This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in JMIR mhealth and uhealth, is properly cited. The complete bibliographic information, a link to the original publication on http://mhealth.jmir.org/, as well as this copyright and license information must be included.
Mobile technology gives researchers unimagined opportunities to design new interventions to increase physical activity. Unfortunately, it is still unclear which elements are useful to initiate and maintain behavior change.
In this meta-analysis, we investigated randomized controlled trials of physical activity interventions that were delivered via mobile phone. We analyzed which elements contributed to intervention success.
After searching four databases and science networks for eligible studies, we entered 50 studies with N=5997 participants into a random-effects meta-analysis, controlling for baseline group differences. We also calculated meta-regressions with the most frequently used behavior change techniques (behavioral goals, general information, self-monitoring, information on where and when, and instructions on how to) as moderators.
We found a small overall effect of the Hedges
Overall, mHealth interventions to increase physical activity have a small to moderate effect. However, including behavioral goals or self-monitoring can lead to greater intervention success. More research is needed to look at more behavior change techniques and their interactions. Reporting interventions in trial registrations and articles need to be structured and thorough to gain accurate insights. This can be achieved by basing the design or reporting of interventions on taxonomies of behavior change.
People spend much time with their mobile phones (on average 2 hours and 27 minutes per day using apps and the Web [
For researchers who strive to change health-related behavior, mHealth shows excellent promise for building life-changing interventions, but their efficacy is currently uncertain [
One way to describe the content of interventions is to identify the behavior change techniques (BCTs) on which they rely. These are theory-based methods to change psychological determinants of behavior, such as agreeing on a behavioral contract or facilitating social comparison. Until now, it is unclear which BCTs contribute more to mHealth intervention success than others. This is mostly because until recently, only a limited number of studies have been available. Previous reviews and meta-analyses identified n=11 [
In the present meta-analysis, we decided to use a taxonomy by Michie and colleagues [
To find suitable studies, we searched Google scholar and 4 databases: PubMed, PsycINFO, ScienceDirect, and the ISI Web of Knowledge using search terms related to physical activity, mHealth and study design. An example for a search syntax is
We did not restrict inclusion to specific populations. Instead, we accepted all studies that targeted physical activity—be it for healthy populations, sick populations, during pregnancy, or for children (
We identified a total of 2067 studies. After removing duplicates, we screened the remaining 1817 records for eligibility according to their title and abstract. We assessed the full text of 205 studies and from these, 50 met the inclusion criteria [
The intervention was automatically delivered via mobile phone by either an app or texting
The intervention targeted an increase in physical activity
There was a control group, which was more passive than the intervention group and did not have personal communications with medical staff or researchers instead of receiving the mHealth intervention
Allocation to experimental and control groups was randomized, though we accepted stratification (eg, by gender)
At least one of the outcomes measured actual physical activity (via electronic trackers or self-reported) or assessed objective indicators of physical fitness (eg, peak oxygen intake)
Study designs were nonexperimental (eg, observational studies, reports and comments, case studies)
The data necessary to calculate an effect size was not available
Flow diagram of the study selection process. RCT: randomized controlled trial; PA: physical activity; mHealth (mobile health).
In the present meta-analysis, we wanted to investigate how selected interventions influence physical activity; hence we focused on physical activity as outcomes. There are 2 main methods to measure physical activity: having people self-report their physical activity and electronically tracking people’s movement. Self-reported and tracked physical activity generally correlate poorly or moderately with each other, but with a range from
Since our pool of studies includes many pilot and feasibility tests, we decided to account for their small sample size by using the Hedges
The extracted studies were initially screened for eligibility and then articles were classified according to title and abstract. The categories are depicted in the list of “Records excluded” (
To estimate study quality, we used the Effective Public Health Practice Project (EPHPP) Quality Assessment Tool [
Further, some of the studies we included did not (only) target physical activity, but health or weight, with an increase of physical activity being one of multiple intervention goals. We expected that mHealth interventions directly targeting physical activity rather than health or weight would be more efficient in increasing physical activity. Therefore, we assessed the studies’ main objective as a moderator of intervention success.
