Enhancing Physical Activity using Virtual Communities

(1)

(2)

Enhancing Physical Activity using Virtual

Communities

(3)

Ph.D. Dissertation Committee

Chairman and Secretary

Prof. dr. P.M.G. Apers University of Twente, NL

Supervisor

Prof. dr. ir. H.J. Hermens University of Twente, NL

Co-Supervisor

Dr. ir. B.J.F. van Beijnum University of Twente, NL

Members

Prof. dr. M.M.R. Vollenbroek-Hutten University of Twente, NL

Prof. dr. D.K.J. Heylen University of Twente, NL

Prof. dr. P. Maret University Jean Monnet, FR

Prof. dr. ir. A.T. van Halteren Vrije Universiteit Amsterdam, NL

Prof. dr. ir. M.R. van Steen University of Twente, NL

CTIT Ph.D. Thesis Series No. 17-454

Centre for Telematics and Information Technology P.O. Box 217, 7500 AE

Enschede, The Netherlands.

The research reported in this dissertation was carried out at the Biomedical Signals and Systems group of the University of Twente.

ISBN: 978-90-365-4463-4

ISSN 1381-3617 (CTIT Ph.D. thesis Series No. 17-454) DOI: 10.3990/1.9789036544634

https://doi.org/10.3990/1.9789036544634

Typeset with LA_{TEX. Cover design by Lamia Elloumi. Printed by Ipskamp Printing.}

All rights reserved. No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without prior permission from the copyright owner.

(4)

ENHANCING PHYSICAL ACTIVITY USING

VIRTUAL COMMUNITIES

DISSERTATION

to obtain

the degree of doctor at the University of Twente,

on the authority of the rector magnificus

Prof. dr. T.T.M. Palstra

on account of the decision of the graduation committee,

to be publicly defended

on Friday the 26thof January 2018 at 12:45

by

Lamia Elloumi

born on the 3rdof September 1986

(5)

This dissertation has been approved by: Supervisor: Prof. dr. ir. Hermie Hermens Co-Supervisor: Dr. ir. Bert-Jan van Beijnum

(6)

List of figures

1.1 Examples of old generation pedometers . . . 4

1.2 Examples of accelerometers . . . 4

1.3 Examples of new generation pedometers . . . 4

1.4 The user-centered design process, ISO-13407 (ISO 1999) . . . 7

1.5 Iterative design . . . 7

1.6 Development cycles of the thesis . . . 8

2.1 Terms for the search queries . . . 14

2.2 Revised model for telemedicine system success by Hu [51] . . . 17

2.3 Overview of the search and selection process . . . 21

2.4 Included studies per year (n=37) . . . 22

2.5 Number of included publications per country where the studies were done (n=37) . . . 23

2.6 Health condition or lifestyle of the studies’ target population (n=17) . 23 2.7 Duration of the non-clinical trial studies (N=14) . . . 24

2.8 Number of participants for the non-clinical trial studies (N=14) . . . . 24

2.9 Type of clinical trials (1-arm, 2-arm and 3-arm) (N=23) . . . 24

2.10 Duration of the clinical trial studies (N=23) . . . 25

2.11 Number of participants for the clinical trial studies (N=23) . . . 25

2.12 PA assessment methods used in the included studies (n=37) . . . 26

3.1 TogetherActive Architecture Overview . . . 59

3.2 TogetherActive Functional Architecture . . . 63

3.3 Main Pages’ Portlets . . . 65

(11)

vi List of figures

3.5 Informational Model . . . 68

3.6 IMA Values - Recommended versus Recorded . . . 69

3.7 Physical Activity Level over time – Recommended level, High Threshold and Low Threshold . . . 70

3.8 Use of Social Networks . . . 74

4.1 Overview of the Architecture of the TogetherActive System [11] . . . . 83

4.2 CONSORT diagram of the study design . . . 90

4.3 Average number of days (in percentage) with over 3000 steps per day . 95 4.4 Distributions of the activity levels for the intervention group and the control group . . . 96

4.5 Average steps (HA) per week . . . 97

5.1 Community Coaching Functional Architecture . . . 114

5.2 Community Coaching Portal Pages and Portlets . . . 115

5.3 Integration of Community Coaching System in TogetherActiveSystem 118 5.4 Total number of activity invitation notifications. . . 122

6.1 Revised model for telemedicine system success by Hu [14] . . . 130

D.1 TogetherActive Home pages 1 . . . 177

D.4 TogetherActive Group Page 1 . . . 179

D.7 TogetherActive Personal Page 1 . . . 180

(12)

List of tables

2.1 System creation phase . . . 28

2.2 System use phase . . . 29

2.3 System impacts phase . . . 29

2.4 Behaviour change theories used in the included studies . . . 30

2.5 Functions provided in the included studies . . . 32

2.6 Included publications . . . 38

3.1 Satisfaction Results. . . 73

4.1 Points attribution . . . 87

4.2 Participants’ demographics . . . 91

4.3 Health-related indexes . . . 92

4.4 Usability results for the intervention and control groups and for the usability study of the TogetherActive system [11] . . . 93

4.5 Average number of days (in percentage) based on the final number of steps achieved by different groups . . . 94

4.6 Average steps (HA) per week and change in the number of steps (com-pared to week 2) for each week of the study . . . 96

4.7 Correlation Analysis . . . 97

5.1 Demographics. . . 110

5.2 Support group creation and data sharing . . . 111

5.3 Context of use and activity types . . . 111

5.4 Using and managing physical activities . . . 112

(13)

viii List of tables

5.6 Usability Results. . . 120

5.7 The number of invited, registered and real Helpers for the different Main

Users. . . 122

C.1 Main user Usefulness Questionnaire . . . 173

(14)

1

CHAPTER

(15)

2 Introduction

1.1 Background on Physical Activity and Physical

Activity Assessment

According to the World Health Organization (WHO), physical inactivity is the fourth leading risk factor for global mortality causing an estimated 3.2 million deaths globally [1]. Moderate and regular physical activity (PA) has significant benefits on health and can reduce the risk of cardiovascular diseases, diabetes, colon and breast cancer, and depression [1, 2]. It is important for all age groups and health conditions. It is defined as any bodily movement produced by skeletal muscles that require energy expenditure (EE) [1]. EE is the amount of energy (or calories) that a person needs to carry out a physical function. PA includes physical exercises, active transportation (e.g., walking and cycling), labour work, house chores, and stated more generally it comprises of activities of daily living. As a general recommendation, people should aim for at least 30 minutes of PA every day or 300 minutes a week. This recommendation changes depending on age (5-17 years old, 18-64 years old and 65 years and above) [3] and on the health condition (healthy, acute diseases and chronic diseases) [1]. Despite all recommendations and well-known health benefits of regular PA, physical inactivity remains a global health problem [1]. Many researchers from different disciplines (e.g., from social sciences and computer sciences) and health organizations are performing investigations to reduce physical inactivity and to promote healthy PA. For example, the WHO Member States have agreed to reduce physical inactivity by 10% in 2025.

