mobile health care apps for patients in sweden and ... · mobile health care application, often...
TRANSCRIPT
Mobile health care apps for patients in
Sweden and Singapore:
What do physicians recommend?
Author: Yiping Zhang
Master's Programme in Health Informatics
Spring Semester 2014
Degree thesis, 30 Credits
Author: Yiping Zhang
Supervisor: Prof. Dr. Sabine Koch, Department of LIME, Karolinska Institutet
Co supervisor: Prof. Dr. Hock-Hai Teo, Department of IS, National University of Singapore
Dr. Sharon Tan, Department of IS, National University of Singapore
Examiner: Dr. Klas Karlgren, Department of LIME, Karolinska Institutet
2
Master's Programme in Health Informatics Spring Semester 2014 Degree thesis, 30 Credits
Affirmation I hereby affirm that this Master thesis was composed by me, that the work
contained herein is my own except where explicitly stated otherwise in the text.
This work has not been submitted for any other degree or professional
qualification except as specified; nor has it been published.
Stockholm, August 31, 2014
__________________________________________________________
Yiping, Zhang
3
Abstract
Background: Up to now, over 50,000 health apps for mobile devices are
available worldwide. Health apps are considered as innovations that potentially
deliver benefits to patients. However, the quality of health apps is uneven due to
the health apps market is still in its infancy. Physicians are considered as potential
channels to patients by recommending qualified health apps. However, several
factors existing in the real world that create significant effects, which influence
the diffusion of health apps.
Aim: The aim of the study is to capture physicians’ attitudes towards
recommending health apps to patients and to describe factors that influence
physicians to recommend apps or not, taking the specifics of two early adopter
countries, Singapore and Sweden, into account.
Method: This comparative study is exploratory and adopted both deductive and
inductive approaches. It initially constructs a theoretical background including
elements of relevant theories (namely Health app maturity model, “Six hurdles”
model, Diffusion of Innovation) and elements found from literature review. Based
on the theoretical background, a web survey questionnaire is constructed and used
together with follow-up interviews for data collection. Descriptive statistics are
used for analysis of the quantitative and recursive abstraction for analysis of
qualitative data.
Results: A total of 44 physicians in Sweden (SE) and 27 in Singapore (SG)
completed the questionnaire, giving a total study population of 71 physicians.
Only a small group of progressive physicians recommended apps to patients. Most
respondents keep “wait-and-see” attitudes. Driving factors and barriers are
distinctly different between the two countries’ results.
Conclusion: Innovators and early adopters play an important role in the diffusion
of health apps. Interpersonal communication is the most effective way for
physicians gaining information and also motivates them to recommend health
apps to patients. Physicians are seeking health apps with good quality to
demonstrate strong evidence in order to improve their confidences and overcome
barriers.
Key words: mobile health app, physician, Diffusion of Innovation, health
informatics, Sweden, Singapore
4
Acknowledgements
First, I would like to express my gratitude and appreciations to my supervisor
Sabine Koch, who supported me through the whole project with her immense
helps and great insights. This study would have even been possible without her!
I also would like to express my thanks to Prof. Teo and Dr. Tan in department of
Information System of National University of Singapore (NUS) and all colleagues
in Health Informatics Lab of NUS. Thank you for your help and warm welcome! I
appreciate that Karolinksa Institutet exchange program offered me the
scholarship and gave me the opportunity to have the unforgettable experiences in
Singapore.
In addition, I would like to thank all participants helped me spread and answered
the survey questionnaire. I am grateful for the four interviewees that allowed me
to conduct follow-up interviews with them to significantly complement the findings.
Finally, I would like to thank my parents, friends, and colleagues for their
continuous supports that accompany with me every day and forever!
Table of Contents 1. Introduction ...................................................................................................... 1
1.1 Mobile Health Apps ............................................................................... 1
1.2 Growth of health apps ............................................................................ 2
1.3 Theoretical Background ......................................................................... 3
1.3.1 Quality assurance of health apps ................................................. 3
1.3.2 Physicians and health apps .......................................................... 4
1.3.3 Diffusion of Innovation Theory and health apps ......................... 6
1.4 Problem Description .............................................................................. 7
1.5 Research Aim and objectives ................................................................. 8
1.6 Research Questions ................................................................................ 8
2 Method .............................................................................................................. 9
2.1 Research approach ................................................................................. 9
2.2 Study setting ........................................................................................ 10
2.2.1 Swedish and Singaporean health care system ........................... 10
2.3 Participants ........................................................................................... 11
2.3.1 Participant selection .................................................................. 12
2.4 Data collection ..................................................................................... 12
2.4.1 Survey questionnaire ................................................................. 12
2.4.2 Follow-up interview .................................................................. 14
2.5 Data analysis ........................................................................................ 14
2.5.1 Quantitative data analysis .......................................................... 14
2.5.2 Qualitative data analysis ............................................................ 15
2.6 Ethical considerations .......................................................................... 16
3 Results ............................................................................................................ 16
3.1 Findings from questionnaires ............................................................... 16
3.1.1 General information of study populations ................................. 16
3.1.2 Adopter groups .......................................................................... 17
3.1.3 Recommendation decisions groups ........................................... 19
3.1.4 Adopter categories and Recommendation decision groups ...... 20
3.1.5 Different situations and Recommendation decision groups ...... 21
3.1.6 Positive Factors (Motivations and Benefits) ............................. 22
3.1.7 Negative factors (Weaknesses and Barriers) ............................. 26
3.1.8 Quality factors ........................................................................... 29
3.2 Findings from follow-up interviews .................................................... 32
4 Discussion ...................................................................................................... 37
4.1 Main findings from both countries ...................................................... 37
4.1.1 Physicians’ current knowledge and attitudes ............................ 37
4.1.2 Driving factors and Barriers ...................................................... 38
4.2 Different findings from two countries ................................................. 41
4.2.1 Main findings from Swedish physicians ................................... 41
4.2.2 Main findings from Singaporean physicians ............................. 44
4.3 Strength and Limitations ...................................................................... 46
4.4 Future research ..................................................................................... 49
vi
5 Conclusion ...................................................................................................... 49
References ............................................................................................................. 51
Appendix ............................................................................................................... 54
Appendix 1 Data Requirements table ............................................................ 54
Appendix 2 Questionnaire ............................................................................. 56
Appendix 3 statistic from SPSS .................................................................... 66
1. Case summary .................................................................................. 66
2. Adopter groups associated with decision groups ............................. 68
3. Channels of getting information about health apps .......................... 69
4. Health apps categories ...................................................................... 70
5. Patients categories ............................................................................ 71
6. Motivation Factors ........................................................................... 72
7. Positive Factors associated with Recommendation decision groups 73
8. Negative factors (barriers) associated with RA and WRA groups ... 74
9. Quality Factors associated with recommendation decision groups .. 75
Appendix 4 Interview Questions ................................................................... 76
vii
Table 1 Five Adopter Categories According To Diffusion Of Innovation Theory [32] ........... 7
Table 2 Perceived Characteristics Of Innovations [35][36] ...................................................... 7
Table 3 Major Differences Between Swedish And Singaporean Health Care Systems [22] . 11
Table 4 Details Of Participants From Sweden And Singapore ............................................... 11
Table 5 Data Requirements Table (See Details In Appendix 1) ............................................. 14
Table 6 Main Measurement Scales (Adopted According To [50]) ......................................... 15
Table 7 Descriptive Summaries Of General Information........................................................ 17
Table 8 Population Distribution In Different Adopter Groups ............................................... 17
Table 9 Comparison Of SE And SG Adopter Groups’ Mean Ratings .................................... 18
Table 10 Physicians’ Decisions About Recommending Health App To Patients (SE And SG)
........................................................................................................................................ 19
Table 11 Summary Of Responds Faced Different Situations And Top Three Channels ........ 21
Table 12 “Physicians Tested App” Associated With “Patients’ Feedback” (Country=SE) .... 21
Table 13 “Physicians Tested App” Associated With “Patients’ Feedback” (Country=SG) ... 22
Table 14 Summary Of Benefits Factors Categorized Into Two Countries.............................. 24
Table 15 Benefit Factors Associated With RA Groups .......................................................... 24
Table 16 Benefit Factors Associated With WRA Groups ....................................................... 25
Table 17 Benefit Factors Associated With NRA Groups ........................................................ 25
Table 18 Top Three Weaknesses Regarding The Design Of Health Apps ............................. 26
Table 19 Top Three Weaknesses Ranked By RA Groups ...................................................... 26
Table 20 Top Three Weaknesses Ranked By NRA Groups.................................................... 26
Table 21 Barrier Factors Associated To RA And WRA Groups ............................................ 27
Table 22 Barrier Factors Associated With RA Group ............................................................ 28
Table 23 Barrier Factors Associated With WRA Group ......................................................... 29
Table 24 Barrier Factors Associated With NRA Group ......................................................... 29
Table 25 Summary Of Medians Of Quality Factors ............................................................... 30
Table 26 Quality Factors Associated With RA Group ............................................................ 30
Table 27 Quality Factors Associated With WRA Group ........................................................ 31
Table 28 Quality Factors Associated With NRA Group ......................................................... 31
Table 29 Basic Information About Interviewees .................................................................... 32
Table 30 Contributive Factors Of Health Apps Diffusion From Physician’S Perspective ..... 39
Table 31 Contributive Factors Of Health Apps Diffusion From Swedish Physicians’
Perspective ..................................................................................................................... .42
Table 32 Contributive Factors Of Health Apps Diffusion From Singaporean Physicians’
Perspective ...................................................................................................................... 45
Figure 1 Seven major categories of health app functionalities [5] .................................... 1 Figure 2 Existing solutions for ensuring the quality of health apps [9][24-29] ................ 4 Figure 3 IMS’s health app maturity model [5] ................................................................. 5 Figure 4 “Six hurdles” for physicians to recommend health apps to patients ................... 5 Figure 5 Adopter distribution and changes of marketing share of an innovation
according to Diffusion of Innovation [32] ................................................................. 6 Figure 6 Comparisons of DOI adopter groups in sweden and singapore....................... 18 Figure 7 Comparison of distribution of adopter groups in each age level (SE and SG) . 19 Figure 8 Recommendation decision groups distribution in two countries ...................... 20 Figure 9 Adopter distributions in RA, WRA and NRA groups ..................................... 20 Figure 10 Top three MFs in each country and in total populations ................................. 23 Figure 11 Diffusion of health apps and contributive factors from physicians’ perspective
.................................................................................................................................. 37
viii
ABBREVIATIONS
DOI Diffusion of Innovation Theory
HA Healthcare apps/health apps
SE Sweden
SG Singapore
RD Recommendation decision
RA Physicians who recommended health apps to patients
WRA Physicians who will recommend health apps to patients
NRA Physicians who never recommend health apps to patients
PFs Positive Factors
NFs Negative Factors
MFs Motivation Factors
BFs Benefit Factors
QFs Quality Factors
1
1. Introduction
1.1 Mobile Health Apps
Mobile technologies have been ubiquitous and indispensable in our daily life. The spread
of smartphone technologies enable the potential for healthcare delivery. As a sub-segment
of e-Health, mobile health is stemming from the rapid rise of mobile device penetration. It
represents the provision of health care service by using modern mobile information and
communication technology[1], which potentially changes when, where, and how health
care services are provided and further improve the health outcomes.
