International Journal of Medical Informatics 133 (2020) 104029

Contents lists available at ScienceDirect

International Journal of Medical Informatics

journal homepage: www.elsevier.com/locate/ijmedinf

An analysis and evaluation of quality and behavioral change techniques T

among physical activity apps in China
Yan Wanga,b, Yanling Wangb, Brian Greenec, Liu Sunb,*
a
University of Pittsburgh, School of Nursing, Pittsburgh, PA, United States
b
Capital Medical University, School of Nursing, Beijing, China
c
University of Pittsburgh, Asian Studies Center, Pittsburgh, PA, United States

A R T I C LE I N FO A B S T R A C T

Keywords: Background: Physical activity (PA) smartphone applications (apps) featuring at least one behavioral change
Physical activity technique (BCT) are ubiquitous. Although BCTs in PA apps and their quality have been evaluated in Western
Smartphone apps countries, such research in China is new. This study (1) characterizes the extent that BCTs are implemented in
Health Chinese PA apps and (2) evaluates their features and quality.
MARS
Methods: Of 5,253 PA apps identified in five Android app stores in China, 51 top-ranked PA apps with more than
Behavioral change
China
2 million total downloads and user rating above 4 (out of 5) were selected and assessed; their BCTs were scored
using the BCT taxonomy(V1), and app quality was evaluated with the Mobile App Rating Scale (MARS).
Correlations among the number of BCTs, app quality, app features, app downloads, and user ratings were ex-
amined.
Results: Of the top-ranked apps, 62.7% allow behavioral tracking (i.e., pedometer), 9.8% promote Chinese
dancing, and 17.6% offer monetary incentives. The average number of BCT categories included in the apps
analyzed was 11 (SD = 3.51; range = 3–16), with the most common ones comprising feedback and monitoring
(88.2%), goals and planning (82.4%), social support (81.4%), and reward and threat (80.4%). The average
scores for objective and subjective quality of these apps were 3.90 (SD = 0.44) and 2.27 (SD = 0.61), respec-
tively, as assessed by MARS. Moderate positive correlations were observed among the number of BCT in-
corporated, apps quality, app features, and their total downloads.
Conclusions: Chinese PA apps vary in the number of BCTs incorporated, with goals setting, feedback, social
support, and reward being typical. App objective quality is moderate, with relatively low subjective quality.
Generally, higher quality PA apps in China include more BCTs with more app features and downloads. These
findings can guide the development of PA apps to mitigate physical inactivity in China—and elsewhere.

1. Introduction become priorities in China.

Currently, the use of smartphone is ubiquitous worldwide. As of the
Insufficient physical activity (PA) is a modifiable risk factor asso- end of 2017, 753 million Chinese were engaging the Internet via
ciated with numerous non-communicable diseases (NCDs), which ac- smartphones [6], and, as of 2019, the number of smartphone users is an
count for 3.2 million deaths globally each year. Indeed, approximately estimated 2.5 billion worldwide [7]. Additionally, a myriad of mobile
25% adults and 80% adolescents are not active enough physically health (mHealth) technologies (e.g., mHealth applications [apps]) are
worldwide [1–3], and this trend can be observed in China. According to helping improve lifestyles and altering healthcare delivery and as a
the literature, 30.4% of men, 36.8% of women [4], and 80% of ado- result of pervasive attitudes toward individual health management
lescents in China are physically inactive [5]. Because physical inactivity [8,9]. Meanwhile, mHealth apps are being developed at a tremendous
and sedentary lifestyle has become a major public health issue, in- rate. A 2017 report identified a total of 325,000 health-related apps
creasing the PA level of individuals and preventing NCDs both have available on all major app stores [10]. Meanwhile, in addition to the

Abbreviations: PA, physical activity; BCT, behavior change technique; MARS, mobile app rating scale; NCDs, non-communicable diseases; DBCIs, digital behavior
change interventions; BCTTv1, behavior change technique taxonomy v1; GP, goals and planning; FM, feedback and monitoring; SS, social support; RT, reward and
threat
⁎
Corresponding author.
E-mail address: [email protected] (L. Sun).

https://doi.org/10.1016/j.ijmedinf.2019.104029
Received 20 May 2018; Received in revised form 21 October 2019; Accepted 30 October 2019
1386-5056/ © 2019 Elsevier B.V. All rights reserved.
Y. Wang, et al. International Journal of Medical Informatics 133 (2020) 104029

Fig. 1. Flowchart of the app selection process.