To identify the factors of mHealth intervention success, we used a taxonomy of 40 BCTs to code intervention contents [
To assure quality in our meta-analysis, we consulted with experts, and followed the recommendations in Borenstein and colleagues [
For the meta-analysis with postintervention group differences and meta-regression we used robumeta, as recommended in a review of meta-analysis packages [
To evaluate a possible publication bias in the field, we created a funnel plot and performed the Egger asymmetry test [
For our meta-analysis, we had a final pool of 50 studies with N=5997 people. The mean age of included samples was 40.6 (SD 16.7), and on average there were 62.7% women in each study (SD 29.2). We found that 40/50 (80%) studies were of good quality, 5 of the 50 (10%) were of moderate quality, and the remaining 10% (5/50) were of poor quality. Twenty-nine of the 50 (58%) studies targeted physical activity, 13/50 (26%) targeted health and 16% (8/50) targeted weight. Furthermore, 20 of the 50 (40%) studies used self-reported measures of physical activity, 17/50 (34%) tracked physical activity, and 13/50 (26%) studies used both methods. We coded a maximum of 4 outcomes per study (
Not all BCTs were used equally often. Therefore, some BCTs are featured in more than half of the studies whereas others are not featured at all. Information for each BCT of the taxonomy is available in
A random-effects meta-analysis with baseline group differences revealed that the estimated effect
Correlating group differences at baseline with group differences postintervention showed a positive association of
We assumed a correlation of
All moderator analyses were conducted using baseline group differences as a covariate. They did not reveal a significant influence of study quality (Δ
Overview of included studies showing the Hedges
Study | N | Baseline, |
Postintervention, |
Abraham et al 2015 [ |
32 | 0.41 | 0.00 (–0.68 to 0.68) |
Adams et al 2013 [ |
20 | –0.77 | 0.44 (–0.41 to 1.29) |
Allen et al 2013 [ |
23 | –0.02 | –0.11 (–0.89 to 0.67) |
Allman-Farinelli et al 2016 [ |
248 | –0.12 | 0.28 (0.03 to 0.52) |
Cadmus-Bertram et al 2015 [ |
51 | –0.10 | 0.13 (–0.31 to 0.57) |
Choi et al 2016 [ |
29 | 0.11 | 0.58 (–0.15 to 1.31) |
Chow et al 2015 [ |
710 | –0.13 | 0.24 (0.04 to 0.44) |
Cotten and Prapavessis 2016 [ |
56 | 0.02 | 0.34 (–0.11 to 0.78) |
Cowdery et al 2015 [ |
39 | –0.34 | 0.17 (–0.45 to 0.79) |
Direito et al 2015 [ |
34 | –0.54 | –0.14 (–0.72 to 0.43) |
Eckerstorfer et al (unpublished data) | 95 | –0.16 | 0.02 (–0.37 to 0.41) |
Fassnacht et al 2015 [ |
45 | –0.31 | 0.00 (–0.59 to 0.59) |
Fjeldsoe et al 2016 [ |
216 | –0.09 | 0.09 (0.56 to 1.25) |
Fjeldsoe et al 2015 [ |
266 | 0.01 | 0.38 (0.09 to 0.68) |
Fjeldsoe et al 2010 [ |
88 | 0.28 | 0.91 (–0.13 to 0.30) |
Frederix et al 2015 [ |
139 | –0.17 | 0.44 (0.14 to 0.73) |
Fukuoka et al 2015 [ |