Mobile health care application, often called mHealth app or health app, which is the
manifestation of mHealth that deliver service to users. Health apps are application
software programs providing mobile solutions for healthcare and prevention, which run
on smartphones, tablet computers and other mobile communication devices [2][3]. They
are available for download in various mobile app digital distribution platforms (i.e. app
stores) and websites such like NHS’s Health App Library. In the early years, health apps
appearing in the world were mainly designed for health care professionals, which aimed
to provide continuing medical education by offering health or medical information
resources and reference[4]. With the progress of technologies evolution and increased
needs from different stakeholders, more and more functionalities are added into health
apps. Currently, there are seven categories of functionalities (Figure 1). A health app can
be equipped with single function or multi-functionalities according to the intended use.
Different apps designs have transformed the mobile health app family into diversity.
Figure 1 Seven major categories of health app functionalities [5]
On one hand, some advanced health apps with serious purposes are called “medical apps”.
Some of them are used as an accessory to a regulated medical device such as viewing a
medical image from picture archiving and communication system (PACS) on a
smartphone or a tablet by running a medical app. Some medical apps can even transform
a mobile platform into a regulated medical device like an Electrocardiogram (ECG) in
•Provide information in a variety of formats (text, photo, video) Inform
•Provide instructions to the user Instruct
•Capture user entered data Record
•Graphically display user entered data/output user entered data Display
•Provide guidance based on user entered information, may further offer a diagnosis, or a consultation with a physician/a course of treatment
Guide
•Provide reminders to the user Remind/Alert
•Provide communication with HCP/patients and/or provide links to social networks
Communicate
2
order to detect the vital signs of patients [2]. Those serious medical apps aim at assisting
health care provider in diagnostics and other clinical purposes within their daily work. On
the other hand, there are large amount of health apps designed for consumers, which are
mostly categorized into “health & fitness apps” in digital app stores. Such kind of
consumer health apps also include rich medical knowledge and references about health
promotion, fitness, nutrition, and commonly have scheduling functions and medical
calculators. Besides that, there are many consumer health apps randomly distributed in
“Health & Fitness” or “Medical” categories in app stores because of their miscellaneous
purposes and features. Most of them are designed for patients or individuals who are
suffering from chronic conditions. Sensors as accessories can be paired with some apps to
detect, collect, and analyze users’ health condition at any time anywhere in order to
improve their self-management abilities[6][7][8]. Physicians can even receive those data
in order to provide real-time assistances to their patients remotely [9][10]. Hence, such
kind of consumer health apps are widely used in supporting chronic disease treatment or
outpatient rehabilitation, for instance, obesity patients can use weight control and calorie
counting apps[11]; hypertension patients are recommended a blood pressure monitoring
app[12]. As a result, patients can stay healthier meanwhile the efficiency of healthcare
resources can be increased significantly.
1.2 Growth of health apps
Up to now, over 50,000 [13] health apps are available worldwide [14] compared to the
number 17,000 in 2010 [15], which also means that individuals located in different
countries will have more options of health apps for download. With the advent of
mHealth, the United States as one major mobile health market has indicated that a long-
expected mobile revolution in healthcare is getting started. According to industry
estimates, the health app market in the United States was worth about 150 million dollars
in 2011, and the number continues growing rapidly [16]. What is more, it is estimated that
about 500 million smartphone users worldwide will download and use at least one health
app by 2015, furthermore, there will be 1.7 billion mobile devices downloading health
apps by 2018 [15]. As increased numbers of healthcare providers join the mHealth market,
the mHealth area is becoming broader to involve healthcare service, sensors, advertising,
and drug sales. According to the results of the Pew Research Internet Project1, 31 percent
of American cell phone owners have used their phone for health care related purpose in
2012, which equals a double growth since 2010 [17]. Meanwhile, the PwC2’s “2012 the
U.S. health market survey” reports that there are 60 percent of consumers believe that
mobile health will significantly change the way they solve health issues in next three
years [18]. Looking around the world, European countries also have noticed the
importance of mHealth. The EU health care system is facing the burden of the prevalence
of chronic disease and the growing ageing population which will increase healthcare
1 Pew Research Center’s Internet & American Life Project aims to be an authoritative source on the evolution of the
internet through surveys that examine how Americans use the internet and how their activities affect their lives. 2 PricewaterhouseCoopers (trading as PwC) is the biggest multinational professional services firm provides industry-
focused assurance, tax and advisory services.
3
demands and raising healthcare costs [19]. As health apps and related services have the
potential to change individuals’ behaviors in order to further improve their self-
management and lifestyle, but also support remote treatment and increase the efficiency
of health care, mHealth is deemed as a cost-efficient solution to support the long-term
sustainability of healthcare delivery. According to the GSMA mobile economy report [19],
mHealth services alone have the potential to save 99 billion euros in healthcare delivery
and increase 93 billion euros to European GDP by 2017. As mHealth solution is showing
its huge potential benefits, some Asian countries also express their interests in adoption of
mHealth to improve quality of life and decrease mortality rate[20].
Compared to the United States, European and Asian countries are also showing their
advantages in mHealth. For instance, Sweden and Singapore have higher smartphone
penetration rates which are ranked as top two among forty two countries reported in a
global survey [21]. Like the United State, Sweden and Singapore have well-known robust
welfare and healthcare systems. As both two counties have put a lot of efforts in
promoting healthy lifestyle in their entire population, their citizens have a high level of
health awareness in general [22]. In addition, researchers have noticed that both countries
have larger numbers of innovators and early adopters who are keen on new technologies
[4]. A high IT literacy exists even in the older population in Sweden [23]. Since different
countries have different cultural backgrounds, the development of health app and its
market trend would be affected in order to satisfy different demands and needs in
different healthcare systems.
1.3 Theoretical Background
1.3.1 Quality assurance of health apps
Health apps are growing at an explosive rising rate. However, the quality of health apps is
uneven since the health app market is still in its infancy. There are significant concerns
about certain amount of health apps are missing clinical practice evidence-based content,
released without clinic trial, lack of protection of data security, and lack of regulations
from governments or market regulators [24]. Since 2011, some solutions and standards
have emerged successively to solve the quality issues and protect patient safety as shown
in Figure 2. Happtique3 launched an in-house Health App Certification Program (HACP)
that aims to help the health app market feel confident in security and functionality but also
to provide a standard that guide developers to follow best practices [25]. People can
download those certified apps from Happtique app formulary website. However, this
certification program has been suspended because of some detected fatal defects of HACP
standards and testing methodologies. Actually, health app certification is not a novelty. In
the United Kingdom, the National Health Service (NHS) has founded their “Health Apps
Library” (HAL) website aiming at discovering, reviewing, and evaluating health apps.
The HAL encourages developers to submit their health apps to the HAL website for
evaluation. Those submitted health apps are evaluated by NHS’s review team to ensure
3 Happtique is a subsidiary of the Greater New York Hospital association’s for-profit arm GNYHA Ventures.
4
they are clinically safe and quality vetting [26]. In addition, “Peer-Review” is also a
popular approach[27] adopted by many health app aggregator websites such as
“iMedicalApps”, and “MedialApp Journal”. Beyond that, the Food and Drug
Administration (FDA) also took action to oversee health apps. The FDA issued a
guideline that regulates health apps which could potentially present risks to patients[9]
[28], which mainly focused on medical apps and mobile medical devices. Nevertheless, in
fact, those aforementioned solutions are little-known by people including patients,
developers, and even health care providers. There is also a lack of studies that exam the
impacts and efficacy of those solutions[29].
Figure 2 Existing solutions for ensuring the quality of health apps [9][24-29]
1.3.2 Physicians and health apps
In February of 2014, Research2Guidance4 launched a web-survey to explore health apps
developers’ opinions in this regard. Based on the survey questions, it is interesting to see
that hospitals, pharmacies and physicians are considered as distribution channels for
health apps by recommending when patients came in for treatment [30]. Indeed, it is
proven on some health IT related online communities where few physicians expressed
they were using mobile devices and medical apps clinically within daily workflow[8][31]
and sometime recommended some suitable health apps to their patients [14]. In 2013,
IMS Institute for Health Informatics had made a contribution to this area by conducting a
huge inductive study in the United States. According to a comprehensive market analysis
and in-depth interviews with physicians, IMS has constructed a health apps maturity
model (figure 3) describing three levels from “recommending of health apps by small
groups of physicians” to “fully integration of health app with healthcare IT systems”. The
current situation for American physicians still rests on the promoting adoption stage,
because there are six identified hurdles (figure 4) which hinder physicians from
recommending health apps to their patients [5].
4 Research2Guidance is an independent mobile industry market research and consultancy provider.
•FDA regulates medical apps and apps which can transform mobile devices to medical devices such as ECG FDA clearance
•Assessing the quality of heath apps through professional judgments made by medical experts Peer-reivew
•Apps are evaluated and rated by users; users shared their experiences to publics. User feedback
•Patients involved in app development process in order to improve user experiences
Patient engagement
•Healthcare institutions or organizations evaluate apps based on certain standards and certify passed apps with an in-house certification
in-house certification
•Apps developed/tested by hospitals based on own needs or certain purposes Hospital-branded
•App released after carried out a proper clinical trial Clinical proven
5
Figure 3 IMS’s Health App Maturity model [5]
As IMS states, it is a long way to pass the six hurdles. There are many challenges and
limitations defined that obstruct the way to fully integration of health apps with healthcare
system. There is still lack of strong evidence to show the efficacy of health apps in
healthcare. Physicians as their patients are also facing the same situation that is hard to
find the right evidence and useful information to make choices. Even if they have found
some useful health apps, they still would consider many issues such as lack of legal
agreement to guide their recommendation behaviors, lack of regulations to control the
quality of health apps, and so on. Therefore, IMS made a significant contribution to
disclose those issues. As this model is not fully generalized yet, it provides opportunities
for researchers in other regions to verify those elements and to complement new findings
in this area.
Figure 4 “Six hurdles” for physicians to recommend health apps to patients [5]
Six Hurdles
to Physicians
to
Recommend
HA to
Patients
Do those apps ensure my
patient’s health data is
secure?
Data
privacey/
security
Are these apps endorsed by
my institution with legal
agreement? Am I liable?
Legal
Issues
Will patients pay or
will insurance
reimburse the cost?
Reimbursement
How do I go about
recommending a health
app to my patient?
Infrastructure
to recommend
How can I find the most relevant
app for my patient? Can I trust
they are good?
Choice/
Evidences
Are there any regulations for
health apps? Are they cleared
by FDA?
Regulations
6
1.3.3 Diffusion of Innovation Theory and health apps
“Diffusion is the process by which an innovation is communicated through certain
channels over time among the members of a social system.” So says E.M. Rogers[32] [33],
who popularizes his Diffusion of Innovation Theory (DOI) to explain how, why and what
rate new ideas and innovations spread through communication channels in social systems
overtime. The crucial components of DOI theory contain the innovation, adopters,
communication and information channels, time, the social system, and the members
including cosmopolite and heterogeneous individuals [34]. Therefore, a diffusion process
consists of five adopter groups (Table 1) which start with a few individuals are called
innovators who are willing to take risks to adopt a new idea and some early adopters who
play the role as opinion leaders, and spread the word later on among their circle of
acquaintances may including early majority, late majority, and laggards who are located
in the subsequent positions along with the blue curve (figure 5). With successive groups
of adopters accepting the new innovation, its market share (shown in yellow) will finally
reach the saturation level.