sheer volume of mHealth apps and the colossal market for them, studies users with better ways of improving their health through PA, and (3)
have shown that mHealth apps and digital behavioral change inter- examine the relationships between the number of BCTs embedded in a
ventions (DBCIs) have the potential to promote PA [11–15]. For ex- specific PA app and app quality, features, and downloads.
ample, in the UK, a 2016 randomized controlled trial concluded that
apps that provide step counts have gradational effect on the PA of
2. Materials and methods
young men [11]. Indeed, this strategy of providing steps counts in the
apps is a key behavioral change technique (BCT) (i.e., feedback and
2.1. Selection of apps
monitoring).
A BCT is “an observable, replicable and irreducible part of a beha-
Currently, Android smartphones are the most widely used smart-
vioral change intervention; a technique that is an active ingredient
phones in China, with a 74.6% mobile operating system market share
(e.g., self-monitoring) within the intervention” [16]. An international
[27]. Therefore, our sample comprised apps for these Android smart-
research effort developed and established the BCT taxonomy v1
phones. The initial search for apps took place from 25 April to 4 May
(BCTTv1), a theory-based classification system that consists of 93 in-
2019. Using the keywords health, exercise, physical activity, pedometer,
dividual techniques, classified into 16 BCT categories [17]. This tax-
step-count, weight loss, obesity, and fit, written either in Chinese or
onomy can be used to identify and measure the behavioral change ef-
English, apps were searched for on the five major online marketplaces
forts in DBCIs and apps [18–20]. Therefore, we assume that BCTs
for Android smartphones in China: Tencent MyApp (https://android.
present in mHealth apps not only can be assessed with the BCTTv1
myapp.com), 360 Mobile Assistant (http://zhushou.360.cn/), Baidu
[17], but also may be an efficient means for delivering DBCIs to pro-
Mobile Assistant (https://mobile.baidu.com/), the Huawei app store
mote PA.
(http://a.vmall.com/uowap/index.html), and the Vivo app store
With regard to assessing the quality of mHealth apps and BCT ca-
(https://dev.vivo.com.cn/distribute/appStore). These online sources
tegories systematically, a few studies have been conducted in developed
comprise 59% of the Android app market in China [28]. The inclusion
countries, and the results suggest that app quality is associated with app
criteria were as follows: (1) Chinese-language smartphone app; (2) app
features and with the number of BCTs incorporated in it [21–25].
available through all five app stores; (3) app promotes physical activity.
However, in other countries, such as China, the analysis of behavioral
The exclusion criteria were the following: (1) non-English or non-Chi-
content in popular PA apps and their quality in China is very limited in
nese language app; (2) app downloads below 400,000 in any one app
scope. Indeed, Android PA apps in China have not been systematically
store; (3) app total downloads in all five app stores below 2,000,000;
evaluated using tools such as the BCTTv1 [17] and the Mobile App
(4) app average rating for five stores was below 4 (out of 5); (5) PA not
Rating Scale (MARS) [26]. Therefore, in the study described herein,
included in the app. Once the apps were selected (n = 51), the same
BCTTv1 was used to identify BCTs embedded in Chinese PA apps—-
version of each app was installed onto Android smartphones and in-
those BCTs that influence mechanisms that underpin behavior change
dependently analyzed by two authors (YW and LS). Fig. 1 depicts the
of users. Simultaneously, MARS, which has been validated in mental
app selection process.
health apps, weight management apps, and fitness apps [21,22,26], was
used to assess the quality of the Chinese PA apps across five domains:
engagement, functionality, aesthetics, information quality, and sub- 2.2. Assessment of BCT categories
jective quality [26]. These tools provided a qualitative assessment of
not only the ratings of and feedback about the PA apps, but also the The content of a specific app was coded using the BCTTv1 [17].
instruments developed to evaluate the design, usability, and quality of BCTs were identified as either present or absent. The coding process for
the PA apps. each app was completed by two authors (YW and LS) independently.
Due to huge differences in culture, policy, regulations, and en- The number of BCT categories and the frequency of each BCT used in
vironment, the quality and characteristics of PA in China may differ selected apps were measured. The higher the score, the higher the
from those in Western contexts. To address this disparity, this study was presence of BCTs within the app. Codes from each researcher were
conducted to (1) characterize the number and common categories of documented on a standardized form. Additional information, com-
BCTs that have been implemented in smartphone apps targeting PA in prising downloads in the five app stores, user rating and app features
China, (2) evaluate the app quality of these apps in providing Chinese (i.e., behavioral monitoring, semi-automated tracking, instructional
coaching, money or monetary incentives, offer sharing, app

2
Y. Wang, et al. International Journal of Medical Informatics 133 (2020) 104029

community, password, login requirement, reminders, and Internet re- Table 1

quired) also were recorded on the form (for detail, see Supplementary Descriptive Data for the Reviewed Apps.
File 1). N(%) M(SD) Md(IQR) Range