61 | 0.06 | 0.57 (0.17 to 0.97) |
Garde et al 2015 [ |
47 | –0.21 | –0.07 (–0.55 to 0.42) |
Gell and Wadsworth 2015 [ |
87 | 0.01 | 0.30 (–0.14 to 0.74) |
Glynn et al 2014 [ |
66 | –0.23 | 0.25 (–0.23 to 0.73) |
Hales et al 2016 [ |
43 | –0.07 | 0.00 (–0.59 to 0.59) |
Hartman et al 2016 [ |
50 | 0.39 | 0.43 (–0.08 to 0.93) |
Hebden et al 2014 [ |
51 | 0.05 | 0.01 (–0.47 to 0.49) |
Hurling et al 2007 [ |
77 | 0.17 | 0.36 (–0.08 to 0.80) |
Johnston et al 2016 [ |
151 | 0.07 | 0.07 (–0.23 to 0.36) |
Joseph et al 2015 [ |
28 | –0.04 | 0.17 (–0.41 to 0.75) |
Kim and Glanz 2013 [ |
41 | 0.49 | 1.14 (0.55 to 1.72) |
Kim et al 2015 [ |
196 | –0.10 | 0.14 (–0.14 to 0.42) |
Kim et al 2016 [ |
95 | 0.08 | –0.06 (–0.45 to 0.33) |
Kinnafick et al 2016 [ |
65 | –0.12 | –0.13 (–0.55 to 0.29) |
Laing et al 2014 [ |
211 | –0.12 | 0.23 (–0.05 to 0.51) |
Lubans et al 2016 [ |
157 | 0.30 | 0.14 (–0.16 to 0.43) |
Maddison et al 2015 [ |
143 | 0.05 | 0.25 (–0.02 to 0.52) |
Maher et al 2015 [ |
98 | 0.05 | 0.54 (0.20 to 0.87) |
Martin et al 2015 [ |
32 | 0.00 | 1.35 (0.73 to 1.97) |
Nguyen et al 2013 [ |
84 | 1.30 | 1.88 (1.36 to 2.40) |
Pfaeffli et al 2015 [ |
123 | 0.63 | 0.55 (0.07 to 1.03) |
Poirier et al 2016 [ |
217 | –0.15 | 0.32 (0.04 to 0.60) |
Prestwich et al 2010 [ |
94 | –0.14 | 0.36 (0.02 to 0.69) |
Rubinstein et al 2016 [ |
553 | 0.03 | 0.05 (–0.15 to 0.25) |
Schwerdtfeger et al 2012 [ |
43 | –0.04 | 0.56 (–0.03 to 1.15) |
Silveira et al 2013 [ |
31 | 1.32 | 1.36 (0.70 to 2.02) |
Suggs et al 2013 [ |
158 | 0.17 | 1.18 (0.84 to 1.52) |
Tabak et al 2014 [ |
29 | 0.54 | 1.05 (0.29 to 1.81) |
van der Weegen et al 2015 [ |
117 | –0.25 | 0.43 (0.09 to 0.77) |
van Drongelen et al 2014 [ |
390 | 0.12 | 0.19 (0.02 to 0.35) |
Vorrink et al 2016 [ |
157 | –0.08 | –0.08 (–0.42 to 0.26) |
Walsh et al 2016 [ |
55 | 0.23 | 0.29 (–0.23 to 0.81) |
Wang et al 2016 [ |
59 | 0.30 | –0.22 (–0.64 to 0.20) |
Zach et al 2016 [ |
100 | 0.49 | 0.30 (–0.09 to 0.69) |
aHedges
The number of studies using each tested behavior change technique overall and split for the way in which physical activity was measured.
Behavior change technique | N | Self-reported, n, (%) | Tracked, n (%) | Both, n (%) |
Behavioral goals | 30 | 10 (33) | 8 (27) | 12 (40) |
Self-monitoring | 26 | 5 (19) | 8 (31) | 13 (50) |
General information | 24 | 9 (38) | 11 (46) | 4 (17) |
Information on where and when | 19 | 6 (32) | 11 (58) | 2 (11) |
Instructions on how to | 18 | 4 (22) | 8 (44) | 6 (33) |
Scatterplot of group differences before and after the intervention for each outcome separately. Point size indicates the number of participants in each study. ES: effect size.
Forest plot for physical activity postintervention. The larger the values, the more active the intervention group was compared to the control group. Horizontal lines depict 95% CI and line thickness indicates each number's impact on the summary effect.
