567004 research-article2015
JHI0010.1177/1460458214567004Health Informatics JournalMuntaner et al.
Original Article
Increasing physical activity through mobile device interventions: A systematic review
Health Informatics Journal 1–19 © The Author(s) 2015 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/1460458214567004 jhi.sagepub.com
Adrià Muntaner, Josep Vidal-Conti and Pere Palou University of the Balearic Islands, Spain
Abstract Physical inactivity is a health problem that affects people worldwide and has been identified as the fourth largest risk factor for overall mortality (contributing to 6% of deaths globally). Many researchers have tried to increase physical activity levels through traditional methods without much success. Thus, many researchers are turning to mobile technology as an emerging method for changing health behaviours. This systematic review sought to summarise and update the existing scientific literature on increasing physical activity through mobile device interventions, taking into account the methodological quality of the studies. The articles were identified by searching the PubMed, SCOPUS and SPORTDiscus databases for studies published between January 2003 and December 2013. Studies investigating efforts to increase physical activity through mobile phone or even personal digital assistant interventions were included. The search results allowed the inclusion of 11 studies that gave rise to 12 publications. Six of the articles included in this review reported significant increases in physical activity levels. The number of studies using mobile devices for interventions has increased exponentially in the last few years, but future investigations with better methodological quality are needed to draw stronger conclusions regarding how to increase physical activity through mobile device interventions.
Keywords cellular phone, mobile application, mobile phone, personal digital assistant, physical activity, short message service, text message
Introduction The positive effects of physical activity (PA) on health and wellness are widely established and documented for all ages. These effects include an improvement in cardiovascular disease, diabetes, osteoporosis and some cancers.1–4 However, despite widespread messages regarding the beneficial impacts of modest amounts of PA, 6 per cent of all deaths globally are attributable to physical Corresponding author: Adrià Muntaner, Physical Activity and Sport Science Research Group, University of the Balearic Islands, Valldemossa Road, km 7.5 Majorca, E-07122, Spain. Email: [email protected]
Downloaded from jhi.sagepub.com at UNIVERSITE LAVAL on March 4, 2015
2
Health Informatics Journal
inactivity;5 in fact, physical inactivity is the fourth leading risk factor for global mortality. This evidence supports the conclusion that physical inactivity is one of the most important public health problems of the 21st century.6 Rates of physical inactivity are growing rapidly in all parts of the world, which is a phenomenon that has attracted the attention of the international healthcare community. The positive effects of PA on people’s health have been described in numerous reviews,7,8 and a variety of behavioural modification programmes have been developed with the intention of reducing rates of physical inactivity. Although there are currently PA interventions that target behaviourally sedentary healthy adults at reasonable costs,9 new tools to reduce costs and increase the effectiveness of interventions are emerging, such as mobile phones and personal digital assistants (PDAs). In particular, the field of telehealth (eHealth) has seen growth as a paradigm involving the concepts of health and the use of technology as tools in health services. In this context and thanks to advances in communications technology, a related new concept has arisen: mHealth (mobile health), a component of eHealth.10 The World Health Organization (WHO) defines mHealth as a ‘medical and public health practice supported by mobile devices, such as mobile phones, patient monitoring devices, PDAs, and other wireless devices’.11 The technological capabilities of mobile technologies are continuing to advance at a rapid pace. Current technological capabilities allow for low-cost interventions.12 At present, there are over 5 billion mobile phones in the world, of which 1.08 billion are smartphones. Approximately 80 per cent of all the people in the world have a mobile phone.13 Mobile phones have the potential to improve population health due to their multiple functions. Many interventions have used mobile technology to engage and motivate healthy behaviours, such as the treatment of diabetes through text messages, smoking cessation, controlling blood pressure and diet control.14–17 Another emerging approach in the primary prevention of disease is encouraging increased PA through mobile device interventions. We have found only one review that specifically focused on increasing PA in the general population through the use of mobile devices.18 This earlier review requires updating because it did not provide a detailed review of interventions via mobile devices (e.g. accurate descriptions of the intervention and study groups and detailed descriptions of the study characteristics). Our review adds a qualitative assessment of the included studies using a previously validated checklist and also provides four new articles; moreover, it includes a detailed study protocol, a literature search for suitable studies based on inclusion criteria and an analysis plan.19,20 The aim of this systematic review was to identify, retrieve, critically appraise and synthesise the existing scientific literature on PA mobile device interventions. The research question was ‘Can these mobile device interventions increase PA?’