7 7
2.3. Evaluation of app quality Total Downloads 51(100%) 8.65862*10 1.123*10 2.54*106
(3.7187*108) (2942*107) -2.64915*109
Average user 51(100%) 4.35(0.49) 4.42(0.64) 3.10–4.95
To evaluate app quality, we used MARS [26] (for detail, see Sup- rating
plementary File 4), which comprises 23 items distributed over five
categories: engagement, functionality, aesthetics, information quality, MARS app quality rating (1–5)
and subjective quality. Each individual item was measured on a 5-point Engagement 51(100%) 3.90(0.62) 3.90(0.90) 2.3–5.0
Functionality 51(100%) 4.08(0.49) 4.13(0.75) 2.6–5.0
scale (i.e., 1—inadequate to 5—excellent). A mean score was calculated
Aesthetics 51(100%) 3.90(0.53) 4.00(0.67) 25–4.8
for the engagement, functionality, aesthetics, and information domains, Information 51(100%) 3.67(0.36) 3.79(0.50) 2.8–4.2
and an objective app quality score was calculated as an average across quality
these four domains. The subjective quality domain was scored as a Objective quality 51(100%) 3.90(0.44) 3.94(0.69) 2.8–4.5
Subjective quality 51(100%) 2.27(0.61) 2.25(1.25) 1.3–3.8
mean subjective quality score. Two authors (YW and LS) completed the
Evaluation 51(100%) 3.43(0.73) 3.5(0.83) 1.3–4.6
coding independently, and codes from each author were documented criteria
on a standardized form. Additional information, comprising app-spe-
cific evaluation criteria (i.e., awareness, knowledge, attitudes, intention Presence of technical app features
to change, help seeking, and behavior change), also was documented on Allows sharing 49(96.1%) 0.95(0.18) – –
Has app 46(90.2%) 0.89(2.51) – –
the form.
community
Password 26(51.0%) 0.51(0.34) – –
2.4. Intercoder reliability Requires login 49(96.1%) 0.95(0.18) – –
Sends reminder 48(94.1%) 0.94(0.22) – –
Krippendorf’s alphas (Kalpha) were used to check the inter-rater Needs internet to 50(98.0%) 0.98(0.14) – –
work
reliability for the MARS and BCT category coding at the initial stage. Behavioral 50(98.0%) 0.97(0.12) – –
The inter-rater reliability coefficient of Krippendorff’s alpha was cal- monitoring
culated using SPSS to evaluate the agreement between raters. All dis- Semi-automated 38(74.5%) 0.75(0.44) – –
agreements in coding (i.e., if Kalpha < 0.75) were resolved by discus- tracking
Instructional 36(70.6%) 0.71(0.46) – –
sion and confirmed by a senior reviewer (YLW). Reliability estimates
coaching
and correlations are provided in Supplementary File 2 and 3. Offers money and 9(17.6%) 0.18(0.39) – –
monetary
2.5. Analyses incentives

All statistical analyses were conducted using SPSS. Frequencies of

each of the 16 BCT categories implemented in the 51 apps were cal- Table 2
Behavioral Change Technique (BCT) Categories Included in Apps (N = 51).
culated; the total numbers of BCTs present in each app were counted. A
Pearson or Spearman correlation was used to analyze the relationships BCT Taxonomy Grouping Category Number of Apps with BCT %
among total downloads, the number of BCTs, MARS scores, and app
Feedback and monitoring 45 88.2
features.
Social support 42 82.4
Goals and planning 42 81.4
3. Results Reward and threat 41 80.4
Comparison of behavior 40 77.5
Antecedents 40 77.5
3.1. Sample characteristics
Shaping knowledge 39 75.5
Identity 39 76.5
Sample descriptive data are summarized in Table 1.Of the 51 apps Comparison of outcomes 37 71.6
selected that were downloaded more than 2 million times in total, (1) Natural consequences 35 68.6
74.5% allowed semi-automated tracking, including gathering activity Repetition and substitution 34 66.7
Covert learning 34 66.7
information from built-in motion sensor or GPS (e.g., Yidong GPS and
Scheduled consequences 33 63.7
Joyrun) and monitoring heart rate and weight through the phone or Associations 32 62.7
third party devices, such as digital scale or wristband (e.g., Hua Wei Regulation 19 36.3
Exercise and PICOOC), (2) 96.1% provided social interaction (e.g., Self-belief 19 36.3
Xingzhe and Jianyue), (3) 9.8% promoted popular types of Chinese
Note. Number and percentage of apps to use the 16 BCT categories as derived
dancing (e.g, Tangdou and 99 Square Dance), and (4) 17.6% offered
from the BCT taxonomy (v1).
monetary incentives (e.g., Zhaole Pedometer and Weibu Exercise).
15 BCT categories—in fact, Hot Work Out incorporates 16 BCTs. In
3.2. Behavioral change techniques included in apps
addition, some of the BCT categories were clustered in the apps studied.
For example, goals and planning (GP), feedback and monitoring (FM),
The inter-rater reliability for our BCT coding was good, with
and social support (SS) and reward and threat (RT) always appeared
agreement above 75% (see Supplementary File 2). The average number
together in the apps; similarly, the combination BCTs identity, com-
of BCTs in the apps was 11 (range = 3–16; SD = 3.51). Table 2 shows
parison of behavior, antecedents, and shaping knowledge was com-
the frequencies of the 16 BCTs present in the apps. For example, the Le
monly observed among the apps.
Dong Live and Hot Workout apps include features such as interactive
private coaching, step-count, and calories burned. Hao Xiangshou pro-
vides private dietitian and trainer services. These three apps present the 3.3. Evaluation of app quality and app features
highest number of BCTs (i.e., 16). The top four apps with the best MARS
score, Keep, Daily Yoga, and Bohe Health, and Hot Work Out all include Inter-rater reliability estimates for quality were desirable (see