Figure 5 Adopter distribution and changes of marketing share of an innovation according
to Diffusion of Innovation [32]
Adopter Definition
Innovators Play key role in launching new ideas in a system and willing to take risks, have the highest social status and good financial liquidity, are easy to reach sources and have interactions with other innovators.
Early adopters Individuals have higher social status and good financial liquidity, are more discreet than innovators on checking the performances of innovations before adopting, usually perform as “opinion leadership” in the system to help others adopt innovation.
Early majority Having above average social status and interactions with early adopters. They are cautious and adopt innovations before average members in the system but deliberate for some time. Seldom perform as “opinion leader”.
Late majority Adopting innovations after the average members in the system. They are typically skeptical on innovations, have below average social status and
7
little financial liquidity.
Laggards The last to adopt innovations. Having an aversion to innovations and change agents. They have lowest social status and financial liquidity, usually are oldest among adopters.
Table 1 Five Adopter Categories According to Diffusion of Innovation Theory [32]
Moreover, Rogers et al. [32] emphasizes a social system is a very important component in
a diffusion process, because most people would like to seek out relative advantage of an
innovation from others who have already experienced it in order to understand and reduce
the uncertainty. As Rogers also believe that direct interpersonal communication is the
most effective way of diffusion, it is very important to have the right person to spread the
word to audiences about an innovation. It just likes a progressive physician demonstrates
a new medical device to beginners, or likes an experienced physician recommends a new
medicine or a new treatment plan to a patient. However, different innovations may have
different speeds of spread among all individuals, because there are five perceived
characteristics influencing adopters’ decision making (Table 2). As a result, innovations
that can get high degree of perceptions of all five characteristics are adopted more rapidly.
Health apps as innovations of technology can be initially perceived as uncertain and even
risky. According to DOI, social systems are needed to build communication channels to
enable the diffusion process of health apps. Actually, physicians and patients can
constitute a social system for health apps diffusion. As expertise in medical profession
distinguishes physicians from others, it also makes a significant difference between
physicians and patients. Meanwhile, it builds a trust relationship between two groups [32].
Thus, when people are facing health problems, they will consult physicians and are
willing to accept their recommendations. According to this argument, a physician can
help his or her patient judge a health app from medical perspectives in order to reduce its
uncertainty. As a result, physicians’ attitudes about health apps can significantly influence
their patients. Therefore, it is necessary to investigate physicians’ attitudes about health
apps to understand the diffusion of health apps in this specific social system.
Concepts Definition
Observability The degree to which the results of an innovation are observable or tangible to others
Trialability The degree to which an innovation may be divisible (sharable) for trial
Simplicity The degree to which an innovation is perceived relatively easy to understand and use
Compatibility The degree to which an innovation is perceived compatible with existing values, benefits, past experiences and needs of potential adopters
Relative advantage
The degree to which an innovation is perceived better than the idea or practices it supersedes
Table 2 Perceived characteristics of innovations [35][36]
1.4 Problem Description
Previous research on the diffusion of health apps [5,8,9,14,15,24-31] has shown that
physicians are considered as a potential channel for diffusion of health apps to patients
8
and they are facing many challenges and “six hurdles” [5]. However there exist a number
of knowledge gaps:
Firstly, there are no similar studies conducted in Sweden and Singapore to investigate
physicians’ attitudes towards recommending health apps to patients. Most studies have
been performed in the United States and study results might not easily be transferred to
other countries with different preconditions regarding health systems [22], e-Health
literacy [4,23] and smartphone penetration [21].
Secondly, physicians are considered as a potential channel for diffusion of health apps to
patients [5,8,14,30,31]. These studies have however not investigated physicians’
knowledge about health apps and attitudes in depth. Thus we have little knowledge about
how physicians think about health apps and how they act in reality. There is a lack of
studies if physicians really recommend health apps to patients and what their reasons are
for doing so or not.
Furthermore, we have little knowledge if physicians realize the quality issues of health
apps and how they judge the quality. With the advent of mobile health app quality issues
and new regulations regarding certification of mobile health apps, this knowledge is
crucial for getting a better understanding of the diffusion for health apps.
Hence, a study is needed to disclose the real situation in order to understand how
physicians think about health apps and what drive or hinder their decision making about
recommending health apps to patients, and further to think about what we can help with.
1.5 Research Aim and objectives
The aim is to capture physicians’ attitudes towards recommending health apps to patients
and to describe factors that influence physicians to recommend apps or not, taking the
specifics of two early adopter countries, Singapore and Sweden, into account.
In order to achieve this aim, there are several objectives made:
1) To adopt relevant theories from the literature review to design this study - namely the
Theory of Diffusion of Innovations [32], the IMS Health App Maturity model [5],
and the “six hurdles ” model [5];
2) To explore the attitudes and opinions of physicians of two early adopter countries
(Sweden and Singapore) towards patient adoption of health apps;
3) To analyse and summarize the driving factors and barriers regarding physicians’
recommendations of health apps to patients based on own results and literature
review;
4) To identify physicians’ knowledge about health app quality and regulatory issues in
this regard.
1.6 Research Questions
According to Saunders et al.[37], research questions should be linked to research purposes
which are categorized as exploratory, descriptive and explanatory study. An exploratory
9
study is generally adopted as a valuable means to explore “what is happening” in order to
seek new insights and also to ask questions and to evaluate phenomena in a new light [36]
[37]. Descriptive study is to describe an precise profile of persons, events or situations
[38]. It is necessary to ensure a clear view of the phenomena on which you desire to
collect data. A pure descriptive study is considered as a lack of insight, thus it is
commonly combined with the other two types of study [37]. Explanatory study aims to
establish causal relationship between variables. According to the definitions of research
purposes, research questions are structured as following:
1. What characterizes physicians’ current knowledge of and attitudes towards
health apps?
2. What are the driving factors and barriers that influence health app diffusion
from physicians to their patients?
3. What are the similarities and differences between Sweden’s and Singapore’s
results?
2 Method
2.1 Research approach
Easterby-Smith et al. [39] suggest that choosing an appropriate approach for the research
is important as it will significantly influence the research design, strategies and smooth
constraints. According to Saunders et al., [37] inductive and deductive are the most
commonly adopted research approaches for reasoning in general.
On one hand, a pure deductive reasoning starts with a theory and involves hypothesis
formulation to conduct the development and rigorous test of the theory with empirical
data, which is most often implemented with quantitative methodological choice that
enable concepts and facts to be operationalized and measured [37]. This implies that a
deductive research can generalize statistically about regularities in human social behavior
from sufficient and rigorous selected samples. It holds the problem as a whole by
following the principle of reductionism in order to make things simplest and easy to
understand.
One the other hand, a pure inductive research starts with an observed problem or an
unfamiliar phenomenon. The purpose here is to understand what is going on and have
better knowledge of the nature of the problem. Qualitative data is predominant in the
inductive research, which would be collected through various data collection techniques
such as interviews, observation, and so on. As the inductive study concerns with the
context in which such events is taking place, the study of small sample of objects might
be appropriate compared to a large number as with the deductive approach [37]. The
10
result of this analysis would be built a theory. The inductive study provides a more
flexibilities that allow changes of research rather emphasis than a rigid highly-structured
research design as the deductive study had [37].
This study adopts abduction approach which combined both deduction and induction
approaches [37] that initially adopt a survey strategy to test the elements extracted from
theories, and further to investigate new findings by using interview strategy. In this study,
theoretical background presents in the introduction section is constructed based on
literature reviews, which include IMS’s health app maturity model [5], “six hurdles”
model [5], diffusion of innovation theory [32-36], and other elements from relevant
literature studies [11,41-45]. Research questions were constructed based on problem
description. The theoretical background also provides a basis to generate indicators
formed the Data Requirements Table (Table 5) that is used to ensure the appropriate data
will be collected refer to the relevant research questions. The survey questionnaire is
constructed by the indicators from Data Requirements Table (Table 5) and runs with logic
to take all the possibilities into account. Then mix-method is adopted to analyze collected
data. Quantitative data from survey questionnaire is analyzed in statistic manner by SPSS
software to test and confirm indicators. Qualitative data from interviews is primarily
analyzed through recursive abstractions data analysis technique. The results of this
analysis are used to confirm and develop the adopted theories.
2.2 Study setting
2.2.1 Swedish and Singaporean health care system
As shown in Table 3, Swedish and Singaporean health care systems have many
similarities and differences in term of health care strategies, health care organization
structures, and welfare systems. According to Olsson et al.[22], there are two fundamental
difference. First, Singaporean health care system tends to a free market system, while
Sweden's shows strong public control. Second, the citizens in Singapore have a higher
degree of independence, social security and a healthy system of health promotion.
However, the Swedish welfare system enhances the performance of health care system, as
a result, Swedish citizens have greater guarantee in health care, social security and
rehabilitation.
Sweden (SE) Singapore (SG)
Total number of physicians (2014)
Approximately 37.000 Approximately 12.100
Healthcare strategy
Provision of high quality healthcare Easy access Equality
Provision of a health service that offers services to users at a low cost Heavy state subsides Allowing market forces to improve the efficiency of the providing of services
Healthcare Provider
Public sector, County council , National government,
Independency managed hospital and private hospitals, health centers and
11
municipalities physicians
components Only western medicine Traditional Chinese healers & western medicine
Population Old, aging population, need extra resources support
Young, less resource-intensive
Expenditure Public expenditure Individual pays lots
Other differences
Focus on prevention and early rehabilitation Good system has small side-effects to date
Focus on health promotion Individuals have more responsibilities for their own health Avoidance of additional dependence on the state
Table 3 Major differences between Swedish and Singaporean health care systems [22]
2.3 Participants
This study focuses on the two countries’ physician populations working in hospital and/or
primary care. There were 44 Swedish physicians and 27 Singaporean physicians
contributed to this study, giving a total study population of 71 physicians. Details
regarding each country’s participants are summarized in Table 4.