Methods Data sources and search strategy The electronic databases searched were PubMed, SCOPUS and SPORTDiscus. These databases were searched for studies conducted between January 2003 and December 2013. The search was limited to English and Spanish language publications. We started the literature search in May 2013 and conducted updates until December 2013. We searched for relevant studies using a combination of three different categories. For these three categories, all relevant keyword variations were used, including both keyword variations in the controlled vocabularies of the various databases and free textword variations in these concepts. The first category included key words that focused on the type of intervention and included (1) ‘text messaging’ (MeSH); ‘cellular phone’ (MeSH); ‘telephone’ (textword); ‘mobile device’ (textword) and ‘SMS’ (textword). The second category included the intended outcome of the intervention and included (2) ‘PA’ (MeSH); ‘exercise’ (MeSH); ‘physical fitness’ (MeSH) and ‘sports’ (MeSH). The third category was in regard to the type of study
Downloaded from jhi.sagepub.com at UNIVERSITE LAVAL on March 4, 2015
3
Muntaner et al.
design and included (3) ‘intervention’ (textword); ‘program’ (textword) and ‘application’ (textword). This search strategy was used for all the consulted databases, taking into account the differences of the various controlled vocabularies. We also systematically searched the reference lists of the included studies.
Inclusion criteria and exclusion criteria To be included in the review, an article had to meet the following criteria: 1. 2. 3. 4.
Reviewed and published in English or Spanish. Used mobile devices such as phones or PDAs as an intervention to increase PA. PA was an outcome. Published between January 2003 and December 2013.
The reference lists of the selected articles were checked for additional eligible articles using the same inclusion criteria. The criteria for exclusion from the review were as follows: 1. 2. 3. 4.
The article focused on a goal other than increasing or promoting PA. The study did not use mobile devices to deliver the intervention. The reviews did not meet pre-defined criteria for methodological quality. The research participants were diseased.
The articles that consisted of proof-of-concept trials, conference proceedings or review articles were also excluded from this analysis. Outcomes from studies that were not explicitly related to PA were also excluded.
Study selection The process used to identify and select the articles for review is shown in Figure 1. In total, 422 articles (228 identified via PubMed, 160 identified via Scopus and 34 identified via SPORTDiscus) were found, of which 128 were duplicates. During the review, all citations from each database were imported into a bibliographic management program (RefWorks 2.0),21 which facilitated the removal of duplicates. In the first stage, the reviewers searched titles for publications that referenced mobile phone interventions that sought to affect PA, and based on this review of titles, 198 publications did not meet the inclusion criteria, with the majority of these (183) being excluded because they did not include a PA mobile device intervention. In the following stage, the reviewers examined the abstracts of the remaining 96 publications. Forty-six of these abstracts were excluded because they did not match the inclusion criteria; the majority of those articles did not include a PA mobile device intervention. In the final stage, 50 full-text citations were reviewed to ensure that all the criteria were met. Thirty-eight of the 50 full-text citations did not meet the inclusion criteria. After completely reviewing the corresponding full-text articles, the total number of accepted articles was reduced to 12. The remaining 38 studies were excluded because their interventions were not predominantly delivered by mobile phone (n = 22), they included no reporting of PA outcomes (n = 10), they consisted of a review article (n = 1) or they consisted of protocol studies (n = 5).
Data extraction Data extraction was completed by three reviewers (A.M., J.V.-C. and P.P.) for accuracy. The fulltext articles were extracted and examined using a methodology quality checklist that was guided
Downloaded from jhi.sagepub.com at UNIVERSITE LAVAL on March 4, 2015
4
Health Informatics Journal
Search of electronic databases (n= 422 publications) • Pubmed (n= 228) • Scopus (n= 160) • SPORTDiscus (n= 34) Excluded duplications across databases (n= 128) 294 publications
Excluded after titles screened (n=198) • No physical activity mobile device intervention (n= 183) • Review articles (n= 6) • Protocol studies (n= 9)
96 publications
Excluded after abstracts screened (n=46) • No physical activity mobile device intervention (n= 38) • Review articles (n= 2) • Protocol studies (n= 6)
50 publications
Excluded after full-text articles screened (n=38) • No predominantly delivered by mobile device (n=22) • No reporting physical activity outcome (n= 10) • Review article (n=1) • Protocol studies (n=5)
12 publications
Figure 1. Flow chart outlining the article selection process for the present review.
by items from an available quality assessment tool.22 Information was extracted regarding study characteristics (year, author, study design, country, number of participants, age of participants and gender of participants), targeted health behaviours (PA, weight loss or other) and characteristics of the intervention (duration, description of contents, methods, statistics and the intervention and control groups). Whenever possible, the pre-test and post-test results of any PA measurements were also included. The reviewers were blinded to the authors or journals involved when they were extracting the data. The screening for eligible articles and the data extraction from the selected articles were performed independently.