The ability to assess EE and estimate PA in free-living individuals is extremely important. It enables:

• Measuring the impact and effectiveness of programs and interventions designed to increase PA

• Monitoring temporal trends in PA

There are a large number of techniques for the assessment of PA. The accepted criterion to validate techniques of estimating habitual PA is calorimetry [4]. The doubly labelled water (DLW) method is currently the most relevant, albeit expensive, method for measuring free-living EE [5]. It is regarded as the gold standard for measuring free-living EE [5]. DLW can be used as a method to measure the average daily metabolic rate of an organism over a period of time (generally from 4 to 21 days is the optimal metabolic period). However, this method is not useful for the identification of activity types, duration or distribution, in addition to the involved high costs and discomfort of subjects.

In addition to the DLW method, different PA assessment methods are being used to assess the PA of individuals; subjective methods and objective methods.

Subjective PA assessment methods include activity diaries, questionnaires, inter-views, observations and self-reports. Researchers have been investigating a large number

(16)

1.1 Background on Physical Activity and Physical Activity Assessment 3

of questionnaires to measure physical activity. Commonly used self-report question-naires are: Global Physical Activity Questionnaire (GPAQ) [6], Modifiable Activity Questionnaire (MAQ) [7], Previous Week Modifiable Activity Questionnaire (PWMAQ) [8], Recent Physical Activity Questionnaire (RPAQ) [9], International Physical Activity Questionnaires (IPAQ) [10], Previous Day Physical Activity Recall (PDPAR) [11], and 7-day Physical Activity Recall (PAR) [12]. These methods rely on the perception of the questionnaire by the subjects themselves therefore usually lead to under or over estimation of PA. Depending on the questionnaire used, the following outcome measures are possible:

• Number of hours or MET hours per week of PA (MAQ and PWMAQ); one MET is equal to the metabolic equivalent of a task (1 MET represents 3.5 ml/kg/min oxygen consumption)

• Total EE (RPAQ, PDPAR and PAR) • PA EE (RPAQ)

• Total PA scores for each category (IPAQ)

Objective PA assessment methods involve the use of a wearable sensor-based devices that measure a PA level. These devices include pedometers, accelerometers, activity monitors and heart rate monitors.

A pedometer is a device that senses body motion and counts number of steps. The pedometers use a small lever arm that moves up and down in response to the movement of hips that occurs with each taken step, thereby counting steps while moving. Steps are used to quantify the PA level, which is limited to activities such as walking, running and climbing. These pedometers are referred as the old generation pedometers. Yamax DigiWalker [13], StepWatch [14] are examples of pedometers (figure 1.1).

An accelerometer is a device that measures the acceleration. An accelerometer in smartphones and activity monitors measures the acceleration of the device, in lateral, longitudinal, and vertical directions. Accelerometers allow detecting the type, duration, and intensity of PAs which covers a larger number of daily living activities such as sitting, walking, cycling and running. ActivPal [15], Tritrac R3D [16], Tracmor(D) [17] and Actigraph [18] are examples of accelerometers (figure 1.2).

New generation pedometers are equipped with an accelerometer that transfers the acceleration data into number of steps. Fitbit activity trackers [19] are examples of new generation pedometers (figure 1.3).

These accelerometers and pedometers can be worn on thigh, hip, waist, ankle and lower back. Depending on the objective of the assessment methods, different outcome measures are possible: steps, time spent on PA, EE, IMA values, distance travelled and list of activities performed.

(17)

4 Introduction

(a) Yamax DigiWalker (b) StepWatch

Figure. 1.1 Examples of old generation pedometers

(a) ActivPal (b) Actigraph

Figure. 1.2 Examples of accelerometers

Figure. 1.3 Examples of new generation pedometers

Combining multiple PA assessment methods has the potential to provide new insights into the benefits of PA [20, 21]. However, if we are only interested in one facet of PA or aim at a global level of information, one assessment method will be sufficient [22]. When selecting the assessment method we need to pay close attention to each methods’ strengths and limitations [22]. Additionally, we must consider different acceptability and feasibility factors in selecting the appropriate PA assessment methods in relation to the purpose of the study. Subjective methods may produce a bias, influenced by the population characteristics or cultural aspects (e.g. engagement in religious practice, or atypical lifestyle), or introduced by under- or overestimation of the PA [23]. Tarasenko et al. [24] suggested that accelerometers may produce a more modest effect on participants’ behaviour than pedometers. Other studies [25, 26] suggest that when participants become accustomed to checking their pedometers, they work towards increasing their PA. Additionally, participant’s characteristics and context

(18)

1.2 Current Telemedicine applications for Physical Activity Support 5

(e.g. athletes obliged to take off accelerometers and pedometers before engaging in sports activities) must be anticipated and addressed to ensure acceptable and feasible measurement [24]. In general, study characteristics including study design, population, sample size, gender, study objectives and geographical location are the main factors to consider during the process of selection of an optimal PA assessment method(s). Finally, some non-functional constraints such as budget should be considered during the process of selection of optimal PA assessment method(s).

1.2 Current Telemedicine applications for Physical

Activity Support

Current research in ICT-mediated healthcare interventions and systems (i.e. Telemedicine applications) is investigating solutions to address motivation, monitor PA and pro-mote PA. These applications aim to provide support and to manage patients with chronic diseases, and to support healthy population in adopting a healthier lifestyle including regular PA. Some of the ICT-mediated healthcare interventions and systems provide some form of social support, which contributes positively to PA behaviour change [27–29]. Social support for PA can be instrumental (e.g., the provision of functionalities that help in performing certain PA tasks), informational (e.g., sharing information about the benefits of PA), emotional (e.g., calling or messaging to see how the person is doing with a new PA plan), or appraisal (e.g., providing encouragement or reinforcement) [30].

people to become more physically active, e.g., apps to promote PA (the review by Middelweerd [31] provides some examples), active video games (the review by Peng et al. [32] provides some examples) and Internet-based PA interventions (examples provided in the reviews by Vandelanotte et al. [33] and van den Berg et al. [34]). Internet-based solutions include the use of social networks, blogs, forums, newsletters and support groups which provide a means to seek and offer support, and connect with others. Examples of the online health communities are PatientsLikeMe [35], FaceToFaceHealth [36], WebMD [37] and DailyStrength [38], where they focus on providing emotional and informational (peer) support. Persuasive technology in healthcare is also targeting at behavioural changes through persuasion and social influence. PersonA [39], UbiFit [40] and ActiveLifestyle [41] are some examples of systems using persuasive technology in order to change PA behaviours. However, these systems are limited in terms of social support, since they mainly focus on the appraisal support.

Additionally, Telemedicine applications couple ICT-based systems with PA assess-ment tools to monitor the PA and give appropriate feedback to reach a personal target, taking into account the health situation of the person [42–44].

The main drawbacks noticed in these ICT-based solutions are that they provide a limited form of social support. Some of these ICT-based interventions showed a

(19)

6 Introduction

decrease of PA level relative to the baseline week (around 4 weeks in [43, 45]). In order to overcome these drawbacks, online virtual communities (i.e. online community and social network) offer opportunities to implement health interventions to motivate individuals to adopt and maintain healthier PA level. Virtual communities offer the medium to provide functionalities conveying the different forms of social support. The combination of the potential functionalities of virtual communities and the integration of PA assessment methods could make virtual communities a promising intervention platform for increasing PA and maintaining it over the long term.