Sweden (SE) Singapore (SG)
Total number n=44 n=27
Age range a (Average)
43-48 years old 31-36 years old
Gender Female (n=20) Male (n=24)
Female (n=14) Male (n=13)
Workplace b
(percentage with country)
Hospital (17%) Polyclinic (22.6%) Primary care (56.6%) Disease research center (3.8%)
Hospital (29.3%) Polyclinic (31%) Primary care (29.7%) Disease research center (10%)
Working years c
(Average) Over 10 years 6-10 years
Specialty General practitioner/Family medicine (n=38) Cardiologist (n=2) Child Diabetes (n=1) Diabetologist (n=1) Chiropractor (n=1) Pediatrics (n=1)
General practitioner/Family medicine (n=18) General surgery (n=1) Ophthalmology/Allergy (n=2) Palliative medicine/Geriatric Medicine (n=1) Pediatrics (n=2) Psychiatry (n=1) Public Health (n=1) Urology (n=1)
Degree of commutation with patients d
(Average)
Usually Always
a The scale ran from 0 to 7 (0=under 25 years old; 1=25-30 years old; 2=31-36 years old; 3=37-42 years old; 4=43-48
years old; 5= 49-54 years old; 6=55-60 years old; 7=more than 60 years old)
b A physician may have more than one workplace. c
The scale ran from 0 to 3 (0 =Less than 1 year; 1 =1-5 years; 2= 6-10 years; 3=over 10 years.) d
The scale ran from 0 to 5 (0=never; 1= seldom; 2=sometimes; 3= often; 4=Usually; 5= Always)
Table 4 Details of participants from Sweden and Singapore
12
2.3.1 Participant selection
In Sweden, participants were approached through (1) the IT interest group of the Swedish
Association of Family Medicine (SFAM), (2) an article named “Läkares attityd till
hälsoappar undersöks” about the research project published in Dagens Medicin on 7th
of
May, 2014 [40], and (3) adopting snowball sampling through sending emails, embedded
with survey links to several physicians. They further spread the link to their acquaintances
(physicians) in order to reach a larger population. Cluster sampling techniques were
adopted to collect data from Singaporean physicians. Their email address was obtained
from public sources in term of Singapore medical association (SMA) website, 3 hospitals’
websites and 6 polyclinics’ websites.
2.4 Data collection
2.4.1 Survey questionnaire
Surveys strategy was adopted to collect empirical data by sending web-based
questionnaires. As this study is a comparative research which needs participants from two
countries, a web-based questionnaire can easily approach to different populations in a
short time. Thus adopting a web-based questionnaire is suitable in this circumstance. As
suggested by Saunders et al. [37], one way to ensure the collection of important data is to
form a Data Requirements Table (Table 5). The questionnaire (see Appendix 2) is then
developed with the help of the Data Requirements Table (details including question no.
see Appendix 1). As shown in Table 5, adopted theories has presented in the introduction
section are extracted to details corresponding to each question. For instance, participants
are classified into different adopter groups by answering one of the survey questions
regarding their attitudes towards new IT innovation in healthcare, which is designed
according to the definitions of adopter groups of the theory of Diffusion of Innovation
[32]. Survey logic has been well designed to ensure it covers all the possibilities and
direction of data collection. Open questions and follow-up interview requests have been
set in the questionnaire that aims at collect qualitative data for seeking out deeper
understandings. Pilot tests have been done within three weeks by sending the
questionnaire links to seven people including three health informatics experts and four
healthcare professionals to ensure the data validity. According to the feedback from
participants of pilot tests, questions are structured in a natural way and easy to understand.
The language is also revised to decrease misunderstanding. Descriptions are added under
some questions to support the respondents understand the meaning. The survey software
generates a link which can be delivered to physicians via their email address. All data
created by respondents can be recorded in the software in real-time.
Research questions Adopted theory Operational definitions Measures
RQ1: What characterizes physicians’ current knowledge of and attitudes towards health apps?
key elements of Diffusion of Innovation (DOI) [32]: the innovation, its adopters, channels,
[the Innovation] What is the innovation (health apps)
Do physicians know about health apps
[Five Adopters]The physician’s characteristics correspond to five DOI
Attitude to new IT innovation in health care
13
a) Knowledge of
health apps b) Attitudes
towards recommending health apps to patients
social system, and its members; The health app maturity model [5]: recognition of health apps/ systematic use of apps/ fully integration
adopter groups
[social system/channel] If patient asked for health app
[Channel]Information channel about health apps
Where do physicians get the information about health apps?
[Action of diffusion] Interpersonal communication is the most effective way of diffusion [current situation] corresponding to health app maturity model
Do physicians recommend mobile health apps to their patients (Yes, I’ve recommended/ I will recommend/I don’t recommend)
[members] (If recommend=yes/will) Target apps, Target patient groups
Which type of health app is recommended to which patient group?
RQ2: What are the driving factors and hinders influencing the diffusion of health app from physicians to patients? a) Positive Factors:
Motivation factors; Benefits factors
DOI (five characteristic); Benefits defined from literature review: Continuity of care, seamless care, integrated care, Patient-centered care [41] [42] [43] [11] [44] [45]
[Motivation factors] (If recommend=yes/will) some elements are related to [Channels]
[Five characteristic of DOI]& [Benefits]physicians perceived benefits factors and feedbacks
What motivate physicians recommend health apps to patients?
Did physicians test app before they recommended them to patients?
How were the feedbacks from patients?
What are the Benefits of health apps are delivered to patients?
b) Negative Factors (NFs): weakness of design; Barriers of recommending apps to patients
DOI (five characteristic); Six Hurdles model; Other NFs from literature review: Lack of clinical evidence-based practices, Lack of patient engagement, Lack of patients [42] [43] [11] [44] [45]
[Five characteristic ] & [six hurdles][Barriers] (If don’t recommend=yes) What are the Reasons? (If recommend=yes/will) What are the perceived barriers?
What are the barriers regarding recommending health apps to patients? Patients’ motivation; Design of health app; Patient Privacy; Reimbursement; incentives; Endorsement; infrastructure of choosing apps; clinical data integration; personal reason
[Five characteristic ]& [Weakness] (If recommend=yes/will) physicians perceived Drawbacks regarding to health app design
Content, functions, language, service
14
c) Quality factors (QFs)
DOI (five characteristic); QFs from literature review: Quality / Regulation issue of mHealth [11] [43][46] [27] [47][48][49][25][26] [10]
[Five characteristic ]& [Quality factors] Identify Factors that ensure the quality of health app & Rate its importance
What are the factors ensuring the quality of health apps? Clinical proven; FDA; ISO; Peer-review; Endorsement/legal agreement; Hospital-branded; user experience; patient engagement of health app development; Protection of patients’ privacy; Customer services
Table 5 Data requirements Table (see details in appendix 1)
2.4.2 Follow-up interview
As the questionnaire respondents are allowed to leave their contacts for follow-up
interviews, qualitative data are obtained from interviews. The interview questions (see
appendix 4) design is initially based on the responses of questionnaire filled out by each
interviewee. As all interviewees have filled out the questionnaire before conducting
interviews, they can have well-understandings about the research topic and problems.
According to analyzing their feed-forward information from their answered
questionnaires, some general impressions of their experiences and attitudes are depicted.
Some further questions can be generated to investigate their thoughts in depth and also to
get more comprehensive information, thus to complement findings. The steps in preparing
for a follow-up interview are to: (1) analyze research problem and research questions, (2)
analyze interviewee’s feed-forward information based on his/her completed questionnaire;
(3) get a summary of interviewee’s feed-forward information; (4) Comparing to the Data
Requirements Table (Table 5) to understand what information and data really needed to
get from an interviewee (5) draft interview questions; (6) pre-test the draft with a health
informatics expert (supervisor) and two healthcare professionals to check and ensure the
understanding of the interviewee regarding the research problem and interview
questions;(7) prepare the final draft of the interview questions according to suggestions
and feedbacks from pilot-test.
By contacting all interviewees for scheduling appointment, all interviewees preferred
email-interviews. Therefore, all interviews were conducted by sending interview
questions to interviewees and receiving answers in the same manner. As there is no time-
limitation for each email interview, interviewees replied the questions adapted to their
schedules.
2.5 Data analysis
2.5.1 Quantitative data analysis
This study adopts both qualitative and quantitative data analysis procedures. In general,
quantitative data is initially collected from survey questionnaire and the analysis is
15
conducted by using of diagrams and statistics. All data is exported from survey software
in a file format of “sav.”, which then are inputted to licensed SPSS statistic software. All
variables such as options of each question are defined numerically in SPSS statistic
software. Variables are given proper scales of measurement (Table 6). “Compare Means”
and “Crosstabs” functions are applied to get descriptive statistics, to make comparisons,
and also to build associations.
Table 6 main measurement scales (adopted according to [50])
2.5.2 Qualitative data analysis
Qualitative data from essay questions and follow-up interviews are analyzed by primarily
adopting recursive abstraction qualitative data analysis by highlighting, categorizing and
coding key words or relevant phrases. However, in this study, coded findings are not
presented in table manner which differs from other typical recursive abstraction analysis
due to the small amount of qualitative data [51]. As only four follow-up interviews are
conducted through email interview manner, questions and answers are well-structured and
easy to categorize. There is no need to transcribe data due to all data are written in text
form by interviewees. There are four steps in the analysis. Initially, interview questions
are already coded corresponding to different categories in term of channels, motivations,
benefits, barriers, and quality factors according to Data Requirements Table (Table 5);
then, everything of interest and relevant answers are highlighted; Next, Re-arrange some
highlighted data into relevant categories; Next, to code some data within the sentence to
avoid change the meaning of the data and sentence, in order to keep a sense of the
interviewee’s original comment; finally, to revise and refine the findings and coded data.
The whole qualitative analysis is conducted by author and member-checked with
supervisor, and further developed, revised and refined.
In the discussion section, “diffusion of innovation (DOI)” theory is adopted throughout
the whole study. It is initially used for identifying different adopter groups from two
countries’ samples. The second step is to analyze the results in order to explain the health
apps diffusion between physicians and patients. The differences and similarities between
two countries’ respondents are explained combined with DOI. The other theories from
literature studies such as “health app maturity model” and “six hurdles model” were
employed to explain the driven factors and barriers found via the result.
Scale Definition Typical use Measures of averages
Nominal Determination of equality and difference
Classification: e.g. gender, occupations, counties, app categories, channels
mode
Ordinal Determination of greater or less
Ranking: e.g. Age ranges Attitude measures/preference
Median
16
2.6 Ethical considerations
This study would not require an ethical approval since it would not involve any sensitive
personal data or handle any intervention affecting a person. Research data regarding
physicians’ attitude are collected through anonymous web-survey questionnaire. Follow-
up interviews are voluntary and requested by respondents of survey questionnaire, thus
contact information are provided by respondents.
3 Results
3.1 Findings from questionnaires
The findings from questionnaires are predominantly quantitative data. There are six main
aspects are presented with statistics data: 1) general information of study population; 2)
adopter groups and relevant associations; 3) recommendation decision (RD) groups and
relevant associations; 4) positive factors including motivation factors and benefit factors
associated with RD groups; 5) negative factors including weakness and barriers associated
with RD groups; 6) qualities factors associated with RD groups.
3.1.1 General information of study populations
A total of 44 physicians in Sweden (SE) and 27 in Singapore (SG) completed the
questionnaire, giving a total study population of 71 physicians. Descriptive characteristics
of two studies are summarized separately in Table 7. General information of study
populations in term of age levels, working years, and the degree of communication with
patients are calculated in percentages and average value. Based on the result, SE
participants have an older age (3.95) in average compared to the SG participants (2.48).
Moreover, SE participants have a longer average working year (2.50) as physician
compared to SG participants (2.11). However, SG participants have higher average value
(4.11) of communication degree than SE participants (2.82). Overall, both two study
populations have similar age distribution at 43-48 years old. Most participants have over
10 years work experiences as physicians and always fully communicate with their patients.