Study quality assessment The article quality was rated independently by three authors (A.M., J.V.-C. and P.P.) who separately scored the methodological quality of the included articles using a validated 26-point checklist.22 The methodological quality of each study was assessed based on four dimensions (e.g.
Downloaded from jhi.sagepub.com at UNIVERSITE LAVAL on March 4, 2015
5
Muntaner et al. Table 1. Methodological quality of studies. Reference
Reporting (11)
External validity (3)
Bias (7)
Confounding (6)
Total score (27)
%
Conroy et al.25 (2011) Cheung et al.26 (2008) David et al.27 (2012) Fjeldsoe et al.28 (2010) Hurling et al.29 (2007) Kim and Glanz30 (2013) Kirwan et al.31 (2012) Prestwich et al.32 (2009) Prestwich et al.33 (2010) Shapiro et al.34 (2008) Sirriyeh et al.35 (2010) Turner-McGrievy and Tate36 (2011)
7 8 7 9 8 8 6 6 9 9 7 8
2 3 2 2 2 1 0 2 1 0 1 1
3 2 3 3 2 4 6 3 4 4 4 3
3 1 4 4 3 3 3 4 4 4 4 5
15 14 16 19 15 16 15 15 18 17 16 17
58 54 62 69 58 62 58 58 69 65 62 65
reporting (10 items), which assessed whether the information provided in the article was sufficient to allow a reader to make an unbiased assessment of the findings of the study; external validity (3 items), which addressed the extent to which the findings from the study could be generalised to the population from which the study subjects were derived; bias (7 items), which addressed biases in the measurement of the intervention and outcome and confounding (6 items), which addressed bias in the selection of study subjects). The maximum score a study could receive was 27, with higher scores indicating higher quality. The answers were scored as 0 or 1, except for one item in the reporting subscale, which was scored as 0 to 2. Items were rated ‘1’ if the requested information was present in the article and the criteria were met. Items were rated ‘0’ if the requested information was not present in the article and the criteria were not met. Items were also rated ‘0’ if insufficient information was provided. A parallel assessment was conducted regarding the quality criteria for applying PA interventions and mHealth interventions (Table 2). These criteria were derived according to review literature on PA assessments in general and on evaluation methods for behavioural intervention technologies.23,24 All criteria were scored as ‘yes’, ‘no’ or ‘unable to determine’.
Results Methodological quality assessment The full methodological quality scoring of the included studies is presented in Table 1. The mean ± standard deviation (SD) Downs and Black score was 16.08 ± 1.44 (range: 14–19). The mean ± SD of the percentage of the analysed studies was 61.66 ± 4.69. One study met 19 criteria,25 implying a higher methodological quality, while the other studies were rated with various scores.26–36 Regarding the reporting dimension, three studies were rated with 9 points, indicating the highest score,28,33,34 and only two studies25,34 reported the adverse events that may have been a consequence of the intervention. Additionally, four studies25,27,31,32 did not report the number of participants lost to follow-up. Regarding external validity, one study26 obtained the maximum score, and two studies31,34 received 0 points. Regarding internal validity, one study31 scored 6 points, four studies30,33–35 scored 4 points, five studies25,27,28,32,36 scored 3 points and two studies26,29
Downloaded from jhi.sagepub.com at UNIVERSITE LAVAL on March 4, 2015
6
Health Informatics Journal
Table 2. Characteristics of intervention, process and outcome of studies. Study authors
Adapted interventions to subject characteristics
Feedback Based on Support Combination on theoretical tools of PA progress guidelines assessment measures
Objective PA assessment methods
Conroy et al.25 (2011) Cheung et al.26 (2008) David et al.27 (2012) Fjeldsoe et al.28 (2010) Hurling et al.29 (2007) Kim and Glanz30 (2013) Kirwan et al.31 (2012) Prestwich et al.32 (2009) Prestwich et al.33 (2010) Shapiro et al.34 (2008) Sirriyeh et al.35 (2010) Turner-McGrievy and Tate36 (2011)
Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Yes No No No Yes No No No No Yes No Yes
Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
No Yes Yes Yes Yes No No Yes Yes Yes Yes Yes
Yes Yes Yes Yes Yes Yes Yes No No No No Yes
No No No No No Yes No No No No No No
PA: physical activity.