1.3 Thesis Objectives

Taking into consideration the current status of telemedicine applications for PA, several scientific opportunities and challenges remain. To this end, the main research goal addressed in this thesis is:

How can virtual communities contribute to the enhancement of the support for physical activity?

In order to achieve this goal and to guide the different studies, the following sub goals were defined:

Goal 1: Understand the state of the art of virtual communities for physical activity

support, learn about their characteristics and what influences the PA outcomes, and determine the status of the social support provision.

Goal 2: Design a virtual community aiming at providing social support, implement

the first prototype and evaluate the prototype.

Goal 3: Implement a second prototype and perform an explorative study to

compare the proposed virtual community system to a reduced version without the social support functionalities.

Goal 4: Extend the provision of instrumental support within the virtual community.

1.4 Research Methodology

In this section, we present the high-level overview of the research methodology used in the thesis. The specific studies, executed to achieve the above-mentioned research goals, require specific methods and instruments. They will be presented in each chapter on a more detailed level.

With the purpose of achieving the first research goal, we conducted a comprehensive literature review as the research methodology.

In order to develop the virtual community system, we used a user-centered iterative engineering approach; a combination of the iterative engineering design cycle and the user-centered methodology.

(20)

1.4 Research Methodology 7

The user-centered design (figure 1.4) is an approach that involves end-users through-out the development process of a product. The iterative design (figure 1.5) is a design methodology based on a cyclic process of prototyping, testing, analyzing, and refining of a product. Lessons learned from the previous iteration are implemented in the next iteration.

Figure. 1.4 The user-centered design process, ISO-13407 (ISO 1999)

Analysis and Design

Implementation Testing Evaluation Requirements Planning Initial planning Deployment

Figure. 1.5 Iterative design

Over the years, researchers became aware of the importance of involving the end-users in the development and evaluation process of the health-related applications, and it became a widely used methodology [46–50]. In our research, we followed the same methodology where we involved users in different stages of the process: specifying the requirements and evaluating the design. The different studies performed followed the iterative design methodology. For the scope of our research, we focused on the

(21)

8 Introduction

three phases: design, prototype and evaluate. Figure 1.6 shows the three iterative cycles implemented in this thesis. Each cycle represents a study performed: study 1 is performed in order to achieve research goal 2, study 3 is performed in order to achieve research goal 3, and study 3 is performed in order to achieve research goal 4. The lessons learned from the first iteration were implemented in iterations 2 and 3.

Design

Prototype

Evaluate

Iteration 1

-Study 1

Design

Prototype

Evaluate

Iteration 2

-Study 2

Design

Prototype

Evaluate

Iteration 3

-Study 3

Figure. 1.6 Development cycles of the thesis

1.5 Thesis Outline

In order to address these goals, the following studies were carried out and written down in this thesis.

In chapter 2, we present a literature survey investigating the characteristics of existing virtual communities providing support for PA. In this literature survey, scientific databases are searched for all virtual communities in which social interaction and PA assessment is provided. The results are presented and discussed. (Goal 1)

In Chapter 3, we present the concepts and the design of the virtual community system called TogetherActive for PA support using the objective PA assessment method ProMove. We present the first prototype of the TogetherActive and the performed usability study. (Goal 2)

In chapter 4, we present the second prototype of the TogetherActive virtual commu-nity system. We also present an explorative study to observe the differences between

(22)

1.5 Thesis Outline 9

using the TogetherActive and a reduced version without the social support functionali-ties. (Goal 3)

In chapter 5, in order to extend the provision of instrumental support, we propose the ICT-mediated Community Coaching system as a strategy to turn PA into a mixed physical/social activity to study its impact on motivation. The chapter outlines the requirement elicitation, the design, the implementation, and the evaluation of the Community Coaching system. (Goal 4)

Finally, chapter 6 presents a summary of the contributions presented in the thesis and discussion regarding the implications for future work.

(23)

(24)

2

CHAPTER

Towards an understanding of existing

online communities for physical activity

support: A literature review

L. Elloumi, B.J. van Beijnum, and H. Hermens, "Towards an understanding of existing online

(25)

12

Towards an understanding of existing online communities for physical activity support: A literature review

Abstract

Background: Physical inactivity is increasingly becoming part of today’s

lifestyle, leading to the rapid rise of such diseases as cardiovascular diseases, diabetes, and obesity. These chronic diseases are, for the most part, entirely preventable by adopting healthy lifestyles, including regular physical activity. To help people adopt an appropriate physical activity level, researchers are developing health interventions based on ICT solutions as online communities.

Purpose: This study provides a comprehensive review of existing online

virtual communities for physical activity support targeting the adult population, as well as providing information regarding factors that may have influenced physical activity outcomes. These factors include the following: demographic characteristics, physical activity assessment methods and outcomes, evaluation phases, behavioural change theories and provided functionalities.

Method: A systematic review was performed investigating the research on

online communities for physical activity support. A literature search was conducted in three electronic databases, PubMed, Scopus and ScienceDirect, examining articles published up to August 2016.

Results: Thirty-seven unique studies on online communities promoting

physi-cal activity met the inclusion criteria for the review [1–37]. Studies with positive effects on physical activity outcomes showed that the use of objective methods for PA assessment, the provision of functions within the categories (physical activity dashboard, feedback, social interaction, system interaction, and facts and information), and the consideration of all the three phases of evaluation (system creation, system use and system impacts) endorse these positive out-comes. Additionally, findings within this review uncovered new opportunities to investigate the provision of social support with respect to instrumental support.

Conclusions: We identified several factors that should be considered in the

design of systems and studies delivering social support via online communities in order to promote their success. Future research is needed to explore new functionalities for instrumental support to maximize the online communities’ positive effects on physical activity.

(26)

2.1 Introduction 13

2.1 Introduction

Physical activity (PA) has beneficial effects on an individual’s health and well-being [38]. PA contributes to the prevention and treatment of chronic conditions, such as cardiovascular diseases, diabetes, cancer, obesity, depression and emotional stress [38]. Given the positive health benefits associated with PA, researchers have investigated methods and interventions to promote PA. A growing body of research has employed information and communication technologies (ICT), such as the Internet and smart-phones as tools for disseminating and supporting PA interventions. The benefits of using ICT-based interventions to promote PA include the potential to reach a large number of people at a relatively low cost [39]. Although there have been numerous studies on the use of ICT-based interventions to promote PA [40], online communities and social media offer a platform for widely delivering these interventions. In addition to the advantages of ICT-based interventions, online community-based interventions offer social support. It is well-known that social support contributes positively to PA behaviour change [41]; in particular, social support from family and friends has been shown to be consistently and positively related to regular PA [42, 43].

Past reviews on social media and online communities for PA support showed that studies of social media interventions relating to healthy lifestyles tend to show low levels of participation and do not show significant differences between groups in key outcomes [44], and most of the Internet-based PA interventions (61.9%) were associated with significant increases in PA [39]. To the best of our knowledge, the existing reviews provide an overview of published Internet-based PA interventions, either restricted to randomized controlled clinical trials [45–47] or considering different research designs [39] (randomized controlled trials, randomized trials without a control condition, non-randomized controlled trials, and single group design). These reviews also provide few outcomes to consider such PA levels (eventually clinical outcomes), assessment methods, and Internet-based provided functionalities. However, they did not present different knowledge gained from these studies.