There are no SG respondents belong to the age level 5 (49-54 years old) neither the age
level 7 (over 60 years old).
Ratios
Age SE (n=44) SG (n=27) Total (n=71)
(L0) Under 25 years old 4.5% 11.1% 7.0%
(L1) 25-30 years old 6.8% 14.8% 9.9%
(L2) 31-36 years old 13.6% 25.9% 18.3%
(L3) 37-42 years old 11.4% 18.5% 14.1%
(L4) 43-48 years old 25% 25.9% 25.4%
(L5) 49-54 years old 15.9% 0% 9.9%
(L6) 55-60 years old 11.4% 3.7% 8.5%
17
Table 7 Descriptive summaries of general information
3.1.2 Adopter groups
The Table 8 and the bar chart (Figure 6) show the descriptive characteristics of defined
adopter groups according to Diffusion of Innovation Theory. On one hand, according to
the results, SE participants have a higher proportion of early adopter compared to SG
participants. On the other hand, even though both countries have same numbers of
innovators, the SG participants have a higher proportion than the SE study population. In
general, most respondents are early adopters and followed by innovator group. There is no
laggard identified by respondents.
Table 8 population distribution in different adopter groups
(L7) 60+ years old 11.4% 0% 7%
Averagea (Media) 4.00 2.00 4.00
Working years
(L0) Less than 1 year 4.5% 7.4% 5.6%
(L2) 1-5 years 9.1% 22.2% 14.1%
(L3) 6-10 years 20.5% 22.2% 21.1%
(L4) Over 10 years 63.6% 48.1% 57.5%
Averageb (Media) 3.00 2.00 3.00
Fully Communication
(L0) Never 2.3% 0% 1.4%
(L1) Seldom 2.3% 7.4% 4.2%
(L2) Sometimes 13.6% 7.4% 11.3%
(L3) Often 13.6% 3.7% 9.9%
(L4) Usually 29.5% 29.6% 29.6%
(L5) Always 38.6% 51.9% 43.7%
Averagec (Media) 4.00 5.00 4.00
a The scale ran from 0 to 7 (0=under 25 years old; 1=25-30 years old; 2=31-36 years old; 3=37-42 years old; 4=43-48 years old; 5= 49-54 years old; 6=55-60 years old; 7=more than 60 years old)
b The scale ran from 0 to 3 (0 =Less than 1 year; 1 =1-5 years; 2= 6-10 years; 3=over 10 years.)
c The scale ran from 0 to 5 (0=never; 1= seldom; 2=sometimes; 3= often; 4=Usually; 5= Always)
Adopter groups SE (n=44) SG (n=27) Total
Innovator 20.5% n=9 33.3% n=9 25.4% n=18
Early adopter 56.8% n=25 22.2% n=6 43.7% n=31
Early majority 13.6% n=6 25.9% n=7 18.3% n=13
Late majority 9.1% n=4 18.5% n=5 12.7% n=9
Laggard 0% n=0 0% n=0 0% n=4
Average (median a) 4.00 4.00 4.00 a
The scan ran from 1 to 5 ( Laggards=1; Late majority=2;Early majority=3;Early adopter=4; Innovator=5)
18
Figure 6 Comparisons of DOI adopter groups in two countries
3.1.2.1 Adopter groups associated with age level and country differences
Table 9 and Figure 7 reveal that the adopters distribution of among different age levels in
two countries. On one hand, according to the observation of SE physician samples, early
adopters exist in all levels of age, especially has a larger numbers in the older age levels.
The 43 years old to 48 years old population have the largest number (n=9) of early
adopters but have no innovators. On the other hand, the SG physician samples have larger
number of early adopters among the populations from 25 years old to 42 years old. The
distributions of innovators are not as Swedish samples which spread almost among all age
groups, but they mostly exist in the younger age populations. The similarity of medians is
found in the age range 43-48 years old adopters. However, the bar chart disclose the
differences that that the SE group has more early adopters but lacked of innovators; The
SG group at the same age level has more early majorities and innovators but lacked of
early adopters. An obvious similarity is found in the bar chart that the 31-36 years old
group is composed of all four adopter categories in both countries.
Table 9 Comparison of SE and SG adopter groups’ mean ratings
Median Ratinga
Attributes Adopters inSE (n=44) Adopters in SG(n=27)
Age Under 25 4.50 (n=2) 5.00 (n=3)
25-30 5.00 (n=3) 3.00 (n=4)
31-36 3.50 (n=6) 4.00 (n=7)
37-42 4.00 (n=5) 4.00 (n=5)
43-48 4.00 (n=11) 3.00 (n=7)
49-54 4.00 (n=7) No input ( n=0)
55-60 4.00 (n=5) 2.00 (n=1)
60+ 4.00 (n=5) No input (n=0)
Country*Age 4.00 4.00 a
The scale ran from 1 to 5 ( Laggards=1; Late majority=2;Early Majority=3;Early adopter=4;Innovator=5)
19
Figure 7 Comparison of distribution of adopter groups in each age level (SE and SG)
3.1.3 Recommendation decisions groups
As data presented in Table 10 and Figure 8, there are 16 out of 44 Swedish physicians
chose “I’ve recommended health apps to my patients” (RA). Correspondingly, there are 5
out of 27 Singaporean physicians made the same decision (RA). More than half of
respondents chose “I will recommend health app to patients” (WRA). Only few
respondents chose “I never recommend” (NRA) option.
However, comparing to Swedish physicians, Singaporean physicians have lager ratios in
WRA and NRA group. The medians for both countries are equal to 1, which are
considered as most respondent chosen “I will recommend” option. As the Swedish RD
group has a larger mean value than the other, it means that more Swedish respondents
recommended health apps than Singaporean respondents; conversely, there is relatively a
larger proportion for Singaporean respondents chose “I never recommend” option
compared to the other.
Recommendation Decisions groups (RD)
SE(n=44) SG(n=27) Total (n=71)
Recommended (RA) 36.7% (n=16) 18.5% (n=5) 29.6% (n=21)
Will Recommend (WRA) 56.8% (n=25) 59.3% (n=16) 57.7% (n=41)
Never Recommend (NRA) 6.8% (n=3) 22.2% (n=6) 12.7% (n=9)
Median a 1.00 1.00 1.00 a The scale ran from 0 to 2 (0=NRA; 1=WRA; 2=RA)
Table 10 Physicians’ decisions about recommending health app to patients (SE and SG)
20
Figure 8 Recommendation decision groups distribution in two countries
3.1.4 Adopter categories and Recommendation decision groups
Firstly, we investigate the relationships between adopters and three recommendation
decision groups. As shown in Figure 9, most physicians recommended health apps to
patients are innovators and early adopters. Compared to Singaporean RA group, the
Swedish RA group has a larger number of early adopters than innovator. However, in the
WRA groups, there is a different situation that more Swedish early adopters compared to
Singaporean sample would like to recommend health apps in the future. (Details of
statistics see Appendix 3.2)
Figure 9 Adopter distributions in RA, WRA and NRA groups
21
3.1.5 Different situations and Recommendation decision groups
Table 11 presents the number of respondent corresponded to each situation. In general,
95.5% of Swedish physicians and 92.6% of Singaporean physicians know health apps
(HA), giving a total 90.1% of respondents know HA through several channels. The SE
groups have a lager ratio than the SG groups that their respondents are using HA
themselves. Moreover, the SE groups also have a larger ratio (59.1%) at the situation of “I
was asked by patients about HA” compared to SG group’s ratio (37%). The following
sections will present the relations between different situations and three recommendation
groups separately.
SE (n=44) SG (n=27) Total (n=71)
Know HA 42 (95.5%) 25 (92.6%) 67 (90.1%)
Use HA 34 (77.2%) 18 (66.7%) 52 (73.2%)
Asked by patients
26 (59.1%) 10 (37%) 36 (50.7%)
Top three Channels of getting information a
Mobile application stores (n=34) Patients (n=21) Colleagues (n=20)
Mobile application stores (n=23) Search engines (n=11) Health app websites (n=10)
Mobile application stores (n=57) Search engines (n=30) Patients (n=29)
Top three HA categories b
Physical training (n=32) Health lifestyle (n=30) Cognitive training (n=27)
Physical training (n=20) Health lifestyle (n=18) Healthcare provider locator (n=15)
Physical training (n=52) Health lifestyle (n=48) Cognitive training (n=38)
Top three target patient categories c
Chronic conditions (n=35) Unhealthy lifestyles (n=33) Cognitive problems (n=25)
Unhealthy lifestyles (n=20) Chronic conditions (n=19) Emergency situation (n=17) Remote assistance (n=17)
Chronic conditions (n=54) Unhealthy lifestyles (n=53) Cognitive problems (n=37)
a statistic of channels can be found in Appendix 3.3
b statistic of HA categories can be found in Appendix 3.4
c statistic of patients categories can be found in Appendix 3.5
Table 11 Summary of responds faced different situations and top three channels
1) Physicians who recommended apps (RA)
For Swedish RA group, all 16 respondents know HA via mobile application stores (n=14),
patients (n=10), and colleagues (n=10) as their major channels. There are 15 respondents
used HA themselves and asked by patients about HA. There are 11 out of 16 respondents
test each HA before they recommended to patients. The median of feedback about
recommended health apps they received from patients is 3.00 which means the overall
feedback from patients is good. The relations between “Physician tested app” and
“Patients’ feedback” in Swedish RA samples are presented in Table 12.
Feedback from patients Total number of respondents I don’t know Good Very good
Test Apps few n=1 n=0 n=0 n=1
most n=0 n=4 n=0 n=4
each n=4 n=4 n=3 n=11
Table 12 “Physicians tested app” associated with “Patients’ feedback” (Country=SE)
22
For Singaporean RA group, all 5 respondents in Singapore RA group used HA
themselves. Mobile application stores (n=5), health app websites (n=4), patients and
colleagues (n=4) are their main channels to get information about HA. They were also
asked by patients about HA. All 5 respondents tested each HA before they recommended
to patients. The median value of patient’s feedback is 3.00, which means the overall
feedback from patients is good. The relations between “Physician tested app” and
“patients’ feedback” in Singapore RA samples were presented in Table 13.
Feedback from patients Total number of
respondents I don’t know Good Very good
Test Apps few n=0 n=0 n=0 n=0
most n=0 n=0 n=0 n=0
each n=1 n=3 n=1 n=5
Table 13 “Physicians tested app” associated with “Patients’ feedback” (Country=SG)
2) Physicians who will recommend apps (WRA)
For Swedish WRA group, 23 out of total 25 respondents know HA via mobile application
stores (n=18), search engines (n=11), and patients (n=10) as their main channels. There
are 17 respondents used HA themselves and 11 respondents asked by patients about HA.
For Singaporean WRA group, 15 of total 16 respondents group know HA via mobile
application stores (n=14), search engines (n=7), health app websites (n=5), and
Scientific journals (n=5) as their main channels. There are 11 respondents used HA
themselves. Only 4 people were asked by patients about HA.
3) Physicians who never recommend apps (NRA)
For Swedish NRA group, all 3 respondents know HA via the same channels as the
Swedish RA group. There are 2 out of 3 respondents used HA themselves but none
respondents were asked by patients about HA.