Also, we analyzed the effect of the five most frequently used BCTs. Two studies could not be coded for BCTs because they used very diverse interventions and from the text, it was not clear which participants received which intervention components. Thus, we conducted the analyses with a pool of 48 studies, using 84 outcomes. The moderator analyses are visually presented (see
Differences in intervention efficacy depending on the use of the 5 most common behavior change techniques (BCTs). In each panel, “no” means that the BCT was not used and “yes” means that the BCT was used. The “n” next to “yes” and “no” indicates the number of studies in each group. The black line shows the mean intervention efficacy with a 95% CI in white. The curved areas depict density of data points (ie, fine-grained vertical histograms) for all included studies. The last panel shows intervention efficacy depending on a combination of behavioral goals and self-monitoring (n(none)=9, n(either)=22, n(both)=17). These depictions are not controlled for baseline group differences.
Funnel plot to assess publication bias. Point size indicates the number of participants in each study. The dotted and dashed lines show a 95% and 99% credibility region respectively and the full lines represent a 95% CI for the summary effect.
To assess possible publication bias visually, we drew a funnel plot (see
This meta-analysis aimed to identify which BCTs contribute to behavior change in mHealth interventions targeting physical activity. We found an association between heightened efficacy and behavioral goals and the same for self-monitoring. Using either of these techniques pushed the effect from
To our knowledge, this is the most comprehensive assessment of mHealth interventions to increase physical activity so far. Further, this is the first one to address baseline group differences as covariates and to analyze moderating effects of the most frequently used BCTs while retaining sufficient power. However, 34/50 (68%) of all included studies were published in the years 2015 and 2016. This confirms the rapid growth of the mHealth literature concerned with increasing physical activity. To facilitate sequential meta-analyses and to increase transparency, we are sharing the information we coded and the annotated R script in
There are 2 caveats we would like to address. Firstly, we did not look at actual changes within groups over time because the correlation between baseline and postintervention, which is necessary for this analysis, was generally not reported. Instead, we assessed group differences at baseline and postintervention. Using this approach, an increase of physical activity in the intervention group compared to the control group led us to assume intervention success. However, we also assumed intervention success if physical activity in the control group decreased during the intervention while being stable in the intervention group. Of course, the effects we analyzed might also have been a combination of both (ie, an increase of physical activity in the intervention group and a decrease of physical activity in the control group). Secondly, 4 studies in our pool of 50 (8%) studies used clustered randomization methods (for example by school) [
When more studies are available, we will have the statistical power for more sophisticated moderator analyses like meta-CART [
In this meta-analysis, we only looked at BCTs, but of course, there is much more to think about when designing successful mHealth interventions. Intervention efficacy does not only depend on BCTs as intervention components, but also on the target population, the intervention design, and duration, as well as the intervention objectives. For example, participants who suffer from illness might be more motivated to use an mHealth intervention because they suffer more than healthy participants. Additionally, healthy participants might already be quite physically active and would therefore not gain a lot from an mHealth intervention targeting motivation to be physically active. Further, BCTs need to be carefully matched with intervention objectives. For example, when an intervention targets capability for physical activity, BCTs related to self-belief might have a greater impact on intervention success than behavioral goals and self-monitoring.
Despite a small to moderate overall success in increasing physical activity with mHealth interventions, setting behavioral goals or enabling self-monitoring, as well as a combination thereof, might be beneficial. With increasing technological possibilities, interventions will become ever more complex, and it is crucial to report their content thoroughly. However, let us not forget: BCTs are not everything. It is also important that people like and use the interventions. Elements of gamification and appealing visual presentation could be considered to address this issue.
Description of included studies (population, country, duration, and outcomes).
Data spreadsheet and annotated analysis script.
behavior change technique
Effective Public Health Practice Project
effect size
mobile health
physical activity
randomized controlled trial
This project was financed by the Austrian Science Fund (FWF) P 28393. The authors acknowledge the financial support from the University of Graz. Many thanks to Martin Aresin, Justus Baltzer and Robert Spörk for helping to code the studies. We would also like to thank Elizabeth Tipton for a fruitful discussion on how to handle baseline group differences.
None declared.
The authors LVE, NKT, and KC conceived the presented idea. LVE developed the theory, supervised the data collection, performed the computations, and wrote the first draft of the manuscript. NKT verified the analytical methods. KC supervised the findings of this work. All authors discussed the results and contributed to the final manuscript.