scored 2 points. Only two studies31,35 attempted to blind the study subjects to the interventions received. One study31 blinded those individuals responsible for measuring the main outcome of the intervention. Regarding confounding, one study36 scored 5 points, six studies27,28,32–35 scored 4 points, four studies25,29–31 scored 3 points and one study26 scored 1 point. Furthermore, in 11 studies,25–31,33–36 it was impossible to determine whether the randomised intervention was concealed from both the patients and the healthcare staff until the final recruitment. Regarding the quality criteria that apply to PA interventions and mHealth interventions, the results showed that 12 studies were adapted to the characteristics of the participants,25–36 and four articles25,29,34,36 provided feedback to the participants throughout the given intervention (Table 2). Ten studies used one or more theoretical models to compose the information transmitted to the intervention group.25,27,29,32–36 These models were the Social Cognitive Theory,27,34,36 the Protection Motivation Theory,32,33 the Transtheorical Model,26,27 the Theory of Planned Behaviour,29,35 the Goal Setting Theory27 and the Problem Solving Theory.27 Two of the included studies used a combination of methods to assess PA.29,30 Eight articles25–31,36 used support tools to deliver the interventions (e.g. paper diaries,25 pedometers,26,27,30,31,34 accelerometers,29 interactive voice response,27 face-to face sessions,28 mobile applications31 and podcasts26). Two studies30,31 used a pedometer and a validated questionnaire to assess PA measures. All the studies25–36 used objective methods, such as accelerometers, pedometers or questionnaires, to assess the outcomes.
Review characteristics Characteristics of the study populations and selected studies The characteristics of the selected studies and study populations are described in Table 3. Five of the 12 studies were conducted in the United States,25,27,30,34,36 four were conducted in the United Kingdom,29,32,33,35 two were conducted in Australia28,31 and one was conducted in Hong Kong.26 The resulting articles were published over a 7-year period from 2007 to 2013. The durations of the
Downloaded from jhi.sagepub.com at UNIVERSITE LAVAL on March 4, 2015
United Kingdom
United States
United Kingdom
United Kingdom
United States
United Kingdom
Hurling et al.29 (2007)
Kim and Glanz30 (2013) Kirwan et al.31 (2012) Prestwich et al.32 (2009)
Prestwich et al.33 (2010)
Shapiro et al.34 (2008)
Sirriyeh et al.35 (2010)
Downloaded from jhi.sagepub.com at UNIVERSITE LAVAL on March 4, 2015
PA
48 58
36 (IG: 26, CG: 10) 200 (IG-1: 50, IG-2: 150) 154 (IG-1: 29, IG-2: 31, IG-3: 29, CG-1: 31, CG-2: 34) 134 (IG-1: 40, IG-2: 48, CG: 46)
45 (IG: 30, CG: 15) 200 (IG-1: 50, IG-2: 50) –
31 (IG-1: 13, IG-2: 7, CG: 11) 120 (IG-1: 31, IG-2: 30, IG-3: 31, CG: 28) 86 (IG-1: 44, IG-2: 42)
58 (IG-1: 18, IG-2: 18, CG: 22)
128(IG-1: 32, IG-2: 31, IG-3: 33, C: 32)
96 (IG-1: 49, IG-2: 47)
149 (IG-1: 47, IG-2: 52, CG: 50)
PA, walking
80.5
77 (IG: 47, CG: 30)
77 (IG: 47, CG: 30)
75
70
62
64
66
PA, weight loss
PA
PA
PA, walking
PA, walking
PA
PA
88 (IG: 45, CG: 43)
88 (IG: 45; CG: 43) 100
39 (IG-1: 21, IG-2: 18) PA, walking
PA, weight loss PA
Health behaviour
71 (IG-1: 35, IG-2: 36) 100
78.8
84
Gender (% female)
SD: standard deviation; IG: intervention group; CG: control group; PA: physical activity; BMI: body mass index. The bold values indicate the overall numbers of participants in each study.
Turner-McGrievy United States and Tate36 (2011)
Australia
Australia
Fjeldsoe et al.28 (2010)
United States
Hong Kong
189 (IG-1: 61, IG-2: 64, IG-3: 64) 52 (IG: 38, CG: 14)
210 (IG-1: 72, IG-2: 68, IG-3: 70) 52 (IG: 38, CG: 14)
Conroy et al.25 (2011) Cheung et al.26 (2008) David et al.27 (2012)
United States
Treat to analysis/ participants with complete data
Participants randomised
Participants
Country
Table 3. Study characteristics.
24
2
8
4
4
12
6
9
12
12
6
24
Length (weeks)
IG-1: 43.2 ± 11.7 IG-2: 42.6 ± 10.7
17.3 ± 0.68
8.7 ± 2.3
23.44 ± 5.63
IG: 70.55 ± 7.5 CG: 69.31 ± 7.3 IG-1: 39.3 ± 12.8 IG-2: 40.1 ± 12.1 23.76 ± 4.64
40.4 ± 7.6
30 ± 6
IG: 38.9 ± 10.8 CG: 26.5 ± 1.9 57 ± 5
47.3 ± 8.8
Age (mean ± SD; years)
Inclusion criteria: exercise
JHI0010.1177/1460458214567004Health Informatics JournalMuntaner et al.