For Singaporean NRA group, 5 of total 6 respondents group know HA via mobile
application stores. Only 1 respondent used HA. There are 2 physicians asked by patients
about HA.
3.1.6 Positive Factors (Motivations and Benefits)
The positive factors include motivation factors (MFs) and benefit factors (BFs). The
following sections describe the results of MFs and BFs which rated by respondents in
different recommendation decision groups in term of RA, WRA, and NRA groups.
3.1.6.1 Motivation factors (MFs) and Recommendation Decision Groups
As shown in the bar chart (Figure 10), in general, the top three factors motivated
respondents to recommend HA to patients, are “I use health apps myself”(MF5), “My
patients are interested in health apps” (MF1), and “patients recommend health apps to
me” (MF2). However, compared the details in country level, there are some difference
regarding sequence of MF5 and MF1. In SE, MF1 shows a slight advantage as the top
23
motivation factor compared to MF5. Conversely, in SG, MF5 has a higher rating than
MF1. The following sections present the relations between three RD groups and MFs
separately (details of statics see appendix 3.6).
Figure 10 Top three MFs in each country and in total populations
1) Physicians who recommended apps (RA) associated with MFs
For Swedish RA group, the top three motivation factors (MF) are chosen by RA group
were “patients are interested in HA” (n=13), “I use HA myself” (n=13), and “patients
recommended HA to me” (n=11).
For Singaporean RA group, the top three MFs are “patients are interested in HA” (n=5),
“I use HA myself” (n=5), and “patients recommended HA to me” (n=5).
2) Physicians who will recommend apps (WRA) associated with MFs
For Swedish WRA group, the top three MFs are “patients are interested in HA” (n=15),
“I use HA myself” (n=13), and “Colleagues recommended HA to me” (n=9).
For Singaporean WRA group, the top three MFs are “I use HA myself” (n=13)” “patients
are interested in HA” (n=9), and “Colleagues recommended HA to me” (n=5).
3) Physicians who never recommend apps (NRA) group associated with MFs
NRA group could not trigger the questionnaire question regarding motivations factors of
recommending HA to patients due to the survey logic. Therefore, no data is presented in
this group.
3.1.6.2 Benefit Factors (BFs) and Recommendation Decision Groups
As shown in Table 14, the maximum median is 4.00 at “Self-management”, thus this PF
is rated as “Strongly Agree”. The minimum median is 2.50 at “Improve clinical data
24
quality”, thus this PF is rated as “Disagree”. Some differences between two countries are
found at “Disease prevention” and “free/low charge”, which present most Singaporean
respondents rated those PFs as “Strongly Agree”, but most Swedish physicians rated them
as “Agree”. The Benefit Factors is also built associations with recommendation decisions
groups which are presented as following sub-sections (details of statistics see Appendix
3.7).
Self-management
Continuity of care
Patient-centered care
Disease Prevention
Clinical data quality
SE 4.00 3.00 3.00 3.00 2.50
SG 4.00 3.00 3.00 3.00 2.50
Total 4.00 3.00 3.00 3.00 2.50
Remote Medical support
Patient encounter
Save operational costs
Free/low charge
Treatment plan
SE 3.00 3.00 3.00 3.00 3.00
SG 4.00 3.00 3.00 4.00 3.00
Total 3.00 3.00 3.00 3.00 3.00 The scale ran from 1 to 4 (1= strongly disagree; 2=disagree; 3= agree; 4=strongly agree)
Table 14 Summary of Benefits Factors categorized into two countries
1) Physicians who recommended apps (RA) associated with Benefit Factors
Table 15 shows the results of benefits factors which rated by respondents who
recommended apps to patients. The similarities are highlighted in the table.
RA groups SE n=16 SG n=5 Strongly Agree (median =4.00)
Improve patient self-management ability
Improve Patient self-management ability;
Improve Continuity of care; Improve patient-centered care; Improve Prevention and health
education; Improve treatment plan; Remote medical support; Efficient encounter; Patients can use health app for free/low
charge
Agree (median =3.00)
Improve Clinical data Quality Save operational costs Improve Continuity of care; Improve Patient-centered care; Improve Prevention and health
education; Improve Treatment plan; Improve Remote medical support; Efficient encounter; Patients can use health app for
free/low charge
Improve Clinical data Quality Save operational costs
Disagree (median =2.00)
none none
Strongly Disagree (median=1.00)
none none
Table 15 Benefit factors associated with RA groups
25
2) Physicians who will recommend apps (WRA) associated with Benefit Factors
Table 16 shows the results of benefits factors which rated by respondents who will
recommend health apps to patients. The similarities are highlighted in the table.
WRA groups SE n=25 SG n=16 Strongly Agree (Median=4.00)
Improve patient self-management ability;
Improve Prevention and health education;
Improve Patient self-management ability;
Improve Prevention and health education;
Remote medical support; Efficient encounter; Patients can use health app for free/low
charge
Agree (median=3.00)
Save operational costs Improve Continuity of care; Improve Patient-centered care; Improve Treatment plan; Improve Remote medical support; Efficient encounter; Patients can use health app for
free/low charge
Save operational costs Improve Continuity of care; Improve patient-centered care; Improve treatment plan; Improve Clinical data Quality
Disagree (median=2.00)
Improve Clinical data Quality none
Strongly Disagree (median=1.00)
none none
Table 16 Benefit factors associated with WRA groups
3) Physicians who never recommend apps (NRA) associated with BFs
Table 17 shows the results of benefits factors which rated by respondents who never
recommended health apps to patients. The similarities are highlighted in the table.
Table 17 Benefit factors associated with NRA groups
NRA groups SE n=3 SG n=6 Strongly Agree (Median =4.00)
none none
Agree (median =3.00)
Patients can use health app for free/low charge
Improve Patient self-management ability;
Improve Patient-centered care; Improve Prevention and health
education; Improve Treatment plan; Improve Remote medical support; Efficient encounter; Improve Clinical data Quality
Patients can use health app for free/low charge
Disagree (median =2.00)
Improve Continuity of care; Save operational costs
Improve Patient self-management ability; Improve Remote medical support;
Strongly Disagree (median=1.00)
none Improve Continuity of care; Improve Patient-centered care; Improve Prevention and health education; Improve Treatment plan; Efficient encounter; Save operational costs
26
3.1.7 Negative factors (Weaknesses and Barriers)
Negative Factors include weaknesses of the design of health apps (W) and barriers
factors (BFs) that influence physicians’ decision making on recommending HA to their
patients. The results are shown in the following sections.
3.1.7.1 Weaknesses of the design of health apps
All respondents ranked the top three weaknesses among seven given options while
answering questionnaires. According to the statistics, both Swedish and Singaporean
physicians made the same ranking orders as shown in Table 18. However, for different
recommendation decision groups, they have different results.
Rank Weakness
Top three weaknesses
1st W1: Lack of evidence-based content
2nd W2: Lack of personalized design
3rd W3: Unclear instruction/information
Table 18 Top three weaknesses regarding the design of health apps
1) Physicians who recommended apps (RA) associated with Weaknesses
As shown in Table 18, there was one difference marked in Swedish RA group. “Lack of
multi-language support” (W4) was considered as the second weakness. There was
nothing changed in Singaporean RA groups.
Rank SE SG
Top three weaknesses
1st W1: Lack of evidence-based content W1:Lack of evidence-based content
2nd W4: Lack of multi-language support W2:Lack of personalized design
3rd W3:Unclear instruction/information W3:Unclear instruction/information
Table 19 Top three weaknesses ranked by RA groups
2) Physicians who will recommend apps (WRA) associated with Weaknesses
There was nothing changed in WRA groups. The results were as same as the data
presented in Table 18.
3) Physicians who don’t recommend apps (NRA) associated with Weaknesses
In NRA groups, there are several changes (Table 20). Firstly, for Swedish NRA group,
“Lack of multi-functionality” (W5) was ranked by respondents as the third considerable
weakness compared to the results in Table 18. Second, for Singaporean NRA group, the
elements are same, but orders were different. W2 was ranked as the most considerable
weakness. Conversely, W1 was ranked in the third position, and meanwhile, W3 was
raised to the second position.
Rank SE SG
Top three weaknesses
1st W1: Lack of evidence-based content W2:Lack of personalized design
2nd W2: Lack of personalized design W3:Unclear instruction/information
3rd W5: Lack of multi-functionality W1:Lack of evidence-based content
Table 20 Top three weaknesses ranked by NRA groups
27
3.1.7.2 Barriers Factors and Recommendation Decision Groups
This section is not given a summary for the whole recommendation groups due to
different questions were triggered by different recommendation decision groups in the
questionnaire. As “physicians who recommended apps” (RA) and “physicians who will
recommend apps” (WRA) were given the same questions about NFs regard barriers of
recommending HA to patients, the following Table 21 presents the summary of mean
values in these two groups in different countries (details of statistics see Appendix 3.8).
In total, both countries have their identical maximums (median=4.00) at “HA lacks
integration with clinical data”, thus this NF were rated as “Strongly Agree” by both
countries’ respondents. The minimum is 2.00 at “Too time-consuming to recommend
HA to patients”, thus this NFs is rated as “Disagree”. None of them were rated as
“Strongly Disagree”. An obvious difference is found at “Data overload if HA integrated
with clinical data” that Swedish respondents rated it as “Disagree”, but Singaporean
respondents rated it as “Agree”.
Patients Lack motivations
Physicians lack incentives
Poor quality and design
Poor protection of patient Information
Too time consuming to recommend
Difficult to find specific one fit patient
SE 3.00 3.00 3.00 3.00 2.00 3.00
SG 4.00 3.00 3.00 3.00 2.00 3.00
Total 3.00 3.00 3.00 3.00 2.00 3.00
Lack of endorsement
Lack of surveillance From regulators
Lack of reimbursement
Lack of integration with clinical data
Data overload if integrated
Increased workload if integrated
SE 3.00 3.00 3.00 4.00 2.00 3.00
SG 3.00 4.00 4.00 4.00 3.00 3.00
Total 3.00 4.00 3.00 4.00 3.00 3.00 The scale ran from 1 to 4 (1= strongly disagree; 2=disagree; 3= agree; 4=strongly agree)
Table 21 Barrier factors associated to RA and WRA groups
1) Physicians who recommended apps (RA) associated to barrier factors
The following Table 22 shows the barriers which rated by respondents who recommended
health apps to their patients. The similarities are highlighted in the table.
RA groups SE n=16 SG n=5 Strongly Agree (median=4.00)
HA lacks integration with clinical data
HA lacks integration with clinical data Patients lack motivations to use HA; Physicians lack incentives to recommend
HA; HA has poor protection of patients’ health
data; HA Lack of endorsements from healthcare
institutions; HA lack reimbursement standards; HA lack surveillance by government or
28
regulators; Data overload if HA integrated with
clinical data; Increase workload if HA integrated with
clinical data
Agree (median=3.00)
HA has poor design and poor quality;
Patients lack motivations to use HA; Physicians lack incentives to
recommend HA; HA has poor protection of patients’
health data; Difficult to find HA fit patients; HA Lack of endorsements from
healthcare institutions; HA lack reimbursement standards; HA lack surveillance by government
or regulators;
HA has poor design and poor quality;
Disagree (median=2.00)
Too time-consuming to recommend HA to patients;
Data overload if HA integrated with clinical data;
Increase workload if HA integrated with clinical data
Difficult to find HA fit patients;
Strongly Disagree (median=1.00)
none Too time-consuming to recommend HA to patients;
Table 22 Barrier factors associated with RA group
2) Physicians who will recommend apps (WRA) associated to barrier factors
The following Table 23 shows the barriers which rated by respondents who will
recommend health apps to their patients. The similarities are highlighted in the table.