3
Y. Wang, et al. International Journal of Medical Informatics 133 (2020) 104029

Supplementary File 3). The average scores for MARS objective quality assessment in other analyses of mHealth apps (i.e., social support, goal
and its four domains, subjective quality, and number of app features of setting, self-monitoring, and feedback techniques) [21,29,30], and (3)
the PA apps are shown in Table 1. The Keep app scored highest in ob- some BCT categories were clustered in the apps.
jective quality (4.5), information quality (4.2), and app-specific eva- Most of the Chinese apps (i.e., over 82%) possess features of
luation (27.5), while the Zhaole Pedometer app scored the lowest, with tracking behaviors (e.g., step count, calories burned, distance covered,
scores of 2.8, 3, and 8 in these three categories, respectively. The Miao and weight change) and offer not only individual goal and planning
Health app scored the highest in subjective quality (3.75), and the options, but also encouragement for users to follow the goal plan (e.g.,
Weibu Exercise app scored the lowest (1.25) in this category. The Keep options to choose how many miles to cover, how many calories to burn,
app features semi-automated tracking and monitoring, feedback, in- or a personal training plan). The features also include socialization,
structional coaching, app community, and monetary incentives. Con- sending reminders, and offering rewards (i.e., money or congratula-
versely, the Miao Health app possesses other technical app elements tions). These commonly seen features support some techniques (i.e.,
(e.g., analyzing stress level) in addition to the 10 features assessed in FM, GP, SS, and RT), which explains (1) why the BCT categories GP,
our analysis (see Table 1). Meanwhile, the Zhaole Pedometer and Weibu FM, and SS were the most common BCTs in the apps studied and (2)
Exercise apps possess the features of step count and direct money offers why some BCTs tend to be clustered among the apps.
in addition to the basic features of sharing, password, login, notifica- Additionally, studies have shown that interactive and social func-
tions, and network needed. Moreover, four other apps (i.e., Jianshen tions in mHealth apps can better facilitate life style change, PA pro-
Baodian, Hupu Run, Runtastic Pro, and Home Yoga), which exhibited low motion, and better quality of life among users [31,32]. However, these
subjective quality scores (i.e., 1.5), possessed limited app features (e.g., functions among the 51 PA apps were basic, compared to WeChat (i.e.,
Jianshen Baodian possessed only instructional coaching in addition to the most popular social app in China, which features its own pedometer
the basic features of sharing, login, notification, and network needed). and over 1 billion active users [33]). This not only might account for
All categories across the MARS were positively correlated, in a statis- the limited use of such functions among people in China, but also re-
tically significant fashion, with one another (see Table 3). veals some features that could be addressed in future app design.
Moreover, the BCT category reward and threat was common to the top-
3.4. Correlation ranked Chinese PA apps (80%), with nine of these apps offering
monetary incentives. These incentives could explain the huge number
The number of behavior change categories incorporated in each app of downloads of some apps (i.e., Qu Zou, Zhaole Pedometer, Yundong
was associated with (1) the objective quality score (r = 0.423, Zhuan, and Weibu Exercise) that possess limited app features and BCTs.
p < 0.001), (2) each of the domains app subjective quality (r = 0.545, Although Sullivan and Lachman attach great importance to the
p < 0.001), (3) app evaluation criteria (r = 0.518, p < 0.001), (4) the psychosocial-factor-related BCTs (e.g., reconstructing social environ-
number of app features (r = 0.362, p < 0.05), and (5) the app total ment and self-belief), they report that few apps in developed countries
downloads from the five app stores (r = 0.498, p < 0.001). In addition, incorporate these BCTs [34]. In contrast, among the Chinese PA apps
the number of app total downloads was positively correlated with app that we evaluated, approximately 77% include reconstructing the social
objective quality (r = 0.634, p < 0.001) and each of the domains. environment (e.g., advise to spent more time with a gym buddy to boost
Moreover, the number of technical features in the apps was positively the motivation to work out), and over 76% incorporate role modeling
associated with app total downloads, app objective quality, subjective and competition (e.g., users stay on the top of the ranking list and set
quality, and each of the domains. Furthermore, app subjective quality good examples for their children and friends). These differences may
and MARS evaluation criteria were not only positively correlated with result from (1) culture and social development (e.g., elderly people in
each other, but also both associated with the number of app features China tend to socialize with non-family members through activities
and app total downloads. Other relationships were not statistically such as group dancing in public squares [i.e., the term square dance seen
significant (see Table 3). in the app titles below] [35] to promote mental and physical well-
being), (2) physical environment and social environment (i.e., popu-
4. Discussion lation density is high, especially in developed large cities [36]; more-
over, public space and free gyms are available in the community in
4.1. Behavioral change techniques included in apps and app features China [37]), and (3) many Chinese PA app communities are linked with
being ranked in the WeChat app exercise section, which offers users the
Previous research on health-related apps has targeted the analysis of opportunity to compete with friends on WeChat [38]. These differences
their content and the BCTs included, especially for PA apps [24,29]. In also may explain why the group dancing in public square apps included
the study descried herein, we found (1) the average number of BCTs in our analysis (i.e., Tangdou, Jiu Ai Square Dance, Square Dance Duoduo,
included in the 51 apps is higher than those of previous research, (2) 99 Square Dance, and Tiaoba Square Dance) were so popular among
the most common BCT categories implemented in these apps mirror the Chinese female seniors [35].