Original Article
Increasing physical activity through mobile device interventions: A systematic review
Health Informatics Journal 1–19 © The Author(s) 2015 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/1460458214567004 jhi.sagepub.com
Adrià Muntaner, Josep Vidal-Conti and Pere Palou University of the Balearic Islands, Spain
Abstract Physical inactivity is a health problem that affects people worldwide and has been identified as the fourth largest risk factor for overall mortality (contributing to 6% of deaths globally). Many researchers have tried to increase physical activity levels through traditional methods without much success. Thus, many researchers are turning to mobile technology as an emerging method for changing health behaviours. This systematic review sought to summarise and update the existing scientific literature on increasing physical activity through mobile device interventions, taking into account the methodological quality of the studies. The articles were identified by searching the PubMed, SCOPUS and SPORTDiscus databases for studies published between January 2003 and December 2013. Studies investigating efforts to increase physical activity through mobile phone or even personal digital assistant interventions were included. The search results allowed the inclusion of 11 studies that gave rise to 12 publications. Six of the articles included in this review reported significant increases in physical activity levels. The number of studies using mobile devices for interventions has increased exponentially in the last few years, but future investigations with better methodological quality are needed to draw stronger conclusions regarding how to increase physical activity through mobile device interventions.
Keywords cellular phone, mobile application, mobile phone, personal digital assistant, physical activity, short message service, text message
Introduction The positive effects of physical activity (PA) on health and wellness are widely established and documented for all ages. These effects include an improvement in cardiovascular disease, diabetes, osteoporosis and some cancers.1–4 However, despite widespread messages regarding the beneficial impacts of modest amounts of PA, 6 per cent of all deaths globally are attributable to physical Corresponding author: Adrià Muntaner, Physical Activity and Sport Science Research Group, University of the Balearic Islands, Valldemossa Road, km 7.5 Majorca, E-07122, Spain. Email: [email protected]
Downloaded from jhi.sagepub.com at UNIVERSITE LAVAL on March 4, 2015
2
Health Informatics Journal
inactivity;5 in fact, physical inactivity is the fourth leading risk factor for global mortality. This evidence supports the conclusion that physical inactivity is one of the most important public health problems of the 21st century.6 Rates of physical inactivity are growing rapidly in all parts of the world, which is a phenomenon that has attracted the attention of the international healthcare community. The positive effects of PA on people’s health have been described in numerous reviews,7,8 and a variety of behavioural modification programmes have been developed with the intention of reducing rates of physical inactivity. Although there are currently PA interventions that target behaviourally sedentary healthy adults at reasonable costs,9 new tools to reduce costs and increase the effectiveness of interventions are emerging, such as mobile phones and personal digital assistants (PDAs). In particular, the field of telehealth (eHealth) has seen growth as a paradigm involving the concepts of health and the use of technology as tools in health services. In this context and thanks to advances in communications technology, a related new concept has arisen: mHealth (mobile health), a component of eHealth.10 The World Health Organization (WHO) defines mHealth as a ‘medical and public health practice supported by mobile devices, such as mobile phones, patient monitoring devices, PDAs, and other wireless devices’.11 The technological capabilities of mobile technologies are continuing to advance at a rapid pace. Current technological capabilities allow for low-cost interventions.12 At present, there are over 5 billion mobile phones in the world, of which 1.08 billion are smartphones. Approximately 80 per cent of all the people in the world have a mobile phone.13 Mobile phones have the potential to improve population health due to their multiple functions. Many interventions have used mobile technology to engage and motivate healthy behaviours, such as the treatment of diabetes through text messages, smoking cessation, controlling blood pressure and diet control.14–17 Another emerging approach in the primary prevention of disease is encouraging increased PA through mobile device interventions. We have found only one review that specifically focused on increasing PA in the general population through the use of mobile devices.18 This earlier review requires updating because it did not provide a detailed review of interventions via mobile devices (e.g. accurate descriptions of the intervention and study groups and detailed descriptions of the study characteristics). Our review adds a qualitative assessment of the included studies using a previously validated checklist and also provides four new articles; moreover, it includes a detailed study protocol, a literature search for suitable studies based on inclusion criteria and an analysis plan.19,20 The aim of this systematic review was to identify, retrieve, critically appraise and synthesise the existing scientific literature on PA mobile device interventions. The research question was ‘Can these mobile device interventions increase PA?’
