 ACM People with Mobility Impairments. Disabil Health J. 2015;8(4):547-556. International Joint Conference on Pervasive and Ubiquitous Computing. ACM; 2015:647-658. doi:10.1016/j.dhjo.2015.04.007. 36. Epstein D, Caraway M, Johnston C, Ping A, Fogarty J, Munson S. Beyond Abandonment to Next Steps. 14. García-Pérez, L. E., Álvarez, M., Dilla, T., Gil-Guillén, V., & Orozco-Beltrán D. Adherence to Therapies in In: CHI ’16. ACM; :1109-1113. doi:10.1145/2858036.2858045. Patients with Type 2 Diabetes. Diabetes Ther. 2013;4(2):175-194. doi:10.1007/s13300-013-0034-y. 37. Griebel L, Kolominsky-Rabas P, Schaller S, et al. Acceptance by laypersons and medical professionals of 15. Justine M, Azian A, Hassan V, Manaf H. Barriers to participation in physical activity and exercise among the personalized eHealth platform, eHealthMonitor. Informatics Heal Soc Care. 2016:1-18. middle-aged and elderly individuals. Singapore Med J. 2013;54(10):581-586. 38. Paton C, Hansen M, Fernandez-Luque L, Lau a YS. Self-Tracking, Social Media and Personal Health 16. Van Gemert-Pijnen, J.E.W.C, Peters,O., & Ossebaard HC. Improving eHealth. Eleven International Records for Patient Empowered Self-Care. Contribution of the IMIA Social Media Working Group. Publishing; 2013. http://lib.myilibrary.com?ID=673579. Yearb Med Inform. 2012;7(1):16-24. doi:me12010016 [pii]. 17. Steinhubl SR, Muse ED, Topol EJ. Can Mobile Health Technologies Transform Health Care? Jama. 39. Rollo M, Aguiar E, Williams R. eHealth technologies to support nutrition and physical activity behaviors 2013;310(22):2395. doi:10.1001/jama.2013.281078. in diabetes self-management. Obes Targets …. 2016;9:381. 18. Fawcett T. Mining the Quantified Self: Personal Knowledge Discovery as a Challenge for Data Science. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5104301/. Big Data. 2015;3(4):249-266. internal-pdf://228.60.152.96/Fawcett. 40. McMillan KA, Kirk A, Hewitt A, MacRury S. A systematic and integrated review of mobile-based 19. Almalki M, Gray K, Martin-Sanchez F. Activity Theory as a Theoretical Framework for Health Self- technology to promote active lifestyles in people with type 2 diabetes. J Diabetes Sci Technol. Quantification: A Systematic Review of Empirical Studies. J Med Internet Res. 2016;18(5):e131. 2017;11(2):299-307. 20. Grnvall E, Verdezoto N. Beyond self-monitoring: understanding non-functional aspects of home-based

healthcare technology. In: Proceedings of the 2013 ACM International Joint Conference on Pervasive and Ubiquitous Computing. ACM; 2013:587-596. 21. Almalki M, Gray K, Martin-Sanchez F. Refining the Concepts of Self-quantification Needed for Health Self-management: A Thematic Literature Review. Computer (Long Beach Calif). 2015;79:1-5. 22. Bandura A. Health promotion from the perspective of social cognitive theory. Psychol Heal. 1998;13(4):623-649. 23. Greaves CJ, Sheppard KE, Abraham C, et al. Systematic review of reviews of intervention components

14 General introduction

References associated with increased effectiveness in dietary and physical activity interventions. BMC Public Health. 2011;11(1):1. 24. Teixeira PJ, Carraa E V, Marques MM, et al. Successful behavior change in obesity interventions in 1 adults: a systematic review of self-regulation mediators. BMC Med. 2015;13(1):1. 1. Arnrich B, Mayora O, Bardram J, Trster G. Pervasive healthcare. Methods Inf Med. 2010;49(1):67-73. 25. Michie S, Ashford S, Sniehotta FF, Dombrowski SU, Bishop A, French DP. A refined taxonomy of 2. Swan M. Emerging patient-driven health care models: an examination of health social networks, behaviour change techniques to help people change their physical activity and healthy eating consumer personalized medicine and quantified self-tracking. Int J Environ Res Public Health. behaviours: The CALO-RE taxonomy. Psychol Health. 2011;26(11):1479-1498. 2009;6(2):492-525. doi:10.1080/08870446.2010.540664. 3. El-Gayar O, Timsina P, Nawar N, Eid W. A systematic review of IT for diabetes self-management: Are 26. Michie S, Richardson M, Johnston M, et al. The behavior change technique taxonomy (v1) of 93 we there yet? Int J Med Inform. 2013;82(8):637-652. doi:10.1016/j.ijmedinf.2013.05.006. hierarchically clustered techniques: building an international consensus for the reporting of behavior 4. Huber M, André Knottnerus J, Green L, et al. How should we define health? BMJ. 2011;343(7817). change interventions. Ann Behav Med. 2013;46(1):81-95. doi:10.1136/bmj.d4163. 27. Lyons EJ, Lewis ZH, Mayrsohn BG, Rowland JL. Behavior change techniques implemented in electronic 5. Sharon T. Self-Tracking for Health and the Quantified Self: Re-Articulating Autonomy, Solidarity, and lifestyle activity monitors: A systematic content analysis. J Med Internet Res. 2014;16(8). Authenticity in an Age of Personalized Healthcare. Philos Technol. 2016:1-29. doi:10.2196/jmir.3469. 6. Panagioti M, Richardson G, Small N, et al. Self-management support interventions to reduce health 28. Menninga K. Learning abstinence theory - PhD thesis under supervision of A. Dijkstra. Univ Groningen. care utilisation without compromising outcomes: a systematic review and meta-analysis. BMC Health 2012. Serv Res. 2014;14(1):356. doi:10.1186/1472-6963-14-356. 29. Carver CS, Scheier MF. Control theory: A useful conceptual framework for personality-social, clinical, 7. Tamayo T, Rosenbauer J, Wild SH, et al. Diabetes in Europe: An update. Diabetes Res Clin Pract. and health psychology. Psychol Bull. 1982;92(1):111-135. doi:10.1037/0033-2909.92.1.111. 2014;103(2):206-217. doi:10.1016/j.diabres.2013.11.007. 30. Kluger AN, DeNisi A. The effects of feedback interventions on performance: A historical review, a 8. Eurostat. European Health Interview Survey: Almost 1 adult in 6 in the EU is considered obese. Prem meta-analysis, and a preliminary feedback intervention theory. Psychol Bull. 1996;119(2):254-284. Off News. 2016;2014(October):7-11. http://ec.europa.eu/eurostat/documents/2995521/7700898/3- doi:10.1037/0033-2909.119.2.254. 20102016-BP-EN.pdf/c26b037b-d5f3-4c05-89c1-00bf0b98d646. 31. Perski O, Blandford A, West R, Michie S. Conceptualising engagement with digital behaviour change 9. Lee IM, Shiroma EJ, Lobelo F, et al. Effect of physical inactivity on major non-communicable diseases interventions: a systematic review using principles from critical interpretive synthesis. Transl Behav worldwide: An analysis of burden of disease and life expectancy. Lancet. 2012;380(9838):219-229. Med. 2016:1-14. doi:10.1016/S0140-6736(12)61031-9. 32. Fritz T, Huang EM, Murphy GC, Zimmermann T. Persuasive technology in the real world: a study of 10. Musto AA. The effects of an incremental pedometer program on metabolic syndrome components in long-term use of activity sensing devices for fitness. In: Proceedings of the SIGCHI Conference on sedentary overweight women. Diss Abstr Int Sect B Sci Eng. 2008;69(3-B):1598. Human Factors in Computing Systems. ACM; 2014:487-496. http://search.ebscohost.com/login.aspx?direct=true&AuthType=ip,shib&db=psyh&AN=2008-99180- 33. Gouveia R, Karapanos E, Hassenzahl M. How do we engage with activity trackers?: a longitudinal study 343&site=ehost-live&custid=s4121186. of Habito. In: Proceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous 11. Haskell WL, Lee IM, Pate RR, et al. Physical activity and public health: Updated recommendation for Computing. ACM; 2015:1305-1316. adults from the American College of Sports Medicine and the American Heart Association. Med Sci 34. Lazar A, Koehler C, Tanenbaum J, Nguyen DH. Why we use and abandon smart devices. In: Proceedings Sports Exerc. 2007;39(8):1423-1434. doi:10.1249/mss.0b013e3180616b27. of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing. ACM; 12. Tudor-Locke C, Craig CL, Brown WJ, et al. How many steps/day are enough? for adults. Int J Behav Nutr 2015:635-646. Phys Act. 2011;8(1):79. doi:10.1186/1479-5868-8-79. 35. Clawson J, Pater JA, Miller AD, Mynatt ED, Mamykina L. No longer wearing: investigating the 13. Vasudevan V, Rimmer JH, Kviz F. Development of the Barriers to Physical Activity Questionnaire for abandonment of personal health-tracking technologies on craigslist. In: Proceedings of the 2015 ACM People with Mobility Impairments. Disabil Health J. 2015;8(4):547-556. International Joint Conference on Pervasive and Ubiquitous Computing. ACM; 2015:647-658. doi:10.1016/j.dhjo.2015.04.007. 36. Epstein D, Caraway M, Johnston C, Ping A, Fogarty J, Munson S. Beyond Abandonment to Next Steps. 14. García-Pérez, L. E., Álvarez, M., Dilla, T., Gil-Guillén, V., & Orozco-Beltrán D. Adherence to Therapies in In: CHI ’16. ACM; :1109-1113. doi:10.1145/2858036.2858045. Patients with Type 2 Diabetes. Diabetes Ther. 2013;4(2):175-194. doi:10.1007/s13300-013-0034-y. 37. Griebel L, Kolominsky-Rabas P, Schaller S, et al. Acceptance by laypersons and medical professionals of 15. Justine M, Azian A, Hassan V, Manaf H. Barriers to participation in physical activity and exercise among the personalized eHealth platform, eHealthMonitor. Informatics Heal Soc Care. 2016:1-18. middle-aged and elderly individuals. Singapore Med J. 2013;54(10):581-586. 38. Paton C, Hansen M, Fernandez-Luque L, Lau a YS. Self-Tracking, Social Media and Personal Health 16. Van Gemert-Pijnen, J.E.W.C, Peters,O., & Ossebaard HC. Improving eHealth. Eleven International Records for Patient Empowered Self-Care. Contribution of the IMIA Social Media Working Group. Publishing; 2013. http://lib.myilibrary.com?ID=673579. Yearb Med Inform. 2012;7(1):16-24. doi:me12010016 [pii]. 17. Steinhubl SR, Muse ED, Topol EJ. Can Mobile Health Technologies Transform Health Care? Jama. 39. Rollo M, Aguiar E, Williams R. eHealth technologies to support nutrition and physical activity behaviors 2013;310(22):2395. doi:10.1001/jama.2013.281078. in diabetes self-management. Obes Targets …. 2016;9:381. 18. Fawcett T. Mining the Quantified Self: Personal Knowledge Discovery as a Challenge for Data Science. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5104301/. Big Data. 2015;3(4):249-266. internal-pdf://228.60.152.96/Fawcett. 40. McMillan KA, Kirk A, Hewitt A, MacRury S. A systematic and integrated review of mobile-based 19. Almalki M, Gray K, Martin-Sanchez F. Activity Theory as a Theoretical Framework for Health Self- technology to promote active lifestyles in people with type 2 diabetes. J Diabetes Sci Technol. Quantification: A Systematic Review of Empirical Studies. J Med Internet Res. 2016;18(5):e131. 2017;11(2):299-307. 20. Grnvall E, Verdezoto N. Beyond self-monitoring: understanding non-functional aspects of home-based healthcare technology. In: Proceedings of the 2013 ACM International Joint Conference on Pervasive and Ubiquitous Computing. ACM; 2013:587-596. 21. Almalki M, Gray K, Martin-Sanchez F. Refining the Concepts of Self-quantification Needed for Health Self-management: A Thematic Literature Review. Computer (Long Beach Calif). 2015;79:1-5. 22. Bandura A. Health promotion from the perspective of social cognitive theory. Psychol Heal. 1998;13(4):623-649. 23. Greaves CJ, Sheppard KE, Abraham C, et al. Systematic review of reviews of intervention components

15 Chapter 2 | Reliability and validity of ten consumer activity trackers

Thea J.M. Kooiman Manon L. Dontje Siska R. Sprenger Wim P. Krijnen Cees P. van der Schans Martijn de Groot

BMC Sports Science, Medicine and Rehabilitation (2015) 7:24

Chapter 2 | Reliability and validity of ten consumer activity trackers

Thea J.M. Kooiman Manon L. Dontje Siska R. Sprenger Wim P. Krijnen Cees P. van der Schans Martijn de Groot

BMC Sports Science, Medicine and Rehabilitation (2015) 7:24

Chapter 2

Abstract Introduction

Background Activity trackers are developed to increase an individual’s awareness about physical activity Activity trackers can potentially stimulate users to increase their physical activity behavior. behavior throughout the day. It is well known that regular physical activity decreases the risk The aim of this study was to examine the reliability and validity of ten consumer activity of many chronic diseases and can improve quality of life.1-3 A commonly used physical trackers for measuring step count in both laboratory and free-living conditions. activity guideline is the 10,000 steps/day norm: healthy adults are recommended to take 10,000 steps per day to maintain physical fitness and health.4 However, many people 1 Method worldwide are not aware if they comply with this recommendation. In addition, previous Healthy adult volunteers (n=33) walked twice on a treadmill (4.8 km/h) for 30 minutes while research has indicated that most people tend to overestimate their level of physical activity wearing ten different activity trackers (i.e. Lumoback, Fitbit Flex, Jawbone Up, Nike+ [5, 6]. Activity trackers may potentially overcome this issue. Fuelband SE, Misfit Shine, Withings Pulse, Fitbit Zip, Omron HJ-203, Yamax Digiwalker SW- Over the past five to ten years, an increasing number and variety of activity trackers 200 and Moves mobile application). In free-living conditions, 56 volunteers wore the same have become available on the consumer market. Activity trackers are small and user-friendly activity trackers for one working day. Test-retest reliability was analyzed with the Intraclass devices that measure the number of steps taken and/or the amount of time spent Correlation Coefficient (ICC). Validity was evaluated by comparing each tracker with the gold performing physical activities at different intensities. Most activity trackers also convert the standard (Optogait system for laboratory and ActivPAL for free-living conditions), using number of steps with algorithms into measures such as the distance covered and the paired samples t-tests, mean absolute percentage errors, correlations and Bland-Altman number of calories burned. Associated (mobile) applications provide users with insight into plots. their individual physical activity behavior over a certain period of time. This might work as a motivator to increase physical activity.7,8 Consumer activity trackers might also be beneficial Results for scientific research, due to their ease of usability and relatively low cost. Examples of Test-retest analysis revealed high reliability for most trackers except for the Omron (ICC .14), popular devices are the Fitbit, Jawbone Up, and Withings Pulse. Moves app (ICC .37) and Nike+ Fuelband (ICC .53). The mean absolute percentage errors of For accurate measurement and interpretation of the data, these devices must be the trackers in laboratory and free-living conditions respectively, were: Lumoback (-0.2, -0.4), reliable and valid. A number of studies have examined consumer tracker accuracy,6,9-18 Fibit Flex (-5.7, 3.7), Jawbone Up (-1.0, 1.4), Nike+ Fuelband (-18, -24), Misfit Shine (0.2, 1.1), however, six studies were based upon earlier versions of Fitbit devices, and the methodology Withings Pulse (-0.5, -7.9), Fitbit Zip (-0.3, 1.2), Omron (2.5, -0.4), Digiwalker (-1.2, -5.9), and for assessing reliability and validity varied considerably. For example, different types of Moves app (9.6, -37.6). Bland-Altman plots demonstrated that the limits of agreement varied activity were used (walking on a treadmill at different speeds, lab cycling, walking stairs, from 46 steps (Fitbit Zip) to 2422 steps (Nike+ Fuelband) in the laboratory condition, and 866 daily activities), and different gold standards were utilized (energy expenditure [EE] steps (Fitbit Zip) to 5150 steps (Moves app) in the free-living condition. measured by breath-to-breath analysis, self-reported physical activity translated to EE [in

METs], and real step count). Five studies were performed in a laboratory condition,9-11, 14,16 Conclusion and six studies examined the reliability or validity of activity trackers during (semi- The reliability and validity of most trackers for measuring step count is good. The Fitbit Zip is structured) free-living conditions.6,12,13,15,17,18 The validity of activity trackers may differ in the most valid whereas the reliability and validity of the Nike+ Fuelband is low. free-living conditions compared to standardized lab conditions because of the increased variety in walking speeds, directions, intensities, etc. in free-living. To date, no studies have assessed reliability and validity of consumer trackers in both laboratory and free-living conditions. The aim of this study was to determine the reliability and validity of ten consumer activity trackers, in both a standardized laboratory condition and in free-living conditions.

18 Reliability and Validity of ten consumer activity trackers

Abstract Introduction

Background Activity trackers are developed to increase an individual’s awareness about physical activity Activity trackers can potentially stimulate users to increase their physical activity behavior. behavior throughout the day. It is well known that regular physical activity decreases the risk The aim of this study was to examine the reliability and validity of ten consumer activity of many chronic diseases and can improve quality of life.1-3 A commonly used physical 2 trackers for measuring step count in both laboratory and free-living conditions. activity guideline is the 10,000 steps/day norm: healthy adults are recommended to take 10,000 steps per day to maintain physical fitness and health.4 However, many people 1 Method worldwide are not aware if they comply with this recommendation. In addition, previous Healthy adult volunteers (n=33) walked twice on a treadmill (4.8 km/h) for 30 minutes while research has indicated that most people tend to overestimate their level of physical activity wearing ten different activity trackers (i.e. Lumoback, Fitbit Flex, Jawbone Up, Nike+ [5, 6]. Activity trackers may potentially overcome this issue. Fuelband SE, Misfit Shine, Withings Pulse, Fitbit Zip, Omron HJ-203, Yamax Digiwalker SW- Over the past five to ten years, an increasing number and variety of activity trackers 200 and Moves mobile application). In free-living conditions, 56 volunteers wore the same have become available on the consumer market. Activity trackers are small and user-friendly activity trackers for one working day. Test-retest reliability was analyzed with the Intraclass devices that measure the number of steps taken and/or the amount of time spent Correlation Coefficient (ICC). Validity was evaluated by comparing each tracker with the gold performing physical activities at different intensities. Most activity trackers also convert the standard (Optogait system for laboratory and ActivPAL for free-living conditions), using number of steps with algorithms into measures such as the distance covered and the paired samples t-tests, mean absolute percentage errors, correlations and Bland-Altman number of calories burned. Associated (mobile) applications provide users with insight into plots. their individual physical activity behavior over a certain period of time. This might work as a motivator to increase physical activity.7,8 Consumer activity trackers might also be beneficial Results for scientific research, due to their ease of usability and relatively low cost. Examples of Test-retest analysis revealed high reliability for most trackers except for the Omron (ICC .14), popular devices are the Fitbit, Jawbone Up, and Withings Pulse. Moves app (ICC .37) and Nike+ Fuelband (ICC .53). The mean absolute percentage errors of For accurate measurement and interpretation of the data, these devices must be the trackers in laboratory and free-living conditions respectively, were: Lumoback (-0.2, -0.4), reliable and valid. A number of studies have examined consumer tracker accuracy,6,9-18 Fibit Flex (-5.7, 3.7), Jawbone Up (-1.0, 1.4), Nike+ Fuelband (-18, -24), Misfit Shine (0.2, 1.1), however, six studies were based upon earlier versions of Fitbit devices, and the methodology Withings Pulse (-0.5, -7.9), Fitbit Zip (-0.3, 1.2), Omron (2.5, -0.4), Digiwalker (-1.2, -5.9), and for assessing reliability and validity varied considerably. For example, different types of Moves app (9.6, -37.6). Bland-Altman plots demonstrated that the limits of agreement varied activity were used (walking on a treadmill at different speeds, lab cycling, walking stairs, from 46 steps (Fitbit Zip) to 2422 steps (Nike+ Fuelband) in the laboratory condition, and 866 daily activities), and different gold standards were utilized (energy expenditure [EE] steps (Fitbit Zip) to 5150 steps (Moves app) in the free-living condition. measured by breath-to-breath analysis, self-reported physical activity translated to EE [in

19 Chapter 2

Methods Testing under free-living conditions In order to examine the validity of the ten trackers in free-living conditions during a working

day, the activity behavior of the participants was measured during one working day between Study design 9.00 am and 4:30 pm. The participants wore each ten different trackers and the ActivPAL The following ten activity trackers were examined: the Lumoback, Fitbit Flex, Nike+ Fuelband simultaneously. During the specified day, participants performed their normal daily SE, Jawbone Up, Misfit Shine, Withings Pulse, Fitbit Zip, Omron HJ-203, Yamax Digiwalker activities; however, they were requested to abstain from cycling or driving a vehicle during SW-200 and the Moves mobile application. The Optogait system (OPTOGait, Microgate S.r.I, the test period. This was required in order to be able to make a realistic comparison Italy, 2010) was used as the gold standard on the treadmill in the laboratory condition. This between the trackers; because the different wearing positions of the trackers might system consists of two beams attached to the sides of the treadmill. The system uses an LED influence step measurements during these activities. The primary outcome measure was the lighting system to precisely measure the number of steps which is a reliable and valid total number of steps measured between 9 am to 4:30 pm. 19 (PAL Technologies Ltd., Glasgow, method for measuring step count (cadence). The ActivPAL UK) was used as the gold standard in the free-living condition. The ActivPAL was worn on the Activity trackers thigh underneath the clothing. Previous research has demonstrated that the ActivPAL is a All devices utilized in this study are able to track step count. reliable and valid tool for measuring the number of steps taken both on a treadmill and in free-living conditions.20-22 Lumoback: The Lumoback™ (Lumo BodyTech, Inc. Palo Alto, California, USA) was worn around the lower back and was calibrated to the user by utilizing the associated application. Fitbit Flex: The Fitbit Flex™ (Fitbit, Inc., San Francisco, CA, USA) is a wrist-worn tri-axial Study sample accelerometer and was worn on the non-dominant arm. Only healthy adult volunteers (age ≥18, <65 years) were included in the study. Participants were recruited through advertisements within the Hanze University and by using the Jawbone UP: The Jawbone UP™ (JAWBONE, San Francisco, CA, USA, is a wrist-worn three- individual networks of the researchers. Subscribers were excluded from participation if they dimensional activity tracker and was worn on the non-dominant arm. experienced problems with standing or normal ambulation as well as if they performed daily Nike+ Fuelband: The Nike+ Fuelband SE ™ (Nike Inc., Beaverton, OR, USA) is a wrist-worn activities which could possibly damage the activity trackers while being worn (when three-dimensional activity tracker and was worn on the non-dominant arm. participating in the free-living study). All components of the study are described below in more detail. The study was in accordance with the principles as outlined in the Declaration of Misfit Shine: The Misfit Shine™ (Misfit Wearables, Burlingame, California, USA) is a small tri- Helsinki and an exemption was obtained by the Medical Ethical Committee of the University axial accelerometer which was carried in the front pocket of the trousers. Medical Center of Groningen for a comprehensive application. All participants were Pulse: The Withings Pulse™ (Withings, Issy les Moulineaux, France) is a small tri-axial informed about the study procedures and provided informed consent prior to the initiation accelerometer which was carried in the front pocket of the trousers. of this study. Fitbit Zip: The Fitbit Zip™ (Fitbit, Inc., San Francisco, CA, USA) is a small tri-axial accelerometer which was carried in the front pocket of the trousers. Testing under laboratory conditions In order to examine the test-retest reliability and the validity of the ten trackers in a Omron: The Omron Walking Style III™ (type HJ-203) (OMRON Healthcare Europe B.V., standardized situation, the participants walked for 30 minutes on a treadmill at a walking Hoofddorp, the Netherlands) is a pedometer with a two-dimensional sensor which was speed of 4.8 km/h. This walking velocity was similar to velocities used in previous treadmill carried in the front pocket of the trousers. 14,23 studies and is based on an average walking speed. During the treadmill test, the Digiwalker: The Yamax Digiwalker SW-200™ (YAMAX Health & Sports, Inc. San Antonio, USA) participants wore all ten activity trackers and the ActivPAL. The Optogait system on the is a two-dimensional pedometer that was attached to the participant’s waistband. treadmill was used as the gold standard. The primary outcome measure was the total R number of steps measured within the duration of the 30-minute treadmill test. All Moves: The Moves is a smartphone application. It uses acceleration sensors from a participants repeated this test one week later. smartphone and GPS to measure the number of steps taken. The mobile phone used in the laboratory study was an Iphone 4S (Iphone 4S, Apple Inc., USA). During the free-living study the smartphone of the participant was used (IOS/Android) and carried in the front pocket of the trousers.

20 Reliability and Validity of ten consumer activity trackers

Methods Testing under free-living conditions In order to examine the validity of the ten trackers in free-living conditions during a working

day, the activity behavior of the participants was measured during one working day between Study design 9.00 am and 4:30 pm. The participants wore each ten different trackers and the ActivPAL The following ten activity trackers were examined: the Lumoback, Fitbit Flex, Nike+ Fuelband simultaneously. During the specified day, participants performed their normal daily SE, Jawbone Up, Misfit Shine, Withings Pulse, Fitbit Zip, Omron HJ-203, Yamax Digiwalker activities; however, they were requested to abstain from cycling or driving a vehicle during 2 SW-200 and the Moves mobile application. The Optogait system (OPTOGait, Microgate S.r.I, the test period. This was required in order to be able to make a realistic comparison Italy, 2010) was used as the gold standard on the treadmill in the laboratory condition. This between the trackers; because the different wearing positions of the trackers might system consists of two beams attached to the sides of the treadmill. The system uses an LED influence step measurements during these activities. The primary outcome measure was the lighting system to precisely measure the number of steps which is a reliable and valid total number of steps measured between 9 am to 4:30 pm. 19 (PAL Technologies Ltd., Glasgow, method for measuring step count (cadence). The ActivPAL UK) was used as the gold standard in the free-living condition. The ActivPAL was worn on the Activity trackers thigh underneath the clothing. Previous research has demonstrated that the ActivPAL is a All devices utilized in this study are able to track step count. reliable and valid tool for measuring the number of steps taken both on a treadmill and in free-living conditions.20-22 Lumoback: The Lumoback™ (Lumo BodyTech, Inc. Palo Alto, California, USA) was worn around the lower back and was calibrated to the user by utilizing the associated application. Fitbit Flex: The Fitbit Flex™ (Fitbit, Inc., San Francisco, CA, USA) is a wrist-worn tri-axial Study sample accelerometer and was worn on the non-dominant arm. Only healthy adult volunteers (age ≥18, <65 years) were included in the study. Participants were recruited through advertisements within the Hanze University and by using the Jawbone UP: The Jawbone UP™ (JAWBONE, San Francisco, CA, USA, is a wrist-worn three- individual networks of the researchers. Subscribers were excluded from participation if they dimensional activity tracker and was worn on the non-dominant arm. experienced problems with standing or normal ambulation as well as if they performed daily Nike+ Fuelband: The Nike+ Fuelband SE ™ (Nike Inc., Beaverton, OR, USA) is a wrist-worn activities which could possibly damage the activity trackers while being worn (when three-dimensional activity tracker and was worn on the non-dominant arm. participating in the free-living study). All components of the study are described below in more detail. The study was in accordance with the principles as outlined in the Declaration of Misfit Shine: The Misfit Shine™ (Misfit Wearables, Burlingame, California, USA) is a small tri- Helsinki and an exemption was obtained by the Medical Ethical Committee of the University axial accelerometer which was carried in the front pocket of the trousers. Medical Center of Groningen for a comprehensive application. All participants were Pulse: The Withings Pulse™ (Withings, Issy les Moulineaux, France) is a small tri-axial informed about the study procedures and provided informed consent prior to the initiation accelerometer which was carried in the front pocket of the trousers. of this study. Fitbit Zip: The Fitbit Zip™ (Fitbit, Inc., San Francisco, CA, USA) is a small tri-axial accelerometer which was carried in the front pocket of the trousers. Testing under laboratory conditions In order to examine the test-retest reliability and the validity of the ten trackers in a Omron: The Omron Walking Style III™ (type HJ-203) (OMRON Healthcare Europe B.V., standardized situation, the participants walked for 30 minutes on a treadmill at a walking Hoofddorp, the Netherlands) is a pedometer with a two-dimensional sensor which was speed of 4.8 km/h. This walking velocity was similar to velocities used in previous treadmill carried in the front pocket of the trousers. 14,23 studies and is based on an average walking speed. During the treadmill test, the Digiwalker: The Yamax Digiwalker SW-200™ (YAMAX Health & Sports, Inc. San Antonio, USA) participants wore all ten activity trackers and the ActivPAL. The Optogait system on the is a two-dimensional pedometer that was attached to the participant’s waistband. treadmill was used as the gold standard. The primary outcome measure was the total R number of steps measured within the duration of the 30-minute treadmill test. All Moves: The Moves is a smartphone application. It uses acceleration sensors from a participants repeated this test one week later. smartphone and GPS to measure the number of steps taken. The mobile phone used in the laboratory study was an Iphone 4S (Iphone 4S, Apple Inc., USA). During the free-living study the smartphone of the participant was used (IOS/Android) and carried in the front pocket of the trousers.

21 Chapter 2

Statistical analysis age (±SD) 37.1 (± 10.6), mean BMI (±SD) 24.1 (± 2) kg/m², and 38 females, mean age (±SD) 30 A sample size analysis was conducted to calculate the number of required participants. As (± 9.5) years, mean BMI (±SD) 23.1 (± 2.5) kg/m²). Most of the participants were university previous data on relevant differences for sample size calculation does not exist, we reasoned employees, with an office job. Activities performed by the participants during the test day that a difference of 10% for the laboratory condition and 15% for the free-living condition included sitting (e.g., at the computer), standing (e.g., teaching activities) and walking. A seemed appropriate. Using these relevant differences and expected mean number of steps number of participants were highly active (e.g., took a long walk during lunch time) whereas in both conditions, it was calculated that at least 24 participants were necessary for others were mainly sedentary during the test day. The Nike+ Fuelband and Moves app were participation in the laboratory condition and 58 participants for the free-living condition to tested with a fewer number of participants in the free-living study (N=20 and N=11 enable substantiation of a relevant difference between the trackers and the gold standards respectively). The Nike+ Fuelband was not available at the beginning but was included during with a power of 80% and a significance level of 5 %. This number of participants is the study. The Moves app was unavailable at no cost for most participants in the free-living comparable to other validation studies.12,14,15 This reassured our reasoned choice for using study. In all 11 cases, the Moves app was operating on an Android device. 10% and 15% as cut-off points for the mean difference.

Descriptive statistics were used to characterize the sample. Normality of the outcome Descriptive statistics measures was tested by Shapiro Wilk for all activity trackers in both parts of the study. Figure 1 depicts the descriptive statistics (mean number of steps, 95% CI) as measured by the gold standards and by the ten activity trackers in both the laboratory (A) and free-living Test-retest reliability of the trackers in the laboratory study was assessed by calculating the condition (B). The mean number of steps (±SD) measured by the Optogait in the laboratory Intraclass Correlation Coefficient (ICC) (two-way random, absolute agreement, single condition was 3314 (± 162), and the mean number of steps (±SD) measured by the ten measures with a 95% confidence interval). Common cut-off points for reliability assessment trackers ranged from 2716 (± 672) [Nike+ Fuelband] to 3633 (± 286) [Moves app]. The mean were used; >.90 (excellent), .75-.90 (good), .60-.75 (moderate), and <.60 (low).24 number of steps (±SD) measured by the ActivPAL in the free-living condition was 4070 (± The validity of the ten trackers was determined by several statistical tests. First, systematic 2430), and the mean number of steps (±SD) measured by the ten trackers ranged from 3271 differences between the activity trackers and the gold standards were assessed by the (± 2136) [Nike+ Fuelband] to 4372 (± 2562) [Fitbit Flex]. As shown in Figure 1, the Nike+ paired samples t-test. In the event of non-normally distributed data, the Wilcoxon Signed Fuelband and Moves app provide a relatively large confidence interval for the mean number Rank test was used. Mean absolute percentage errors (c) compared to the gold standards of steps in the free-living condition, which is partly due to a lower number of measurements were calculated with the following formula: mean difference activity tracker-gold standard x of these devices. Therefore, additional power analyses were executed, which are shown 100 / mean gold standard. Second, in order to examine the correlation between the trackers below. and the gold standards, the ICC was calculated (absolute agreement, two-way random, single measures, 95% confidence interval). Third, to examine the level of agreement between the trackers and the gold standard, Bland-Altman plots were constructed with their associated limits of agreement. In addition, the ActivPAL scores from the laboratory study were compared with the corresponding Optogait scores by use of the three previously mentioned statistical tests, in order to assess the degree of consensus between the two gold standards used in this study.

Results

For the laboratory study, 33 participants were included (16 males, mean age (±SD) 39 (± 13.1) years, mean BMI (±SD) 23.6 (± 2.2) kg/m², and 17 females, mean age (±SD) 35 (± 11.2), mean BMI 22.5 (± 2.1) kg/m²). Thirty of the 33 participants performed the test again one week later. Most individuals who participated in the laboratory study also participated in the free-living study (N= 23) wherein a total of 56 participants were included (18 males, mean

22 Reliability and Validity of ten consumer activity trackers

Statistical analysis age (±SD) 37.1 (± 10.6), mean BMI (±SD) 24.1 (± 2) kg/m², and 38 females, mean age (±SD) 30 A sample size analysis was conducted to calculate the number of required participants. As (± 9.5) years, mean BMI (±SD) 23.1 (± 2.5) kg/m²). Most of the participants were university previous data on relevant differences for sample size calculation does not exist, we reasoned employees, with an office job. Activities performed by the participants during the test day that a difference of 10% for the laboratory condition and 15% for the free-living condition included sitting (e.g., at the computer), standing (e.g., teaching activities) and walking. A seemed appropriate. Using these relevant differences and expected mean number of steps number of participants were highly active (e.g., took a long walk during lunch time) whereas in both conditions, it was calculated that at least 24 participants were necessary for others were mainly sedentary during the test day. The Nike+ Fuelband and Moves app were 2 participation in the laboratory condition and 58 participants for the free-living condition to tested with a fewer number of participants in the free-living study (N=20 and N=11 enable substantiation of a relevant difference between the trackers and the gold standards respectively). The Nike+ Fuelband was not available at the beginning but was included during with a power of 80% and a significance level of 5 %. This number of participants is the study. The Moves app was unavailable at no cost for most participants in the free-living comparable to other validation studies.12,14,15 This reassured our reasoned choice for using study. In all 11 cases, the Moves app was operating on an Android device. 10% and 15% as cut-off points for the mean difference.

Results

23 Chapter 2

living condition - c) and free - e a

f). The middle line shows the mean difference between the tracker and the gold standard, and the dashed lines lines dashed the and standard, gold the and the tracker between difference the mean shows line middle The f). - e difference scores).

Figure 1.

Descriptive Statistics (mean number of steps, 95% CI) as measured by the gold standards and the ten activity

trackers in the laboratory and free-living condition. Pulse, figur Withings and Lumoback, Zip, Fitbit vs. (Optogait condition laboratory the in trackers activity top three the of plots Altman - (ActivPal vs. Fitbit Zip, Misfit Shine, and Lumoback, figure d indicate the limits of agreement * (±1.96 SD of th Figure 2. Bland

24 Reliability and Validity of ten consumer activity trackers

living condition - 2 c) and free - e a Pulse, figur

Figure 1.

Descriptive Statistics (mean number of steps, 95% CI) as measured by the gold standards and the ten activity

trackers in the laboratory and free-living condition. Withings and Lumoback, Zip, Fitbit vs. (Optogait condition laboratory the in trackers activity top three the of plots Altman - indicate the limits of agreement * (±1.96 SD of th (ActivPal vs. Fitbit Zip, Misfit Shine, and Lumoback, figure d Figure 2. Bland

25 Chapter 2

Agreement between the two gold standards Systematic differences and mean absolute percentage error The ActivPAL was compared with the Optogait in the laboratory condition using the same In the laboratory condition, there was a significant difference between the number of steps statistical tests that were used for the ten activity trackers. The ActivPAL demonstrated a measured by the Optogait (gold standard) and those measured by the Lumoback, Fitbit Flex, mean difference of 9 ± 6 steps [0.3%] with the Optogait (P<0.001, N=25). The effect size of Nike+ Fuelband, Withings Pulse, Fitbit Zip, Omron, and the Moves app (Table 2). However, this significant difference was calculated using Cohens effect size25 and indicated an effect the size of the mean difference was less than 34 steps (MAPE = 1%) or close to this MAPE for size of 0.02, which is negligibly small. The ICC between the ActivPAL and the Optogait is 1. most of the trackers. There was a more substantial MAPE between the Optogait and Fitbit The Bland-Altman plot revealed a difference between the lower and upper limit of Flex; (188 steps [5.7%]), the Moves app (319 steps [9.6%]), and the Nike+ Fuelband (598 agreement of 24 steps. These results indicate excellent agreement of the two gold standards steps [18%]). The Misfit Shine demonstrated the smallest MAPE compared with the Optogait used in this study. [i.e., 0.18%].

In the free-living condition, there was a significant difference in the number of steps Test-retest reliability between the ActivPAL (gold standard) and the Fitbit Flex, Nike+ Fuelband, Fitbit Zip, Withings The ICCs between the first test and the second test (one week later) in the laboratory Pulse, Digiwalker, and the Moves app (Table 2). Again, the MAPE values of the trackers were condition varied between 0.14 and 0.96 (Table 1). The gold standards used in this study small (less than 10%), except for the Nike+ Fuelband and the Moves app (24% and 37.6% (Optogait and ActivPAL), demonstrated excellent test-retest reliability. Test-retest reliability respectively). The smallest MAPE values were between the ActivPAL and the Omron (0.4%) of the Lumoback, Fitbit Zip, and Withings Pulse was excellent as well (i.e., ICC > .90). Test- and Lumoback (0.4%). The power for the calculation of the Nike+ Fuelband and Moves app retest reliability of the Jawbone Up, Fitbit Flex, and Misfit Shine was good (ICC .75 - .90); was 62% and 39%, respectively. The power for the remaining devices was high, i.e., greater test-retest reliability of the Digiwalker was moderate (ICC .60 - .75); and test-retest reliability than 99%. of the Nike+ Fuelband, Omron, and Moves app was low (ICC < 0.60).

Table 1. Intraclass correlation coefficients between Test 1 and Test 2 of the treadmill walking test (N=30). Activity tracker ICC 95% confidence Interval

Optogait 0.92** 0.85 - 0.96 ActivPAL 0.96** 0.90 - 0.99 Lumoback 0.90** 0.79 - 0.95 Fitbit Flex 0.81** 0.64 - 0.91 Jawbone UP 0.83** 0.66 - 0.91 Nike+ Fuelband 0.53** 0.22 - 0.75 Misfit Shine 0.86** 0.73 - 0.93 Withings Pulse 0.92** 0.83 - 0.96 Fitbit Zip 0.90** 0.80 - 0.95 Omron 0.14 -0.24 - 0.47 Digiwalker 0.71 ** 0.47 - 0.86 Moves app 0.37* 0.02 - 0.64 * P < 0.05 ** P < 0.01

26 Reliability and Validity of ten consumer activity trackers

Agreement between the two gold standards Systematic differences and mean absolute percentage error The ActivPAL was compared with the Optogait in the laboratory condition using the same In the laboratory condition, there was a significant difference between the number of steps statistical tests that were used for the ten activity trackers. The ActivPAL demonstrated a measured by the Optogait (gold standard) and those measured by the Lumoback, Fitbit Flex, mean difference of 9 ± 6 steps [0.3%] with the Optogait (P<0.001, N=25). The effect size of Nike+ Fuelband, Withings Pulse, Fitbit Zip, Omron, and the Moves app (Table 2). However, this significant difference was calculated using Cohens effect size25 and indicated an effect the size of the mean difference was less than 34 steps (MAPE = 1%) or close to this MAPE for size of 0.02, which is negligibly small. The ICC between the ActivPAL and the Optogait is 1. most of the trackers. There was a more substantial MAPE between the Optogait and Fitbit 2 The Bland-Altman plot revealed a difference between the lower and upper limit of Flex; (188 steps [5.7%]), the Moves app (319 steps [9.6%]), and the Nike+ Fuelband (598 agreement of 24 steps. These results indicate excellent agreement of the two gold standards steps [18%]). The Misfit Shine demonstrated the smallest MAPE compared with the Optogait used in this study. [i.e., 0.18%].

Table 1. Intraclass correlation coefficients between Test 1 and Test 2 of the treadmill walking test (N=30). Activity tracker ICC 95% confidence Interval

27 Chapter 2

Correlations

Value Table 3 illustrates the Intraclass Correlation Coefficients between the ten activity trackers - p 0.332 * 0.026 0.851 * 0.000 0.719 * 0.000 * 0.008 0.479 * 0.041 * 0.004 and the gold standard, for both the laboratory study and the free-living study. In the normality normality -

laboratory study, the ICCs ranged from -.13 (Moves) to .99 (Lumoback, Withings Pulse, and c Positive values

a Fitbit Zip). The ICCs in the free-living study ranged from 0.80 (Moves) to 1 (Fitbit Zip). value - 0.97 2.23 0.24 3.55 0.36 5.24 2.66 0.71 2.04 2.85 Z ------In case of non of In case

. Table 3.

b Intraclass Correlation Coefficients between the activity trackers and gold standards in the laboratory and free-

living study.

.7 MAPE 0.4 3 1.4 24 1.1 7.9 1.2 0.4 5.9 37.6 Laboratory 95 % Free-living 95 %

study (N=33) confidence study (N=56) confidence a (Optogait) interval (ActivPAL) interval

ActivPAL 1 0.94 - 1

150 58 43 49 Mean difference 17 - - 977 - 323 - 17 240 1529 Lumoback 0.99 ** 0.98 - 0.99 0.99 ** 0.98 - 0.99 Fitbit Flex 0.22 * -0.08 - 0.5 0.96 ** 0.94 - 0.98 living condition ² -

Jawbone UP 0.98 ** .095 - 0.99 0.94 ** 0.90 - 0.97

Free N 55 51 54 53 20 55 51 55 55 55 11

MAPE = mean absolute percentage error Nike+ Fuelband 0.12 -0.1 - 0.37 0.83 ** 0.37 - 0.94

b 0.97 ** 0.93 - 0.98 0.99 ** 0.98 - 0.99 Misfit Shine

* living condition.

0 * Withings Pulse 0.99 ** 0.95 - 0.97 0.96 ** 0.91 - 0.98 value value indicating a systematic difference of the activity tracker. - Fitbit Zip 0.99 ** 0.96 - 0.99 1 ** 0.99 - 1 - p * 0.000 * 0.033 * 0.000 0.119 * 0.000 0.430 0.001 * 0.000 * 0.006 0.153 0.00 Omron 0.59 ** 0.27 - 0.78 0.98 ** 0.96 - 0.99

c Digiwalker 0.65 ** 0.39 - 0.81 0.96 ** 0.93 - 0.98

Moves app -0.13 -0.32 - 0.15 0.80 ** 0.05 - 0.99 value value / - 4.36 0.80 2.96 4.36

t Z- 7.19 2.24 4.93 1.60 - - 3.70 5.44 - 1.46 - * P < 0.05 ** P < 0.01 test. -

Level of agreement 4070(±2430) significant* p

MAPE 0.3 0.2 5.7 1.0 18.0 0.2 0.5 0.3 2.5 1.2 9.6 = Bland-Altman plots indicate the differences between the tracker and the gold standard (y-

axis) against the average of the two methods (x-axis). Table 4 indicates the mean differences a

and negative values indicate an overestimation. overestimation. an indicate values negative and with the gold standard and the limits of agreement for all activity trackers. In the laboratory

condition, the plots showed the narrowest limits for the Fitbit Zip (46 steps), Lumoback (78

steps), and Withings Pulse (92 steps). The broadest limits were for the Nike+ Fuelband (2422 219) activity tracker) and MAPE in the laboratory and free and and laboratory the in MAPE tracker) activity 157)

123) 23) 43) – steps), Moves app (1436 steps), and Fitbit Flex (855 steps). In the free-living condition, the 6 ( 82 ( (366) 319 Mean difference (SD) 9 (6) 8 (20) ( 188 34 ( (618) 598 - 15 ( 11 (12) - 38 (145) - plots showed the narrowest limits for the Fitbit Zip (866 steps), Misfit Shine (1400 steps), and the Lumoback (1590 steps). The broadest limits of agreement were determined for the

Moves app (5150 steps), Nike+ Fuelband (4528 steps), and Jawbone Up (3350 steps). Figure 3314 (±162) ² Mean (±SD) ActivPAL

N 25 32 33 32 33 33 32 32 32 32 33 = 2 illustrates the Bland-Altman plots for the top three activity trackers (narrowest limits of Laboratory Condition ¹

agreement) for both the laboratory (Fitbit Zip, Lumoback, and Withings Pulse) and for the

free-living condition (Fitbit Zip, Misfit Shine and Lumoback).

indicate an underestimation of the activity tracker tracker activity of the underestimation an indicate

28 Reliability and Validity of ten consumer activity trackers

Correlations

laboratory study, the ICCs ranged from -.13 (Moves) to .99 (Lumoback, Withings Pulse, and c Positive values

a Fitbit Zip). The ICCs in the free-living study ranged from 0.80 (Moves) to 1 (Fitbit Zip). value

- 2 0.97 2.23 0.24 3.55 0.36 5.24 2.66 0.71 2.04 2.85 Z ------In case of non of In case c

. Table 3.

b Intraclass Correlation Coefficients between the activity trackers and gold standards in the laboratory and free-

living study.

.7 MAPE 0.4 3 1.4 24 1.1 7.9 1.2 0.4 5.9 37.6 Laboratory 95 % Free-living 95 %

study (N=33) confidence study (N=56) confidence a (Optogait) interval (ActivPAL) interval

ActivPAL 1 0.94 - 1

150 58 43 49 Mean difference 17 - - 977 - 323 - 17 240 1529 Lumoback 0.99 ** 0.98 - 0.99 0.99 ** 0.98 - 0.99 Fitbit Flex 0.22 * -0.08 - 0.5 0.96 ** 0.94 - 0.98 living condition ² -

Jawbone UP 0.98 ** .095 - 0.99 0.94 ** 0.90 - 0.97

Free N 55 51 54 53 20 55 51 55 55 55 11

MAPE = mean absolute percentage error Nike+ Fuelband 0.12 -0.1 - 0.37 0.83 ** 0.37 - 0.94 b 0.97 ** 0.93 - 0.98 0.99 ** 0.98 - 0.99 Misfit Shine

* living condition.

Moves app -0.13 -0.32 - 0.15 0.80 ** 0.05 - 0.99 value value / - 4.36 0.80 2.96 4.36

t Z- 7.19 2.24 4.93 1.60 - - 3.70 5.44 - 1.46 - * P < 0.05 ** P < 0.01 test. -

Level of agreement 4070(±2430) significant* p

MAPE 0.3 0.2 5.7 1.0 18.0 0.2 0.5 0.3 2.5 1.2 9.6 = Bland-Altman plots indicate the differences between the tracker and the gold standard (y-

axis) against the average of the two methods (x-axis). Table 4 indicates the mean differences a with the gold standard and the limits of agreement for all activity trackers. In the laboratory

condition, the plots showed the narrowest limits for the Fitbit Zip (46 steps), Lumoback (78

123) 23) 43) steps), Moves app (1436 steps), and Fitbit Flex (855 steps). In the free-living condition, the 6 ( 82 ( (366) 319 Mean difference (SD) 9 (6) 8 (20) ( 188 34 ( (618) 598 - 15 ( 11 (12) - 38 (145) - plots showed the narrowest limits for the Fitbit Zip (866 steps), Misfit Shine (1400 steps), and the Lumoback (1590 steps). The broadest limits of agreement were determined for the

Moves app (5150 steps), Nike+ Fuelband (4528 steps), and Jawbone Up (3350 steps). Figure 3314 (±162) ² Mean (±SD) ActivPAL

N 25 32 33 32 33 33 32 32 32 32 33 = 2 illustrates the Bland-Altman plots for the top three activity trackers (narrowest limits of Laboratory Condition ¹

agreement) for both the laboratory (Fitbit Zip, Lumoback, and Withings Pulse) and for the

free-living condition (Fitbit Zip, Misfit Shine and Lumoback).

ActivPAL Lumoback Flex Fitbit Jawbone UP Nike+ Fuelband Misfit Shine Withings Pulse Zip Fitbit Omron Digiwalker Moves app the Wilcoxon Signed Rank Test was used instead of the Paired Samples T Samples Paired the of instead used was Test Rank Signed Wilcoxon the Table 2. – standard (gold scores difference Mean ¹ Mean (±SD) Optogait (±SD) Mean ¹ indicate an underestimation of the activity tracker and negative values indicate an overestimation. overestimation. an indicate values negative and tracker activity of the underestimation an indicate

29 Chapter 2

Table 4. differences using comparable conditions. Melanson et al 26 found an accuracy of 97.8% of Mean difference scores with the gold standards and limits of agreement of the activity trackers in the the Digiwalker SW-200 during walking on the treadmill with speeds between 3.0 and 3.5 laboratory and free-living study. mph (4.8 – 5.6 km/h), which is in accordance with our finding of 1.2% error. In the study of Mean Limits of Mean Limits of De Cocker et al,27 the Omron differed on an average of 6.7% compared to the gold standard. difference Agreement difference Agreement (Optogait – (ActivPAL- The slightly smaller difference of 2.5% determined in our study could possibly be explained tracker, lab tracker, free- by the longer duration of the treadmill test in this study (30 minutes vs. 5 minutes) which a a study) living study) decreases the relative size of measurement error. Case et al 16 found an error of +6.2% for Lower Upper Lower Upper the Moves app installed on an IOS device and an error of -6.7% for the Moves app installed ActivPAL 9 -3 21 on an Android device. The MAPE found for the IOS device was a bit lower than the +9.6% Lumoback 8 -31 47 17 -778 812 difference in our study. An explanation could be the different version of the Iphone that was Fitbit Flex 188 -240 615 -150 -1424 1124 Jawbone UP 34 -54 81 -58 -1732 1618 utilized (Iphone 5S compared to the 4S in our study). For the Nike+ Fuelband, Case et al Nike+ Fuelband 598 -613 1809 977 -1288 3240 found a mean underestimation of 22.7%. This was in line with our finding of 18% Misfit Shine -6 -91 85 -43 -743 657 underestimation. Withings Pulse 15 -31 61 323 -864 1510 The second method to determine validity was to examine the ICCs between the Fitbit Zip 11 -12 34 -49 -482 384 trackers and the gold standard. In the laboratory study, all trackers demonstrated a good to Omron -82 -390 226 17 -1006 1040 Digiwalker 38 -248 323 240 -1028 1508 excellent agreement with the gold standard, with the exception of the Moves app, Nike+ Moves app -319 -1037 399 1529 -1046 4104 Fuelband, and Fitbit Flex. Two other studies also examined correlations between the activity a Positive values indicate an underestimation of the activity tracker and negative values indicate an trackers and the gold standard in laboratory conditions. For the Fitbit One, Tacacs et al 14 overestimation. ascertained concordance correlations between 0.97 and 1.0 for five different speeds on the treadmill with manual steps counting as the gold standard. This was in accordance with our finding for the Fitbit Zip (ICC .99). For the Digiwalker SW-200, Beets et al determined an ICC Discussion of .99 compared to real step count for children walking on a treadmill at the same speed (4.8 28 km/h). This is somewhat higher than the ICC found in our study (ICC .65). However, if we Ten popular consumer activity trackers were tested for their reliability and validity for removed the four outliers in our analyses our ICC increased to .94, which is more in line with measuring step count. Seven out of ten trackers were reliable (Lumoback, Fitbit Flex, the findings of Beets et al. Jawbone UP, Misfit Shine, Withings Pulse, Fitbit Zip, and Digiwalker), and five of these The third and last way to examine validity was to assess the level of agreement by trackers also demonstrated high validity in laboratory conditions (Lumoback, Jawbone Up, visualizing the data with Bland-Altman plots.29 The difference between the lower and upper Misfit Shine, Withings Pulse, and Fitbit Zip). The Moves app and Nike+ Fuelband exhibited limit of agreement (Mean difference ± 1.96SD of difference scores) ranged from 46 steps low reliability and a low validity in laboratory conditions. In free-living conditions, the Fitbit Zip showed the highest validity and the Nike+ Fuelband indicated a low validity. (Fitbit Zip) to 2422 steps (Nike+ Fuelband). The Lumoback, Jawbone Up, Misfit Shine, Withings Pulse, and Fitbit Zip indicated the narrowest limits of agreement (less than 300 The validity of the ten activity trackers in laboratory conditions was examined with steps) which equals less than 10% and less than 3 minutes walking. This can be considered as three methods of which the first was to assess systematic differences. According to Tudor- a relatively small range. Taken together with the small systematic differences of these 23 Locke et al, activity monitors should not exceed a 1% error deviation (MAPE) from the gold trackers (less than 1%), it is suggested that the Lumoback, Jawbone Up, Misfit Shine, standard during walking on a treadmill at a speed of 3 mph (4.8 km/h) in order to be Withings Pulse, and Fitbit Zip can be used interchangeably with the gold standard when considered accurate. In the controlled lab-condition, five trackers achieved this condition: walking on a treadmill. The systematic differences and the range between the upper and the Lumoback, Jawbone Up, Misfit Shine, Withings Pulse, and Fitbit Zip. The Digiwalker and lower limits of agreement of the Moves app (1436 steps) and the Nike+ Fuelband (2422 Omron had an error deviation slightly higher than the 1% threshold, e.g., 1.2% and 2.5%, steps) are considered to be too large to be used interchangeably with the gold standard. respectively, which still represents a very low MAPE. The Fitbit Flex (5.6%), Moves app (9.6%) and Nike+ Fuelband (18%) exhibited greater deviation errors whereby the Fitbit Flex To summarize, the lab results show that most trackers are valid with the Lumoback, Jawbone and Nike+ Fuelband underestimated the number of steps, and the Moves app overestimated Up, Misfit Shine, Withings Pulse, and Fitbit Zip demonstrating the highest validity. The the number of steps. Some trackers were examined in other studies as well for systematic Moves app and Nike+ Fuelband are clearly invalid. It should be noted that, in a controlled lab

30 Reliability and Validity of ten consumer activity trackers

Table 4. differences using comparable conditions. Melanson et al 26 found an accuracy of 97.8% of Mean difference scores with the gold standards and limits of agreement of the activity trackers in the the Digiwalker SW-200 during walking on the treadmill with speeds between 3.0 and 3.5 laboratory and free-living study. mph (4.8 – 5.6 km/h), which is in accordance with our finding of 1.2% error. In the study of Mean Limits of Mean Limits of De Cocker et al,27 the Omron differed on an average of 6.7% compared to the gold standard. difference Agreement difference Agreement (Optogait – (ActivPAL- The slightly smaller difference of 2.5% determined in our study could possibly be explained tracker, lab tracker, free- by the longer duration of the treadmill test in this study (30 minutes vs. 5 minutes) which 2 a a study) living study) decreases the relative size of measurement error. Case et al 16 found an error of +6.2% for Lower Upper Lower Upper the Moves app installed on an IOS device and an error of -6.7% for the Moves app installed ActivPAL 9 -3 21 on an Android device. The MAPE found for the IOS device was a bit lower than the +9.6% Lumoback 8 -31 47 17 -778 812 difference in our study. An explanation could be the different version of the Iphone that was Fitbit Flex 188 -240 615 -150 -1424 1124 Jawbone UP 34 -54 81 -58 -1732 1618 utilized (Iphone 5S compared to the 4S in our study). For the Nike+ Fuelband, Case et al Nike+ Fuelband 598 -613 1809 977 -1288 3240 found a mean underestimation of 22.7%. This was in line with our finding of 18% Misfit Shine -6 -91 85 -43 -743 657 underestimation. Withings Pulse 15 -31 61 323 -864 1510 The second method to determine validity was to examine the ICCs between the Fitbit Zip 11 -12 34 -49 -482 384 trackers and the gold standard. In the laboratory study, all trackers demonstrated a good to Omron -82 -390 226 17 -1006 1040 Digiwalker 38 -248 323 240 -1028 1508 excellent agreement with the gold standard, with the exception of the Moves app, Nike+ Moves app -319 -1037 399 1529 -1046 4104 Fuelband, and Fitbit Flex. Two other studies also examined correlations between the activity a Positive values indicate an underestimation of the activity tracker and negative values indicate an trackers and the gold standard in laboratory conditions. For the Fitbit One, Tacacs et al 14 overestimation. ascertained concordance correlations between 0.97 and 1.0 for five different speeds on the treadmill with manual steps counting as the gold standard. This was in accordance with our finding for the Fitbit Zip (ICC .99). For the Digiwalker SW-200, Beets et al determined an ICC Discussion of .99 compared to real step count for children walking on a treadmill at the same speed (4.8 28 km/h). This is somewhat higher than the ICC found in our study (ICC .65). However, if we Ten popular consumer activity trackers were tested for their reliability and validity for removed the four outliers in our analyses our ICC increased to .94, which is more in line with measuring step count. Seven out of ten trackers were reliable (Lumoback, Fitbit Flex, the findings of Beets et al. Jawbone UP, Misfit Shine, Withings Pulse, Fitbit Zip, and Digiwalker), and five of these The third and last way to examine validity was to assess the level of agreement by trackers also demonstrated high validity in laboratory conditions (Lumoback, Jawbone Up, visualizing the data with Bland-Altman plots.29 The difference between the lower and upper Misfit Shine, Withings Pulse, and Fitbit Zip). The Moves app and Nike+ Fuelband exhibited limit of agreement (Mean difference ± 1.96SD of difference scores) ranged from 46 steps low reliability and a low validity in laboratory conditions. In free-living conditions, the Fitbit Zip showed the highest validity and the Nike+ Fuelband indicated a low validity. (Fitbit Zip) to 2422 steps (Nike+ Fuelband). The Lumoback, Jawbone Up, Misfit Shine, Withings Pulse, and Fitbit Zip indicated the narrowest limits of agreement (less than 300 The validity of the ten activity trackers in laboratory conditions was examined with steps) which equals less than 10% and less than 3 minutes walking. This can be considered as three methods of which the first was to assess systematic differences. According to Tudor- a relatively small range. Taken together with the small systematic differences of these 23 Locke et al, activity monitors should not exceed a 1% error deviation (MAPE) from the gold trackers (less than 1%), it is suggested that the Lumoback, Jawbone Up, Misfit Shine, standard during walking on a treadmill at a speed of 3 mph (4.8 km/h) in order to be Withings Pulse, and Fitbit Zip can be used interchangeably with the gold standard when considered accurate. In the controlled lab-condition, five trackers achieved this condition: walking on a treadmill. The systematic differences and the range between the upper and the Lumoback, Jawbone Up, Misfit Shine, Withings Pulse, and Fitbit Zip. The Digiwalker and lower limits of agreement of the Moves app (1436 steps) and the Nike+ Fuelband (2422 Omron had an error deviation slightly higher than the 1% threshold, e.g., 1.2% and 2.5%, steps) are considered to be too large to be used interchangeably with the gold standard. respectively, which still represents a very low MAPE. The Fitbit Flex (5.6%), Moves app (9.6%) and Nike+ Fuelband (18%) exhibited greater deviation errors whereby the Fitbit Flex To summarize, the lab results show that most trackers are valid with the Lumoback, Jawbone and Nike+ Fuelband underestimated the number of steps, and the Moves app overestimated Up, Misfit Shine, Withings Pulse, and Fitbit Zip demonstrating the highest validity. The the number of steps. Some trackers were examined in other studies as well for systematic Moves app and Nike+ Fuelband are clearly invalid. It should be noted that, in a controlled lab

31 Chapter 2

condition, there is no variation in walking speed, intensity, direction, etc. which is in contrast Ferguson et al reported similar correlations for the Jawbone UP, Nike+ Fuelband, Misfit to real life. Therefore, validity was also tested in free-living conditions. Shine, Withings Pulse, and Fitbit Zip in their free-living study of 48 hours.17

The first way to validate activity trackers in free-living conditions was to assess Finally, the level of agreement of the activity trackers with the gold standard during systematic differences. In free-living conditions, an acceptable mean deviation from free-living conditions was assessed by Bland-Altman plots. The difference between the lower the gold standard is 10%.23 Eight activity trackers achieved this criterion. The Nike+ and upper limit of agreement ranged from 861 steps (Fitbit Zip) to 5150 steps (Moves app). Fuelband and Moves app showed larger percentages of underestimation: 24.0% and For the Fitbit Zip, the range of 861steps (less than 1000 steps, e.g., 10 minutes walking) 37.6%, respectively. Lee et al 12 investigated various consumer trackers during appears to be sufficiently low enough to be a valid measure in scientific research. The Misfit different semi-structured activities (the participants followed a 69-minute protocol) Shine and Lumoback demonstrated slightly larger limits of agreement (1400 and 1590 steps, and compared total energy expenditure with the gold standard (breath-to-breath respectively) which still demonstrates a good validity. For the other trackers, the limits of analysis). The Fitbit Zip, Jawbone Up, and Nike+ Fuelband differed 10.1%, 12.2%, and agreement show that, despite the relatively small systematic error (below 400 steps [10%] 13.0%, respectively, from the gold standard. The differences are greater for the Fitbit for eight of the ten trackers), larger individual differences are evident, resulting in a lower Zip and Jawbone Up compared to the results of our study which could possibly be validity. explained by the different outcome measure that was utilized in the study of Lee et al To summarize, the validity of eight of the ten trackers was good during free-living (energy expenditure vs. step count). The difference between the Nike+ Fuelband and conditions whereby the Fitbit Zip showed the best validity. The validity of the Nike+ the gold standard is smaller compared to the present study (24%). However, Lee et al Fuelband is low for measuring steps in free-living conditions. has already mentioned inconsistent results for the Nike+ Fuelband (a relatively small MAPE but also a low correlation with the gold standard) and, therefore, advised Our study has some limitations. First, in the laboratory condition, only one type of interpreting these results with caution. Ferguson et al [17] investigated five similar activity was examined (walking), however, activity trackers can possibly perform differently devices (Jawbone UP, Nike+ Fuelband, Misfit Shine, Withings Pulse and Fitbit Zip) in during different activities or velocities (such as walking slow). The advantage of the 30- free-living conditions for 48 hours. They ascertained differences of 8.1%, 25.6%, minute measurement was that reliable data for average walking speed was obtained. 10.1%, 6.3% and 4.3%, respectively. These values are in line with our findings in Second, for examining free-living activity, we used a time span of 9:00-16:30 in which which the somewhat larger differences can be explained by the longer period of ‘occupational activity’ was mostly measured. The advantage of this method was that we measurement. De Cocker et al,27 investigated the Omron during free-living conditions were able to make a realistic comparison between the different trackers with different and used the Digiwalker as a criterion measure. They reported a more substantial wearing positions because cycling was excluded. Cycling could have biased the results difference between the two devices compared to the findings of the present study between centrally worn and wrist-worn trackers. However, the trackers might perform (36.9% vs. 0.4%) which can be a result of non-walking activities, a longer period of differently during a greater variety of activities such as more intensive exercise. These measurement, and the different gold standard. activities were not measured in this study. The third limitation was, that in the free-living condition, the Nike+ Fuelband and Moves app were tested with fewer number of The second way to determine the validity of the activity trackers during free- participants. Because of a reasonable power (62%), consistent results with the laboratory living conditions was to calculate ICCs. All activity trackers were highly correlated to condition, and consistent results with other studies,12,16,17 the results of the Nike+ Fuelband the gold standard (ActivPAL). The Nike+ Fuelband and the Moves app showed ICCs are considered reliable. For the Moves app, only preliminary conclusions can be drawn on which were a bit lower and had broad confidence intervals (.83 [CI .37; .94] and .80 the validity in free-living conditions. This is due to the low N, consequently a lower power of [CI .05 - .99] respectively). The high ICCs in the free-living study can be partially 39%, and because the Moves app was tested on different types of phones compared to the attributed to the differences in activity patterns between the participants during the laboratory study (Android vs. IOS devices). Therefore, the results of the free-living condition test day; more variation increases the chances of a high ICC. Lee et al indicated cannot be compared with the lab condition because the different types of firmware may similar results for the Fitbit Zip, Jawbone Up, and the Nike+ Fuelband, i.e., high have influenced the results. However, our results for the Moves app on the different types correlations for the Fitbit Zip and Jawbone Up and a lower correlation for the Nike+ of phones are comparable with the study of Case et al16 who showed that Android devices Fuelband.12 Tully et al, investigated the validity of the Fitbit Zip in free-living are associated with a modest underestimation, and IOS devices show a modest conditions; the Fitbit Zip was worn for seven days along with the Actigraph overestimation of step counting, which is in line with our results. accelerometer. They reported a high correlation (Spearman Rho= .91) between steps/day when measured by the Fitbit Zip and by the Actigraph.15 In addition,