Table 3
Correlations among App Average Ratings, MARS, Number of BCTs, Total Downloads, and App Features.
Variables 1 2 3 4 5 6 7 8 9 10 11

1. Average user rating 1.00

2. MARS engagement 0.135 1.00
3. MARS functionality 0.204 0.542** 1.00
4. MARS aesthetics 0.287* 0.700** 0.703** 1.00
5. MARS information quality 0.352* 0.705** 0.722** 0.799** 1.00
6. MARS objective quality 0.273 0.856** 0.829** 0.912** 0.903** 1.00
7. MARS subjective quality 0.446** 0.719** 0.606** 0.681** 0.757** 0.806** 1.00
8. MARS evaluation criteria 0.286* 0.808** 0.535** 0.608** 0.602** 0.750** 0.716** 1.00
9. Number of BCTs 0.174 0.449** 0.236 0.374** 0.351* 0.423** 0.545** 0.518** 1.00
10. Number of features 0.210 0.515** 0.350* 0.540** 0.577** 0.585** 0.574** 0.529** 0.362* 1.00
11. Total downloads 0.325* 0.474** 0.534** 0.562** 0.638** 0.634** 0.651** 0.611** 0.498** 0.504** 1.00

Note. MARS mobile app rating scale. **p < 0.01; *p < 0.05.

4
Y. Wang, et al. International Journal of Medical Informatics 133 (2020) 104029

Moreover, we found that some individual techniques have attributes we did not have access to such data, which theoretically could be
that are impractical to be incorporated in DBCIs in actual practice, such provided by either the app developers or an independent third-party
as BCT behavior cost and BCT remote punishment. Therefore, app de- data collection app. Nevertheless, the attitudes of users and their in-
signers and health researchers should consider the limitations of im- tention to change a behavior and repeated use of a given app are the
plementing the BCTs in the development of—and leverage the benefits premise underlying the effectiveness of a given digital behavioral in-
of—mHealth technologies and BCTs as part of the active components in tervention. Therefore, subsequent studies should examine (1) the
developing more complex behavioral interventions. Subsequent studies duration and frequency of app usage among users and (2) the re-
should the establishment of effective BCTs for DBCIs to support and lationship between this usage and the BCTs present in the apps, the
sustain behavioral change. features of the apps, the quality of the apps, and the number of app
downloads to provide guidance for future PA app design. Moreover,
4.2. Quality of apps subsequent studies should take into account our observation of the
combinations of BCTs present in the apps that we examined, which is
We measured PA app objective quality with MARS and included consistent with the literature and suggests the potential effectiveness of
four domains: evaluating engagement, functionality, aesthetics, and the apps [21,24]. Overall, the links among the number of BCTs present,
information quality. We also assessed subjective quality. The objective app features, app quality, and number of app downloads suggests that
quality of 51 top-ranked PA apps available in China is moderate, with app designers should concentrate on providing evidence-based content
information quality scoring lower than the categories of engagement, and incorporating more features that support effective PA techniques to
functionality, and aesthetics. This corresponds with the idea that users not only improve app quality, but also encourage more PA app usage.
tend to choose fun, easy-to-use, functional apps, which offer visual
appeal. The information quality domain achieved the lowest score 5. Conclusion
across the MARS domains, which suggests an overall lack of high-
quality evidence-based content. This finding mirrors the results of other Popular PA apps in China, which overall possess relatively low
studies [21,24]. This low information quality highlights the need for subjective quality and moderate objective quality, score highest in
app designers to invest more in providing better evidence-based in- functionality and lowest in information quality. The number of BCTs
formation from credible sources in addition to addressing app func- that are incorporated in these apps is associated with their quality and
tionality, aesthetics, and engagement. Most of the 51 popular apps se- their number of downloads. Moreover, we observed disparities in the
lected were developed by commercial companies or personal extent to which BCTs were included among these apps, which high-
programmers. We observed little evidence of the involvement of health lights an opportunity for app developers, given the large amount of PA
professionals in the design of the PA apps in China, which could ac- apps available to the Chinese public. Indeed, PA app developers can and
count for the poor information quality of the apps. This result is con- should involve more health professionals, integrate more evidence-
sistent with studies that demonstrate (1) inadequate healthcare pro- based content, and incorporate more effective BCTs in app design in
fessional participation in app design is the reason of the poor credibility China. Furthermore, subsequent studies on the topic should examine
of PA apps [39,40] and (2) the more healthcare professionals are in- not only the association of specific app features and effective BCTs, but
volved in mHealth app development, the more the resultant apps are also which combination of different BCTs in PA apps is more appealing
downloaded [41]. Had more healthcare professionals been involved in to and more effective among Chinese users in encouraging them to
the creation of our sample of 51 top-rated PA apps, the perceived maintain adherence to behavioral change interventions targeting phy-
credibility of the apps among the users might have been greater, which sical activity.
might have improved the users’ adherence to—and the effectiveness
of—the given behavioral intervention. Moreover, this greater perceived Contributions of authors
credibility could have improved subjective quality score among the
users, which was low. In terms of the information quality scores of the YW participated in the design of the study, carried out the statistical
PA apps, we suggest that these scores could be improved through in- analyses, interpreted the data, and drafted the manuscript. YLW was
creased supervision by the government-managed Network Supervision involved in the design of the study and analysis and interpretation of
Department regarding the accountability of app provider to provide a the data. BG co-authored, edited, revised, and provided feedback on
quality product for not only the general public, but also the care pro- several manuscript drafts to facilitate the completion of the final draft
viders who choose to include one of the apps as part of an intervention featured herein. LS was involved in the design of the study, statistical
to promote PA among patients. analyses, and revised the manuscript drafts. All authors read and ap-
proved the manuscript, and all authors agree to be accountable for all
4.3. Correlation aspects of the work in ensuring that questions related to the accuracy or
integrity of any part of the work are appropriated investigated and
Our results reveal the relationship of PA app quality with the pre- resolved.
sence of BCT categories, app features, and total downloads. The better
the subjective and objective quality of a given PA app, the more BCTs Funding
and features are incorporated in the PA app, and the more likely that
the app will be downloaded. Our findings related to app features as- This research was supported by a grant from Foundation for Capital
sessed in this paper (e.g., semi-automated tracking and step count, app Nursing Research, Beijing, China17HL20.
community and chat group, instructional teaching and coaching, and Summary Points
notification and offering financial incentives support some BCTs [i.e.,
FM, GP, SS, and RT]) suggest that offering features that support BCTs
would likely improve not only the user’s perception of the app’s in- What was already known on the topic?
formation quality, but also his or her engagement with and personal
experience of using the app, and this could result in more downloads of • Internationally recognized BCT taxonomy V1 or MARS has
been used to evaluate BCTs in various apps and their quality
the app.
Although it would be reasonable to assume that the number of respectively, including PA apps, in Western countries.
downloads might be associated with the usage of the apps, we do not • The correlation between BCTs, mHealth app objective quality,
and app features in developed countries
advocate such a conclusion. We did not investigate app usage because