WRA groups SE n=25 SG n=16 Strongly Agree (Median=4.00)
HA lack surveillance by government or regulators;
HA lacks integration with clinical data
HA lack surveillance by government or regulators;
HA lacks integration with clinical data
Physicians lack incentives to recommend HA;
Agree (median=3.00)
HA has poor design and poor quality; Patients lack motivations to use HA; HA has poor protection of patients’
health data; Difficult to find HA fit patients; HA Lack of endorsements from
healthcare institutions; HA lack reimbursement standards; Increase workload if HA integrated
with clinical data Physicians lack incentives to
recommend HA;
HA has poor design and poor quality; Patients lack motivations to use HA; HA has poor protection of patients’
health data; Difficult to find HA fit patients; HA Lack of endorsements from
healthcare institutions; HA lack reimbursement standards; Increase workload if HA integrated
with clinical data Data overload if HA integrated with
clinical data
Disagree (median=2.00)
Too time-consuming to recommend HA to patients;
Data overload if HA integrated with
Too time-consuming to recommend HA to patients;
29
clinical data;
Strongly Disagree (median=1.00)
none none
Table 23 Barrier factors associated with WRA group
3) Physicians who never recommend apps (NRA) associated to barrier factors
The following Table 24 shows the barriers which rated by respondents who never
recommend health apps to their patients. The similarities are highlighted in the table.
NRA groups SE n=3 SG n=6 Strongly Agree (Median=4.00)
It just did not occur to me to recommend HA;
HA has poor quality and design; HA has poor protection of patient
health data; HA lack surveillance by government
or regulators; HA lacks integration with clinical
data
It just did not occur to me to recommend HA;
HA has poor quality and design; HA has poor protection of patient
health data; HA lack surveillance by government or
regulators; HA lacks integration with clinical data; Can’t find specific app fit patient; Physicians lack incentives to
recommend apps HA lack endorsements from healthcare
institutions Lack of reimbursement standard
Agree (median=3.00)
Patients lack motivations to use HA Physicians lack incentives to
recommend apps HA lack endorsements from
healthcare institutions Can’t find specific app fit patient; Lack of reimbursement standard
Patients lack motivations to use HA Too time-consuming to recommend HA
to patients Increased workload if HA integrated
with clinical data
Disagree (median=2.00)
Data overload if HA integrated with clinical data
Too time-consuming to recommend HA to patients
Increased workload if HA integrated with clinical data
Data overload if HA integrated with clinical data
Strongly Disagree (median=1.00)
none none
Table 24 Barrier factors associated with NRA group
3.1.8 Quality factors
Table 25 shows the summary of all quality factors (QFs) with median separately in two
countries. Median was calculated to compare the levels of importance. As shown, both
countries considered these five factors in term of “clinically proven”, “Peer-reviewed”,
“Good user feedback”, “good protection of patient privacy, confidentiality”, and “good
customer service” as very important factors that guaranteed the quality of health app.
Conversely, “Top ranking in app stores” was considered as less important factor.
Moreover, “Developed by IT companies” was considered as unimportant factor to
30
influence the quality of health apps by both countries’ respondents. There is an obvious
differences found at “hospital-banded”. Most Swedish respondents considered this factor
as a less important QF, however, Singaporean respondents considered it as an important
QF. The PFs is also built associations with recommendation decisions groups which will
be presented as following (More detailed statistics were presented in Appendix 3.9)
Clinically proven (CP)
FDA Hospital-branded
ISO Endorsement Peer-review
SE 4.00 3.00 2.00 3.00 3.00 4.00
SG 4.00 4.00 3.00 4.00 4.00 4.00
Total 4.00 4.00 2.00 4.00 3.00 4.00 User
Feedback Top ranking Patients
engagement (PE)
Confidentiality, security, privacy (CSP)
Customer services
Developed by IT companies
SE 4.00 2.00 3.00 4.00 4.00 1.00
SG 4.00 2.00 4.00 4.00 4.00 1.00
Total 4.00 2.00 3.00 4.00 4.00 1.00
The scale ran from 1 to 4 ( 1=unimportant; 2=less important; 3=important; 4=very important)
Table 25 Summary of medians of quality factors
3.1.8.1 Quality factors and Recommendation decision groups
1) Physicians who recommended apps (RA) associated to QFs
The following Table 26 shows the QFs which rated by respondents who recommended
health apps to their patients. The similarities are highlighted in the table.
NRA groups SE n=3 SG n=6 Very important (Median=4.00)
Clinically Proven Evaluated by user feedback Protection of patient health data Good customer service and technical
support
Clinically Proven Evaluated by user feedback Protection of patient health data Good customer service and technical
support Cleared by FDA Certified by ISO Endorsed by healthcare organizations Evaluated by “Peer-review” Patient involved in app development
Important (median=3.00)
Endorsed by healthcare organizations Patient involved in app development Evaluated by “Peer-review” Cleared by FDA Certified by ISO
Hospital-branded HA
Less important (median=2.00)
Top ranking in app stores Hospital-branded HA
Top ranking in app stores
Unimportant (median=1.00)
Developed by famous IT companies Developed by famous IT companies
Table 26 Quality Factors associated with RA group
2) Physicians who will recommend apps (WRA) associated to QFs
The following Table 27 shows the QFs which rated by respondents who will recommend
health apps to their patients. The similarities are highlighted in the table.
31
NRA groups SE n=3 SG n=6 Very important (Median=4.00)
Clinically Proven Evaluated by “Peer-review” Evaluated by user feedback Protection of patient health data Good customer service and technical
support
Clinically Proven Evaluated by “Peer-review” Evaluated by user feedback Protection of patient health data Good customer service and technical
support Cleared by FDA Certified by ISO
Important (median=3.00)
Endorsed by healthcare organizations Patient involved in app development Cleared by FDA Certified by ISO
Endorsed by healthcare organizations Patient involved in app development Hospital-branded HA
Less important (median=2.00)
Top ranking in app stores Hospital-branded HA
Top ranking in app stores
Unimportant (median=1.00)
Developed by famous IT companies Developed by famous IT companies
Table 27 Quality Factors associated with WRA group
3) Physicians who never recommend apps (NRA) associated to QFs
The following Table 28 shows the QFs which rated by respondents who never
recommend health apps to their patients. The similarities are highlighted in the table.
NRA groups SE n=3 SG n=6 Very important (Median=4.00)
Clinically Proven Cleared by FDA Certified by ISO Evaluated by “Peer-review” Evaluated by user feedback Patient involved in app
development Protection of patient health data Good customer service and technical
support
Clinically Proven Cleared by FDA Certified by ISO Evaluated by “Peer-review” Evaluated by user feedback Patient involved in app development Protection of patient health data Good customer service and technical
support Hospital-branded HA HA endorsed by healthcare
organizations Top ranking in app stores
Important (median=3.00)
Hospital-branded HA HA endorsed by healthcare
organizations Top ranking in app stores
Developed by famous IT companies
Less important (median=2.00)
Developed by famous IT companies none
Unimportant (median=1.00)
none none
Table 28 Quality Factors associated with NRA group
32
3.2 Findings from follow-up interviews
There were 4 out of 71 respondents given follow-up interviews via e-mail after they
completed the questionnaire, thus both countries had 2 interviewees (Table 29) answered
interview questions which were designed based on their questionnaire answers.
ID Specialty Adopter category Recommendation decision
Countries
178 pediatrician Innovator Recommended (RA) SE
233 Child Diabetes Early Adopter Recommended (RA) SE
237 GP/family medicine Innovator Recommended (RA) SG
202 GP/family medicine Early Adopter Recommended (RA) SG
Table 29 Basic information about interviewees
The interview results were summarized into different questions as shown below:
1) [Motivation factors-Patients’ interests]
How do you perceive your patients’ interests/knowledge about health apps?
“Younger people are more prone to use smartphones actively. I wouldn’t say
children, rather teenagers and adolescents [trialabiltiy; target group].” (ID178)
“The majorities think it s perfect to be able to use apps instead of paper lists etc. as
they always carry their phones with them [simplicity].” (ID233)
“When my patients choose health apps, e.g. fitness app, they don't know what they
really need. Some dieters only use weight record and fat intake calculator without
concerning about their health conditions [patients lack medical knowledge].” (ID
237)
“Actually my patients know more about health apps rather than me. Younger people,
they know more about using health apps…however, the seniors have a lot of
problems on it… [Trialability; target group]” (ID202)
2) [Motivation factors and Channels - Patients’ Feedback]
How do know and measure patients’ feedback about the recommended health apps?
“I don t measure that scientifically. I get the response when I meet the families again
and they tell me how they use the apps and how it has helped them in different ways
[patients show interests in health apps; interpersonal communication].” (ID233)
“I always try to get feedback from patients what works and not. I don’t have the time
to do a structured feedback so it is qualitative… [Interpersonal communication]”
(ID178)
33
“I asked when I met or called them. I usually trust what they have told to me
[Interpersonal communication]. I also can get the truth when I check their health
conditions [judge the utility of a health app based on own knowledge and
experience].” (ID 237)
3) [Channel/motivation factors]
Which channel helps you most?
“I have several high qualitative web sites [health app website] I follow. Usually I test
the app myself as well.” (ID178)
“I trust the information from colleagues most… The patients are the users of the
apps and their point of view is therefore very important. They know what they need
and what will help them manage their diseases. [Interpersonal communication]”
(ID233)
“App stores [searching on app store] have a lot of good health apps… Patients
[Interpersonal communication] also have some good recommendations, but I really
need to test them by myself... [judge the utility of a health app based on own
knowledge and experience]” (ID 237)
“Some websites [health website (NHS)]…most time I asked my colleagues for trying,
and then I collected feedback. [Interpersonal communication]” (ID202)
4) [Benefits – observed benefits]
Have you found any remarkable changes happened on your patients while using the
recommended health apps?
“Yes, some younger patients get greater freedom as they can use apps for
carbohydrate counting and also get dose suggestions as a help when they re in
school, with friends etc. They also get trained in knowing the carbohydrate content
in common food. I think the apps sometimes are the reason some patients start using
carbohydrate counting [Improve self-management ability; driving factors].” (ID233)
“Some of my patients are adopting some fitness apps with calorie calculator.... Some
remarkable things I've seen that they use apps proactively when they are jogging,
work-out or eating food. It is a good way for improving self-care. [Improve self-
management ability; driving factors]” (ID202)
5) [Benefits/Barrier - integration]
Can we trust patient input data? Would you like to integrate those apps with clinical
system?