The objective of this review is to gain knowledge from previous studies that aimed at delivering social support via online communities, such as web-based interventions for PA support. We focus exclusively on online community-based interventions targeting adult populations.

The following questions guide the analysis towards the realization of this review objective:

• What are the reported effects on PA outcomes?

• What is the goal and purpose of these studies? Were they technically and/or clinically evaluated?

• Are these studies based on behaviour change theory? How did it influence the outcomes?

(27)

14

Towards an understanding of existing online communities for physical activity support: A literature review • What are the provided (social) functions by these systems/platforms? How did

it influence the outcomes?

• What characteristics of online communities/studies have been investigated? (e.g., setting, duration, participants, PA assessment methods)

• What are the lessons learned for a successful online community-based intervention for PA?

This review paper is organized as follows. The methodology used for this literature search is presented in Section 2.2. The results of the literature search are described in Section 2.3 and discussed in Section 2.4. Finally, the study’s conclusions are given in Section 2.5.

2.2 Method

2.2.1 Literature search strategy

Electronic literature searches were conducted on the following databases: PubMed, Scopus and ScienceDirect. The queries were based on terms related to virtual com-munities and PA. The terms and combinations that were used are listed in Figure 2.1. Publications were included up to August 2016 with no restrictions on the year of publication. online OR virtual OR social OR web OR web-based OR internet-based OR support OR electronic community OR group OR network physical activity OR physical inactivity AND AND

e-support OR e-group OR peer OR e-community

web OR internet OR online OR web-based OR website OR web page OR portal OR social media OR forum OR wiki OR blog OR

facebook OR twitter

OR

(28)

2.2 Method 15

2.2.2 Inclusion and exclusion criteria for research article

selec-tion

Studies and clinical trials that fulfilled the following eligibility criteria were included in this review:

• The target population is adults (over 18), regardless of their health diagnoses or conditions

• The target primary or secondary outcome is PA

• The PA is recorded by an electronic device (e.g., pedometer, accelerometer) or is self-reported

• Studies incorporate social interaction that uses Internet-supported technology • Studies include technical evaluation or clinical trials

• The publication language is English

The exclusion criteria in this review were as follows: • The target population is teenagers or children

• The study methods or algorithms examine activity recognition • Studies use only telephone or face-to-face interventions

• Studies are review papers

2.2.3 Selection method

Two independent researchers performed the selection on the basis of titles and abstracts (LE and BJB). Each abstract was categorized into either an included, excluded or doubt group. The results from the included and doubt group were compared between researchers. For publications with doubt, the content was scanned by the researchers. After the scan, the researchers made the decision of inclusion or exclusion of the publication. By the end of the process, a final selection of publications was established.

2.2.4 Methods of data collection and analysis

Relevant data from the selected publications were summarized in review table previously agreed on by the 2 researchers (LE and BJB). The review table captured the following: study characteristics (purpose and design), demographic characteristics (target group, sample size, sex, age and country), assessment method for PA and study outcomes regarding the PA level. Extra information was captured, if available, such as the evaluation stages of the studies, provided functionalities in the studies and behavioural change theories deployed in the studies.

(29)

16

Study and demographic characteristics

The publications were categorized by year, countries, and health condition of the participants in the studies.

A differentiation was made between clinical trials and non-clinical trials. Clinical trial studies aim to investigate the effectiveness of a treatment or intervention with respect to the primary outcomes, PA in this case. Non-clinical studies may include this metric, as well, with a focus being placed on the technical evaluation of systems. The non-clinical trial studies were categorized according to the following study designs: design and early evaluation studies, exploratory studies, field studies, and pilot studies. Both clinical trial and non-clinical trial studies were categorized by year, duration length of studies and number of participants.

Physical activity assessment methods and outcomes

Different PA assessment methods can be used to capture the PA of individuals: objective methods and subjective methods. Subjective methods include activity diaries, questionnaires, interviews, observations and self-report. These methods rely on the questions and perceptions of PA behaviour by the subject itself, which can lead to under- and/or overestimation of PA [48]. Objective methods include accelerometers (uniaxial or triaxial) and pedometers. Depending on the assessment methods, different resulting outcome measures are possible: steps, minutes of PA, energy expenditure, IMA values and list of activities performed. For each study included in this review, the PA assessment methods and outcomes were identified.

Finally, the effect of the performed studies on the PA outcomes was investigated. The possible effects (for either clinical trial or non-clinical trial studies) are as follows: • Positive effect: where the PA level for an intervention group was higher than the

PA level for a control group by the end of the performed study

• Potential effect: where the PA level for an intervention group showed a tendency that it would be higher than the PA level for a control group by the end of the performed study, but this finding was yet to be confirmed

• No effect: where the PA level for an intervention group did not differ from the PA level for a control group by the end of the performed study

• Short-term effect: where the PA level for an intervention group was higher than the PA level for a control group during the first period of the performed study, and by the end of the study, both groups had no differences in PA levels

Evaluation phases

The literature shows that proper evaluation of telemedicine applications is challenging and can be very complex. Hermens and Vollenbroek-Hutten [49] emphasized the

(30)

2.2 Method 17

importance of technical evaluation of the systems before investigating their clinical effectiveness in order to increase the likelihood of success. In their contribution, the evaluation framework by DeChant was suggested [50]. This approach has four stages of assessments that depends on the maturity of the system: stages 1 and 2 focus on the technical and clinical feasibility of the systems, and stages 3 and 4 focus on the global impacts on health care of the systems. However, according to the authors [51, 52], the DeChant framework failed to provide a holistic and comprehensive view of the systems’ success or to offer practical guidance on its evaluations.

After reviewing studies focusing on providing evaluation frameworks applicable to telemedicine applications, and after taking into consideration the frameworks that focus on technical as well as clinical evaluation, we selected the model by Hu [51], adapted from information systems to telemedicine applications (Figure 2.2).

Figure. 2.2 Revised model for telemedicine system success by Hu [51]

This model and the staged approach to the evaluation of telemedicine by DeChant et al. [50] share a similar vision, in which both consider the evaluation from two perspectives: technical system evaluation and system impact. It is based on the information systems success model by DeLone and McLean [53, 54] which provides comprehensive and practical guidance to evaluating systems, and it is widely used in different domains. In the model by Hu, there are three evaluation phases (system creation, system use and system impacts) and several distinct success categories (e.g., input data quality and system use) that have both a temporal and causal relation (Figure 2.2). The success categories are as follows:

• Input data quality: quality of acquired data or signals from various sources (e.g., patients and medical equipment/devices).

• System quality: qualities valued by users (e.g., usability, reliability, response time, and availability); technical success.

(31)

18

Towards an understanding of existing online communities for physical activity support: A literature review • Information quality: quality of graphical interfaces; relevant information provided to the patients and any information produced by the system; semantic success. • System Use: behavioural aspect of system use; attitudinal or cognitive

measure-ments common to technology acceptance/adoption or effectiveness assessmeasure-ments may be adequate.