5
Y. Wang, et al. International Journal of Medical Informatics 133 (2020) 104029

What this study added to our knowledge interventions, J. Pediatr. Psychol. 40 (4) (2015) 373–384, https://doi.org/10.1093/
jpepsy/jsu108.

• The characterization of BCT categories available in popular

PA apps for Android smartphones in China and a comparison
[15] A.L. Roberts, A. Fisher, L. Smith, M. Heinrich, H. Potts, Digital health behaviour
change interventions targeting physical activity and diet in cancer survivors: a
systematic review and meta-analysis, J. Cancer Survivorship: Res. Pract. 11 (6)
of the differences regarding PA app quality and BCTs in- (2017) 704–719, https://doi.org/10.1007/s11764-017-0632-1.
cluded in the apps between China and Western countries. [16] S. Michie, M. Johnston, Behavior change techniques, in: M.D. Gellman, J.R. Turner
• The nature of some cultural characteristics present among
popular PA apps in China.
(Eds.), Encyclopaedia of Behavioral Medicine, Springer, New York, 2013, pp.
182–187, , https://doi.org/10.1007/978-1-4419-1005-9_1661.

•
[17] S. Michie, M. Richardson, M. Johnston, C. Abraham, J. Francis, W. Hardeman, et al.,
The identification of correlations between the presence of The behavior change technique taxonomy (v1) of 93 hierarchically clustered
BCTs in popular PA apps and their quality, features, and techniques: building an international consensus for the reporting of behavior
downloads; the presentation of recommendations related to change interventions, Ann. Behav. Med. 46 (1) (2013) 81–95, https://doi.org/10.
PA apps and their BCTs to guide advancement of higher 1007/s12160-013-9486-6.
quality and more effective PA apps in China. [18] J. Martin, A. Chater, F. Lorencatto, Effective behaviour change techniques in the
prevention and management of childhood obesity, Int. J. Obes. 37 (10) (2013)
1287–1294, https://doi.org/10.1038/ijo.2013.107.
[19] B. Gardner, L. Smith, F. Lorencatto, M. Hamer, S.J. Biddle, How to reduce sitting
time? A review of behaviour change strategies used in sedentary behaviour re-
Declaration of Competing Interest duction interventions among adults, Health Psychol. Rev. 10 (1) (2016) 89–112,
https://doi.org/10.1080/17437199.2015.1082146.
[20] K. Mercer, M. Li, L. Giangregorio, C. Burns, K. Grindrod, Behavior change techni-
None. ques present in wearable activity trackers: a critical analysis, JMIR mHealth
uHealth 4 (2) (2016) e40, https://doi.org/10.2196/mhealth.4461.
Acknowledgements [21] M. Bardus, S.B. van Beurden, J.R. Smith, C. Abraham, A review and content analysis
of engagement, functionality, aesthetics, information quality, and change techni-
ques in the most popular commercial apps for weight management, Int. J. Behav.
The authors would like to thank Professor Dianxu Ren, University of Nutr. Phys. Act. 13 (2016) 35, https://doi.org/10.1186/s12966-016-0359-9.
Pittsburgh, for supporting the data analysis and statistics, Professor [22] R. Martín Payo, J. Harris, J. Armes, Prescribing fitness apps for people with cancer:
a preliminary assessment of content and quality of commercially available apps, J.
Karen Lamb, Rush University, for her proofreading and meaningful Cancer Survivorship (2019), https://doi.org/10.1007/s11764-019-00760-2.
suggestion, and Huan Fan, PhD, Capital Medical University, for her [23] C.E. Mauch, T.P. Wycherley, R.A. Laws, B.J. Johnson, L.K. Bell, R.K. Golley, Mobile
valuable suggestions. apps to support healthy family food provision: systematic assessment of popular,
commercially available apps, JMIR mHealth uHealth 6 (12) (2018) e11867, ,
https://doi.org/10.2196/11867.
References [24] S. Schoeppe, S. Alley, A.L. Rebar, M. Hayman, N.A. Bray, W. Van Lippevelde, et al.,
Apps to improve diet, physical activity and sedentary behaviour in children and