32 Reliability and Validity of ten consumer activity trackers

The first way to validate activity trackers in free-living conditions was to assess Finally, the level of agreement of the activity trackers with the gold standard during systematic differences. In free-living conditions, an acceptable mean deviation from free-living conditions was assessed by Bland-Altman plots. The difference between the lower the gold standard is 10%.23 Eight activity trackers achieved this criterion. The Nike+ and upper limit of agreement ranged from 861 steps (Fitbit Zip) to 5150 steps (Moves app). 2 Fuelband and Moves app showed larger percentages of underestimation: 24.0% and For the Fitbit Zip, the range of 861steps (less than 1000 steps, e.g., 10 minutes walking) 37.6%, respectively. Lee et al 12 investigated various consumer trackers during appears to be sufficiently low enough to be a valid measure in scientific research. The Misfit different semi-structured activities (the participants followed a 69-minute protocol) Shine and Lumoback demonstrated slightly larger limits of agreement (1400 and 1590 steps, and compared total energy expenditure with the gold standard (breath-to-breath respectively) which still demonstrates a good validity. For the other trackers, the limits of analysis). The Fitbit Zip, Jawbone Up, and Nike+ Fuelband differed 10.1%, 12.2%, and agreement show that, despite the relatively small systematic error (below 400 steps [10%] 13.0%, respectively, from the gold standard. The differences are greater for the Fitbit for eight of the ten trackers), larger individual differences are evident, resulting in a lower Zip and Jawbone Up compared to the results of our study which could possibly be validity. explained by the different outcome measure that was utilized in the study of Lee et al To summarize, the validity of eight of the ten trackers was good during free-living (energy expenditure vs. step count). The difference between the Nike+ Fuelband and conditions whereby the Fitbit Zip showed the best validity. The validity of the Nike+ the gold standard is smaller compared to the present study (24%). However, Lee et al Fuelband is low for measuring steps in free-living conditions. has already mentioned inconsistent results for the Nike+ Fuelband (a relatively small MAPE but also a low correlation with the gold standard) and, therefore, advised Our study has some limitations. First, in the laboratory condition, only one type of interpreting these results with caution. Ferguson et al [17] investigated five similar activity was examined (walking), however, activity trackers can possibly perform differently devices (Jawbone UP, Nike+ Fuelband, Misfit Shine, Withings Pulse and Fitbit Zip) in during different activities or velocities (such as walking slow). The advantage of the 30- free-living conditions for 48 hours. They ascertained differences of 8.1%, 25.6%, minute measurement was that reliable data for average walking speed was obtained. 10.1%, 6.3% and 4.3%, respectively. These values are in line with our findings in Second, for examining free-living activity, we used a time span of 9:00-16:30 in which which the somewhat larger differences can be explained by the longer period of ‘occupational activity’ was mostly measured. The advantage of this method was that we measurement. De Cocker et al,27 investigated the Omron during free-living conditions were able to make a realistic comparison between the different trackers with different and used the Digiwalker as a criterion measure. They reported a more substantial wearing positions because cycling was excluded. Cycling could have biased the results difference between the two devices compared to the findings of the present study between centrally worn and wrist-worn trackers. However, the trackers might perform (36.9% vs. 0.4%) which can be a result of non-walking activities, a longer period of differently during a greater variety of activities such as more intensive exercise. These measurement, and the different gold standard. activities were not measured in this study. The third limitation was, that in the free-living condition, the Nike+ Fuelband and Moves app were tested with fewer number of The second way to determine the validity of the activity trackers during free- participants. Because of a reasonable power (62%), consistent results with the laboratory living conditions was to calculate ICCs. All activity trackers were highly correlated to condition, and consistent results with other studies,12,16,17 the results of the Nike+ Fuelband the gold standard (ActivPAL). The Nike+ Fuelband and the Moves app showed ICCs are considered reliable. For the Moves app, only preliminary conclusions can be drawn on which were a bit lower and had broad confidence intervals (.83 [CI .37; .94] and .80 the validity in free-living conditions. This is due to the low N, consequently a lower power of [CI .05 - .99] respectively). The high ICCs in the free-living study can be partially 39%, and because the Moves app was tested on different types of phones compared to the attributed to the differences in activity patterns between the participants during the laboratory study (Android vs. IOS devices). Therefore, the results of the free-living condition test day; more variation increases the chances of a high ICC. Lee et al indicated cannot be compared with the lab condition because the different types of firmware may similar results for the Fitbit Zip, Jawbone Up, and the Nike+ Fuelband, i.e., high have influenced the results. However, our results for the Moves app on the different types correlations for the Fitbit Zip and Jawbone Up and a lower correlation for the Nike+ of phones are comparable with the study of Case et al16 who showed that Android devices Fuelband.12 Tully et al, investigated the validity of the Fitbit Zip in free-living are associated with a modest underestimation, and IOS devices show a modest conditions; the Fitbit Zip was worn for seven days along with the Actigraph overestimation of step counting, which is in line with our results. accelerometer. They reported a high correlation (Spearman Rho= .91) between steps/day when measured by the Fitbit Zip and by the Actigraph.15 In addition,

33 Chapter 2

By combining the results of both conditions, it can be concluded that the validity of References most activity trackers is good (Fitbit Zip, followed by Misfit Shine and Lumoback) or acceptable (Fitbit Flex, Jawbone Up, Withings Pulse, Omron, and Digiwalker). Looking at the wearing position of the trackers (wrist-worn for the Fitbit Flex, Jawbone UP, and 1. Lee IM, Shiroma EJ, Lobelo F, Puska P, Blair SN, Katzmarzyk PT, Lancet Physical Activity Series Working Nike+Fuelband and centrally worn, e.g. close to the pelvis or trunk, for the remaining Group: Effect of physical inactivity on major non-communicable diseases worldwide: an analysis of burden of disease and life expectancy. Lancet 2012, 380(9838):219-229. devices), our results indicate that activity trackers worn close to the body exhibit a better 2. Warburton DE, Nicol CW, Bredin SS: Health benefits of physical activity: the evidence. CMAJ 2006, validity than the wrist-worn activity trackers, especially during free-living conditions. For 174(6):801-809. wrist-worn activity trackers, more measurement error can occur due to more variation in the 3. Haskell WL, Lee IM, Pate RR, Blair SN, Franklin BA, Macera CA, Heath GW., Thompson., PD., Bauman A.: Physical activity and public health: updated recommendation for adults from the American College way the arms are used in free-living conditions. This finding is supported by the research of of Sports Medicine and the American Heart Association. Med Sci Sports Exerc 2007, 39(8):1423-1434. Atallah et al.30 4. Tudor-Locke C, Craig CL, Brown WJ, Clemes SA, De Cocker K, Giles-Corti B, Hatano, Y, Inoue, S, Matsudo, SM, Mutrie, N, Oppert J, Rowe DA, Schmidt MD, Schofield GM, Spence JC, Teixeira PJ, Tully For the choice of a device, different considerations can be taken into account. MA, Blair SN.: How many steps/day are enough? For adults. Int J Behav Nutr Phys Act 2011, 8:79- 5868-8-79. First, the goal of physical activity measurement should be considered. For individual 5. Godino JG, Watkinson C, Corder K, Sutton S, Griffin SJ, Van Sluijs EM: Awareness of physical activity in users, it is most important that the change in physical activity is clearly displayed, healthy middle-aged adults: a cross-sectional study of associations with sociodemographic, biological, therefore, devices should be reliable. For large-scale research, the validity of a tracker behavioural, and psychological factors. BMC Public Health 2014, 14(1):421. 6. Vooijs M, Alpay LL, Snoeck-Stroband JB, Beerthuizen T, Siemonsma PC, Abbink, J.J.Sont JK, Rövekamp is important in order to be able to compare physical activity levels of different TA.: Validity and usability of low-cost accelerometers for internet-based self-monitoring of physical groups. In addition, the type of activity that will be measured should be considered activity in patients with chronic obstructive pulmonary disease. Interact J Med Res 2014, 3(4):e14. 7. Bravata DM, Smith-Spangler C, Sundaram V, Gienger AL, Lin N, Lewis R, Stave CD., Olkin I., Sirard JR.: so a choice for the wearing position can be made. For example, wrist-worn activity Using pedometers to increase physical activity and improve health: a systematic review. JAMA 2007, trackers are better able to measure higher limb activity, and ankle worn trackers are 298(19):2296-304. . better able to measure lower limb activity (e.g. cycling).31 Furthermore, a consumer 8. El-Gayar O, Timsina P, Nawar N, Eid W: A systematic review of IT for diabetes self-management: are we there yet? Int J Med Inform 2013, 82(8):637-652. can choose between a more advanced -and mostly more expensive device-, or a 9. Adam Noah J, Spierer DK, Gu J, Bronner S: Comparison of steps and energy expenditure assessment in more simple and affordable device. This study demonstrated that less expensive adults of Fitbit Tracker and Ultra to the Actical and indirect calorimetry. J Med Eng Technol 2013, devices are not necessarily less valid. 37(7):456-462. 10. Dannecker KL, Sazonova NA, Melanson EL, Sazonov ES, Browning RC: A comparison of energy expenditure estimation of several physical activity monitors. Med Sci Sports Exerc 2013, 45(11):2105- 2112. 11. Fortune E, Lugade V, Morrow M, Kaufman K: Validity of using tri-axial accelerometers to measure Conclusions human movement - Part II: Step counts at a wide range of gait velocities. Med Eng Phys 2014, 36(6):659-669. 12. Lee JM, Kim Y, Welk GJ: Validity of consumer-based physical activity monitors. Med Sci Sports Exerc In conclusion, the reliability of the Lumoback, Fitbit Flex, Jawbone UP, Misfit Shine, Withings 2014, 46(9):1840-1848. Pulse, Fitbit Zip, and Digiwalker is good. These trackers are suitable for consumer usage and 13. Stahl ST, Insana SP: Caloric expenditure assessment among older adults: criterion validity of a novel accelerometry device. J Health Psychol 2014, 19(11):1382-1387. health enhancing programs. Of all ten trackers the Fitbit Zip shows the highest validity 14. Takacs J, Pollock CL, Guenther JR, Bahar M, Napier C, Hunt MA: Validation of the Fitbit One activity whereas the Nike+ Fuelband shows the lowest validity. The results of this study can assist monitor device during treadmill walking. J Sci Med Sport 2014, 17(5):496-500. consumers, researchers, and health care providers to make an evidence-based choice for an 15. Tully MA, McBride C, Heron L, Allen W, Hunter RF: The validation of Fibit ZipTM physical activity monitor as a measure of free-living physical activity. BMC research notes 2014, 7(1):952. activity tracker to measure step count. 16. Case MA, Burwick HA, Volpp KG, Patel MS: Accuracy of smartphone applications and wearable devices for tracking physical activity data. JAMA 2015, 313(6):625-626. 17. Ferguson T, Rowlands AV, Olds T, Maher C: The validity of consumer-level, activity monitors in healthy adults worn in free-living conditions: a cross-sectional study. Int J Behav Nutr Phys Act 2015, 12:42-015-0201-9. 18. Dontje ML, de Groot M, Lengton RR, van der Schans, Cees P, Krijnen WP: Measuring steps with the Fitbit activity tracker: an inter-device reliability study. J Med Eng Technol 2015, 39(5):286-290. 19. Lee M, Song C, Lee K, Shin D, Shin S: Agreement between the spatio-temporal gait parameters from treadmill-based photoelectric cell and the instrumented treadmill system in healthy young adults and stroke patients. Med Sci Monit 2014, 20:1210-1219. 20. Dahlgren G, Carlsson D, Moorhead A, Hager-Ross C, McDonough SM: Test-retest reliability of step counts with the ActivPAL device in common daily activities. Gait Posture 2010, 32(3):386-390.

34 Reliability and Validity of ten consumer activity trackers

By combining the results of both conditions, it can be concluded that the validity of References most activity trackers is good (Fitbit Zip, followed by Misfit Shine and Lumoback) or acceptable (Fitbit Flex, Jawbone Up, Withings Pulse, Omron, and Digiwalker). Looking at the wearing position of the trackers (wrist-worn for the Fitbit Flex, Jawbone UP, and 1. Lee IM, Shiroma EJ, Lobelo F, Puska P, Blair SN, Katzmarzyk PT, Lancet Physical Activity Series Working Nike+Fuelband and centrally worn, e.g. close to the pelvis or trunk, for the remaining Group: Effect of physical inactivity on major non-communicable diseases worldwide: an analysis of burden of disease and life expectancy. Lancet 2012, 380(9838):219-229. 2 devices), our results indicate that activity trackers worn close to the body exhibit a better 2. Warburton DE, Nicol CW, Bredin SS: Health benefits of physical activity: the evidence. CMAJ 2006, validity than the wrist-worn activity trackers, especially during free-living conditions. For 174(6):801-809. wrist-worn activity trackers, more measurement error can occur due to more variation in the 3. Haskell WL, Lee IM, Pate RR, Blair SN, Franklin BA, Macera CA, Heath GW., Thompson., PD., Bauman A.: Physical activity and public health: updated recommendation for adults from the American College way the arms are used in free-living conditions. This finding is supported by the research of of Sports Medicine and the American Heart Association. Med Sci Sports Exerc 2007, 39(8):1423-1434. Atallah et al.30 4. Tudor-Locke C, Craig CL, Brown WJ, Clemes SA, De Cocker K, Giles-Corti B, Hatano, Y, Inoue, S, Matsudo, SM, Mutrie, N, Oppert J, Rowe DA, Schmidt MD, Schofield GM, Spence JC, Teixeira PJ, Tully For the choice of a device, different considerations can be taken into account. MA, Blair SN.: How many steps/day are enough? For adults. Int J Behav Nutr Phys Act 2011, 8:79- 5868-8-79. First, the goal of physical activity measurement should be considered. For individual 5. Godino JG, Watkinson C, Corder K, Sutton S, Griffin SJ, Van Sluijs EM: Awareness of physical activity in users, it is most important that the change in physical activity is clearly displayed, healthy middle-aged adults: a cross-sectional study of associations with sociodemographic, biological, therefore, devices should be reliable. For large-scale research, the validity of a tracker behavioural, and psychological factors. BMC Public Health 2014, 14(1):421. 6. Vooijs M, Alpay LL, Snoeck-Stroband JB, Beerthuizen T, Siemonsma PC, Abbink, J.J.Sont JK, Rövekamp is important in order to be able to compare physical activity levels of different TA.: Validity and usability of low-cost accelerometers for internet-based self-monitoring of physical groups. In addition, the type of activity that will be measured should be considered activity in patients with chronic obstructive pulmonary disease. Interact J Med Res 2014, 3(4):e14. 7. Bravata DM, Smith-Spangler C, Sundaram V, Gienger AL, Lin N, Lewis R, Stave CD., Olkin I., Sirard JR.: so a choice for the wearing position can be made. For example, wrist-worn activity Using pedometers to increase physical activity and improve health: a systematic review. JAMA 2007, trackers are better able to measure higher limb activity, and ankle worn trackers are 298(19):2296-304. . better able to measure lower limb activity (e.g. cycling).31 Furthermore, a consumer 8. El-Gayar O, Timsina P, Nawar N, Eid W: A systematic review of IT for diabetes self-management: are we there yet? Int J Med Inform 2013, 82(8):637-652. can choose between a more advanced -and mostly more expensive device-, or a 9. Adam Noah J, Spierer DK, Gu J, Bronner S: Comparison of steps and energy expenditure assessment in more simple and affordable device. This study demonstrated that less expensive adults of Fitbit Tracker and Ultra to the Actical and indirect calorimetry. J Med Eng Technol 2013, devices are not necessarily less valid. 37(7):456-462. 10. Dannecker KL, Sazonova NA, Melanson EL, Sazonov ES, Browning RC: A comparison of energy expenditure estimation of several physical activity monitors. Med Sci Sports Exerc 2013, 45(11):2105- 2112. 11. Fortune E, Lugade V, Morrow M, Kaufman K: Validity of using tri-axial accelerometers to measure Conclusions human movement - Part II: Step counts at a wide range of gait velocities. Med Eng Phys 2014, 36(6):659-669. 12. Lee JM, Kim Y, Welk GJ: Validity of consumer-based physical activity monitors. Med Sci Sports Exerc In conclusion, the reliability of the Lumoback, Fitbit Flex, Jawbone UP, Misfit Shine, Withings 2014, 46(9):1840-1848. Pulse, Fitbit Zip, and Digiwalker is good. These trackers are suitable for consumer usage and 13. Stahl ST, Insana SP: Caloric expenditure assessment among older adults: criterion validity of a novel accelerometry device. J Health Psychol 2014, 19(11):1382-1387. health enhancing programs. Of all ten trackers the Fitbit Zip shows the highest validity 14. Takacs J, Pollock CL, Guenther JR, Bahar M, Napier C, Hunt MA: Validation of the Fitbit One activity whereas the Nike+ Fuelband shows the lowest validity. The results of this study can assist monitor device during treadmill walking. J Sci Med Sport 2014, 17(5):496-500. consumers, researchers, and health care providers to make an evidence-based choice for an 15. Tully MA, McBride C, Heron L, Allen W, Hunter RF: The validation of Fibit ZipTM physical activity monitor as a measure of free-living physical activity. BMC research notes 2014, 7(1):952. activity tracker to measure step count. 16. Case MA, Burwick HA, Volpp KG, Patel MS: Accuracy of smartphone applications and wearable devices for tracking physical activity data. JAMA 2015, 313(6):625-626. 17. Ferguson T, Rowlands AV, Olds T, Maher C: The validity of consumer-level, activity monitors in healthy adults worn in free-living conditions: a cross-sectional study. Int J Behav Nutr Phys Act 2015, 12:42-015-0201-9. 18. Dontje ML, de Groot M, Lengton RR, van der Schans, Cees P, Krijnen WP: Measuring steps with the Fitbit activity tracker: an inter-device reliability study. J Med Eng Technol 2015, 39(5):286-290. 19. Lee M, Song C, Lee K, Shin D, Shin S: Agreement between the spatio-temporal gait parameters from treadmill-based photoelectric cell and the instrumented treadmill system in healthy young adults and stroke patients. Med Sci Monit 2014, 20:1210-1219. 20. Dahlgren G, Carlsson D, Moorhead A, Hager-Ross C, McDonough SM: Test-retest reliability of step counts with the ActivPAL device in common daily activities. Gait Posture 2010, 32(3):386-390.

35 Chapter 2

21. Dowd KP, Harrington DM, Donnelly AE: Criterion and concurrent validity of the activPAL professional physical activity monitor in adolescent females. PLoS One 2012, 7(10):e47633. 22. Ryan CG, Grant PM, Tigbe WW, Granat MH: The validity and reliability of a novel activity monitor as a measure of walking. Br J Sports Med 2006, 40(9):779-784. 23. Tudor-Locke C, Sisson SB, Lee SM, Craig CL, Plotnikoff RC, Bauman A: Evaluation of quality of commercial pedometers. CAN J PUBLIC HEALTH 2006, S10-S15. 24. Portney L, Watkins M: Foundations of clinical research: applications to practice. Upper Saddle River, N.J, Pearson/Prentice Hall 2009. 25. Cohen J: A power primer. Psychol Bull 1992, 112(1):155. 26. Melanson EL, Knoll JR, Bell ML, Hill JO, Nysse LJ, Lanningham-Foster, L. Peters JC, Levine, JA.: Commercially available pedometers: considerations for accurate step counting. Prev Med 2004, 39(2):361-368. 27. De Cocker KA, De Meyer J, De Bourdeaudhuij IM, Cardon GM: Non-traditional wearing positions of pedometers: Validity and reliability of the Omron HJ-203-ED pedometer under controlled and free- living conditions. J Sci Med Sport 2012, 15(5):418-424. 28. Beets MW, Patton MM, Edwards S: The accuracy of pedometer steps and time during walking in children. Med Sci Sports Exerc 2005, 37(3):513-520. 29. Martin Bland J, Altman D: Statistical methods for assessing agreement between two methods of clinical measurement. The lancet 1986, 327(8476):307-310. 30. Atallah L, Lo B, King R, Yang G: Sensor positioning for activity recognition using wearable accelerometers. Biomedical Circuits and Systems, IEEE Transactions 2011, 5(4):320-329. 31. Mannini A, Intille SS, Rosenberger M, Sabatini AM, Haskell W: Activity recognition using a single accelerometer placed at the wrist or ankle. Med Sci Sports Exerc 2013, 45(11):2193-2203.

Chapter 3 | Reliability and validity of ten consumer activity trackers depend on walking speed

Tryntsje Fokkema Thea J.M. Kooiman Wim P. Krijnen Cees P. van der Schans Martijn de Groot

Medicine and Science in Sports and Exercise (2017) 49(4):793-800

Chapter 3 | Reliability and validity of ten consumer activity trackers depend on walking speed

Tryntsje Fokkema Thea J.M. Kooiman Wim P. Krijnen Cees P. van der Schans Martijn de Groot

Medicine and Science in Sports and Exercise (2017) 49(4):793-800

Chapter 3

Abstract Introduction

Purpose Consumer activity trackers are an inexpensive and feasible method for estimating daily To examine the test-retest reliability and validity of ten activity trackers for step counting at physical activity. As the availability of these devices has increased, so has their use in daily three different walking speeds. life, health care, and medical science. Two commonly used physical activity guidelines are the 30-minutes of moderate to vigorous activity (MVPA) per day for at least five days a 1 2 Methods week. and the 10.000 steps/day norm. Research to a healthy amount of physical activity per day shows that engagement in at least 8000 to 11000 steps a day is related to many Thirty-one healthy participants walked twice on a treadmill for 30 minutes while wearing ten health benefits, like a better physical fitness, body composition, and glycemic control.2,3 activity trackers (Polar Loop, Garmin Vivosmart, Fitbit Charge HR, Apple Watch Sport, Pebble When 3000 steps are taken at moderate to vigorous intensity, both guidelines correspond Smartwatch, Samsung Gear S, Misfit Flash, Jawbone Up Move, Flyfit and Moves). with each other.4 For physically inactive people (e.g. people who take on average less than Participants walked three walking speeds for ten minutes each; slow (3.2 km·h-1), average 5000 steps/day), an increment of 2000 steps per day already relates to health improvements (4.8 km·h-1), and vigorous (6.4 km·h-1). To measure test-retest reliability, intraclass like a better body composition and decrement of BMI.5 Therefore, activity trackers have a correlations (ICCs) were determined between the first and second treadmill test. Validity large value in objectifying ones physical activity pattern and demonstrating changes in one’s was determined by comparing the trackers with the gold standard (hand counting), using activity behavior. Activity trackers should therefore be reliable and valid. mean differences, mean absolute percentage errors, and ICCs. Statistical differences were calculated by paired-sample t-tests, Wilcoxon signed-rank tests, and by constructing Bland- Many trackers demonstrate acceptable validity and reliability of step counting, Altman plots. however, other activity trackers perform relatively inadequately.6,7 The accuracy of activity trackers that were recently released into the market is currently unknown. A common Results challenge of activity trackers is their validity for tracking activities at different walking speeds 8,9 Test-retest reliability varied with ICCs ranging from -0.02 to 0.97. Validity varied between including a slower walking speed. The latter could be an issue when self-tracking is used trackers and different walking speeds with mean differences between the gold standard and for the assessment of daily physical activity of patients with limited physical abilities or the 10,11 activity trackers ranging from 0.0 to 26.4%. Most trackers showed relatively low ICCs and elderly population. Validation of activity trackers at different speeds is thus important. broad limits of agreement of the Bland-Altman plots at the different speeds. For the slow This certainly accounts for wearables that have recently entered the market. To achieve this, walking speed, the Garmin Vivosmart and Fitbit Charge HR showed the most accurate the aim of this study is to examine the test-retest reliability and validity of ten relatively new results. The Garmin Vivosmart and Apple Watch Sport demonstrated the best accuracy at an activity trackers when walking at three different speeds. average walking speed. For vigorous walking, the Apple Watch Sport, Pebble Smartwatch, and Samsung Gear S exhibited the most accurate results. Methods

Conclusion Test-retest reliability and validity of activity trackers depends on walking speed. In general, Research design consumer activity trackers perform better at an average and vigorous walking speed than at A prospective study was conducted in a laboratory setting. Healthy adult volunteers were a slower walking speed. invited to walk two times for 30 minutes on a treadmill on different days (with approximately one week between the first and the second measurement). Each participant wore ten activity trackers. During the measurement phase, participants walked for half an hour at three different speeds (ten minutes each). First, they walked at a slow walking speed (3.2 km·h-1), next at a speed that is usually experienced as a comfortable walking speed (4.8 km·h-1), and finally at a vigorous walking speed (6.4 km·h-1).12 Participants were instructed to walk in a natural way with a normal intuitive arm swing. During the measurements, the number of steps was counted with a manual hand counter by one observer; the number subsequently functioned as the gold standard. The measurements were also recorded with a

40 Reliability and validity of ten consumer activity trackers depend on walking speed

Abstract Introduction

Purpose Consumer activity trackers are an inexpensive and feasible method for estimating daily To examine the test-retest reliability and validity of ten activity trackers for step counting at physical activity. As the availability of these devices has increased, so has their use in daily three different walking speeds. life, health care, and medical science. Two commonly used physical activity guidelines are the 30-minutes of moderate to vigorous activity (MVPA) per day for at least five days a 1 2 Methods week. and the 10.000 steps/day norm. Research to a healthy amount of physical activity per day shows that engagement in at least 8000 to 11000 steps a day is related to many Thirty-one healthy participants walked twice on a treadmill for 30 minutes while wearing ten 3 health benefits, like a better physical fitness, body composition, and glycemic control.2,3 activity trackers (Polar Loop, Garmin Vivosmart, Fitbit Charge HR, Apple Watch Sport, Pebble When 3000 steps are taken at moderate to vigorous intensity, both guidelines correspond Smartwatch, Samsung Gear S, Misfit Flash, Jawbone Up Move, Flyfit and Moves). with each other.4 For physically inactive people (e.g. people who take on average less than Participants walked three walking speeds for ten minutes each; slow (3.2 km·h-1), average 5000 steps/day), an increment of 2000 steps per day already relates to health improvements (4.8 km·h-1), and vigorous (6.4 km·h-1). To measure test-retest reliability, intraclass like a better body composition and decrement of BMI.5 Therefore, activity trackers have a correlations (ICCs) were determined between the first and second treadmill test. Validity large value in objectifying ones physical activity pattern and demonstrating changes in one’s was determined by comparing the trackers with the gold standard (hand counting), using activity behavior. Activity trackers should therefore be reliable and valid. mean differences, mean absolute percentage errors, and ICCs. Statistical differences were calculated by paired-sample t-tests, Wilcoxon signed-rank tests, and by constructing Bland- Many trackers demonstrate acceptable validity and reliability of step counting, Altman plots. however, other activity trackers perform relatively inadequately.6,7 The accuracy of activity trackers that were recently released into the market is currently unknown. A common Results challenge of activity trackers is their validity for tracking activities at different walking speeds 8,9 Test-retest reliability varied with ICCs ranging from -0.02 to 0.97. Validity varied between including a slower walking speed. The latter could be an issue when self-tracking is used trackers and different walking speeds with mean differences between the gold standard and for the assessment of daily physical activity of patients with limited physical abilities or the 10,11 activity trackers ranging from 0.0 to 26.4%. Most trackers showed relatively low ICCs and elderly population. Validation of activity trackers at different speeds is thus important. broad limits of agreement of the Bland-Altman plots at the different speeds. For the slow This certainly accounts for wearables that have recently entered the market. To achieve this, walking speed, the Garmin Vivosmart and Fitbit Charge HR showed the most accurate the aim of this study is to examine the test-retest reliability and validity of ten relatively new results. The Garmin Vivosmart and Apple Watch Sport demonstrated the best accuracy at an activity trackers when walking at three different speeds. average walking speed. For vigorous walking, the Apple Watch Sport, Pebble Smartwatch, and Samsung Gear S exhibited the most accurate results. Methods

41 Chapter 3

video camera as a backup. The three times ten minutes time slots of the treadmill test were Statistical analyses measured with software from the Optogait system (OPTOGait, Microgate S.r.I, Italy, 2010). Descriptive statistics and their corresponding 95% confidence intervals were determined for Before and after each time slot the number of steps as recorded by the trackers was all variables. manually entered in a dedicated research form. During registration participants were asked Test-retest reliability was determined by calculating the ICCs between Session 1 and to stand still with their hands on the handrails of the treadmill. The number of steps was Session 2 (two-way random, absolute agreement, single measures) with 95% confidence read either directly from the trackers display or from the corresponding application, which intervals. An ICC > 0.90 was considered as excellent, 0.75 - 0.90 as good, 0.60 - 0.75 as were installed on an iPod touch (2014, Model A1509, Apple Inc., Cupertino, CA, USA). This moderate, and < 0.60 as low.13 Because negative values of the ICC theoretically don’t exist was the case for the Misfit Flash, Jawbone Up Move and the Flyfit. Observers typically (7,23), negative values were set to zero. Additionally, test-retest was assessed by calculating waited for one or two minutes before registration to allow the trackers to make Bluetooth or the mean differences and the mean absolute percentage errors (MAPE) between the Wi-Fi connection with the iPod for synchronization. The registration phase usually took no sessions. Significant mean differences were investigated by paired-sample t-tests and more than five minutes. Wilcoxon signed-rank tests.

Participants Validity was assessed by the mean difference and the mean absolute percentage Thirty-one healthy adults volunteered to participate in this study (16 males and 15 females; errors (MAPE) between the gold standard and the activity trackers. According to Feito et mean ± SD age 32 ± 12 years; mean ± SD; BMI 22.6 ± 2.4 kg·m-2). Participants were recruited al,14,15 a MAPE exceeding 5% can be considered as a practically relevant difference. by flyer advertisement and by word of mouth within the Hanze University of Applied Therefore, a 5% cut-off criterion was utilized for the MAPE. To determine the agreement Sciences, Groningen, the Netherlands. Participants were informed about the test procedures between the gold standard and activity trackers, Bland-Altman plots with the associated and signed an informed consent form prior to the study. The research was performed in limits of agreement were constructed. In addition, the agreement between the gold accordance with the Declaration of Helsinki and an exemption for a comprehensive standard and the activity trackers was determined by calculating intraclass correlation application was obtained by the Medical Ethical Committee of the University Medical Center coefficients (ICC) (two-way random, absolute agreement, single measures with 95% of Groningen. confidence intervals).

All statistical analyses were performed using SPSS 23 (SPSS Inc., Chicago, IL, USA), Activity trackers with a significance level of 5%. To correct for multiple testing, the significance level for the In this study nine activity trackers and one smartphone application were examined. A paired-sample t-tests and Wilcoxon tests was adjusted by using the posthoc correction manual hand counter (Voltcraft, Conrad Electronic SE, Hirschau, Germany) was used as the method of Bonferroni.16 This resulted in an alpha of 0.0045 for the test-retest analyzes, and gold standard. On their right wrist participants wore the Garmin Vivosmart (2014, Garmin an alpha of 0.005 for the validity analyzes. International Inc., Olathe, KS, USA) at the distal side, the Fitbit Charge HR (2014, Fitbit Inc., San Francisco, CA, USA) in the middle and the Polar Loop (2013, Polar Electro Oy, Kempele, Finland) at the proximal side. Three smartwatches were placed on their left wrist; the Apple Watch Sport (2015, Apple Inc., Cupertino, CA, USA) at the distal side, the Pebble Smartwatch Results (2014, Pebble Technology Corp., Redwood City, CA, USA) in the middle, and the Samsung Gear S (2014, Samsung Electronics Co, Ltd., Seoul, South Korea) at the proximal side. The On average participants walked 947 ± 54 steps at 3.2 km·h-1, 1112 ± 45 steps at 4.8 km·h-1 Misfit Flash (2014, Misfit Wearables, Burlingame, CA, USA) and Jawbone Up Move (2014, and 1254 ± 53 steps at 6.4 km·h-1, as measured with the gold standard. The mean number of Jawbone Inc., Beverly Hills, CA, USA) were attached at their right hip to the belt of their steps measured by the activity trackers and their 95% confidence intervals during both trousers. The Flyfit (2014, Flyfit Inc., San Francisco, CA, USA) was worn on their right ankle. sessions are depicted in Figures 1, 2, and 3. No sex differences were found in the results. The Finally, a smartphone (Samsung S5 Active, Samsung Electronics Co, Ltd., Seoul, South Korea) number of participants in the different conditions varied (from n=31 to n=21). This variation on which the Moves application (ProtoGeo, Helsinki, Finland) was installed was placed in the was mainly due to a number of occasions in which there was a delay in synchronization, front or back pocket of their trousers. leading to underestimation of the number of steps in the preceding time slot and occasionally an overestimation in the next. Every observed delay was recorded in a diary and involved metrics were excluded from data analysis. This resulted in a lower number of observations for some trackers (especially Flyfit, Misfit Flash, Jawbone up Move).

42 Reliability and validity of ten consumer activity trackers depend on walking speed

video camera as a backup. The three times ten minutes time slots of the treadmill test were Statistical analyses measured with software from the Optogait system (OPTOGait, Microgate S.r.I, Italy, 2010). Descriptive statistics and their corresponding 95% confidence intervals were determined for Before and after each time slot the number of steps as recorded by the trackers was all variables. manually entered in a dedicated research form. During registration participants were asked Test-retest reliability was determined by calculating the ICCs between Session 1 and to stand still with their hands on the handrails of the treadmill. The number of steps was Session 2 (two-way random, absolute agreement, single measures) with 95% confidence read either directly from the trackers display or from the corresponding application, which intervals. An ICC > 0.90 was considered as excellent, 0.75 - 0.90 as good, 0.60 - 0.75 as were installed on an iPod touch (2014, Model A1509, Apple Inc., Cupertino, CA, USA). This moderate, and < 0.60 as low.13 Because negative values of the ICC theoretically don’t exist was the case for the Misfit Flash, Jawbone Up Move and the Flyfit. Observers typically (7,23), negative values were set to zero. Additionally, test-retest was assessed by calculating waited for one or two minutes before registration to allow the trackers to make Bluetooth or 3 the mean differences and the mean absolute percentage errors (MAPE) between the Wi-Fi connection with the iPod for synchronization. The registration phase usually took no sessions. Significant mean differences were investigated by paired-sample t-tests and more than five minutes. Wilcoxon signed-rank tests.

43 Chapter 3

Figure 3. Mean number of steps and 95% confidence interval (95% CI) of the activity trackers at 6.4 km·h-1 during session 1 (a) and session 2 (b). The horizontal lines represent the mean number of steps of the gold standard (1259 ± 53 and 1251 ± 54 steps respectively). Figure 1. Mean number of steps and 95% confidence interval (95% CI) of the activity trackers at 3.2 km·h-1 during session 1 (a) and session 2 (b). The horizontal lines represent the mean number of steps of the gold standard (953 ± 46 Test-retest reliability and 940 ± 61 steps respectively). -1 The outcome measures of test-retest reliability are shown in Table 1. At 3.2 km·h , the mean differences between Sessions 1 and 2 varied from seven steps (MAPE 0.7%, Apple) to 75 steps (MAPE 9.0%, Flyfit). At 4.8 km·h-1, the mean differences varied from three steps (MAPE -0.3%, Moves) to 93 steps (MAPE 8.6%, Polar Loop) and differed significantly for the Apple Watch Sport. At 6.4 km·h-1, the mean differences varied from zero steps (MAPE 0.0%, Pebble Smartwatch) to 40 steps (MAPE 3.5%, Garmin Vivosmart). The ICCs of the gold standard were good at slow and average walking speeds (0.76 and 0.87 respectively) and excellent at a vigorous walking speed (0.93). At 3.2 km·h-1, the ICCs of the trackers ranged from -0.02 (Moves) to 0.97 (Samsung Gear S). At this slowest walking speed, most of the trackers demonstrated low ICCs. The Moves showed a very low ICC, while the Polar Loop and Fitbit Charge HR showed moderate ICCs, the Garmin Vivosmart exhibited a good ICC, and the Samsung Gear S showed an excellent ICC. At 4.8 km·h-1, the ICCs of the trackers ranged from 0.00 (Jawbone) to 0.86 (Samsung Gear S). The Fitbit Charge HR demonstrated a moderate ICC and Samsung Gear S showed a good ICC. All of the other Figure 2. Mean number of steps and 95% confidence interval (95% CI) of the activity trackers at 4.8 km·h-1 during trackers showed low ICCs at this average walking speed. At 6.4 km·h-1, the ICCs of the session 1 (a) and session 2 (b). The horizontal lines represent the mean number of steps of the gold standard trackers ranged from 0.14 (Misfit) to 0.93 (Samsung Gear S). Here the Polar Loop, Misfit (1117 ± 44 and 1108 ± 46 steps respectively). Flash, and Flyfit showed low ICCs, while the Garmin Vivosmart, Fitbit Charge HR, Jawbone Up Move, and Moves showed moderate ICCs. There were two trackers (Apple Watch Sport and Pebble Smartwatch) that indicated good ICCs and one tracker (Samsung Gear S) that showed an excellent ICC at the vigorous walking speed.

44 Reliability and validity of ten consumer activity trackers depend on walking speed

45 Chapter 3

Table 1. Validity Test-retest reliability measures of session 1 versus session 2: mean differences (session 1 - session 2) ± The outcome measures of the validity tests are shown in Tables 2 and 3. Each column standard error (SE), mean absolute percentage error (MAPE), scores on the paired samples t-test and Wilcoxon signed-rank test, intraclass correlation coefficient (ICC), and the corresponding 95% confidence intervals (95% contains 30 measurements (ten trackers at three different speeds). A total number of 12 out CI). of 30 measurements showed a significant difference of the mean number of steps compared to the gold standard in Session 1, and 12 measurements showed a significant difference in Activity tracker Speed N Mean difference ± SE MAPE (%) t-valuea/ ICC 95% CIc (km·h -1) Z-valueb Session 2 assessed by either the paired samples t-test or the Wilcoxon signed-rank test. With increasing speed, the MAPE decreased for the Polar Loop, Pebble Smartwatch, Samsung Hand counter 3.2 31 13 ± 6 1.3 1.98a 0.76** 0.56 - 0.88 Gear S, Misfit Flash, Jawbone UP Move, Flyfit, and Moves. It was fairly constant for the Apple 4.8 31 10 ± 4 0.9 2.59a 0.87** 0.72 - 0.94 Watch Sport at all three speeds. The MAPE increased for the Garmin Vivosmart and the a 6.4 30 4 ± 4 0.3 1.01 0.93** 0.86 - 0.97 Fitbit Charge HR with accelerating speed. At a walking speed of 3.2 km·h-1, the Polar Loop, Polar Loop 3.2 31 9 ± 46 1.3 -0.01b 0.74** 0.52 - 0.87 Misfit Flash, Jawbone Up Move, Flyfit, and Moves had a MAPE exceeding 5%. The MAPE of 4.8 30 93 ± 41 8.6 -2.66b 0.15 -0.17 - 0.46 the Pebble Smartwatch and Samsung Gear S was higher than 5% during Session 1, but under 6.4 29 -2 ± 19 -0.1 -0.13b 0.49** 0.15 - 0.72 5% during Session 2. All other trackers had a MAPE less than 5% at the slowest walking Garmin Vivosmart 3.2 31 12 ± 7 1.2 1.73a 0.79** 0.60 - 0.89 speed. At 4.8 km·h-1, the MAPE of the Misfit Flash, Jawbone Up Move, and Flyfit was more 4.8 31 16 ± 7 1.4 2.16a 0.51** 0.20 - 0.72 than 5% during both sessions. The Jawbone Up Move had a MAPE over 5% only during 6.4 30 40 ± 22 3.5 -1.58b 0.72** 0.49 - 0.86 Session 1 while the Polar Loop obtained this only during Session 2. Finally, at 6.4 km·h-1, Fitbit Charge HR 3.2 31 12 ± 10 1.2 -1.28b 0.73** 0.51 - 0.86

4.8 31 7 ± 10 0.6 -1.47b 0.70** 0.46 - 0.84 most of the trackers had a MAPE under 5% except for the Garmin Vivosmart, Fitbit Charge

6.4 30 33 ± 14 2.8 -2.37b 0.65** 0.38 - 0.82 HR, and Misfit Flash. The Flyfit had a MAPE of less than 5% during Session 1, however, it Apple Watch Sport 3.2 30 7 ± 16 0.7 -0.73b 0.38* 0.02 - 0.65 exceeded 5% during Session 2.

4.8 28 41 ± 13 3.7 -3.22b # 0.48** 0.12 - 0.73 The limits of agreement of the Bland-Altman plots are presented in Tables 2 and 3. At a 6.4 28 -2 ± 8 -0.1 -0.18 0.80** 0.61 - 0.90 3.2 km·h-1, the Garmin Vivosmart had the narrowest limits of agreement (49 steps, Session Pebble Smartwatch 3.2 31 -16 ± 17 -1.8 -0.12b 0.56** 0.26 - 0.76 1), while the Polar Loop exhibited the broadest limits of agreement (1298 steps, Session 1). 4.8 31 -7 ± 20 -0.6 -1.70b 0.33* -0.03 - 0.61 At 4.8 km·h-1, the Garmin Vivosmart had the narrowest limits of agreement (35 steps, 6.4 30 0 ± 5 0.0 0.06a 0.89** 0.79 - 0.95 Session 2) and the Misfit Flash had the broadest limits of agreement (1104 steps, Session 2). Samsung Gear S 3.2 29 10 ± 9 1.1 -0.98b 0.97** 0.93 - 0.98 -1 4.8 30 4 ± 10 0.3 -1.88b 0.86** 0.73 - 0.93 At 6.4 km·h , the Samsung Gear S had the narrowest limits of agreement (73 steps, Session

6.4 30 3 ± 3 0.2 0.73a 0.93** 0.86 - 0.97 2) and the Misfit Flash had the broadest limits of agreement (1029 steps, Session 2). Misfit Flash 3.2 22 74 ± 56 9.1 -0.92b 0.48** 0.10 - 0.74 The ICCs (Tables 2 and 3) at 3.2 km·h-1 ranged from 0 (Samsung Gear S, session 2 and 4.8 23 25 ± 60 2.4 -0.42b 0.03 -0.40 - 0.44 Moves, Session 2) to 0.95 (Garmin Vivosmart, Sessions 1 and 2) while, at 4.8 km·h-1, ICCs b 6.4 22 -2 ± 62 -0.1 -1.25 0.14 -0.31 - 0.53 ranged from 0 (Moves, Session 1 and 2, Polar Loop session 2) to 0.98 (Garmin Vivosmart, Jawbone Up Move 3.2 28 37 ± 38 4.2 -0.47b 0.07 -0.31 - 0.42 Session 2). ICCs at 6.4 km·h-1 ranged from 0 (Garmin Vivosmart, Session 2) to 0.92 (Samsung 4.8 30 -29 ± 35 -2.7 -0.82a 0.00 -0.36 - 0.36 Gear S, Session 2). Generally, ICCs were higher at vigorous walking speed compared to the 6.4 29 10 ± 10 0.8 -1.10b 0.65** 0.38 - 0.82 slow walking speed except for Garmin Vivosmart and Fitbit Charge HR which showed the Flyfit 3.2 23 75 ± 62 9.0 -1.25b 0.15 -0.26 - 0.52 highest ICCs at the slow walking speed. 4.8 22 32 ± 32 3.0 -1.67b 0.58** 0.23 - 0.80

6.4 18 11 ± 42 0.9 -0.63b 0.46* -0.01 - 0.76 Moves 3.2 25 -57 ± 75 -6.9 -0.76a -0.02 -0.42 - 0.37

4.8 26 -3 ± 30 -0.3 -0.11a 0.49** 0.13 - 0.74

6.4 28 -8 ± 22 -0.7 -0.38a 0.66** 0.38 - 0.83 # p<0.0045; *p<0.05; **p<0.01; a paired samples t-test; b Wilcoxon signed-rank test in case of a non-normal distribution; c 95% CI of the ICC.

46 Reliability and validity of ten consumer activity trackers depend on walking speed

Table 1. Validity Test-retest reliability measures of session 1 versus session 2: mean differences (session 1 - session 2) ± The outcome measures of the validity tests are shown in Tables 2 and 3. Each column standard error (SE), mean absolute percentage error (MAPE), scores on the paired samples t-test and Wilcoxon signed-rank test, intraclass correlation coefficient (ICC), and the corresponding 95% confidence intervals (95% contains 30 measurements (ten trackers at three different speeds). A total number of 12 out CI). of 30 measurements showed a significant difference of the mean number of steps compared to the gold standard in Session 1, and 12 measurements showed a significant difference in Activity tracker Speed N Mean difference ± SE MAPE (%) t-valuea/ ICC 95% CIc (km·h -1) Z-valueb Session 2 assessed by either the paired samples t-test or the Wilcoxon signed-rank test. With increasing speed, the MAPE decreased for the Polar Loop, Pebble Smartwatch, Samsung Hand counter 3.2 31 13 ± 6 1.3 1.98a 0.76** 0.56 - 0.88 Gear S, Misfit Flash, Jawbone UP Move, Flyfit, and Moves. It was fairly constant for the Apple 4.8 31 10 ± 4 0.9 2.59a 0.87** 0.72 - 0.94 Watch Sport at all three speeds. The MAPE increased for the Garmin Vivosmart and the a 3 6.4 30 4 ± 4 0.3 1.01 0.93** 0.86 - 0.97 Fitbit Charge HR with accelerating speed. At a walking speed of 3.2 km·h-1, the Polar Loop, Polar Loop 3.2 31 9 ± 46 1.3 -0.01b 0.74** 0.52 - 0.87 Misfit Flash, Jawbone Up Move, Flyfit, and Moves had a MAPE exceeding 5%. The MAPE of 4.8 30 93 ± 41 8.6 -2.66b 0.15 -0.17 - 0.46 the Pebble Smartwatch and Samsung Gear S was higher than 5% during Session 1, but under 6.4 29 -2 ± 19 -0.1 -0.13b 0.49** 0.15 - 0.72 5% during Session 2. All other trackers had a MAPE less than 5% at the slowest walking Garmin Vivosmart 3.2 31 12 ± 7 1.2 1.73a 0.79** 0.60 - 0.89 speed. At 4.8 km·h-1, the MAPE of the Misfit Flash, Jawbone Up Move, and Flyfit was more 4.8 31 16 ± 7 1.4 2.16a 0.51** 0.20 - 0.72 than 5% during both sessions. The Jawbone Up Move had a MAPE over 5% only during 6.4 30 40 ± 22 3.5 -1.58b 0.72** 0.49 - 0.86 Session 1 while the Polar Loop obtained this only during Session 2. Finally, at 6.4 km·h-1, Fitbit Charge HR 3.2 31 12 ± 10 1.2 -1.28b 0.73** 0.51 - 0.86

4.8 31 7 ± 10 0.6 -1.47b 0.70** 0.46 - 0.84 most of the trackers had a MAPE under 5% except for the Garmin Vivosmart, Fitbit Charge

6.4 18 11 ± 42 0.9 -0.63b 0.46* -0.01 - 0.76 Moves 3.2 25 -57 ± 75 -6.9 -0.76a -0.02 -0.42 - 0.37

4.8 26 -3 ± 30 -0.3 -0.11a 0.49** 0.13 - 0.74

6.4 28 -8 ± 22 -0.7 -0.38a 0.66** 0.38 - 0.83 # p<0.0045; *p<0.05; **p<0.01; a paired samples t-test; b Wilcoxon signed-rank test in case of a non-normal distribution; c 95% CI of the ICC.

47 Chapter 3

Table 2. Table 3. Validity measures of session 1: mean differences (hand counter – activity tracker) ± standard error (SE), mean Validity measures of session 2: mean differences (hand counter – activity tracker) ± standard error (SE), mean absolute percentage error (MAPE), scores on the paired samples t-test and Wilcoxon signed-rank test, limits of absolute percentage error (MAPE), scores on the paired samples t-test and Wilcoxon signed-rank test, limits of agreement of the Bland-Altman plots, intraclass correlation coefficient (ICC), and the corresponding 95% agreement of the Bland-Altman plots, intraclass correlation coefficient (ICC), and the corresponding 95% confidence intervals (95% CI). confidence intervals (95% CI).

Activity tracker Speed N Mean difference MAPE (%) t-valuea/ Limits of agreement ICC 95% CIc Activity tracker Speed N Mean MAPE (%) t-valuea/ Limits of ICC 95% CIc (km·h-1) ± SE Z-valueb (km·h-1) difference ± Z-valueb agreement Lower Upper SE Polar Loop 3.2 31 252 ± 59 26.4 -3.86b # -397 901 0.08 -0.15 - 0.35 Lower Upper b # b Polar Loop 3.2 31 248 ± 58 26.3 -3.34 -387 882 0.09 -0.14 - 0.36 4.8 31 34 ± 15 3.0 -2.06 -127 195 0.26 -0.06 - 0.54 # # 4.8 30 119 ± 44 10.7 -3.53b -350 588 0 -0.28 - 0.30 6.4 31 45 ± 20 3.6 -3.08b -174 264 0.24 -0.09 - 0.53 b Garmin Vivosmart 3.2 31 10 ± 2 1.0 4.36a # -15 34 0.95** 0.78 - 0.98 6.4 29 38 ± 12 3.0 -2.74 -92 168 0.42** 0.07 - 0.68 a a Garmin Vivosmart 3.2 31 9 ± 3 0.9 2.51 -29 46 0.95** 0.88 - 0.98 4.8 31 -2 ± 7 -0.2 -0.32 -81 77 0.57** 0.27 - 0.77 a # 4.8 31 4 ± 2 0.3 2.41 -14 21 0.98** 0.95 - 0.99 6.4 31 114 ± 27 9.0 -4.44b -177 404 0.10 -0.14 - 0.36 b # Fitbit Charge HR 3.2 31 -7 ± 9 -0.7 -0.81 -101 87 0.62** 0.35 - 0.80 6.4 30 149 ± 34 11.9 -4.17 -220 518 0 -0.26 - 0.20 b a Fitbit Charge HR 3.2 31 -8 ± 9 -0.9 -0.13 -108 92 0.74** 0.54 - 0.87 4.8 31 22 ± 13 2.0 1.70 -118 162 0.20 -0.14 - 0.50 b # 4.8 31 19 ± 13 1.7 -1.93 -122 160 0.27 -0.07 - 0.56 6.4 31 65 ± 14 5.2 -4.61b -83 214 0.31** -0.05 - 0.60 b # Apple Watch Sport 3.2 30 18 ± 9 1.9 2.14a -74 111 0.57** 0.27 - 0.77 6.4 30 96 ± 20 7.7 -4.26 -116 308 0.15 -0.10 - 0.42 b a Apple Watch Sport 3.2 31 13 ± 10 1.4 -0.83 -98 124 0.73** 0.52 - 0.86 4.8 29 0 ± 3 0.0 -0.09 -36 35 0.93** 0.86 - 0.97 b a 4.8 30 29 ± 12 2.6 -2.20 -101 159 0.52** 0.21 - 0.74 6.4 30 6 ± 5 0.5 1.24 -45 56 0.91** 0.82 - 0.95 a Pebble Smartwatch 3.2 31 57 ± 19 6.0 -4.29b # -154 269 0.28* -0.03 - 0.56 6.4 29 1 ± 6 0.1 0.15 -65 67 0.86** 0.72 - 0.93 # # Pebble Smartwatch 3.2 31 29 ± 6 3.0 4.67a -38 96 0.78** 0.33 - 0.92 4.8 31 32 ± 19 2.9 -4.08b -179 244 0.34* 0.00 - 0.61 a # 4.8 31 16 ± 6 1.4 2.65 -48 80 0.77** 0.54 - 0.89 6.4 31 16 ± 5 1.3 3.37a -36 67 0.86** 0.65 - 0.94 a # Samsung Gear S 3.2 31 53 ± 32 5.6 -2.01b -300 406 0.04 -0.30 - 0.37 6.4 30 13 ± 4 1.0 3.51 -27 52 0.91** 0.72 - 0.96 b a Samsung Gear S 3.2 29 45 ± 41 4.8 -0.16 -387 477 0 -0.49 - 0.22 4.8 31 45 ± 22 4.0 2.05 -195 285 0.02 -0.29 - 0.34 # # 4.8 30 38 ± 17 3.5 -4.08b -149 225 0.17 -0.15 - 0.48 6.4 31 14 ± 5 1.1 3.15a -36 65 0.85** 0.63 - 0.93 a Misfit Flash 3.2 25 144 ± 45 15.2 -3.84b # -298 586 0.06 -0.21 - 0.38 6.4 30 9 ± 3 0.8 2.74 -27 46 0.92** 0.81 - 0.97 b # b Misfit Flash 3.2 27 170 ± 47 18.1 -4.16 -312 652 0.15 -0.13 - 0.45 4.8 25 60 ± 31 5.4 -1.71 -241 362 0.26 -0.10 - 0.58 b a 4.8 27 94 ± 54 8.5 -0.86 -458 646 0.07 -0.27 - 0.42 6.4 28 75 ± 47 6.0 1.62 -409 560 0.11 -0.24 - 0.45 b Jawbone Up Move 3.2 29 83 ± 26 8.7 3.17a # -193 358 0.12 -0.16 - 0.42 6.4 25 93 ± 53 7.5 -2.29 -421 608 0.08 -0.28 - 0.44 b # a Jawbone Up Move 3.2 29 110 ± 26 11.7 -4.12 -160 381 0.17 -0.11 - 0.45 4.8 31 65 ± 30 5.9 2.16 -265 396 0.09 -0.22 - 0.41 a a 4.8 30 29 ± 11 2.6 2.71 -85 143 0.56** 0.23 - 0.77 6.4 30 15 ± 8 1.2 1.78 -75 105 0.71** 0.47 - 0.85 b # Flyfit 3.2 28 154 ± 30 16.1 5.13a # -157 464 0.18 -0.10 - 0.47 6.4 30 21 ± 8 1.6 -3.40 -60 101 0.72** 0.45 - 0.86 b # b Flyfit 3.2 26 185 ± 47 19.5 -4.20 -281 651 0.17 -0.11 - 0.47 4.8 26 60 ± 25 5.3 -2.25 -192 312 0.31* -0.04 - 0.60 b # a 4.8 27 77 ± 26 7.0 -3.36 -185 339 0.32* -0.02 - 0.61 6.4 21 29 ± 39 2.3 0.75 -317 375 0.27 -0.18 - 0.62 b Moves 3.2 29 133 ± 51 14.0 -2.01b -406 671 0.15 -0.16 - 0.46 6.4 27 85 ± 46 6.8 -2.45 -386 556 0.08 -0.26 - 0.43 b a Moves 3.2 27 119 ± 63 12.6 -1.57 -524 762 0 -0.41 - 0.27 4.8 29 -29 ± 23 -2.6 -1.30 -268 209 0 -0.52 - 0.17 b a 4.8 28 -9 ± 43 -0.8 -1.32 -460 442 0 -0.49 - 0.26 6.4 29 4 ± 24 0.3 0.15 -253 261 0.25 -0.14 - 0.56 6.4 30 -3 ± 20 -0.2 -1.51b -220 215 0.37* 0.06 - 0.64 #p<0.005; *p<0.05; **p<0.01; a paired samples t-test; b Wilcoxon signed-rank test in case of a non-normal distribution. c 95% CI of the ICC. # p<0.005; * p<0.05; ** p<0.01; a paired samples t-test. b Wilcoxon signed-rank test in case of a non-normal distribution. c 95% CI of the ICC.

48 Reliability and validity of ten consumer activity trackers depend on walking speed

49 Chapter 3

Discussion rather than slow walking, which may well explain why the Polar Loop was inadequate for tracking steps at a slower walking speed. Only a small number of trackers demonstrated

good validity at 3.2 km·h-1. When examining both the test-retest reliability and validity, the The aim of this study was to examine the test-retest reliability and validity of ten relatively Garmin Vivosmart showed the best results. The Fibit Charge HR also indicated good results. new activity trackers at three different walking speeds. In general, the results showed that The other trackers had inadequate results on the test-retest reliability and/or the validity. validity and reliability are strongly influenced by walking speed. Though most trackers showed acceptable validity scores at an average walking speed, none of the trackers had Most participants confirmed that the average walking speed of 4.8 km·h-1 was normal valid step counting measures at all three walking speeds. Most trackers seem to for them, as described previously.12 At this speed, accelerations of the body are higher than underestimate the number of steps at each walking speed (with a few exceptions; Fitbit at at 3.2 km·h-1 and, therefore, better results on test-retest reliability and validity were the slowest speed and the Moves app at average and fastest speed). A systematic expected. For the test-retest reliability, this was the case for five out of ten trackers. Only underestimation at slow walking speed has been described before in literature.17 Also, three trackers showed a better test-retest reliability at 4.8 km·h-1 than at 3.2 km·h-1 (Apple mobile applications have been shown to be associated with high variation when tracking Watch Sport, Flyfit and Moves) and two remained approximately the same (Fitbit Charge HR walking on a treadmill.7,17 The general underestimation here may be the result of a and Jawbone Up Move). The other five trackers had lower test-retest reliability at 4.8 km·h-1. systematic bias that affected all devices. This is probably due to the onset and offset of the Most trackers did exhibit a profound improvement in the results at 4.8 km·h-1 than at 3.2 treadmill. Each step is recorded by the gold standard, but the first and last steps are km·h-1 in regard to validity. Eight trackers had a MAPE smaller than 5% during one of the associated with limited acceleration which was probably not enough to be registered by sessions, and six trackers remained below this cut-off point during both sessions. All other accelerometry and thus leading to a small systematic underestimation. trackers that did not meet the 5% criterion still had a MAPE lower than 6%. Only the Polar Loop showed an unexpected high MAPE (10.7%) during the second session. In general, it can In this study, the slowest walking speed was 3.2 km·h-1. At this speed, it is difficult for be concluded that, at this average speed, most trackers show acceptable validity results. the activity trackers to detect accelerations.8,9 It is not only recognized as being problematic Notably, these validity results are accompanied with relatively broad limits of agreement of for activity trackers to do a valid count of steps at a slow walking speed. Our study also the Bland-Altman plots (an average of 39.4% of the total number of steps of the gold showed that the participants had difficulty maintaining a constant pace at 3.2 km·h-1. The standard). The low MAPE in combination with broad LOAs indicate that, although these gold standard had an ICC of only 0.76 which indicates that there were possibly actual trackers, on average, have acceptable validity, this performance varies between individual differences between Sessions 1 and 2 in the number of steps taken by the participants. This participants. Only the Garmin Vivosmart, Apple Watch Sport and Pebble Smartwatch had a can plausibly be explained by the fact that a speed of 3.2 km·h-1 was too slow for many of MAPE less than 5% and narrow limits of agreement (within 20% of the average number of the healthy participants which made it difficult for them to walk in a natural way. This was, steps of the gold standard). However, all of these four trackers had a low test-retest for example, visible in a very small step length or only minimal arm swing during walking. reliability whereby those of the Garmin Vivosmart and Apple Watch Sport were generally The participants probably selected slightly different strategies to compensate for the slow acceptable. speed between Sessions 1 and 2 which resulted in a different number of steps between the sessions even though the speed and distance covered were equal. The Samsung Gear S Most trackers showed the best results at 6.4 km·h-1. For the test-retest reliability, showed the highest ICC (ICC=0.96) at 3.2 km·h-1. However, this differs from the gold there were only three trackers that had low ICCs (Polar Loop, Misfit Flash and Flyfit). The standard which demonstrates that the actual test-retest reliability of the Samsung Gear S at three smartwatches (Apple Watch Sport, Pebble Smartwatch and Samsung Gear S) had the 3.2 km·h-1 may be not that good. The ICCs of the Polar Loop, Garmin Vivosmart, and Fitbit best test-retest reliability. The validity was also generally better than at the slower speeds. Charge HR (ICCs of 0.74, 0.79 and 0.73, respectively) are more equal to the ICC of the gold However, there were two trackers (Garmin Vivosmart and Fitbit Charge HR) that had the standard indicating that these trackers had the best test-retest reliability at 3.2 km·h-1. The poorest validity at the highest walking speed. The explanation for this finding is unclear, validity was probably also influenced by the unnatural walking pattern at 3.2 km·h-1. The however, these two trackers most likely just perform best at slow walking speeds. This is slow walking speed and the unnatural walking pattern at 3.2 km·h-1 resulted in low validity of especially remarkable for the Garmin Vivosmart since Garmin is a sports brand and, most trackers. Only three trackers had a MAPE smaller than 5%. The limits of agreement of therefore, better results at higher speeds were expected. The other trackers had a lower the Bland-Altman plots were high with an average of 65.8% of the total number of steps of and, in a number of cases, a similar MAPE when compared to 3.2 and 4.8 km·h-1. However, the gold standard. The Polar Loop was particularly not able to make a valid measurement of for most trackers the Bland-Altman plots showed relatively broad limits of agreement (on step counting as illustrated by an exceptionally high MAPE of 26% during both sessions. This average, 32.9% of the average number of steps of the gold standard). This indicates that may be explained by the fact that this activity tracker was developed for sports activities there were many individual differences in the results of the trackers also at 6.4 km·h-1, and

50 Reliability and validity of ten consumer activity trackers depend on walking speed

Discussion rather than slow walking, which may well explain why the Polar Loop was inadequate for tracking steps at a slower walking speed. Only a small number of trackers demonstrated

good validity at 3.2 km·h-1. When examining both the test-retest reliability and validity, the The aim of this study was to examine the test-retest reliability and validity of ten relatively Garmin Vivosmart showed the best results. The Fibit Charge HR also indicated good results. new activity trackers at three different walking speeds. In general, the results showed that The other trackers had inadequate results on the test-retest reliability and/or the validity. validity and reliability are strongly influenced by walking speed. Though most trackers showed acceptable validity scores at an average walking speed, none of the trackers had Most participants confirmed that the average walking speed of 4.8 km·h-1 was normal valid step counting measures at all three walking speeds. Most trackers seem to for them, as described previously.12 At this speed, accelerations of the body are higher than underestimate the number of steps at each walking speed (with a few exceptions; Fitbit at at 3.2 km·h-1 and, therefore, better results on test-retest reliability and validity were 3 the slowest speed and the Moves app at average and fastest speed). A systematic expected. For the test-retest reliability, this was the case for five out of ten trackers. Only underestimation at slow walking speed has been described before in literature.17 Also, three trackers showed a better test-retest reliability at 4.8 km·h-1 than at 3.2 km·h-1 (Apple mobile applications have been shown to be associated with high variation when tracking Watch Sport, Flyfit and Moves) and two remained approximately the same (Fitbit Charge HR walking on a treadmill.7,17 The general underestimation here may be the result of a and Jawbone Up Move). The other five trackers had lower test-retest reliability at 4.8 km·h-1. systematic bias that affected all devices. This is probably due to the onset and offset of the Most trackers did exhibit a profound improvement in the results at 4.8 km·h-1 than at 3.2 treadmill. Each step is recorded by the gold standard, but the first and last steps are km·h-1 in regard to validity. Eight trackers had a MAPE smaller than 5% during one of the associated with limited acceleration which was probably not enough to be registered by sessions, and six trackers remained below this cut-off point during both sessions. All other accelerometry and thus leading to a small systematic underestimation. trackers that did not meet the 5% criterion still had a MAPE lower than 6%. Only the Polar Loop showed an unexpected high MAPE (10.7%) during the second session. In general, it can In this study, the slowest walking speed was 3.2 km·h-1. At this speed, it is difficult for be concluded that, at this average speed, most trackers show acceptable validity results. the activity trackers to detect accelerations.8,9 It is not only recognized as being problematic Notably, these validity results are accompanied with relatively broad limits of agreement of for activity trackers to do a valid count of steps at a slow walking speed. Our study also the Bland-Altman plots (an average of 39.4% of the total number of steps of the gold showed that the participants had difficulty maintaining a constant pace at 3.2 km·h-1. The standard). The low MAPE in combination with broad LOAs indicate that, although these gold standard had an ICC of only 0.76 which indicates that there were possibly actual trackers, on average, have acceptable validity, this performance varies between individual differences between Sessions 1 and 2 in the number of steps taken by the participants. This participants. Only the Garmin Vivosmart, Apple Watch Sport and Pebble Smartwatch had a can plausibly be explained by the fact that a speed of 3.2 km·h-1 was too slow for many of MAPE less than 5% and narrow limits of agreement (within 20% of the average number of the healthy participants which made it difficult for them to walk in a natural way. This was, steps of the gold standard). However, all of these four trackers had a low test-retest for example, visible in a very small step length or only minimal arm swing during walking. reliability whereby those of the Garmin Vivosmart and Apple Watch Sport were generally The participants probably selected slightly different strategies to compensate for the slow acceptable. speed between Sessions 1 and 2 which resulted in a different number of steps between the sessions even though the speed and distance covered were equal. The Samsung Gear S Most trackers showed the best results at 6.4 km·h-1. For the test-retest reliability, showed the highest ICC (ICC=0.96) at 3.2 km·h-1. However, this differs from the gold there were only three trackers that had low ICCs (Polar Loop, Misfit Flash and Flyfit). The standard which demonstrates that the actual test-retest reliability of the Samsung Gear S at three smartwatches (Apple Watch Sport, Pebble Smartwatch and Samsung Gear S) had the 3.2 km·h-1 may be not that good. The ICCs of the Polar Loop, Garmin Vivosmart, and Fitbit best test-retest reliability. The validity was also generally better than at the slower speeds. Charge HR (ICCs of 0.74, 0.79 and 0.73, respectively) are more equal to the ICC of the gold However, there were two trackers (Garmin Vivosmart and Fitbit Charge HR) that had the standard indicating that these trackers had the best test-retest reliability at 3.2 km·h-1. The poorest validity at the highest walking speed. The explanation for this finding is unclear, validity was probably also influenced by the unnatural walking pattern at 3.2 km·h-1. The however, these two trackers most likely just perform best at slow walking speeds. This is slow walking speed and the unnatural walking pattern at 3.2 km·h-1 resulted in low validity of especially remarkable for the Garmin Vivosmart since Garmin is a sports brand and, most trackers. Only three trackers had a MAPE smaller than 5%. The limits of agreement of therefore, better results at higher speeds were expected. The other trackers had a lower the Bland-Altman plots were high with an average of 65.8% of the total number of steps of and, in a number of cases, a similar MAPE when compared to 3.2 and 4.8 km·h-1. However, the gold standard. The Polar Loop was particularly not able to make a valid measurement of for most trackers the Bland-Altman plots showed relatively broad limits of agreement (on step counting as illustrated by an exceptionally high MAPE of 26% during both sessions. This average, 32.9% of the average number of steps of the gold standard). This indicates that may be explained by the fact that this activity tracker was developed for sports activities there were many individual differences in the results of the trackers also at 6.4 km·h-1, and

51 Chapter 3

most trackers cannot be used interchangeably with the gold standard. Only the Apple Watch the highest walking speed, but also because these trackers seem to meet the current trend Sport, Pebble Smartwatch, Samsung Gear S and Jawbone Up Move had narrow limits of in technological developments. However, smartwatches are usually more expensive than agreement (within 20% of the average number of steps of the gold standard) and a MAPE most activity trackers, which can be a serious drawback. Fortunately, most trackers showed smaller than 5%. Since the Apple Watch Sport, Pebble Smartwatch and Samsung Gear S also a reasonably good test-retest reliability, and validity was increasing with increasing walking had the best test-retest reliability, it can be concluded that the three smartwatches speed. So, for individual users who want to measure changes in their physical activity exhibited the best results at 6.4 km·h-1. pattern, most trackers are suitable.