• User satisfaction: the degree to which an individual is satisfied with his or her satisfaction with specific functions and/or overall use of the system under evaluation.

• Service impacts: effects of a system on the resultant service rendered and delivered to remote participants.

• Individual impacts: telemedicine system’s individual impact on formal care providers (e.g., cardiologists and nursing staff) and recipients (e.g., patients), as well as on the informal caregivers (e.g., patients’ family and friends). It may include improvements in the domain or medical knowledge and clinical decision making.

• Organizational impact: the telemedicine system’s impacts on the overall organi-zation at either end of a connection link is relatively challenging. It may include financial aspects and clinical outcomes (e.g., morbidity and mortality).

To gain an understanding of the maturity of the systems in the included publications, the publications were mapped into the model by Hu. Within each success category, the mapping was based on objective and subjective measurements of the categories.

2.2.5 Behavioural change theories

Behavioural change theories are attempts to help in understanding and explaining why and how people do or do not practice health promoting behaviours, to help identify what information is needed to design an effective intervention strategy and to provide insight into how to design a behaviour change program so that it is successful. Adopting healthy PA behaviours represents one of the behaviour changes. The five most widely used theoretical models of health behaviour change are as follows:

• The Health Belief Model [55]: For people to change a health behaviour, their perception of the threat of a health problem and the appraisal of the recommended behaviour for preventing or managing the problem must outweigh their perceived barriers to action.

• The Transtheoretical Model (a.k.a. Stages of Change Model) [55, 56]: In adopting healthy PA behaviours or eliminating unhealthy ones, individuals make progress through five levels related to their readiness to change—pre-contemplation,

con-templation, preparation, action, and maintenance. At each stage, different

(32)

2.2 Method 19

• Social Learning (a.k.a. Social Cognitive Theory) [55, 57]: Individuals’ health behavioural change is the result of reciprocal relationships with the environment, personal factors, and attributes of the behaviour itself. Self-efficacy is one of the most important characteristics that determine behavioural change.

• Theory of Planned Behaviour [55, 58]: An individual’s perceived control over the opportunities, resources, and skills needed to perform a behaviour affect behavioural intentions. Intentions are determined by two factors—attitudes towards the behaviour and beliefs regarding other individuals’ support of the behaviour.

• Social Support [59]: Often incorporated into health promotion interventions, social support can be instrumental, informational, emotional, or appraising (providing feedback and reinforcement of new behaviours).

To gain an understanding of the included publications, we identified the health behaviour change theories used.

2.2.6 Social support functionalities

Various studies investigated potential functionalities to include telemedicine applications that focus on the encouragement of physical activities and being more physically active. Different methodologies have been used to define these functionalities: requirements elicitation through prototypes tested with real users [60–62], requirements elicitation through participatory action research design involving different stakeholders (end users, caregivers, designers and researchers) [63] and requirements elicitation through leveraging a systematic design framework from existing theories and concepts [64, 65].

Taking into account the results from these studies [60–65] and our interest in telemedicine applications using virtual communities for PA support, we propose the following categories, with examples of functions in each category:

• PA dashboard: User is able to be self-aware about activity level (history of past behaviour, current status, and activity level performance) and set and plan personal goals

• Feedback: User is able to get feedback on personal and peer performance • System interaction: User is able to receive reminders or rewards/punishments,

suggestions, similarities and likings from the system • Social interaction: User is able to do the following:

– Give and receive encouragements and support from peers to reach PA goals – Share and compare information related to their PA with peers

(33)

20

– Cooperate with peers

– Be part of an online competition and challenge peers. – Be recognized for achieved PA

• Facts and information: User is able to have access to texts and links to up-to-date information about PA

Social support is expressed in informational support (advice, suggestions, and information), emotional support (expressions of empathy, love, trust and caring), appraisal support (information that is useful for self-evaluation) and instrumental support (tangible aid and service). The functions within the facts and information category contribute to the provision of the informational support. Some functions within the social interaction category contribute to emotional support. Functions within the feedback and system interaction categories contribute to the appraisal support. Functions within the PA dashboard and social interaction categories contribute to the instrumental support.

The included publications in this review were mapped onto these categories in order to gain an understanding of how the functions and social support are provided to their users.

2.3 Results

2.3.1 Database search

The titles and abstracts of the publications were retrieved from the databases PubMed, Scopus and ScienceDirect. In total, 971 publications were found from the search. After the removal of duplications and the fast identification of publications with exclusion criteria (review papers, non-English papers, studies/clinical trials targeting non-adults), 623 publications remained. Two independent reviewers screened the title and abstract and applied the criteria accordingly. The results from both reviewers were combined. Mismatches in "included" publications and publications in the "doubt" group were discussed. In total, 69 abstracts remained. The retrieved 69 full-text publications were read before detailed analysis. After the reading process, 32 publications were excluded. The excluded publications did not meet two of the inclusion criteria: social interaction and/or PA assessment. Finally, 37 full-text publications were included for this review [1–37]. Figure 2.3 illustrates the process performed.

2.3.2 Characteristics of the included studies

Table 2.6 describes the primary characteristics of interest (authorship, study purpose, study setting, sample characteristics, PA metric and assessment method, social function provided, and main outcomes related to PA) of each publication.

(34)

2.3 Results 21

Publications identified through database searching

Total n=971 PubMed: 267 results

Scopus: 390 results ScienceDirect: 314 results

Titles and abstracts screened by two independent researchers

n=623

Search

Studies on virtual communities for physical activity support

Removal of duplicates

n=226

Removal of publications with exclusion criteria (review, language and under target age)

n=122

Full-text publications reviewed for relevance

n=69

Abstracts that did not meet the inclusion criteria

n=554

Final publications included

n=37

Full-text publications that did not meet inclusion criteria

n=32

Figure. 2.3 Overview of the search and selection process

Geographic regions and publication year

Publications were grouped by year of publication (Figure 2.4) and demographic regions where the studies were performed (Figure 2.5). Most of the publications were published from 2012 to August 2016 (29 studies out of 37 studies, approximately 75%); the oldest study included dates from 2001. Most studies were performed in the USA (23 studies out of 37 studies, approximately 60%) and two studies were performed jointly with Canada [24,28]. Five studies occurred in Australia [7,9,12–14], four studies in Europe (one in UK [32], one in Sweden [37], and two in the Netherlands [31,36]) and three studies in Asia (one in Malaysia [23], one in Korea [30], and one in UAE [29]).

Health conditions

Most of the studies did not focus on specific health conditions or lifestyles (20 studies out of 37 studies). The 17 studies with on specific health conditions or lifestyles were grouped based on the health condition or lifestyle of their target population (Figure 2.6). Several of the studies focus on multiple health conditions or lifestyles [1, 5, 12, 13]. From the studies with a focus on health conditions, there were studies focusing

(35)

22

Towards an understanding of existing online communities for physical activity support: A literature review 0 1 2 3 4 5 6 7 8 9 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Nu m b er of p u b lica ti on s Year Trial studies Non-trial studies

Figure. 2.4 Included studies per year (n=37)

on diabetes [1, 24], coronary artery disease [1], COPD [11, 17], cancer survivors [8, 16], rheumatoid arthritis [37], serious mental illness [19], metabolic syndrome [23] and obesity [12, 13, 35]. Other studies focused on a lifestyle: smokers [30], sedentary behaviour [4, 5], weight concerns 6, and overweight [1, 5, 12, 13, 25].