[1] World Health Organization (WHO), Global Status Report on non-Communicable adolescents: a review of quality, features, and behaviour change techniques, Int. J.
Diseases 2014, WHO, Geneva, 2014 Retrieved from http://apps.who.int/iris/bit- Behav. Nutr. Phys. Act. 14 (2017) 83, https://doi.org/10.1186/s12966-017-
stream/10665/148114/1/9789241564854_eng.pdf?ua=1 (Accessed 18 May 0538-3.
2019). [25] P. Tinschert, R. Jakob, F. Barata, J.N. Kramer, T. Kowatsch, The potential of Mobile
[2] World Health Organization (WHO), February). Physical Inactivity: A Global Public apps for improving asthma self-management: a review of publicly available and
Health Problem, WHO, Geneva, 2018 Retrieved from http://www.who.int/media- Well-adopted asthma apps, JMIR mHealth and uHealth 5 (8) (2017) e113, https://
centre/factsheets/fs385/en/ (Accessed 18 May 2019). doi.org/10.2196/mhealth.7177.
[3] World Health Organization (WHO), Global Health Risks: Mortality and Burden of [26] S.R. Stoyanov, L. Hides, D.J. Kavanagh, O. Zelenko, D. Tjondronegoro, M. Mani,
Disease Attributable to Selected Major Risks, March, Retrieved from http:// Mobile app rating scale: a new tool for assessing the quality of health mobile apps,
www.who.int/healthinfo/global_burden_disease/GlobalHealthRisks_report_full.pdf JMIR mHealth and uHealth 3 (1) (2015) e27, https://doi.org/10.2196/mhealth.
(Accessed 18 May 2019) (2018). 3422.
[4] P. Muntner, D. Gu, R.P. Wildman, J. Chen, W. Qan, P.K. Whelton, J. He, Prevalence [27] Statcounter GlobalStats, Mobiel Operating System Market Share China Apr 2018-
of physical activity among Chinese adults: results from the international colla- Apr 2019, Retrieved from http://gs.statcounter.com/os-market-share/mobile/
borative study of cardiovascular disease in Asia, Am. J. Public Health 95 (9) (2005) china (Accessed 18 May 2019) (2019).
1631–1636, https://doi.org/10.2105/AJPH.2004.044743. [28] iiMedia Research, 2018–2019 China Mobile App Store Market Annual Monitoring
[5] Y. Chen, Z. Zheng, J. Yi, S. Yao, Associations between physical inactivity and se- Report, December 24, Retrieved from https://www.iimedia.cn/c400/63240.html
dentary behaviors among adolescents in 10 cities in China, BMC Public Health 14 (Accessed 18 May 2019) (2018).
(2014) 744, https://doi.org/10.1186/1471-2458-14-744. [29] C.H. Yang, J.P. Maher, D.E. Conroy, Implementation of behavior change techniques
[6] China Internet Network Information Center (CNNIC), Statistical Report on Internet in mobile applications for physical activity, Am. J. Prev. Med. 48 (4) (2015) 452,
Development in China, CNNIC, Beijing, 2018 Retrieved from https://cnnic.com.cn/ https://doi.org/10.1016/j.amepre.2014.10.010.
IDR/ReportDownloads/201807/P020180711391069195909.pdf (Accessed 18 May [30] A. Middelweerd, J.S. Mollee, C.N. van der Wal, J. Brug, S.J. te Velde, Apps to
2019). promote physical activity among adults: a review and content analysis, Int. J.
[7] Statista, Number of Smartphone Users Worldwide from 2014 to 2020 (in Billions), Behav. Nutr. Phys. Act. 11 (2014) 97, https://doi.org/10.1186/s12966-014-
Statista GmbH, Hamburg, 2018 Retrieved from https://www.statista.com/statis- 0097-9.
tics/330695/number-of-smartphone-users-worldwide/ (Accessed 18 May 2019). [31] N. Johnston, J. Bodegard, S. Jerström, J. Åkesson, H. Brorsson, J. Alfredsson, et al.,
[8] L. Sun, Y. Wang, B. Greene, Q. Xiao, C. Jiao, M. Ji, Y. Wu, Facilitators and barriers Effects of interactive patient smartphone support app on drug adherence and life-
to using physical activity smartphone apps among Chinese patients with chronic style changes in myocardial infarction patients: a randomized study, Am. Heart J.