In our study, the placement of the trackers was always at the same position on the In conclusion, there is variation in test-retest reliability and validity of consumer hip, wrist or ankle for each participant during each session. Wrist trackers were also activity trackers which largely depends on walking speed. At a slower walking speed, the consistently placed in the same order from distal to proximal. It should be noted there is an Garmin Vivosmart and Fitbit Charge HR showed the most favorable results. For average open debate about possible differences in accelerations between the dominant and non- walking speed, the Garmin Vivosmart and Apple Watch Sport indicated the best results. For dominant wrist. One study showed that trackers at the non-dominant wrist have a higher vigorous walking, the Apple Watch Sport, Pebble Smartwatch, and Samsung Gear S validity than trackers at the dominant wrist.18 Another study showed no differences in demonstrated the most favorable results. In general, it can be concluded that the activity validity between the dominant and non-dominant wrist.19 These contradicting results trackers are more valid at higher walking speeds, and smartwatches showed slightly better indicate that in a (semi) free-living setting, a counterbalanced placement would have been results than the wearables that are solely developed for activity tracking. more appropriate to exclude possible different acceleration values between the dominant and non-dominant wrist. In our study participants were tested in a controlled lab setting.

Walking on a treadmill is typically characterized with symmetrical movement of the dominant and non-dominant arm. Thus, an absence of counterbalanced placement in our study design is unlikely to affect our results. Furthermore, it should be noted that two trackers we tested (Misfit Flash and Jawbone Up Move) can be worn on both the hip and the wrist. However, the results from hip and wrist are not interchangeable.20,21 In this study we choose to place them on the hip because research showed that this positioning is associated with higher validity compared to the wrist.7,18 Therefore, the results of the Misfit Flash and Jawbone Up Move in this study do only apply when wearing them on the hip.

Some caution should be taken regarding the generalizability of these results. First, the test-retest reliability and validity were only examined during treadmill walking at a constant speed. This is not a natural situation in which there is variation in walking speed and activities other than walking are also performed. Second, to translate these results into clinical practice, it should be taken into account that only healthy volunteers participated in this study. The walking pattern of elderly people or those with limited physical abilities may be different. A similar study with the specific target group should be performed to provide actual proof. Despite these limitations, this study demonstrates that most activity trackers perform better as the walking speed increases. In addition, the trackers in this study were relatively new and, to our knowledge, most of them have not been studied before. Therefore, this study provides an initial insight into the test-retest reliability and validity of the activity trackers which can be useful when selecting an activity tracker for a certain target group with a specific level of daily physical activity. For people with a low walking speed, the Garmin Vivosmart or the Fitbit Charge HR seem most suitable. For people with a higher walking speed, the smartwatches (especially the Apple Watch Sport and the Pebble Smartwatch) seem more appropriate. Not only because these trackers were most valid at

52 Reliability and validity of ten consumer activity trackers depend on walking speed

In our study, the placement of the trackers was always at the same position on the In conclusion, there is variation in test-retest reliability and validity of consumer hip, wrist or ankle for each participant during each session. Wrist trackers were also activity trackers which largely depends on walking speed. At a slower walking speed, the 3 consistently placed in the same order from distal to proximal. It should be noted there is an Garmin Vivosmart and Fitbit Charge HR showed the most favorable results. For average open debate about possible differences in accelerations between the dominant and non- walking speed, the Garmin Vivosmart and Apple Watch Sport indicated the best results. For dominant wrist. One study showed that trackers at the non-dominant wrist have a higher vigorous walking, the Apple Watch Sport, Pebble Smartwatch, and Samsung Gear S validity than trackers at the dominant wrist.18 Another study showed no differences in demonstrated the most favorable results. In general, it can be concluded that the activity validity between the dominant and non-dominant wrist.19 These contradicting results trackers are more valid at higher walking speeds, and smartwatches showed slightly better indicate that in a (semi) free-living setting, a counterbalanced placement would have been results than the wearables that are solely developed for activity tracking. more appropriate to exclude possible different acceleration values between the dominant and non-dominant wrist. In our study participants were tested in a controlled lab setting.

53 Chapter 3

References

1. Abellan Van Kan G, Rolland Y, Andrieu S et al. Gait speed at usual pace as a predictor of adverse outcomes in community-dwelling older people An International Academy on Nutrition and Aging (IANA) Task Force. J Nutr Health Aging. 2009;13(10):881-9. 2. Case MA, Burwick HA, Volpp KG, Patel MS. Accuracy of smartphone applications and wearable devices for tracking physical activity data. JAMA. 2015; 313(6):625-6. 3. Dieu O, Mikulovic J, Fardy PS, Bui-Xuan G, Béghin L, Vanhelst J. Physical activity using wrist-worn accelerometers: comparison of dominant and non-dominant wrist. Clin Physiol Funct Imaging. 2016. 4. Evenson KR, Goto MM, Furberg RD. Systematic review of the validity and reliability of consumer- wearable activity trackers. Int J Behav Nutr Phys Act. 2015;12(1):159. 5. Feito Y, Bassett DR, Thompson DL. Evaluation of activity monitors in controlled and free-living environments. Med Sci Sports Exerc. 2012;44(4):733-41. 6. Feito Y, Garner HR, Bassett DR. Evaluation of ActiGraph's low-frequency filter in laboratory and free- living environments. Med Sci Sports Exerc. 2015;47(1):221-7. 7. Giraudeau B. Negative values of the intraclass correlation coefficient are not theoretically possible. J Clin Epidemiol. 1996;49(10):1205. 8. Gjoreski M, Gjoreski H, Luštrek M, Gams M. How accurately can your wrist device recognize daily activities and detect falls? Sensors. 2016;16(6):800. 9. Haskell WL, Lee IM, Pate RR, 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. 10. Hasson RE, Haller J, Pober DM, Staudenmayer J, Freedson PS. Validity of the Omron HJ-112 Pedometer during Treadmill Walking. Med Sci Sports Exerc. 2009;41(4):805-9. 11. Hildebrand M, Van Hees VT, Hansen BH, Ekelund U. Age-group comparability of raw accelerometer output from wrist-and hip-worn monitors. Med Sci Sports Exerc. 2014;46(9):1816-24. 12. Holm S. A simple sequentially rejective multiple test procedure. Scand J Statist. 1979;6(2):65-70. 13. Kooiman TJM, Dontje ML, Sprenger SR, Krijnen WP, van der Schans, CP, de Groot M. Reliability and validity of ten consumer activity trackers. BMC Sports Sci Med Rehabil. 2015; 7:24. 14. Middleton A, Fritz SL, Lusardi M. Walking speed: the functional vital sign. J Aging Phys Act. 2015;23(2):314-22. 15. Musto A, Jacobs K, Nash M, DelRossi G, Perry A. The effects of an incremental approach to 10,000 steps/day on metabolic syndrome components in sedentary overweight women. J Phys Act Health. 2010;7(6):737. 16. Portney L, Watkins M. Foundations of clinical research: applications to practice. Upper Saddle River, NJ: Pearson/Prentice Hall; 2009; Chapter 5. 17. Ryan CG, Grant PM, Tigbe WW, Granat MH. The validity and reliability of a novel activity monitor as a measure of walking. Br J Sports Med. 2006;40(9):779-84. 18. Takacs J, Pollock CL, Guenther JR, Bahar M, Napier C, Hunt MA. Validation of the Fitbit One activity mo nitor device during treadmill walking. J Sci Med Sport. 2014; 17:496-500. 19. Tudor-Locke C, Ainsworth BE, Whitt MC, Thompson RW, Addy CL, Jones DA. The relationship between pedometer-determined ambulatory activity and body composition variables. Int J Obes Relat Metab Di sord. 2001;25(11):1571-8. 20. Tudor-Locke C, Craig CL, Brown WJ, et al. How many steps/day are enough? For adults. Int J Behav Nut r Phys Act. 2011;8:79. 21. Tudor-Locke C, Barreira TV, Schuna Jr JM. Comparison of step outputs for waist and wrist acceleromet er attachment sites. Med Sci Sports Exerc. 2015;47(4):839-42. 22. Tudor-Locke C, Craig CL, Thyfault JP, Spence JC. A step-defined sedentary lifestyle index:< 5000 steps/d ay. Appl Physiol Nutr Metab. 2012;38(2):100-14. 23. Weir JP. Quantifying test-retest reliability using the intraclass correlation coefficient and the SEM. J Strength Cond Res. 2005;19(1):231-40.

54 Reliability and validity of ten consumer activity trackers depend on walking speed

References

1. Abellan Van Kan G, Rolland Y, Andrieu S et al. Gait speed at usual pace as a predictor of adverse outcomes in community-dwelling older people An International Academy on Nutrition and Aging (IANA) Task Force. J Nutr Health Aging. 2009;13(10):881-9. 2. Case MA, Burwick HA, Volpp KG, Patel MS. Accuracy of smartphone applications and wearable devices for tracking physical activity data. JAMA. 2015; 313(6):625-6. 3. Dieu O, Mikulovic J, Fardy PS, Bui-Xuan G, Béghin L, Vanhelst J. Physical activity using wrist-worn accelerometers: comparison of dominant and non-dominant wrist. Clin Physiol Funct Imaging. 2016. 3 4. Evenson KR, Goto MM, Furberg RD. Systematic review of the validity and reliability of consumer- wearable activity trackers. Int J Behav Nutr Phys Act. 2015;12(1):159. 5. Feito Y, Bassett DR, Thompson DL. Evaluation of activity monitors in controlled and free-living environments. Med Sci Sports Exerc. 2012;44(4):733-41. 6. Feito Y, Garner HR, Bassett DR. Evaluation of ActiGraph's low-frequency filter in laboratory and free- living environments. Med Sci Sports Exerc. 2015;47(1):221-7. 7. Giraudeau B. Negative values of the intraclass correlation coefficient are not theoretically possible. J Clin Epidemiol. 1996;49(10):1205. 8. Gjoreski M, Gjoreski H, Luštrek M, Gams M. How accurately can your wrist device recognize daily activities and detect falls? Sensors. 2016;16(6):800. 9. Haskell WL, Lee IM, Pate RR, 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. 10. Hasson RE, Haller J, Pober DM, Staudenmayer J, Freedson PS. Validity of the Omron HJ-112 Pedometer during Treadmill Walking. Med Sci Sports Exerc. 2009;41(4):805-9. 11. Hildebrand M, Van Hees VT, Hansen BH, Ekelund U. Age-group comparability of raw accelerometer output from wrist-and hip-worn monitors. Med Sci Sports Exerc. 2014;46(9):1816-24. 12. Holm S. A simple sequentially rejective multiple test procedure. Scand J Statist. 1979;6(2):65-70. 13. Kooiman TJM, Dontje ML, Sprenger SR, Krijnen WP, van der Schans, CP, de Groot M. Reliability and validity of ten consumer activity trackers. BMC Sports Sci Med Rehabil. 2015; 7:24. 14. Middleton A, Fritz SL, Lusardi M. Walking speed: the functional vital sign. J Aging Phys Act. 2015;23(2):314-22. 15. Musto A, Jacobs K, Nash M, DelRossi G, Perry A. The effects of an incremental approach to 10,000 steps/day on metabolic syndrome components in sedentary overweight women. J Phys Act Health. 2010;7(6):737. 16. Portney L, Watkins M. Foundations of clinical research: applications to practice. Upper Saddle River, NJ: Pearson/Prentice Hall; 2009; Chapter 5. 17. Ryan CG, Grant PM, Tigbe WW, Granat MH. The validity and reliability of a novel activity monitor as a measure of walking. Br J Sports Med. 2006;40(9):779-84. 18. Takacs J, Pollock CL, Guenther JR, Bahar M, Napier C, Hunt MA. Validation of the Fitbit One activity mo nitor device during treadmill walking. J Sci Med Sport. 2014; 17:496-500. 19. Tudor-Locke C, Ainsworth BE, Whitt MC, Thompson RW, Addy CL, Jones DA. The relationship between pedometer-determined ambulatory activity and body composition variables. Int J Obes Relat Metab Di sord. 2001;25(11):1571-8. 20. Tudor-Locke C, Craig CL, Brown WJ, et al. How many steps/day are enough? For adults. Int J Behav Nut r Phys Act. 2011;8:79. 21. Tudor-Locke C, Barreira TV, Schuna Jr JM. Comparison of step outputs for waist and wrist acceleromet er attachment sites. Med Sci Sports Exerc. 2015;47(4):839-42. 22. Tudor-Locke C, Craig CL, Thyfault JP, Spence JC. A step-defined sedentary lifestyle index:< 5000 steps/d ay. Appl Physiol Nutr Metab. 2012;38(2):100-14. 23. Weir JP. Quantifying test-retest reliability using the intraclass correlation coefficient and the SEM. J Strength Cond Res. 2005;19(1):231-40.

Chapter 4 | Behavioral determinants for the adoption of self-tracking devices by adults – a longitudinal study

Thea J.M. Kooiman Arie Dijkstra Justin Timmer Wim P. Krijnen Adriaan Kooy Cees P. van der Schans Martijn de Groot

Submitted

Chapter 4 | Behavioral determinants for the adoption of self-tracking devices by adults – a longitudinal study

Thea J.M. Kooiman Arie Dijkstra Justin Timmer Wim P. Krijnen Adriaan Kooy Cees P. van der Schans Martijn de Groot

Submitted

Chapter 4

Abstract Introduction

Background Consumer based self-tracking devices such as activity trackers, sleep trackers, smart Consumer based self-tracking devices may be used as effective tools in health enhancing bodyweight scales, glucose monitors, and heart rate monitors have become increasingly programs. However, not much is known about behavioral determinants for the adoption of popular over the past several years.1–3 These devices may support self-regulative health these devices and whether they differ among self-tracking functions. behavior which is crucial for the maintenance of health and prevention of lifestyle related diseases.4,5 For instance, self-tracking of physical activity improves physical activity behavior Purpose and regular self-weighing has a positive impact on weight loss for individuals who are This study aimed to identify behavioral determinants for the adoption of an activity and overweight.6,7 The adoption or sustained use of such devices, therefore, is essential for their sleep tracker and a weight scale. usefulness and clinical relevance.

Methods Several studies have been conducted regarding how well self-tracking devices are 8–15 Healthy adults (N=95) received two devices for self-tracking of activity, sleep, and weight. being used and which factors impact long-term adoption. An extensive recent study After six months, behavioral factors, among which included self-regulation capacity, were found that the average days of use of an activity tracker was 129 days with 50% of all of the assessed for the adoption of the activity, sleep, and weight tracking function (i.e., the participants no longer using their device within six months.15 A few small and short term number of days that activity and sleep was measured, and number of self-weighing’s) using follow-up studies found that the percentage of people who stopped using their device was Poisson regression analysis. 59%, 62%, and 75% respectively.10,12,13 Another study found that only 14% of the study population still actively used their activity tracking app after two weeks.16 Adoption also Results varies among the type of monitoring function, e.g., in the study of Kim et al, the adoption of Usage of the activity and sleep tracking function declined over time, whereas number of self- activity monitoring was twice as high compared to sleep monitoring; 44 and 22 days, weighing’s stabilized over time. One subscale of self-regulation, i.e., goal orientation, respectively, within a 90-day study period. Diet was monitored for an average of 17 days.9 contributed to number of days that activity or sleep was measured. Other determinants for use of the activity function were activity level and intention to change physical activity. Most People may have several reasons to use self-tracking devices. Some people use new important determinants for use of the sleep function were motive for self-tracking, and devices simply out of curiosity about new technology, which is usually quickly satisfied.17 social norm for sleep tracking. For weight tracking, most important determinants were BMI, Other people use devices to become more active, which indeed has been found to be a intention to monitor weight, and motive for self-tracking. significant factor in the sustained use of activity trackers.11,16 This is in accordance with the research of Rooksby et al who propose different motives, such as a directive motive (using Conclusions technology to improve health), for engaging in the self-tracking of health.18 Age has also Behavioral determinants are related differently to the usage of various self-tracking been found as predictor for sustained use of an activity tracker, with people with a higher functions. Goal orientation as sub domain of self-regulation is an important factor when age showing a longer duration of use.15 This may also be well explained by different motives using self-tracking devices within health care. to engage in self-tracking; i.e., it has been found that that young individuals use activity trackers primarily for fitness optimization while the older population uses them mainly for improving overall health and extending their lifespan.19

Several reasons have also been determined as to why people discontinue using a self- tracking device. First, an important reason is the failure of a device due to technical problems such as limited battery functioning.13–15 Second, a recurring explanation in the literature is that a device (no longer) meets the expectations of the user.9–11,20–23 For example, users noticed that a certain activity could not be registered by the device,11 the purchased data was perceived as inaccurate,13,14 or expectations about the design10 or esthetics10,23 (especially for females), time investments,10, ease of use,9,20–22 or perceived usefulness9,20–22 were not satisfied. Third, certain behavioral conditions to keep using a

58 Behavioral determinants for the adoption of self-tracking devices by adults

Abstract Introduction

Methods Several studies have been conducted regarding how well self-tracking devices are 4 8–15 Healthy adults (N=95) received two devices for self-tracking of activity, sleep, and weight. being used and which factors impact long-term adoption. An extensive recent study After six months, behavioral factors, among which included self-regulation capacity, were found that the average days of use of an activity tracker was 129 days with 50% of all of the assessed for the adoption of the activity, sleep, and weight tracking function (i.e., the participants no longer using their device within six months.15 A few small and short term number of days that activity and sleep was measured, and number of self-weighing’s) using follow-up studies found that the percentage of people who stopped using their device was Poisson regression analysis. 59%, 62%, and 75% respectively.10,12,13 Another study found that only 14% of the study population still actively used their activity tracking app after two weeks.16 Adoption also Results varies among the type of monitoring function, e.g., in the study of Kim et al, the adoption of Usage of the activity and sleep tracking function declined over time, whereas number of self- activity monitoring was twice as high compared to sleep monitoring; 44 and 22 days, weighing’s stabilized over time. One subscale of self-regulation, i.e., goal orientation, respectively, within a 90-day study period. Diet was monitored for an average of 17 days.9 contributed to number of days that activity or sleep was measured. Other determinants for use of the activity function were activity level and intention to change physical activity. Most People may have several reasons to use self-tracking devices. Some people use new important determinants for use of the sleep function were motive for self-tracking, and devices simply out of curiosity about new technology, which is usually quickly satisfied.17 social norm for sleep tracking. For weight tracking, most important determinants were BMI, Other people use devices to become more active, which indeed has been found to be a intention to monitor weight, and motive for self-tracking. significant factor in the sustained use of activity trackers.11,16 This is in accordance with the research of Rooksby et al who propose different motives, such as a directive motive (using Conclusions technology to improve health), for engaging in the self-tracking of health.18 Age has also Behavioral determinants are related differently to the usage of various self-tracking been found as predictor for sustained use of an activity tracker, with people with a higher functions. Goal orientation as sub domain of self-regulation is an important factor when age showing a longer duration of use.15 This may also be well explained by different motives using self-tracking devices within health care. to engage in self-tracking; i.e., it has been found that that young individuals use activity trackers primarily for fitness optimization while the older population uses them mainly for improving overall health and extending their lifespan.19

59 Chapter 4

device were not always present, such as making usage a habit,10,13,14 or social factors related functions. After six months, the adoption of the devices will be evaluated in terms of days of to acceptance or satisfaction.10,11,17,22 Fourth, personal reasons were indicated to stop using use (activity and sleep) or number of times of use (weight). The primary aim of this study is a device such as having achieved personal goals or having learned enough, people were to identify behavioral determinants for the adoption of an activity and sleep tracker and upgrading to newer models, or changes in priorities.16,23 weight scale.

Most of the above-mentioned reasons to use or stop using a self-tracking device can be interpreted by behavioral theories.24–26 However, only a small number of studies have Methods incorporated a theoretical framework in their research design to investigate which known behavioral factors impact the adoption of self-tracking devices. The use of a theoretical framework is needed in order to gain a broader understanding of potential determinants for Study design their use. The behavioral concept of interest in this study is self-regulation of health This is a six-month study using a within subjects’ design within the Lifelines Cohort Study. behavior. For this purpose, the Temporal Self-regulation Theory (TST) will be used.26,27 The Lifelines is a multi-disciplinary prospective population-based cohort study examining in a TST is a model for specific individual health behavior that states that the intention to unique three-generation design the health and health-related behaviors of 167,729 persons perform a certain behavior (e.g., using a self-tracking device) is influenced by connectedness living in the North of The Netherlands. Healthy adults (aged ≥25 years) were provided with beliefs and temporal valuations. The latter represent a perceived time gap between the an activity tracker and a digital weight scale. The participants were instructed to install the costs and benefits of a certain behavior. Connectedness beliefs represent motivational devices as soon as possible after receiving them. During the study period, the participants factors such as interest in using new technology. These beliefs fit with the above described were free to use the devices as much as desired. No instructions concerning frequency of reasons for using a device. Whether the intention for using self-tracking technology then usage were provided in order to investigate the natural course of the use of the devices. actually leads to a sustained use of the device depends on behavioral prepotency and self- regulation capacity. Behavioral prepotency refers to the probability that the usual behavior occurs, for example, habitual or social norm behavior. The described reasons to stop using a Participants device fit within this construct. Prepotent behavior can be overruled by self-regulative Participants from the Lifelines Cohort Study were recruited by e-mail after pre-selection behavior. Self-regulation capacity describes an individual’s ability to set, implement, and based on age and postal number for logistical reasons. Potential participants subsequently monitor goals in order to successfully regulate their own behavior.26–28 It is known that the received a small questionnaire concerning inclusion and exclusion questions. Inclusion monitoring of behavior and feedback on it are both essential to detect goal attainment and, criteria were being ≥25 years and access to a smartphone with internet (IOS or Android). thereafter, for regulation of an individual’s behavior, i.e., to continue to strive for one’s Participants were excluded if they were already in the possession of an activity monitor or goals.24,28,29 Therefore, the use of self-tracking technology fits within the principles of self- smart weight scale or were not able to engage in self-tracking of activity, sleep, or weight regulation because these devices allow for goal-setting, self-monitoring of behavior, and due to physical, social, cognitive- and/or mental problems. Included participants received an provide the user with personalized feedback. Although people have already been self- invitation to collect their devices at the research office of Lifelines. Informed consent was monitoring their health for a long time, e.g., by keeping a diary for their sleep quality or obtained from all of the participants. Ethical approval was granted within the Lifelines coffee intake, the rise of consumer self-tracking technology has made self-monitoring much Cohort Study by the University Medical Center Groningen (METc 2007/152) based on the easier. In the context of the use of self-tracking devices, it is important to assess whether declaration of Helsinki of Ethical Principles for Medical Research Involving Human Subjects. self-regulation capacity for healthy behavior is related to the adoption of such technology. Self-tracking devices Altogether, the adoption of self-tracking devices variably depends on motivational, The Nokia Pulse, (Nokia, Nozay, France, previously Withings, Issy les Moulineaux, France) personal, or device related factors. However, it is unknown whether self-regulation capacity measures activity and sleep. The Nokia WS-30 (Nokia, Nozay, France), measures weight and is (directly) related to the use of self-tracking devices. Also, it is unknown whether self- body mass index (BMI). The devices were connected with a smartphone application (Nokia regulation and other behavioral factors have a different influence on the adoption of Health Mate), which exhibited the course of a participant’s activity pattern, sleeping pattern, different types of tracking functions such as physical activity and weight tracking. In this six- and body weight over time. In addition, the application provided automated personalized month longitudinal study, we will follow healthy adults who receive two devices for self- feedback messages concerning progression towards the self-selected goals of the tracking of health behavior. At the start of the study they complete a TST-based participants. Participants had the possibility to engage with friends within the app. In case of questionnaire with possible determinants for the adoption of different self-tracking

60 Behavioral determinants for the adoption of self-tracking devices by adults

Most of the above-mentioned reasons to use or stop using a self-tracking device can be interpreted by behavioral theories.24–26 However, only a small number of studies have Methods incorporated a theoretical framework in their research design to investigate which known behavioral factors impact the adoption of self-tracking devices. The use of a theoretical framework is needed in order to gain a broader understanding of potential determinants for Study design their use. The behavioral concept of interest in this study is self-regulation of health This is a six-month study using a within subjects’ design within the Lifelines Cohort Study. behavior. For this purpose, the Temporal Self-regulation Theory (TST) will be used.26,27 The Lifelines is a multi-disciplinary prospective population-based cohort study examining in a 4 TST is a model for specific individual health behavior that states that the intention to unique three-generation design the health and health-related behaviors of 167,729 persons perform a certain behavior (e.g., using a self-tracking device) is influenced by connectedness living in the North of The Netherlands. Healthy adults (aged ≥25 years) were provided with beliefs and temporal valuations. The latter represent a perceived time gap between the an activity tracker and a digital weight scale. The participants were instructed to install the costs and benefits of a certain behavior. Connectedness beliefs represent motivational devices as soon as possible after receiving them. During the study period, the participants factors such as interest in using new technology. These beliefs fit with the above described were free to use the devices as much as desired. No instructions concerning frequency of reasons for using a device. Whether the intention for using self-tracking technology then usage were provided in order to investigate the natural course of the use of the devices. actually leads to a sustained use of the device depends on behavioral prepotency and self- regulation capacity. Behavioral prepotency refers to the probability that the usual behavior occurs, for example, habitual or social norm behavior. The described reasons to stop using a Participants device fit within this construct. Prepotent behavior can be overruled by self-regulative Participants from the Lifelines Cohort Study were recruited by e-mail after pre-selection behavior. Self-regulation capacity describes an individual’s ability to set, implement, and based on age and postal number for logistical reasons. Potential participants subsequently monitor goals in order to successfully regulate their own behavior.26–28 It is known that the received a small questionnaire concerning inclusion and exclusion questions. Inclusion monitoring of behavior and feedback on it are both essential to detect goal attainment and, criteria were being ≥25 years and access to a smartphone with internet (IOS or Android). thereafter, for regulation of an individual’s behavior, i.e., to continue to strive for one’s Participants were excluded if they were already in the possession of an activity monitor or goals.24,28,29 Therefore, the use of self-tracking technology fits within the principles of self- smart weight scale or were not able to engage in self-tracking of activity, sleep, or weight regulation because these devices allow for goal-setting, self-monitoring of behavior, and due to physical, social, cognitive- and/or mental problems. Included participants received an provide the user with personalized feedback. Although people have already been self- invitation to collect their devices at the research office of Lifelines. Informed consent was monitoring their health for a long time, e.g., by keeping a diary for their sleep quality or obtained from all of the participants. Ethical approval was granted within the Lifelines coffee intake, the rise of consumer self-tracking technology has made self-monitoring much Cohort Study by the University Medical Center Groningen (METc 2007/152) based on the easier. In the context of the use of self-tracking devices, it is important to assess whether declaration of Helsinki of Ethical Principles for Medical Research Involving Human Subjects. self-regulation capacity for healthy behavior is related to the adoption of such technology. Self-tracking devices Altogether, the adoption of self-tracking devices variably depends on motivational, The Nokia Pulse, (Nokia, Nozay, France, previously Withings, Issy les Moulineaux, France) personal, or device related factors. However, it is unknown whether self-regulation capacity measures activity and sleep. The Nokia WS-30 (Nokia, Nozay, France), measures weight and is (directly) related to the use of self-tracking devices. Also, it is unknown whether self- body mass index (BMI). The devices were connected with a smartphone application (Nokia regulation and other behavioral factors have a different influence on the adoption of Health Mate), which exhibited the course of a participant’s activity pattern, sleeping pattern, different types of tracking functions such as physical activity and weight tracking. In this six- and body weight over time. In addition, the application provided automated personalized month longitudinal study, we will follow healthy adults who receive two devices for self- feedback messages concerning progression towards the self-selected goals of the tracking of health behavior. At the start of the study they complete a TST-based participants. Participants had the possibility to engage with friends within the app. In case of questionnaire with possible determinants for the adoption of different self-tracking

61 Chapter 4

loss or technical failure the devices could be replaced instantly at the research office during resulted in a lower alpha. the whole study period. Physical activity level was measured with the Nokia Pulse. For each participant, the average steps per day were calculated from all of the available measurement days within the study

period, and subsequently categorized in sedentary (<5000 steps/d), somewhat active Measures (≥5000-7500 steps/d), active (≥7500-10000 steps/d), or very active (≥10.000 steps/d).32 This The primary outcome variable was the duration of usage of the devices that were separately variable was only assessed for adoption of the activity tracking function. measured for the three tracking functions: activity, sleep, and weight.

Adoption of the physical activity function was measured by counting the total number of Behavioral factors - specific self-tracking factors days the Pulse was worn within the study period. For each participant, the 180-day study Connectedness beliefs were assessed by measuring attitude (six items) and self-efficacy (four period began on the second day after the Pulse was received. One day of use was only items) towards self-tracking of activity, sleep, and body weight on a 5-point Likert scale.25,31 included if ≥500 steps were measured in order to exclude test or accidental measurements. For example, for ‘Attitude’, participants were asked “What is your opinion about measuring Adoption of the sleep function was measured as the total number of days the Pulse was used your steps regularly?” with answers ranging from ‘very useless’ to ‘very useful’. Questions as a sleep monitor. A cut-off point of ≤ 1 hour of sleep per day was used to exclude test or were summed and divided by the number of items. accidental measurements. Adoption of the weight scale was measured as the total number Motive for self-tracking was assessed as a nominal variable by a 1-item questionnaire in of times body weight was measured beginning on day the weight scale was received. In which the participant could indicate one out of five motives as suggested by Rooksby et al addition, adoption was classified in different weighing frequencies: monthly (self-weighing (2014); documentary (to gain information about one’s health), diagnostic (to explain one’s less than once a week), weekly (self-weighing at a minimum of once per week), or daily (self- subjective health), directive (to improve health), collecting rewards (to receive positive weighing at a minimum of six days per week).7,30 feedback about one’s health behavior ), and fetishized (interested in new gadgets).18

Intention to engage in self-tracking was separately assessed by a 1-item statement on a 5- The behavioral factors for adoption were assessed at baseline by using a digital point Likert scale for the three self-tracking behaviors.25,31 For example, the intention to self- questionnaire. These factors were subdivided into specific variables for the outcome of monitor steps was measured by; “I intend to measure my steps regularly within one month”. healthy behavior itself and those for engagement in self-tracking behavior, for the three A score of 1 indicated no intention at all, and a score of 5 indicated a definite intention to do tracking functions separately. so.

Behavioral prepotency was measured separately with a questionnaire about habit (six items) Behavioral factors - healthy behavior factors and social norm (three items) for self-tracking of activity, sleep, and weight. Habit questions Intention to alter physical activity, sleep, and body weight was measured using three 1-item were translated from the validated Self Report Habit Index.33 The social norm questions questionnaires.25,31 The participant could indicate 1) the intention to move/sleep more or were based on Boudreaux et al (2014). gain weight, 2) no intention to change, or 3) the intention to move/sleep or weigh less.

Self-regulation towards health was measured with the self-regulation questionnaire from Personal factors Brown et al (1999). This questionnaire was translated and adapted to increase specificity for Age, gender, education, and BMI were measured to be included as covariates. Age, gender, self-regulation towards health behavior (physical activity, sleep, food intake, and body and education were assessed in a questionnaire. Education was classified as low (finished weight). Participants could indicate their extent of agreement on the different items on a 5- primary education or preparatory secondary vocational education), moderate (finished point Likert scale ranging from strongly disagree (score 1) to strongly agree (score 5). The secondary education or intermediate vocational education), or high (college education or mean scores on the different subscales were calculated, according to the factor structure of higher). BMI was assessed using the height and first self-measurement on the weight scale, Gavora et al (2015). These subscales were goal orientation (e.g., “I can stick to a health plan and subsequently categorized in <25, 25-30, or >30. that’s working well”), self-direction (e.g., “I don’t seem to learn from my unhealthy behavior”), decision-making (e.g., “As soon as I see a problem or challenge, I start looking for possible solutions”), and impulse control (e.g., “I get easily distracted from my plans”). Statistical analyses Cronbach’s alpha was .69, .74, .66, and .83 respectively for goal orientation (five items), self- Descriptive analyses and ANOVA repeated measures analyses were used first to examine the direction (seven items), decision-making (seven items), and impulse-control (eight items). overall using patterns of the tracking functions over the six-month period. The number of The item ‘I am set in my ways’ was deleted from the goal orientation scale, because this item self-measurements per month were extracted separately from the dataset for the three

62 Behavioral determinants for the adoption of self-tracking devices by adults

Adoption of the physical activity function was measured by counting the total number of Behavioral factors - specific self-tracking factors days the Pulse was worn within the study period. For each participant, the 180-day study Connectedness beliefs were assessed by measuring attitude (six items) and self-efficacy (four period began on the second day after the Pulse was received. One day of use was only items) towards self-tracking of activity, sleep, and body weight on a 5-point Likert scale.25,31 included if ≥500 steps were measured in order to exclude test or accidental measurements. For example, for ‘Attitude’, participants were asked “What is your opinion about measuring Adoption of the sleep function was measured as the total number of days the Pulse was used your steps regularly?” with answers ranging from ‘very useless’ to ‘very useful’. Questions 4 as a sleep monitor. A cut-off point of ≤ 1 hour of sleep per day was used to exclude test or were summed and divided by the number of items. accidental measurements. Adoption of the weight scale was measured as the total number Motive for self-tracking was assessed as a nominal variable by a 1-item questionnaire in of times body weight was measured beginning on day the weight scale was received. In which the participant could indicate one out of five motives as suggested by Rooksby et al addition, adoption was classified in different weighing frequencies: monthly (self-weighing (2014); documentary (to gain information about one’s health), diagnostic (to explain one’s less than once a week), weekly (self-weighing at a minimum of once per week), or daily (self- subjective health), directive (to improve health), collecting rewards (to receive positive weighing at a minimum of six days per week).7,30 feedback about one’s health behavior ), and fetishized (interested in new gadgets).18

63 Chapter 4

tracking functions. Because of a non-normal distribution of the dependent variables (i.e., the total number of days/times that activity, sleep, or weight was measured) and because the (dependent) outcome data is of the `count’ type, a Poisson regression analysis was conducted separately for each self-tracking function (activity, sleep, and weight tracking).

The dependent variable was the total number of measurements counted during the study period. The independent variables were added in the analysis as either categorical (age class, BMI class, gender, education, physical activity level, motive for self-tracking, and intention to change physical activity, sleep, or weight) or as continuous variables (the self-regulation subscales of goal orientation, self-direction, decision making, and impulse control; the intention to engage in self-tracking of activity, sleep, or weight; and attitude, self-efficacy, social norm, and habit for self-tracking of activity, sleep, and weight). First, descriptive statistics were used to examine for patterns and balance of the data. Also, assumptions to conduct Poisson Regression were examined. Next, to identify significant predictors, univariate Poisson analyses were conducted separately for each of the predictive variables. After this, all of the identified predictors were analysed with a multivariate Poisson regression. Subsequently, manual back fitting was used to determine a model in which all variables contribute significantly and for at least 15% to the outcome (i.e., all variables with an odds ratio between 0.85 and 1.15 were removed). This percentage was selected in order to be able to detect not only significant but also relevant determinants; a contribution of a variable of less than 15% for the number of measurements was considered to be less relevant. Figure 1. Flow of participants through the study. Analyses were conducted using SPSS for Windows (version 22, 2010, IBM-SPSS Inc). A cutoff value of α<.05 was utilized to assess statistical significance.

Table 1. Baseline characteristics of the study population (N=84). Results Mean ± SD

In total, 95 participants of the Lifelines cohort study signed for informed consent and Age 48.3 ± 6.8 received the self-tracking devices. The Pulse data became available for 84 participants and Gender the weight data was available from 81 participants within the study period. Therefore, 11 Male 34.5 % participants were assigned as ‘inclusion failure’, resulting in a study population of 84 Female 65.5 % participants. Figure 1 describes the reasons for the inclusion failures. Table 1 shows baseline Weight (kg) 78.9 ± 14.9 characteristics of the study population. At baseline, 56% of the study population intended to BMI 26.0 ± 3.7 increase physical activity, and 22.6% wanted to increase sleeping time. No participants <25 45.2 % indicated that they would like to decrease physical activity or sleep. With regard to 25-30 39.3 % intentions to change weight, 56% of the study population wanted to lose weight at baseline, >30 15.5 % 29.8% did not want to change weight, and 14.3% wanted to gain weight. Education Lower education 10.7 %

Medium education 34.5 % Higher education 54.8 %

64 Behavioral determinants for the adoption of self-tracking devices by adults

The dependent variable was the total number of measurements counted during the study period. The independent variables were added in the analysis as either categorical (age class, BMI class, gender, education, physical activity level, motive for self-tracking, and intention to change physical activity, sleep, or weight) or as continuous variables (the self-regulation subscales of goal orientation, self-direction, decision making, and impulse control; the intention to engage in self-tracking of activity, sleep, or weight; and attitude, self-efficacy, social norm, and habit for self-tracking of activity, sleep, and weight). First, descriptive statistics were used to examine for patterns and balance of the data. Also, assumptions to 4 conduct Poisson Regression were examined. Next, to identify significant predictors, univariate Poisson analyses were conducted separately for each of the predictive variables. After this, all of the identified predictors were analysed with a multivariate Poisson regression. Subsequently, manual back fitting was used to determine a model in which all variables contribute significantly and for at least 15% to the outcome (i.e., all variables with an odds ratio between 0.85 and 1.15 were removed). This percentage was selected in order to be able to detect not only significant but also relevant determinants; a contribution of a variable of less than 15% for the number of measurements was considered to be less relevant. Figure 1. Flow of participants through the study. Analyses were conducted using SPSS for Windows (version 22, 2010, IBM-SPSS Inc). A cutoff value of α<.05 was utilized to assess statistical significance.

Table 1. Baseline characteristics of the study population (N=84). Results Mean ± SD

Medium education 34.5 % Higher education 54.8 %

65 Chapter 4

Motive for self-tracking Data processing Documentary 38.1 % Due to a strong association of intention to engage in a specific self-tracking behavior with Diagnostic 1.2 % attitude and self-efficacy towards self-tracking them (r >.6, p<.01), attitude and self-efficacy Directive 23.8 % were excluded from the analyses to avoid multicollinearity. One variable had unbalanced Positive feedback 19.0 % Fetishized 15.5 % numbers across the categories, i.e., only one participant indicated the motive for self- Other 2.4 % tracking as ‘diagnostic’. Therefore, the results for ‘diagnostic’ are not displayed in the analyses.

Adoption of the self-tracking devices The number of days of use of the step tracking function varied from two to 180 days Determinants of use of the activity tracker (median 148, IQR 90-168 days, N=84). The days of use of the sleep tracking function varied Eight variables were determined to be independently significantly related to the number of from zero to 178 days (median 70, IQR 20-141, N=84). The number of weight measurements activity measurements by a univariate Poisson regression. Table 2 shows all univariate varied from one to 422 weight measurements (median 62, IQR 28-134 measurements, significant determinants with associated odds ratios and the final multivariate model for the N=81). A percentage of 22% of the participants weighed themselves monthly, 62% weekly, use of the activity tracking function. Goal orientation as subscale of self-regulation was and 16% weighed daily. Figure 2.a and 2.b show the adoption of the different self-tracking significant in the final model (OR 1.20 [CI 1.16; 1.25], p=.000). Other significant determinants functions per month over the six-month period. Activity tracking decreased significantly found by the final model were BMI class, physical activity level, and intention to change from an average of 24 ± 9 days in the first month to 17 ± 11 days in the sixth month (F=15.5, activity level (Table 2). p=.000). Sleep tracking dropped from 16 ± 11 days of use to 9 ± 11 days per month (F=29.3, p=.000). The number of weight measurements was also reduced from the first month (on Determinants of use of the sleep tracker average 19 ± 15 self-weighing’s) till the sixth month (F=20.2, p=.000), however, this number Eleven variables were significantly related to the number of sleep measurements in the stabilized from the third until the sixth month with an average of 12 ± 13 self-weighing’s per univariate Poisson regression. Table 3 demonstrates all of the univariate determinants with month. associated odds ratios and the final multivariate model for determinants for the use of the sleep tracking function. All four self-regulation scales were significant in the univariate Poisson regression. In the final model, goal-orientation scale remained significant (OR 1.29 [CI 1.23;1.35], p=.000). Five other determinants remained in the final model; age class, BMI class, education, motive for self-tracking, and social norm for self-tracking of sleep (Table 3).

Figure 2. Number of days of activity [circles] and sleep [triangles] measurements (A, N=84), and number of weight measurements per month (B, N=81).

66 Behavioral determinants for the adoption of self-tracking devices by adults

Adoption of the self-tracking devices The number of days of use of the step tracking function varied from two to 180 days Determinants of use of the activity tracker (median 148, IQR 90-168 days, N=84). The days of use of the sleep tracking function varied Eight variables were determined to be independently significantly related to the number of from zero to 178 days (median 70, IQR 20-141, N=84). The number of weight measurements activity measurements by a univariate Poisson regression. Table 2 shows all univariate 4 varied from one to 422 weight measurements (median 62, IQR 28-134 measurements, significant determinants with associated odds ratios and the final multivariate model for the N=81). A percentage of 22% of the participants weighed themselves monthly, 62% weekly, use of the activity tracking function. Goal orientation as subscale of self-regulation was and 16% weighed daily. Figure 2.a and 2.b show the adoption of the different self-tracking significant in the final model (OR 1.20 [CI 1.16; 1.25], p=.000). Other significant determinants functions per month over the six-month period. Activity tracking decreased significantly found by the final model were BMI class, physical activity level, and intention to change from an average of 24 ± 9 days in the first month to 17 ± 11 days in the sixth month (F=15.5, activity level (Table 2). p=.000). Sleep tracking dropped from 16 ± 11 days of use to 9 ± 11 days per month (F=29.3, p=.000). The number of weight measurements was also reduced from the first month (on Determinants of use of the sleep tracker average 19 ± 15 self-weighing’s) till the sixth month (F=20.2, p=.000), however, this number Eleven variables were significantly related to the number of sleep measurements in the stabilized from the third until the sixth month with an average of 12 ± 13 self-weighing’s per univariate Poisson regression. Table 3 demonstrates all of the univariate determinants with month. associated odds ratios and the final multivariate model for determinants for the use of the sleep tracking function. All four self-regulation scales were significant in the univariate Poisson regression. In the final model, goal-orientation scale remained significant (OR 1.29 [CI 1.23;1.35], p=.000). Five other determinants remained in the final model; age class, BMI class, education, motive for self-tracking, and social norm for self-tracking of sleep (Table 3).

Figure 2. Number of days of activity [circles] and sleep [triangles] measurements (A, N=84), and number of weight measurements per month (B, N=81).

67 Chapter 4

Table 2. Table 3. Significant determinants for use of the activity tracking function (N=84). Significant determinants for use of the sleep tracking function (N=84).

Univariate results OR Confidence Interval p-value Univariate results OR Confidence Interval p-value Lower Upper Lower Upper

Age class BMI class 50-59 years .84 .79 .91 .000 >30 1.05 .99 1.12 .091 40-49 years .88 25-30 1.28 1.23 1.34 .000 .82 .95 .001 < 25 (Ref) 30-39 years (Ref)

Education* BMI class High 1.03 .96 1.09 .443 >30 .99 .93 1.08 .974 Medium .945 .89 1.02 .128 25-30 1.35 1.29 1.43 .000 Low (Ref) <25 Activity class Education Very active 1.48 1.36 1.60 .000 Active 1.19 1.10 1.29 .000 High 1.28 1.18 1.40 .000 Somewhat active 1.20 1.11 1.29 .000 Medium 1.04 .95 1.14 .442 Sedentary (Ref) Low (Ref)

Intention to change activity level Motive for self-tracking Want to increase activity 1.28 1.23 1.33 .000 Fetishized 1.34 1.24 1.44 .000 No intention to change (Ref) Collecting rewards 1.38 1.29 1.48 .000 Directive Motive for self-tracking 1.24 1.16 1.32 .000 Fetishized 1.15 1.08 1.21 .000 Diagnostic N.A. # N.A. # N.A. # N.A. # Collecting rewards 1.13 1.08 1.20 .000 Documentary (Ref) Directive 1.14 1.09 1.20 .000 Diagnostic N.A N.A N.A. N.A. Intention for self-tracking of sleep 1.34 1.28 1.39 .000 Documentary (Ref) SR – goal orientation 1.23 1.18 1.29 .000 SR - goal orientation 1.12 1.08 1.15 .000 SR – self-direction 1.14 1.09 1.19 .000 SR – decision-making 1.41 1.32 1.50 .000 Intention for self-tracking of steps 1.18 1.14 1.23 .000 SR – impulse control 1.05 1.01 1.10 .010 Habit for self-tracking of steps 1.12 1.09 1.15 .000 Social norm for self-tracking of sleep 1.67 1.59 1.75 .000 Multivariate final results OR Confidence Interval p-value Habit for self-tracking of sleep 1.30 1.26 1.34 .000 BMI class >30 1.10 1.04 1.17 .002 Multivariate final results OR Confidence Interval p-value 25-30 1.22 1.17 1.27 .000 < 25 (Ref) Age class 50-59 years .76 .70 .82 .000 Activity level 40-49 years .96 .89 1.04 .355 Very active 1.58 1.46 1.72 .000 Active 1.22 1.13 1.32 .000 30-39 years (Ref) Somewhat active 1.25 1.16 1.35 .000 Sedentary (Ref) BMI class >30 1.03 1.17 1.40 .446 Intention to change activity level 25-30 1.28 1.21 1.36 .000 Want to increase activity 1.32 1.27 1.38 .000 <25 No intention to change (Ref) Education SR – Goal orientation 1.20 1.16 1.25 .000 High 1.33 1.20 1.47 .000 OR= odds ratio SR=self-regulation * education was included in this table as significant univariate predictor Medium 1.16 1.05 1.29 .005 because the main model was significant. Low (Ref)

68 Behavioral determinants for the adoption of self-tracking devices by adults

Table 2. Table 3. Significant determinants for use of the activity tracking function (N=84). Significant determinants for use of the sleep tracking function (N=84).

Univariate results OR Confidence Interval p-value Univariate results OR Confidence Interval p-value Lower Upper Lower Upper

Age class BMI class 50-59 years .84 .79 .91 .000 >30 1.05 .99 1.12 .091 40-49 years .88 25-30 1.28 1.23 1.34 .000 .82 .95 .001 < 25 (Ref) 30-39 years (Ref)

Education* BMI class High 1.03 .96 1.09 .443 >30 .99 .93 1.08 .974 Medium .945 .89 1.02 .128 25-30 1.35 1.29 1.43 .000 Low (Ref) <25 4 Activity class Education Very active 1.48 1.36 1.60 .000 Active 1.19 1.10 1.29 .000 High 1.28 1.18 1.40 .000 Somewhat active 1.20 1.11 1.29 .000 Medium 1.04 .95 1.14 .442 Sedentary (Ref) Low (Ref)

69 Chapter 4

Motive for self-tracking Table 4. Fetishized 1.25 1.16 1.35 .000 Significant determinants for use of the weight tracking function (N=78). Collecting rewards 1.27 1.19 1.36 .000 Directive 1.33 1.24 1.43 .000 Univariate results OR Confidence Interval p-value Diagnostic N.A.# N.A. # N.A. # N.A. # Lower Upper Documentary (Ref)

SR – goal orientation 1.29 1.23 1.35 .000 Age class Social norm for self-tracking of sleep 1.71 1.62 1.80 .000 50-59 years .77 .72 .82 .000 40-49 years .50 .46 .54 .000 OR = odds ratio SR=self-regulation N.A. # = not applicable due to an unbalanced count. 30-39 years (Ref)

Gender Women .75 .71 .79 .000 Men (Ref)

Determinants of use of the weight scale BMI class Three participants were extreme outliers for the number of weight measurements, i.e., they >30 .62 .56 .69 .000 25-30 1.45 1.37 1.53 .000 measured their weight 353, 380, and 422 times compared to 198 times or less for the rest of <25 (Ref) the participants. Therefore, these participants were excluded from the analysis. Ten Education variables were statistically significant in the univariate Poisson regression for the number of High .92 .85 .99 .034 weight measurements. The self-regulation scales ‘impulse-control’ and ‘goal orientation’ Medium .75 .69 .82 .000 were negatively related to the number of weight measurements in the univariate analysis. In Low (Ref) the final multivariate model, both scales became non-significant. Seven determinants Intention to change weight remained in the final multivariate model; age class, gender, BMI class, education, intention Want to gain weight 1.49 1.38 1.61 .000 to change weight, motive for self-tracking, and intention to monitor weight (Table 4). Want to lose weight 1.05 .00 1.12 .105 No intention to change (Ref)

Motive for self-tracking Fetishized .78 .71 .85 .000 Collecting rewards 1.44 1.34 1.54 .000 Directive 1.12 1.05 1.20 .001 Diagnostic N.A. # N.A. # N.A. # N.A. # Documentary (Ref)

Intention for self-tracking of weight 1.31 1.25 1.38 .000

SR – goal orientation .90 .86 .94 .000 SR – impulse control .94 .90 .98 .004

Habit for self-tracking of weight 1.06 1.03 1.10 .001

Multivariate results OR Lower Upper p-value

Age class 50-59 years .71 .66 .77 .000 40-49 years .42 .39 .46 .000 30-39 years (Ref)

Gender Female .66 .61 .70 .000 Male (Ref)

70 Behavioral determinants for the adoption of self-tracking devices by adults

Gender Women .75 .71 .79 .000

Men (Ref) 4 Determinants of use of the weight scale BMI class Three participants were extreme outliers for the number of weight measurements, i.e., they >30 .62 .56 .69 .000 25-30 1.45 1.37 1.53 .000 measured their weight 353, 380, and 422 times compared to 198 times or less for the rest of <25 (Ref) the participants. Therefore, these participants were excluded from the analysis. Ten Education variables were statistically significant in the univariate Poisson regression for the number of High .92 .85 .99 .034 weight measurements. The self-regulation scales ‘impulse-control’ and ‘goal orientation’ Medium .75 .69 .82 .000 were negatively related to the number of weight measurements in the univariate analysis. In Low (Ref) the final multivariate model, both scales became non-significant. Seven determinants Intention to change weight remained in the final multivariate model; age class, gender, BMI class, education, intention Want to gain weight 1.49 1.38 1.61 .000 to change weight, motive for self-tracking, and intention to monitor weight (Table 4). Want to lose weight 1.05 .00 1.12 .105 No intention to change (Ref)

Motive for self-tracking Fetishized .78 .71 .85 .000 Collecting rewards 1.44 1.34 1.54 .000 Directive 1.12 1.05 1.20 .001 Diagnostic N.A. # N.A. # N.A. # N.A. # Documentary (Ref)

Intention for self-tracking of weight 1.31 1.25 1.38 .000

SR – goal orientation .90 .86 .94 .000 SR – impulse control .94 .90 .98 .004

Habit for self-tracking of weight 1.06 1.03 1.10 .001

Multivariate results OR Lower Upper p-value

Age class 50-59 years .71 .66 .77 .000 40-49 years .42 .39 .46 .000 30-39 years (Ref)

Gender Female .66 .61 .70 .000 Male (Ref)

71 Chapter 4

BMI class weight and obtain positive feedback are more likely to monitor their weight more often. This >30 .70 .62 .78 .000 is in accordance with the Feedback Theory which states that positive feedback is crucial for 25-30 1.24 1.17 1.33 .000 34 <25 (Ref) maintenance of motivation and continued goal-striving behavior. Furthermore, intention to increase activity and change weight contributed to the number of days or number of Education High .73 .66 .80 .000 times that activity or weight was measured, however, the intention to increase sleep did not Medium .66 .60 .73 .000 affect number of sleep measurements. This may be explained by fewer people intending to Low (Ref) change sleep (N= 19), but participants may still have been curious about their sleeping Intention to change weight patterns. For weight tracking, we found that only the intention to gain weight contributed to Want to gain weight 1.45 1.33 1.59 .000 the number of weight measurements (N=11). An intention to lose weight did not result in Want to lose weight .92 .86 .99 .107 No intention to change (Ref) more measurements compared to no intention to change weight. With regard to intentions to use technology, the intention to monitor weight contributed as the only variable to the

Motive for self-tracking use of a specific tracking function. This might indicate that, in the case of physical activity Fetishized .90 .82 .99 .032 Collecting rewards 1.40 1.30 1.51 .000 tracking and sleep tracking, other factors overrule or mediate this intention that individuals Directive .95 .88 1.03 .226 had at baseline. Diagnostic N.A. # N.A. # N.A. # N.A. # Documentary (Ref) Out of the four domains of self-regulation, we determined goal orientation to be

Intention for self-tracking of weight 1.27 1.20 1.34 .000 significantly related to the number of activity and sleep measurements. Thus, people who are more focused on realizing their goals and plan how to accomplish them show a higher OR= odds ratio SR = self-regulation N.A.# = not applicable due to an unbalanced count. adoption of a device that quantifies activity and sleep. However, this was not the case for self-tracking of weight. This may be explained by the differences between the outcomes of the devices with weight being a less controllable outcome. For activity and sleep tracking, a Discussion discrepancy reducing loop (i.e., increasing the steps per day until the step goal is achieved or going to bed earlier) may be easier to reach than for weight tracking (i.e., losing weight). This study aimed at identifying behavioral determinants for the adoption of an activity and Also, the larger variety in number of weight measurements that were probably induced by sleep tracker and a weight scale. Overall, there was considerable variation between the personal opinions on how many measurements are needed to successfully monitor weight, tracking functions with regard to the degree of usage (total number of measurements and may explain why goal orientation as well as the intention to lose weight did not contribute number of measurements per month), the using pattern over time, and the factors to the number of weight measurements. associated with their use. The physical activity tracking function was associated with a higher Lastly, we found that BMI and activity level were significant factors in the adoption of usage compared to the sleep tracking function, although the usage of both functions the devices. People with a BMI of >30 engaged in fewer weight measurements, but, on the declined over time. For weight tracking, after an initial decline, self-weighing behavior other hand, engaged in more physical activity measurements compared to people with a stabilised over time with three out of four individuals weighing themselves regularly (i.e., healthy weight. This may be explained in two ways. First, again, the difference in the type of daily or weekly). Our results show that behavioral factors related to motivation (both outcome may explain this finding with weight being a less easily controlled outcome motives and specific intentions), self-regulation, and social support differently explain the compared to activity. Second, this finding may also be explained by the Feedback Theory. adoption of the different self-tracking functions. Our finding is in line with research that showed that people who recently gained weight tend 35 Personal motives are determinants for the adoption of self-tracking of sleep and to decrease their frequency of self-weighing, and another study that showed that self- weight. In sleep tracking, the ‘directive’, ‘collecting rewards’, and ‘fetishized’ motives reported weighing frequency was significantly lower in obese adults compared to 36 contributed to at least 25% more measurements compared to the ‘documentary’ motive. overweight adults. This implies that positive feedback is required in order for people to This indicates that for sleep monitoring, people need a specific motive in order to continue continue with self-weighing, especially for people with obesity. With regard to activity level, with sleep tracking. For weight tracking, ‘positive feedback’ is a predominant motive. From we found that a higher activity level does contribute to the adoption of the activity function. the participants who indicated ‘positive feedback’ as the most important motive, 75% This may also be well explained by the positive feedback that is induced by walking more indicated that they wanted to lose weight. This suggests that people who both want to lose than 10.000 steps/d. This amount is the default step goal in many activity trackers and, when

72 Behavioral determinants for the adoption of self-tracking devices by adults

Motive for self-tracking use of a specific tracking function. This might indicate that, in the case of physical activity 4 Fetishized .90 .82 .99 .032 Collecting rewards 1.40 1.30 1.51 .000 tracking and sleep tracking, other factors overrule or mediate this intention that individuals Directive .95 .88 1.03 .226 had at baseline. Diagnostic N.A. # N.A. # N.A. # N.A. # Documentary (Ref) Out of the four domains of self-regulation, we determined goal orientation to be

73 Chapter 4

reached, the Health Mate application sends positive feedback messages and rewards. This self-weighing is used in weight loss interventions, attempts should be made to help probably reinforces the sustained usage of the activity tracking function. individuals to receive positive feedback instead of negative feedback.

This study has some strengths and limitations. This is the first follow-up study that includes several behavioral determinants, including self-regulation capacity, for the adoption Acknowledgements of an activity and sleep tracker and weight scale. A limitation was that some participants The authors would like to thank Nokia Health (previously Withings) for the donation of the self- experienced technical problems with the activity and sleep tracking device. Technical tracking devices and collaboration during the study. We also thank the Lifelines Cohort Study for failures, the perceived ease of use of a device, and perceived usefulness have earlier been granting access to their study cohort and accompanying support. found to impact the adoption of self-tracking technology.13,20,22 However, both devices could be replaced during the complete study period, therefore, technical failures could have had only limited influence on the adoption of the devices. The study was undertaken in a general population within the Lifelines Cohort with a high variation in age and BMI, which increases generalizability. However, the population of the Lifelines Cohort consists mainly of Caucasian people who live in the Northern part of the Netherlands, which may limit the generalisability to other parts of Europe and other parts of the world. Finally, the sample size of this study was large enough to conduct this exploration, however, attrition in combination with the small size of some subgroups limit the generalizability of the results. Therefore, the conclusions should be read with caution and within the context of these limitations.

In conclusion, the three self-tracking functions physical activity, sleep, and weight, showed different degrees of adoption. Also, behavioral determinants for adoption differ between the different tracking functions. An intention to become more active is an important behavioral determinant for the adoption of an activity tracker. For adoption of a sleep tracker, important behavioral determinants are social norm for self-tracking of sleep, and a specific personal motive to engage in self-tracking. For weight tracking, the most important determinants are an individual’s intention to monitor weight and having a specific motive for self-tracking. Finally, goal orientation as a sub domain of self-regulation capacity is related to a higher usage of the activity and sleep tracking function.

The results of this study may assist health practitioners who intend to suggest the use of self-tracking devices to their patients. It is important to note that a difference exists between asking patients to engage in self-monitoring or supporting patients who initiated self-tracking themselves. In the latter case, a patient is probably more motivated.28 In the first and most important case from a health practitioner perspective, the health practitioner might, before all else, explore what the individual health goals of the patient are, how the individual feels about self-tracking, what type of feedback is needed, and how goal- orientated this patient is. Based on the results of our study, this may increase the probability of a higher adoption rate and greater engagement with the device, ultimately leading to a more effective health behavior change. When a client does not have a specific health goal or has problems with adhering to goals, other strategies might be needed to begin with (such as a more extensive individual counselling) before self-tracking tools can be deployed. When

74 Behavioral determinants for the adoption of self-tracking devices by adults

This study has some strengths and limitations. This is the first follow-up study that includes several behavioral determinants, including self-regulation capacity, for the adoption Acknowledgements of an activity and sleep tracker and weight scale. A limitation was that some participants The authors would like to thank Nokia Health (previously Withings) for the donation of the self- experienced technical problems with the activity and sleep tracking device. Technical tracking devices and collaboration during the study. We also thank the Lifelines Cohort Study for failures, the perceived ease of use of a device, and perceived usefulness have earlier been granting access to their study cohort and accompanying support. found to impact the adoption of self-tracking technology.13,20,22 However, both devices could be replaced during the complete study period, therefore, technical failures could have had only limited influence on the adoption of the devices. The study was undertaken in a general population within the Lifelines Cohort with a high variation in age and BMI, which increases 4 generalizability. However, the population of the Lifelines Cohort consists mainly of Caucasian people who live in the Northern part of the Netherlands, which may limit the generalisability to other parts of Europe and other parts of the world. Finally, the sample size of this study was large enough to conduct this exploration, however, attrition in combination with the small size of some subgroups limit the generalizability of the results. Therefore, the conclusions should be read with caution and within the context of these limitations.