2.3.3 Study and demographic characteristics

The included studies have different study settings. The scope of the studies is as follows:

• 14 non-clinical trial studies [24–37], with 7 design and early evaluation studies [30–33,35–37] involving the design and early evaluation of systems/interventions, 2 exploratory studies [27,29], 1 field study [28] and 4 pilot studies [24–26,34] • 23 clinical trial studies [1–23], which represent more than half of the total number

of included studies

Regarding the 14 non-clinical trial studies [24–37], these studies occurred between 2001 and 2016 (Figure 2.4). The experiments/evaluations performed within these studies have different lengths of duration (Figure 2.7), varying from 1 week [30,36] to 16 weeks [26]. Similarly, the studies vary greatly in the number of participants (Figure 2.8), from 5 participants [30] to 5333 participants [26]. Out of the 14 studies, 6 studies [24,25,27,28,32,34] were based on randomization of their participants into two groups: the intervention group with the social component and the control group with the traditional system. The goal of these studies was varied: evaluating the short-term benefits of adding social components traditional intervention on PA [24], examining the feasibility and efficacy of additional social components on PA [25,27], investigating

(36)

2.3 Results 23 0 5 10 15 20 25 Korea Malaysia Sweden UAE UK Netherlands Multiple countries* Australia USA Number of publications C ou n tr ie s

* multi-countries studies where participants were recruited from USA and Canada

Figure. 2.5 Number of included publications per country where the studies were done (n=37)

0 1 2 3 4 5

Diabetes Coronary Artery Disease COPD Cancer survivors Rheumatoid Arthritis Mental illness Metabolic syndrome Obesity Number Health Condition 0 1 2 3 4 5 Smoker Sedentary With weight concerns Overweight

Number Lifestyle

Figure. 2.6 Health condition or lifestyle of the studies’ target population (n=17)

the added values of social networks and exploring the effectiveness of additional social components on PA [28,32]. Two studies divided their participants into two groups (friends vs. co-workers in the study) [29] and into teams (652 teams in the study) [26]. One study [29] explored the influence of social functionalities on PA, and another study [26] examined the effect of team-based social components on PA.

The 23 clinical trial studies were conducted between 2010 and 2016 (Figure 2.4) with most of them between 2013 and 2016 (5 clinical trials per year). The clinical trials are 1-arm, 2-arm, and 3-arm trials (Figure 2.9), with most of them being 2-arm trials (18 out of 23 clinical trials, which represents ∼80% of the studies). Most of these studies (the 18 2-arm studies) compare an intervention group with social components to a control group with no social components. Five studies were performed in Australia [7,9,12–14], one study in Malaysia [23] and twelve studies (∼74%) in the USA. The clinical trials duration varies from 8 weeks [15] to 2 years [18,19] (Figure 2.10). Similarly, the clinical trials have different numbers of participants, varying from 32 participants [19] to 2894 participants [9] (Figure 2.11). The study [18,19] with the longest duration

(37)

24

Towards an understanding of existing online communities for physical activity support: A literature review occurred in the USA, and the study [9] with the largest number of participants occurred in Australia. 0 2 4 6 8 10 12 14 16 Mckay et al. (2001) [24] Pullen et al. (2008) [25] Leahey et al. (2010) [26] Kharrazi et al. (2011) [27] Kamal et al. (2013) [28] Khalil et al. (2013) [29] Sohn et al. (2007) [30] Fialho et al. (2009) [31] Foster et al. (2010) [32] Munson et al. (2012) [33] Gotsis et al. (2013) [34] Ayubi et al. (2014) [35] Elloumi et al. (2015) [36] Revenas et al. (2016) [37] Length in weeks St u d ie s

Figure. 2.7 Duration of the non-clinical trial studies (N=14)

78 21 5333 60 32 8 5 12 10 14 23 87 10 28 1 10 100 1000 Mckay et al. (2001) [24] Pullen et al. (2008) [25] Leahey et al. (2010) [26] Kharrazi et al. (2011) [27] Kamal et al. (2013) [28] Khalil et al. (2013) [29] Sohn et al. (2007) [30] Fialho et al. (2009) [31] Foster et al. (2010) [32] Ayubi et al. (2014) [33] Munson et al. (2012) [34] Gotsis et al. (2013) [35] Elloumi et al. (2015) [36] Revenas et al. (2016) [37] Number of participants

Figure. 2.8 Number of participants for the non-clinical trial studies (N=14)

1-arm Trial, 1 Randomized 2-arm Control Trial, 18 Randomized 3-arm Control Trial, 4

(38)

2.3 Results 25 0 20 40 60 80 100 Richardson et al. (2010) [1] Cavallo et al. (2012) [2] Glasgow et al. (2012) [3] Carr et al. (2013) [4] Carr et al. (2013) [5] Greene et al. (2013) [6] Kolt et al. (2013) [7] Valle et al. (2013) [8] Caperchione et al. (2014) [9] Cavallo et al. (2014) [10] Martinez et al. (2014) [11] Morgan et al. (2014) [12] Jane et al. (2015) [13] Maher et al. (2015) [14] Rote et al. (2015) [15] Valle et al. (2015) [16] Nguyen et al. (2016) [17] Laska et al. (2016) [18] Aschbrenner et al. (2016) [19] Monroe et al. (2016) [20] Willis et al. (2016) [21] Munson et al. (2015) [22] Chee et al. (2014) [23] Length in weeks St u d ies

Figure. 2.10 Duration of the clinical trial studies (N=23)

324 135 463 53 40 349 504 86 2894 134 239 159 120 110 63 86 1650 441 32 63 73 165 120 1 10 100 1000 Richardson et al. (2010) [1] Cavallo et al. (2012) [2] Glasgow et al. (2012) [3] Carr et al. (2013) [4] Carr et al. (2013) [5] Greene et al. (2013) [6] Kolt et al. (2013) [7] Valle et al. (2013) [8] Caperchione et al. (2014) [9] Cavallo et al. (2014) [10] Martinez et al. (2014) [11] Morgan et al. (2014) [12] Jane et al. (2015) [13] Maher et al. (2015) [14] Rote et al. (2015) [15] Valle et al. (2015) [16] Nguyen et al. (2016) [17] Laska et al. (2016) [18] Aschbrenner et al. (2016) [19] Monroe et al. (2016) [20] Willis et al. (2016) [21] Munson et al. (2015) [22] Chee et al. (2014) [23] Number of participants St u d ies

Figure. 2.11 Number of participants for the clinical trial studies (N=23)

2.3.4 PA assessment methods and outcomes

The studies used different physical assessment methods (Figure ??, either with objective methods (independent accelerometers or pedometers or integrated into participants’ phones) connected directly to the system or self-reported methods, or with subjective methods (questionnaires, survey and diaries). Ten studies (out of the 37 studies) used subjective PA assessment methods (of which five are non-clinical trial studies and five are clinical trial studies), and twenty-seven studies used objective PA assessment methods (of which nine are non-clinical trial studies and eighteen are clinical trial studies). Three studies used Fitbit as an objective PA assessment method [19,21,22]; these studies are clinical trials.