Methods Data sources and search strategy The electronic databases searched were PubMed, SCOPUS and SPORTDiscus. These databases were searched for studies conducted between January 2003 and December 2013. The search was limited to English and Spanish language publications. We started the literature search in May 2013 and conducted updates until December 2013. We searched for relevant studies using a combination of three different categories. For these three categories, all relevant keyword variations were used, including both keyword variations in the controlled vocabularies of the various databases and free textword variations in these concepts. The first category included key words that focused on the type of intervention and included (1) ‘text messaging’ (MeSH); ‘cellular phone’ (MeSH); ‘telephone’ (textword); ‘mobile device’ (textword) and ‘SMS’ (textword). The second category included the intended outcome of the intervention and included (2) ‘PA’ (MeSH); ‘exercise’ (MeSH); ‘physical fitness’ (MeSH) and ‘sports’ (MeSH). The third category was in regard to the type of study
Downloaded from jhi.sagepub.com at UNIVERSITE LAVAL on March 4, 2015
3
Muntaner et al.
design and included (3) ‘intervention’ (textword); ‘program’ (textword) and ‘application’ (textword). This search strategy was used for all the consulted databases, taking into account the differences of the various controlled vocabularies. We also systematically searched the reference lists of the included studies.
Inclusion criteria and exclusion criteria To be included in the review, an article had to meet the following criteria: 1. 2. 3. 4.
Reviewed and published in English or Spanish. Used mobile devices such as phones or PDAs as an intervention to increase PA. PA was an outcome. Published between January 2003 and December 2013.
The reference lists of the selected articles were checked for additional eligible articles using the same inclusion criteria. The criteria for exclusion from the review were as follows: 1. 2. 3. 4.
The article focused on a goal other than increasing or promoting PA. The study did not use mobile devices to deliver the intervention. The reviews did not meet pre-defined criteria for methodological quality. The research participants were diseased.
The articles that consisted of proof-of-concept trials, conference proceedings or review articles were also excluded from this analysis. Outcomes from studies that were not explicitly related to PA were also excluded.
Study selection The process used to identify and select the articles for review is shown in Figure 1. In total, 422 articles (228 identified via PubMed, 160 identified via Scopus and 34 identified via SPORTDiscus) were found, of which 128 were duplicates. During the review, all citations from each database were imported into a bibliographic management program (RefWorks 2.0),21 which facilitated the removal of duplicates. In the first stage, the reviewers searched titles for publications that referenced mobile phone interventions that sought to affect PA, and based on this review of titles, 198 publications did not meet the inclusion criteria, with the majority of these (183) being excluded because they did not include a PA mobile device intervention. In the following stage, the reviewers examined the abstracts of the remaining 96 publications. Forty-six of these abstracts were excluded because they did not match the inclusion criteria; the majority of those articles did not include a PA mobile device intervention. In the final stage, 50 full-text citations were reviewed to ensure that all the criteria were met. Thirty-eight of the 50 full-text citations did not meet the inclusion criteria. After completely reviewing the corresponding full-text articles, the total number of accepted articles was reduced to 12. The remaining 38 studies were excluded because their interventions were not predominantly delivered by mobile phone (n = 22), they included no reporting of PA outcomes (n = 10), they consisted of a review article (n = 1) or they consisted of protocol studies (n = 5).
Data extraction Data extraction was completed by three reviewers (A.M., J.V.-C. and P.P.) for accuracy. The fulltext articles were extracted and examined using a methodology quality checklist that was guided
Downloaded from jhi.sagepub.com at UNIVERSITE LAVAL on March 4, 2015
4
Health Informatics Journal
Search of electronic databases (n= 422 publications) • Pubmed (n= 228) • Scopus (n= 160) • SPORTDiscus (n= 34) Excluded duplications across databases (n= 128) 294 publications
Excluded after titles screened (n=198) • No physical activity mobile device intervention (n= 183) • Review articles (n= 6) • Protocol studies (n= 9)
96 publications
Excluded after abstracts screened (n=46) • No physical activity mobile device intervention (n= 38) • Review articles (n= 2) • Protocol studies (n= 6)
50 publications
Excluded after full-text articles screened (n=38) • No predominantly delivered by mobile device (n=22) • No reporting physical activity outcome (n= 10) • Review article (n=1) • Protocol studies (n=5)
12 publications
Figure 1. Flow chart outlining the article selection process for the present review.
by items from an available quality assessment tool.22 Information was extracted regarding study characteristics (year, author, study design, country, number of participants, age of participants and gender of participants), targeted health behaviours (PA, weight loss or other) and characteristics of the intervention (duration, description of contents, methods, statistics and the intervention and control groups). Whenever possible, the pre-test and post-test results of any PA measurements were also included. The reviewers were blinded to the authors or journals involved when they were extracting the data. The screening for eligible articles and the data extraction from the selected articles were performed independently.