75 Chapter 4

References 23. Epstein DA, Caraway M, Johnston C, Ping A, Fogarty J, Munson SA. Beyond Abandonment to Next Steps: Understanding and Designing for Life after Personal Informatics Tool Use. Proc 2016 CHI Conf Hum Factors Comput Syst. 2016:1109-1113. doi:10.1145/2858036.2858045. 24. Bandura A. Health Promotion by Social Cognitive Means. Heal Educ Behav. 2004;31(2):143-164. 1. Fawcett T. Mining the Quantified Self: Personal Knowledge Discovery as a Challenge for Data Science. doi:10.1177/1090198104263660. Big Data. 2015;3(4):249-266. internal-pdf://228.60.152.96/Fawcett. 25. Ajzen I. The theory of planned behavior. Organ Behav Hum Decis Process. 1991;50(2):179-211. 2. Lupton D. Quantifying the body: monitoring and measuring health in the age of mHealth technologies. doi:10.1016/0749-5978(91)90020-T. Crit Public Health. 2013;23(4):393-403. doi:10.1080/09581596.2013.794931. 26. Hall PA, Fong GT. Temporal self-regulation theory: A model for individual health behavior. Health 3. Sanders JP, Loveday A, Pearson N, et al. Devices for self-monitoring sedentary time or physical activity: Psychol Rev. 2007;1(1):6-52. doi:10.1080/17437190701492437. a scoping review. J Med Internet Res. 2016;18(5). 27. Hall PA, Fong GT. Temporal self-regulation theory: Integrating biological, psychological, and ecological 4. Panagioti M, Richardson G, Small N, et al. Self-management support interventions to reduce health care determinants of health behavior performance. In: Social neuroscience and public health. Springer; utilisation without compromising outcomes: a systematic review and meta-analysis. BMC Health Serv 2013:35-53. Res. 2014;14(1):356. doi:10.1186/1472-6963-14-356. 28. Mann T, de Ridder D, Fujita K. Self-Regulation of Health Behavior. Heal Psychol. 2013;32(5):487-498. 5. Whitehead L, Seaton P. The effectiveness of self-management mobile phone and tablet apps in long- doi:10.1037/a0028533. term condition management: a systematic review. J Med Internet Res. 2016;18(5). 29. Carver CS, Scheier MF. Control theory: A useful conceptual framework for personality-social, clinical, 6. de Vries HJ, Kooiman TJM, van Ittersum MW, van Brussel M, de Groot M. Do activity monitors increase and health psychology. Psychol Bull. 1982;92(1):111-135. doi:10.1037/0033-2909.92.1.111. physical activity in adults with overweight or obesity? A systematic review and meta-analysis. Obesity. 30. Rosenbaum DL, Espel HM, Butryn ML, Zhang F, Lowe MR. Daily self-weighing and weight gain 2016;24(10):2078-2091. doi:10.1002/oby.21619. prevention: a longitudinal study of college-aged women. J Behav Med. 2017:1-8. 7. Zheng Y, Klem M Lou, Sereika SM, Danford CA, Ewing LJ, Burke LE. Self-weighing in weight management: 31. Boudreau F, Godin G. Participation in Regular Leisure-Time Physical Activity Among Individuals with A systematic literature review. Obesity. 2015;23(2):256-265. Type 2 Diabetes Not Meeting Canadian Guidelines: the Influence of Intention, Perceived Behavioral 8. Clawson J, Pater JA, Miller AD, Mynatt ED, Mamykina L. No longer wearing: investigating the Control, and Moral Norm. Int J Behav Med. 2014;21(6):918-926. doi:10.1007/s12529-013-9380-4. abandonment of personal health-tracking technologies on craigslist. In: Proceedings of the 2015 ACM 32. Tudor-Locke C, Craig CL, Brown WJ, et al. How many steps/day are enough? for adults. Int J Behav Nutr International Joint Conference on Pervasive and Ubiquitous Computing. ACM; 2015:647-658. Phys Act. 2011;8(1):79. doi:10.1186/1479-5868-8-79. 9. Kim J. Analysis of Health Consumers’ Behavior Using Self-Tracker for Activity, Sleep, and Diet. Telemed 33. Verplanken B, Orbell S. Reflections on Past Behavior: A Self-Report Index of Habit Strength1. J Appl Soc J E Health. 2014;20(6):552-558. doi:10.1089/tmj.2013.0282. Psychol. 2003;33(6):1313-1330. 10. Shih PC, Han K, Poole ES, Rosson MB, Carroll JM. Use and adoption challenges of wearable activity 34. Kluger AN, DeNisi A. The effects of feedback interventions on performance: A historical review, a meta- trackers. iConference 2015 Proc. 2015. analysis, and a preliminary feedback intervention theory. Psychol Bull. 1996;119(2):254-284. 11. Fritz T, Huang EM, Murphy GC, Zimmermann T. Persuasive technology in the real world: a study of long- doi:10.1037/0033-2909.119.2.254. term use of activity sensing devices for fitness. In: Proceedings of the SIGCHI Conference on Human 35. Sperrin M, Rushton H, Dixon WG, et al. Who Self-Weighs and What Do They Gain From It? A Factors in Computing Systems. ACM; 2014:487-496. Retrospective Comparison Between Smart Scale Users and the General Population in England. J Med 12. Fausset CB, Mitzner TL, Price CE, Jones BD, Fain BW, Rogers WA. Older adults’ use of and attitudes Internet Res. 2016;18(1). toward activity monitoring technologies. In: Proceedings of the Human Factors and Ergonomics Society 36. Gavin KL, Linde JA, Pacanowski CR, French SA, Jeffery RW, Ho Y-Y. Weighing frequency among working Annual Meeting. Vol 57. SAGE Publications; 2013:1683-1687. adults: Cross-sectional analysis of two community samples. Prev Med reports. 2015;2:44-46. 13. Lazar A, Koehler C, Tanenbaum J, Nguyen DH. Why we use and abandon smart devices. In: Proceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing. ACM; 2015:635-646. 14. Harrison D, Marshall P, Bianchi-Berthouze N, Bird J. Activity tracking: barriers, workarounds and customisation. In: Proceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing. ACM; 2015:617-621. 15. Hermsen S, Moons J, Kerkhof P, Wiekens C, Groot M De. Determinants for Sustained Use of an Activity Tracker: Observational Study. JMIR mHealth uHealth. 2017;5(10):e164. 16. Gouveia R, Karapanos E, Hassenzahl M. How do we engage with activity trackers?: a longitudinal study of Habito. In: Proceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing. ACM; 2015:1305-1316. 17. Canhoto AI, Arp S. Exploring the factors that support adoption and sustained use of health and fitness wearables. J Mark Manag. 2017;33(1-2):32-60. 18. Rooksby J, Rost M, Morrison A, Chalmers MC. Personal tracking as lived informatics. In: Proceedings of the 32nd Annual ACM Conference on Human Factors in Computing Systems - CHI ’14. ; 2014:1163-1172. doi:10.1145/2556288.2557039. 19. Dan L. Inside Wearables - Part 2. Endeavour Partners; 2014. http://search.credoreference.com/content/entry/gwyouth/wearables. 20. Kim J. A qualitative analysis of user experiences with a self-tracker for activity, sleep, and diet. Interact J Med Res. 2014;3(1). 21. Garavand A, Mohseni M, Asadi H, Etemadi M, Moradi-Joo M, Moosavi A. Factors influencing the adoption of health information technologies: a systematic review. Electron physician. 2016;8(8):2713. 22. Lunney A, Cunningham NR, Eastin MS. Wearable fitness technology: A structural investigation into acceptance and perceived fitness outcomes. Comput Human Behav. 2016;65:114-120.

76 Behavioral determinants for the adoption of self-tracking devices by adults

References 23. Epstein DA, Caraway M, Johnston C, Ping A, Fogarty J, Munson SA. Beyond Abandonment to Next Steps: Understanding and Designing for Life after Personal Informatics Tool Use. Proc 2016 CHI Conf Hum Factors Comput Syst. 2016:1109-1113. doi:10.1145/2858036.2858045. 24. Bandura A. Health Promotion by Social Cognitive Means. Heal Educ Behav. 2004;31(2):143-164. 1. Fawcett T. Mining the Quantified Self: Personal Knowledge Discovery as a Challenge for Data Science. doi:10.1177/1090198104263660. Big Data. 2015;3(4):249-266. internal-pdf://228.60.152.96/Fawcett. 25. Ajzen I. The theory of planned behavior. Organ Behav Hum Decis Process. 1991;50(2):179-211. 2. Lupton D. Quantifying the body: monitoring and measuring health in the age of mHealth technologies. doi:10.1016/0749-5978(91)90020-T. Crit Public Health. 2013;23(4):393-403. doi:10.1080/09581596.2013.794931. 26. Hall PA, Fong GT. Temporal self-regulation theory: A model for individual health behavior. Health 3. Sanders JP, Loveday A, Pearson N, et al. Devices for self-monitoring sedentary time or physical activity: Psychol Rev. 2007;1(1):6-52. doi:10.1080/17437190701492437. a scoping review. J Med Internet Res. 2016;18(5). 27. Hall PA, Fong GT. Temporal self-regulation theory: Integrating biological, psychological, and ecological 4. Panagioti M, Richardson G, Small N, et al. Self-management support interventions to reduce health care determinants of health behavior performance. In: Social neuroscience and public health. Springer; utilisation without compromising outcomes: a systematic review and meta-analysis. BMC Health Serv 2013:35-53. Res. 2014;14(1):356. doi:10.1186/1472-6963-14-356. 28. Mann T, de Ridder D, Fujita K. Self-Regulation of Health Behavior. Heal Psychol. 2013;32(5):487-498. 5. Whitehead L, Seaton P. The effectiveness of self-management mobile phone and tablet apps in long- doi:10.1037/a0028533. term condition management: a systematic review. J Med Internet Res. 2016;18(5). 29. Carver CS, Scheier MF. Control theory: A useful conceptual framework for personality-social, clinical, 6. de Vries HJ, Kooiman TJM, van Ittersum MW, van Brussel M, de Groot M. Do activity monitors increase and health psychology. Psychol Bull. 1982;92(1):111-135. doi:10.1037/0033-2909.92.1.111. 4 physical activity in adults with overweight or obesity? A systematic review and meta-analysis. Obesity. 30. Rosenbaum DL, Espel HM, Butryn ML, Zhang F, Lowe MR. Daily self-weighing and weight gain 2016;24(10):2078-2091. doi:10.1002/oby.21619. prevention: a longitudinal study of college-aged women. J Behav Med. 2017:1-8. 7. Zheng Y, Klem M Lou, Sereika SM, Danford CA, Ewing LJ, Burke LE. Self-weighing in weight management: 31. Boudreau F, Godin G. Participation in Regular Leisure-Time Physical Activity Among Individuals with A systematic literature review. Obesity. 2015;23(2):256-265. Type 2 Diabetes Not Meeting Canadian Guidelines: the Influence of Intention, Perceived Behavioral 8. Clawson J, Pater JA, Miller AD, Mynatt ED, Mamykina L. No longer wearing: investigating the Control, and Moral Norm. Int J Behav Med. 2014;21(6):918-926. doi:10.1007/s12529-013-9380-4. abandonment of personal health-tracking technologies on craigslist. In: Proceedings of the 2015 ACM 32. Tudor-Locke C, Craig CL, Brown WJ, et al. How many steps/day are enough? for adults. Int J Behav Nutr International Joint Conference on Pervasive and Ubiquitous Computing. ACM; 2015:647-658. Phys Act. 2011;8(1):79. doi:10.1186/1479-5868-8-79. 9. Kim J. Analysis of Health Consumers’ Behavior Using Self-Tracker for Activity, Sleep, and Diet. Telemed 33. Verplanken B, Orbell S. Reflections on Past Behavior: A Self-Report Index of Habit Strength1. J Appl Soc J E Health. 2014;20(6):552-558. doi:10.1089/tmj.2013.0282. Psychol. 2003;33(6):1313-1330. 10. Shih PC, Han K, Poole ES, Rosson MB, Carroll JM. Use and adoption challenges of wearable activity 34. Kluger AN, DeNisi A. The effects of feedback interventions on performance: A historical review, a meta- trackers. iConference 2015 Proc. 2015. analysis, and a preliminary feedback intervention theory. Psychol Bull. 1996;119(2):254-284. 11. Fritz T, Huang EM, Murphy GC, Zimmermann T. Persuasive technology in the real world: a study of long- doi:10.1037/0033-2909.119.2.254. term use of activity sensing devices for fitness. In: Proceedings of the SIGCHI Conference on Human 35. Sperrin M, Rushton H, Dixon WG, et al. Who Self-Weighs and What Do They Gain From It? A Factors in Computing Systems. ACM; 2014:487-496. Retrospective Comparison Between Smart Scale Users and the General Population in England. J Med 12. Fausset CB, Mitzner TL, Price CE, Jones BD, Fain BW, Rogers WA. Older adults’ use of and attitudes Internet Res. 2016;18(1). toward activity monitoring technologies. In: Proceedings of the Human Factors and Ergonomics Society 36. Gavin KL, Linde JA, Pacanowski CR, French SA, Jeffery RW, Ho Y-Y. Weighing frequency among working Annual Meeting. Vol 57. SAGE Publications; 2013:1683-1687. adults: Cross-sectional analysis of two community samples. Prev Med reports. 2015;2:44-46. 13. Lazar A, Koehler C, Tanenbaum J, Nguyen DH. Why we use and abandon smart devices. In: Proceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing. ACM; 2015:635-646. 14. Harrison D, Marshall P, Bianchi-Berthouze N, Bird J. Activity tracking: barriers, workarounds and customisation. In: Proceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing. ACM; 2015:617-621. 15. Hermsen S, Moons J, Kerkhof P, Wiekens C, Groot M De. Determinants for Sustained Use of an Activity Tracker: Observational Study. JMIR mHealth uHealth. 2017;5(10):e164. 16. Gouveia R, Karapanos E, Hassenzahl M. How do we engage with activity trackers?: a longitudinal study of Habito. In: Proceedings of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing. ACM; 2015:1305-1316. 17. Canhoto AI, Arp S. Exploring the factors that support adoption and sustained use of health and fitness wearables. J Mark Manag. 2017;33(1-2):32-60. 18. Rooksby J, Rost M, Morrison A, Chalmers MC. Personal tracking as lived informatics. In: Proceedings of the 32nd Annual ACM Conference on Human Factors in Computing Systems - CHI ’14. ; 2014:1163-1172. doi:10.1145/2556288.2557039. 19. Dan L. Inside Wearables - Part 2. Endeavour Partners; 2014. http://search.credoreference.com/content/entry/gwyouth/wearables. 20. Kim J. A qualitative analysis of user experiences with a self-tracker for activity, sleep, and diet. Interact J Med Res. 2014;3(1). 21. Garavand A, Mohseni M, Asadi H, Etemadi M, Moradi-Joo M, Moosavi A. Factors influencing the adoption of health information technologies: a systematic review. Electron physician. 2016;8(8):2713. 22. Lunney A, Cunningham NR, Eastin MS. Wearable fitness technology: A structural investigation into acceptance and perceived fitness outcomes. Comput Human Behav. 2016;65:114-120.

Chapter 5 | Do activity monitors increase physical activity in adults with overweight or obesity? A systematic review and meta- analysis

Herman J. de Vries Thea J.M. Kooiman Miriam W. van Ittersum Marco van Brussel Martijn de Groot

Obesity (2016) 24, 2078–2091

Chapter 5 | Do activity monitors increase physical activity in adults with overweight or obesity? A systematic review and meta- analysis

Herman J. de Vries Thea J.M. Kooiman Miriam W. van Ittersum Marco van Brussel Martijn de Groot

Obesity (2016) 24, 2078–2091

Chapter 5

Abstract Introduction

Objective Worldwide, 1.46 billion adults suffered from overweight and 502 million had obesity in To systematically assess contemporary knowledge regarding behavioral physical activity 2008.1 The global rising prevalence of these conditions is expected to further increase both 2 interventions including an activity monitor (BPAI+) in adults with overweight or obesity. the health and economic burdens in the following decades. Overweight and obesity are frequently caused by a chronic imbalance involving dietary and physical activity patterns.3 Methods Behavioral interventions involving alterations in both physical activity and diet can lead to PubMed/MEDLINE, Embase, CINAHL, PsycINFO, CENTRAL and PEDro were searched for clinically important weight loss (≥5% of baseline weight) in adults with overweight or 4 eligible full text articles up to July 1st 2015. Studies eligible for inclusion were (randomized) obesity. Physical activity should be facilitated in intervention programs to enhance the controlled trials describing physical activity outcomes in adults with overweight or obesity. likelihood of not only successful weight loss and weight maintenance but also for health 5 Methodological quality was independently assessed employing the Cochrane Collaboration's benefits regardless of weight loss. A recent systematic review concluded that physical tool for risk of bias. PROSPERO registration: CRD42015024086. activity was included in 88% of studies that achieved clinically important weight loss, 6 whereby behavioral training (such as self-monitoring) was included in 92% of these studies. Results Over the previous decades, there has been increasing interest in the therapeutic Fourteen studies (1157 participants) were included for systematic review and eleven for application of objective measures of self-monitoring. One of the first objective measuring meta-analysis. A positive trend in BPAI+ effects on several measures of physical activity was instruments for physical activity was introduced in 1965 with the release of the Japanese ascertained compared to both waitlist or usual care (WL/UC) and behavioral physical activity manpo-kei pedometer, meaning “10,000 steps meter”.7 Currently, devices such as tri-axial interventions without an activity monitor (BPAI-). No convincing evidence of BPAI+ accelerometers, gyroscopes and global positioning systems are combined to create activity effectiveness on weight loss was found compared to BPAI-. monitors that are more accurate8, 9 and even integrate behavior change techniques (BCTs) such as social support, prompts/cues, rewards and behavioral outcome self-monitoring.10 Conclusions Pedometer employment increases physical activity in adults with type 2 diabetes, 11-14 Behavioral physical activity interventions with an activity monitor increase physical activity musculoskeletal diseases and several other outpatient populations. Furthermore, in adults with overweight or obesity. Also, adding an activity monitor to behavioral physical pedometer based walking interventions have assisted in achieving moderate weight loss in 11, 13, 15 activity interventions appears to increase the effect on physical activity, although current adults. Although recent meta-analysis regarding the effects of activity monitors evidence has not yet provided conclusive evidence for its effectiveness. indicates positive outcomes on physical activity, HbA1c, systolic blood pressure and body mass index (BMI) in patients with type 2 diabetes,16 no systematic review regarding the

effects of activity monitor based interventions on physical activity in adults with overweight or obesity is yet available. Therefore, the aim of the current systematic review was twofold. The first aim was to establish whether receiving a behavioral physical activity intervention with an activity monitor (BPAI+) increases physical activity in adults with overweight or obesity compared to both waitlist and usual care (WL/UC). The rationale was to determine whether offering a BPAI+ has clinical relevance when increasing physical activity is targeted. The second aim is to establish the added value of activity monitoring in existing interventions on increment of physical activity. Therefore, BPAI+ was compared with behavioral physical activity interventions without an activity monitor (BPAI-). Additionally, the BPAI+ effect on body weight compared to BPAI- will be examined.

80 Do activity monitors increase physical activity in adults with overweight or obesity?

Abstract Introduction

Objective Worldwide, 1.46 billion adults suffered from overweight and 502 million had obesity in To systematically assess contemporary knowledge regarding behavioral physical activity 2008.1 The global rising prevalence of these conditions is expected to further increase both 2 interventions including an activity monitor (BPAI+) in adults with overweight or obesity. the health and economic burdens in the following decades. Overweight and obesity are frequently caused by a chronic imbalance involving dietary and physical activity patterns.3 Methods Behavioral interventions involving alterations in both physical activity and diet can lead to PubMed/MEDLINE, Embase, CINAHL, PsycINFO, CENTRAL and PEDro were searched for clinically important weight loss (≥5% of baseline weight) in adults with overweight or 4 eligible full text articles up to July 1st 2015. Studies eligible for inclusion were (randomized) obesity. Physical activity should be facilitated in intervention programs to enhance the controlled trials describing physical activity outcomes in adults with overweight or obesity. likelihood of not only successful weight loss and weight maintenance but also for health 5 Methodological quality was independently assessed employing the Cochrane Collaboration's benefits regardless of weight loss. A recent systematic review concluded that physical tool for risk of bias. PROSPERO registration: CRD42015024086. activity was included in 88% of studies that achieved clinically important weight loss, 6 whereby behavioral training (such as self-monitoring) was included in 92% of these studies. 5 Results Over the previous decades, there has been increasing interest in the therapeutic Fourteen studies (1157 participants) were included for systematic review and eleven for application of objective measures of self-monitoring. One of the first objective measuring meta-analysis. A positive trend in BPAI+ effects on several measures of physical activity was instruments for physical activity was introduced in 1965 with the release of the Japanese ascertained compared to both waitlist or usual care (WL/UC) and behavioral physical activity manpo-kei pedometer, meaning “10,000 steps meter”.7 Currently, devices such as tri-axial interventions without an activity monitor (BPAI-). No convincing evidence of BPAI+ accelerometers, gyroscopes and global positioning systems are combined to create activity effectiveness on weight loss was found compared to BPAI-. monitors that are more accurate8, 9 and even integrate behavior change techniques (BCTs) such as social support, prompts/cues, rewards and behavioral outcome self-monitoring.10 Conclusions Pedometer employment increases physical activity in adults with type 2 diabetes, 11-14 Behavioral physical activity interventions with an activity monitor increase physical activity musculoskeletal diseases and several other outpatient populations. Furthermore, in adults with overweight or obesity. Also, adding an activity monitor to behavioral physical pedometer based walking interventions have assisted in achieving moderate weight loss in 11, 13, 15 activity interventions appears to increase the effect on physical activity, although current adults. Although recent meta-analysis regarding the effects of activity monitors evidence has not yet provided conclusive evidence for its effectiveness. indicates positive outcomes on physical activity, HbA1c, systolic blood pressure and body mass index (BMI) in patients with type 2 diabetes,16 no systematic review regarding the

81 Chapter 5

Methods assessment, incomplete outcome data, selective reporting and other biases. Since the blinding of participants is practically infeasible in a self-monitoring intervention, this was not assessed and blinding of personnel was scored separately. Discrepancies between the raters The protocol for this systematic review and meta-analysis was based on the PRISMA-P were resolved in a consensus meeting. The strength of inter-rater agreement was measured statement17 and registered at PROSPERO (CRD42015024086).18 The review was executed by Cohen’s Coefficient.19 according to the Cochrane Handbook for Systematic Reviews of Interventions19 following the PRISMA statement.20 Synthesis of results Data extraction was performed by the reviewers utilizing a standard extraction form. Search strategy Extracted data from the articles included: (a) first author, publication year, and study Electronic databases were searched using the sensitivity-maximizing version of the Cochrane location, (b) participants age and BMI; (c) intervention characteristics; (d) outcome Search Strategy to filter for Randomized Controlled Trials (RCT’s) and Controlled Clinical measures; and (e) study results. When multiple (≥2) studies compared analogous groups and Trials (CCT’s).19 MEDLINE, Embase, CINAHL, PsycINFO, CENTRAL and PEDro were searched reported the same outcome measures, results were pooled in RevMan 5.3 software for for eligible articles published prior to July 1st 2015. The employed MeSH terms and keywords random-effects meta-analysis using the Inverse Variance method. For statistical pooling, included overweight, obesity, accelerometry, actigraphy, physical activity, exercise and extracted data from Intention To Treat (ITT) analyses was preferred over completer analyses energy expenditure. Furthermore, a reference tracking strategy was performed by searching and data measured by an objective instrument was favored over data from subjective the reference lists and citations of included articles in Web of Science and Scopus. The instruments. complete search strategy can be found in the protocol.18

Study selection Articles were eligible for inclusion if (a) they had a RCT or CCT design; (b) the majority of the participants were human adults with overweight or obesity (mean baseline BMI ≥27.0 kg/m2 for Caucasians or ≥25.0 kg/m2 for Asians); (c) the intervention included the application of activity monitors; (d) the control group was on a waitlist, received usual care or were provided with a similar physical activity intervention as that of the intervention group but without activity monitor feedback; (e) physical activity changes for both intervention- and control group were described; and (f) the article was full-text available in English. Articles were excluded if (g) the document was a conference abstract, research letter, editorial note or commentary; (h) the intervention included non-spontaneous physical activity or a workplace environment modification; (i) participants were primarily older adults (mean age ≥60 years) or pregnant women; (j) participants were possibly limited in the ability to modify physical activity due to serious comorbidity caused by a chronic disease or it’s treatment; or (k) the intervention period was <2 weeks.

Two independent content area experts [authors HJdV and TJMK] screened potentially eligible articles for inclusion based on titles and abstract. Full-text articles were subsequently screened for final inclusion. Differences in appraisal were resolved by reaching consensus. The strength of inter-rater agreement was measured by Cohen’s coefficient.19

Methodological quality (risk of bias) The risk of bias was scored by two independent reviewers [HJdV and TJMK] using the Cochrane Collaboration’s tool.19 This tool reviews the random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome

82 Do activity monitors increase physical activity in adults with overweight or obesity?

Methods assessment, incomplete outcome data, selective reporting and other biases. Since the blinding of participants is practically infeasible in a self-monitoring intervention, this was not assessed and blinding of personnel was scored separately. Discrepancies between the raters The protocol for this systematic review and meta-analysis was based on the PRISMA-P were resolved in a consensus meeting. The strength of inter-rater agreement was measured statement17 and registered at PROSPERO (CRD42015024086).18 The review was executed by Cohen’s Coefficient.19 according to the Cochrane Handbook for Systematic Reviews of Interventions19 following the PRISMA statement.20 Synthesis of results Data extraction was performed by the reviewers utilizing a standard extraction form. Search strategy Extracted data from the articles included: (a) first author, publication year, and study Electronic databases were searched using the sensitivity-maximizing version of the Cochrane location, (b) participants age and BMI; (c) intervention characteristics; (d) outcome Search Strategy to filter for Randomized Controlled Trials (RCT’s) and Controlled Clinical measures; and (e) study results. When multiple (≥2) studies compared analogous groups and Trials (CCT’s).19 MEDLINE, Embase, CINAHL, PsycINFO, CENTRAL and PEDro were searched reported the same outcome measures, results were pooled in RevMan 5.3 software for for eligible articles published prior to July 1st 2015. The employed MeSH terms and keywords random-effects meta-analysis using the Inverse Variance method. For statistical pooling, included overweight, obesity, accelerometry, actigraphy, physical activity, exercise and extracted data from Intention To Treat (ITT) analyses was preferred over completer analyses energy expenditure. Furthermore, a reference tracking strategy was performed by searching and data measured by an objective instrument was favored over data from subjective 5 the reference lists and citations of included articles in Web of Science and Scopus. The instruments. complete search strategy can be found in the protocol.18

83 Chapter 5

Results

Study selection The search strategy identified 1,645 articles, of which 14 full-text articles (including 1,157 participants) that satisfied the inclusion criteria were identified and reviewed.21-34 Selection agreement between the reviewers was excellent (κ=0.75).19 A flowchart of the study selection process is depicted in Figure 1. Seven studies compared patients receiving BPAI+ to WL/UC21-24, 28, 31, 33 and seven different studies compared patients receiving BPAI+ to patients receiving BPAI-.25-27, 29, 30, 32, 34 Five studies comparing patients receiving BPAI+ to WL/UC used an objective measure to determine physical activity outcomes21-23, 31, 33 and two used a self-administered questionnaire24, 28. Additionally, four studies comparing patients receiving BPAI+ to BPAI- used an objective physical activity measure25-27, 34 while three employed a questionnaire.29, 30, 32 Data from eleven studies21-26, 28-30, 33, 34 were pooled for meta-analysis including: steps per day, total moderate to vigorous physical activity (MVPA) minutes per time unit, walking MET·minutes/week, physical activity kilocalories per week, weight change in kilograms, relative weight change and BMI change outcomes. No major adverse events related to the interventions were reported. Four studies executed a gender-specific intervention.23-26 All of the individual study characteristics can be found in Table 1.

Methodological quality (risk of bias) The agreement between the reviewers in the risk of bias assessment was fair (κ=0.52),19 however, consensus was reached. Among the 14 included studies, several increased risks of bias were assessed. One study24 used a non-random component in the sequence generation process by randomizing work crews instead of individual participants (cluster randomized trial). It is unclear in seven studies25, 26, 29, 30, 32-34 if random sequence generation was used and in ten studies22, 24-30, 33, 34 whether allocation concealment was applied. Blinding of Figure 1. intervention personnel was absent in seven studies21, 22, 24, 27, 30, 31, 33 and unclear in three Flowchart of selected studies. studies.29, 32, 34 Blinding of outcome assessment was absent in one study26 and unclear in seven studies.22, 24, 25, 29, 30, 32, 34 Three studies had a high risk of attrition bias due to incomplete outcome data22, 29, 32 while the risk of attrition bias was unclear in two studies.30, 34 Two studies had a high risk of reporting bias due to selective reporting.31, 34 Another high risk of bias was introduced by two studies which allowed control group participants to register their own weekly step totals.25, 26 Lastly, an unclear risk of bias may have been introduced by a study that removed outliers without being able to verify if the data was erroneous31 and a study with a significant between-group age differences at baseline.34 The distribution of the risks of bias is graphically depicted in Figure 2.

84 Do activity monitors increase physical activity in adults with overweight or obesity?

Results

Study selection The search strategy identified 1,645 articles, of which 14 full-text articles (including 1,157 participants) that satisfied the inclusion criteria were identified and reviewed.21-34 Selection agreement between the reviewers was excellent (κ=0.75).19 A flowchart of the study selection process is depicted in Figure 1. Seven studies compared patients receiving BPAI+ to WL/UC21-24, 28, 31, 33 and seven different studies compared patients receiving BPAI+ to patients receiving BPAI-.25-27, 29, 30, 32, 34 Five studies comparing patients receiving BPAI+ to WL/UC used an objective measure to determine physical activity outcomes21-23, 31, 33 and two used a self-administered questionnaire24, 28. Additionally, four studies comparing patients receiving BPAI+ to BPAI- used an objective physical activity measure25-27, 34 while three employed a questionnaire.29, 30, 32 Data from eleven studies21-26, 28-30, 33, 34 were pooled for meta-analysis including: steps per day, total moderate to vigorous physical activity (MVPA) minutes per 5 time unit, walking MET·minutes/week, physical activity kilocalories per week, weight change in kilograms, relative weight change and BMI change outcomes. No major adverse events related to the interventions were reported. Four studies executed a gender-specific intervention.23-26 All of the individual study characteristics can be found in Table 1.

85 Chapter 5

based based

ined until until ined

Meter. 200).

- - O

- x Digiwalker). Unknown 7 days 7 Unknown day. 1 ch week in the

Participants were shown and steps daily Record Unknown Record pedometer and steps daily Monitor

109E Step determined SMART goal -

- : Daily steps + i1: paper paper i1: + steps Daily : : Daily: steps PA+ diary log + monitoring steps daily : : Daily steps. online logbook, steps, Daily : log.

Use of the Activity Monitor the Activity of Use : Individually discussed during first first during discussed Individually : : 1 self : Increase: steps/d by 3,000 on 5 :1) moderate intensity walking in : 1) ≥10,000 steps/d on ≥5 d/w

k, online i2: diary entries. related walking minutes in a - rd the data on the website to assist assist to website the on data the rd : Omron HJ : model) (unknown Pedometer : : Pedometer (Yamax SW : Pedometer (model unknown). : Pedometer (Yama

Data recall from AM: from recall Data per month. 2 PA related weight loss reduce and steps daily increase messages; sitting time. feedback AM logboo steps for four days ea a checkbox off and to tick books support when goals are achieved. PA goal AM AM: Instruction step daily monitor to AM the use to how count. PA goal goals sub with reached was goal This d/w. mainta then and 7 week until week 12. feedback AM AM: from recall Data AM AM: Instruction bout monitoring PA goal bouts ≥10 min for 30 min/d and 2) increase steps/d by 5,000/d. This goal was reached in smaller sub goals per week. feedback AM AM: from recall Data AM AM: Ins truction PA goal was the goal whether unknown session, specifically for steps. feedback AM AM: from recall Data AM AM: Instruction AM AM: Instruction reco with goal setting. PA goal d/w targeting ≥2 training strength and 2) body areas. ≥2 feedback AM behaviors. weekly graphical feedback of PA

selected

- mail - setting, setting,

- Intervention group Intervention

setting, use of a mails as as mails -

- feedback of other

mails about their

setting, plan making. based logbook. logbook. based -

- Determination nd website user guide. guide. user nd website

- 45 min) -

selected goal .

) - BPAI Discussion of motivation and motivation of Discussion Increase knowledge, goal knowledge, Increase Self - Education, goal Education,

- - -

: Weekly: graphical : : : : individualized e ≤7 : : care provider, optional e optional provider, care

loss DVD, weight loss handbook and and handbook loss weight DVD, loss ning, PA cues, review of progress, - related questions. questions. related

Cognitive Theory.

None Achieve a negative energy balance of None

Same as resources group, with additional additional with group, resources as Same - mail or telephone contact to facilitate facilitate to telephone or contact mail 7 individualized feedback e feedback individualized 7 : Session handouts Session : : Weight : CaloryKing website, website guide, user : :

- : Transtheoretical Model. Model. Transtheoretical : Planned of Theory Model, Transtheoretical : Social: Social: Cognitive Theory, BDM. solving. 1 individual session (75 min). - : 1 individual session at baseline (30 min). min). (30 at baseline session individual 1 : (30 sessions individual 6 : : None: the of instruction with conversation Orientation :

: Social: on Cognitive Theory.

setting, advice for social social strategies. support for advice setting, - eight loss book. loss eight barriers towards PA, individual goal feedback Additional Behavioral components: Behavioral Based on Based Contact on Based Behavior, Social Cognitive and Self Theory. Contact components: Behavioral setting, action plan problem feedback Additional Resources Advice: Dietary : 6 weeks Duration on Based Contact components: Behavioral credible source, small financial incentives. feedback Additional (i1): group Resources Based Contact Tailoring components: for men, self Behavioral goal feedback Additional Resources book. counter kilojoule measure, book, tape support advice: Dietary per2000 kJ day. 7 diet related weight loss messages. months 3 : Duration (i2): group Online Resources: website use (CaloryKing website) a paper of instead monitoring Online Feedback: entries. diary exercise and food on Based Contact website, e discussion troubleshooting, and website with interaction of goals with health health for contact feedback Additional Resources w loss. weight and diet about Education advice: Dietary : 14 weeks Duration Resources : advice Dietary : 12 weeks Duration Behavioral components: Behavioral tailoring of website content (informed by assessed health behavioral skills. of use behaviors), response on weight, eating and exercise diaries.

- : :

list control list control list control list control list - - - - Control group Control Wait group careUsual pre Standard surgical care; adopt to advice active lifestyle in engage and and walking similar activities but no formal or prescription to strategies change PA behavior. Wait group Wait group Wait , withgroup an to access alternate website containing general health of information interest to men to likely not but changes to lead

: 80: (c: 80: (c: 110 : 159 : 441 : : 100 : : 13.5: : 20.3: 100 : 100 :

: 45.0 ± 6.5 45.0 : ± 3.6 30.5 : ± : 29.0 ± 29.0: ± 3.5 32.7 : ± : 34.3 ± 34.3: ± : 46.0 ± 8.9 46.0 : ± 8.6 44.4 : ± : 49.2 ± 8.8 49.2 : ± 47.5: ± 8.0 43.9 : ± 44.9 7.8) ±

Patients Patients Male shift Men.

2: 53)2:

ber (randomized) ber Participants (M ± SD) ± (M Participants

i: 42) i: workers. Number (randomized) Number 40, 39) i: Community group: Target sample. (randomized) Number 38, (randomized) % male 11.4,(c: i: 15.0) (randomized) Age 48.1 (c: 8.1,± 44.2 i: 9.2) ± (randomized) I BM 44.4 (c: 5.8,± 45.6 i: 7.0) ± group: Target awaiting bariatric surgery. Num 45,(c: 65) i: (randomized) % male (randomized) Age (randomized) Number 52,(c: i1: 54, i (randomized) Number 217,(c: i: 224) % male (randomized) % male 43.7 (c: 9.1,± 44.8 i: 8.3) ± (randomized) BMI 30.2 (c: 3.5,± 30.7 i: 3.6) ± group: Target (randomized) % male (randomized) % male (c: 20.0%,(c: 20.5%) i: (randomized) Age (randomized) Age Age (randomized) Age 51.2 (c: 7.9,± 47.3 i: 9.3) ± (randomized) BMI (c:5.6* 29.4 6.3,± 28.5 i: ± 4.8) 48.011.0 (c: 1.2, ± 48.0 i1: ± 10.8, 46.5 i2: 11.1) ± (randomized) BMI 3.9,(33.1 ± 32.4 i1: 3.3, ± i2: 3.4)32.8 ± group: Target 42.8 (c: 8.0,± i: BMI (randomized) BMI (c:4.1* 34.3 4.0,± 34.2 i: ± 4.2)

24 23 21

22 28

Study

Scotland Australia USA WL/UC versus BPAI+ Comparison: G. Baker, 2008 D. Bond, S. 2014 USA Morgan, P.J. 2011 Australia Morgan, P.J. 2013 Patrick, K. 2011 Table 1 BPAI versus and WL/UC versus (BPAI+ by comparison divided studies, of included Characteristics

86 Do activity monitors increase physical activity in adults with overweight or obesity?

based based

ined until until ined

Meter. 200).

- - O

- x Digiwalker). 7 days 7 Unknown day. 1 Unknown ch week in the

Participants were shown and steps daily Record Unknown Record pedometer and steps daily Monitor

109E Step determined SMART goal -

- : Daily: steps PA+ diary log + monitoring steps daily : : Daily steps. paper i1: + steps Daily : online logbook, steps, Daily : log.

Use of the Activity Monitor the Activity of Use : Individually discussed during first first during discussed Individually : : Increase: steps/d by 3,000 on 5 :1) moderate intensity walking in : 1 self : 1) ≥10,000 steps/d on ≥5 d/w

AM AM: Instruction step daily monitor to AM the use to how count. PA goal goals sub with reached was goal This d/w. mainta then and 7 week until week 12. feedback AM AM: from recall Data AM AM: Instruction bout monitoring PA goal bouts ≥10 min for 30 min/d and 2) increase steps/d by 5,000/d. This goal was reached in smaller sub goals per week. feedback AM AM: from recall Data AM AM: Ins truction PA goal was the goal whether unknown session, specifically for steps. feedback AM AM: from recall Data AM AM: Instruction steps for four days ea a checkbox off and to tick books support when goals are achieved. PA goal per month. 2 PA related weight loss reduce and steps daily increase messages; sitting time. feedback AM logboo AM: from recall Data AM AM: Instruction reco with goal setting. PA goal d/w targeting ≥2 training strength and 2) body areas. ≥2 feedback AM behaviors. weekly graphical feedback of PA

selected

- mail - setting, setting,

- Intervention group Intervention

setting, use of a 5 mails as as mails -

- feedback of other

mails about their

- setting, plan making. based logbook. logbook. based -

- Determination nd website user guide. guide. user nd website

- 45 min) -

selected goal .

) - BPAI Discussion of motivation and motivation of Discussion Increase knowledge, goal knowledge, Increase Self - Education, goal Education,

- - -

: Weekly: graphical : : : : individualized e ≤7 : : care provider, optional e optional provider, care

loss DVD, weight loss handbook and and handbook loss weight DVD, loss ning, PA cues, review of progress, - related questions. questions. related

Cognitive Theory.

None Achieve a negative energy balance of None

Same as resources group, with additional additional with group, resources as Same - mail or telephone contact to facilitate facilitate to telephone or contact mail 7 individualized feedback e feedback individualized 7 : Session handouts Session : : CaloryKing website, website guide, user : Weight : :

: Social: on Cognitive Theory.

setting, advice for social social strategies. support for advice setting, - eight loss book. loss eight barriers towards PA, individual goal feedback Additional Behavioral components: Behavioral Based on Based Contact on Based Behavior, Social Cognitive and Self Theory. Contact components: Behavioral setting, action plan problem feedback Additional Resources Advice: Dietary : 6 weeks Duration on Based Contact components: Behavioral credible source, small financial incentives. feedback Additional Resources w loss. weight and diet about Education advice: Dietary : 14 weeks Duration (i1): group Resources Based Contact Tailoring components: for men, self Behavioral goal feedback Additional Resources book. counter kilojoule measure, book, tape support advice: Dietary per2000 kJ day. 7 diet related weight loss messages. months 3 : Duration (i2): group Online Resources: website use (CaloryKing website) a paper of instead monitoring Online Feedback: entries. diary exercise and food on Based Contact website, e discussion troubleshooting, and website with interaction of goals with health health for contact feedback Additional Resources : advice Dietary : 12 weeks Duration Behavioral components: Behavioral tailoring of website content (informed by assessed health behavioral skills. of use behaviors), response on weight, eating and exercise diaries.

- : :

list control list control list control list control list - - - - Control group Control Wait group careUsual pre Standard Wait group Wait group Wait , withgroup an to access alternate website containing general health of information interest to men to likely not but changes to lead surgical care; adopt to advice active lifestyle in engage and and walking similar activities but no formal or prescription strategies to strategies change PA behavior.

: 80: (c: 80: (c: 110 : 159 : 441 : : 100 : : 100 : : 20.3: 13.5: 100 :

: 45.0 ± 6.5 45.0 : ± 3.6 30.5 : ± 3.5 32.7 : ± : 29.0 ± 29.0: ± : 34.3 ± 34.3: ± : 46.0 ± 8.9 46.0 : ± 8.6 44.4 : ± 47.5: ± : 49.2 ± 8.8 49.2 : ± 8.0 43.9 : ± 44.9 7.8) ±

Men. Patients Patients Male shift

2: 53)2:

ber (randomized) ber Participants (M ± SD) ± (M Participants

i: 42) i: workers. Number (randomized) Number 40, 39) i: (randomized) Number 38, Num 45,(c: 65) i: (randomized) % male (randomized) Age (randomized) Number 52,(c: i1: 54, i (randomized) % male (randomized) Age 48.011.0 (c: 1.2, ± 48.0 i1: ± 10.8, 46.5 i2: 11.1) ± (randomized) BMI 3.9,(33.1 ± 32.4 i1: 3.3, ± i2: 3.4)32.8 ± group: Target (randomized) Number 217,(c: i: 224) % male (randomized) % male (randomized) % male (randomized) Age 48.1 (c: 8.1,± 44.2 i: 9.2) ± (randomized) I BM 44.4 (c: 5.8,± 45.6 i: 7.0) ± group: Target 43.7 (c: 9.1,± 44.8 i: 8.3) ± (randomized) BMI 30.2 (c: 3.5,± 30.7 i: 3.6) ± group: Target (randomized) % male (c: 20.0%,(c: 20.5%) i: 11.4,(c: i: 15.0) (randomized) Age Age (randomized) Age 51.2 (c: 7.9,± 47.3 i: 9.3) ± (randomized) BMI (c:5.6* 29.4 6.3,± 28.5 i: ± 4.8) Community group: Target sample. awaiting bariatric surgery. 42.8 (c: 8.0,± i: BMI (randomized) BMI (c:4.1* 34.3 4.0,± 34.2 i: ± 4.2)

24 23 21

22 28

Study

Comparison: BPAI+ versus WL/UC versus BPAI+ Comparison:

Scotland USA USA Baker, G. Baker, 2008 D. Bond, S. 2014 Morgan, P.J. 2011 Australia Morgan, P.J. 2013 Australia Patrick, K. 2011 Characteristics of included studies, divided by comparison (BPAI+ versus WL/UC and versus BPAI versus and WL/UC versus (BPAI+ by comparison divided studies, of included Characteristics Table 1

87 Chapter 5

200) -

Unknown setting and self 1 day 1 day ame as c. ame as c.

1,000 steps/w, until - Wear the pedometer bothUse pedometer

200) 200). -

- : Daily: and steps PA : Daily steps, online logbook. online logbook. steps, Daily :

2) 10,0002) steps/d.

accumulated on accelerometer,

1 day. 1 1 day. 1

: Unknown : : Create weekly step goals,

: Pedometer (Yamax SW : Pedometer (model unknown) Wear the AM and monitor daily steps. daily monitor and AM the Wear Monitor daily steps, record on a record steps, daily Monitor

ial Cognitive Theory.

AM feedback AM calendar. AM: from recall Data monitoring. PA goal Data recall from AM: from recall Data AM AM: Instruction the on steps daily record and daily website. PA goal increasing by 500 reaching ultimate walking goal of 10,000 steps/d. feedback AM into steps tracked converted website The and journey virtual a on traveled distance motivation. increase to targets provided AM: from recall Data AM AM: Instruction goal for and calendar

- - Soc

: Daily steps and calendar. and steps Daily : : Activity: units : Daily steps and calendar. and steps Daily : : : : :

axial accelerometer (BioTrainer). : Brisk walking or exercise of equivalent intensity - Use of the Activity Monitor added in BPAI+ in added Monitor Activity the of Use - : Pedometer (Yamax SW : Pedometer (Yamax SW : Tri

AM AM calendar. a on steps/d Record AM: Instruction PA goal: ≥10,000 steps/d. AM Based on Based Same as c. components: Contact/Behavioral on Based S components: Contact/Behavioral Feedback on Based S components: Contact/Behavioral Data recall from AM: from recall Data Same as c. advice/duration: Resources/dietary Data recall from AM: from recall Data same as c. advice/duration: Resources/dietary Instruction AM: Instruction PA goal ≥3 times pw. Feedback monthly history displayed graphically on a computer computer a on graphically displayed history monthly the of Discussion diary. paper printed, was and data. accelerometer 28 days. AM: from recall Data Instruction AM: Instruction calendar. PA goal: Set small achievable goals, weekly increase to 1) walking ≥30 min/d and Same as c. advice/duration: Resources/dietary Feedback -

Theory, Transtheoretical Transtheoretical Theory, management Goal setting and problem and setting Goal Encouragement, strategies for for strategies Encouragement, -

reported exercise.

- es website. es f the study. the f

rded weekly steps. steps. weekly rded

: Positive online feedback. : e

based learning. based - Dietary goals and information about a and goals information Dietary Non

aseline, mid study and post study. study. post and study mid aseline, : Web : New Lifestyl Program Step manual. First : : Theory of Planned Behavior. Planned of Theory : Social: Cognitive BPAI : Online message board, three data collection data collection board, three message Online : : 4 group sessions in the first four weeks.

- - Basic behavioralBasic self

sed on

health behaviors. behaviors. health Resources advice: Dietary website. the through diet healthy 12 months: Duration on Based Contact sessions at b components: Behavioral board. message online the through improvement feedback Additional Nutritional advice: Dietary guidance, same c. as Resources : 12 weeks Duration Ba Model. Contact components: Behavioral solving exercises. Resources feedback Additional Dietary advice: Dietary : 16 weeks Duration

l achievable goals like 10 min walks, gradually

Maintain normal diet. diet. normal Maintain

page manual with exercise instructions and self and instructions exercise with manual page -

onthly counseling by behavior therapist (15 min). (15 therapist by behavior counseling onthly : : - - : National: Australian Physical Activity Guidelines. : National Australian Physical Activity Guidelines. Activity Physical Australian National : : 24:

list control control list - : A sealed pedometer. Control participants wore this sealed : A sealed pedometer. Participants reco : Tri change in in change Resources AM pedometer for 12 weeks with weekly recording. PA goal were set. goals min/d. No step ≥30 to weekly goal increase Feedback o duration for diet Maintain advice: Dietary : 12 weeks Duration advice: Dietary Behavioral components: Behavioral components: Behavioral AM Resources : 12 weeks. Duration in diet or physical physical or diet in activity (e.g., on information stress, hair loss, injury worksite prevention). careUsual "Usual Lifestyle", no advised. activity data three During collection sessions, 30 minutes were on spent nutritional in guidance, order to minimise biases. dietary Wait group on Based Contact on Based Contact PA goal Feedback on Based Contact and encountered problems goals, of discussion principles, solving. problem Behavioral components: Behavioral AM times pw. Feedback Resources None. advice: Dietary PA goal management.

: 106 : 60: (c: 30: (c: 32: (c: 30: (c:

: 0 : 0 : 46.2: : 20.2 (c: 20.2: (c: 55.3: (c:

- : 48.1 ± 7.1 48.1 : ± : 32.5 ± 5.4 32.5 : ± : 33.3 ± 5.6 33.3 : ± 2.6* 29.3: ± 4.1* 29.3: ± : 50.0 ± 9.3 50.0: ± 5.2 52.7: ± 28.8* 43 ± : 9.1 43.5: ±

Women. Military Patients with with Patients Community Men.

BPAI

#, 29.9 i: 2.5#,±

± 8.4#,± 41.4 i: 9.8#, ±

r (randomized) : 52):

Unknown (inclusion: (inclusion: Unknown

54, I

type 2 diabetes. 2 type sample Target group: Target Target group: Target (randomized) Number (c: (completers) % male 27.3%, 13.3%) i: (completers) Age 50 ± (c: 8.4, 50 i: 10.1) ± (completers) BMI 32.6 (c: 6.2,± 32.3 i: 4.6, ± p=.915) group: Target beneficiaries. (randomized) Number 30, 30) i: (completers) Age (randomized) Number 15, 15) i: p=.423) (randomized) Number 16, 16) i: Numbe 15, 15) i: % male (completers) % male 60.9%, 50%) i: 52.5 (c: 4.8,± 52.8 i: 5.7) ± (completers) BMI 32.5 (c: 5.0,± 34.1 i: 6.1) ± group: Target (randomized) % male (completers) Age 44 ± (c: 24.9#, 33.2#, 42 ± i: p=.287) (completers) BMI 28.6 (c: 2.7 ± (randomized) % male (completers) Age 45.27(c: p>.05) (completers) BMI 29.7 (c: 4.1#,± 28.9 i: 4.3#,± p>.05) Women. group: Target (randomized) % male (c: 47%,(c: 47%, i: p>.05) Age (randomized) Age 47.0 (c: 7.2,± 48.8 i: 6.1, ± p>.05) BMI: ≥30) group: Target

25 26 33

31 27

Australia Australia Staudter, Staudter, M. 2011 USA Tudor C. Locke, versus BPAI+ Comparison: Pal, S. 2009 Pal, S. 2011 Paschali, A. A. 2004 Canada 2005 USA

88 Do activity monitors increase physical activity in adults with overweight or obesity?

200) -

Unknown setting and self 1 day 1 day ame as c. ame as c.

1,000 steps/w, until - Wear the pedometer bothUse pedometer

200) 200). -

- : Daily: and steps PA : Daily steps, online logbook. online logbook. steps, Daily :

2) 10,0002) steps/d.

accumulated on accelerometer,

1 day. 1 1 day. 1

: Unknown : : Create weekly step goals,

: Pedometer (Yamax SW : Pedometer (model unknown) Wear the AM and monitor daily steps. daily monitor and AM the Wear Monitor daily steps, record on a record steps, daily Monitor

ial Cognitive Theory.

AM feedback AM calendar. AM: from recall Data monitoring. PA goal Data recall from AM: from recall Data AM AM: Instruction the on steps daily record and daily website. PA goal increasing by 500 reaching ultimate walking goal of 10,000 steps/d. feedback AM AM AM: Instruction goal for and calendar The website converted tracked steps into into steps tracked converted website The and journey virtual a on traveled distance motivation. increase to targets provided AM: from recall Data

- - Soc

: Daily steps and calendar. and steps Daily : : Activity: units : Daily steps and calendar. and steps Daily : : : : :

AM AM Based on Based Same as c. components: Contact/Behavioral on Based S components: Contact/Behavioral AM calendar. a on steps/d Record AM: Instruction PA goal: ≥10,000 steps/d. Feedback AM: from recall Data Same as c. advice/duration: Resources/dietary on Based S components: Contact/Behavioral Data recall from AM: from recall Data same as c. advice/duration: Resources/dietary Instruction AM: Instruction PA goal ≥3 times pw. Feedback monthly history displayed graphically on a computer computer a on graphically displayed history monthly the of Discussion diary. paper printed, was and data. accelerometer 28 days. AM: from recall Data Instruction AM: Instruction calendar. PA goal: Set small achievable goals, weekly increase to 1) walking ≥30 min/d and Same as c. advice/duration: Resources/dietary Feedback

- 5

Theory, Transtheoretical Transtheoretical Theory, management Goal setting and problem and setting Goal Encouragement, strategies for for strategies Encouragement, -

reported exercise.

- es website. es f the study. the f

rded weekly steps. steps. weekly rded

: Positive online feedback. : e

based learning. based - Dietary goals and information about a and goals information Dietary Non

- aseline, mid study and post study. study. post and study mid aseline, : Web : New Lifestyl Program Step manual. First : : Theory of Planned Behavior. Planned of Theory : Social: Cognitive BPAI : Online message board, three data collection data collection board, three message Online : : 4 group sessions in the first four weeks.

- - behavioralBasic self

sed on health behaviors. behaviors. health Resources advice: Dietary website. the through diet healthy 12 months: Duration on Based Contact sessions at b components: Behavioral board. message online the through improvement feedback Additional Resources : 12 weeks Duration Ba Model. Contact components: Behavioral solving exercises. Resources feedback Additional Dietary advice: Nutritional advice: Dietary guidance, same c. as Dietary advice: Dietary : 16 weeks Duration

l achievable goals like 10 min walks, gradually

Maintain normal diet. diet. normal Maintain

page manual with exercise instructions and self and instructions exercise with manual page -

: Recording: of total weekly steps. : Recording: of total weekly steps. : Paper diary with discussion of self of discussion with diary Paper : : : : Social: Cognitive Theory Axial accelerometer (Bio trainer), display turned off. : Review of PA guidelines at baseline. Discussion: at baseline about PA guidelines. M : : Set smal : ≥30 min/d walking on top of baseline activity. baseline on top walking of min/d ≥30 : risk walking or exercise of equivalent intensity ≥3 intensity of equivalent exercise or walking B risk : - list control control list - : A sealed pedometer. Control participants wore this sealed : A sealed pedometer. Participants reco : Tri change in in change Resources AM pedometer for 12 weeks with weekly recording. PA goal were set. goals min/d. No step ≥30 to weekly goal increase Feedback o duration for diet Maintain advice: Dietary : 12 weeks Duration Behavioral components: Behavioral in diet or physical physical or diet in activity (e.g., on information stress, hair loss, injury worksite prevention). careUsual "Usual Lifestyle", no advised. activity data three During collection sessions, 30 minutes were on spent nutritional in guidance, order to minimise biases. dietary Wait group on Based Contact on Based Contact components: Behavioral AM PA goal Feedback Resources advice: Dietary : 12 weeks. Duration on Based Contact and encountered problems goals, of discussion principles, solving. problem Behavioral components: Behavioral AM times pw. Feedback Resources None. advice: Dietary PA goal management.

: 106 : 60: (c: 30: (c: 32: (c: 30: (c:

: 0 : : 0 : 46.2: : 20.2 (c: 20.2: (c: 55.3: (c:

- : 48.1 ± 7.1 48.1 : ± : 32.5 ± 5.4 32.5 : ± 4.1* 29.3: ± : 33.3 ± 5.6 33.3 : ± 2.6* 29.3: ± : 50.0 ± 9.3 50.0: ± 5.2 52.7: ± 28.8* 43 ± : 9.1 43.5: ±

Military Women. Patients with with Patients Community Men.

BPAI

#, 29.9 i: 2.5#,±

± 8.4#,± 41.4 i: 9.8#, ±

r (randomized) : 52):

Unknown (inclusion: (inclusion: Unknown

54, I type 2 diabetes. 2 type sample (c: 32.6 (c: 6.2,± 32.3 i: 4.6, ± p=.915) group: Target beneficiaries. group: Target Target group: Target (randomized) Number (c: (completers) BMI (randomized) Number 30, 30) i: (completers) Age (randomized) Number 15, 15) i: p=.423) (randomized) Number 16, 16) i: (randomized) % male (completers) Age 45.27(c: p>.05) (completers) BMI 29.7 (c: 4.1#,± 28.9 i: 4.3#,± p>.05) Women. group: Target Numbe 15, 15) i: % male (completers) % male 27.3%, 13.3%) i: (completers) Age 50 ± (c: 8.4, 50 i: 10.1) ± (completers) % male 60.9%, 50%) i: 52.5 (c: 4.8,± 52.8 i: 5.7) ± (completers) BMI 32.5 (c: 5.0,± 34.1 i: 6.1) ± group: Target (randomized) % male (completers) Age 44 ± (c: 24.9#, 33.2#, 42 ± i: p=.287) (completers) BMI 28.6 (c: 2.7 ± (randomized) % male (c: 47%,(c: 47%, i: p>.05) Age (randomized) Age 47.0 (c: 7.2,± 48.8 i: 6.1, ± p>.05) BMI: ≥30) group: Target

25 26 33

31 27

Australia Staudter, Staudter, M. Tudor C. Locke, versus BPAI+ Comparison: Pal, S. 2009 Pal, S. 2011 Australia Paschali, A. A. 2011 USA 2004 Canada 2005 USA

89 Chapter 5

log. -

ssion. ssion. 200 (pedometer).

- Same as c. Same as c.

Based Behavioral Weight Weight Behavioral Based : Same as c. - Based Behavioral Weight Weight Behavioral Based - Daily in week 1,5,9. Unlimited. Unlimited. day. 1

No weekly meetings but weekly

n to 250 mins/w of exercise.

rol group), (Dietary DG Guidelines), EE Wear AM and upload data daily. daily. data upload and AM Wear Wear AM and digital display daily, daily, display digital and AM Wear Wear the AM, upload AM data on data upload AM AM, the Wear in steps daily record daily, AM Wear page handout summarizing benefits of of benefits summarizing handout page

- week calendar to use as step - DigiWalker SW ice/Duration Website.

- - - : Website. One: : Website. : Daily steps and EE, internet monitoring of PA PA of monitoring internet EE, and steps Daily : 9 : : Digital: display (real feedback), time website, : Website where PA data was shown, and food : : : : Social: Cognitive Theory. : :

300 min/w : Progressio: : Moderate intensity exercise increase from 20 Increase: daily average of steps by 400, each : Progressive engagement in moderate intensity

recall from AM: from recall .

uction AM: uction : Sensewear Pro armband (only weeks 1, 5 and 9). : Yamax watch). digital (armband, Fit BodyMedia : : BodyMedia Fit and additional digital display. display. digital additional and Fit BodyMedia :

Instruction AM: Instruction PA goal Dietary advice/duration: Dietary AM Contact/Behavioral components: same as c. components: Contact/Behavioral AM: Instruction Based on Based website. PA goal to 40 min/d, 5 d/w. Feedback other the during diaries paper 9), 5, 1, (week diet and weeks. AM: from recall Data Resources Same as c. advice/duration: Dietary of availability Continuous : except i1, as Same monitoring. internet and armband Pro Sensewear AM AM: Instruction calendar. PA the PA goal week. Feedback AM: from recall Data Resources PA. Dietary adv AM Standard Behavioral Weight Loss plus technology based based technology plus Loss Weight Behavioral Standard system (i1): on Based Same as c. components: Contact/Behavioral AM Instr daily. website the on data PA download PA goal PA 100 - Feedback weekly feedback from interventionist. Data Resources advice/duration: Dietary (i2): only system based Technology : except Same as i1, mailed behavioral lessons. At baseline 1 instruction and 1 weight loss information se Technology Intermittent (i1): Program Control Technology Continuous (i2): Program Control on Based same as c. components: Contact/Behavioral on Based same as c. components: Contact/Behavioral Feedback intake and weight was recorded, weekly written feedback from interventionist. AM: from recall Data Resources:

ied).

s (afters week 1500 kcal/d, 1800 kcal/week,

, PA (Physical Activity).

1800 kcal/d.

monitoring of diet and PA PA and diet of monitoring monitoring PA and diet, weekly and diet, PA monitoring

- -

Behavioral strategies (not specified). specified). (not strategies Behavioral

20% of total intake. of intake. 20% total

uce intake to 1500 -

page handout summarizing the benefits of of PA. benefits the summarizing handout page

reduce caloric intake to 1200 - - Red -

per diary (diet, PA and weight), weekly written written weekly weight), and PA (diet, diary per week calendar, with own recording of PA.

Person Behavioral Weight Control Program (c): Program Control Weight Behavioral Person - - - -

- - - Social Cognitive Theory (SCT). : : : One: : : None.

: Pa: : 9 : : Paper diary: self for : Paper diary: self for : : : : : : : 3 months. 3 : : Weekly: meetings (3 group meetings, 1 individual : A brief scripted statement endorsing the benefits of : 7 individualized counseling sessions. counseling individualized 7 : : Progressive engagement in moderate intensity PA to : Increase PA level by 10% each week. : Progression toward 250 mins/week of exercise. : Moderate: intensity exercise progressing from 20 to 40 back None. 300 min/w.

: None. : None: : None. ontact : Weekly group meetings. Duration (c): Loss Weight Behavioral Standard on Based Contact meeting p/m) components: Behavioral AM Feedback feedback from interventionist. PA goal 100 - Resources advice: Dietary total of calories. Elicit 20% to fats and and dietary energy deficit of 500 kcal/day. months. 6 : Duration In Standard on Based on Based Contact call phone Two min), (1 activity physical increased on Based C 1 and in week 5). - components: Behavioral Contact AM PA goal Feed specif (not approaches Behavioral components: Behavioral AM PA goal Resources Behavioral components: Constructs components: of SCT (notBehavioral specified). advice: Dietary AM: PA goal min/d during 5 d/w. Feedback : 9 weeks. Duration behaviors. Resources Reduce advice: energyDietary intake to 1200 - reduce saturated fat to : 12 weeks. Duration Feedback written feedback from interventionis Resources advice: Dietary : 6 months. 6 : Duration

: 51: (c: 58: (c: 94: (c: 29: (c:

: 13.7: : 1.6: 33.0,: 18:

: 33.7 ± 3.6 33.7 : ± : 33.1 ± 2.8 33.1 : ± 8.5 31.0 : ± 3.9 45.0: ± : 44.2 ± 8.7 44.2 : ± : 41.3 ± 8.7 41.3 : ± 40.9: ± 41.5: ±

Community Community a of Patients Community Community

5%, 17.6, i2:

edicine clinic. i: 38.0)i:

33.4 3.6,± p=.42)

sample. sample. Number (randomized) Number 17, 17, i1: 17) i2: (randomized) % male 0%, i1:(c: 23. p=.12) (randomized) Age 9.4,(45.1 ± 43.3 i1: 9.1, ± i2: 8.1,44.1 ± p=.85) (randomized) BMI 33.1 (c: 3.8,± 34.7 i1: 3.4, ± i2: Community group: Target sample. (randomized) Number 19, 19, i1: 19) i2: (randomized) Number 44, 50) i: (randomized) Number 14, 1 5) i: % male (randomized) % male (randomized) % male (randomized) % male Age (randomized) Age 27.3, (c: (randomized) Age (c: 40.2 (c: 8.0,± 41.1 i1: 8.3, ± 42.6i2: 10.0,± p=.71) (randomized) BMI 33.6 (c: 2.7,± 33.4 i1: 2.8, ± 32.6i2 : 2.7,± p=.51) group: Target (randomized) Age 44.313.4 (c: 13.8, ± 38.0 i: ± 12.4) (randomized) BMI 31.5 (c: 9.8,± 30.5 i: 7.3) ± group: Target family m (c:9.8* 46.1 9.1,± 38.7 i: ± 9.3, p=.04) (randomized) BMI Target group: Target

ations: AM (Activity Monitor), BDM (Behavioral Determinants Model), BPAI (Behavioral Physical Activity Intervention), c (cont

29 30 32 34

USA Pellegrini, C. A. 2012 USA Polzien, K. M. Stovitz, S. D. Unick, J. L. 2012 2007 USA 2005 USA (energy expenditure), i (intervention group), MVPA (Moderate to Vigorous Physical Activity) Physical Vigorous to (Moderate MVPA group), (intervention i expenditure), (energy 7.7.a section Handbook Cochrane in formula the using groups separate two from data the on based calculated was SD * Abbrevi

90 Do activity monitors increase physical activity in adults with overweight or obesity?

log. -

ssion. ssion. 200 (pedometer).

- Same as c. Same as c.

Based Behavioral Weight Weight Behavioral Based : Same as c. - Based Behavioral Weight Weight Behavioral Based - Unlimited. Daily in week 1,5,9. Unlimited.

1 day. 1

No weekly meetings but weekly

n to 250 mins/w of exercise.

rol group), (Dietary DG Guidelines), EE Wear AM and digital display daily, daily, display digital and AM Wear daily. data upload and AM Wear Wear the AM, upload AM data on data upload AM AM, the Wear Wear AM daily, record daily steps in in steps daily record daily, AM Wear page handout summarizing benefits of of benefits summarizing handout page

- week calendar to use as step - DigiWalker SW ice/Duration Website.

- - - : Website. : Website. : One: : Digital: display (real feedback), time website, : Daily steps and EE, internet monitoring of PA PA of monitoring internet EE, and steps Daily : : 9 : : Website where PA data was shown, and food : : : : : Social: Cognitive Theory.

300 min/w : Progressio: : Progressive engagement in moderate intensity : Moderate intensity exercise increase from 20 : Increase: daily average of steps by 400, each

recall from AM: from recall . uction AM: uction : Sensewear Pro armband (only weeks 1, 5 and 9). : BodyMedia Fit and additional digital display. display. digital additional and Fit BodyMedia : watch). digital (armband, Fit BodyMedia : : Yamax

Instruction AM: Instruction PA goal Dietary advice/duration: Dietary Feedback weekly feedback from interventionist. AM PA goal PA 100 - Data Resources advice/duration: Dietary (i2): only system based Technology same as c. components: Contact/Behavioral AM: Instruction Based on Based Same as c. components: Contact/Behavioral AM Instr daily. website the on data PA download : except Same as i1, mailed behavioral lessons. At baseline 1 instruction and 1 weight loss information se on Based website. PA goal to 40 min/d, 5 d/w. Feedback other the during diaries paper 9), 5, 1, (week diet and weeks. AM: from recall Data Resources Same as c. advice/duration: Dietary of availability Continuous : except i1, as Same monitoring. internet and armband Pro Sensewear AM Standard Behavioral Weight Loss plus technology based based technology plus Loss Weight Behavioral Standard system (i1): Technology Intermittent (i1): Program Control Technology Continuous (i2): Program Control on Based same as c. components: Contact/Behavioral AM AM: Instruction calendar. PA the PA goal week. Feedback AM: from recall Data Resources PA. Dietary adv on Based same as c. components: Contact/Behavioral Feedback intake and weight was recorded, weekly written feedback from interventionist. AM: from recall Data Resources:

5 ied).

s (afters week 1500 kcal/d, 1800 kcal/week,

, PA (Physical Activity).