The outcome measures were as follows:

(39)

26

Towards an understanding of existing online communities for physical activity support: A literature review 5 5 9 18 0 2 4 6 8 10 12 14 16 18 20 22 24

Non-Trials studies Trial studies

To

ta

l

Type of study Subjective methods Objective methods

Figure. 2.12 PA assessment methods used in the included studies (n=37)

• Minutes of PA in 6 studies [2,10,18,24,28,34] • Energy expenditure in three studies [3,25,31] • IMA in one study [36]

• Activities performed in one study [33]

• Multiple outcomes (minutes of PA, energy expenditure and/or steps) in four studies [5,9,35,37]

When looking at the outcomes of PA levels of the included studies, out of the 37 studies, 23 studies reported actual PA outcomes:

• 11 studies [3,5,6,8,14,15,23,24,26,29,32] showed a positive effect, in which inter-vention groups had higher and increased PA levels compared to control groups with no social components

• 3 studies [25,28,31] showed a potential effect on intervention group PA levels compared to those of control groups

• 7 studies [1,2,12,16,18,20,34] showed no effect on or difference in PA levels for both groups

• 2 studies [4,22] showed a short-term positive effect of intervention group PA levels compared to those of control groups; there was an increase in the PA levels in the intervention groups during the first period, followed by similar PA levels for both groups

Out of the 14 studies with no direct outcomes on PA, 8 clinical trial studies were studies with not-yet-published or processed results or with PA data as a secondary outcome.

When considering the effect of the various types of studies (clinical vs. non-clinical), we observed the following. Of the 11 studies that reported a positive effect on PA, 64%

(40)

2.3 Results 27

of these studies were clinical trials [3,5,6,8,14,15,23]. Of the 3 studies that reported a potential effect, none of these were clinical trials. Of the 7 studies that reported no effect, 86% of these studies were clinical trials [1,2,12,16,18,20]. Finally, the 2 studies that reported a short-term effect were both clinical trials [4,22].

When considering the effect of various durations of studies (10 weeks or less vs. more than 10 weeks), we observed the following. Of the 11 studies that reported a positive effect on PA, 36% of these studies had a duration of 10 weeks or less [15,24,29,32]. Of the 3 studies that reported a potential effect, 67% of these studies had a duration of 10 weeks or less [28,31]. Of the 7 studies that reported no effect, 14% of these studies had a duration of 10 weeks or less [34]. Finally, the 2 studies that reported a short-term effect both had a duration of more than 10 weeks.

When considering the effect of various numbers of participants included in the studies (100 participants or fewer vs. more than 100 participants), we observed the following. Of the 11 studies that reported a positive effect on PA, 55% of these studies included 100 participants or fewer [5,8,15,24,29,32]. Of the 3 studies that reported a potential effect, all of these studies included 100 participants or fewer [25,28,31]. Of the 7 studies that reported no effect, 14% of these studies included 100 participants or fewer [16,20,34]. Finally, of the 2 studies that reported a short-term effect, 50% of these studies included 100 participants or fewer [4]. When considering the effect of the different PA assessment methods (objective vs. subjective), we observed the following. Of the 11 studies that reported a positive effect on PA, 71% of these studies used an objective method [5,6,8,14,15,23,26,29,32]. Of the 3 studies that reported a potential effect, 33% of these studies used an objective method [31]. Of the 7 studies that reported no effect, 57% of these studies used an objective method [1,12,16,20]. Finally, of the 2 studies that reported a short-term effect, none of them used an objective method.

2.3.5 Evaluations phases

The included studies were mapped to the three evaluation phases: system creation (Table 2.1), system use (Table 2.2) and system impacts (Table 2.3). One study can be part of one or more phases. Within each phase, each study was mapped to the appropriate success category based on objective and/or subjective measures. Four studies addressed measures within all the three phases [7–9,24].

System creation

When looking into the system creation phase, we can observe that few studies consider the success category system quality, and none addressed the input data quality or the information quality categories. Eleven studies out of the 37 studies addressed the

(41)

28

Towards an understanding of existing online communities for physical activity support: A literature review system quality category with the use of the measures usability, ease of use, feasibility, acceptability and safety (all measures are shown in Table 2.1).

System use

Out the 37 included studies, only one study [17] did not address any success category within the system use phase. Within the other 36 studies, 23 studies addressed the system use success category and 28 studies addressed the user satisfaction success category. Most of the studies address one or two measures within the system use phase, but one study [31] had 6 measures, two studies [20,37] had 4 measures and four studies [7,9,12] had 3 measures (all measures are shown in Table 2.2).

As a measure for the success category system use, most of the studies focused on the frequency of use of specific functions offered by their system (21 studies out of the 36 studies). In addition, two studies measured the frequency of access to the system, two studies measured the adherence, and one study examined the appropriate use of measures for the system use category.

As a measure for the success category user satisfaction, most of the studies focused on overall satisfaction (19 studies of the 36 studies). Within this category, one study [31] addressed 5 measures, three studies [20,21,37] addressed 3 measures, and the rest of studies addressed one or 2 measures.

System impacts

Within the system impacts phases, studies focused on the service impacts and organi-zational impacts success categories. Service impacts were measured via the PA level in 23 studies [1–6,8,12,14–16,18,20,22–26,28,29,31,32,34], physical health outcomes in 15 studies [5,6,9,11–13,16–19,21,23,25,27,34], and quality of life in 5 studies [7,8,11,14,21]. Organizational impacts were measured via the hospital visits (emergency visits or hospitalization or observational visits) in two studies [11,17], financial costs for one study [21] and mortality in one study [17] (all measures are shown in Table 2.3).

Table 2.1 System creation phase Individual study results mapped to the Hu model

System

creation

Input data quality

System quality Ease of use [24] Usability [7,9,35,36]

Safety [17] Feasibility [35,37] Acceptability [8]

Information qual-ity

Of the 11 studies that reported positive effects on PA, 2 studies addressed all 3 evalu-ation phases [8,24], and 9 studies addressed only phases 2 and 3 [3,5,6,14,15,23,26,29,32].

(42)

2.3 Results 29

Table 2.2 System use phase Individual study results mapped to the Hu model

System

use System Use

Frequency of use of specific functions [1,3, 6–9,11–15,20,22–24,26,28,31,32,36,37]

Adherence [8,12] Frequency of ac-cess [7,34] Appropriate use [27] User satisfaction Overall satisfaction [5,7,9,11–14, 20–22, 24,25,28,30,31,33, 35–37] Self-efficacy [3,4,9,16,20,21] Enjoyment [31,37] Competence [1,18,31] Intention [10,29] Perceived social support [2,16, 19–21,37]

Attitude [4,10,29] Perceived useful-ness [25,31]

Table 2.3 System impacts phase Individual study results mapped to the Hu model

System impacts Service Impacts Physical activity level [1–6,8,12,14–16, 18,20,22–26,28,29, 31,32,34] Physical health outcomes [5,6,9,11–13,16– 19,21,23,25,27,34] Quality of life [7,8,11,14,21] Individual impacts Organizational impacts Emergency visits, Hospitalization, Observational stays [11,17] Cost [21] Mortality [17]

Of the studies that reported potential effects on PA, all the studies addressed only evaluation phases 2 and 3 [25,28,31]. Of the 7 studies that reported no effects on PA, all studies addressed only evaluation phases 2 and 3 [1,2,12,16,18,20,34]. Of the 2 studies that reported short-term effects on PA, all studies addressed only evaluation phases 2 and 3 [4,22].