Study quality assessment The article quality was rated independently by three authors (A.M., J.V.-C. and P.P.) who separately scored the methodological quality of the included articles using a validated 26-point checklist.22 The methodological quality of each study was assessed based on four dimensions (e.g.
Downloaded from jhi.sagepub.com at UNIVERSITE LAVAL on March 4, 2015
5
Muntaner et al. Table 1. Methodological quality of studies. Reference
Reporting (11)
External validity (3)
Bias (7)
Confounding (6)
Total score (27)
%
Conroy et al.25 (2011) Cheung et al.26 (2008) David et al.27 (2012) Fjeldsoe et al.28 (2010) Hurling et al.29 (2007) Kim and Glanz30 (2013) Kirwan et al.31 (2012) Prestwich et al.32 (2009) Prestwich et al.33 (2010) Shapiro et al.34 (2008) Sirriyeh et al.35 (2010) Turner-McGrievy and Tate36 (2011)
7 8 7 9 8 8 6 6 9 9 7 8
2 3 2 2 2 1 0 2 1 0 1 1
3 2 3 3 2 4 6 3 4 4 4 3
3 1 4 4 3 3 3 4 4 4 4 5
15 14 16 19 15 16 15 15 18 17 16 17
58 54 62 69 58 62 58 58 69 65 62 65
reporting (10 items), which assessed whether the information provided in the article was sufficient to allow a reader to make an unbiased assessment of the findings of the study; external validity (3 items), which addressed the extent to which the findings from the study could be generalised to the population from which the study subjects were derived; bias (7 items), which addressed biases in the measurement of the intervention and outcome and confounding (6 items), which addressed bias in the selection of study subjects). The maximum score a study could receive was 27, with higher scores indicating higher quality. The answers were scored as 0 or 1, except for one item in the reporting subscale, which was scored as 0 to 2. Items were rated ‘1’ if the requested information was present in the article and the criteria were met. Items were rated ‘0’ if the requested information was not present in the article and the criteria were not met. Items were also rated ‘0’ if insufficient information was provided. A parallel assessment was conducted regarding the quality criteria for applying PA interventions and mHealth interventions (Table 2). These criteria were derived according to review literature on PA assessments in general and on evaluation methods for behavioural intervention technologies.23,24 All criteria were scored as ‘yes’, ‘no’ or ‘unable to determine’.
Results Methodological quality assessment The full methodological quality scoring of the included studies is presented in Table 1. The mean ± standard deviation (SD) Downs and Black score was 16.08 ± 1.44 (range: 14–19). The mean ± SD of the percentage of the analysed studies was 61.66 ± 4.69. One study met 19 criteria,25 implying a higher methodological quality, while the other studies were rated with various scores.26–36 Regarding the reporting dimension, three studies were rated with 9 points, indicating the highest score,28,33,34 and only two studies25,34 reported the adverse events that may have been a consequence of the intervention. Additionally, four studies25,27,31,32 did not report the number of participants lost to follow-up. Regarding external validity, one study26 obtained the maximum score, and two studies31,34 received 0 points. Regarding internal validity, one study31 scored 6 points, four studies30,33–35 scored 4 points, five studies25,27,28,32,36 scored 3 points and two studies26,29
Downloaded from jhi.sagepub.com at UNIVERSITE LAVAL on March 4, 2015
6
Health Informatics Journal
Table 2. Characteristics of intervention, process and outcome of studies. Study authors
Adapted interventions to subject characteristics
Feedback Based on Support Combination on theoretical tools of PA progress guidelines assessment measures
Objective PA assessment methods
Conroy et al.25 (2011) Cheung et al.26 (2008) David et al.27 (2012) Fjeldsoe et al.28 (2010) Hurling et al.29 (2007) Kim and Glanz30 (2013) Kirwan et al.31 (2012) Prestwich et al.32 (2009) Prestwich et al.33 (2010) Shapiro et al.34 (2008) Sirriyeh et al.35 (2010) Turner-McGrievy and Tate36 (2011)
Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Yes No No No Yes No No No No Yes No Yes
Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
No Yes Yes Yes Yes No No Yes Yes Yes Yes Yes
Yes Yes Yes Yes Yes Yes Yes No No No No Yes
No No No No No Yes No No No No No No
PA: physical activity.