1800 kcal/d.

monitoring of diet and PA PA and diet of monitoring monitoring PA and diet, weekly and diet, PA monitoring

- -

Behavioral strategies (not specified). specified). (not strategies Behavioral

20% of total intake. of intake. 20% total

uce intake to 1500 -

page handout summarizing the benefits of of PA. benefits the summarizing handout page reduce caloric intake to 1200 - - Red -

per diary (diet, PA and weight), weekly written written weekly weight), and PA (diet, diary per week calendar, with own recording of PA.

Person Behavioral Weight Control Program (c): Program Control Weight Behavioral Person - - - -

- - - Social Cognitive Theory (SCT). : : : One: : : None.

: Pa: 9 : : Paper diary: self for : Paper diary: self for : : : : : : : 3 months. 3 : : A brief scripted statement endorsing the benefits of : Weekly: meetings (3 group meetings, 1 individual : 7 individualized counseling sessions. counseling individualized 7 : : Progressive engagement in moderate intensity PA to : Increase PA level by 10% each week. : Progression toward 250 mins/week of exercise. : Moderate: intensity exercise progressing from 20 to 40 back None. 300 min/w.

: None. None: : None. ontact : Weekly group meetings. PA goal 100 - Resources advice: Dietary Duration (c): Loss Weight Behavioral Standard on Based AM Feedback feedback from interventionist. total of calories. Elicit 20% to fats and and dietary energy deficit of 500 kcal/day. months. 6 : Duration In Standard on Based on Based Contact call phone Two min), (1 activity physical increased 1 and in week 5). - components: Behavioral AM PA goal Feed Resources advice: Dietary : 9 weeks. Duration on Based C Contact meeting p/m) components: Behavioral Contact specif (not approaches Behavioral components: Behavioral AM PA goal Behavioral components: Constructs components: of SCT (notBehavioral specified). AM: PA goal min/d during 5 d/w. Feedback behaviors. Resources Reduce advice: energyDietary intake to 1200 - reduce saturated fat to : 12 weeks. Duration Feedback written feedback from interventionis Resources advice: Dietary : 6 months. 6 : Duration

: 51: (c: 58: (c: 94: (c: 29: (c:

: 33.0,: : 13.7: 1.6: 18:

: 31.0 ± 8.5 31.0 : ± : 33.7 ± 3.6 33.7 : ± 2.8 33.1 : ± 3.9 45.0: ± : 40.9 ± 40.9: ± : 44.2 ± 8.7 44.2 : ± 8.7 41.3 : ± 41.5: ±

Patients of a of Patients Community Community Community Community

5%, 17.6, i2:

edicine clinic. i: 38.0)i:

33.4 3.6,± p=.42) sample. sample. Number (randomized) Number 17, 17, i1: 17) i2: (randomized) Number 19, 19, i1: 19) i2: (randomized) Number 44, 50) i: (randomized) % male 27.3, (c: (randomized) Age 44.313.4 (c: 13.8, ± 38.0 i: ± 12.4) (randomized) BMI 31.5 (c: 9.8,± 30.5 i: 7.3) ± group: Target family m (randomized) Number 14, 1 5) i: % male (randomized) % male (randomized) % male (randomized) % male (c: 0%, i1:(c: 23. p=.12) (randomized) Age (randomized) Age (randomized) Age (45.1 ± 9.4,(45.1 ± 43.3 i1: 9.1, ± i2: 8.1,44.1 ± p=.85) (randomized) BMI 33.1 (c: 3.8,± 34.7 i1: 3.4, ± i2: Community group: Target sample. 40.2 (c: 8.0,± 41.1 i1: 8.3, ± 42.6i2: 10.0,± p=.71) (randomized) BMI 33.6 (c: 2.7,± 33.4 i1: 2.8, ± 32.6i2 : 2.7,± p=.51) group: Target (c:9.8* 46.1 9.1,± 38.7 i: ± 9.3, p=.04) (randomized) BMI Target group: Target

ations: AM (Activity Monitor), BDM (Behavioral Determinants Model), BPAI (Behavioral Physical Activity Intervention), c (cont

29 30 32 34

USA Pellegrini, C. A. Polzien, K. M. Stovitz, S. D. 2005 USA Unick, J. L. 2012 2012 USA 2007 USA (energy expenditure), i (intervention group), MVPA (Moderate to Vigorous Physical Activity) Physical Vigorous to (Moderate MVPA group), (intervention i expenditure), (energy 7.7.a section Handbook Cochrane in formula the using groups separate two from data the on based calculated was SD * Abbrevi

91 Chapter 5

BPAI+ versus BPAI-. A meta-analysis using the SMD for total MVPA minutes per time-unit was performed on three studies25, 26, 34 (Figure 4A). A positive (SMD 0.43, 95%CI 0.00-0.87) but not significant (p=.05) intervention effect estimate of moderate quality (Table 2) without heterogeneity (I2=0%, p=.66) was found. A meta-analysis using the Mean Difference (MD) for walking MET·minutes/week was performed on two studies25, 26 (Figure 4B). A significant (p=.002) positive (MD 282.00 walking MET·minutes/week, 95%CI 103.82-460.18) intervention effect estimate and low35 (I2=4%), not significant (p=.31) heterogeneity was found. The two studies included had a 100% female sample. One study33 reported insufficient data on walking MET·minutes/week to be included in meta-analysis. They used Figure 2. the nonparametric Mann-Whitney U Test and found a significant (p=.03) BPAI+ effect on Risk of bias graph for included studies (n=14). total walking MET·minutes/week compared to BPAI-. A meta-analysis using the SMD for physical activity kilocalories per week was performed on two studies29, 30 (Figure 4C). One Synthesis of results study30 that compared two relevant BPAI+ groups to a similar BPAI- group was entered BPAI+ versus WL/UC. A meta-analysis using the Standardized Mean Difference (SMD) for twice. A significant (p=.02) positive (SMD 0.45, 95%CI 0.07-0.83) intervention effect estimate 21-23, 33 steps per day was performed on four studies comparing BPAI+ to WL/UC (Figure 3A). without heterogeneity (I2=0%, p=.57) was found. Meta-analyses using MD were also 23 One study that compared two relevant intervention groups to waitlist controls was entered performed for studies comparing BPAI+ to BPAI- that reported outcomes for changes in body twice. Another study that described a non-significant (p=.167) positive effect on steps/day weight. All analyses found a positive but not significant intervention effect estimate with 31 could not be included as steps/day outcomes were only graphically displayed. A significant low-to-moderate35 but not significant heterogeneity. Three studies25, 29, 30 reported (p<.00001) positive (SMD 0.90, 95%CI 0.61-1.19) intervention effect estimate with outcomes for weight change in kilograms (MD -0.86, 95%CI -2.93-1.20, p=.41 and I2=45%, 35 2 moderate (I =49%) but not significant (p=0.10) heterogeneity was found. A meta-analysis p=.14), two30, 34 for relative weight change in percentage (MD -0.75, 95%CI -3.10-1.59, p=.53 22, 24, 28 using the SMD for total MVPA minutes per time-unit was performed on three studies and I2=46%, p=.16) and three25, 26, 29 for BMI change (MD -0.39, 95%CI -1.53-.75, p=.51 and comparing BPAI+ to WL/UC (Figure 3B). A significant (p=.01) positive (SMD 0.50, 95%CI 0.11- I2=27%, p=.26). 0.88) intervention effect estimate with substantial35 (I2=74%) and significant (p=.02) heterogeneity was found.

Figure 4. Forrest plots of (A) the standardized mean difference in minutes of moderate to vigorous physical activity, (B) Figure 3. the mean difference in walking MET-minutes per week, and (C) the standardized mean difference in physical Forrest plot of standardized mean differences of (A) steps per day and (B) total moderate to vigorous physical activity kilocalories per week in studies comparing patients receiving a behavioral physical activity intervention activity in studies comparing patients receiving a behavioral physical activity intervention with an activity with an activity monitor (BPAI+) with patients receiving a behavioral physical activity intervention without an monitor (BPAI1) to patients on a wait list or receiving usual care (WL/UC). activity monitor (BPAI-).

92 Do activity monitors increase physical activity in adults with overweight or obesity?

BPAI+ versus BPAI-. A meta-analysis using the SMD for total MVPA minutes per time-unit was performed on three studies25, 26, 34 (Figure 4A). A positive (SMD 0.43, 95%CI 0.00-0.87) but not significant (p=.05) intervention effect estimate of moderate quality (Table 2) without heterogeneity (I2=0%, p=.66) was found. A meta-analysis using the Mean Difference (MD) for walking MET·minutes/week was performed on two studies25, 26 (Figure 4B). A significant (p=.002) positive (MD 282.00 walking MET·minutes/week, 95%CI 103.82-460.18) intervention effect estimate and low35 (I2=4%), not significant (p=.31) heterogeneity was found. The two studies included had a 100% female sample. One study33 reported insufficient data on walking MET·minutes/week to be included in meta-analysis. They used Figure 2. the nonparametric Mann-Whitney U Test and found a significant (p=.03) BPAI+ effect on Risk of bias graph for included studies (n=14). total walking MET·minutes/week compared to BPAI-. A meta-analysis using the SMD for physical activity kilocalories per week was performed on two studies29, 30 (Figure 4C). One Synthesis of results study30 that compared two relevant BPAI+ groups to a similar BPAI- group was entered BPAI+ versus WL/UC. A meta-analysis using the Standardized Mean Difference (SMD) for twice. A significant (p=.02) positive (SMD 0.45, 95%CI 0.07-0.83) intervention effect estimate 21-23, 33 steps per day was performed on four studies comparing BPAI+ to WL/UC (Figure 3A). without heterogeneity (I2=0%, p=.57) was found. Meta-analyses using MD were also 5 23 One study that compared two relevant intervention groups to waitlist controls was entered performed for studies comparing BPAI+ to BPAI- that reported outcomes for changes in body twice. Another study that described a non-significant (p=.167) positive effect on steps/day weight. All analyses found a positive but not significant intervention effect estimate with 31 could not be included as steps/day outcomes were only graphically displayed. A significant low-to-moderate35 but not significant heterogeneity. Three studies25, 29, 30 reported (p<.00001) positive (SMD 0.90, 95%CI 0.61-1.19) intervention effect estimate with outcomes for weight change in kilograms (MD -0.86, 95%CI -2.93-1.20, p=.41 and I2=45%, 35 2 moderate (I =49%) but not significant (p=0.10) heterogeneity was found. A meta-analysis p=.14), two30, 34 for relative weight change in percentage (MD -0.75, 95%CI -3.10-1.59, p=.53 22, 24, 28 using the SMD for total MVPA minutes per time-unit was performed on three studies and I2=46%, p=.16) and three25, 26, 29 for BMI change (MD -0.39, 95%CI -1.53-.75, p=.51 and comparing BPAI+ to WL/UC (Figure 3B). A significant (p=.01) positive (SMD 0.50, 95%CI 0.11- I2=27%, p=.26). 0.88) intervention effect estimate with substantial35 (I2=74%) and significant (p=.02) heterogeneity was found.

93 Chapter 5

Table 2. The small number of studies included in this meta-analysis could not be amended. Summary of findings table for studies used in meta-analysis (BPAI+ vs. WL/UC and vs. BPAI-). Only 0.85% of the studies identified by the search strategy (14 out of 1,645) could be Outcome Number of Anticipated absolute effect of included as the search funnel was likely sensitive. Analysis on funnel plot asymmetry in order participants BPAI+ (95% CI) to detect publication bias was not feasible as the number of studies that could be included in

(studies) 19 Comparison: BPAI+ versus WL/UC the five individual meta-analyses was insufficient (<10). The search strategy identified Steps per day 417 (4 RCT’s) SMD 0.9 more (0.61 more to 1.19 seven conference abstracts,36-42 one research letter43 and one commentary article44 more) describing studies without a full article publication; these were excluded. Although Total MVPA minutes per week 651 (3 RCT’s) SMD 0.5 more (0.11 more to 0.88 publication bias appears to be present based on these excluded studies, their results are in more) Comparison: BPAI+ versus BPAI- accordance with the results of this systematic review and meta-analysis. Positive BPAI+ Total MVPA minutes per week 83 (3 RCT’s) SMD 0.43 more (0 more to 0.87 effects on steps per day have also been meta-analyzed in healthy adults,11 in adults with more) type 2 diabetes12, 16 and patients with musculoskeletal disorders,14 as well as being based on 54 (2 RCT’s) MD 282 more (103.82 more to (Brisk)walking minutes per week pooled data from all available populations.45 The small number of studies that could be 460.18 more) Kilocalories per week 110 (2 RCT’s) SMD 0.45 more (0.07 more to 0.83 included in our meta-analyses can be further explained by the apparent lack of consensus on more) a preferred physical activity outcome measure and the limited number of studies reporting Abbreviations: CI (Confidence Interval), MD (Mean Difference), SMD (Standardized Mean Difference). physical activity outcomes as 81 studies could not be included for those reasons (Figure 1). As a result, the effect sizes of the performed meta-analyses, particularly those of analyses comparing BPAI+ to BPAI-, were associated with relatively wide confidence intervals, Discussion meaning the accuracy of the estimated effect sizes are low. Therefore, future clinical studies are encouraged to report physical activity outcomes and preferably include objectively measured MVPA and steps per day as these have clinical relevance regarding both current Activity monitors may serve as a tool to enhance self-awareness of daily physical activity and international physical activity guidelines and the feedback provided by activity monitors. To to support individual behavioral physical activity interventions. To the authors’ knowledge, establish the isolated effect of activity monitors, future clinical studies should also consider this is the first systematic review (including meta-analysis) that describes the effects of a simply issuing their intervention participants an activity monitor with regular consumer behavioral physical activity intervention with an activity monitor (BPAI+) in adults with instructions or to compare a well-described BPAI+ to an identical intervention without the overweight or obesity. The first aim was to establish whether receiving a BPAI+ increases activity monitor by mimicking either consumer use or the added value for existing clinical physical activity in adults with overweight or obesity compared to both waitlist and usual programs. care (WL/UC). A positive effect of BPAI+ on physical activity was ascertained when compared to WL/UC, which demonstrates that offering adults with overweight or obesity a BPAI+ has When compared to WL/UC, the effect of BPAI+ on steps per day (SMD 0.90, 95%CI clinical relevance when increasing physical activity is targeted. However, clinical diversity 0.61-1.19) is larger than on MVPA (SMD 0.50, 95%CI 0.11-0.81). In the analysis comparing within the included populations and applied interventions resulted in statistical the effects BPAI+ to BPAI-, a positive trend on different MVPA outcomes was ascertained heterogeneity. This heterogeneity makes it difficult to attribute the causes of the positive with moderate SMD’s. Unfortunately, no studies could be pooled for meta-analysis effects of BPAI+, including the added value of activity monitors. The second aim was to comparing the effect of BPAI+ on steps per day to BPAI-. All steps per day outcomes were establish the added value of an activity monitor in existing interventions on increment of objectively measured, but two out of three studies included in both MVPA meta-analyses physical activity. A positive effect was determined based on a positive trend in the used a subjective measure. This increases the likelihood that the effect on steps per day is performed meta-analyses and results of individual studies comparing BPAI+ to BPAI-. The indeed larger than on MVPA because adults with obesity tend to overestimate their physical magnitude of this effect remains uncertain due to wide confidence intervals in all analyses as activity in subjective measures.46 Only one study22 measured both steps and MVPA within a result of the low number of studies that could be pooled for meta-analysis. Although the same study population. Therefore, it is unknown whether there are actual differences in conclusive evidence cannot be derived from these results, all results generally indicate a the effect of BPAI+ on steps per day and MVPA or if the differences were caused by the positive effect in favor of the use of activity monitors. Additional insight, study limitations presence of the various study populations and designs. For both MVPA and steps per day, and clinical implications will be discussed below. two broad accepted physical activity norms are available. The current American College of Sports Medicine (ACSM) guideline recommends that adults perform 30 minutes of MVPA per day for at least five days per week.47 Another norm prescribes 10,000 steps per day, which

94 Do activity monitors increase physical activity in adults with overweight or obesity?

(studies) 19 Comparison: BPAI+ versus WL/UC the five individual meta-analyses was insufficient (<10). The search strategy identified Steps per day 417 (4 RCT’s) SMD 0.9 more (0.61 more to 1.19 seven conference abstracts,36-42 one research letter43 and one commentary article44 more) describing studies without a full article publication; these were excluded. Although Total MVPA minutes per week 651 (3 RCT’s) SMD 0.5 more (0.11 more to 0.88 publication bias appears to be present based on these excluded studies, their results are in more) Comparison: BPAI+ versus BPAI- accordance with the results of this systematic review and meta-analysis. Positive BPAI+ Total MVPA minutes per week 83 (3 RCT’s) SMD 0.43 more (0 more to 0.87 effects on steps per day have also been meta-analyzed in healthy adults,11 in adults with more) type 2 diabetes12, 16 and patients with musculoskeletal disorders,14 as well as being based on 54 (2 RCT’s) MD 282 more (103.82 more to (Brisk)walking minutes per week pooled data from all available populations.45 The small number of studies that could be 460.18 more) Kilocalories per week 110 (2 RCT’s) SMD 0.45 more (0.07 more to 0.83 included in our meta-analyses can be further explained by the apparent lack of consensus on more) a preferred physical activity outcome measure and the limited number of studies reporting Abbreviations: CI (Confidence Interval), MD (Mean Difference), SMD (Standardized Mean Difference). physical activity outcomes as 81 studies could not be included for those reasons (Figure 1). 5 As a result, the effect sizes of the performed meta-analyses, particularly those of analyses comparing BPAI+ to BPAI-, were associated with relatively wide confidence intervals, Discussion meaning the accuracy of the estimated effect sizes are low. Therefore, future clinical studies are encouraged to report physical activity outcomes and preferably include objectively measured MVPA and steps per day as these have clinical relevance regarding both current Activity monitors may serve as a tool to enhance self-awareness of daily physical activity and international physical activity guidelines and the feedback provided by activity monitors. To to support individual behavioral physical activity interventions. To the authors’ knowledge, establish the isolated effect of activity monitors, future clinical studies should also consider this is the first systematic review (including meta-analysis) that describes the effects of a simply issuing their intervention participants an activity monitor with regular consumer behavioral physical activity intervention with an activity monitor (BPAI+) in adults with instructions or to compare a well-described BPAI+ to an identical intervention without the overweight or obesity. The first aim was to establish whether receiving a BPAI+ increases activity monitor by mimicking either consumer use or the added value for existing clinical physical activity in adults with overweight or obesity compared to both waitlist and usual programs. care (WL/UC). A positive effect of BPAI+ on physical activity was ascertained when compared to WL/UC, which demonstrates that offering adults with overweight or obesity a BPAI+ has When compared to WL/UC, the effect of BPAI+ on steps per day (SMD 0.90, 95%CI clinical relevance when increasing physical activity is targeted. However, clinical diversity 0.61-1.19) is larger than on MVPA (SMD 0.50, 95%CI 0.11-0.81). In the analysis comparing within the included populations and applied interventions resulted in statistical the effects BPAI+ to BPAI-, a positive trend on different MVPA outcomes was ascertained heterogeneity. This heterogeneity makes it difficult to attribute the causes of the positive with moderate SMD’s. Unfortunately, no studies could be pooled for meta-analysis effects of BPAI+, including the added value of activity monitors. The second aim was to comparing the effect of BPAI+ on steps per day to BPAI-. All steps per day outcomes were establish the added value of an activity monitor in existing interventions on increment of objectively measured, but two out of three studies included in both MVPA meta-analyses physical activity. A positive effect was determined based on a positive trend in the used a subjective measure. This increases the likelihood that the effect on steps per day is performed meta-analyses and results of individual studies comparing BPAI+ to BPAI-. The indeed larger than on MVPA because adults with obesity tend to overestimate their physical magnitude of this effect remains uncertain due to wide confidence intervals in all analyses as activity in subjective measures.46 Only one study22 measured both steps and MVPA within a result of the low number of studies that could be pooled for meta-analysis. Although the same study population. Therefore, it is unknown whether there are actual differences in conclusive evidence cannot be derived from these results, all results generally indicate a the effect of BPAI+ on steps per day and MVPA or if the differences were caused by the positive effect in favor of the use of activity monitors. Additional insight, study limitations presence of the various study populations and designs. For both MVPA and steps per day, and clinical implications will be discussed below. two broad accepted physical activity norms are available. The current American College of Sports Medicine (ACSM) guideline recommends that adults perform 30 minutes of MVPA per day for at least five days per week.47 Another norm prescribes 10,000 steps per day, which

95 Chapter 5

can be obtained by both low intensity physical activity and MVPA. The 10,000 steps per day Although additional well-designed studies are needed to improve the accuracy of recommendation corresponds to that of 30 minutes of MVPA per day when at least 3,000 of our findings, the apparent positive effect of activity monitor use in behavioral physical the taken steps are taken at MVPA intensity.48 Modern activity monitors are able to activity interventions in patients with overweight or obesity is consistent with findings in determine if steps are taken at a moderate-to-vigorous intensity. These modern activity other patient populations. Even more so, a natural transition to the use of an activity monitors also include BCTs such as social support, prompts/cues and rewards,10 all of which monitor appears to be ongoing as an increasing number of patients are using activity are all associated with positive physical activity changes in obese adults.49 Since modern monitors to self-monitor physical activity.55 Based on both the indications of a positive effect activity monitors can objectively measure the quantity and intensity of physical activity in of activity monitor use and the growing popularity of consumer-available activity monitors, the context of physical activity guidelines and incorporate BCTs that have proven to be clinicians are advised to develop methods to incorporate activity monitor feedback into their effective, they should be preferred over older models for future studies and clinical use. behavioral physical activity interventions. Specifically, we advise to use modern activity monitors that can determine if physical activity was executed at moderate to vigorous This study found no convincing evidence of an effect on changes in bodyweight intensity and include BCTs which have been substantiated to successfully improve physical when an activity monitor is added to the behavioral intervention (BPAI+ versus BPAI-). This activity. Several studies included in this meta-analysis also indicate the importance of goal- accords to that of other studies that found only modest effects of pedometer interventions setting that can preferably be monitored by the activity monitor feedback (e.g., 10,000 steps on weight loss in adults with both overweight or obesity and type 2 diabetes.15 A possible per day in pedometer interventions) and can be achieved by an incremental increase of explanation for the limited effect on body weight changes is that lower intensity physical physical activity.21, 22, 25, 26 Also, Morgan et al.24 mention that the increase in physical activity activity is associated with less impact on weight loss,50 although it remains ambiguous if the should be placed in the context of weight loss. For example, when using energy expenditure use of an activity monitor does indeed particularly increase low intensity physical activity data in combination with caloric intake data, the patient can learn about the caloric balance such as steps per day that are not taken at a moderate to vigorous intensity. However, more and amount of physical activity that is necessary to lose weight. One or more physical factors could have influenced the lack of effect found on body weight. First, only five studies activity consultations may be needed to discuss individual goal-setting, self-efficacy and described effects on body weight or BMI, limiting generalization. Second, in addition to physical activity barriers,21, 22 especially for adults who experience difficulties with self- inactivity, a variety of other factors are related to the onset and retaining of obesity, such as management. In addition, healthcare providers might simply advise their patients to use an ethnicity, pre-birth health, eating habits and sleeping patterns.51 It should also be noted, activity tracking application on their smartphones when purchasing an activity monitor is a however, that higher levels of physical activity and fitness are favorably associated with barrier.28 biomarkers and anthropometric markers in adults with overweight or obesity, even causing overweight adults to have similar cardiovascular profiles to their normal weight counterparts; the so called “fat but fit” paradigm.52 Because of the strong evidence for reduced rates of multiple disease outcomes and a variety of health benefits,53 increasing Conclusion physical activity has clinical relevance and should be stimulated for inactive adults with overweight or obesity, even if it has no effect on body weight. Behavioral physical activity interventions with an activity monitor increase physical activity in As described in Table 1, three studies included sub-populations with a specific adults with overweight or obesity. Also, adding an activity monitor to behavioral physical characteristic in patients awaiting bariatric surgery22, patients with diabetes mellitus type 233 activity interventions seems to increase the effect on physical activity, although current and military beneficiaries.31 While it cannot be precluded that these characteristics could evidence does not yet provide conclusive evidence for its effectiveness. have influenced the study outcomes, either positively (the characteristic as an extra motivator) or negatively (the characteristic as a limitation to increase physical activity levels), we do not expect that the inclusion of these participants has influenced the drawn conclusions. This systematic review excluded original contributions that aimed at increasing the physical activity levels of older adults. This exclusion was necessary because lower effect sizes are typical in intervention studies comprising older adults (mean age ≥60 years) due to a lower compliance to BCTs that are often applied in BPAI+, such as setting behavioral goals, prompting self-monitoring of behavior, planning for relapses, providing normative information and providing feedback on performance.54

96 Do activity monitors increase physical activity in adults with overweight or obesity?

can be obtained by both low intensity physical activity and MVPA. The 10,000 steps per day Although additional well-designed studies are needed to improve the accuracy of recommendation corresponds to that of 30 minutes of MVPA per day when at least 3,000 of our findings, the apparent positive effect of activity monitor use in behavioral physical the taken steps are taken at MVPA intensity.48 Modern activity monitors are able to activity interventions in patients with overweight or obesity is consistent with findings in determine if steps are taken at a moderate-to-vigorous intensity. These modern activity other patient populations. Even more so, a natural transition to the use of an activity monitors also include BCTs such as social support, prompts/cues and rewards,10 all of which monitor appears to be ongoing as an increasing number of patients are using activity are all associated with positive physical activity changes in obese adults.49 Since modern monitors to self-monitor physical activity.55 Based on both the indications of a positive effect activity monitors can objectively measure the quantity and intensity of physical activity in of activity monitor use and the growing popularity of consumer-available activity monitors, the context of physical activity guidelines and incorporate BCTs that have proven to be clinicians are advised to develop methods to incorporate activity monitor feedback into their effective, they should be preferred over older models for future studies and clinical use. behavioral physical activity interventions. Specifically, we advise to use modern activity monitors that can determine if physical activity was executed at moderate to vigorous This study found no convincing evidence of an effect on changes in bodyweight intensity and include BCTs which have been substantiated to successfully improve physical when an activity monitor is added to the behavioral intervention (BPAI+ versus BPAI-). This activity. Several studies included in this meta-analysis also indicate the importance of goal- accords to that of other studies that found only modest effects of pedometer interventions setting that can preferably be monitored by the activity monitor feedback (e.g., 10,000 steps on weight loss in adults with both overweight or obesity and type 2 diabetes.15 A possible per day in pedometer interventions) and can be achieved by an incremental increase of explanation for the limited effect on body weight changes is that lower intensity physical physical activity.21, 22, 25, 26 Also, Morgan et al.24 mention that the increase in physical activity activity is associated with less impact on weight loss,50 although it remains ambiguous if the 5 should be placed in the context of weight loss. For example, when using energy expenditure use of an activity monitor does indeed particularly increase low intensity physical activity data in combination with caloric intake data, the patient can learn about the caloric balance such as steps per day that are not taken at a moderate to vigorous intensity. However, more and amount of physical activity that is necessary to lose weight. One or more physical factors could have influenced the lack of effect found on body weight. First, only five studies activity consultations may be needed to discuss individual goal-setting, self-efficacy and described effects on body weight or BMI, limiting generalization. Second, in addition to physical activity barriers,21, 22 especially for adults who experience difficulties with self- inactivity, a variety of other factors are related to the onset and retaining of obesity, such as management. In addition, healthcare providers might simply advise their patients to use an ethnicity, pre-birth health, eating habits and sleeping patterns.51 It should also be noted, activity tracking application on their smartphones when purchasing an activity monitor is a however, that higher levels of physical activity and fitness are favorably associated with barrier.28 biomarkers and anthropometric markers in adults with overweight or obesity, even causing overweight adults to have similar cardiovascular profiles to their normal weight counterparts; the so called “fat but fit” paradigm.52 Because of the strong evidence for reduced rates of multiple disease outcomes and a variety of health benefits,53 increasing Conclusion physical activity has clinical relevance and should be stimulated for inactive adults with overweight or obesity, even if it has no effect on body weight. Behavioral physical activity interventions with an activity monitor increase physical activity in As described in Table 1, three studies included sub-populations with a specific adults with overweight or obesity. Also, adding an activity monitor to behavioral physical characteristic in patients awaiting bariatric surgery22, patients with diabetes mellitus type 233 activity interventions seems to increase the effect on physical activity, although current and military beneficiaries.31 While it cannot be precluded that these characteristics could evidence does not yet provide conclusive evidence for its effectiveness. have influenced the study outcomes, either positively (the characteristic as an extra motivator) or negatively (the characteristic as a limitation to increase physical activity levels), we do not expect that the inclusion of these participants has influenced the drawn conclusions. This systematic review excluded original contributions that aimed at increasing the physical activity levels of older adults. This exclusion was necessary because lower effect sizes are typical in intervention studies comprising older adults (mean age ≥60 years) due to a lower compliance to BCTs that are often applied in BPAI+, such as setting behavioral goals, prompting self-monitoring of behavior, planning for relapses, providing normative information and providing feedback on performance.54

97 Chapter 5

References health outcomes: a 12-week randomized controlled trial. International Journal of Behavioral Nutrition and Physical Activity. 2008;5. 22. Bond DS, Vithiananthan S, Thomas JG, Trautvetter J, Unick JL, Jakicic JM, et al. Bari-Active: a randomized controlled trial of a preoperative intervention to increase physical activity in bariatric 1. Swinburn BA, Sacks G, Hall KD, McPherson K, Finegood DT, Moodie ML, et al. The global obesity surgery patients. Surgery for obesity and related diseases : official journal of the American Society for pandemic: shaped by global drivers and local environments. Lancet. 2011;378(9793):804-14. Bariatric Surgery. 2014;11(1):169-77. 2. Wang YC, McPherson K, Marsh T, Gortmaker SL, Brown M. Health and economic burden of the 23. Morgan PJ, Callister R, Collins CE, Plotnikoff RC, Young MD, Berry N, et al. The SHED-IT community projected obesity trends in the USA and the UK. Lancet. 2011;378(9793):815-25. trial: a randomized controlled trial of internet- and paper-based weight loss programs tailored for 3. Gortmaker SL, Swinburn BA, Levy D, Carter R, Mabry PL, Finegood DT, et al. Changing the future of overweight and obese men. Annals of behavioral medicine : a publication of the Society of Behavioral obesity: science, policy, and action. Lancet. 2011;378(9793):838-47. Medicine. 2013;45(2):139-52. 4. Peirson L, Douketis J, Ciliska D, Fitzpatrick-Lewis D, Ali MU, Raina P. Treatment for overweight and 24. Morgan PJ, Collins CE, Plotnikoff RC, Cook AT, Berthon B, Mitchell S, et al. Efficacy of a workplace- obesity in adult populations: a systematic review and meta-analysis. CMAJ open. 2014;2(4):E306-17. based weight loss program for overweight male shift workers: the Workplace POWER (Preventing 5. Swift DL, Johannsen NM, Lavie CJ, Earnest CP, Church TS. The role of exercise and physical activity in Obesity Without Eating like a Rabbit) randomized controlled trial. Preventive medicine. weight loss and maintenance. Prog Cardiovasc Dis. 2014;56(4):441-7. 2011;52(5):317-25. 6. Ramage S, Farmer A, Eccles KA, McCargar L. Healthy strategies for successful weight loss and weight 25. Pal S, Cheng C, Egger G, Binns C, Donovan R. Using pedometers to increase physical activity in maintenance: a systematic review. Applied physiology, nutrition, and metabolism = Physiologie overweight and obese women: a pilot study. BMC public health. 2009;9:309. appliquee, nutrition et metabolisme. 2014;39(1):1-20. 26. Pal S, Cheng C, Ho S. The effect of two different health messages on physical activity levels and health 7. Tudor-Locke C. Manpo-Kei: The Art and Science of Step Counting: How to Be Naturally Active and Lose in sedentary overweight, middle-aged women. BMC public health. 2011;11:204. Weight: Trafford Publishing; 2003. 27. Paschali AA, Goodrick GK, Kalantzi-Azizi A, Papadatou D, Balasubramanyam A. Accelerometer feedback 8. Appelboom G, Yang AH, Christophe BR, Bruce EM, Slomian J, Bruyere O, et al. The promise of wearable to promote physical activity in adults with type 2 diabetes: a pilot study. Perceptual and motor skills. activity sensors to define patient recovery. Journal of clinical neuroscience : official journal of the 2005;100(1):61-8. Neurosurgical Society of Australasia. 2014;21(7):1089-93. 28. Patrick K, Calfas KJ, Norman GJ, Rosenberg D, Zabinski MF, Sallis JF, et al. Outcomes of a 12-month 9. Kooiman TJM, Dontje ML, Sprenger SR, Krijnen WP, van der Schans CP, de Groot M. Reliability and web-based intervention for overweight and obese men. Annals of behavioral medicine : a publication validity of ten consumer activity trackers. BMC Sports Science, Medicine and Rehabilitation. of the Society of Behavioral Medicine. 2011;42(3):391-401. 2015;7(1):1-11. 29. Pellegrini CA, Verba SD, Otto AD, Helsel DL, Davis KK, Jakicic JM. The comparison of a technology- 10. Lyons EJ, Lewis ZH, Mayrsohn BG, Rowland JL. Behavior change techniques implemented in electronic based system and an in-person behavioral weight loss intervention. Obesity (Silver Spring, Md). lifestyle activity monitors: a systematic content analysis. Journal of medical Internet research. 2012;20(2):356-63. 2014;16(8):e192. 30. Polzien KM, Jakicic JM, Tate DF, Otto AD. The efficacy of a technology-based system in a short-term 11. Bravata DM, Smith-Spangler C, Sundaram V, Gienger AL, Lin N, Lewis R, et al. Using pedometers to behavioral weight loss intervention. Obesity (Silver Spring, Md). 2007;15(4):825-30. increase physical activity and improve health: a systematic review. Jama. 2007;298(19):2296-304. 31. Staudter M, Dramiga S, Webb L, Hernandez D, Cole R. Effectiveness of pedometer use in motivating 12. Qiu S, Cai X, Chen X, Yang B, Sun Z. Step counter use in type 2 diabetes: a meta-analysis of randomized active duty and other military healthcare beneficiaries to walk more. US Army Medical Department controlled trials. BMC medicine. 2014;12:36. journal. 2011:108-19. 13. Richardson CR, Newton TL, Abraham JJ, Sen A, Jimbo M, Swartz AM. A meta-analysis of pedometer- 32. Stovitz SD, VanWormer JJ, Center BA, Bremer KL. Pedometers as a means to increase ambulatory based walking interventions and weight loss. Annals of family medicine. 2008;6(1):69-77. activity for patients seen at a family medicine clinic. The Journal of the American Board of Family 14. Mansi S, Milosavljevic S, Baxter GD, Tumilty S, Hendrick P. A systematic review of studies using Practice. 2005;18(5):335-43. pedometers as an intervention for musculoskeletal diseases. BMC musculoskeletal disorders. 33. Tudor-Locke C, Bell RC, Myers AM, Harris SB, Ecclestone NA, Lauzon N, et al. Controlled outcome 2014;15:231. evaluation of the First Step Program: a daily physical activity intervention for individuals with type II 15. Cai X, Qiu SH, Yin H, Sun ZL, Ju CP, Zugel M, et al. Pedometer intervention and weight loss in diabetes. International journal of obesity and related metabolic disorders : journal of the International overweight and obese adults with Type 2 diabetes: a meta-analysis. Diabetic medicine : a journal of Association for the Study of Obesity [Internet]. 2004; 28(1):[113-9 pp.]. Available from: the British Diabetic Association. 2016. http://onlinelibrary.wiley.com/o/cochrane/clcentral/articles/355/CN-00473355/frame.html 16. Vaes AW, Cheung A, Atakhorrami M, Groenen MT, Amft O, Franssen FM, et al. Effect of 'activity http://www.nature.com/ijo/journal/v28/n1/pdf/0802485a.pdf. monitor-based' counseling on physical activity and health-related outcomes in patients with chronic 34. Unick JL, O'Leary KC, Bond DS, Wing RR. Physical activity enhancement to a behavioral weight loss diseases: A systematic review and meta-analysis. Ann Med. 2013;45(5-6):397-412. program for severely obese individuals: A preliminary investigation. ISRN obesity. 2012;2012. 17. Moher D, Shamseer L, Clarke M, Ghersi D, Liberati A, Petticrew M, et al. Preferred reporting items for 35. Schünemann HJ, Brożek J, Guyatt GH, Oxman AD. Handbook for grading the quality of evidence and systematic review and meta-analysis protocols (PRISMA-P) 2015 statement. Systematic reviews. the strength of recommendations using the GRADE approach. 2013. Available from: 2015;4:1. http://www.guidelinedevelopment.org/handbook/. 18. Does an activity monitor based intervention increase daily physical activity of adults with overweight 36. Aguiar E, Morgan P, Collins C, Plotnikoff R, Callister R. Improvements in weight, HbA1C and fitness or obesity? A systematic review and meta-analysis. [Internet]. PROSPERO 2015:CRD42015024086. following lifestyle intervention: The PULSE trial for type 2 diabetes prevention in men. Journal of Available from: http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42015024086. Science and Medicine in Sport. 2014;18:e68. 19. Higgins JPTG, S. Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [updated 37. Ahmed SM, Sho S, Changchien E, Wu E, Arslan Z, Morton JM. Do pedometers increase physical activity March 2011]: The Cochrane Collaboration; 2011. Available from: http://www.cochrane- following RYGB? Obesity (Silver Spring, Md) [Internet]. 2011; 19:[S57 p.]. Available from: handbook.org/. http://onlinelibrary.wiley.com/o/cochrane/clcentral/articles/825/CN-01034825/frame.html 20. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta- http://onlinelibrary.wiley.com/store/10.1038/oby.2011.222/asset/oby.2011.222.pdf?v=1&t=i6gyv7ze analyses: the PRISMA statement. BMJ (Clinical research ed). 2009;339:b2535. &s=06924ad11d9826d147fe6a469b5b7936456e3b4e. 21. Baker G, Gray SR, Wright A, Fitzsimons C, Nimmo M, Lowry R, et al. The effect of a pedometer-based 38. Alduhishy A, Baxendale R. 10,000 Step per day programme among Saudi Arabian overweight. community walking intervention "Walking for Wellbeing in the West" on physical activity levels and Physiotherapy (United Kingdom). 2011;97:eS45-eS6.

98 Do activity monitors increase physical activity in adults with overweight or obesity?

References health outcomes: a 12-week randomized controlled trial. International Journal of Behavioral Nutrition and Physical Activity. 2008;5. 22. Bond DS, Vithiananthan S, Thomas JG, Trautvetter J, Unick JL, Jakicic JM, et al. Bari-Active: a randomized controlled trial of a preoperative intervention to increase physical activity in bariatric 1. Swinburn BA, Sacks G, Hall KD, McPherson K, Finegood DT, Moodie ML, et al. The global obesity surgery patients. Surgery for obesity and related diseases : official journal of the American Society for pandemic: shaped by global drivers and local environments. Lancet. 2011;378(9793):804-14. Bariatric Surgery. 2014;11(1):169-77. 2. Wang YC, McPherson K, Marsh T, Gortmaker SL, Brown M. Health and economic burden of the 23. Morgan PJ, Callister R, Collins CE, Plotnikoff RC, Young MD, Berry N, et al. The SHED-IT community projected obesity trends in the USA and the UK. Lancet. 2011;378(9793):815-25. trial: a randomized controlled trial of internet- and paper-based weight loss programs tailored for 3. Gortmaker SL, Swinburn BA, Levy D, Carter R, Mabry PL, Finegood DT, et al. Changing the future of overweight and obese men. Annals of behavioral medicine : a publication of the Society of Behavioral obesity: science, policy, and action. Lancet. 2011;378(9793):838-47. Medicine. 2013;45(2):139-52. 4. Peirson L, Douketis J, Ciliska D, Fitzpatrick-Lewis D, Ali MU, Raina P. Treatment for overweight and 24. Morgan PJ, Collins CE, Plotnikoff RC, Cook AT, Berthon B, Mitchell S, et al. Efficacy of a workplace- obesity in adult populations: a systematic review and meta-analysis. CMAJ open. 2014;2(4):E306-17. based weight loss program for overweight male shift workers: the Workplace POWER (Preventing 5. Swift DL, Johannsen NM, Lavie CJ, Earnest CP, Church TS. The role of exercise and physical activity in Obesity Without Eating like a Rabbit) randomized controlled trial. Preventive medicine. weight loss and maintenance. Prog Cardiovasc Dis. 2014;56(4):441-7. 2011;52(5):317-25. 6. Ramage S, Farmer A, Eccles KA, McCargar L. Healthy strategies for successful weight loss and weight 25. Pal S, Cheng C, Egger G, Binns C, Donovan R. Using pedometers to increase physical activity in maintenance: a systematic review. Applied physiology, nutrition, and metabolism = Physiologie overweight and obese women: a pilot study. BMC public health. 2009;9:309. appliquee, nutrition et metabolisme. 2014;39(1):1-20. 26. Pal S, Cheng C, Ho S. The effect of two different health messages on physical activity levels and health 7. Tudor-Locke C. Manpo-Kei: The Art and Science of Step Counting: How to Be Naturally Active and Lose in sedentary overweight, middle-aged women. BMC public health. 2011;11:204. Weight: Trafford Publishing; 2003. 27. Paschali AA, Goodrick GK, Kalantzi-Azizi A, Papadatou D, Balasubramanyam A. Accelerometer feedback 8. Appelboom G, Yang AH, Christophe BR, Bruce EM, Slomian J, Bruyere O, et al. The promise of wearable to promote physical activity in adults with type 2 diabetes: a pilot study. Perceptual and motor skills. 5 activity sensors to define patient recovery. Journal of clinical neuroscience : official journal of the 2005;100(1):61-8. Neurosurgical Society of Australasia. 2014;21(7):1089-93. 28. Patrick K, Calfas KJ, Norman GJ, Rosenberg D, Zabinski MF, Sallis JF, et al. Outcomes of a 12-month 9. Kooiman TJM, Dontje ML, Sprenger SR, Krijnen WP, van der Schans CP, de Groot M. Reliability and web-based intervention for overweight and obese men. Annals of behavioral medicine : a publication validity of ten consumer activity trackers. BMC Sports Science, Medicine and Rehabilitation. of the Society of Behavioral Medicine. 2011;42(3):391-401. 2015;7(1):1-11. 29. Pellegrini CA, Verba SD, Otto AD, Helsel DL, Davis KK, Jakicic JM. The comparison of a technology- 10. Lyons EJ, Lewis ZH, Mayrsohn BG, Rowland JL. Behavior change techniques implemented in electronic based system and an in-person behavioral weight loss intervention. Obesity (Silver Spring, Md). lifestyle activity monitors: a systematic content analysis. Journal of medical Internet research. 2012;20(2):356-63. 2014;16(8):e192. 30. Polzien KM, Jakicic JM, Tate DF, Otto AD. The efficacy of a technology-based system in a short-term 11. Bravata DM, Smith-Spangler C, Sundaram V, Gienger AL, Lin N, Lewis R, et al. Using pedometers to behavioral weight loss intervention. Obesity (Silver Spring, Md). 2007;15(4):825-30. increase physical activity and improve health: a systematic review. Jama. 2007;298(19):2296-304. 31. Staudter M, Dramiga S, Webb L, Hernandez D, Cole R. Effectiveness of pedometer use in motivating 12. Qiu S, Cai X, Chen X, Yang B, Sun Z. Step counter use in type 2 diabetes: a meta-analysis of randomized active duty and other military healthcare beneficiaries to walk more. US Army Medical Department controlled trials. BMC medicine. 2014;12:36. journal. 2011:108-19. 13. Richardson CR, Newton TL, Abraham JJ, Sen A, Jimbo M, Swartz AM. A meta-analysis of pedometer- 32. Stovitz SD, VanWormer JJ, Center BA, Bremer KL. Pedometers as a means to increase ambulatory based walking interventions and weight loss. Annals of family medicine. 2008;6(1):69-77. activity for patients seen at a family medicine clinic. The Journal of the American Board of Family 14. Mansi S, Milosavljevic S, Baxter GD, Tumilty S, Hendrick P. A systematic review of studies using Practice. 2005;18(5):335-43. pedometers as an intervention for musculoskeletal diseases. BMC musculoskeletal disorders. 33. Tudor-Locke C, Bell RC, Myers AM, Harris SB, Ecclestone NA, Lauzon N, et al. Controlled outcome 2014;15:231. evaluation of the First Step Program: a daily physical activity intervention for individuals with type II 15. Cai X, Qiu SH, Yin H, Sun ZL, Ju CP, Zugel M, et al. Pedometer intervention and weight loss in diabetes. International journal of obesity and related metabolic disorders : journal of the International overweight and obese adults with Type 2 diabetes: a meta-analysis. Diabetic medicine : a journal of Association for the Study of Obesity [Internet]. 2004; 28(1):[113-9 pp.]. Available from: the British Diabetic Association. 2016. http://onlinelibrary.wiley.com/o/cochrane/clcentral/articles/355/CN-00473355/frame.html 16. Vaes AW, Cheung A, Atakhorrami M, Groenen MT, Amft O, Franssen FM, et al. Effect of 'activity http://www.nature.com/ijo/journal/v28/n1/pdf/0802485a.pdf. monitor-based' counseling on physical activity and health-related outcomes in patients with chronic 34. Unick JL, O'Leary KC, Bond DS, Wing RR. Physical activity enhancement to a behavioral weight loss diseases: A systematic review and meta-analysis. Ann Med. 2013;45(5-6):397-412. program for severely obese individuals: A preliminary investigation. ISRN obesity. 2012;2012. 17. Moher D, Shamseer L, Clarke M, Ghersi D, Liberati A, Petticrew M, et al. Preferred reporting items for 35. Schünemann HJ, Brożek J, Guyatt GH, Oxman AD. Handbook for grading the quality of evidence and systematic review and meta-analysis protocols (PRISMA-P) 2015 statement. Systematic reviews. the strength of recommendations using the GRADE approach. 2013. Available from: 2015;4:1. http://www.guidelinedevelopment.org/handbook/. 18. Does an activity monitor based intervention increase daily physical activity of adults with overweight 36. Aguiar E, Morgan P, Collins C, Plotnikoff R, Callister R. Improvements in weight, HbA1C and fitness or obesity? A systematic review and meta-analysis. [Internet]. PROSPERO 2015:CRD42015024086. following lifestyle intervention: The PULSE trial for type 2 diabetes prevention in men. Journal of Available from: http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42015024086. Science and Medicine in Sport. 2014;18:e68. 19. Higgins JPTG, S. Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [updated 37. Ahmed SM, Sho S, Changchien E, Wu E, Arslan Z, Morton JM. Do pedometers increase physical activity March 2011]: The Cochrane Collaboration; 2011. Available from: http://www.cochrane- following RYGB? Obesity (Silver Spring, Md) [Internet]. 2011; 19:[S57 p.]. Available from: handbook.org/. http://onlinelibrary.wiley.com/o/cochrane/clcentral/articles/825/CN-01034825/frame.html 20. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta- http://onlinelibrary.wiley.com/store/10.1038/oby.2011.222/asset/oby.2011.222.pdf?v=1&t=i6gyv7ze analyses: the PRISMA statement. BMJ (Clinical research ed). 2009;339:b2535. &s=06924ad11d9826d147fe6a469b5b7936456e3b4e. 21. Baker G, Gray SR, Wright A, Fitzsimons C, Nimmo M, Lowry R, et al. The effect of a pedometer-based 38. Alduhishy A, Baxendale R. 10,000 Step per day programme among Saudi Arabian overweight. community walking intervention "Walking for Wellbeing in the West" on physical activity levels and Physiotherapy (United Kingdom). 2011;97:eS45-eS6.

99 Chapter 5

39. Miyachi M, Ohmori Y, Morita A, Aiba N, Watanabe S. Effects of pedometer-based physical activity intervention on abdominal fat and blood pressure: Saku communitybased randomized crossover intervention study. Journal of Clinical Hypertension. 2010;12:A14. 40. Moon YJ, Park SW, Oh KW, Lee WY, Park CY, Rhee EJ, et al. The effects of exercise education with accelerometer on glucose control, lipid profile and obesity in type 2 diabetic patients. Diabetes. 2013;62:A624. 41. Scanlan B, Conroy MB, Tudorascu DL, Karpov I, Hess R, Fischer G, et al. Association of adherence measures with physical activity outcomes in an online weight loss trial: Results from the ocelot study. Journal of General Internal Medicine. 2014;29:S32-S3. 42. Yates T, Davies M, Gorely T, Bull F, Troughton J, Mandalia P, et al. Twelve-month follow-up from the PREPARE (Prediabetes Risk Education and Physical Activity Recommendation and Encouragement) programme study: A randomized controlled trial. Diabetic Medicine. 2009;26:17. 43. Dicken-Kano R, Bell MM. Pedometers as a means to increase walking and achieve weight loss. Journal of the American Board of Family Medicine [Internet]. 2006; 19(5):[524-5 pp.]. Available from: http://onlinelibrary.wiley.com/o/cochrane/clcentral/articles/992/CN-00571992/frame.html. 44. Mateo KF, Jay M. Access to a behavioral weight loss website with or without group sessions increased weight loss in statewide campaign. Journal of Clinical Outcomes Management. 2014;21(8):345-8. 45. Kang M, Marshall SJ, Barreira TV, Lee JO. Effect of pedometer-based physical activity interventions: a meta-analysis. Research quarterly for exercise and sport. 2009;80(3):648-55. 46. Tully MA, Panter J, Ogilvie D. Individual characteristics associated with mismatches between self- reported and accelerometer-measured physical activity. PloS one. 2014;9(6):e99636. 47. Medicine ACoS. ACSM's guidelines for exercise testing and prescription: Lippincott Williams & Wilkins; 2013. 48. Tudor-Locke C, Craig CL, Brown WJ, Clemes SA, De Cocker K, Giles-Corti B, et al. How many steps/day are enough? For adults. The international journal of behavioral nutrition and physical activity. 2011;8:79. 49. Olander EK, Fletcher H, Williams S, Atkinson L, Turner A, French DP. What are the most effective techniques in changing obese individuals' physical activity self-efficacy and behaviour: a systematic review and meta-analysis. The international journal of behavioral nutrition and physical activity. 2013;10:29. 50. Pellegrini CA, Song J, Chang R, Semanik PA, Lee J, Ehrlich-Jones L, et al. Change in Physical Activity and Sedentary Time Associated With 2-Year Weight Loss in Obese Adults With Osteoarthritis. Journal of physical activity & health. 2015. 51. Wahi G, Anand SS. Race/Ethnicity, Obesity, and Related Cardio-Metabolic Risk Factors: A Life-Course Perspective. Curr Cardiovasc Risk Rep. 2013;7:326-35. 52. Loprinzi P, Smit E, Lee H, Crespo C, Andersen R, Blair SN. The "fit but fat" paradigm addressed using accelerometer-determined physical activity data. N Am J Med Sci. 2014;6(7):295-301. 53. Lee IM, Shiroma EJ, Lobelo F, Puska P, Blair SN, Katzmarzyk PT. Effect of physical inactivity on major non-communicable diseases worldwide: an analysis of burden of disease and life expectancy. Lancet. 2012;380(9838):219-29. 54. French DP, Olander EK, Chisholm A, Mc Sharry J. Which behaviour change techniques are most effective at increasing older adults' self-efficacy and physical activity behaviour? A systematic review. Annals of behavioral medicine : a publication of the Society of Behavioral Medicine. 2014;48(2):225- 34. 55. Fawcett T. Mining the Quantified Self: Personal Knowledge Discovery as a Challenge for Data Science. Big Data. 2015;3(4):249-66.

100

Chapter 6 | Self-tracking of physical activity in people with type 2 diabetes - a randomized controlled trial

Thea J.M. Kooiman Martijn de Groot Klaas Hoogenberg Wim P. Krijnen Cees P. van der Schans Adriaan Kooy

Computers, Informatics, Nursing (2018) 36(7): 340-349

Chapter 6 | Self-tracking of physical activity in people with type 2 diabetes - a randomized controlled trial

Thea J.M. Kooiman Martijn de Groot Klaas Hoogenberg Wim P. Krijnen Cees P. van der Schans Adriaan Kooy

Computers, Informatics, Nursing (2018) 36(7): 340-349

Chapter 6

Abstract Introduction

Aim More than 400 million people worldwide have diabetes carrying an increased risk of 1 This study aimed to determine the efficacy of an online self-tracking program on physical cardiovascular disease, cancer, and dementia. Physical inactivity and being overweight due activity, HbA1c, and other health measures in patients with type 2 diabetes. to an unhealthy lifestyle are key factors in both the onset as well as the progression of type 2 diabetes.2,3 Increasing physical activity and the adoption of a healthy lifestyle are essential Methods for preventing long term complications and co-morbidity as this improves glycemic control Seventy-two patients with type 2 diabetes were randomized into an intervention or control and reduces weight.3–6 Therefore, stimulating physical activity is of great importance within group. All participants received usual care. The intervention group received an activity daily clinical practice for people with type 2 diabetes, especially in nursing care. Physical tracker (Fitbit Zip) connected to an online lifestyle program. Physical activity was analyzed in activity guidelines for this population recommend engagement in progressive moderate to average steps per day from week 0 until week 12. Health outcome measurements occurred vigorous resistance training in addition to a minimum of 150 minutes of moderate to in both groups at baseline (T0) and after 13 weeks (T1). vigorous physical activity (MVPA) per week and avoidance of prolonged sedentary activities.6,7 Since walking is generally an appropriate activity for those with diabetes, these Results recommendations can be translated into taking at least 7500 steps per day (steps/d) of Results indicated that the intervention group significantly increased physical activity with 1.5 which 3000 steps should be at a moderate to vigorous intensity.8 A recent report suggests ± 3 days per week of engagement in 30 minutes of moderate-vigorous physical activity vs. no that beginning with ten minutes of MVPA per day (1000 steps/d) would be a feasible start increase in the control group (p = 0.047). Intervention participants increased with 1255 ± towards achieving these guidelines for sedentary individuals in midlife.9 1500 steps per day compared to their baseline (p < 0.010). No significant differences were found in HbA1c with the intervention group decreasing -0.28 ± 1.03% and the control group - Up to now, many people worldwide, including people with type 2 diabetes, do not 0.0 ± 0.69% (p = 0.206). Responders (56%, increasing minimally 1000 steps/day) significantly comply with physical activity guidelines.3 Moreover, people who are overweight tend to decreased HbA1c compared to non-responders (-0.69 ± 1.18% vs. 0.22 ± 0.47%, p = 0.007). overestimate their level of physical activity compared to people who have a healthy weight.10 Adherence to physical activity recommendations from health care professionals is Conclusion low in people with diabetes,11 or may not have sustainable effects on physical activity Self-tracking of physical activity is effective for increment of physical activity in people with behavior and glycemic control. A large trial of an intensive lifestyle program found significant type 2 diabetes. To improve effectiveness of eHealth programs on health-related outcomes, improvements on health outcomes after one year of follow-up. However, these effects additional strategies are needed. diminished after ten years of follow-up.12 Factors that influence the absence of (long-term) adherence to exercise advices can vary, e.g., a low health literacy, lack of motivation, negative beliefs towards physical activity, inconvenience of being active, lack of time, lack of

an adequate exercise plan, and overly vigorous building up leading to injuries.11,13

Several behavioral intervention components have been determined as being crucial for facilitating incremental physical activity, including those for individuals with type 2 diabetes. According to Social Cognitive Theory, certain beliefs about the desired behavior and oneself are needed in order to accomplish behavioral change. These beliefs include a positive attitude about the desired behavior, positive outcome expectations, and adequate self-efficacy beliefs for performing the behavior. The latter plays a significant role because self-efficacy beliefs directly impact behavior and the goals that people set for themselves.14,15 Positive beliefs about physical activity and self-efficacy for exercise can be influenced by education, tailoring of health information, use of a trusted source, and a gradual building up of activity aiming at small goals.14,15 In accordance with this, goal-setting is an important component in many interventions because goals motivate individuals to decrease discrepancy between a person’s current state and the desired state.16 Goals should

104 Self-tracking of physical activity in people with type 2 diabetes

Abstract Introduction

Aim More than 400 million people worldwide have diabetes carrying an increased risk of 1 This study aimed to determine the efficacy of an online self-tracking program on physical cardiovascular disease, cancer, and dementia. Physical inactivity and being overweight due activity, HbA1c, and other health measures in patients with type 2 diabetes. to an unhealthy lifestyle are key factors in both the onset as well as the progression of type 2 diabetes.2,3 Increasing physical activity and the adoption of a healthy lifestyle are essential Methods for preventing long term complications and co-morbidity as this improves glycemic control Seventy-two patients with type 2 diabetes were randomized into an intervention or control and reduces weight.3–6 Therefore, stimulating physical activity is of great importance within group. All participants received usual care. The intervention group received an activity daily clinical practice for people with type 2 diabetes, especially in nursing care. Physical tracker (Fitbit Zip) connected to an online lifestyle program. Physical activity was analyzed in activity guidelines for this population recommend engagement in progressive moderate to average steps per day from week 0 until week 12. Health outcome measurements occurred vigorous resistance training in addition to a minimum of 150 minutes of moderate to in both groups at baseline (T0) and after 13 weeks (T1). vigorous physical activity (MVPA) per week and avoidance of prolonged sedentary activities.6,7 Since walking is generally an appropriate activity for those with diabetes, these Results recommendations can be translated into taking at least 7500 steps per day (steps/d) of Results indicated that the intervention group significantly increased physical activity with 1.5 which 3000 steps should be at a moderate to vigorous intensity.8 A recent report suggests ± 3 days per week of engagement in 30 minutes of moderate-vigorous physical activity vs. no that beginning with ten minutes of MVPA per day (1000 steps/d) would be a feasible start increase in the control group (p = 0.047). Intervention participants increased with 1255 ± towards achieving these guidelines for sedentary individuals in midlife.9 1500 steps per day compared to their baseline (p < 0.010). No significant differences were 6 found in HbA1c with the intervention group decreasing -0.28 ± 1.03% and the control group - Up to now, many people worldwide, including people with type 2 diabetes, do not 0.0 ± 0.69% (p = 0.206). Responders (56%, increasing minimally 1000 steps/day) significantly comply with physical activity guidelines.3 Moreover, people who are overweight tend to decreased HbA1c compared to non-responders (-0.69 ± 1.18% vs. 0.22 ± 0.47%, p = 0.007). overestimate their level of physical activity compared to people who have a healthy weight.10 Adherence to physical activity recommendations from health care professionals is Conclusion low in people with diabetes,11 or may not have sustainable effects on physical activity Self-tracking of physical activity is effective for increment of physical activity in people with behavior and glycemic control. A large trial of an intensive lifestyle program found significant type 2 diabetes. To improve effectiveness of eHealth programs on health-related outcomes, improvements on health outcomes after one year of follow-up. However, these effects additional strategies are needed. diminished after ten years of follow-up.12 Factors that influence the absence of (long-term) adherence to exercise advices can vary, e.g., a low health literacy, lack of motivation, negative beliefs towards physical activity, inconvenience of being active, lack of time, lack of

an adequate exercise plan, and overly vigorous building up leading to injuries.11,13

105 Chapter 6

be both behavioral (e.g., increase steps/d) and outcome related (e.g., weight loss).16–18 Methods Subsequently, after an individual has made adjustments in behavior, positive feedback on this new behavior is very important for maintaining motivation. Hereby, self-monitoring of Study design behavior is recognized as an essential strategy to gain personalized feedback and to This study was designed as an RCT. In addition to usual care, the intervention group received stimulate positive learning experiences.15,17,19 The above mentioned strategies for behavior an activity tracker connected to an online program that intended to encourage them to change are part of a well-known taxonomy of Behavioral Change Techniques (BCTs).17,18 initiate a healthy lifestyle. The control group received only usual care, i.e., visits every three These BCTs enable intervention designers to use an evidence based, reproducible, and months with their diabetes nurse and/or internist for monitoring HbA1c and advice uniform intervention description. regarding medication, lifestyle, and weight reduction in order to normalize blood glucose eHealth systems such as mobile health technology are increasingly used in the levels.29 No restrictions were specified in the prescription of (extra) medication before or treatment of people with type 2 diabetes, in order to optimize diabetes self-management within the study period. The study had a total duration of 13 weeks (baseline week 0 and behaviors.20 These systems offer the possibility to design interventions that include the intervention weeks 1-12). Primary and secondary outcome measures were assessed at above mentioned evidence based strategies to improve physical activity and glycemic baseline (T0) and at the end of the trial (T1). The research and intervention were executed control in people with diabetes.20–22 In the past several years, different activity monitoring by diabetes nurses (specialized nurses with an additional training for diabetes), who were devices have been rapidly developing.23 Modern consumer level activity trackers have the employed in two hospitals in the Netherlands. The diabetes nurses received a study protocol ability to increase awareness about the actual individual physical activity behavior, facilitate and training by the research team before the beginning of the study. The study protocol goal-setting, and facilitate personalized feedback.17,18,23 Previous studies have shown that described the enrollment, measurements, and intervention procedures in detail. The early versions of activity monitors, such as simple pedometers, can incite an increase in diabetes nurses were supported by the research team during the complete study period. physical activity in people with diabetes.24,25 However, there is a gap between the introduction of these newly developed consumer level technologies and the evidence for the Participants effectiveness of their use in clinical care.26 In addition, stimulation of physical activity within Eligible participants were patients with type 2 diabetes, aged ≥ 18 years, HbA1c ≥ 7.5% [58 nursing care is not yet considered as usual care and requires further exploration.27,28 mmol/mol]) with access to the internet and the ability to use a computer. Exclusion criteria Therefore, we designed an online behavioral intervention program that was connected to a were pregnancy, already engaging in >3 hours of intensive exercise per week, or co- modern consumer level activity tracker. This program aimed to assist individuals with type 2 morbidity/ cognitive dysfunction interfering with physical activity. Participants received diabetes to establish a healthy lifestyle. outpatient care and were not hospitalized. All of the participants signed an informed consent. The complete study protocol was approved by the medical ethical committee of the

University Medical Center Groningen (file number 2014-334) and was published in the Dutch Study aim Trial Register (NTR5215). The aim of this Randomized Controlled Trial (RCT) was to evaluate this online self-tracking program for physical activity, glycemic control (HbA1c), and other health outcome measures Recruitment, randomization and allocation (AGEs, weight, BMI, Waist-Hip-Ratio, and self-reported health) in patients with type 2 Participants were recruited at the Bethesda General Hospital (Bethesda Diabetes Research diabetes. We hypothesized that the online self-tracking program would positively affect Center, Hoogeveen, the Netherlands) and the Martini Hospital (Groningen, the Netherlands). physical activity, HbA1c, and other health outcome measures in the intervention group Recruitment methods were flyers, letters, an advertisement in a local paper, and eligible compared to the control group. In addition, we hypothesized that within the intervention patients being asked by their diabetes nurse. After stratification for HbA1c and BMI (based group, participants increasing with a minimum of 1000 steps per day (steps/d) would on the mean values of the patients included thus far), patients were randomly assigned to demonstrate a greater reduction in HbA1c compared to participants increasing less than the intervention or control group using block randomization.30 A predetermined formula per 1000 steps/d. block (e.g., ICCI) determined to which group a patient was assigned. Figure 1 illustrates the

flowchart of recruitment of the participants in the study.