2.3.6 Behavioural change theories

Table 2.4 represents the behaviour change theories used in the included studies. The social cognitive theory and the theory of planned behaviour were the most used theories. Out of the 37 studies, 24 studies were based on one or multiple behaviour change theories, in which 16 of these studies were clinical trial studies and 8 of these studies were non-clinical studies. Of the 11 studies that showed positive effects on the PA level, 4 studies were not based on a behavioural change theory: two clinical trial studies and two design and early evaluation studies. Within these 11 studies, three studies used social cognitive theory and three studies used a combination of 2 theories.

(43)

30

Table 2.4 Behaviour change theories used in the included studies

So cial cognitiv e theory Theory of planned b eha viour So cial influence theories So cial ecological theory Self-managemen t mo del T ranstheoretical mo del Theory of self-regulation F un theory Ecological theory of health b eha viours So cial net w orking theory Health Promotion Mo del So cial learning theories Gratifications Theories Theory of reasoned action So cial supp ort Theory Multiple b eha viour c hange theories Richardson et al. (2010) [1] x x Cavallo et al. (2012) [2] x Glasgow et al. (2012) [3] x x Carr et al. (2013) [4] x Carr et al. (2013) [5] x Greene et al. (2013) [6] Kolt et al. (2013) [7] x x x Valle et al. (2013) [8] x Caperchione et al. (2014) [9] Cavallo et al. (2014) [10] x Martinez et al. (2014) [11] x Morgan et al. (2014) [12] x Jane et al. (2015) [13] Maher et al. (2015) [14] x x Rote et al. (2015) [15] Valle et al. (2015) [16] x Nguyen et al. (2016) [17] Laska et al. (2016) [18] x x x Aschbrenner et al. (2016) [19] Monroe et al. (2016) [20] x Willis et al. (2016) [21] x Munson et al. (2015) [22] Chee et al. (2014) [23] x Mckay et al. (2001) [24] Pullen et al. (2008) [25] x Leahey et al. (2010) [26] x x Kharrazi et al. (2011) [27] x Kamal et al. (2013) [28] x x Khalil et al. (2013) [29] x Sohn et al. (2007) [30] Fialho et al. (2009) [31] Foster et al. (2010) [32] Munson et al. (2012) [33] x Gotsis et al. (2013) [34] Ayubi et al. (2014) [35] x Elloumi et al. (2015) [36] x Revenas et al. (2016) [37]

(44)

2.3 Results 31

When considering the effect of the use of a behavioural change theory (based on a behaviour change theory vs. not based on a behaviour change theory), we observed the following. Of the 11 studies that reported a positive effect on PA, 3 of these studies were based on multiple theories [3,14,26], 2 studies were based on social cognitive theory [5,8], one study was based on the theory of reasoned action [29] and one study was based on the transtheoretical model [23]. Of the 3 studies that reported potential effects on PA, one study was not based on a specified theory [31], one study was based on multiple theories [28], and on study was based on the health promotion model [25]. Of the 7 studies that reported no effect on PA, one study was not based on a specified theory [34], two studies were based on multiple theories [1,18], three studies were based on social cognitive theory [12,16,20], and one study was based on the theory of planned behaviour [2]. Of the 2 studies that reported short-term effects on PA, one study had no specified theory [22], and one study was based on the social cognitive model [4].

2.3.7 Social support functionalities

Table 2.5 gives a summary of the provided social functions in the studies based on the following categories: recording, visualization, dialogue support, social support, and facts and information.

Thirteen studies used the existing social media platform Facebook to deliver social support to the participants [2,8,10,13–16,19,21–23,27,32]. In 9 studies, administrator or moderator or coach (occupational therapist) roles were introduced to moderate the communication between the participants or to communicate directly with the participants [8,10,13,16,18,21,24,27,31]. In 13 studies, peers were able to view their peers’ PA levels [4,6,7,14,17,28,29,31,32,34–37]; 8 of these studies were clinical trials. A total of 14 studies included virtual competition and virtual rewards as deployed strategies [4–6,12,14,26–28,30,31,33–36]; 9 of these studies were design and early evaluation studies.

When considering the effect of the use of specific functions, we observed the following. Of the 11 studies that reported a positive effect on PA, 3 of these studies were based on multiple theories [3,14,26], all studies provided PA dashboard functionalities [3,5,6,8,14,15,23,24,26,29,32], 10 studies provided feedback [3,5,6,8,14,15,24,26,29,32] and social interaction [3,5,6,8,14,15,23,24,26,32] functionalities, 8 studies provided system interaction functionalities [5,6,8,14,15,23,26,32] and 7 studies provided facts and information functionalities. For the system interaction functionalities, 5 studies used Facebook [8,14,15,23,32] and 4 studies used virtual competitions and rewards [5,6,14,26]. For social interaction, 4 studies used community resources [3,5,24,26]; 3 studies used private IM [5,6,15] and invitation of external friends [6,8,32] and sharing with peers [6,14,32]; 2 studies used email [5,14], phone or face-to-face meetings [3,24], and involvement of administrators or moderators or coaches [8,24]; and one study invited real friends as participants [14]. When combining functionalities from system

Enhancing Physical Activity using Virtual Communities

Enhancing Physical Activity using Virtual

Communities

ENHANCING PHYSICAL ACTIVITY USING

VIRTUAL COMMUNITIES

DISSERTATION

Lamia Elloumi

Table of contents

List of figures

List of tables

1

1.1

Background on Physical Activity and Physical

Activity Assessment

1.2

Current Telemedicine applications for Physical

Activity Support

1.3

Thesis Objectives

1.4

Research Methodology

Design

Prototype

Evaluate

Iteration 1

-Study 1

Design

Prototype

Evaluate

Iteration 2

-Study 2

Design

Prototype

Evaluate

Iteration 3

-Study 3

1.5

Thesis Outline

2

Towards an understanding of existing

online communities for physical activity

support: A literature review

Abstract

2.1

Introduction

2.2

Method

2.2.1

Literature search strategy

2.2.2

Inclusion and exclusion criteria for research article

selec-tion

2.2.3

Selection method

2.2.4

Methods of data collection and analysis

2.2.5

Behavioural change theories

2.2.6

Social support functionalities

2.3

Results

2.3.1

Database search

2.3.2

Characteristics of the included studies

2.3.3

Study and demographic characteristics

2.3.4

PA assessment methods and outcomes

2.3.5

Evaluations phases

2.3.6

Behavioural change theories

2.3.7

Social support functionalities