diseases, BMC Med. Inf. Decis. Making 17 (2017) 44, https://doi.org/10.1186/ 178 (2016) 85–94, https://doi.org/10.1016/j.ahj.2016.05.005.
s12911-017-0446-0. [32] R. Nightingale, A. Hall, C. Gelder, S. Friedl, E. Brennan, V. Swallow, Desirable
[9] S. Fox, M. Duggan, Pew Research Center. Mobile Health 2012: Half of Smartphone components for a customized, home-based, digital care-management app for chil-
Owners Use Their Devices to Get Health Information, One-Fifth of Smartphone dren and young people with long-term, chronic conditions: a qualitative explora-
Owners Have Health Apps, Retrieved from http://www.pewinternet.org/2012/11/ tion, J. Med. Internet Res. 19 (7) (2017) e235, https://doi.org/10.2196/jmir.7760.
08/mobile-health-2012/ (Accessed 18 May 2019) (2012). [33] Tencent.com, 2018 WeChat Data Report, Tencent.com, Beijing, 2018 Retrieved
[10] Research2Guidance, 84,000 Health App Publishers in 2017-Newcomers Differ in from https://xw.qq.com/amphtml/20190109A0Q6MJ00 (Accessed 18 May 2019).
Their Go-to-Market Approach, Retrieved from https://research2guidance.com/ [34] A.N. Sullivan, M.E. Lachman, Behavior change with fitness technology in sedentary
84000-health-app-publishers-in-2017/ (Accessed 18 May 2019) (2017). adults: a review of the evidence for increasing physical activity, Front. Public
[11] T. Harries, P. Eslambolchilar, R. Rettie, C. Stride, S. Walton, H.C. van Woerden, Health 4 (2016) 289, https://doi.org/10.3389/fpubh.2016.00289.
Effectiveness of a smartphone app in increasing physical activity amongst male [35] The Straits Times. East Asia, China’S Square-Dancing Aunties Now Have an App to
adults: a randomised controlled trial, BMC Public Health 16 (1) (2016) 925, https:// Help Perfect Their Routines, November, Retrieved from https://www.strait-
doi.org/10.1186/s12889-016-3593–3599. stimes.com/asia/east-asia/chinas-square-dancing-aunties-now-have-an-app-to-
[12] J. Stephens, J. Allen, Mobile phone interventions to increase physical activity and help-perfect-their-routines (Accessed 18 May 2019) (2016).
reduce weight: a systematic review, J. Cardiovasc. Nurs. 28 (4) (2013) 320–329, [36] Paihang bang 123, Population Density Ranking in Cities and Provinces in China,
https://doi.org/10.1097/JCN.0b013e318250a3e7. Feburary, Retrieved from https://web.phb123.com/city/renkou/22253.html
[13] Y. Wang, L. Sun, Y. Xu, Q. Xiao, P. Chang, Y. Wu, Comparision and analysis of top- (Accessed 18 May 2019 (2019).
10 exercise android apps in mainland China, Stud. Health Technol. Inf. 216 (2015) [37] Renming Daily, Sports in China, Retrieved from http://www.xinhuanet.com//
1024. 2017-10/14/c_1121801971.htm (Accessed 18 May 2019) (2017).
[14] E.E. Brannon, C.C. Cushing, A systematic review: Is there an app for that? [38] Baidu.com, How to Use Step Count and Ranking Features in App Wechat?
Translational science of pediatric behavior change for physical activity and dietary November, Retrieved form https://jingyan.baidu.com/article/
b24f6c827c5a6d86bfe5dae4.html (Accessed 18 May 2019) (2018).

6
Y. Wang, et al. International Journal of Medical Informatics 133 (2020) 104029

[39] J. Rivera, A. McPherson, J. Hamilton, C. Birken, M. Coons, S. Iyer, et al., Mobile https://doi.org/10.2196/mhealth.4334.
apps for weight management: a scoping review, JMIR Mhealth Uhealth 4 (3) (2016) [41] N. Pereira-Azevedo, L. Osório, V. Cavadas, A. Fraga, E. Carrasquinho, E. Cardoso de
e87, https://doi.org/10.2196/mhealth.5115. Oliveira, et al., Expert involvement predicts mHealth app downloads: multivariate
[40] J. Chen, J.E. Cade, M. Allman-Farinelli, The most popular smartphone apps for regression analysis of urology apps, JMIR Mhealth Uhealth 4 (3) (2016) e86,
weight loss: a quality assessment, JMIR mHealth uHealth 3 (4) (2015) e104, https://doi.org/10.2196/mhealth.5738.