106 Self-tracking of physical activity in people with type 2 diabetes

be both behavioral (e.g., increase steps/d) and outcome related (e.g., weight loss).16–18 Methods Subsequently, after an individual has made adjustments in behavior, positive feedback on this new behavior is very important for maintaining motivation. Hereby, self-monitoring of Study design behavior is recognized as an essential strategy to gain personalized feedback and to This study was designed as an RCT. In addition to usual care, the intervention group received stimulate positive learning experiences.15,17,19 The above mentioned strategies for behavior an activity tracker connected to an online program that intended to encourage them to change are part of a well-known taxonomy of Behavioral Change Techniques (BCTs).17,18 initiate a healthy lifestyle. The control group received only usual care, i.e., visits every three These BCTs enable intervention designers to use an evidence based, reproducible, and months with their diabetes nurse and/or internist for monitoring HbA1c and advice uniform intervention description. regarding medication, lifestyle, and weight reduction in order to normalize blood glucose eHealth systems such as mobile health technology are increasingly used in the levels.29 No restrictions were specified in the prescription of (extra) medication before or treatment of people with type 2 diabetes, in order to optimize diabetes self-management within the study period. The study had a total duration of 13 weeks (baseline week 0 and behaviors.20 These systems offer the possibility to design interventions that include the intervention weeks 1-12). Primary and secondary outcome measures were assessed at above mentioned evidence based strategies to improve physical activity and glycemic baseline (T0) and at the end of the trial (T1). The research and intervention were executed control in people with diabetes.20–22 In the past several years, different activity monitoring by diabetes nurses (specialized nurses with an additional training for diabetes), who were devices have been rapidly developing.23 Modern consumer level activity trackers have the employed in two hospitals in the Netherlands. The diabetes nurses received a study protocol ability to increase awareness about the actual individual physical activity behavior, facilitate and training by the research team before the beginning of the study. The study protocol goal-setting, and facilitate personalized feedback.17,18,23 Previous studies have shown that described the enrollment, measurements, and intervention procedures in detail. The early versions of activity monitors, such as simple pedometers, can incite an increase in diabetes nurses were supported by the research team during the complete study period. physical activity in people with diabetes.24,25 However, there is a gap between the 6 introduction of these newly developed consumer level technologies and the evidence for the Participants effectiveness of their use in clinical care.26 In addition, stimulation of physical activity within Eligible participants were patients with type 2 diabetes, aged ≥ 18 years, HbA1c ≥ 7.5% [58 nursing care is not yet considered as usual care and requires further exploration.27,28 mmol/mol]) with access to the internet and the ability to use a computer. Exclusion criteria Therefore, we designed an online behavioral intervention program that was connected to a were pregnancy, already engaging in >3 hours of intensive exercise per week, or co- modern consumer level activity tracker. This program aimed to assist individuals with type 2 morbidity/ cognitive dysfunction interfering with physical activity. Participants received diabetes to establish a healthy lifestyle. outpatient care and were not hospitalized. All of the participants signed an informed consent. The complete study protocol was approved by the medical ethical committee of the

flowchart of recruitment of the participants in the study.

107 Chapter 6

Intervention Physical activity The intervention group received usual care plus an activity tracker (Fitbit Zip, Fitbit Inc, CA) Physical activity was the intermediate outcome measure and was measured with a 1-item and access to the online self-tracking (eHealth) program. The activity tracker was linked to physical activity questionnaire that indicated how many days per week a participant engaged the personal accounts of the participants in the eHealth program. The program was designed in 30 minutes of MVPA. This questionnaire has strong reliability and good validity compared by a project group which consisted of members from healthcare organizations, health to expert classification of subjects in the 30 minutes of MVPA recommendation.31 In training institutes, and technology companies. The complete content of the program aimed addition, within the intervention group, physical activity was measured with the Fitbit Zip in to optimize knowledge about a healthy lifestyle, increase awareness of an individual’s own steps/d. This activity tracker could be clipped to the clothing or placed in the pants pocket. physical activity behavior, increase self-efficacy for exercise, and ultimately to optimize and The Fitbit Zip has previously been shown to be reliable and valid.32 Participants were asked maintain a healthy lifestyle. Important BCTs of the intervention were providing information to wear the activity tracker consistently in the preferred wearing position in order to about health consequences, behavioral goal-setting, goal-setting of outcome, barrier optimize validity and reliability of the data.32 identification/problem solving and action planning (through the eHealth program), behavioral self-monitoring and review of behavioral goals (through the Fitbit device), Primary outcome providing feedback on behavior, habit formation, habit reversal and graded tasks (both The primary outcome was HbA1c (% and mmol/mol), as measured during usual care in through the Fitbit device and eHealth program).14–18 In more detail, the participants were specialized laboratories. instructed to maintain their usual activity pattern in Week 0 for determining their baseline activity level. Beginning in Week 1, they were encouraged by their diabetes nurse and the Secondary health outcome measures eHealth program to set incremental activity goals, based on their individual baseline activity 1) Advanced Glycation Endproducts (AGEs). AGEs were measured in skin autofluorescence level (i.e., behavioral goals) and outcome goals (e.g., losing weight). Participants were (fluorescent properties of AGEs in the skin) with the AGE Reader mu, according to the device encouraged to begin with small goals such as increasing 500 or 1000 steps/d, and, protocol (Diagnoptics, Groningen, the Netherlands). This device has been shown to be depending on their individual capabilities, continuing to increase to the norm of a minimum reliable and valid.33,34 of 7500 steps/d or 150 minutes of MVPA per week.6,8 The participants could easily contact (2) Weight (kg) / BMI (kg/m2); body weight was measured on a regular scale without shoes the diabetes nurse throughout the program for questions and support. The program or extra clothing and was translated to BMI using the height of the participant. Height was provided weekly information about physical activity, a healthy diet with example recipes, measured in a standing position without shoes using a measuring bar. diabetes, and videos of strength exercises with an explanation of how to build muscle 3) Waist-Hip-Ratio in cm/cm was calculated by dividing the waist circumference by the hip strength. The information explained the benefits of healthy behavior and addressed circumference. Waist circumference was measured according to a protocol in a standing frequent barriers that people experience when engaging in physical activity.11,13 For position at the midpoint between the lowest rib and iliac crest.35 example, in order to counteract a frequent barrier ‘lack of time’, information instructions on 4) Subjective health score was measured on a Visual Analogue Scale (VAS) ranging from 0-10 physical activity included several strategies to increase steps per day that the participant with 10 indicating the best health an individual could imagine and 0 the worst. This score has could integrate during the (working) day. In addition, tailored feedback messages were been shown to be reliable and valid.36 provided through the program once per week based on the number of steps taken in the 5) Changes in use and dosage of medication. Per participant, all diabetes-related past week. These messages were based on whether the participants had increased their medications, including oral medication (Metformin, SU, acarbose, and DPP4 – inhibition), steps/d with an average minimum of 500 steps/d, stayed the same, or decreased their GLP-1 therapy, and units of insulin use per day were assessed at baseline and at T1. An steps/d compared to the previous week. All messages, also in the event of a decrement in adjustment for (changes of) HbA1c influencing medication is needed to reliably study the activity, had a positive tone, included fun facts, and were aimed to encourage the participant effects of the intervention on the primary endpoint (HbA1c). Clinically relevant extra to increase activity levels. medication at T0 was defined as a dose increase of 25% of oral medication, an increase of a minimum of four units of insulin, or additional diabetes medication. A clinically relevant Outcome measures change in medication at T1 was defined as a change of at least four units of insulin or a Data collection was undertaken by the research nurse at T0 and T1. The self-reported change of >25% of other glucose-lowering medication. behavioral measures were filled in by the participants at T0 and T1 by using digital questionnaires through the eHealth program. Self-reported behavioral measures

Four domains for behavioral factors were discriminated including intention, attitude, self-

108 Self-tracking of physical activity in people with type 2 diabetes

Intervention Physical activity The intervention group received usual care plus an activity tracker (Fitbit Zip, Fitbit Inc, CA) Physical activity was the intermediate outcome measure and was measured with a 1-item and access to the online self-tracking (eHealth) program. The activity tracker was linked to physical activity questionnaire that indicated how many days per week a participant engaged the personal accounts of the participants in the eHealth program. The program was designed in 30 minutes of MVPA. This questionnaire has strong reliability and good validity compared by a project group which consisted of members from healthcare organizations, health to expert classification of subjects in the 30 minutes of MVPA recommendation.31 In training institutes, and technology companies. The complete content of the program aimed addition, within the intervention group, physical activity was measured with the Fitbit Zip in to optimize knowledge about a healthy lifestyle, increase awareness of an individual’s own steps/d. This activity tracker could be clipped to the clothing or placed in the pants pocket. physical activity behavior, increase self-efficacy for exercise, and ultimately to optimize and The Fitbit Zip has previously been shown to be reliable and valid.32 Participants were asked maintain a healthy lifestyle. Important BCTs of the intervention were providing information to wear the activity tracker consistently in the preferred wearing position in order to about health consequences, behavioral goal-setting, goal-setting of outcome, barrier optimize validity and reliability of the data.32 identification/problem solving and action planning (through the eHealth program), behavioral self-monitoring and review of behavioral goals (through the Fitbit device), Primary outcome providing feedback on behavior, habit formation, habit reversal and graded tasks (both The primary outcome was HbA1c (% and mmol/mol), as measured during usual care in through the Fitbit device and eHealth program).14–18 In more detail, the participants were specialized laboratories. instructed to maintain their usual activity pattern in Week 0 for determining their baseline activity level. Beginning in Week 1, they were encouraged by their diabetes nurse and the Secondary health outcome measures eHealth program to set incremental activity goals, based on their individual baseline activity 1) Advanced Glycation Endproducts (AGEs). AGEs were measured in skin autofluorescence level (i.e., behavioral goals) and outcome goals (e.g., losing weight). Participants were (fluorescent properties of AGEs in the skin) with the AGE Reader mu, according to the device 6 encouraged to begin with small goals such as increasing 500 or 1000 steps/d, and, protocol (Diagnoptics, Groningen, the Netherlands). This device has been shown to be depending on their individual capabilities, continuing to increase to the norm of a minimum reliable and valid.33,34 of 7500 steps/d or 150 minutes of MVPA per week.6,8 The participants could easily contact (2) Weight (kg) / BMI (kg/m2); body weight was measured on a regular scale without shoes the diabetes nurse throughout the program for questions and support. The program or extra clothing and was translated to BMI using the height of the participant. Height was provided weekly information about physical activity, a healthy diet with example recipes, measured in a standing position without shoes using a measuring bar. diabetes, and videos of strength exercises with an explanation of how to build muscle 3) Waist-Hip-Ratio in cm/cm was calculated by dividing the waist circumference by the hip strength. The information explained the benefits of healthy behavior and addressed circumference. Waist circumference was measured according to a protocol in a standing frequent barriers that people experience when engaging in physical activity.11,13 For position at the midpoint between the lowest rib and iliac crest.35 example, in order to counteract a frequent barrier ‘lack of time’, information instructions on 4) Subjective health score was measured on a Visual Analogue Scale (VAS) ranging from 0-10 physical activity included several strategies to increase steps per day that the participant with 10 indicating the best health an individual could imagine and 0 the worst. This score has could integrate during the (working) day. In addition, tailored feedback messages were been shown to be reliable and valid.36 provided through the program once per week based on the number of steps taken in the 5) Changes in use and dosage of medication. Per participant, all diabetes-related past week. These messages were based on whether the participants had increased their medications, including oral medication (Metformin, SU, acarbose, and DPP4 – inhibition), steps/d with an average minimum of 500 steps/d, stayed the same, or decreased their GLP-1 therapy, and units of insulin use per day were assessed at baseline and at T1. An steps/d compared to the previous week. All messages, also in the event of a decrement in adjustment for (changes of) HbA1c influencing medication is needed to reliably study the activity, had a positive tone, included fun facts, and were aimed to encourage the participant effects of the intervention on the primary endpoint (HbA1c). Clinically relevant extra to increase activity levels. medication at T0 was defined as a dose increase of 25% of oral medication, an increase of a minimum of four units of insulin, or additional diabetes medication. A clinically relevant Outcome measures change in medication at T1 was defined as a change of at least four units of insulin or a Data collection was undertaken by the research nurse at T0 and T1. The self-reported change of >25% of other glucose-lowering medication. behavioral measures were filled in by the participants at T0 and T1 by using digital questionnaires through the eHealth program. Self-reported behavioral measures

Four domains for behavioral factors were discriminated including intention, attitude, self-

109 Chapter 6

efficacy, and social norm towards engagement in exercise (minimally five days a week for 30 5000 steps which is established as the cut-off point for sedentary behavior.39 It was minutes of MVPA) which is consistent with behavioral literature.15,37 These domains were reasoned that, when an individual had even 90% less steps than the cut-off point of measured with a 17-item questionnaire that was based on the items of Boudreau and Godin, sedentary behavior, the activity tracker was probably either not worn or not worn for the 2014.38 Participants could indicate their extent of agreement on a 5-points Likert scale. For complete day. example, with social norm, participants could respond to the statement, “The most important persons in my immediate area advise me to increase my physical activity”. Answers could range from strongly disagree (a score of 1) to strongly agree (a score of 5). Results Per domain (e.g., for attitude, self-efficacy, and social norm) all scores were added and subsequently divided by the number of items. To determine the internal consistency of the Inclusion different items, Cronbach’s alpha was calculated for attitude (seven items), self-efficacy (six The participants were recruited between April 2015 and July 2016. A total of 465 patients items), and social norm (three items). were screened of which 105 appeared eligible for inclusion. From these, 72 adults (47.2% After completion of the program intervention participants were contacted by a member of females) were randomly assigned to the intervention or control group (Figure 1 ‘flowchart’). the research team to inquire about the perceived usefulness of the activity tracker and the During the study period, two participants of the control group, and one participant of the eHealth program as well as the impact of the program on their physical activity and intervention group dropped out. Three intervention participants were unavailable for the perceived health. follow-up primary outcome measurement at T1. Therefore, 66 participants could be

included in the analysis for the primary outcome (Figure 1). The baseline characteristics of Statistical analyses the participants are depicted in Table 1. At baseline, the mean age was 56 ± 11 years, mean A sample size computation with an expected mean HbA1c of 7.5 ± 0.34% [59 ± 4 mmol/mol] diabetes duration 15.3 ± 6.7 years, mean HbA1c 8.6 ± 1.0% [70.0 ± 11.3 mmol/mol], and at baseline indicated the need for the inclusion of a minimum of 28 participants per group in mean BMI 32.9 ± 5 kg/m2. The participants were primarily Caucasians (98.6%). No significant order to demonstrate a minimal relevant reduction of 0.27% [3 mmol/mol] with a statistical differences existed between the intervention group and control group at T0. Ten participants power of 80% and a significance level of 5%. (six of the intervention and four of the control group) received extra anti-hyperglycemic Analysis occurred for all of the participants (intention to treat analysis) using ANOVA medication at baseline (Table 2). repeated measures analysis, with self-reported physical activity, HbA1c, AGEs, weight, BMI, and Waist-Hip-Ratio as dependent variables. Two time levels were included, T0 (Week 0) and Adherence T1 (Week 12), to investigate for statistical differences between the intervention and control Adherence to the intervention program was defined as having worn the Fitbit on >75% of group over time. Age, gender, extra medication at T0, perceived health at T0, intention to intervention days and having read >50% of program content. The latter was verified digitally increase physical activity at T0, attitude, self-efficacy, and social norm were included as and with the telephonic evaluation. In this way 82.5% of intervention participants were covariates. defined as being adherent.

Within the intervention group, mixed models analysis was used to analyze the change of physical activity over time, measured as average steps/d from Week 0 (baseline) until Week 12. The advantage of mixed models is that this method can handle missing data, e.g., a participant with a missing week of average steps/d data will still be included in the analysis. The intervention group was categorized into responders (e.g., increasing a minimum of 1000 steps/d compared to baseline) and non-responders (increasing with less than 1000 steps/d), to further examine the effect on HbA1c and other outcome measures.7–9 The average steps/d were calculated by dividing the total steps of the specific week by the number of days the participant had measured steps with the activity tracker. It was determined that it was necessary for any given participant to have at least four valid measurement days per week in order to make a reliable calculation. It was decided that at least 500 steps had to be measured during a day for it to be included in the analysis, since 500 steps represents 10% of

110 Self-tracking of physical activity in people with type 2 diabetes

included in the analysis for the primary outcome (Figure 1). The baseline characteristics of Statistical analyses the participants are depicted in Table 1. At baseline, the mean age was 56 ± 11 years, mean A sample size computation with an expected mean HbA1c of 7.5 ± 0.34% [59 ± 4 mmol/mol] diabetes duration 15.3 ± 6.7 years, mean HbA1c 8.6 ± 1.0% [70.0 ± 11.3 mmol/mol], and 6 at baseline indicated the need for the inclusion of a minimum of 28 participants per group in mean BMI 32.9 ± 5 kg/m2. The participants were primarily Caucasians (98.6%). No significant order to demonstrate a minimal relevant reduction of 0.27% [3 mmol/mol] with a statistical differences existed between the intervention group and control group at T0. Ten participants power of 80% and a significance level of 5%. (six of the intervention and four of the control group) received extra anti-hyperglycemic Analysis occurred for all of the participants (intention to treat analysis) using ANOVA medication at baseline (Table 2). repeated measures analysis, with self-reported physical activity, HbA1c, AGEs, weight, BMI, and Waist-Hip-Ratio as dependent variables. Two time levels were included, T0 (Week 0) and Adherence T1 (Week 12), to investigate for statistical differences between the intervention and control Adherence to the intervention program was defined as having worn the Fitbit on >75% of group over time. Age, gender, extra medication at T0, perceived health at T0, intention to intervention days and having read >50% of program content. The latter was verified digitally increase physical activity at T0, attitude, self-efficacy, and social norm were included as and with the telephonic evaluation. In this way 82.5% of intervention participants were covariates. defined as being adherent.

111 Chapter 6

Table 1. Baseline characteristics and differences in changes of health outcome measures between intervention and control group. Intervention group (N=40) Control group (N=32) F-value b p-value b

ITT Baseline a ∆ Baseline a ∆

Age (years) 56.8 ± 11.4 N.A. 55.8 ± 11.4 N.A. N.A. N.A.

Diabetes duration 15.5 ± 7.7 N.A. 14.9 ± 5.3 N.A. N.A. N.A. (years) Medication use (%) Oral medication 77.5 N.A. 65.6 N.A. N.A. N.A. GLP-1 therapy 25 21.9 Insulin 55 53.1 Insulin (units) (based 62.9 ± 41.8 4.7 ± 15.2 76.4 ± 54.8 5.0 ± 10.3 N.A. N.A. on users) HbA1c (%) 8.5 ± 0.87 -0.28 ± 1.03 8.6 ± 1.22 -0.0 ± 0.69 1.634 0.206

HbA1c (mmol/mol) 69.9 ± 9.5 -3.1 ± 11.3 70.2 ± 13.3 -0.03 ± 7.5 1.634 0.206

AGEs (SAF) 2.46 ± 0.57 0.14 ± 0.34 2.60 ± 0.4 0.05 ± 0.35 0.661 0.421

Weight (kg) 102.2 ± 19.3 -0.1 ± 3.2 99.8 ± 16.3 0.5 ± 2.5 0.602 0.441

BMI (kg/m2) 33.2 ± 5.3 -0.02 ± 1.1 32.6 ± 4.5 0.17 ± 0.8 0.550 0.462

Waist circumference 112.1 ± 11.6 0.05 ± 3.4 116.4 ± 13.2 0.17 ± 5.2 0.011 0.918 (cm) Hip circumference 115.3 ± 8.9 0.0 ± 3.2 114.6 ± 11.5 -0.4 ± 4.1 0.140 0.709 (cm) Waist Hip Ratio 0.96 ± 0.17 0.03 ± 0.03 1.00 ± 0.17 0.04 ± 0.04 0.008 0.928 (cm/cm) Self-perceived health 5.7 ± 2 1.1 ± 2.3 5.6 ± 1.8 -0.3 ± 1.4 5.874 0.020 * Figure 1. (points)

Flowchart of recruitment of participants in the study. a All variables were checked for normality and influence of non-normality was checked with cook’s distance. b ANOVA repeated measures analysis, for differences in change over time between intervention and control group. ∆=Change between T0 and T1. N.A.= not applicable SAF= skin auto fluorescence * P<0.05

112 Self-tracking of physical activity in people with type 2 diabetes

Table 1. Baseline characteristics and differences in changes of health outcome measures between intervention and control group. Intervention group (N=40) Control group (N=32) F-value b p-value b

ITT Baseline a ∆ Baseline a ∆

Age (years) 56.8 ± 11.4 N.A. 55.8 ± 11.4 N.A. N.A. N.A.

HbA1c (mmol/mol) 69.9 ± 9.5 -3.1 ± 11.3 70.2 ± 13.3 -0.03 ± 7.5 1.634 0.206

AGEs (SAF) 2.46 ± 0.57 0.14 ± 0.34 2.60 ± 0.4 0.05 ± 0.35 0.661 0.421 Weight (kg) 102.2 ± 19.3 -0.1 ± 3.2 99.8 ± 16.3 0.5 ± 2.5 0.602 0.441 6 BMI (kg/m2) 33.2 ± 5.3 -0.02 ± 1.1 32.6 ± 4.5 0.17 ± 0.8 0.550 0.462

113 Chapter 6

Table 2. Medication use at baseline and T1. Extra medication at T0 (N=10) * Intervention group Control group

Started with Metformin N=1 N=0 Started with SUs N=0 N=0 Started with acarbose N=0 N=0 Started with DPP-4 I N=0 N=0 Started with GLP-1 N=0 N=0 Started with insulin N=1 N=4 Increased insulin > 4 units N=3 N=0 Started with more than 1 type of N=1 N=0 medication Change in medication at T1 Intervention group (N=40) Control group (N=32) No change in medication N=32 N=26 Increased insulin > 4 units N=4 N=6 Decreased insulin > 4 units N=1 N=0 Increased dosage of other N=3 N=0 Figure 2. medication Change in physical activity from T0 to T1 in both groups (A) and change in steps/d from baseline till week 12 Decreased dosage of other N=0 N=0 within the intervention group (B). MVPA= moderate to vigorous physical activity. medication

* Six participants of the intervention and four of the control group received extra medication at baseline.

Table 3. Physical activity Change in physical activity (average steps/d) over time. All intervention weeks are compared The intervention group demonstrated an increase of 1.5 ± 3 days per week of self-reported with the baseline week (N=36). engagement in a minimum of 30 minutes of MVPA, while the control group showed no Week Estimate (SE) # Confidence Interval p-value increase; 0.0 ± 1.8 (F = 4.164, p = 0.047, Figure 2.a). Within the intervention group, physical Lower Upper activity data from the activity tracker were available for 36 out of 40 participants (due to one 1 1284 (372) 552 2016 0.001 * drop-out, one non-adherence to the intervention procedure, and two participants 2 1172 (371) 442 1902 0.002 * experiencing technical problems with pairing their activity tracker to the eHealth program). 3 1536 (378) 793 2280 0.000 * The activity tracker disclosed a mean of 5975 ± 2982 steps/d at the baseline week that 4 1351 (378) 607 2095 0.000 * significantly increased during all of the intervention weeks (p < 0.010, mixed models analysis, 5 1399 (375) 660 2137 0.005 * Table 3). Figure 2.b illustrates the average steps/d of the participants during all of the 6 1116 (372) 385 1847 0.003 * intervention weeks. On average, participants increased with 1255 ± 1500 steps/d during the 7 1063 (375) 326 1800 0.005 * intervention period. The control participants -who received the activity tracker after finishing 8 1384 (378) 641 2127 0.000 * their control period at T1- walked on average 6113 ± 2478 steps/d during their baseline 9 1031 (372) 301 1771 0.006 * week. This was approximately the same as the baseline average steps/d of the intervention 10 959 (372) 227 1690 0.010 * group (mean difference = 138 steps/d, p = 0.859). 11 1435 (375) 697 2173 0.000 * 12 1135 (369) 409 1860 0.002 * # Linear mixed models. The estimates from week 1-12 indicate the estimated changes compared to baseline. SE= Standard Error. * p<0.010.

114 Self-tracking of physical activity in people with type 2 diabetes

Table 2. Medication use at baseline and T1. Extra medication at T0 (N=10) * Intervention group Control group

* Six participants of the intervention and four of the control group received extra medication at baseline. 6

115 Chapter 6

Glycemic control English, and including content (i.e., mostly in case of physical strength building videos) that is There was no significant difference in HbA1c change between the intervention group and more tailored to individual needs. the control group (F = 1.634, p = .206). The intervention group showed a (non-significant) mean decrease of -0.28 ± 1.03% [-3.1 ± 11.3 mmol/mol], and the control group showed no decrease; -0.0 ± 0.69% [-0.03 ± 7.5 mmol/mol] (Table 1). Table 4. Results on health outcome measures for responders vs. non-responders within the intervention group (N=36).

Responders Non-responders F-value a p-value a Secondary health outcome measures (N=20) (N=16) No significant differences on the health-related outcomes (AGEs, weight/BMI and Hip-Waist-

Ratio) were found between the intervention and control groups (p > .05) except for the HbA1c (%) -0.69 ± 1.18 0.22 ± 0.47 8.430 0.007 * subjective health score (p = .02). No differences existed in the change in medication HbA1c (mmol/mol) -7.6 ± 12.9 2.4 ± 5.3 8.430 0.007 * prescription at T1, with eight intervention participants and six control participants receiving AGEs (SF) 0.06 ± 0.33 0.26 ± 0.32 1.816 0.192 extra medication at T1. One intervention participant decreased medication at T1 (Table 2). Weight (kg) -0.3 ± 3.6 0.3 ± 2.7 0.388 0.538 All of the results on the health outcome measures are presented in Table 1. BMI (kg/m2) -0.08 ± 1.3 0.06 ± 0.9 0.416 0.524 Waist circumference (cm) -0.8 ± 3.7 1.4 ± 3.9 0.563 0.459 Hip circumference (cm) -0.3 ± 2.8 -0.4 ± 3.8 0.062 0.805 Subgroup analyses and covariates Waist Hip Ratio (cm/cm) -0.08 ± 3.7 1.4 ± 3.9 0.796 0.380 Fifty-six % of participants were defined as ‘responders’ (increasing ≥ 1000 steps/d compared *=p<0.01 a ANOVA repeated measures analysis, for differences in change over time between responders and to baseline). When ‘being a responder’ was included in a sub analysis of intervention non-responders within the intervention group. participants, a significant interaction effect was found for being a responder and HbA1c over time: -0.69 ± 1.18% [-7.6 ± 12.9 mmol/mol] for responders and 0.22 ± 0.47% [2.4 ± 5.3 mmol/mol] for non-responders (F = 8.430, p = 0.007). No significant results were determined Discussion for being a responder on the other health outcome measures (Table 4). Cronbach’s Alpha for the different domains of the behavioral questionnaire was .845, .860, The goal of the present study was to examine the effects of an online self-tracking program and .683 for attitude, self-efficacy, and social norm, respectively. on physical activity, glycemic control, and other health outcome measures in people with Social norm was a significant covariate when added to the main analysis (F = 7.475, p = type 2 diabetes. Physical activity significantly increased in the intervention group and, from a 0.009). Age, gender, BMI, intention to increase physical activity, attitude, and self-efficacy certain level of increase (≥ 1000 steps per day), a clinically relevant decrease of HbA1c was showed no significant effects on HbA1c change. Also, extra medication at T0 had no determined. Overall, for responders and non-responders, no effects were found on HbA1c or significant main effect on HbA1c (F = 3.102, p = 0.088). Within the intervention group, the other health outcome measures except on the subjective health score. Changes in responders had a significantly higher social norm score at baseline (p = 0.020) compared to medication at baseline did not affect the results. non-responders. To explain the lack of an overall effect on HbA1c, a more thorough analysis was made

of the activity tracker data measured within the intervention group. The average steps/d Patient evaluations increased compared to the individual baselines of the intervention participants with 1255 ± Ninety percent of the intervention participants perceived the activity tracker as being useful 1500 steps/d. Hereby, 39% of participants complied to the guideline of taking ≥7500 steps/d or very useful. A lower percentage (46%) qualified the eHealth program as being useful or during the intervention, and 61% did not. A substantial variability was present in the very useful, 28% as neutral, and 24 % as not useful. Three out of four participants (74%) increase of steps/d between the participants; -1355 till 5049 steps/d. This variability was indicated that they had increased their physical activity behavior due to the intervention supported by the evaluation results; 74% of the intervention participants indicated having program, whereby 41% of participants indicated that they increased their activity levels a lot increased their physical activity, i.e., 41% daily and 33% weekly activity, while 26% had not. (daily changes), and 33% indicated a moderate change (weekly changes). One out of two Therefore, this inter-individual variability may well explain the absence of an overall participants (51%) indicated that they felt more fit or healthier since their participation in significant decline in HbA1c in the intervention group. Indeed, subgroup analysis showed the program, 26% were uncertain, and 20% indicated that they did not feel more fit or that responders had a significant and clinically relevant decline in HbA1c (-0.69 % vs. 0.22%, healthy. Suggestions for improvement were indicated such as improving the ease of use of p = 0.007). Thus, for the complete intervention group the increase in physical activity was the eHealth program, providing an activity tracker with a Dutch mobile application instead of

116 Self-tracking of physical activity in people with type 2 diabetes

Responders Non-responders F-value a p-value a Secondary health outcome measures (N=20) (N=16) No significant differences on the health-related outcomes (AGEs, weight/BMI and Hip-Waist-

Ratio) were found between the intervention and control groups (p > .05) except for the HbA1c (%) -0.69 ± 1.18 0.22 ± 0.47 8.430 0.007 * subjective health score (p = .02). No differences existed in the change in medication HbA1c (mmol/mol) -7.6 ± 12.9 2.4 ± 5.3 8.430 0.007 * prescription at T1, with eight intervention participants and six control participants receiving AGEs (SF) 0.06 ± 0.33 0.26 ± 0.32 1.816 0.192 extra medication at T1. One intervention participant decreased medication at T1 (Table 2). Weight (kg) -0.3 ± 3.6 0.3 ± 2.7 0.388 0.538 All of the results on the health outcome measures are presented in Table 1. BMI (kg/m2) -0.08 ± 1.3 0.06 ± 0.9 0.416 0.524 Waist circumference (cm) -0.8 ± 3.7 1.4 ± 3.9 0.563 0.459 Hip circumference (cm) -0.3 ± 2.8 -0.4 ± 3.8 0.062 0.805 Subgroup analyses and covariates Waist Hip Ratio (cm/cm) -0.08 ± 3.7 1.4 ± 3.9 0.796 0.380 Fifty-six % of participants were defined as ‘responders’ (increasing ≥ 1000 steps/d compared *=p<0.01 a ANOVA repeated measures analysis, for differences in change over time between responders and to baseline). When ‘being a responder’ was included in a sub analysis of intervention non-responders within the intervention group. participants, a significant interaction effect was found for being a responder and HbA1c over 6 time: -0.69 ± 1.18% [-7.6 ± 12.9 mmol/mol] for responders and 0.22 ± 0.47% [2.4 ± 5.3 mmol/mol] for non-responders (F = 8.430, p = 0.007). No significant results were determined Discussion for being a responder on the other health outcome measures (Table 4). Cronbach’s Alpha for the different domains of the behavioral questionnaire was .845, .860, The goal of the present study was to examine the effects of an online self-tracking program and .683 for attitude, self-efficacy, and social norm, respectively. on physical activity, glycemic control, and other health outcome measures in people with Social norm was a significant covariate when added to the main analysis (F = 7.475, p = type 2 diabetes. Physical activity significantly increased in the intervention group and, from a 0.009). Age, gender, BMI, intention to increase physical activity, attitude, and self-efficacy certain level of increase (≥ 1000 steps per day), a clinically relevant decrease of HbA1c was showed no significant effects on HbA1c change. Also, extra medication at T0 had no determined. Overall, for responders and non-responders, no effects were found on HbA1c or significant main effect on HbA1c (F = 3.102, p = 0.088). Within the intervention group, the other health outcome measures except on the subjective health score. Changes in responders had a significantly higher social norm score at baseline (p = 0.020) compared to medication at baseline did not affect the results. non-responders. To explain the lack of an overall effect on HbA1c, a more thorough analysis was made

117 Chapter 6

probably not enough to improve glycemic control. However, the combined improvement of programs should incorporate a systematic approach to include all of these factors, including physical activity and glycemic control in responders with advanced type 2 diabetes is an analysis for appropriate intervention strategies.18 clinically relevant and promising for the future. The prediction of responsiveness might be a Our study has a number of strengths and limitations. The first limitation is the target of further research. Compared to the literature, the increase in steps/d was lower relatively short-term follow-up of three months which was selected in order to afford the than the SMD of 1822 steps/d found in the meta-analysis by Qiu et al.40 This may be control group the opportunity to engage in the program after finishing their control period. explained by the additional support, such as counseling or telephonic support, that was In this manner, an increased withdrawal rate in the control group could be prevented. provided in the studies reviewed by Qiu et al. In a recent comparable, but larger, study of However, a longer duration may have been beneficial for incorporating lifestyle habits and Dasgupta et al, physical activity was increased with 1190 steps/d, CI [550-1840], and HbA1c for a longer-term comparison with the control group. Second, this study had a relatively was decreased by 0.38% compared to the control group.41 This is consistent with the results small sample size. Although we met the minimal number of participants necessary from our found in our study. sample size calculation, a larger group would have strengthened the results. Third, this study The secondary health outcomes showed no significant changes, even in the missed an objective physical activity instrument for both groups. Because the Fitbit was an responders vs. non-responders’ analysis. AGEs are complex linkage products measured in important component of the intervention, the control group did not receive a Fitbit. the tissue and because the accumulation of AGEs also depends on other factors such as However, from the baseline step measurements that the control group made after finishing smoking behavior and intake of certain foods,34 it is likely that more long-term lifestyle their control period, it appeared that they walked the same average steps/d compared to changes are necessary to achieve significant results on AGEs. Also, for body weight and the baseline steps/d of the intervention group. This reinforces our finding that the Waist-Hip-Ratio no effects were found. As an increase of at least 2000 steps/d is needed to intervention group increased their average steps/d. A strength was that the study was achieve relevant effects on weight and body composition,8,42 the lack of findings on these conducted in a general hospital setting, embedded in the usual nursing care for patients with measures is probably explained by the smaller increase in the number of steps/d found in type 2 diabetes. This real-life setting simplifies an extrapolation of the data to the general this study. population with type 2 diabetes. Also, as determined from the evaluations, most participants were satisfied with the self-tracking program as 90% indicated that the activity tracker was Interestingly, social norm was a significant confounder for the HbA1c results, and useful or very useful for them. This indicates that patients with type 2 diabetes are willing to responders had a higher social norm score at baseline compared to non-responders. This engage in lifestyle programs based on self-tracking. result emphasizes the importance of taking into account the social support of a patient with type 2 diabetes. In accordance with this, the use of theory based BCTs within self-monitoring devices or lifestyle interventions is receiving increasing attention within physical activity Conclusions research. Several studies pointed out the importance of providing information, goal-setting, action planning, self-monitoring, barrier identification, personalized feedback, and In our study, self-tracking of physical activity did improve physical activity in patients with rewards.17,25,43 These BCTs were present in our study, however, other important BCTs such advanced type 2 diabetes. The intervention did not improve glycemic control overall, due to as facilitating social support, changing environmental factors, and the use of follow-up a large inter-individual variability in responsiveness to the intervention. However, it did prompts were lacking in our intervention at a structural basis.43–46 In addition, the BCT relevantly improve glycemic control in 56% of the participants who increased their physical ‘action planning’ was incorporated within the informational documents, but not structurally activity with a minimum of 1000 steps/d. To improve the effectiveness of online self-tracking tailored for individual participants. Since these BCTs have been found to be associated with programs on health outcomes, more development in the ease of use of eHealth technology, improved intervention outcomes in people with diabetes, the lack of an overall effect may integration of behavioral change techniques, and tailoring of intervention programs is well be explained by insufficient structural implementation of these BCTs in our study.45,46 required, for example, based on presence of social support. This should be improved in future studies; for instance, social support may be enhanced by creating a system in which patients with diabetes are connected to each other, and/or are facilitated to meet for walking groups. Next to the inclusion of additional evidence based Acknowledgements BCTs, effects of future programs may also be enhanced by improving the ease of use of The authors would like to thank all members of the Living Lab project group ‘Active Ageing Diabetes’ digital techniques, incorporating additional advances to use personal generated health data for their help and assistance in developing the eHealth program and its content and the diabetes nurses from the Bethesda Diabetes Research Center and the Martini Hospital (Ellen Wessels, Brenda in a meaningful way,20 and providing technical support from a person other than the nurse. Wierbos, and Anneke Kleine), for their dedicated guidance and treatment of the patients and for 20,29 This will enhance the role of the nurse for providing personal lifestyle support. Future their accurate data collection.

118 Self-tracking of physical activity in people with type 2 diabetes

probably not enough to improve glycemic control. However, the combined improvement of programs should incorporate a systematic approach to include all of these factors, including physical activity and glycemic control in responders with advanced type 2 diabetes is an analysis for appropriate intervention strategies.18 clinically relevant and promising for the future. The prediction of responsiveness might be a Our study has a number of strengths and limitations. The first limitation is the target of further research. Compared to the literature, the increase in steps/d was lower relatively short-term follow-up of three months which was selected in order to afford the than the SMD of 1822 steps/d found in the meta-analysis by Qiu et al.40 This may be control group the opportunity to engage in the program after finishing their control period. explained by the additional support, such as counseling or telephonic support, that was In this manner, an increased withdrawal rate in the control group could be prevented. provided in the studies reviewed by Qiu et al. In a recent comparable, but larger, study of However, a longer duration may have been beneficial for incorporating lifestyle habits and Dasgupta et al, physical activity was increased with 1190 steps/d, CI [550-1840], and HbA1c for a longer-term comparison with the control group. Second, this study had a relatively was decreased by 0.38% compared to the control group.41 This is consistent with the results small sample size. Although we met the minimal number of participants necessary from our found in our study. sample size calculation, a larger group would have strengthened the results. Third, this study The secondary health outcomes showed no significant changes, even in the missed an objective physical activity instrument for both groups. Because the Fitbit was an responders vs. non-responders’ analysis. AGEs are complex linkage products measured in important component of the intervention, the control group did not receive a Fitbit. the tissue and because the accumulation of AGEs also depends on other factors such as However, from the baseline step measurements that the control group made after finishing smoking behavior and intake of certain foods,34 it is likely that more long-term lifestyle their control period, it appeared that they walked the same average steps/d compared to changes are necessary to achieve significant results on AGEs. Also, for body weight and the baseline steps/d of the intervention group. This reinforces our finding that the Waist-Hip-Ratio no effects were found. As an increase of at least 2000 steps/d is needed to intervention group increased their average steps/d. A strength was that the study was achieve relevant effects on weight and body composition,8,42 the lack of findings on these conducted in a general hospital setting, embedded in the usual nursing care for patients with measures is probably explained by the smaller increase in the number of steps/d found in type 2 diabetes. This real-life setting simplifies an extrapolation of the data to the general 6 this study. population with type 2 diabetes. Also, as determined from the evaluations, most participants were satisfied with the self-tracking program as 90% indicated that the activity tracker was Interestingly, social norm was a significant confounder for the HbA1c results, and useful or very useful for them. This indicates that patients with type 2 diabetes are willing to responders had a higher social norm score at baseline compared to non-responders. This engage in lifestyle programs based on self-tracking. result emphasizes the importance of taking into account the social support of a patient with type 2 diabetes. In accordance with this, the use of theory based BCTs within self-monitoring devices or lifestyle interventions is receiving increasing attention within physical activity Conclusions research. Several studies pointed out the importance of providing information, goal-setting, action planning, self-monitoring, barrier identification, personalized feedback, and In our study, self-tracking of physical activity did improve physical activity in patients with rewards.17,25,43 These BCTs were present in our study, however, other important BCTs such advanced type 2 diabetes. The intervention did not improve glycemic control overall, due to as facilitating social support, changing environmental factors, and the use of follow-up a large inter-individual variability in responsiveness to the intervention. However, it did prompts were lacking in our intervention at a structural basis.43–46 In addition, the BCT relevantly improve glycemic control in 56% of the participants who increased their physical ‘action planning’ was incorporated within the informational documents, but not structurally activity with a minimum of 1000 steps/d. To improve the effectiveness of online self-tracking tailored for individual participants. Since these BCTs have been found to be associated with programs on health outcomes, more development in the ease of use of eHealth technology, improved intervention outcomes in people with diabetes, the lack of an overall effect may integration of behavioral change techniques, and tailoring of intervention programs is well be explained by insufficient structural implementation of these BCTs in our study.45,46 required, for example, based on presence of social support. This should be improved in future studies; for instance, social support may be enhanced by creating a system in which patients with diabetes are connected to each other, and/or are facilitated to meet for walking groups. Next to the inclusion of additional evidence based Acknowledgements BCTs, effects of future programs may also be enhanced by improving the ease of use of The authors would like to thank all members of the Living Lab project group ‘Active Ageing Diabetes’ digital techniques, incorporating additional advances to use personal generated health data for their help and assistance in developing the eHealth program and its content and the diabetes nurses from the Bethesda Diabetes Research Center and the Martini Hospital (Ellen Wessels, Brenda in a meaningful way,20 and providing technical support from a person other than the nurse. Wierbos, and Anneke Kleine), for their dedicated guidance and treatment of the patients and for 20,29 This will enhance the role of the nurse for providing personal lifestyle support. Future their accurate data collection.

119 Chapter 6

References 21. Arnrich B, Mayora O, Bardram J, Trster G. Pervasive healthcare. Methods of Information in Medicine. 2010;49(1):67-73. 22. Van Gemert-Pijnen, J.E.W.C, Peters,O., & Ossebaard HC. Improving eHealth. Eleven International Publishing; 2013. http://lib.myilibrary.com?ID=673579. 1. International Diabetes Federation. IDF Diabetes Atlas, 8th edn. Brussels, Belgium: 23. Sanders JP, Loveday A, Pearson N, et al. Devices for self-monitoring sedentary time or physical activity: International Diabetes Federation, 2017. http://www.diabetesatlas.org. Accessed on 24-08-2017. a scoping review. Journal of Medical Internet Research. 2016;18(5). 2. Madden KM. Evidence for the benefit of exercise therapy in patients with type 2 diabetes. Diabetes, 24. El-Gayar O, Timsina P, Nawar N, Eid W. A systematic review of IT for diabetes self-management: Are Metabolic Syndrome and Obesity: Targets and Therapy. 2013;6:233-239. doi:10.2147/DMSO.S32951. we there yet? International Journal of Medical Informatics. 2013;82(8):637-652. 3. Lee IM, Shiroma EJ, Lobelo F, et al. Effect of physical inactivity on major non-communicable diseases doi:10.1016/j.ijmedinf.2013.05.006. worldwide: An analysis of burden of disease and life expectancy. Lancet. 2012;380(9838):219-229. 25. Cotter AP, Durant N, Agne AA, Cherrington AL. Internet interventions to support lifestyle modification doi:10.1016/S0140-6736(12)61031-9. for diabetes management: A systematic review of the evidence. Journal of Diabetes and its 4. Diabetes prevention program research group. Reduction in the Incidence of Type 2 Diabetes With Complications. 2014;28(2):243-251. doi:10.1016/j.jdiacomp.2013.07.003. Lifestyle Intervention or Metformin. The New England Journal of Medicine. 2006;346(6):393-403. 26. McMillan KA, Kirk A, Hewitt A, MacRury S. A systematic and integrated review of mobile-based 5. Diabetes prevention program research group. Long-term effects of lifestyle intervention or metformin technology to promote active lifestyles in people with type 2 diabetes. Journal of Diabetes Science and on diabetes development and microvascular complications over 15-year follow-up: the Diabetes Technology. 2017;11(2):299-307. Prevention Program Outcomes Study. Lancet Diabetes Endocrinology. 2015;3(11):866-875. 27. Old N. Paving the Way for Health Promotion Nurses: An International Perspective. Creative Nursing. doi:10.1016/S2213-8587(15)00291-0. 2014;20(4):222-226. 6. Kirwan JP, Sacks J, Nieuwoudt S. The essential role of exercise in the management of type 2 diabetes. 28. Frank JR, Danoff D. The CanMEDS initiative: implementing an outcomes-based framework of physician Cleveland Clinic Journal of Medicine. 2017;84(7 Suppl 1):S15. competencies. Medical Teacher. 2007;29(7):642-647. 7. Haskell WL, Lee IM, Pate RR, et al. Physical activity and public health: Updated recommendation for 29. American Diabetes Association. Standards of medical care in diabetes-2011. Diabetes Care. adults from the American College of Sports Medicine and the American Heart Association. Medicine & 2011;34(SUPPL.1):11. doi:10.2337/dc11-S011. Science in Sports & Exercise. 2007;39(8):1423-1434. doi:10.1249/mss.0b013e3180616b27. 30. Kernan WN, Viscoli CM, Makuch RW, Brass LM, Horwitz RI. Stratified randomization for clinical trials. 8. Tudor-Locke C, Craig CL, Brown WJ, et al. How many steps/day are enough? for adults. International Journal of Clinical Epidemiology. 1999;52(1):19-26. doi:10.1016/S0895-4356(98)00138-3. Journal of Behavioral Nutrition and Physical Activity. 2011;8(1):79. doi:10.1186/1479-5868-8-79. 31. Milton K, Bull FC, Bauman A. Reliability and validity testing of a single-item physical activity measure. 9. Brannan M, Varney J, Timpson C, Foster C, Murphy M. 10 Minutes Brisk Walking Each Day in Mid-Life British Journal of Sports Medicine. 2011;45(3):203-208. doi:10.1136/bjsm.2009.068395. for Health Benefits and towards Achieving Physical Activity Recommendations. Evidence Summary; 32. Kooiman TJM, Dontje ML, Sprenger SR, Krijnen WP, der Schans van, de Groot M. Reliability and validity 2017. https://www.gov.uk/government/publications/everybody-active-every-day-a-framework-to- of ten consumer activity trackers. BMC Sports Science, Medicine and Rehabilitation. 2015;7(1):1-11. embed-physical-activity-into-daily-life. doi:10.1186/s13102-015-0018-5. 10. Tully MA, Panter J, Ogilvie D. Individual characteristics associated with mismatches between self- 33. Meerwaldt R, Graaff R, Oomen PHN, et al. Simple non-invasive assessment of advanced glycation reported and accelerometer-measured physical activity. PLoS One. 2014;9(6):e99636. endproduct accumulation. Diabetologia. 2004;47(7):1324-1330. doi:10.1007/s00125-004-1451-2. doi:10.1371/journal.pone.0099636. 34. Bos D, de Ranitz-Greven W, de Valk H. Advanced Glycation End Products, Measured as skin 11. García-Pérez, L. E., Álvarez, M., Dilla, T., Gil-Guillén, V., & Orozco-Beltrán D. Adherence to Therapies in autofluorescence and diabetes complications, a systematic review. Diabetes Technology & Patients with Type 2 Diabetes. Diabetes Therapy. 2013;4(2):175-194. doi:10.1007/s13300-013-0034-y. Therapeutics. 2011;13(7):773-779. doi:10.1089/dia.2011.0034. 12. Look AHEAD research group. Cardiovascular Effects of Intensive Lifestyle Intervention in Type 2 35. World Health Organization. Measuring obesity—classification and description of anthropometric data. Diabetes — NEJM. The New England Journal of Medicine. 2013;(369):145-154. Report on a WHO consultation of the epidemiology of obesity. Warsaw 21-23 October 1987. doi:10.1056/NEJMoa1212914. Copenhagen: WHO, 1989. Nutrition Unit Document EUR/ICP/NUT. 1987;123. 13. Justine M, Azian A, Hassan V, Manaf H. Barriers to participation in physical activity and exercise among 36. Janssen MF, Pickard AS, Golicki D, et al. Measurement properties of the EQ-5D-5L compared to the EQ- middle-aged and elderly individuals. Singapore Medical Journal. 2013;54(10):581-586. 5D-3L across eight patient groups: A multi-country study. Quality of Life Research. 2013;22(7):1717- 14. Bandura A. Health Promotion by Social Cognitive Means. Health Education & Behavior. 1727. doi:10.1007/s11136-012-0322-4. 2004;31(2):143-164. doi:10.1177/1090198104263660. 37. Ajzen I. The theory of planned behavior. Organizational Behavior and Human Decision Processes. 15. Bandura A. Health promotion from the perspective of social cognitive theory. Psychology & Health. 1991;50(2):179-211. doi:10.1016/0749-5978(91)90020-T. 1998;13(4):623-649. 38. Boudreau F, Godin G. Participation in Regular Leisure-Time Physical Activity Among Individuals with 16. Mann T, de Ridder D, Fujita K. Self-Regulation of Health Behavior. Health Psychology. 2013;32(5):487- Type 2 Diabetes Not Meeting Canadian Guidelines: the Influence of Intention, Perceived Behavioral 498. doi:10.1037/a0028533. Control, and Moral Norm. International Journal of Behavioral Medicine. 2014;21(6):918-926. 17. Michie S, Ashford S, Sniehotta FF, Dombrowski SU, Bishop A, French DP. A refined taxonomy of doi:10.1007/s12529-013-9380-4. behaviour change techniques to help people change their physical activity and healthy eating 39. Tudor-Locke C, Craig CL, Thyfault JP, Spence JC. A step-defined sedentary lifestyle index: <5000 behaviours: The CALO-RE taxonomy. Psychology &Health. 2011;26(11):1479-1498. steps/day. Applied Physiology, Nutrition, and Metabolism. 2013;38(4):100-114. doi:10.1139/apnm- doi:10.1080/08870446.2010.540664. 2012-0235. 18. Michie S, Richardson M, Johnston M, et al. The behavior change technique taxonomy (v1) of 93 40. Qiu S, Cai X, Chen X, Yang B, Sun Z. Step counter use in type 2 diabetes: a meta-analysis of randomized hierarchically clustered techniques: building an international consensus for the reporting of behavior controlled trials. BMC Medicine. 2014;12(1):36. doi:10.1186/1741-7015-12-36. change interventions. Annals of Behavioral Medicine. 2013;46(1):81-95. 41. Dasgupta K, Rosenberg E, Joseph L, et al. Physician Step prescription and Monitoring to improve 19. Kluger AN, DeNisi A. The effects of feedback interventions on performance: A historical review, a ARTERial health (SMARTER): a randomized controlled trial in type 2 diabetes and hypertension. meta-analysis, and a preliminary feedback intervention theory. Psychological Bulletin. Diabetes, Obesity and Metabolism. 2017. 1996;119(2):254-284. doi:10.1037/0033-2909.119.2.254. 42. Musto AA. The effects of an incremental pedometer program on metabolic syndrome components in 20. Greenwood DA, Gee PM, Fatkin KJ, Peeples M. A Systematic Review of Reviews Evaluating Technology- sedentary overweight women. Dissertation abstracts international. 2008;69(3-B):1598. Enabled Diabetes Self-Management Education and Support. Journal of Diabetes Science and http://search.ebscohost.com/login.aspx?direct=true&AuthType=ip,shib&db=psyh&AN=2008-99180- Technology. 2017:1932296817713506. 343&site=ehost-live&custid=s4121186.

120 Self-tracking of physical activity in people with type 2 diabetes

References 21. Arnrich B, Mayora O, Bardram J, Trster G. Pervasive healthcare. Methods of Information in Medicine. 2010;49(1):67-73. 22. Van Gemert-Pijnen, J.E.W.C, Peters,O., & Ossebaard HC. Improving eHealth. Eleven International Publishing; 2013. http://lib.myilibrary.com?ID=673579. 1. International Diabetes Federation. IDF Diabetes Atlas, 8th edn. Brussels, Belgium: 23. Sanders JP, Loveday A, Pearson N, et al. Devices for self-monitoring sedentary time or physical activity: International Diabetes Federation, 2017. http://www.diabetesatlas.org. Accessed on 24-08-2017. a scoping review. Journal of Medical Internet Research. 2016;18(5). 2. Madden KM. Evidence for the benefit of exercise therapy in patients with type 2 diabetes. Diabetes, 24. El-Gayar O, Timsina P, Nawar N, Eid W. A systematic review of IT for diabetes self-management: Are Metabolic Syndrome and Obesity: Targets and Therapy. 2013;6:233-239. doi:10.2147/DMSO.S32951. we there yet? International Journal of Medical Informatics. 2013;82(8):637-652. 3. Lee IM, Shiroma EJ, Lobelo F, et al. Effect of physical inactivity on major non-communicable diseases doi:10.1016/j.ijmedinf.2013.05.006. worldwide: An analysis of burden of disease and life expectancy. Lancet. 2012;380(9838):219-229. 25. Cotter AP, Durant N, Agne AA, Cherrington AL. Internet interventions to support lifestyle modification doi:10.1016/S0140-6736(12)61031-9. for diabetes management: A systematic review of the evidence. Journal of Diabetes and its 4. Diabetes prevention program research group. Reduction in the Incidence of Type 2 Diabetes With Complications. 2014;28(2):243-251. doi:10.1016/j.jdiacomp.2013.07.003. Lifestyle Intervention or Metformin. The New England Journal of Medicine. 2006;346(6):393-403. 26. McMillan KA, Kirk A, Hewitt A, MacRury S. A systematic and integrated review of mobile-based 5. Diabetes prevention program research group. Long-term effects of lifestyle intervention or metformin technology to promote active lifestyles in people with type 2 diabetes. Journal of Diabetes Science and on diabetes development and microvascular complications over 15-year follow-up: the Diabetes Technology. 2017;11(2):299-307. Prevention Program Outcomes Study. Lancet Diabetes Endocrinology. 2015;3(11):866-875. 27. Old N. Paving the Way for Health Promotion Nurses: An International Perspective. Creative Nursing. doi:10.1016/S2213-8587(15)00291-0. 2014;20(4):222-226. 6. Kirwan JP, Sacks J, Nieuwoudt S. The essential role of exercise in the management of type 2 diabetes. 28. Frank JR, Danoff D. The CanMEDS initiative: implementing an outcomes-based framework of physician Cleveland Clinic Journal of Medicine. 2017;84(7 Suppl 1):S15. competencies. Medical Teacher. 2007;29(7):642-647. 7. Haskell WL, Lee IM, Pate RR, et al. Physical activity and public health: Updated recommendation for 29. American Diabetes Association. Standards of medical care in diabetes-2011. Diabetes Care. adults from the American College of Sports Medicine and the American Heart Association. Medicine & 2011;34(SUPPL.1):11. doi:10.2337/dc11-S011. Science in Sports & Exercise. 2007;39(8):1423-1434. doi:10.1249/mss.0b013e3180616b27. 30. Kernan WN, Viscoli CM, Makuch RW, Brass LM, Horwitz RI. Stratified randomization for clinical trials. 8. Tudor-Locke C, Craig CL, Brown WJ, et al. How many steps/day are enough? for adults. International Journal of Clinical Epidemiology. 1999;52(1):19-26. doi:10.1016/S0895-4356(98)00138-3. Journal of Behavioral Nutrition and Physical Activity. 2011;8(1):79. doi:10.1186/1479-5868-8-79. 31. Milton K, Bull FC, Bauman A. Reliability and validity testing of a single-item physical activity measure. 9. Brannan M, Varney J, Timpson C, Foster C, Murphy M. 10 Minutes Brisk Walking Each Day in Mid-Life British Journal of Sports Medicine. 2011;45(3):203-208. doi:10.1136/bjsm.2009.068395. 6 for Health Benefits and towards Achieving Physical Activity Recommendations. Evidence Summary; 32. Kooiman TJM, Dontje ML, Sprenger SR, Krijnen WP, der Schans van, de Groot M. Reliability and validity 2017. https://www.gov.uk/government/publications/everybody-active-every-day-a-framework-to- of ten consumer activity trackers. BMC Sports Science, Medicine and Rehabilitation. 2015;7(1):1-11. embed-physical-activity-into-daily-life. doi:10.1186/s13102-015-0018-5. 10. Tully MA, Panter J, Ogilvie D. Individual characteristics associated with mismatches between self- 33. Meerwaldt R, Graaff R, Oomen PHN, et al. Simple non-invasive assessment of advanced glycation reported and accelerometer-measured physical activity. PLoS One. 2014;9(6):e99636. endproduct accumulation. Diabetologia. 2004;47(7):1324-1330. doi:10.1007/s00125-004-1451-2. doi:10.1371/journal.pone.0099636. 34. Bos D, de Ranitz-Greven W, de Valk H. Advanced Glycation End Products, Measured as skin 11. García-Pérez, L. E., Álvarez, M., Dilla, T., Gil-Guillén, V., & Orozco-Beltrán D. Adherence to Therapies in autofluorescence and diabetes complications, a systematic review. Diabetes Technology & Patients with Type 2 Diabetes. Diabetes Therapy. 2013;4(2):175-194. doi:10.1007/s13300-013-0034-y. Therapeutics. 2011;13(7):773-779. doi:10.1089/dia.2011.0034. 12. Look AHEAD research group. Cardiovascular Effects of Intensive Lifestyle Intervention in Type 2 35. World Health Organization. Measuring obesity—classification and description of anthropometric data. Diabetes — NEJM. The New England Journal of Medicine. 2013;(369):145-154. Report on a WHO consultation of the epidemiology of obesity. Warsaw 21-23 October 1987. doi:10.1056/NEJMoa1212914. Copenhagen: WHO, 1989. Nutrition Unit Document EUR/ICP/NUT. 1987;123. 13. Justine M, Azian A, Hassan V, Manaf H. Barriers to participation in physical activity and exercise among 36. Janssen MF, Pickard AS, Golicki D, et al. Measurement properties of the EQ-5D-5L compared to the EQ- middle-aged and elderly individuals. Singapore Medical Journal. 2013;54(10):581-586. 5D-3L across eight patient groups: A multi-country study. Quality of Life Research. 2013;22(7):1717- 14. Bandura A. Health Promotion by Social Cognitive Means. Health Education & Behavior. 1727. doi:10.1007/s11136-012-0322-4. 2004;31(2):143-164. doi:10.1177/1090198104263660. 37. Ajzen I. The theory of planned behavior. Organizational Behavior and Human Decision Processes. 15. Bandura A. Health promotion from the perspective of social cognitive theory. Psychology & Health. 1991;50(2):179-211. doi:10.1016/0749-5978(91)90020-T. 1998;13(4):623-649. 38. Boudreau F, Godin G. Participation in Regular Leisure-Time Physical Activity Among Individuals with 16. Mann T, de Ridder D, Fujita K. Self-Regulation of Health Behavior. Health Psychology. 2013;32(5):487- Type 2 Diabetes Not Meeting Canadian Guidelines: the Influence of Intention, Perceived Behavioral 498. doi:10.1037/a0028533. Control, and Moral Norm. International Journal of Behavioral Medicine. 2014;21(6):918-926. 17. Michie S, Ashford S, Sniehotta FF, Dombrowski SU, Bishop A, French DP. A refined taxonomy of doi:10.1007/s12529-013-9380-4. behaviour change techniques to help people change their physical activity and healthy eating 39. Tudor-Locke C, Craig CL, Thyfault JP, Spence JC. A step-defined sedentary lifestyle index: <5000 behaviours: The CALO-RE taxonomy. Psychology &Health. 2011;26(11):1479-1498. steps/day. Applied Physiology, Nutrition, and Metabolism. 2013;38(4):100-114. doi:10.1139/apnm- doi:10.1080/08870446.2010.540664. 2012-0235. 18. Michie S, Richardson M, Johnston M, et al. The behavior change technique taxonomy (v1) of 93 40. Qiu S, Cai X, Chen X, Yang B, Sun Z. Step counter use in type 2 diabetes: a meta-analysis of randomized hierarchically clustered techniques: building an international consensus for the reporting of behavior controlled trials. BMC Medicine. 2014;12(1):36. doi:10.1186/1741-7015-12-36. change interventions. Annals of Behavioral Medicine. 2013;46(1):81-95. 41. Dasgupta K, Rosenberg E, Joseph L, et al. Physician Step prescription and Monitoring to improve 19. Kluger AN, DeNisi A. The effects of feedback interventions on performance: A historical review, a ARTERial health (SMARTER): a randomized controlled trial in type 2 diabetes and hypertension. meta-analysis, and a preliminary feedback intervention theory. Psychological Bulletin. Diabetes, Obesity and Metabolism. 2017. 1996;119(2):254-284. doi:10.1037/0033-2909.119.2.254. 42. Musto AA. The effects of an incremental pedometer program on metabolic syndrome components in 20. Greenwood DA, Gee PM, Fatkin KJ, Peeples M. A Systematic Review of Reviews Evaluating Technology- sedentary overweight women. Dissertation abstracts international. 2008;69(3-B):1598. Enabled Diabetes Self-Management Education and Support. Journal of Diabetes Science and http://search.ebscohost.com/login.aspx?direct=true&AuthType=ip,shib&db=psyh&AN=2008-99180- Technology. 2017:1932296817713506. 343&site=ehost-live&custid=s4121186.

121 Chapter 6

43. Sullivan AN, Lachman ME. Behavior Change with Fitness Technology in Sedentary Adults: A Review of the Evidence for Increasing Physical Activity. Frontiers in Public Health. 2017;4(289):1. doi:10.3389/fpubh.2016.00289. 44. Gardner B, Smith L, Lorencatto F, Hamer M, Biddle SJ. How to reduce sitting time? A review of behaviour change strategies used in sedentary behaviour reduction interventions among adults. Health Psychology Review. 2016;10(1):89-112. doi:10.1080/17437199.2015.1082146. 45. Avery L, Flynn D, Wersch A Van, Sniehotta FF, Trenell MI. Changing Physical Activity Behavior in Type 2 Diabetes. Diabetes Care. 2012;35(12):2681-2689. 46. Cheng L, Sit JWH, Choi K, Chair S, Li X, He X. Effectiveness of Interactive Self-Management Interventions in Individuals With Poorly Controlled Type 2 Diabetes: A Meta-Analysis of Randomized Controlled Trials. Worldviews on Evidence-Based Nursing. 2017;14(1):65-73.

122

Chapter 7 | The role of self-regulation in the effect of self-tracking of physical activity and weight on BMI

Thea J.M. Kooiman Arie Dijkstra Adriaan Kooy Cees P. van der Schans Martijn de Groot

Submitted

Chapter 7 | The role of self-regulation in the effect of self-tracking of physical activity and weight on BMI

Thea J.M. Kooiman Arie Dijkstra Adriaan Kooy Cees P. van der Schans Martijn de Groot

Submitted

Chapter 7

Abstract Introduction

Purpose Overweight and physical inactivity are known risk factors for a number of chronic diseases To assess to what extent a change in self-regulation capabilities can explain weight loss after such as diabetes, cardiovascular disease, and cancer.1 Self-quantification of health has been four and 12 months of self-tracking physical activity and weight. suggested as a possible way to create awareness about individual health and to stimulate optimization of different health behaviors and health outcomes.2,3 Self-tracking of physical Methods activity and weight are two ways of self-quantification that have been previously studied. The study was part of the Lifelines Cohort Study in Northern Netherlands. Healthy adult Several studies determined an increase of physical activity as a result of self-tracking of volunteers (N=95) received a digital weight scale and an activity tracker. physical activity in different populations both with and without additional intervention Personal characteristics as well as the intention to change weight and physical activity were components.4,5 In addition, frequent self-weighing has been found to be an effective measured at T0 (baseline). Self-regulation capabilities (goal orientation, self-direction, stimulation to lose weight.6–9 decision making, and impulse control) and body weight were measured at T0, T1 (after four months), and T2 (after 12 months). Pearson correlation, univariate, and multivariate linear However, although self-tracking of physical activity and weight are considered as regression analysis were used to examine the relationship between BMI and (change in) self- promising intervention strategies, they may not help every person to acquire a more active regulation capabilities. lifestyle or to lose weight. A complete picture about which individual does or does not achieve lifestyle changes and weight loss as a result of using self-tracking technology and the psychological factors that may play a role in weight outcomes by using it is currently lacking.9 Results Therefore, there is a need for research aimed at identifying the psychological working At T0, all four dimensions of self-regulation were negatively related to BMI (p<.01). At T1, mechanism of this technology. weight significantly declined (-2.0 kg/-0.64 kg/m2, p<.001). At T2, weight still declined compared to T0 (-1.8 kg/-0.57 kg/m2, p<.01). At T1, intention to lose weight, self-weighing There are several theoretical models that may be used to understand self-tracking of frequency, and an increase in goal orientation explained weight loss. At T2, an increase in behavior including the Social Cognitive Theory, Temporal Self-regulation Theory, Feedback decision making explained weight loss. Theory, and Control Theory.10–15 According to the Control Theory, people regulate their behavior by discrepancy reducing feedback loops. This is done by a reference value (i.e., an Conclusion individual’s goal or a standard), an input function (i.e., an error detector), a comparator, and 15 Incremental self-regulation capabilities explain weight loss after engaging in self-tracking of an output function (i.e., cognitive and behavioral processes). The Control Theory argues physical activity and weight. that awareness of a person’s own behavior is the first step towards being able to make behavioral changes. Subsequently, this behavioral change leads to a new feedback loop in which progress towards ones’ goal can be seen which enhances learning and motivation. This theory can be effectively applied for self-tracking of health (behavior) whereby a self- tracking device serves as the error detector between one’s current state and the desired state. Also, the other afore-mentioned theories emphasize goal-setting, self-monitoring, and feedback as being important principles for health behavior change. Therefore, these principles are also employed as a basis for Behavior Change Techniques (BCT’s)16,17 that are being increasingly incorporated within consumer self-tracking technology.18,19 We propose that these behavior change techniques within consumer self-tracking devices can impact a person’s self-regulation capabilities and subsequently explain weight loss. Figure 1 illustrates this proposed working mechanism on weight loss induced by self-tracking technology based on the principles of Control Theory.