scored 2 points. Only two studies31,35 attempted to blind the study subjects to the interventions received. One study31 blinded those individuals responsible for measuring the main outcome of the intervention. Regarding confounding, one study36 scored 5 points, six studies27,28,32–35 scored 4 points, four studies25,29–31 scored 3 points and one study26 scored 1 point. Furthermore, in 11 studies,25–31,33–36 it was impossible to determine whether the randomised intervention was concealed from both the patients and the healthcare staff until the final recruitment. Regarding the quality criteria that apply to PA interventions and mHealth interventions, the results showed that 12 studies were adapted to the characteristics of the participants,25–36 and four articles25,29,34,36 provided feedback to the participants throughout the given intervention (Table 2). Ten studies used one or more theoretical models to compose the information transmitted to the intervention group.25,27,29,32–36 These models were the Social Cognitive Theory,27,34,36 the Protection Motivation Theory,32,33 the Transtheorical Model,26,27 the Theory of Planned Behaviour,29,35 the Goal Setting Theory27 and the Problem Solving Theory.27 Two of the included studies used a combination of methods to assess PA.29,30 Eight articles25–31,36 used support tools to deliver the interventions (e.g. paper diaries,25 pedometers,26,27,30,31,34 accelerometers,29 interactive voice response,27 face-to face sessions,28 mobile applications31 and podcasts26). Two studies30,31 used a pedometer and a validated questionnaire to assess PA measures. All the studies25–36 used objective methods, such as accelerometers, pedometers or questionnaires, to assess the outcomes.
Review characteristics Characteristics of the study populations and selected studies The characteristics of the selected studies and study populations are described in Table 3. Five of the 12 studies were conducted in the United States,25,27,30,34,36 four were conducted in the United Kingdom,29,32,33,35 two were conducted in Australia28,31 and one was conducted in Hong Kong.26 The resulting articles were published over a 7-year period from 2007 to 2013. The durations of the
Downloaded from jhi.sagepub.com at UNIVERSITE LAVAL on March 4, 2015
United Kingdom
United States
United Kingdom
United Kingdom
United States
United Kingdom
Hurling et al.29 (2007)
Kim and Glanz30 (2013) Kirwan et al.31 (2012) Prestwich et al.32 (2009)
Prestwich et al.33 (2010)
Shapiro et al.34 (2008)
Sirriyeh et al.35 (2010)
Downloaded from jhi.sagepub.com at UNIVERSITE LAVAL on March 4, 2015
PA
48 58
36 (IG: 26, CG: 10) 200 (IG-1: 50, IG-2: 150) 154 (IG-1: 29, IG-2: 31, IG-3: 29, CG-1: 31, CG-2: 34) 134 (IG-1: 40, IG-2: 48, CG: 46)
45 (IG: 30, CG: 15) 200 (IG-1: 50, IG-2: 50) –
31 (IG-1: 13, IG-2: 7, CG: 11) 120 (IG-1: 31, IG-2: 30, IG-3: 31, CG: 28) 86 (IG-1: 44, IG-2: 42)
58 (IG-1: 18, IG-2: 18, CG: 22)
128(IG-1: 32, IG-2: 31, IG-3: 33, C: 32)
96 (IG-1: 49, IG-2: 47)
149 (IG-1: 47, IG-2: 52, CG: 50)
PA, walking
80.5
77 (IG: 47, CG: 30)
77 (IG: 47, CG: 30)
75
70
62
64
66
PA, weight loss
PA
PA
PA, walking
PA, walking
PA
PA
88 (IG: 45, CG: 43)
88 (IG: 45; CG: 43) 100
39 (IG-1: 21, IG-2: 18) PA, walking
PA, weight loss PA
Health behaviour
71 (IG-1: 35, IG-2: 36) 100
78.8
84
Gender (% female)
SD: standard deviation; IG: intervention group; CG: control group; PA: physical activity; BMI: body mass index. The bold values indicate the overall numbers of participants in each study.
Turner-McGrievy United States and Tate36 (2011)
Australia
Australia
Fjeldsoe et al.28 (2010)
United States
Hong Kong
189 (IG-1: 61, IG-2: 64, IG-3: 64) 52 (IG: 38, CG: 14)
210 (IG-1: 72, IG-2: 68, IG-3: 70) 52 (IG: 38, CG: 14)
Conroy et al.25 (2011) Cheung et al.26 (2008) David et al.27 (2012)
United States
Treat to analysis/ participants with complete data
Participants randomised
Participants
Country
Table 3. Study characteristics.
24
2
8
4
4
12
6
9
12
12
6
24
Length (weeks)
IG-1: 43.2 ± 11.7 IG-2: 42.6 ± 10.7
17.3 ± 0.68
8.7 ± 2.3
23.44 ± 5.63
IG: 70.55 ± 7.5 CG: 69.31 ± 7.3 IG-1: 39.3 ± 12.8 IG-2: 40.1 ± 12.1 23.76 ± 4.64
40.4 ± 7.6
30 ± 6
IG: 38.9 ± 10.8 CG: 26.5 ± 1.9 57 ± 5
47.3 ± 8.8
Age (mean ± SD; years)
Inclusion criteria: exercise