126 The role of self-regulation in the effect of self-tracking of physical activity and weight on BMI

Abstract Introduction

Purpose Overweight and physical inactivity are known risk factors for a number of chronic diseases To assess to what extent a change in self-regulation capabilities can explain weight loss after such as diabetes, cardiovascular disease, and cancer.1 Self-quantification of health has been four and 12 months of self-tracking physical activity and weight. suggested as a possible way to create awareness about individual health and to stimulate optimization of different health behaviors and health outcomes.2,3 Self-tracking of physical Methods activity and weight are two ways of self-quantification that have been previously studied. The study was part of the Lifelines Cohort Study in Northern Netherlands. Healthy adult Several studies determined an increase of physical activity as a result of self-tracking of volunteers (N=95) received a digital weight scale and an activity tracker. physical activity in different populations both with and without additional intervention Personal characteristics as well as the intention to change weight and physical activity were components.4,5 In addition, frequent self-weighing has been found to be an effective measured at T0 (baseline). Self-regulation capabilities (goal orientation, self-direction, stimulation to lose weight.6–9 decision making, and impulse control) and body weight were measured at T0, T1 (after four months), and T2 (after 12 months). Pearson correlation, univariate, and multivariate linear However, although self-tracking of physical activity and weight are considered as regression analysis were used to examine the relationship between BMI and (change in) self- promising intervention strategies, they may not help every person to acquire a more active regulation capabilities. lifestyle or to lose weight. A complete picture about which individual does or does not achieve lifestyle changes and weight loss as a result of using self-tracking technology and the psychological factors that may play a role in weight outcomes by using it is currently lacking.9 Results Therefore, there is a need for research aimed at identifying the psychological working At T0, all four dimensions of self-regulation were negatively related to BMI (p<.01). At T1, mechanism of this technology. weight significantly declined (-2.0 kg/-0.64 kg/m2, p<.001). At T2, weight still declined compared to T0 (-1.8 kg/-0.57 kg/m2, p<.01). At T1, intention to lose weight, self-weighing There are several theoretical models that may be used to understand self-tracking of 7 frequency, and an increase in goal orientation explained weight loss. At T2, an increase in behavior including the Social Cognitive Theory, Temporal Self-regulation Theory, Feedback decision making explained weight loss. Theory, and Control Theory.10–15 According to the Control Theory, people regulate their behavior by discrepancy reducing feedback loops. This is done by a reference value (i.e., an Conclusion individual’s goal or a standard), an input function (i.e., an error detector), a comparator, and 15 Incremental self-regulation capabilities explain weight loss after engaging in self-tracking of an output function (i.e., cognitive and behavioral processes). The Control Theory argues physical activity and weight. that awareness of a person’s own behavior is the first step towards being able to make behavioral changes. Subsequently, this behavioral change leads to a new feedback loop in which progress towards ones’ goal can be seen which enhances learning and motivation. This theory can be effectively applied for self-tracking of health (behavior) whereby a self- tracking device serves as the error detector between one’s current state and the desired state. Also, the other afore-mentioned theories emphasize goal-setting, self-monitoring, and feedback as being important principles for health behavior change. Therefore, these principles are also employed as a basis for Behavior Change Techniques (BCT’s)16,17 that are being increasingly incorporated within consumer self-tracking technology.18,19 We propose that these behavior change techniques within consumer self-tracking devices can impact a person’s self-regulation capabilities and subsequently explain weight loss. Figure 1 illustrates this proposed working mechanism on weight loss induced by self-tracking technology based on the principles of Control Theory.

127 Chapter 7

Methods

Design A 12-month prospective intervention study was conducted within the Lifelines Cohort Study. Eligible participants were provided with an activity tracker and a digital weight scale. Participants completed a digital questionnaire at the beginning of the study (T0), after four months (T1), and after 12 months (T2).

Sample Participants were recruited within the Lifelines Cohort Study in the Netherlands. Lifelines is a multi-disciplinary prospective population-based cohort study examining in a unique three- generation design the health and health-related behaviors of 167,729 persons living in the North of The Netherlands. It employs a broad range of investigative procedures in assessing Figure 1. Proposed working mechanism of self-tracking of health based on the Control Theory. The health goal serves as the biomedical, socio-demographic, behavioral, physical and psychological factors which a reference value, while the self-tracking device functions as the error detector. The self-tracking individual is contribute to the health and disease of the general population, with a special focus on multi- the comparator. Both the possible increment of self-regulation and health outcomes functions as output. morbidity and complex genetics. Inclusion criteria of the participants were: ≥ 25 years and access to a smartphone with internet (IOS or Android). Participants were excluded if they were already in the possession of an activity monitor or smart weight scale or were not able Self-regulation of behavior is defined as an individual’s ability to establish, to engage in self-tracking of physical activity, sleep, or weight due to physical, social, implement, and monitor goals in order to successfully regulate their own behavior.10,13,20 cognitive, and/or mental problems. Participants came to the research office of Lifelines to This encompasses both behavioral, cognitive, and emotional processes.10 According to pick up their devices and an explanatory guide on how to install them. Informed consent was Gavora et al, self-regulation can be divided into four different dimensions; goal orientation obtained from all of the participants. Ethical approval was granted within the Lifelines (the degree to which an individual attempts to fulfill personal goals, e.g., by plan making), program by the University Medical Center Groningen (METc 2007/152) based on the self-direction (the degree in which one can formulate learning goals and learns from declaration of Helsinki of Ethical Principles for Medical Research Involving Human Subjects. previous experiences), decision-making (the ability to make decisions and find multiple ways to achieve goals), and impulse control (the ability for an individual to manage short-term interferences with goals). These dimensions are considered as being different but not fully Self-tracking devices autonomous processes for self-regulation. The different processes may occur simultaneously The Nokia Pulse, (Nokia, Nozay, France, previously Withings, Issy les Moulineaux, France) or at different moments,21,22 and provide the ability to distinguish which specific process of measured physical activity and sleep. The Nokia WS-30 (Nokia, Nozay, France, previously self-regulation is most important in particular situations or for types of behavior. Therefore, Withings, Issy les Moulineaux, France), measured weight and body mass index (BMI). The these different dimensions will be used in this study. devices were connected with a smartphone application (Nokia Health Mate) which In summary, the body of knowledge regarding the effects of self-quantification of graphically showed the individual’s personal health data retrieved from the devices over health is increasing. However, the mechanism behind the effect is still unclear. In this study, time and provided automatically generated personalized feedback messages concerning we will provide participants with two devices for self-quantification of physical activity and progression towards the self-selected goals of the participants. Also, social features such as weight. The primary aim of this study is to assess to what extent a change in self-regulation the possibility to connect with friends existed in this app. If the participants lost or broke capabilities can explain weight loss after four and 12 months of self-tracking physical activity their activity tracker or when technical problems with the Pulse occurred, the Pulse was and weight. Additionally, we aim to examine if weight loss is different for people who did or replaced during the first six months of the study. Thereafter, no replacement was possible did not record self-reported lifestyle changes. due to a restricted availability of the Pulse activity tracker.

128 The role of self-regulation in the effect of self-tracking of physical activity and weight on BMI

Methods

Sample Participants were recruited within the Lifelines Cohort Study in the Netherlands. Lifelines is a multi-disciplinary prospective population-based cohort study examining in a unique three- generation design the health and health-related behaviors of 167,729 persons living in the North of The Netherlands. It employs a broad range of investigative procedures in assessing Figure 1. Proposed working mechanism of self-tracking of health based on the Control Theory. The health goal serves as the biomedical, socio-demographic, behavioral, physical and psychological factors which a reference value, while the self-tracking device functions as the error detector. The self-tracking individual is contribute to the health and disease of the general population, with a special focus on multi- the comparator. Both the possible increment of self-regulation and health outcomes functions as output. morbidity and complex genetics. Inclusion criteria of the participants were: ≥ 25 years and access to a smartphone with internet (IOS or Android). Participants were excluded if they were already in the possession of an activity monitor or smart weight scale or were not able Self-regulation of behavior is defined as an individual’s ability to establish, to engage in self-tracking of physical activity, sleep, or weight due to physical, social, implement, and monitor goals in order to successfully regulate their own behavior.10,13,20 cognitive, and/or mental problems. Participants came to the research office of Lifelines to This encompasses both behavioral, cognitive, and emotional processes.10 According to 7 pick up their devices and an explanatory guide on how to install them. Informed consent was Gavora et al, self-regulation can be divided into four different dimensions; goal orientation obtained from all of the participants. Ethical approval was granted within the Lifelines (the degree to which an individual attempts to fulfill personal goals, e.g., by plan making), program by the University Medical Center Groningen (METc 2007/152) based on the self-direction (the degree in which one can formulate learning goals and learns from declaration of Helsinki of Ethical Principles for Medical Research Involving Human Subjects. previous experiences), decision-making (the ability to make decisions and find multiple ways to achieve goals), and impulse control (the ability for an individual to manage short-term interferences with goals). These dimensions are considered as being different but not fully Self-tracking devices autonomous processes for self-regulation. The different processes may occur simultaneously The Nokia Pulse, (Nokia, Nozay, France, previously Withings, Issy les Moulineaux, France) or at different moments,21,22 and provide the ability to distinguish which specific process of measured physical activity and sleep. The Nokia WS-30 (Nokia, Nozay, France, previously self-regulation is most important in particular situations or for types of behavior. Therefore, Withings, Issy les Moulineaux, France), measured weight and body mass index (BMI). The these different dimensions will be used in this study. devices were connected with a smartphone application (Nokia Health Mate) which In summary, the body of knowledge regarding the effects of self-quantification of graphically showed the individual’s personal health data retrieved from the devices over health is increasing. However, the mechanism behind the effect is still unclear. In this study, time and provided automatically generated personalized feedback messages concerning we will provide participants with two devices for self-quantification of physical activity and progression towards the self-selected goals of the participants. Also, social features such as weight. The primary aim of this study is to assess to what extent a change in self-regulation the possibility to connect with friends existed in this app. If the participants lost or broke capabilities can explain weight loss after four and 12 months of self-tracking physical activity their activity tracker or when technical problems with the Pulse occurred, the Pulse was and weight. Additionally, we aim to examine if weight loss is different for people who did or replaced during the first six months of the study. Thereafter, no replacement was possible did not record self-reported lifestyle changes. due to a restricted availability of the Pulse activity tracker.

129 Chapter 7

The four dimensions of self-regulation were measured with the self-regulation questionnaire.20 This questionnaire was slightly modified to increase specificity for self-

regulation of health behavior (physical activity, sleep, nutrition, and body weight). For example, for ‘decision making’, an item included: ‘normally, I am able to find several ways when I want to change something in my health behavior’, whereby ‘health’ was added to the original item. The average of the scores on the different subscales: goal orientation (5

items), self-direction (7 items) decision-making (7 items), and impulse control (8 items) were calculated according to the grouping of Gavora et al.21 Cronbach’s alpha was calculated with all items belonging to one subscale. These included .69 (when item 31 was deleted), .74, .66,

and .83, respectively. Experienced effects of using the devices were assessed in a self-composed evaluation questionnaire at T1 and at T2. Participants completed questions about whether they had increased their physical activity behavior or changed their eating pattern as a result of using

the devices. Answers could be indicated on a 5-point Likert scale. Subsequently, scores were clustered into three categories for each of the two behaviors; ‘made changes’ (yes, a lot or yes, some), ‘unknown’ (I don’t know) and ‘no changes’ (no, I don’t think so or no, no changes). For the analyses, scores were clustered into ‘made changes in both physical activity and eating pattern, made changes in either physical activity or eating pattern’ and ‘made no changes’.

Analyses Figure 2. Flow of participants through the study All of the variables were evaluated by using descriptive statistics. Then, the relationship between each of the four dimensions of self-regulation capacity (goal-orientation, self-

direction, decision making and impulse control) and BMI at baseline was assessed by using Measures Pearson correlation analysis. BMI change between T0 and T1, T1-T2, and between T0 and T2 Weight and weighing frequency were measured with the Nokia WS-30 weight scale. Weight were assessed by paired samples t-tests. Thereafter, for the time periods when a significant change between T0-T1, T1-T2, and T0-T2 was calculated from the weight self-measurements BMI change was found, it was assessed whether this change was related to an increase of that the participants conducted at these time points. Weighing frequency was calculated the four separate dimensions of self-regulation by using univariate linear regression analysis. from baseline to T1 and from baseline to T2. Subsequently, during those periods, the Hereby, it was also examined whether there were any significant interaction effects for the number of measurements were categorized in a low frequency (self-weighing less than once relationship between the increments of each of the four dimensions of self-regulation and per week), a moderate frequency (self-weighing once or several times per week), and a high weight loss, by baseline weight class (BMI ≥25 vs. <25). Thereafter, predictors for BMI 8 frequency (self-weighing minimally six days per week, i.e., daily self-weighing). All of the changes between T0 and T1 and between T0-T2 were analyzed by assessing personal data (weight and weighing frequency) were retrieved from Nokia Health by Lifelines and characteristics (i.e., age, gender, education, BMI), intention to change weight, intention to anonymously made available for data analyses. change physical activity, self-weighing frequency, and changes in self-regulation capabilities Personal characteristics (age, gender, education, and height) were assessed in a digital in a multivariate linear regression analysis. Significant predictors were analyzed by using the questionnaire at baseline. BMI was assessed using the height and the weight during the first ‘backward’ method. measurement (T0), at T1, and at T2. In addition, to assess our second aim, univariate ANOVA tests were conducted to determine Intention to change weight and intention to change physical activity was measured using two whether weight loss differed among people with differential self-reported changes in 1-item questionnaires. The participant could indicate 1) the intention to gain physical activity and eating patterns as a result of using the devices. All of the analyses were weight/increase activity, 2) no intention to change, or 3) the intention to lose conducted by using SPSS, version 22, 2010, IBM-SPSS Inc. weight/decrease activity.

130 The role of self-regulation in the effect of self-tracking of physical activity and weight on BMI

The four dimensions of self-regulation were measured with the self-regulation questionnaire.20 This questionnaire was slightly modified to increase specificity for self-

Analyses 7 Figure 2. Flow of participants through the study All of the variables were evaluated by using descriptive statistics. Then, the relationship between each of the four dimensions of self-regulation capacity (goal-orientation, self-

131 Chapter 7

Results increase in goal orientation was significantly related to a decrease in BMI at T1. An increase in decision-making was significantly related to a decrease in BMI at T2. A significant

interaction effect was found for BMI class (i.e., BMI <25 vs. ≥25) on the relation between the At baseline, 81 eligible participants filled out the questionnaire and installed both devices. One participant was excluded from the analyses due to her pregnancy during the study increment of self-direction capability and weight loss: for participants with a BMI ≥25, an period. At T1 (four months), 74 participants had completed the questionnaire and, at T2 (12 increase in self-direction was significantly related to weight loss (β= -0.93, p<.01) whereas no months), 59 participants had done so. Together with the number of people who were still significant relationship was found for individuals with a BMI<25 (p=.456, Table 2). measuring their weight at T1 and T2, i.e., had at least one weight measurement at T1/T2 or within a range of two months from T1/T2, this resulted in a study group of N=80 at baseline, Table 2. N=73 at T1, and N=46 at T2 in the combined analyses of weight and questionnaire data. Univariate regression coefficients of change scores in the different self-regulation Figure 2 describes the flow of participants through the study. The mean age (SD) at baseline dimensions on BMI change at T1 and T2. 2 was 48.4 (6.7) years; mean body weight was 78.5 (14.9) kg; and mean BMI 25.9 (3.6) kg/m . BMI change between T0-T1 (N=73)

Change score of specific β SE p-value Association between BMI and self-regulation capabilities at baseline dimension: At baseline, significant negative Pearson correlations were found between BMI and the Goal-orientation -0.45 0.22 .049 different dimensions of the self-regulation questionnaire (r between -.32 and -.43, p<.01). Self-direction BMI < 25 0.25 0.36 .494 Table 1 presents the correlation coefficients of the four dimensions of self-regulation. BMI > 25 -0.93 0.32 .006

Decision making -0.55 0.30 .071 Impulse control -0.22 0.22 .321 Table 1. BMI change between T0-T2 (N=46) Correlations between BMI and self-regulation at T0 (N=80). Change score of specific β SE p-value BMI at T0 dimension: Goal-orientation -0.63 0.33 .067 Self-regulation at T0 Self-direction -0.36 0.37 .338 Goal orientation -.32** Decision making -2.61 0.46 <.001 Self-direction -.43** Impulse control -0.23 0.27 .399 Decision making -.41**

Impulse control -.39** ** p<.01 Multivariate explaining factors for BMI change at T1 and T2 Table 3 depicts the significant predictors for the short- and long-term change in BMI from BMI changes at the different time points the multivariate linear regression analysis. Self-weighing frequency, intention to lose weight, Paired samples t-tests revealed a significant decline in weight and BMI at T1 and T2. Mean and an increase in goal-orientation remained significant in the final model for BMI change at 2 BMI (SD) decreased from 25.9 (3.6) at T0 to 25.2 (3.6) at T1 (Mean difference -0.64 (0.92) T1 (F(4,68) = 4.5, R = .21, p<.01). At T2, the only variable that explained the variance in BMI 2 kg/m2, CI [ -.43; -.85], p<.001). At T2, mean BMI was 25.3 (3.5) (mean difference -0.57 (1.2) change was the increase in decision making between T0 and T2 (F(1,44) = 32.9, R = .43, kg/m2, CI [-0.26; -0.88], p<.01). No significant BMI changes occurred between T1 and T2 p<.001). (mean difference 0.017 ± 0.98, p=.892). Mean weight (SD) decreased from 78.5 kg (14.9) at T0 to 76.4 kg (14.6) at T1 and 77.1 kg (14.2) at T2 (mean difference -2.0 (2.8) kg at T1, and -

1.8 (3.7) kg at T2).

Univariate relations between change in self-regulation and BMI change Table 2 shows the univariate relations between the changes in the four different self- regulation scales and the BMI change between baseline and T1 and baseline and T2. An

132 The role of self-regulation in the effect of self-tracking of physical activity and weight on BMI

Results increase in goal orientation was significantly related to a decrease in BMI at T1. An increase in decision-making was significantly related to a decrease in BMI at T2. A significant

Decision making -0.55 0.30 .071 Impulse control -0.22 0.22 .321 Table 1. BMI change between T0-T2 (N=46) Correlations between BMI and self-regulation at T0 (N=80). Change score of specific β SE p-value BMI at T0 dimension: 7 Goal-orientation -0.63 0.33 .067 Self-regulation at T0 Self-direction -0.36 0.37 .338 Goal orientation -.32** Decision making -2.61 0.46 <.001 Self-direction -.43** Impulse control -0.23 0.27 .399 Decision making -.41**

1.8 (3.7) kg at T2).

133 Chapter 7

Table 3. Discussion Significant multivariate explaining factors for BMI change at T1 and at T2.

β SE p-value BMI change at T1 (N=73) This study aimed to describe the relation between BMI and (change in) self-regulation after Intercept .37 .31 .233 four and 12 months of self-tracking physical activity and weight. After four months of self- Change goal orientation -.53 .22 .017 tracking, body weight and BMI significantly decreased. The reduced weight was maintained Weighing frequency up to 12 months, but no additional weight loss occurred between four to 12 months, Daily -1.02 .33 .003 Weekly -.76 .28 .008 indicating that most weight loss occurs within the first months after beginning with self- Less than weekly (ref) tracking physical activity and weight. We determined that different processes of self- Intention weight loss at T0 regulation, i.e., goal orientation and self-direction in a sub group of people with overweight, Want to lose weight -.48 .22 .034 were related to weight loss after four months whereas an increase in decision making was Want to stay the same (ref) BMI change at T2 (N=46) related to weight loss after 12 months. In addition, we found that six out of ten people indicated that they increased their physical activity, and four out of ten indicated that they Intercept -.57 .13 <.001 modified their eating pattern as a result of using the devices. These self-indicated changes Change decision making -2.61 .46 <.001 were reflected in objectively measured weight loss.

In our study, self-regulation of health behavior was the main variable of interest. We Relationship of self-indicated changes in behavior and BMI change distinguished between goal orientation, self-direction, decision making, and impulse control At T1, 61% of the study group indicated that they changed their physical activity pattern and as dimensions of self-regulation. These dimensions of self-regulation were all negatively 41% indicated that they modified their food intake as a result of using the devices. When related to BMI at baseline, thus, people with a higher self-regulation for health behavior combined, 38% of the study population had changed both physical activity and food intake. have a lower BMI from the start. This confirms that different self-regulation processes for When this group was compared to those who indicated that they only altered one of the two health behavior are related with BMI. Notably, the decrease of BMI was related to an behaviors or to those who indicated that they had not changed any of the behaviors, a increase in goal-orientation after four months and related to an increase in decision making significant difference was found in BMI change on the univariate ANOVA test (F(2,70) = 6.8, after 12 months of self-tracking. This may imply that these self-regulation processes play a p<.01). Table 4 shows the BMI changes of the different groups. At T2, 45% of the study different role at short term and long term. Goal-orientation and decision making reflect both population (63% of participants who completed the questionnaire) indicated that they self-regulatory goal striving processes.10 Goal-orientation comprises the planning and actual changed their physical activity behavior and 33% (46% of participants who completed the implementation of health goals. Decision making reflects the ability to make decisions and to questionnaire) changed their food intake. When combined, 23% indicated that they changed find multiple ways to achieve goals. This is important for dealing with setbacks in the process both behaviors at T2. No significant difference was found at T2 between those groups in BMI of doing so. Thus, our results suggest that an increase in planning and implementation of change (F(2,43) = 1.76, p =.185) (Table 4). health goals contribute to short term weight loss (four months). An increase in the ability to find multiple ways to achieve goals may be more important for a successful long term weight

loss (12 months). Alternatively, it may also be possible that different people have decreased Table 4. their weight at long term compared to short term and that this explains the two different BMI changes in people with different self-indicated changes in physical activity and food intake behavior at T1 (N=73) and T2 (N=46). dimensions explaining weight loss at short term or long term. Changed both Changed either Did not change p-value Our results are in line with and extend the results of other studies concerning weight physical physical activity or either physical loss and self-regulation.23,24 Klieman et al (2017) found that an increase in overall self- activity and food intake activity or food regulation for weight loss (without distinction in sub capabilities) mediated the effect of a food intake intake brief weight loss intervention on weight loss after three months. In line with our short term BMI change at T1 -1.10 ± 1.04 -0.45 ± 0.83 -0.25 ± 0.60 .002 results about goal orientation, they also found that the participants who logged their weight (N=30) (N=22) (N=21) and behavior more often and made more plans for behavior change showed a greater BMI change at T2 -0.94 ± 1.3 -0.50 ± 0.92 -0.21 ± 0.90 .184 weight loss.23 McKee et al found in their qualitative research that people who successfully (N=18) (N=15) (N=13) maintained weight loss differed in self-regulation capabilities compared to people who were

134 The role of self-regulation in the effect of self-tracking of physical activity and weight on BMI

Table 3. Discussion Significant multivariate explaining factors for BMI change at T1 and at T2.

In our study, self-regulation of health behavior was the main variable of interest. We Relationship of self-indicated changes in behavior and BMI change distinguished between goal orientation, self-direction, decision making, and impulse control At T1, 61% of the study group indicated that they changed their physical activity pattern and as dimensions of self-regulation. These dimensions of self-regulation were all negatively 41% indicated that they modified their food intake as a result of using the devices. When related to BMI at baseline, thus, people with a higher self-regulation for health behavior combined, 38% of the study population had changed both physical activity and food intake. have a lower BMI from the start. This confirms that different self-regulation processes for When this group was compared to those who indicated that they only altered one of the two health behavior are related with BMI. Notably, the decrease of BMI was related to an 7 behaviors or to those who indicated that they had not changed any of the behaviors, a increase in goal-orientation after four months and related to an increase in decision making significant difference was found in BMI change on the univariate ANOVA test (F(2,70) = 6.8, after 12 months of self-tracking. This may imply that these self-regulation processes play a p<.01). Table 4 shows the BMI changes of the different groups. At T2, 45% of the study different role at short term and long term. Goal-orientation and decision making reflect both population (63% of participants who completed the questionnaire) indicated that they self-regulatory goal striving processes.10 Goal-orientation comprises the planning and actual changed their physical activity behavior and 33% (46% of participants who completed the implementation of health goals. Decision making reflects the ability to make decisions and to questionnaire) changed their food intake. When combined, 23% indicated that they changed find multiple ways to achieve goals. This is important for dealing with setbacks in the process both behaviors at T2. No significant difference was found at T2 between those groups in BMI of doing so. Thus, our results suggest that an increase in planning and implementation of change (F(2,43) = 1.76, p =.185) (Table 4). health goals contribute to short term weight loss (four months). An increase in the ability to find multiple ways to achieve goals may be more important for a successful long term weight

135 Chapter 7

not successful. People who maintained weight loss were better able to set realistic goals, feedback on performance that state that feedback can enhance motivation.16,25 Another construct a plan or certain routine for their diet, and monitor their progress.24 These skills explanation may be that people who weigh themselves more frequently already had a also correspond to our goal orientation subscale and to the use of the self-tracking devices. greater motivation for weight loss at baseline. As these devices afford the opportunity for people to set goals, offer tips and tricks for Our study has a number of strengths and limitations. This is one of the first studies weight loss, enable people to monitor progress, and provide feedback, these device that elaborated on the role of self-regulation on BMI change when using self-tracking functions or BCTs probably helped a subgroup of our study population to increase their goal technology. We confirmed previous findings that regular self-weighing is associated with orientation and thereby explain weight loss. To our knowledge, no studies thus far have weight loss. In addition, we extended the literature by exploring four different dimensions of reported about specific decision-making capabilities in relation with self-tracking and (long- self-regulation and their impact on weight loss. We also validated self-indicated lifestyle term) weight loss. As we ascertained a substantial explained variance (46%) by increment of changes by comparing them with objective weight changes. A limitation of the study is that decision making on weight loss, this may indicate a need for future research on this topic. our longer-term results may be affected by the relatively low number of participants (N=46) Another dimension of self-regulation, i.e., change in self-direction, did not explain that could be included in the analyses. Another limitation could be the fact that our weight loss in our multivariate analysis. However, we found that an increase in self-direction participants were recruited from within a pre-existing cohort study which may have was univariate significantly related to weight loss for people with overweight whereas this introduced a certain selection bias through the selection of subjects with an above average relationship was not found in people with a healthy weight. Thus, an increment in one’s self- health awareness. Therefore, the generalizability of our results may be limited. A final direction capability, i.e., learning about own mistakes, leads to weight loss only for people limitation is the omission of assessing self-efficacy for losing weight which may also have with overweight. This may be an important finding since learning about one’s own behavior been an explaining factor for weight loss. and how to improve it is a crucial process for accomplishing successful behavior change.10,25 In conclusion, self-regulation capabilities and changes in these capabilities play an People with overweight may have had a higher level of motivation or a higher need to learn important role in weight and weight loss when using health self-quantification technology. from previous behavior which may well explain this differential effect on weight loss. Further Self-tracking of physical activity and weight results in a modest weight loss after four months research is needed on this specific dimension and how to further increase learning by using which is maintained after twelve months. Whether different dimensions of self-regulation self-tracking devices, preferably in a population of people with solely overweight. From our are differently related to weight loss at short term and long term need to be confirmed in results it was also remarkable that a change in impulse-control did not contribute to weight future studies. loss. This is in line with recent developments in self-control research. Milyavskaya et al (2015) showed that the reason that people with a higher level of self-control show more successful outcomes on a variety of measurements - including weight loss – is not because of SO WHAT? Implications for Health Promotion Practitioners and Researchers having a greater ability to resist temptations. Instead, these people experience fewer The results of this study can be translated into different health promotion and research practices. To obstacles and distractions for their goals because they have a higher autonomous motivation achieve weight loss, attempts should be made to stimulate people to weigh themselves weekly or for their goals (i.e., they have so-called ‘want to’ goals, instead of ‘have to’ goals). This leads daily. In addition, strategies should be provided to optimize autonomous motivation and self- to a better routine for accomplishing their goals without increasing effort.26 This may well regulation capabilities. For instance, different behavior change techniques can be deployed to explain why a change in impulse-control (our measurement for self-control in this study) was achieve an increase in goal orientation and decision-making capabilities, such as goal setting of not related to weight loss. The importance of autonomous motivation was also highlighted behavior, goal setting of outcomes, and action planning (e.g., providing a format whereby the user in the study of Schüz et al who determined that the relationship between intention to can construct a plan of how to accomplish a certain goal using different methods). To enhance the ‘self-direction’ capability of self-regulation, feedback from a device or health practitioner should perform a behavior and the planning of the behavior was moderated by the strength of the emphasize learning. The feedback should allow the individual to gain knowledge and obtain personal underlying health motive relative to other motives in life.27 meaning from the information they receive. Ideally, the feedback allows the individual to learn how Our finding that more frequent self-weighing is related to a greater weight loss is the desired behavior is related to positive outcomes such as weight loss, but also to personal similar to the findings of other studies about this topic.6–9 An explanation for the impact of perceptions such as feeling fit or learning that one can still be active on a rainy day. A direction for frequency of self-weighing may be that an individual who daily or weekly self-weighs follow-up research may be to explore effective ways to further enhance self-regulation when using receives feedback on a regular basis and is, therefore, able to detect relationships with one’s self-tracking technology and to assess the impact of different types of self-regulation stimuli on weight loss. Such research would enhance the understanding of the relationship of self-regulation recent behavior and weight. Also, an individual can readily observe lapses and react on them and weight loss. immediately. This explanation is in line with the Feedback Theory and BCTs about providing

136 The role of self-regulation in the effect of self-tracking of physical activity and weight on BMI

not successful. People who maintained weight loss were better able to set realistic goals, feedback on performance that state that feedback can enhance motivation.16,25 Another construct a plan or certain routine for their diet, and monitor their progress.24 These skills explanation may be that people who weigh themselves more frequently already had a also correspond to our goal orientation subscale and to the use of the self-tracking devices. greater motivation for weight loss at baseline. As these devices afford the opportunity for people to set goals, offer tips and tricks for Our study has a number of strengths and limitations. This is one of the first studies weight loss, enable people to monitor progress, and provide feedback, these device that elaborated on the role of self-regulation on BMI change when using self-tracking functions or BCTs probably helped a subgroup of our study population to increase their goal technology. We confirmed previous findings that regular self-weighing is associated with orientation and thereby explain weight loss. To our knowledge, no studies thus far have weight loss. In addition, we extended the literature by exploring four different dimensions of reported about specific decision-making capabilities in relation with self-tracking and (long- self-regulation and their impact on weight loss. We also validated self-indicated lifestyle term) weight loss. As we ascertained a substantial explained variance (46%) by increment of changes by comparing them with objective weight changes. A limitation of the study is that decision making on weight loss, this may indicate a need for future research on this topic. our longer-term results may be affected by the relatively low number of participants (N=46) Another dimension of self-regulation, i.e., change in self-direction, did not explain that could be included in the analyses. Another limitation could be the fact that our weight loss in our multivariate analysis. However, we found that an increase in self-direction participants were recruited from within a pre-existing cohort study which may have was univariate significantly related to weight loss for people with overweight whereas this introduced a certain selection bias through the selection of subjects with an above average relationship was not found in people with a healthy weight. Thus, an increment in one’s self- health awareness. Therefore, the generalizability of our results may be limited. A final direction capability, i.e., learning about own mistakes, leads to weight loss only for people limitation is the omission of assessing self-efficacy for losing weight which may also have with overweight. This may be an important finding since learning about one’s own behavior been an explaining factor for weight loss. and how to improve it is a crucial process for accomplishing successful behavior change.10,25 In conclusion, self-regulation capabilities and changes in these capabilities play an People with overweight may have had a higher level of motivation or a higher need to learn important role in weight and weight loss when using health self-quantification technology. from previous behavior which may well explain this differential effect on weight loss. Further Self-tracking of physical activity and weight results in a modest weight loss after four months research is needed on this specific dimension and how to further increase learning by using which is maintained after twelve months. Whether different dimensions of self-regulation self-tracking devices, preferably in a population of people with solely overweight. From our 7 are differently related to weight loss at short term and long term need to be confirmed in results it was also remarkable that a change in impulse-control did not contribute to weight future studies. loss. This is in line with recent developments in self-control research. Milyavskaya et al (2015) showed that the reason that people with a higher level of self-control show more successful outcomes on a variety of measurements - including weight loss – is not because of SO WHAT? Implications for Health Promotion Practitioners and Researchers having a greater ability to resist temptations. Instead, these people experience fewer The results of this study can be translated into different health promotion and research practices. To obstacles and distractions for their goals because they have a higher autonomous motivation achieve weight loss, attempts should be made to stimulate people to weigh themselves weekly or for their goals (i.e., they have so-called ‘want to’ goals, instead of ‘have to’ goals). This leads daily. In addition, strategies should be provided to optimize autonomous motivation and self- to a better routine for accomplishing their goals without increasing effort.26 This may well regulation capabilities. For instance, different behavior change techniques can be deployed to explain why a change in impulse-control (our measurement for self-control in this study) was achieve an increase in goal orientation and decision-making capabilities, such as goal setting of not related to weight loss. The importance of autonomous motivation was also highlighted behavior, goal setting of outcomes, and action planning (e.g., providing a format whereby the user in the study of Schüz et al who determined that the relationship between intention to can construct a plan of how to accomplish a certain goal using different methods). To enhance the ‘self-direction’ capability of self-regulation, feedback from a device or health practitioner should perform a behavior and the planning of the behavior was moderated by the strength of the emphasize learning. The feedback should allow the individual to gain knowledge and obtain personal underlying health motive relative to other motives in life.27 meaning from the information they receive. Ideally, the feedback allows the individual to learn how Our finding that more frequent self-weighing is related to a greater weight loss is the desired behavior is related to positive outcomes such as weight loss, but also to personal similar to the findings of other studies about this topic.6–9 An explanation for the impact of perceptions such as feeling fit or learning that one can still be active on a rainy day. A direction for frequency of self-weighing may be that an individual who daily or weekly self-weighs follow-up research may be to explore effective ways to further enhance self-regulation when using receives feedback on a regular basis and is, therefore, able to detect relationships with one’s self-tracking technology and to assess the impact of different types of self-regulation stimuli on weight loss. Such research would enhance the understanding of the relationship of self-regulation recent behavior and weight. Also, an individual can readily observe lapses and react on them and weight loss. immediately. This explanation is in line with the Feedback Theory and BCTs about providing

137 Chapter 7

References and automaticity on the effectiveness of a brief weight loss habit-based intervention: Secondary analysis of the 10 top tips randomised trial. Int J Behav Nutr Phys Act. 2017;14(1). doi:10.1186/s12966- 017-0578-8. 24. McKee H, Ntoumanis N, Smith B. Weight maintenance: Self-regulatory factors underpinning success 1. Lee IM, Shiroma EJ, Lobelo F, et al. Effect of physical inactivity on major non-communicable diseases and failure. Psychol Heal. 2013;28(10):1207-1223. doi:10.1080/08870446.2013.799162. worldwide: An analysis of burden of disease and life expectancy. Lancet. 2012;380(9838):219-229. 25. Kluger AN, DeNisi A. The Effects of Feedback Interventions on Performance. Psychol Bull. doi:10.1016/S0140-6736(12)61031-9. 1996;119(2):254-284. doi:10.1037/0033-2909.119.2.254. 2. Almalki M, Gray K, Martin-Sanchez F. Refining the Concepts of Self-quantification Needed for Health 26. Milyavskaya M, Inzlicht M, Hope N, Koestner R. Saying “no” to temptation: Want-to motivation Self-management: A Thematic Literature Review. Computer (Long Beach Calif). 2015;79:1-5. improves self-regulation by reducing temptation rather than by increasing self-control. J Pers Soc 3. Whitehead L, Seaton P. The effectiveness of self-management mobile phone and tablet apps in long- Psychol. 2015;109(4):677-693. doi:10.1037/pspp0000045. term condition management: a systematic review. J Med Internet Res. 2016;18(5). 27. Schüz B, Wurm S, Warner LM, Wolff JK, Schwarzer R. Health motives and health behaviour self- 4. de Vries H, Kooiman T, van Ittersum M, van Brussel M, de Groot M. Does an activity monitor based regulation in older adults. J Behav Med. 2014;37(3):491-500. doi:10.1007/s10865-013-9504-y. intervention increase daily physical activity of adults with overweight or obesity? A systematic review and meta-analysis. http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42015024086. 5. Qiu S, Cai X, Chen X, Yang B, Sun Z. Step counter use in type 2 diabetes: a meta-analysis of randomized controlled trials. BMC Med. 2014;12(1):36. doi:10.1186/1741-7015-12-36. 6. LaRose JG, Lanoye A, Tate DF, Wing RR. Frequency of self-weighing and weight loss outcomes within a brief lifestyle intervention targeting emerging adults. Obes Sci Pract. 2016. 7. Zheng Y, Klem M Lou, Sereika SM, Danford CA, Ewing LJ, Burke LE. Self-weighing in weight management: A systematic literature review. Obesity. 2015;23(2):256-265. 8. Rosenbaum DL, Espel HM, Butryn ML, Zhang F, Lowe MR. Daily self-weighing and weight gain prevention: a longitudinal study of college-aged women. J Behav Med. 2017:1-8. 9. Pacanowski CR, Levitsky DA. Frequent self-weighing and visual feedback for weight loss in overweight adults. J Obes. 2015;2015. 10. Mann T, de Ridder D, Fujita K. Self-Regulation of Health Behavior. Heal Psychol. 2013;32(5):487-498. doi:10.1037/a0028533. 11. Bandura A. Health Promotion by Social Cognitive Means. Heal Educ Behav. 2004;31(2):143-164. doi:10.1177/1090198104263660. 12. Bandura A. Health promotion from the perspective of social cognitive theory. Psychol Heal. 1998;13(4):623-649. 13. Hall PA, Fong GT. Temporal self-regulation theory: A model for individual health behavior. Health Psychol Rev. 2007;1(1):6-52. doi:10.1080/17437190701492437. 14. Kluger AN, DeNisi A. The effects of feedback interventions on performance: A historical review, a meta-analysis, and a preliminary feedback intervention theory. Psychol Bull. 1996;119(2):254-284. doi:10.1037/0033-2909.119.2.254. 15. Suls J, Wallston KA. Social Psychological Foundations of Health and Illness.; 2003. doi:10.1002/9780470753552. 16. Michie S, Richardson M, Johnston M, et al. The behavior change technique taxonomy (v1) of 93 hierarchically clustered techniques: building an international consensus for the reporting of behavior change interventions. Ann Behav Med. 2013;46(1):81-95. 17. Michie S, Ashford S, Sniehotta FF, Dombrowski SU, Bishop A, French DP. A refined taxonomy of behaviour change techniques to help people change their physical activity and healthy eating behaviours: The CALO-RE taxonomy. Psychol Health. 2011;26(11):1479-1498. doi:10.1080/08870446.2010.540664. 18. Lyons EJ, Lewis ZH, Mayrsohn BG, Rowland JL. Behavior change techniques implemented in electronic lifestyle activity monitors: A systematic content analysis. J Med Internet Res. 2014;16(8). doi:10.2196/jmir.3469. 19. Sullivan AN, Lachman ME. Behavior Change with Fitness Technology in Sedentary Adults: A Review of the Evidence for Increasing Physical Activity. Front Public Heal. 2017;4. doi:10.3389/fpubh.2016.00289. 20. Brown JM, Miller WR, Lawendowski LA. The self-regulation questionnaire. 1999. 21. Gavora P, Jakešová J, Kalenda J. The Czech validation of the Self-regulation Questionnaire. Procedia- Social Behav Sci. 2015;171:222-230. 22. Jakešová J, Gavora P, Kalenda J, Vávrová S. Czech Validation of the Self-regulation and Self-efficacy Questionnaires for Learning. Procedia - Soc Behav Sci. 2016;217:313-321. doi:10.1016/j.sbspro.2016.02.092. 23. Kliemann N, Vickerstaff V, Croker H, Johnson F, Nazareth I, Beeken RJ. The role of self-regulatory skills

138 The role of self-regulation in the effect of self-tracking of physical activity and weight on BMI

References and automaticity on the effectiveness of a brief weight loss habit-based intervention: Secondary analysis of the 10 top tips randomised trial. Int J Behav Nutr Phys Act. 2017;14(1). doi:10.1186/s12966- 017-0578-8. 24. McKee H, Ntoumanis N, Smith B. Weight maintenance: Self-regulatory factors underpinning success 1. Lee IM, Shiroma EJ, Lobelo F, et al. Effect of physical inactivity on major non-communicable diseases and failure. Psychol Heal. 2013;28(10):1207-1223. doi:10.1080/08870446.2013.799162. worldwide: An analysis of burden of disease and life expectancy. Lancet. 2012;380(9838):219-229. 25. Kluger AN, DeNisi A. The Effects of Feedback Interventions on Performance. Psychol Bull. doi:10.1016/S0140-6736(12)61031-9. 1996;119(2):254-284. doi:10.1037/0033-2909.119.2.254. 2. Almalki M, Gray K, Martin-Sanchez F. Refining the Concepts of Self-quantification Needed for Health 26. Milyavskaya M, Inzlicht M, Hope N, Koestner R. Saying “no” to temptation: Want-to motivation Self-management: A Thematic Literature Review. Computer (Long Beach Calif). 2015;79:1-5. improves self-regulation by reducing temptation rather than by increasing self-control. J Pers Soc 3. Whitehead L, Seaton P. The effectiveness of self-management mobile phone and tablet apps in long- Psychol. 2015;109(4):677-693. doi:10.1037/pspp0000045. term condition management: a systematic review. J Med Internet Res. 2016;18(5). 27. Schüz B, Wurm S, Warner LM, Wolff JK, Schwarzer R. Health motives and health behaviour self- 4. de Vries H, Kooiman T, van Ittersum M, van Brussel M, de Groot M. Does an activity monitor based regulation in older adults. J Behav Med. 2014;37(3):491-500. doi:10.1007/s10865-013-9504-y. intervention increase daily physical activity of adults with overweight or obesity? A systematic review and meta-analysis. http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42015024086. 5. Qiu S, Cai X, Chen X, Yang B, Sun Z. Step counter use in type 2 diabetes: a meta-analysis of randomized controlled trials. BMC Med. 2014;12(1):36. doi:10.1186/1741-7015-12-36. 6. LaRose JG, Lanoye A, Tate DF, Wing RR. Frequency of self-weighing and weight loss outcomes within a brief lifestyle intervention targeting emerging adults. Obes Sci Pract. 2016. 7. Zheng Y, Klem M Lou, Sereika SM, Danford CA, Ewing LJ, Burke LE. Self-weighing in weight management: A systematic literature review. Obesity. 2015;23(2):256-265. 8. Rosenbaum DL, Espel HM, Butryn ML, Zhang F, Lowe MR. Daily self-weighing and weight gain prevention: a longitudinal study of college-aged women. J Behav Med. 2017:1-8. 9. Pacanowski CR, Levitsky DA. Frequent self-weighing and visual feedback for weight loss in overweight adults. J Obes. 2015;2015. 10. Mann T, de Ridder D, Fujita K. Self-Regulation of Health Behavior. Heal Psychol. 2013;32(5):487-498. doi:10.1037/a0028533. 11. Bandura A. Health Promotion by Social Cognitive Means. Heal Educ Behav. 2004;31(2):143-164. doi:10.1177/1090198104263660. 12. Bandura A. Health promotion from the perspective of social cognitive theory. Psychol Heal. 7 1998;13(4):623-649. 13. Hall PA, Fong GT. Temporal self-regulation theory: A model for individual health behavior. Health Psychol Rev. 2007;1(1):6-52. doi:10.1080/17437190701492437. 14. Kluger AN, DeNisi A. The effects of feedback interventions on performance: A historical review, a meta-analysis, and a preliminary feedback intervention theory. Psychol Bull. 1996;119(2):254-284. doi:10.1037/0033-2909.119.2.254. 15. Suls J, Wallston KA. Social Psychological Foundations of Health and Illness.; 2003. doi:10.1002/9780470753552. 16. Michie S, Richardson M, Johnston M, et al. The behavior change technique taxonomy (v1) of 93 hierarchically clustered techniques: building an international consensus for the reporting of behavior change interventions. Ann Behav Med. 2013;46(1):81-95. 17. Michie S, Ashford S, Sniehotta FF, Dombrowski SU, Bishop A, French DP. A refined taxonomy of behaviour change techniques to help people change their physical activity and healthy eating behaviours: The CALO-RE taxonomy. Psychol Health. 2011;26(11):1479-1498. doi:10.1080/08870446.2010.540664. 18. Lyons EJ, Lewis ZH, Mayrsohn BG, Rowland JL. Behavior change techniques implemented in electronic lifestyle activity monitors: A systematic content analysis. J Med Internet Res. 2014;16(8). doi:10.2196/jmir.3469. 19. Sullivan AN, Lachman ME. Behavior Change with Fitness Technology in Sedentary Adults: A Review of the Evidence for Increasing Physical Activity. Front Public Heal. 2017;4. doi:10.3389/fpubh.2016.00289. 20. Brown JM, Miller WR, Lawendowski LA. The self-regulation questionnaire. 1999. 21. Gavora P, Jakešová J, Kalenda J. The Czech validation of the Self-regulation Questionnaire. Procedia- Social Behav Sci. 2015;171:222-230. 22. Jakešová J, Gavora P, Kalenda J, Vávrová S. Czech Validation of the Self-regulation and Self-efficacy Questionnaires for Learning. Procedia - Soc Behav Sci. 2016;217:313-321. doi:10.1016/j.sbspro.2016.02.092. 23. Kliemann N, Vickerstaff V, Croker H, Johnson F, Nazareth I, Beeken RJ. The role of self-regulatory skills

139

Chapter 8 | General Discussion

Chapter 8

Introduction for the Gamin Vivosmart and Fitbit Charge HR. At the highest speed, the three smartwatches demonstrated the best validity.

Increasing physical activity and weight management are important behaviors for the Meanwhile, additional studies have been published about the reliability and validity prevention of overweight and management of chronic diseases such as type 2 diabetes. A of consumer activity trackers. These studies are generally in accordance with our findings for possible way for facilitating these self-management behaviors is the use of eHealth the reliability and validity of measuring steps.1–4 However, the validity of other indicators of technology such as activity trackers and digital weight scales. This technology is capable of physical activity such as sedentary time, intensity of the physical activities performed, and the aggregating personal health data such as different indicators for physical activity, weight estimated amount of energy expenditure has been found to be (much) lower.2,5–9 Consumer data, and sleeping patterns over time. As this data can be uploaded and shared with others, activity trackers tend to underestimate energy expenditure,2,5,9 both under and overestimate it is beneficial in broad applications for the individual, health care practitioners, and time spent in low, moderate, and vigorous activity,2,6–8 and may both over- and research. However, before this technology can be implemented within health care, certain underestimate sedentary time.5–7 For example, Rosenberger et al determined that different conditions must be satisfied. For instance, the data must be of sufficient quality so that consumer activity trackers overestimated the amount of time spent in moderate-vigorous users, health practitioners, and researchers can rely on these devices. In addition, before activity between 51-91%6 whereas Gomersall et al ascertained both over- and self-tracking technology can have an impact on health behavior and health outcomes, a underestimations of 46 and -50%.8 Therefore, at the moment, most consumer activity certain level of adoption and engagement with the device is needed. Therefore, it is crucial trackers are primarily suitable for individual use but, in many cases, not appropriate for the to know which factors determine the successful use of different self-tracking devices. evaluation of research outcomes concerning time spent in moderate to vigorous physical Subsequently, knowledge about the actual effectiveness of this technology is also needed. activity and energy expenditure. When consumer activity trackers are being selected for Therefore, this dissertation focused on three domains: (1) the validity and reliability of evaluation of specific research outcomes, this must be done based on existing research.6,10 activity trackers, (2) the adoption of devices that quantify physical activity, sleep and weight, and (3) the effectiveness of this technology for people with overweight and/or have type 2 Adoption of self-tracking technology diabetes as well as a general population of healthy adults. In Chapter 4, we assessed the six-month adoption of two devices that quantify physical activity, sleep, and weight. We found that the activity tracking function was used more

frequently compared to sleep tracking over the six-month period, however, the use of both Main findings functions declined over time. The number of self-weighings also declined over time but stabilized from the third until the sixth month of use with over 80% of the study population

weighing themselves weekly (i.e., one to five self-weighings per week) or daily (over six self- Reliability and validity of activity trackers weighing’s per week). Chapter 2 and 3 focused on the reliability and validity of activity trackers. In total, 20 activity We found that different types of factors (e.g., personal, behavioral, and technical) are trackers, smartwatches, and apps were assessed for their reliability and validity for important for the adoption and sustained use of self-tracking devices. These factors were measuring the number of steps taken. In the first study, the Lumoback, Fitbit Flex, Jawbone related differently to the use of different self-tracking functions. Up, Nike+ Fuelband SE, Misfit Shine, Withings Pulse, Fitbit Zip, Omron HJ-203, Yamax Digiwalker SW-200 and Moves mobile application were assessed for walking at an average Personal factors (age, BMI, gender, and education) were not related to the use of an speed (4.8km/h) on a treadmill, and in free-living circumstances during one day. Although activity tracker. For the use of the sleep tracker, younger people, people with a higher differences between the trackers exist, most trackers were reliable and valid in both education, and individuals with a BMI between 25-30 used the sleep tracker more often. For laboratory and free-living circumstances except for the Nike+Fuelband and Moves app. The the use of the weight scale, men, younger people, and individuals with a BMI between 25-30 Fitbit Zip demonstrated the best validity. weighed themselves more often. With regard to behavioral factors, the most important In the second study, we assessed the Polar Loop, Garmin Vivosmart, Fitbit Charge HR, Apple findings were that, in general, having a specific motive for self-tracking (compared to a Watch Sport, Pebble Smartwatch, Samsung Gear S, Misfit Flash, Jawbone Up Move, Flyfit, general motive ‘documentary’, i.e., the wish to know more about own health) or having the and Moves app at three different speeds on a treadmill (slow, average, and fast). We intention to change a specific behavior (e.g., the wish to increase physical activity) help to concluded that the validity depended on walking speed. Most trackers were reliable and increase adoption of a device. In addition, of the four dimensions of self-regulation used in valid at an average walking speed, with the Garmin Vivosmart and Apple Watch Sport this thesis, a higher goal-orientation at baseline contributed to the numbers of activity and demonstrating the best validity. At a slower speed, validity declined for most trackers except sleep measurements during the study period. Also, technical factors are important when

142 General Discussion

Introduction for the Gamin Vivosmart and Fitbit Charge HR. At the highest speed, the three smartwatches demonstrated the best validity.

143 Chapter 8

considering the adoption of self-tracking technology. Technical failures, battery life, the ease associated with a clinically relevant lowering of HbA1c at the 12-month follow-up. of use, and perceived usefulness of a device have all been found to impact adoption.11,12 Responders decreased their HbA1c, on average, with -10.7 ± 9.2 mmol/mol whereas non- From our evaluations with the participants in both Chapters 4 and 6, participants indeed responders showed a change of 0.8 ± 7.7 mmol/mol. These results suggest that, if people identified technical factors, such as the installation procedure and limited battery life as with diabetes initially increase their physical activity level with at least 1000 steps/d, they barriers for usage. may have maintained these increased activity levels leading to a sustained impact on HbA1c at a one-year follow-up. These results are not displayed in Chapter 6 but do strengthen the Our findings are mainly in line with the review of Perski et al who summarized a conclusions of the study. variety of factors associated with the engagement of digital behavioral change interventions. They determined that both individual or population based factors influence engagement In Chapter 7, the role of self-regulation was investigated for the effect of self-tracking (e.g., motivation, expectations, self-efficacy, and demographic characteristics), the setting of physical activity and weight on BMI change in a general population. We found that BMI (e.g., the social and physical environment such as cultural factors or access to the internet), significantly decreased at short term (four months) and that this was maintained at long the content of the intervention (social support features, reminders) and delivery based term (12 months). Change in BMI was explained by the intention to decrease weight, self- factors (e.g., mode of delivery, professional support features, control features, novelty, weighing frequency, and increment of self-regulation capabilities: goal orientation at short complexity, tailoring of content, and interactivity).13 In addition, many studies point out that term and decision making at long term. In total, six out of ten participants indicated to have eHealth literacy is an important factor influencing the adoption of eHealth technology. increased their physical activity behavior as a result of using the devices, and four out of ten eHealth literacy has been defined as “the ability to seek, find, understand and appraise improved their eating pattern. health information from electronic sources and apply the knowledge gained to addressing or These findings on the effectiveness of self-tracking technology accord with other solving a health problem”.14 Men, older people, those having less education, and people studies evaluating these devices. Several studies have found positive effects of self-tracking who are unemployed may have lower eHealth literacy.15,16 People who are more of physical activity on physical activity behavior.17–19 Moreover, a recent study indicated experienced with using the internet and indicate a higher perceived health status show significant long-term effects of their pedometer based intervention within primary care at higher eHealth literacy.16 three and four years of follow-up.20 However, the effect of self-tracking of physical activity on health-related outcome measures is far less certain.21 Effectiveness of self-tracking technology For the effect of self-weighing on weight outcomes, our findings are in line with the We have examined the effectiveness of the use of self-tracking technology in three different literature. Several studies also demonstrated weight loss in different populations with self- studies. In Chapter 5, a systematic review and meta-analysis was conducted to determine weighing frequency being the most significant predictor for weight loss.22,23 the effectiveness of the use of activity monitors for incremental physical activity in people with overweight or obesity. We found moderate evidence that activity monitor based behavioral physical activity interventions increase physical activity in these adults and that Clinical implications and considerations when using eHealth adding an activity monitor to a behavioral physical activity intervention increases the effect on physical activity. In the studies included in this systematic review, however, primarily This thesis showed that, thus far, eHealth interventions based on self-tracking and self- simple pedometers were employed with limited abilities to graphically represent individual tracking devices alone have the potential to improve lifestyle behavior. Below, a number of physical activity patterns over time and limited possibilities for personalized feedback clinical implications and considerations when using self-tracking technology in healthcare towards individual goals. will be discussed. Thereafter, points of improvement and future directions are discussed. Therefore, in Chapter 6, a randomized clinical trial was conducted to ascertain the efficacy of a consumer level activity tracker combined with an online lifestyle program in Use and considerations when using eHealth technology in health care people with type 2 diabetes. In this study, it was found that this intervention was effective Although a self-tracking device can already be considered as an intervention in itself, for use for increasing physical activity. In participants who increased their steps per day with a within healthcare, a more comprehensive approach is needed to increase the effectiveness minimum of 1000 steps (defined as responders), a clinically relevant and significant decline on both behavioral as well as health-related outcome measures. Thus, within health care, in HbA1c was found. Social support was found to be a significant confounder for results on self-tracking devices can best be embedded within intervention programs that are more HbA1c with responders exhibiting a greater social support at baseline compared to non- extensive and theory based, including the use of behavioral counselling strategies. Although responders. Notably, the effect of being a responder within the intervention group was still

144 General Discussion

considering the adoption of self-tracking technology. Technical failures, battery life, the ease associated with a clinically relevant lowering of HbA1c at the 12-month follow-up. of use, and perceived usefulness of a device have all been found to impact adoption.11,12 Responders decreased their HbA1c, on average, with -10.7 ± 9.2 mmol/mol whereas non- From our evaluations with the participants in both Chapters 4 and 6, participants indeed responders showed a change of 0.8 ± 7.7 mmol/mol. These results suggest that, if people identified technical factors, such as the installation procedure and limited battery life as with diabetes initially increase their physical activity level with at least 1000 steps/d, they barriers for usage. may have maintained these increased activity levels leading to a sustained impact on HbA1c at a one-year follow-up. These results are not displayed in Chapter 6 but do strengthen the Our findings are mainly in line with the review of Perski et al who summarized a conclusions of the study. variety of factors associated with the engagement of digital behavioral change interventions. They determined that both individual or population based factors influence engagement In Chapter 7, the role of self-regulation was investigated for the effect of self-tracking (e.g., motivation, expectations, self-efficacy, and demographic characteristics), the setting of physical activity and weight on BMI change in a general population. We found that BMI (e.g., the social and physical environment such as cultural factors or access to the internet), significantly decreased at short term (four months) and that this was maintained at long the content of the intervention (social support features, reminders) and delivery based term (12 months). Change in BMI was explained by the intention to decrease weight, self- factors (e.g., mode of delivery, professional support features, control features, novelty, weighing frequency, and increment of self-regulation capabilities: goal orientation at short complexity, tailoring of content, and interactivity).13 In addition, many studies point out that term and decision making at long term. In total, six out of ten participants indicated to have eHealth literacy is an important factor influencing the adoption of eHealth technology. increased their physical activity behavior as a result of using the devices, and four out of ten eHealth literacy has been defined as “the ability to seek, find, understand and appraise improved their eating pattern. health information from electronic sources and apply the knowledge gained to addressing or These findings on the effectiveness of self-tracking technology accord with other solving a health problem”.14 Men, older people, those having less education, and people studies evaluating these devices. Several studies have found positive effects of self-tracking who are unemployed may have lower eHealth literacy.15,16 People who are more of physical activity on physical activity behavior.17–19 Moreover, a recent study indicated experienced with using the internet and indicate a higher perceived health status show significant long-term effects of their pedometer based intervention within primary care at higher eHealth literacy.16 three and four years of follow-up.20 However, the effect of self-tracking of physical activity on health-related outcome measures is far less certain.21 Effectiveness of self-tracking technology For the effect of self-weighing on weight outcomes, our findings are in line with the We have examined the effectiveness of the use of self-tracking technology in three different literature. Several studies also demonstrated weight loss in different populations with self- 8 studies. In Chapter 5, a systematic review and meta-analysis was conducted to determine weighing frequency being the most significant predictor for weight loss.22,23 the effectiveness of the use of activity monitors for incremental physical activity in people with overweight or obesity. We found moderate evidence that activity monitor based behavioral physical activity interventions increase physical activity in these adults and that Clinical implications and considerations when using eHealth adding an activity monitor to a behavioral physical activity intervention increases the effect on physical activity. In the studies included in this systematic review, however, primarily This thesis showed that, thus far, eHealth interventions based on self-tracking and self- simple pedometers were employed with limited abilities to graphically represent individual tracking devices alone have the potential to improve lifestyle behavior. Below, a number of physical activity patterns over time and limited possibilities for personalized feedback clinical implications and considerations when using self-tracking technology in healthcare towards individual goals. will be discussed. Thereafter, points of improvement and future directions are discussed. Therefore, in Chapter 6, a randomized clinical trial was conducted to ascertain the efficacy of a consumer level activity tracker combined with an online lifestyle program in Use and considerations when using eHealth technology in health care people with type 2 diabetes. In this study, it was found that this intervention was effective Although a self-tracking device can already be considered as an intervention in itself, for use for increasing physical activity. In participants who increased their steps per day with a within healthcare, a more comprehensive approach is needed to increase the effectiveness minimum of 1000 steps (defined as responders), a clinically relevant and significant decline on both behavioral as well as health-related outcome measures. Thus, within health care, in HbA1c was found. Social support was found to be a significant confounder for results on self-tracking devices can best be embedded within intervention programs that are more HbA1c with responders exhibiting a greater social support at baseline compared to non- extensive and theory based, including the use of behavioral counselling strategies. Although responders. Notably, the effect of being a responder within the intervention group was still

145 Chapter 8

we studied people with overweight and/or had type 2 diabetes in this thesis, self-tracking technology can be (and already is) applied by a broad range of target groups such as patients The role of the health care provider with low back pain, COPD, or heart disease. In addition, devices may be used in a variety of For a successful enrolment and engagement of patients in eHealth programs provided by settings in both primary and secondary care; e.g., within hospitals, general care practises, or healthcare institutions, the role of the health practitioner is crucial. To initiate using eHealth, paramedical care such as physical therapy practises. health practitioners will need to first invest time in order to become familiar with the new The timing of health enhancing interventions is obviously important. Since the technology. Also, remote monitoring of and responding to questions or patient generated number of people with overweight, obesity, and type 2 diabetes is still expected to increase data will expend time. Therefore, health practitioners (for example, nurses) should be in the next decades, health enhancing interventions should ideally occur when people are afforded the opportunity to fulfil this specific role. Thereby, the health practitioner should recently diagnosed with type 2 diabetes or even before that. This will ultimately prevent that have high motivation to employ eHealth technology in order to successfully engage diabetes related complications such as feet problems or obesity related comorbidity limit participants in eHealth programs. Also, for the guidance of patients (both face to face and physical activity at a later time. Thus, I suggest that additional emphasis should be placed on remotely), the health care provider needs specific CANMED competences such as being a health enhancing interventions within primary care to prevent or postpone health issues health advocate and communicator27 and have sufficient skills in motivational interviewing that are more severe and need treatment in secondary care. principles.28

An important consideration when using eHealth technology within health care is the To date, wearable activity trackers and associated apps or websites already contain a eHealth literacy of the user.15,16,24 In our eHealth intervention described in Chapter 6, we variety of behavioral change techniques (BCTs).29,30 BCTs that are currently the most noticed that a number of patients with type 2 diabetes did not want to participate because frequently applied are related to behavioral goals, monitoring and feedback, social support, they were not in possession of a computer or smartphone or due to lack of abilities to use a and rewards/prompting cues on past success.31–33 However, for the application in health computer. In addition, a number of participants in our intervention had difficulties with care, the question is whether the patient will optimally benefit from these device features or installing the Fitbit activity tracker and using the eHealth program. Thus, intervention BCTs. Support from the health care professional is most likely necessary for this purpose. designers and health practitioners should take the eHealth literacy of their target group into Therefore, I next recommend a few specific actions for the health care provider when using account and find solutions for optimizing support when needed. activity trackers in healthcare. Within the entire process, it is vital that individual needs for autonomy, competence/self-efficacy, and relatedness are respected.34 Hereby, health care Furthermore, although the use of wearable technologies was primarily studied in this providers should begin with exploring personal motivation (including intrinsic motivation, thesis, the rise of mobile health applications (i.e., mHealth) may also provide opportunities attitude, outcome expectations and self-efficacy towards physical activity), exploring for use in health care. These health apps can be downloaded on an individual’s smartphone, individual capability (e.g., knowledge with regard to physical activity or self-regulation and many of them involve the promotion of physical activity. They use accelerometers from capabilities) and opportunity of engaging in physical activity (e.g., one’s environmental the smartphone to track physical activity and/or sleep levels. Important advantages of health context and social support).30 In this session, outcome goals may already be formulated to apps are that they are highly accessible (i.e., many people currently own a smartphone), and enhance motivation and the use of the device may need to be explained. they are associated with significant lower costs compared to wearable devices.25 A disadvantage may be lower validity of health apps to measure physical activity parameters, In a next session, the health care provider may offer support in the formulation of as we found for one app in Chapter 2 and 3, and as was found for several apps in the feasible goals based on one’s baseline activity pattern, ensure a gradual building up, offer research of Konharn et al.26 Also, it may be more difficult to capture an individual’s complete support with plan-making, and discuss barriers for reaching personal goals. Also, it is physical activity pattern using an app because carrying a smartphone all day during different recommended that health care providers discuss and facilitate social support for their (exercise) activities may not always be feasible or desirable. In addition, Bondaronek et al patients, e.g., by connecting patients or by facilitating walking groups. Many devices already found in their recent review into the quality of health apps that, thus far, health apps have a offer the possibility for social support through the device app or platform, however, as number of short comings such as quality of the content and safety issues.25 Therefore, as discussed in Chapter 6, additional actions are needed for patients to benefit from these with wearable devices, the selection of an health app for use in health care interventions social support functions. Lastly, a health care provider should think about and agree with the should always be done carefully, preferable based on scientific evidence. In line with this, patient on how the personal health data is monitored, how and when feedback is provided, different initiatives have already begun to test both wearables and health apps and when evaluation sessions will be held. Figure 1 illustrates all of these recommendations systematically, for instance, within the National eHealth Living Lab (NeLL) in Leiden. for health care professionals when using consumer activity trackers.

146 General Discussion

An important consideration when using eHealth technology within health care is the To date, wearable activity trackers and associated apps or websites already contain a eHealth literacy of the user.15,16,24 In our eHealth intervention described in Chapter 6, we variety of behavioral change techniques (BCTs).29,30 BCTs that are currently the most noticed that a number of patients with type 2 diabetes did not want to participate because frequently applied are related to behavioral goals, monitoring and feedback, social support, they were not in possession of a computer or smartphone or due to lack of abilities to use a and rewards/prompting cues on past success.31–33 However, for the application in health computer. In addition, a number of participants in our intervention had difficulties with care, the question is whether the patient will optimally benefit from these device features or installing the Fitbit activity tracker and using the eHealth program. Thus, intervention BCTs. Support from the health care professional is most likely necessary for this purpose. designers and health practitioners should take the eHealth literacy of their target group into Therefore, I next recommend a few specific actions for the health care provider when using account and find solutions for optimizing support when needed. activity trackers in healthcare. Within the entire process, it is vital that individual needs for autonomy, competence/self-efficacy, and relatedness are respected.34 Hereby, health care Furthermore, although the use of wearable technologies was primarily studied in this providers should begin with exploring personal motivation (including intrinsic motivation, thesis, the rise of mobile health applications (i.e., mHealth) may also provide opportunities attitude, outcome expectations and self-efficacy towards physical activity), exploring 8 for use in health care. These health apps can be downloaded on an individual’s smartphone, individual capability (e.g., knowledge with regard to physical activity or self-regulation and many of them involve the promotion of physical activity. They use accelerometers from capabilities) and opportunity of engaging in physical activity (e.g., one’s environmental the smartphone to track physical activity and/or sleep levels. Important advantages of health context and social support).30 In this session, outcome goals may already be formulated to apps are that they are highly accessible (i.e., many people currently own a smartphone), and enhance motivation and the use of the device may need to be explained. they are associated with significant lower costs compared to wearable devices.25 A disadvantage may be lower validity of health apps to measure physical activity parameters, In a next session, the health care provider may offer support in the formulation of as we found for one app in Chapter 2 and 3, and as was found for several apps in the feasible goals based on one’s baseline activity pattern, ensure a gradual building up, offer research of Konharn et al.26 Also, it may be more difficult to capture an individual’s complete support with plan-making, and discuss barriers for reaching personal goals. Also, it is physical activity pattern using an app because carrying a smartphone all day during different recommended that health care providers discuss and facilitate social support for their (exercise) activities may not always be feasible or desirable. In addition, Bondaronek et al patients, e.g., by connecting patients or by facilitating walking groups. Many devices already found in their recent review into the quality of health apps that, thus far, health apps have a offer the possibility for social support through the device app or platform, however, as number of short comings such as quality of the content and safety issues.25 Therefore, as discussed in Chapter 6, additional actions are needed for patients to benefit from these with wearable devices, the selection of an health app for use in health care interventions social support functions. Lastly, a health care provider should think about and agree with the should always be done carefully, preferable based on scientific evidence. In line with this, patient on how the personal health data is monitored, how and when feedback is provided, different initiatives have already begun to test both wearables and health apps and when evaluation sessions will be held. Figure 1 illustrates all of these recommendations systematically, for instance, within the National eHealth Living Lab (NeLL) in Leiden. for health care professionals when using consumer activity trackers.

147 Chapter 8

Next to privacy and security related concerns, there may also be ethical concerns with regard to the use of eHealth and wearable technologies. Sharon (2017)36 discusses a number of those concerns with regard to autonomy, solidarity, and authenticity. One concern is that self-tracking may be more or less imposed, e.g., in working environments, creating the emergence of a surveillance culture which may diminish an individual’s autonomy. In addition, concerns have been expressed about a decrement of solidarity due to a shift from a collective towards an individual responsibility for health. With regard to authenticity, concerns have been raised that self-tracking may lead individuals to alienation of their own feelings and intuition, by simplifying complex phenomena, such as health, towards numbers and categories created by other people. On the other hand, advocates of self-tracking technologies point out that the act of self-tracking has the ability to improve autonomy, solidarity, and authenticity. For instance, self-tracking may help individuals to engage in personal experiments which enables them to find out ‘what works for me’. By Figure 1. Practical and counseling recommendations for health care professionals when using consumer activity trackers. sharing their data and experiences, they may help others thereby increasing solidarity. With regard to authenticity, Sharon points out that that the quantified self-community does not

intend to carelessly trust their data but intend to go beyond their data by linking them to Data security, privacy and ethics subjective experiences and feelings. In this, the data is more or less a natural phase towards Finally, the use of wearable devices to gain personalized health data from citizens, including a richer feeling about the self. For health care providers, this may imply that they discuss the patients, has also raised concerns with regard to data security, privacy, and ethics.35,36 act of self-tracking with their patients and encourage them to link their data to other Data security and privacy is especially a current major topic with the introduction of the experiences and feelings. General Data Protection Regulation (GDPR) law and since the recent revelations about privacy breaches from large companies such as Facebook. Questions like ‘what happens to our data’ are asked more and more often. Many researchers and privacy organisations Future directions of eHealth and related research already pointed out the need for privacy policy regulations including public and citizen engagement, clarity and transparency, and even a new regulatory framework in which the Within this thesis, many points for improvement of eHealth technology have emerged. In 35 user sells personal data back to the company. order to further increase the adoption of wearable devices, it is vital that technical abilities Privacy regulations differ between the European Union and the United States. With of consumer level self-tracking devices are being improved. Thereafter, developments in the EU-U.S. Privacy Shield, US companies processing data from EU citizens are obliged to behavioral design are also needed to improve effectiveness of eHealth. Based on our results provide a privacy policy in which they inform their users about what type of personal data and other eHealth related research, recommendations are next provided for future they process and why, the reasons why these data are being processed, whether they intend directions in eHealth and research. to share data to other companies, and reasons for this. In addition, a Privacy Shield company can process the data only for the initial goal for which it was collected, they are obligated to Improvement of technical abilities of self-tracking devices minimize the amount of data that is gathered, keep the data only for the time needed, are First, overall user experience may be enhanced by ensuring a user-friendly installation required to secure the data for misuse, and are obligated to provide users access to their procedure in an individual’s native language and by addressing battery limitations in order to own data and the ability to change, correct, or delete personal data.37 It is recommended avoid data loss and non-wearing time. that health care providers and researchers are aware of the privacy statement of the Second, the reliability and validity of activity trackers should be further improved. In wearable products they utilize for their patients, and, since many digital health care products the event of step counting, this is especially necessary for activities that take place at a are developed in the United States, whether they are registered with the EU-U.S Privacy slower walking pace. As older adults and people with overweight exhibit a different, slower Shield. walking pattern compared to people without overweight,38,39 it is important that their physical activity efforts can still be captured by consumer level activity trackers. In addition,

148 General Discussion

intend to carelessly trust their data but intend to go beyond their data by linking them to Data security, privacy and ethics subjective experiences and feelings. In this, the data is more or less a natural phase towards Finally, the use of wearable devices to gain personalized health data from citizens, including a richer feeling about the self. For health care providers, this may imply that they discuss the patients, has also raised concerns with regard to data security, privacy, and ethics.35,36 act of self-tracking with their patients and encourage them to link their data to other Data security and privacy is especially a current major topic with the introduction of the experiences and feelings. General Data Protection Regulation (GDPR) law and since the recent revelations about privacy breaches from large companies such as Facebook. Questions like ‘what happens to our data’ are asked more and more often. Many researchers and privacy organisations Future directions of eHealth and related research already pointed out the need for privacy policy regulations including public and citizen 8 engagement, clarity and transparency, and even a new regulatory framework in which the Within this thesis, many points for improvement of eHealth technology have emerged. In 35 user sells personal data back to the company. order to further increase the adoption of wearable devices, it is vital that technical abilities Privacy regulations differ between the European Union and the United States. With of consumer level self-tracking devices are being improved. Thereafter, developments in the EU-U.S. Privacy Shield, US companies processing data from EU citizens are obliged to behavioral design are also needed to improve effectiveness of eHealth. Based on our results provide a privacy policy in which they inform their users about what type of personal data and other eHealth related research, recommendations are next provided for future they process and why, the reasons why these data are being processed, whether they intend directions in eHealth and research. to share data to other companies, and reasons for this. In addition, a Privacy Shield company can process the data only for the initial goal for which it was collected, they are obligated to Improvement of technical abilities of self-tracking devices minimize the amount of data that is gathered, keep the data only for the time needed, are First, overall user experience may be enhanced by ensuring a user-friendly installation required to secure the data for misuse, and are obligated to provide users access to their procedure in an individual’s native language and by addressing battery limitations in order to own data and the ability to change, correct, or delete personal data.37 It is recommended avoid data loss and non-wearing time. that health care providers and researchers are aware of the privacy statement of the Second, the reliability and validity of activity trackers should be further improved. In wearable products they utilize for their patients, and, since many digital health care products the event of step counting, this is especially necessary for activities that take place at a are developed in the United States, whether they are registered with the EU-U.S Privacy slower walking pace. As older adults and people with overweight exhibit a different, slower Shield. walking pattern compared to people without overweight,38,39 it is important that their physical activity efforts can still be captured by consumer level activity trackers. In addition,

149 Chapter 8

validity of physical activity measures such as sedentary time, time spent in moderate to Improvement of effectiveness of eHealth technology vigorous physical activity (MVPA), and energy expenditure should also be enhanced. As mentioned above, wearable activity trackers already contain a number of behavioral Sedentary behavior is an independent risk factor for adverse health outcomes40,41 and, change techniques. However, there are also a number of BCTs that have been associated therefore, may be a unique target within health promoting interventions. In addition, the with increment of physical activity or self-efficacy for exercise that are currently rarely amount of time spent in MVPA is the primary physical activity measure used in physical applied within self-tracking technology. Table 1 provides an overview of these BCTs. When activity guidelines.41,42 Also, energy expenditure (EE) is an important physical activity these BCTs are further included within wearable technology (or associated programs), the measure, especially in weight loss interventions that aim to increase EE and decrease caloric potential of these interventions to further increase physical activity behavior -also in more intake. Recent research has shown that GPS provide the most valid method for estimating vulnerable populations such as sedentary individuals, older adults, or people with a low EE43 compared to heart rate combined with accelerometry measurements. Therefore, social economic status (SES)- may increase greatly.31–33 Therefore, these BCTs are discussed combining both accelerometry with heart rate and GPS may be a viable option for improving below using the categorisation of BCTs by Michie et al.30 validity of EE that is measured by consumer activity monitors. In addition, there is an ongoing need for research into the validity and reliability of newly developed consumer Table 1. activity trackers. This need is parallel to the nature of consumer technology since the market Overview of BCTs that are not yet present in most consumer activity trackers. of consumer wearable technology continues to expand. Category Behavioral Change Techniques Third, to increase experienced usefulness of the data and thereby adoption, a device Goals and planning -Goal setting of outcomes 44 -Barrier identification/problem solving should measure what the individual really wants to know. Therefore, more different types -Action planning of activities may be implemented in self-tracking technology such as cycling, swimming, or -Prompt review on outcome goals performing exercises at the gym. It would be most useful for users to know how much time Shaping knowledge -Provide instruction on how to perform the behavior they spent (or how many repetitions they made) when performing these specific activities Natural consequences -Provide information on consequences of behavior in general Repetition and substitution -Behavioral practice and at which intensity. Although there are already developments in the measurements of -Generalization of the target behavior these different activities, a single device that is able to capture all of these activities is not Antecedents -Restructuring the physical environment yet available. Reward and threat -Self-reward Self-belief -Self-talk Lastly, users will benefit from increased possibilities for data integration. Generally, Covert learning -Prompt use of imagery an individual does not just want to measure or change a single behavior or health outcome. Instead, people would like to gain insight into how several aspects of health, health Advances in goals and planning within activity tracker-based interventions may help behavior, and personal daily habits such as ‘what did I do today’ are related to each other. users to increase their self-regulation abilities such as increasing their goal orientation and Thus, ideally, different types of data should be integrated into single apps or graphs in order decision-making capabilities. Goal setting of outcomes, problem solving, and action planning to increase meaningful interpretations of accumulated data. have found to be rarely applied in consumer activity trackers. 31–33 This may be improved by With all of the above mentioned improvements, self-generated health data will gain providing formats into apps or internet platforms associated with the device, to help users certain ‘valances’, such as self-evidence (accuracy of the data with regard to technical to select both behavioral and outcome goals, as well as making a plan on how to achieve 34 abilities), truthfulness of the data (accuracy of the data with regard to user behavior, such as those goals. Offering sufficient meaningful choices is hereby crucial to support autonomy. wearing time), data transparency and discovery (i.e., being able to analyse relations between In addition, users may get the possibility to select individual barriers for physical activity different types of data and discover patterns in the data), and actionability.45 These are which are most common (e.g., lack of time, tiredness, lack of knowledge how to be active), important conditions for people to gain insight into their actual lifestyle habits and and subsequently be prompted to think about ways to overcome those barriers. The app possibility to engage in personal experimentations (e.g., finding out which individual lifestyle may also provide advices based on these barriers. Research is needed in this area to design changes are needed to lose weight), and consequently being able to act upon this personal these suggested behavioral strategies, for instance using focus groups. 45 data. This will both enhance overall user experience and motivation when using self- Although monitoring and feedback related BCTs are mostly present in consumer tracking technology. activity trackers, there is room for improvement in the delivery of feedback trough digital devices.46 The Feedback theory emphasizes that feedback should be aimed at task

150 General Discussion

validity of physical activity measures such as sedentary time, time spent in moderate to Improvement of effectiveness of eHealth technology vigorous physical activity (MVPA), and energy expenditure should also be enhanced. As mentioned above, wearable activity trackers already contain a number of behavioral Sedentary behavior is an independent risk factor for adverse health outcomes40,41 and, change techniques. However, there are also a number of BCTs that have been associated therefore, may be a unique target within health promoting interventions. In addition, the with increment of physical activity or self-efficacy for exercise that are currently rarely amount of time spent in MVPA is the primary physical activity measure used in physical applied within self-tracking technology. Table 1 provides an overview of these BCTs. When activity guidelines.41,42 Also, energy expenditure (EE) is an important physical activity these BCTs are further included within wearable technology (or associated programs), the measure, especially in weight loss interventions that aim to increase EE and decrease caloric potential of these interventions to further increase physical activity behavior -also in more intake. Recent research has shown that GPS provide the most valid method for estimating vulnerable populations such as sedentary individuals, older adults, or people with a low EE43 compared to heart rate combined with accelerometry measurements. Therefore, social economic status (SES)- may increase greatly.31–33 Therefore, these BCTs are discussed combining both accelerometry with heart rate and GPS may be a viable option for improving below using the categorisation of BCTs by Michie et al.30 validity of EE that is measured by consumer activity monitors. In addition, there is an ongoing need for research into the validity and reliability of newly developed consumer Table 1. activity trackers. This need is parallel to the nature of consumer technology since the market Overview of BCTs that are not yet present in most consumer activity trackers. of consumer wearable technology continues to expand. Category Behavioral Change Techniques Third, to increase experienced usefulness of the data and thereby adoption, a device Goals and planning -Goal setting of outcomes 44 -Barrier identification/problem solving should measure what the individual really wants to know. Therefore, more different types -Action planning of activities may be implemented in self-tracking technology such as cycling, swimming, or -Prompt review on outcome goals performing exercises at the gym. It would be most useful for users to know how much time Shaping knowledge -Provide instruction on how to perform the behavior they spent (or how many repetitions they made) when performing these specific activities Natural consequences -Provide information on consequences of behavior in general Repetition and substitution -Behavioral practice and at which intensity. Although there are already developments in the measurements of -Generalization of the target behavior these different activities, a single device that is able to capture all of these activities is not Antecedents -Restructuring the physical environment yet available. Reward and threat -Self-reward Self-belief -Self-talk Lastly, users will benefit from increased possibilities for data integration. Generally, Covert learning -Prompt use of imagery 8 an individual does not just want to measure or change a single behavior or health outcome. Instead, people would like to gain insight into how several aspects of health, health Advances in goals and planning within activity tracker-based interventions may help behavior, and personal daily habits such as ‘what did I do today’ are related to each other. users to increase their self-regulation abilities such as increasing their goal orientation and Thus, ideally, different types of data should be integrated into single apps or graphs in order decision-making capabilities. Goal setting of outcomes, problem solving, and action planning to increase meaningful interpretations of accumulated data. have found to be rarely applied in consumer activity trackers. 31–33 This may be improved by With all of the above mentioned improvements, self-generated health data will gain providing formats into apps or internet platforms associated with the device, to help users certain ‘valances’, such as self-evidence (accuracy of the data with regard to technical to select both behavioral and outcome goals, as well as making a plan on how to achieve 34 abilities), truthfulness of the data (accuracy of the data with regard to user behavior, such as those goals. Offering sufficient meaningful choices is hereby crucial to support autonomy. wearing time), data transparency and discovery (i.e., being able to analyse relations between In addition, users may get the possibility to select individual barriers for physical activity different types of data and discover patterns in the data), and actionability.45 These are which are most common (e.g., lack of time, tiredness, lack of knowledge how to be active), important conditions for people to gain insight into their actual lifestyle habits and and subsequently be prompted to think about ways to overcome those barriers. The app possibility to engage in personal experimentations (e.g., finding out which individual lifestyle may also provide advices based on these barriers. Research is needed in this area to design changes are needed to lose weight), and consequently being able to act upon this personal these suggested behavioral strategies, for instance using focus groups. 45 data. This will both enhance overall user experience and motivation when using self- Although monitoring and feedback related BCTs are mostly present in consumer tracking technology. activity trackers, there is room for improvement in the delivery of feedback trough digital devices.46 The Feedback theory emphasizes that feedback should be aimed at task

151 Chapter 8

motivation or task learning processes.47 Ideally, feedback content should be tailored as much trackers by including messages or notifications in which users are, for example, prompted to as possible to individual users (e.g., based on age, gender, activity level, health beliefs, reward themselves or tell themselves that a walk will be energizing. personal goals, self-efficacy expectations, barriers, or goal progression) and facilitate learning about an individual’s own behavior.48,49 Learning can comprise gaining knowledge about which specific routes in a person’s environment or actions are needed to reach a step Concluding remarks goal or how to cook a healthy meal. In concrete learning situations, positive outcome expectations about the undesired behavior (e.g., having lunch at the computer saves time), eHealth, including the use of self-tracking technology, offers potential for improving quality should be replaced with positive outcome expectations about the desired behavior (e.g., of health care and self-management of patients. In this thesis it was found that effective use taking a walking lunch break is feasible and makes me feel fit). This will enable people to of an activity tracker may significantly reduce the progression of diabetes, by lowering learn about the changes they make in their daily habits and how these changes are related HbA1c both at short term and after one year. For the design of future interventions, it is vital to goal progression and personal perceptions. In addition, the tone of the feedback should that co-creation occurs with all of the relevant stakeholders in order to successfully be empathic, positive, and always be aimed at increasing the self-efficacy of the user for implement eHealth innovations within healthcare.53 In addition, the selection of appropriate engagement in the target behavior.50,33 This may, for instance, be accomplished by providing intervention functions and behavior change techniques is crucial in order to optimise success stories of peers. In addition to personalized or goal-orientated feedback (based on intervention effects.30 All stakeholders (e.g., patients or end-users, health care providers, personal characteristics, personal data and individual goals), normative feedback (comparing small and medium enterprises, researchers, and policy makers / health care ensures) should an individual’s own data with others), iterative feedback (comparing own data with data work together to create a viable eHealth product. This includes a product that offers added from the past), and actionable feedback should also be considered.48,51 Actionable feedback value for the patient and health care provider, includes a positive business case for the small includes feedback that provides multiple cues for action, i.e., when, where, and how to and medium enterprises, and is cost-effective for the health care insurer.24,53 engage in a goal-directed behavior. This type of feedback was found in only 15% of the reviewed studies by Schembre et al,51 and, therefore, affords an opportunity for the improvement of feedback content in future programs. In addition, next to the content of the feedback, efforts may be made to enhance the presentation (i.e., attractiveness), timing, frequency, and duration of feedback.46,52 For instance, providing real-time feedback that accords with the context of the user will likely be more beneficial, e.g., a truck driver will benefit more from encouragement to increase activity levels during a break than receiving activity reminders while driving.

With regard to ‘shaping knowledge’ and ‘natural consequences’, the BCTs ‘providing instruction on how to perform the behavior’ and ‘providing information about health consequences of performing the behavior’ were found to be present in some, but not all activity trackers reviewed in Lyons et al. Both BCTs have been determined to influence physical activity behavior.32 Therefore, additional efforts should be made to include information on benefits of physical activity and ways to increase it. These actions correspond with providing actionable feedback as described above.

Also, ‘repetition and substitution’ and ‘antecendents’ related BCTs (e.g., behavioral practice, generalization of the target behavior, and restructuring the physical environment) have been suggested as important BCTs to integrate within consumer technology.31–33 This may be done by prompting users to perform certain exercises with detailed instructions through an app or including maps of suitable walking routes within the own area of the user. In addition, ‘reward and threat’, ‘self-belief’ and ‘covert learning’ related BCTs (i.e., self- reward, self-talk, and use of imaginary) may be integrated within associated apps of activity

152 General Discussion

motivation or task learning processes.47 Ideally, feedback content should be tailored as much trackers by including messages or notifications in which users are, for example, prompted to as possible to individual users (e.g., based on age, gender, activity level, health beliefs, reward themselves or tell themselves that a walk will be energizing. personal goals, self-efficacy expectations, barriers, or goal progression) and facilitate learning about an individual’s own behavior.48,49 Learning can comprise gaining knowledge about which specific routes in a person’s environment or actions are needed to reach a step Concluding remarks goal or how to cook a healthy meal. In concrete learning situations, positive outcome expectations about the undesired behavior (e.g., having lunch at the computer saves time), eHealth, including the use of self-tracking technology, offers potential for improving quality should be replaced with positive outcome expectations about the desired behavior (e.g., of health care and self-management of patients. In this thesis it was found that effective use taking a walking lunch break is feasible and makes me feel fit). This will enable people to of an activity tracker may significantly reduce the progression of diabetes, by lowering learn about the changes they make in their daily habits and how these changes are related HbA1c both at short term and after one year. For the design of future interventions, it is vital to goal progression and personal perceptions. In addition, the tone of the feedback should that co-creation occurs with all of the relevant stakeholders in order to successfully be empathic, positive, and always be aimed at increasing the self-efficacy of the user for implement eHealth innovations within healthcare.53 In addition, the selection of appropriate engagement in the target behavior.50,33 This may, for instance, be accomplished by providing intervention functions and behavior change techniques is crucial in order to optimise success stories of peers. In addition to personalized or goal-orientated feedback (based on intervention effects.30 All stakeholders (e.g., patients or end-users, health care providers, personal characteristics, personal data and individual goals), normative feedback (comparing small and medium enterprises, researchers, and policy makers / health care ensures) should an individual’s own data with others), iterative feedback (comparing own data with data work together to create a viable eHealth product. This includes a product that offers added from the past), and actionable feedback should also be considered.48,51 Actionable feedback value for the patient and health care provider, includes a positive business case for the small includes feedback that provides multiple cues for action, i.e., when, where, and how to and medium enterprises, and is cost-effective for the health care insurer.24,53 engage in a goal-directed behavior. This type of feedback was found in only 15% of the reviewed studies by Schembre et al,51 and, therefore, affords an opportunity for the improvement of feedback content in future programs. In addition, next to the content of the feedback, efforts may be made to enhance the presentation (i.e., attractiveness), timing, frequency, and duration of feedback.46,52 For instance, providing real-time feedback that accords with the context of the user will likely be more beneficial, e.g., a truck driver will 8 benefit more from encouragement to increase activity levels during a break than receiving activity reminders while driving.

153 Chapter 8

References 22. Rosenbaum DL, Espel HM, Butryn ML, Zhang F, Lowe MR. Daily self-weighing and weight gain prevention: a longitudinal study of college-aged women. J Behav Med. 2017:1-8. 23. Zheng Y, Klem M Lou, Sereika SM, Danford CA, Ewing LJ, Burke LE. Self-weighing in weight management: A systematic literature review. Obesity. 2015;23(2):256-265. 24. van Gemert-Pijnen JEWC, Peters O, Ossebaard HC. Improving eHealth. Eleven International Publishing; 1. An HS, Jones GC, Kang SK, Welk GJ, Lee JM. How valid are wearable physical activity trackers for 2013. http://lib.myilibrary.com?ID=673579. measuring steps? Eur J Sport Sci. 2017;17(3):360-368. doi:10.1080/17461391.2016.1255261. 25. Bondaronek P, Alkhaldi G, Slee A, Hamilton FL, Murray E. Quality of Publicly Available Physical Activity 2. Evenson KR, Goto MM, Furberg RD. Systematic review of the validity and reliability of consumer- Apps: Review and Content Analysis. JMIR mHealth uHealth. 2018;6(3):e53. doi:10.2196/mhealth.9069. wearable activity trackers. Int J Behav Nutr Phys Act. 2015;12(1). doi:10.1186/s12966-015-0314-1. 26. Konharn K, Eungpinichpong W, Promdee K, et al. Validity and Reliability of Smartphone Applications 3. Case MA, Burwick HA, Volpp KG, Patel MS. Accuracy of smartphone applications and wearable devices for the Assessment of Walking and Running in Normal-weight and Overweight/Obese Young Adults. J for tracking physical activity data. JAMA. 2015;313(6):625-626. Phys Act Heal. 2016;13(12):1333-1340. 4. El-Amrawy F, Nounou MI. Are currently available wearable devices for activity tracking and heart rate 27. Frank JR, Danoff D. The CanMEDS initiative: implementing an outcomes-based framework of physician monitoring accurate, precise, and medically beneficial? Healthc Inform Res. 2015;21(4):315-320. competencies. Med Teach. 2007;29(7):642-647. doi:10.4258/hir.2015.21.4.315. 28. Kahan S, Wilson DK, Sweeney AM. The Role of Behavioral Medicine in the Treatment of Obesity in 5. Brooke SM, An HS, Kang SK, Noble JM, Berg KE, Lee JM. Concurrent Validity of Wearable Activity Primary Care. Med Clin North Am. 2018;102(1):125-133. doi:10.1016/j.mcna.2017.09.002. Trackers under Free-Living Conditions. J Strength Cond Res. 2017;31(4):1097-1106. 29. Michie S, Ashford S, Sniehotta FF, Dombrowski SU, Bishop A, French DP. A refined taxonomy of doi:10.1519/JSC.0000000000001571. behaviour change techniques to help people change their physical activity and healthy eating 6. Rosenberger ME, Buman MP, Haskell WL, McConnell M V., Carstensen LL. Twenty-four Hours of Sleep, behaviours: The CALO-RE taxonomy. Psychol Health. 2011;26(11):1479-1498. Sedentary Behavior, and Physical Activity with Nine Wearable Devices. Med Sci Sports Exerc. doi:10.1080/08870446.2010.540664. 2016;48(3):457-465. doi:10.1249/MSS.0000000000000778. 30. Michie S, Atkins L, West R. The Behaviour Change Wheel: A Guide to Designing Interventions.; 2014. 7. Reid RER, Insogna JA, Carver TE, et al. Validity and reliability of Fitbit activity monitors compared to 31. Mercer K, Li M, Giangregorio L, Burns C, Grindrod K. Behavior Change Techniques Present in Wearable ActiGraph GT3X+ with female adults in a free-living environment. J Sci Med Sport. 2017. Activity Trackers: A Critical Analysis. JMIR mHealth uHealth. 2016;4(2):e40. doi:10.2196/mhealth.4461. doi:10.1016/j.jsams.2016.10.015. 32. Lyons EJ, Lewis ZH, Mayrsohn BG, Rowland JL. Behavior change techniques implemented in electronic 8. Gomersall SR, Ng N, Burton NW, Pavey TG, Gilson ND, Brown WJ. Estimating physical activity and lifestyle activity monitors: A systematic content analysis. J Med Internet Res. 2014;16(8). sedentary behavior in a free-living context: A pragmatic comparison of consumer-based activity doi:10.2196/jmir.3469. trackers and actigraph accelerometry. J Med Internet Res. 2016. doi:10.2196/jmir.5531. 33. Sullivan AN, Lachman ME. Behavior Change with Fitness Technology in Sedentary Adults: A Review of 9. BUNN JA, Navalta JW, Fountaine CJ, REECE JD. Current State of Commercial Wearable Technology in the Evidence for Increasing Physical Activity. Front Public Heal. 2017;4(289):1. Physical Activity Monitoring 2015–2017. Int J Exerc Sci. 2018;11(7):503. doi:10.3389/fpubh.2016.00289. 10. Cadmus-Bertram L. Using Fitness Trackers in Clinical Research: What Nurse Practitioners Need to 34. Ryan R, Deci E. Self-determination theory and the facilitation of intrinsic motivation, social Know. J Nurse Pract. 2017;13(1):34-40. doi:10.1016/j.nurpra.2016.10.012. development, and well-being. Am Psychol. 2000;55(1):68-78. doi:10.1037/0003-066X.55.1.68. 11. Kim J. Analysis of Health Consumers’ Behavior Using Self-Tracker for Activity, Sleep, and Diet. Telemed 35. Kostkova P, Brewer H, de Lusignan S, et al. Who Owns the Data? Open Data for Healthcare. Front J E Health. 2014;20(6):552-558. doi:10.1089/tmj.2013.0282. Public Heal. 2016;4. doi:10.3389/fpubh.2016.00007. 12. Kim KJ, Shin D-H. An acceptance model for smart watches: implications for the adoption of future 36. Sharon T. Self-Tracking for Health and the Quantified Self: Re-Articulating Autonomy, Solidarity, and wearable technology. Internet Res. 2015;25(4):527-541. Authenticity in an Age of Personalized Healthcare. Philos Technol. 2017;30(1):93-121. 13. Perski O, Blandford A, West R, Michie S. Conceptualising engagement with digital behaviour change doi:10.1007/s13347-016-0215-5. interventions: a systematic review using principles from critical interpretive synthesis. Transl Behav 37. The European Commission. EU-U.S. Privacy Shield Adequacy Decision. EuropaEu. 2016;(July):2016. Med. 2016:1-14. doi:10.2838/199012. 14. Norman CD, Skinner HA. eHealth literacy: Essential skills for consumer health in a networked world. J 38. Meng H, O’Connor DP, Lee BC, Layne CS, Gorniak SL. Alterations in over-ground walking patterns in Med Internet Res. 2006;8(2). doi:10.2196/jmir.8.2.e9. obese and overweight adults. Gait Posture. 2017;53:145-150. doi:10.1016/j.gaitpost.2017.01.019. 15. Tennant B, Stellefson M, Dodd V, et al. eHealth literacy and Web 2.0 health information seeking 39. Spyropoulos P, Pisciotta JC, Pavlou KN, Cairns MA, Simon SR. Biomechanical gait analysis in obese behaviors among baby boomers and older adults. J Med Internet Res. 2015;17(3). men. Arch Phys Med Rehabil. 1991;72(13):1065-1070. doi:10.1007/s10787-012-0152-6. doi:10.2196/jmir.3992. 40. Song J, Lindquist LA, Chang RW, et al. Sedentary behavior as a risk factor for physical frailty 16. Park H, Cormier E, Glenna G. Health consumers eHealth literacy to decrease disparities in accessing independent of moderate activity: Results from the osteoarthritis initiative. Am J Public Health. eHealth information. In: Studies in Health Technology and Informatics. Vol 225. ; 2016:895-896. 2015;105(7):1439-1445. doi:10.2105/AJPH.2014.302540. doi:10.3233/978-1-61499-658-3-895. 41. Gezondheidsraad. Beweegrichtlijnen 2017. gezondheidsraad.nl. 17. Qiu S, Cai X, Chen X, Yang B, Sun Z. Step counter use in type 2 diabetes: a meta-analysis of randomized https://www.gezondheidsraad.nl/sites/default/files/grpublication/beweegrichtlijnen_2017.pdf. controlled trials. BMC Med. 2014;12(1):36. doi:10.1186/1741-7015-12-36. Published 2017. 18. Kang M, Marshall SJ, Barreira T V., Lee JO. Effect of pedometer-based physical activity interventions: A 42. Haskell WL, Lee IM, Pate RR, et al. Physical activity and public health: Updated recommendation for meta-analysis. Res Q Exerc Sport. 2009;80(3):648-655. doi:10.1080/02701367.2009.10599604. adults from the American College of Sports Medicine and the American Heart Association. Med Sci 19. Bravata DM, Smith-Spangler C, Sundaram V, et al. Using pedometers to increase physical activity and Sports Exerc. 2007;39(8):1423-1434. doi:10.1249/mss.0b013e3180616b27. improve health: a systematic review. JAMA 2007 Nov 21;298(19)2296-304. 43. de Müllenheim PY, Chaudru S, Emily M, et al. Using GPS, accelerometry and heart rate to predict 20. Harris T, Kerry SM, Limb ES, et al. Physical activity levels in adults and older adults 3–4 years after outdoor graded walking energy expenditure. J Sci Med Sport. 2018;21(2):166-172. pedometer-based walking interventions: Long-term follow-up of participants from two randomised doi:10.1016/j.jsams.2017.10.004. controlled trials in UK primary care. PLoS Med. 2018;15(3):e1002526. 44. Fritz T, Huang E, Murphy G, Zimmermann T. Persuasive technology in the real world. In: CHI ’14. ACM; 21. Noah B, Keller MS, Mosadeghi S, et al. Impact of remote patient monitoring on clinical outcomes: an :487-496. doi:10.1145/2556288.2557383. updated meta-analysis of randomized controlled trials. npj Digit Med. 2017;1(1):2. 45. Almalki M, Gray K, Martin-Sanchez F. Refining the Concepts of Self-quantification Needed for Health doi:10.1038/s41746-017-0002-4. Self-management: A Thematic Literature Review. Computer (Long Beach Calif). 2015;79:1-5.

154 General Discussion

References 22. Rosenbaum DL, Espel HM, Butryn ML, Zhang F, Lowe MR. Daily self-weighing and weight gain prevention: a longitudinal study of college-aged women. J Behav Med. 2017:1-8. 23. Zheng Y, Klem M Lou, Sereika SM, Danford CA, Ewing LJ, Burke LE. Self-weighing in weight management: A systematic literature review. Obesity. 2015;23(2):256-265. 24. van Gemert-Pijnen JEWC, Peters O, Ossebaard HC. Improving eHealth. Eleven International Publishing; 1. An HS, Jones GC, Kang SK, Welk GJ, Lee JM. How valid are wearable physical activity trackers for 2013. http://lib.myilibrary.com?ID=673579. measuring steps? Eur J Sport Sci. 2017;17(3):360-368. doi:10.1080/17461391.2016.1255261. 25. Bondaronek P, Alkhaldi G, Slee A, Hamilton FL, Murray E. Quality of Publicly Available Physical Activity 2. Evenson KR, Goto MM, Furberg RD. Systematic review of the validity and reliability of consumer- Apps: Review and Content Analysis. JMIR mHealth uHealth. 2018;6(3):e53. doi:10.2196/mhealth.9069. wearable activity trackers. Int J Behav Nutr Phys Act. 2015;12(1). doi:10.1186/s12966-015-0314-1. 26. Konharn K, Eungpinichpong W, Promdee K, et al. Validity and Reliability of Smartphone Applications 3. Case MA, Burwick HA, Volpp KG, Patel MS. Accuracy of smartphone applications and wearable devices for the Assessment of Walking and Running in Normal-weight and Overweight/Obese Young Adults. J for tracking physical activity data. JAMA. 2015;313(6):625-626. Phys Act Heal. 2016;13(12):1333-1340. 4. El-Amrawy F, Nounou MI. Are currently available wearable devices for activity tracking and heart rate 27. Frank JR, Danoff D. The CanMEDS initiative: implementing an outcomes-based framework of physician monitoring accurate, precise, and medically beneficial? Healthc Inform Res. 2015;21(4):315-320. competencies. Med Teach. 2007;29(7):642-647. doi:10.4258/hir.2015.21.4.315. 28. Kahan S, Wilson DK, Sweeney AM. The Role of Behavioral Medicine in the Treatment of Obesity in 5. Brooke SM, An HS, Kang SK, Noble JM, Berg KE, Lee JM. Concurrent Validity of Wearable Activity Primary Care. Med Clin North Am. 2018;102(1):125-133. doi:10.1016/j.mcna.2017.09.002. Trackers under Free-Living Conditions. J Strength Cond Res. 2017;31(4):1097-1106. 29. Michie S, Ashford S, Sniehotta FF, Dombrowski SU, Bishop A, French DP. A refined taxonomy of doi:10.1519/JSC.0000000000001571. behaviour change techniques to help people change their physical activity and healthy eating 6. Rosenberger ME, Buman MP, Haskell WL, McConnell M V., Carstensen LL. Twenty-four Hours of Sleep, behaviours: The CALO-RE taxonomy. Psychol Health. 2011;26(11):1479-1498. Sedentary Behavior, and Physical Activity with Nine Wearable Devices. Med Sci Sports Exerc. doi:10.1080/08870446.2010.540664. 2016;48(3):457-465. doi:10.1249/MSS.0000000000000778. 30. Michie S, Atkins L, West R. The Behaviour Change Wheel: A Guide to Designing Interventions.; 2014. 7. Reid RER, Insogna JA, Carver TE, et al. Validity and reliability of Fitbit activity monitors compared to 31. Mercer K, Li M, Giangregorio L, Burns C, Grindrod K. Behavior Change Techniques Present in Wearable ActiGraph GT3X+ with female adults in a free-living environment. J Sci Med Sport. 2017. Activity Trackers: A Critical Analysis. JMIR mHealth uHealth. 2016;4(2):e40. doi:10.2196/mhealth.4461. doi:10.1016/j.jsams.2016.10.015. 32. Lyons EJ, Lewis ZH, Mayrsohn BG, Rowland JL. Behavior change techniques implemented in electronic 8. Gomersall SR, Ng N, Burton NW, Pavey TG, Gilson ND, Brown WJ. Estimating physical activity and lifestyle activity monitors: A systematic content analysis. J Med Internet Res. 2014;16(8). sedentary behavior in a free-living context: A pragmatic comparison of consumer-based activity doi:10.2196/jmir.3469. trackers and actigraph accelerometry. J Med Internet Res. 2016. doi:10.2196/jmir.5531. 33. Sullivan AN, Lachman ME. Behavior Change with Fitness Technology in Sedentary Adults: A Review of 9. BUNN JA, Navalta JW, Fountaine CJ, REECE JD. Current State of Commercial Wearable Technology in the Evidence for Increasing Physical Activity. Front Public Heal. 2017;4(289):1. Physical Activity Monitoring 2015–2017. Int J Exerc Sci. 2018;11(7):503. doi:10.3389/fpubh.2016.00289. 10. Cadmus-Bertram L. Using Fitness Trackers in Clinical Research: What Nurse Practitioners Need to 34. Ryan R, Deci E. Self-determination theory and the facilitation of intrinsic motivation, social Know. J Nurse Pract. 2017;13(1):34-40. doi:10.1016/j.nurpra.2016.10.012. development, and well-being. Am Psychol. 2000;55(1):68-78. doi:10.1037/0003-066X.55.1.68. 11. Kim J. Analysis of Health Consumers’ Behavior Using Self-Tracker for Activity, Sleep, and Diet. Telemed 35. Kostkova P, Brewer H, de Lusignan S, et al. Who Owns the Data? Open Data for Healthcare. Front J E Health. 2014;20(6):552-558. doi:10.1089/tmj.2013.0282. Public Heal. 2016;4. doi:10.3389/fpubh.2016.00007. 12. Kim KJ, Shin D-H. An acceptance model for smart watches: implications for the adoption of future 36. Sharon T. Self-Tracking for Health and the Quantified Self: Re-Articulating Autonomy, Solidarity, and 8 wearable technology. Internet Res. 2015;25(4):527-541. Authenticity in an Age of Personalized Healthcare. Philos Technol. 2017;30(1):93-121. 13. Perski O, Blandford A, West R, Michie S. Conceptualising engagement with digital behaviour change doi:10.1007/s13347-016-0215-5. interventions: a systematic review using principles from critical interpretive synthesis. Transl Behav 37. The European Commission. EU-U.S. Privacy Shield Adequacy Decision. EuropaEu. 2016;(July):2016. Med. 2016:1-14. doi:10.2838/199012. 14. Norman CD, Skinner HA. eHealth literacy: Essential skills for consumer health in a networked world. J 38. Meng H, O’Connor DP, Lee BC, Layne CS, Gorniak SL. Alterations in over-ground walking patterns in Med Internet Res. 2006;8(2). doi:10.2196/jmir.8.2.e9. obese and overweight adults. Gait Posture. 2017;53:145-150. doi:10.1016/j.gaitpost.2017.01.019. 15. Tennant B, Stellefson M, Dodd V, et al. eHealth literacy and Web 2.0 health information seeking 39. Spyropoulos P, Pisciotta JC, Pavlou KN, Cairns MA, Simon SR. Biomechanical gait analysis in obese behaviors among baby boomers and older adults. J Med Internet Res. 2015;17(3). men. Arch Phys Med Rehabil. 1991;72(13):1065-1070. doi:10.1007/s10787-012-0152-6. doi:10.2196/jmir.3992. 40. Song J, Lindquist LA, Chang RW, et al. Sedentary behavior as a risk factor for physical frailty 16. Park H, Cormier E, Glenna G. Health consumers eHealth literacy to decrease disparities in accessing independent of moderate activity: Results from the osteoarthritis initiative. Am J Public Health. eHealth information. In: Studies in Health Technology and Informatics. Vol 225. ; 2016:895-896. 2015;105(7):1439-1445. doi:10.2105/AJPH.2014.302540. doi:10.3233/978-1-61499-658-3-895. 41. Gezondheidsraad. Beweegrichtlijnen 2017. gezondheidsraad.nl. 17. Qiu S, Cai X, Chen X, Yang B, Sun Z. Step counter use in type 2 diabetes: a meta-analysis of randomized https://www.gezondheidsraad.nl/sites/default/files/grpublication/beweegrichtlijnen_2017.pdf. controlled trials. BMC Med. 2014;12(1):36. doi:10.1186/1741-7015-12-36. Published 2017. 18. Kang M, Marshall SJ, Barreira T V., Lee JO. Effect of pedometer-based physical activity interventions: A 42. Haskell WL, Lee IM, Pate RR, et al. Physical activity and public health: Updated recommendation for meta-analysis. Res Q Exerc Sport. 2009;80(3):648-655. doi:10.1080/02701367.2009.10599604. adults from the American College of Sports Medicine and the American Heart Association. Med Sci 19. Bravata DM, Smith-Spangler C, Sundaram V, et al. Using pedometers to increase physical activity and Sports Exerc. 2007;39(8):1423-1434. doi:10.1249/mss.0b013e3180616b27. improve health: a systematic review. JAMA 2007 Nov 21;298(19)2296-304. 43. de Müllenheim PY, Chaudru S, Emily M, et al. Using GPS, accelerometry and heart rate to predict 20. Harris T, Kerry SM, Limb ES, et al. Physical activity levels in adults and older adults 3–4 years after outdoor graded walking energy expenditure. J Sci Med Sport. 2018;21(2):166-172. pedometer-based walking interventions: Long-term follow-up of participants from two randomised doi:10.1016/j.jsams.2017.10.004. controlled trials in UK primary care. PLoS Med. 2018;15(3):e1002526. 44. Fritz T, Huang E, Murphy G, Zimmermann T. Persuasive technology in the real world. In: CHI ’14. ACM; 21. Noah B, Keller MS, Mosadeghi S, et al. Impact of remote patient monitoring on clinical outcomes: an :487-496. doi:10.1145/2556288.2557383. updated meta-analysis of randomized controlled trials. npj Digit Med. 2017;1(1):2. 45. Almalki M, Gray K, Martin-Sanchez F. Refining the Concepts of Self-quantification Needed for Health doi:10.1038/s41746-017-0002-4. Self-management: A Thematic Literature Review. Computer (Long Beach Calif). 2015;79:1-5.

155 Chapter 8

46. Hermsen S, Frost J, Renes RJ, Kerkhof P. Using feedback through digital technology to disrupt and change habitual behavior: A critical review of current literature. Comput Human Behav. 2016;57:61- 74. doi:10.1016/j.chb.2015.12.023. 47. Kluger AN, DeNisi A. The effects of feedback interventions on performance: A historical review, a meta-analysis, and a preliminary feedback intervention theory. Psychol Bull. 1996;119(2):254-284. doi:10.1037/0033-2909.119.2.254. 48. Park Eun-jun MAJM. Computerized tailoring of health information. Comput Informatics, Nurs. 2009;27. 49. Menninga K. Learning abstinence theory - PhD thesis under supervision of A. Dijkstra. Univ Groningen. 2012. 50. Bandura A. Health promotion from the perspective of social cognitive theory. Psychol Heal. 1998;13(4):623-649. 51. Schembre SM, Liao Y, Robertson MC, et al. Just-in-Time Feedback in Diet and Physical Activity Interventions: Systematic Review and Practical Design Framework. J Med Internet Res. 2018;20(3):e106. 52. Gouveia R, Pereira F, Caraban A, Munson SA, Karapanos E. You Have 5 Seconds: Designing Glanceable Feedback for Physical Activity Trackers. UbiComp/ISWC ’15. 2015. doi:10.1145/2800835.2809437. 53. Swinkels ICS, Huygens MWJ, Schoenmakers TM, et al. Lessons Learned From a Living Lab on the Broad Adoption of eHealth in Primary Health Care. J Med Internet Res. 2018;20(3):e83.

156

Samenvatting

Aanleiding Adoptie van zelfmeettechnologie Voldoende bewegen en het verkrijgen van een gezond gewicht is van groot belang voor In hoofdstuk 4 is het gebruik van twee apparaten onderzocht die beweging, slaap en gewicht mensen met overgewicht en mensen met diabetes type 2. Een mogelijke manier om kwantificeren over een periode van zes maanden. Ook is onderzocht welke factoren zelfmanagement naar een gezonde leefstijl te bewerkstelligen is het gebruik van eHealth bijdroegen aan het gebruik van de verschillende meetfuncties (beweging, slaap en gewicht). technologie zoals activity trackers en digitale weegschalen. Deze zelfmeettechnologie kan Uit de resultaten bleek dat de beweegfunctie vaker gebruikt werd dan de slaapfunctie, persoonlijke gezondheidsdata zoals beweging, gewicht en slaap bijhouden over langere tijd. echter nam het gebruik van beide functies af over de tijd. Het aantal keren dat mensen op Doordat deze data geüpload kunnen worden en gedeeld met anderen, zijn er veel de weegschaal stonden nam ook af door de tijd, maar dit stabiliseerde vanaf de derde toepassingen mogelijk. Niet alleen voor de eindgebruikers, maar ook voor zorgprofessionals maand tot de zesde maand waarbij ruim 80% zichzelf wekelijks (één tot vijf metingen per en onderzoekers. Echter voordat deze technologie gebruikt kan worden binnen de week) of dagelijks (zes of meer metingen per week) woog. Verschillende typen factoren gezondheidszorg moet er voldaan worden aan bepaalde voorwaarden. De data moet droegen bij aan het gebruik van de zelfmeetapparaten. Deze factoren waren verschillend bijvoorbeeld van voldoende kwaliteit zijn zodat men op deze informatie kan vertrouwen. gerelateerd aan het gebruik van de verschillende zelfmeetfuncties. Verder is een bepaald niveau van adoptie en betrokkenheid met het apparaat nodig voordat Persoonlijke factoren (leeftijd, BMI, geslacht, en educatieniveau) waren niet gerelateerd aan deze een effect kan hebben op de gezondheid. Daarom is het belangrijk om te weten welke het gebruik van de beweegmeter. Voor het meten van slaap waren persoonlijke factoren wel factoren in verband staan met het gebruik van verschillende apparaten voor zelfmeting. Tot gerelateerd aan het gebruik: jongere mensen (leeftijd tussen de 30-40), mensen met een slot is er ook kennis nodig over de daadwerkelijke effectiviteit van deze technologie. hoger educatieniveau en mensen met een BMI tussen de 25-30 gebruikten de slaapmeter Dit proefschrift heeft zich gefocust op drie verschillende domeinen: (1) de betrouwbaarheid vaker. De weegschaal werd vaker gebruikt door mannen, jongere mensen (tussen de 30-40), en validiteit van activity trackers, (2) de adoptie van apparaten die beweging, slaap en en mensen met een BMI tussen de 25-30. gewicht kwantificeren en (3) de effectiviteit van deze technologie voor mensen met Met betrekking tot gedragsfactoren waren de belangrijkste bevindingen dat het hebben van overgewicht en/of diabetes type 2, als ook voor een algemene populatie van gezonde een specifiek motief voor het doen van zelfmetingen (in vergelijking met een algemeen volwassenen. In dit proefschrift wordt niet ingegaan op andere evident belangrijke aspecten motief: de eigen gezondheid in kaart willen brengen), of het hebben van de intentie om een die onderdeel zijn van de ontwikkelingen op het gebied van zelfmeting, zelfmanagement en specifieke gedraging te veranderen (bijvoorbeeld de intentie om meer te gaan bewegen) eHealth, zoals privacy, veiligheid van data en ethische aspecten rondom het verwerven en bijdroegen aan het gebruik van een zelfmeetapparaat. Van de vier dimensies van het gebruik van data door bijvoorbeeld bedrijven, overheid en zorgverzekeraars. zelfregulatie die onderzocht zijn in dit proefschrift, droeg een hogere ‘doelgerichtheid’ op baseline bij aan het aantal activiteiten- en slaapmetingen tijdens de studieperiode. Betrouwbaarheid en validiteit van activity trackers In hoofdstuk 2 en 3 is de betrouwbaarheid en validiteit van in totaal 20 activity trackers, Ook technische factoren zijn belangrijk wanneer het gaat over adoptie van apparaten voor smartwatches en apps onderzocht voor het gemeten aantal stappen. In de eerste studie zijn zelfmeting. Technisch falen, een beperkte batterijduur, het gebruikersgemak en ervaren nut de Lumoback, Fitbit Flex, Jawbone Up, Nike+ Fuelband SE, Misfit Shine, Withings Pulse, Fitbit van een apparaat zijn uit eerder onderzoek al benoemd als belangrijke factoren voor Zip, Omron HJ-203, Yamax Digiwalker SW-200 en Moves app onderzocht voor lopen op een adoptie. Van onze evaluaties met de deelnemers in zowel Hoofstuk 3 en 5, gaven de gemiddelde snelheid (4.8 km/h) op een loopband en tijdens normale dagelijkse activiteiten deelnemers ook aan dat technische factoren zoals de installatieprocedure en beperkte gedurende een werkdag. Hoewel er verschillen tussen de apparaatjes zijn, waren de meeste batterijduur barrières waren voor het (langdurig) gebruik. betrouwbaar en valide: zowel tijdens lopen op de loopband als tijdens de dagelijkse activiteiten. De Nike+ Fuelband en de Moves app waren hierop een uitzondering. De Fitbit Effectiviteit van zelfmeettechnologie Zip liet de beste validiteit zien. De effectiviteit van het gebruik van zelfmeettechnologie is onderzocht in drie verschillende In de tweede studie zijn de Polar Loop, Garmin Vivosmart, Fitbit Charge HR, Apple Watch studies. In hoofdstuk 5 is een systematische review en meta-analyse uitgevoerd om de Sport, Pebble Smartwatch, Samsung Gear S, Misfit Flash, Jawbone Up Move, Flyfit en Moves effectiviteit van het gebruik van een activiteitenmeter te bepalen op de mate van beweging app onderzocht op drie verschillende snelheden op de loopband (langzaam, gemiddeld en bij mensen met overgewicht of obesitas. We hebben enig bewijs gevonden dat snel). De meeste apparaatjes waren betrouwbaar en valide op een gemiddelde snelheid, gedragsmatige beweeginterventies gecombineerd met activiteitenmeters effect hebben op waarbij de Garmin Vivosmart en Apple Watch Sport de beste validiteit lieten zien. Op een het beweeggedrag bij deze doelgroep. Ook het toevoegen van een activiteitenmeter bij een langzamere snelheid nam de validiteit af voor de meeste activity trackers, behalve voor gedragsmatige beweeginterventie verhoogt het effect op de mate van beweging ten Gamin Vivosmart en Fitbit Charge HR. Op de hoogste snelheid lieten de drie smartwatches opzichte van beweeginterventies zonder activiteitenmeter. In de studies die geïncludeerd de beste validiteit zien. konden worden voor deze review werden echter hoofdzakelijk simpele stappentellers

160 Samenvatting

161 Samenvatting

gebruikt met beperkte mogelijkheden om het beweeggedrag over de tijd visueel weer te geven, en beperkte mogelijkheden voor gepersonaliseerde feedback op individuele doelen.

In hoofdstuk 6 is een Randomized Clinical Trial (RCT) uitgevoerd om het effect vast te stellen van een activity tracker in combinatie met een online leefstijlprogramma voor mensen met diabetes type 2. Dit programma was effectief voor het verhogen van fysieke activiteit bij deze doelgroep. Bij deelnemers die hun dagelijkse stappen verhoogden met minimaal 1000 stappen per dag (gedefinieerd als responders) werd een klinisch relevante en significante afname in HbA1c vastgesteld. Sociale norm voor fysieke activiteit bleek een significante confounder voor de resultaten op HbA1c waarbij responders een hogere sociale norm lieten zien op baseline vergeleken met niet-responders.

In hoofdstuk 7 is de rol van zelfregulatie onderzocht bij het effect van het zelfmeten van beweging en gewicht op BMI-verandering in een algemene populatie van gezonde volwassenen. We hebben vastgesteld dat het BMI significant afnam op korte termijn (vier maanden) en dat dit behouden bleef op lange termijn (12 maanden). De afname in BMI werd verklaard door de intentie om af te vallen, de frequentie van het aantal keer zelfwegen, en toename van zelfregulatiecapaciteit: ‘doelgerichtheid’ op korte termijn en ‘beslissingen maken’ op lange termijn. Gewichtsverlies door het gebruik van zelfmeettechnologie wordt dus mede verklaard door een toename van zelfregulatiecapaciteit. Ook gaven zes op de tien deelnemers aan dat zij meer zijn gaan bewegen, en vier op de tien dat zij hun eetpatroon hebben aangepast door het gebruik van de apparaten.

162 Samenvatting

gebruikt met beperkte mogelijkheden om het beweeggedrag over de tijd visueel weer te geven, en beperkte mogelijkheden voor gepersonaliseerde feedback op individuele doelen.

163

Dankwoord

Over de auteur| About the author

Dankwoord

Over de auteur| About the author

Dankwoord

Dankwoord Over de auteur

Toen ik in 2013 een vacature voor promovendus doorgestuurd kreeg was mijn eerste Thea Kooiman werd geboren op 26 februari 1988 te Breukelen. Na gedachte ‘dit is niks voor mij’. Nu ben ik vijf jaar verder en kan ik deze promotieperiode bijna het afronden van de HAVO aan het Niftarlake college te Maarssen afsluiten. Een mooie tijd met ups en downs die ik zeker niet had willen missen. Ik heb begon zij in 2005 aan de opleiding Oefentherapie Mensendieck aan ontzettend veel geleerd, niet alleen op het gebied van onderzoek maar ook op persoonlijk de Hogeschool van Amsterdam. Binnen deze opleiding volgde ze de vlak. Als eerste wil ik daarom Emmy bedanken omdat je me destijds hebt overtuigd om minor ‘internationalization, working in developing countries’ gewoon te solliciteren. Superleuk dat je nu ook m’n paranimf bent! waarvoor ze twee maanden stage liep in een revalidatiecentrum in Zambia. In 2009 begon zij aan haar vervolgstudie Zonder de hulp van mijn (co)-promotoren: Cees van der Schans, Adriaan Kooy en Martijn de Bewegingswetenschappen aan de Rijksuniversiteit Groningen. Groot, was dit boekje er zeker niet gekomen. Bij deze wil ik jullie dan ook hartelijk bedanken. Tegelijkertijd werkte zij als oefentherapeut in verschillende Martijn, jou wil ik in het bijzonder bedanken voor de prettige overleggen, je peptalks en je praktijken in Groningen. Na het afronden van haar master in 2012 heeft ze in Cambodja vernieuwende inzichten. Dit heb ik zeer gewaardeerd! als vrijwillig oefentherapeut gewerkt in een revalidatiecentrum. Eind 2013 startte ze haar Ook wil ik graag Arie Dijkstra bedanken voor je hulp bij het schrijven van twee artikelen promotietraject aan de Hanzehogeschool Groningen. Hierbij was ze aangesloten bij de binnen de gezondheidspsychologie, Wim Krijnen voor je hulp bij de statistiek en alle overige Graduate School for Medical Sciences (GSMS-SHARE) waar ze diverse co-auteurs voor jullie hulp bij het schrijven van de artikelen. onderzoeksgerelateerde cursussen heeft gevolgd. Tijdens haar promotietraject is ze ook als Verder wil ik ook alle mensen bedanken van de verschillende organisaties en bedrijven oefentherapeut blijven werken in haar eigen praktijk, en later in een duopraktijk in waarmee ik heb samengewerkt; de Quantified Self-community, de innovatiewerkplaats Groningen. Diabetes, Lifelines, het Bethesda Diabetes Research Center en het Martini Ziekenhuis. Bedankt voor de fijne samenwerking en al jullie hulp bij het opzetten en uitvoeren van de studies. Ellen, jou wil ik nog even apart benoemen vanwege je geweldige enthousiasme About the author waarmee je patiënten bent gaan werven in het Martini Ziekenhuis, waarmee je de studie gered hebt. Je was voor mij op dat moment echt een engel die uit de hemel kwam vallen. Thea Kooiman was born on February 26th, 1988 in Breukelen, the Netherlands. After the completion of her secondary school she started with the college education ‘Exercise therapy’ Natuurlijk ook alle collega’s van het lectoraat en secretariaat heel erg bedankt voor jullie at the Amsterdam University of Applied Sciences in 2005. During this education she followed gezelligheid en tips en trics op het gebied van onderzoek. Willemke, na alle brainstorm- en the minor ‘internationalization, working in developing countries’ for which she did an inspiratiesessies die we samen gehad hebben kon ik natuurlijk niet anders dan je als internship in a rehabilitation center in Zambia. In 2009 she started her Master education paranimf vragen. Heel erg bedankt voor al je hulp en gezelligheid! program Human Movement Sciences at the University of Groningen. At the same time, she Tot slot, familie, vrienden, en Anne: bedankt voor al jullie support de afgelopen jaren! Lieve worked as an exercise therapist in different practices in Groningen. After the completion of Anne, zonder jouw kalmte en nuchtere blik had ik het zeker niet gered. Heel erg bedankt her master program in 2012, she worked in Cambodia as a voluntary exercise therapist. By daarvoor! the end of 2013 she started her PhD research and training program at the Hanze University of Applied Sciences in Groningen. During this program she was a member of the Graduate

School for Medical Sciences (GSMS-SHARE) where she attended various research-related courses. During her PhD trajectory she also continued to work as an exercise therapist in her own practice, and later on in a duo practice in Groningen.

nl.linkedin.com/in/theakooiman

166 Over de auteur

Dankwoord Over de auteur

nl.linkedin.com/in/theakooiman

167

Research Institute SHARE

Previous dissertations Veen HC van der Articulation issues in total hip arthroplasty This thesis is published within the Research Institute SHARE (Science in Healthy Ageing and (prof SK Bulstra, dr JJAM van Raay, dr IHF Reininga, dr I van den Akker-Scheek) healthcaRE) of the University Medical Center Groningen / University of Groningen. Further information regarding the institute and its research can be obtained from our internet site: Elsenburg LK http://www.share.umcg.nl/. Adverse life events and overweight in childhood, adolescence and young adulthood (prof AC Liefbroer, dr N Smidt)

More recent theses can be found in the list below. Becking K ((co-) supervisors are between brackets) Inflammatory matters; exploring the underlying pathophysiology of unipolar and bipolar disorder (prof RA Schoevers, dr BCM Haarman) 2018 ‘t Hoen EFM Miranda Azevedo R de Practical applications of the flexibilities of the agreement on trade-related aspects of Shades of blue; an epidemiological investigation of depressive symptom dimensions and the intellectual property rights; lessons beyond HIV for access to new essential medicines association with cardiovascular disease (prof HV Hogerzeil, prof BCA Toebes) (prof P de Jonge, dr AM Roest) Stojanovska V Pang C Fetal programming in pregnancy-associated disorders; studies in novel preclinical models Computational methods for data discovery, harmonization and integration; using lexical and (prof SA Scherjon, dr T Plösch) semantic matching with an application to biobanking phenotypes (prof MA Swertz, prof JL Hillege) Eersel MEA van The association of cognitive performance with vascular risk factors across adult life span Arifin B (prof JPJ Slaets, dr GJ Izaks, dr JMH Joosten) Distress and health-related quality of life in Indonesian Type 2 diabetes mellitus outpatients (prof MJ Postma, dr PJM Krabbe, dr J Atthobari) Rolfes L Patient participation in pharmacovigilance Zakiyah N (prof EP van Puijenbroek, prof K Taxis, dr FPAM van Hunsel) Women’s health from a global economic perspective (prof MJ Postma, dr ADI van Asselt) Brandenbarg D The role of the general practitioner in the care for patients with colorectal cancer Metting, EI (prof MY Berger, prof GH de Bock, dr AJ Berendsen) Development of patient centered management of asthma and COPD inf primary care (prof T van der Molen, prof R Sanderman, dr JWH Kocks) Oldenkamp M Caregiving experiences of informal caregivers; the importance of characteristics of the Scheffers WJ informal caregiver, care recipient, and care situation Body experience in patients with mental disorders (prof RP Stolk, prof M Hagedoorn, prof RPM Wittek, dr N Smidt) (prof RA Schoevers, dr JT van Busschbach, dr MAJ van Duijn) Kammen K van Suhoyo Y Neuromuscular control of Lokomat guided gait; evaluation of training parameters Feedback during clerkships: the role of culture (prof LHV van der Woude, dr A den Otter, dr AM Boonstra, dr HA Reinders-Messelink) (prof JBM Kuks, prof J Cohen-Schotanus, dr J Schönrock-Adema) Hornman J Kikkert LHJ Stability of development and behavior of preterm children Gait characteristics as indicators of cognitive impairment in geriatric patients (prof SA Reijneveld, prof AF Bos, dr A de Winter) (prof T Hortobagyi, dr CJ Lamoth, dr N Vuillerme)

170 SHARE

171 SHARE

Vries, YA de Acknowledgements Evidence-b(i)ased psychiatry (prof P de Jonge, dr AM Roest)

Smits KPJ Quality of prescribing in chronic kidney disease and type 2 diabetes (prof P denig, prof GJ Navis, prof HJG Bilo, dr GA Sidorenkov)

Zhan Z Evaluation and analysis of stepped wedge designs; application to colorectal cancer follow-up (prof GH de Bock, prof ER van den Heuvel)

For more 2018 and earlier theses visit our website.

172

Vries, YA de Acknowledgements Evidence-b(i)ased psychiatry (prof P de Jonge, dr AM Roest)

Smits KPJ Quality of prescribing in chronic kidney disease and type 2 diabetes (prof P denig, prof GJ Navis, prof HJG Bilo, dr GA Sidorenkov)

Zhan Z Evaluation and analysis of stepped wedge designs; application to colorectal cancer follow-up (prof GH de Bock, prof ER van den Heuvel)

For more 2018 and earlier theses visit our website.

Uitnodiging Thea Kooiman

Voor het bijwonen van de openbare verdediging van mijn proefschrift

Aansluitend bent u van harte welkom voor een hapje en een drankje in het Goudkantoor,

The use of self-tracking technology for health Waagplein 1 te Groningen.

BMI Steps Graag voor 1 november aanmelden voor de borrel via [email protected]

Thea Kooiman [email protected] [email protected]

Paranimfen

BMI Steps Emmy Wietsma Willemke Nijholt

Contact [email protected]

BMI Steps

BMI Steps BMI

BMI Steps BMI Steps Step

The use of self-tracking s

BMI s Step technology for health BMI Steps

Thea Kooiman s Step BMI BMI Step s