34
“We can of course not totally trust self-reported data anyhow – not to day and not in
the future…. If we would like to engage the patients as an actor in their own care
processes then we need integration with EMRs [integration with clinical data] for
example for fine tuning medication, follow up of rehab etc. I would like to validate
data as close to measurement as possible so it would be a must for the apps. If we
use international standards ISO/HL7/Continua etc., then it would be easier.” (ID178)
“It would be very exciting to integrate apps with clinical systems, e.g. compare the
carbohydrate intake with blood sugar level, HbA1c… [Positive attitudes towards
integration with clinical data]” (ID233)
“We can track their changes and monitor their activities, it is better than just “listen
to” their feedback… [Positive attitudes towards integration with clinical data]”
(ID202)
“I’d like to see the fully integration with our system in the future [Positive attitudes
towards integration with clinical data]. Otherwise there is no meaningful use of
patient input data.” (ID 237)
6) [Barriers – additional barriers]
Before you decide to recommend a health app to your patient, what are the major
issues you will consider about?
“When I have older patients or immigrants’ language can be a problem – most apps
are in English. [Language problem - Trialability]” (ID178)
“Some patients are not used to apps, some children don t have their own mobile
phone, and some don t have the right model for some apps [Trialability]…Something
[Barriers] likes ‘Does the app add anything of value for the patient [relative
advantage]? Is it reliable? Do other patients or healthcare professionals use the app
[interpersonal communication; Compatibility]?’ My experience of apps is limited to
diabetes apps. The carbohydrate-lists are reliable, but the apps that give dose-
recommendations by carbohydrate ratios, carbohydrate intake etc. They have their
limitations that are hard to bridge. You always have to think yourself what s
reasonable [judge the utility of a health app based on own knowledge and experience].
“(ID233)
“Most of my patients are seniors. Some of them don't even use smartphones. So I
usually need to find apps which are easy to use and easy to understand”
[Trialability] (ID 237)
“Are those apps tested based on RCT [Randomized Controlled Trial]? Can they
really help my patients? I need to see the evidence…” [Relative advantage;
Compatibility] (ID202)
35
7) [Barriers – Incentive system]
Do you think that an incentive system for recommending health app can motivate
more doctors to recommend apps to patients?
“It must also be incentivized like other medical activities…” (ID178)
“Yes, maybe some doctors think it s good with an incentive system for health apps.
Perhaps there are risks if the prescribed app will expire [importance of updates] but I
don t think it s a big risk. It could be a bigger risk that the doctor is unaware of
existing apps that the patients use.” (ID233)
“Encourage physicians who recommend health apps to patients by providing
incentives like regular journal updates, free access to medical/health education
apps for them etc.” (ID202)
8) [Barriers - Lack of choice]
Do you think a dedicated health app aggregator website/app store providing
search/filter functions can make things easier?
“Yes, I think it would be of great help with an aggregator website to provide good
quality of the health apps. (ID 233)
“It depends if we see the apps as wellness, preventive care or as tools for chronic
conditions…” (ID 178)
“Yes, it is good. Actually I usually search health apps on websites just like what you
described… if you know NHS, it is trustable.”(ID 202)
9) [Quality Factors]:
How do you think about the quality of health apps in overall? What are the major
issues we need to fix immediately?
“I am not prescribing apps…I would welcome some apps to be regarded as medical
device like blood sugar meters etc…It has to be regulated like other medical devices
(Medical Device Directive) – when someone says that that hinders innovation I object,
if not the innovator dares to take the responsibility for an app, why should a patient
or a physician dare?” (ID178)
“Of course the responsibility would lie on persons who are good at estimate the
quality of an app [surveillance], e.g. health care professionals specialized on the
actual disease.[peer-review]…I think it s hard for the government to guarantee the
quality of health apps even if it would be a good idea. If they would, they should focus
on that the app in serious, what s its intention, no advertising, that it s been clinically
36
tested if possible…I always emphasizes to my patients that the app only can give a
dose recommendation, it doesn t know anything about you, what you re up to etc.
[lack of personalized design]” (ID233)
“Presently, most apps are not smart enough to care each individual's nutrition intake
and daily exercises [lack of personalized design]…Most health apps used by my
patients are not regulated by FDA. It is very difficult to push the SMC (Singapore
Medical Council) to regulate the quality of app presently. Actually it is a very
dangerous situation, because we cannot be sure they are safe for my patients since
most of them are not clinically tested. RCT [Randomized Control trial] is very
necessary!” (ID 237)
“More physicians and patients need to be aware of it through better promotion,
advertisement and awareness campaigns.” (ID202)
37
4 Discussion
Principally, the Diffusion of innovation theory dominates the discussion to answer
research questions. The discussion of main findings in section 4.1 initially focuses on the
similarities found from both countries. The section 4.2 focuses on discussing the
differences of two countries’ results. The main findings of each country are presented
separately in sub-section 4.2.1 (Sweden) and sub-section 4.2.2 (Singapore).
4.1 Main findings from both countries
4.1.1 Physicians’ current knowledge and attitudes
According to the results, the diffusion process is depicted in Figure 11. Since there are
several similarities presented in result section, this diffusion process can be applied to
both countries.
Figure 11 Diffusion of health apps and contributive factors from physician’s perspective
High recognition of adopting health apps
According to the results in section 3.1.5, almost all the respondents know health apps.
Moreover, there are 73.2% physicians using health apps themselves. Therefore, it shows
high recognitions of adoption of health apps among physicians. However, if matching the
results to IMS’s Health App Maturity Model, the current situations still stay on the initial
stage of promoting adoption of health apps for both Swedish and Singaporean physicians
due to only small group of physicians recommending health apps to patients.
“Wait-and-see” attitude is dominant
This study confirmed the existence of diffusion of health apps between physicians and
patients in the real world. For the current situation, more than half of physicians keep
“wait-and-see” attitudes that they consider recommending health apps to their patients in
the future. However, if we compare the results of other recommendation decision groups,
physicians who have recommended health apps have double populations than the numbers
Unhealthy lifestyle
Chronic condition
Cognitive problem
Patients
Deliver Benefits
Quality factors Weakness & Barriers factors
Provide Evidences
Physical training
Major Motivations
Most recommended apps
Health promotion
Cognitive training
Booking & Locator
Major Target
Groups
Provide Feedbacks
Physicians
Main Channels
38
of physicians who don’t recommend health apps. It still shows a positive trend for
diffusing health apps from physicians to patients.
Innovators and Early adopters contribute the diffusion of health apps
According to statistics in section 3.1.2, there are 18 innovators and 31 early adopters
occupied 69% of total 71 participants. Among those innovators and early adopters, half of
them (n=20) have recommended health apps to their patients and the rests will
recommend health apps to patients. Therefore, innovators and early adopters keep positive
attitudes towards recommending health apps, even though half of them have “wait-and-
see” attitudes. As mentioned in the introduction section, innovators and early adopters
performed as “opinion leader” for diffusion of an innovation. In other words, other
adopters would like to ask them for suggestions and evidence of helpfulness of an
innovation. As our results show that physicians also test apps before they recommend to
patients, their experiences about using the health app will be valuable to other adopters. In
addition, considering the overall patients’ feedback of recommended health apps is good,
it shows a good potential for promoting the behavior of recommendation among
physicians by showing good feedback and strong evidence. Therefore, innovators and
early adopters play have an important role influencing the diffusion of health app, but also
influence the diffusion of recommendation behaviors among different adopter groups.
Limited channels: interpersonal communication is the most effective channel
As shown it the figure 11, four main channels that physicians get information about health
apps are listed. No doubt, mobile apps store is rated as the top channel. Search engines
and patients only differ by a single vote. From the follow-up interviews, two early
adopters from two countries have stated that they like to ask colleagues and patients for
information and experiences about a health app, even though they also search apps in app
stores. As patients and colleagues provide feedbacks and evidences that help them reduce
the uncertainties, they can make easier or faster choices to find or choose a suitable health
app. Therefore, interpersonal communication is considered as the most effective way to
know a health app and exam the usefulness of a health app. However, patients and
physicians are not able to know and use all different kinds of health apps, which may limit
the information channel. In addition, mobile app stores also contain a large numbers of
health apps with different quality. It is increase the difficulties to physicians to find
specific ones fit their patients’ needs.
4.1.2 Driving factors and Barriers
DOI Driving factors Barriers Quality Factors
Observability improve patient encounter; Strongly improve patients’ self-management abilities
Lack of integration with clinical data Poor quality and design
High Awareness of Clinical trial (RCT); High awareness of trustable Peer-review & User feedbacks
Trialability Good Trialability, (self-test, patients test, colleagues test )
Cannot find specific one fit patient Unaware of
Relative low awareness of existing aggregator websites;
39
recommendation behavior
Simplicity Good mobility; Easy to run on mobile devices; Easy to download; Replace paper lists
Unclear instruction/information; Lack of personalized design;
Strong demands of good customer services/technical support;
Compatibility Personal judgments; Good user experiences from self-use, colleagues, and patients; Save health care operational costs
Lack of surveillance by government/regulators Lack of incentive system
High awareness of Protection of patient information/health data;
Relative advantage
Strongly improve patients’ self-management abilities; Improve continuity of care; Patient-centered care; Improve disease prevention;
Lack of evidence-based content; Lack of integration with clinical data (did not improve the clinical data quality)
Strong demands of evidence and successful stories
Table 30 Contributive factors of health apps diffusion from physician’s perspective
Driving factors: interpersonal communications motivate the diffusion of health apps
Diffusion of Innovation defines five perceived characteristics which can effects the
success of diffusion of an innovation in a system, thus factors are categorized into
corresponding aspects (Table 30). The driving factors are derived from motivation factors
and benefit factors of health apps, which promote most physicians to have positive
attitudes towards recommending health apps to patients. As shown in figure 11, there are
three major motivation factors in term of self-use health apps, patients’ interests and
recommendations, and colleagues’ recommendation. As 73.2% of participants are using
health apps themselves, their experiences about used health apps may help them judge the
quality of a health app. Moreover, they can share their experiences with patients and other
physicians. As data shows that 50.7% of total participants were asked by their patients
about health apps, Patients’ interests about health apps also can motivate physicians to try
or seek specific health apps for matching their patients’ needs. Additionally, as mentioned,
patients and colleagues provide most efficient information channel for physicians to know
and find health apps, a trust relationship between them is built that can also motivate
physicians to recommend good word of mouth health apps to their patients. Therefore,
interpersonal communication plays an important role influencing the diffusion of health
apps.
Driving factors: observable benefits enhance the relative advantage of health apps
As shown in Table 30, the most important benefit of health apps (i.e. improving patient
self-management ability) is perceived by both countries’ physicians. If we have a look the
results from the interviews, both countries’ interviewees have stated that they tested each
app by themselves and also observed the changes from patients when they met. The
observable benefits from patient encounters can enhance the relative advantages of health
40
apps by demonstrating significant evidences. Combined with good mobility and
trialability, those driving factors encourage innovators and early adopters to recommend
health apps to both cosmopolite and heterogeneous individuals (i.e. their colleagues and
patients).
Barriers: multiple factors obstruct the diffusion of health apps from physicians to
patients
As shown in Figure 11, physicians would like to deliver benefits to patients via health
apps, however, in the current situation, physicians are facing many challenges. Firstly, our
results have confirmed the influences of the elements presented in IMS’s “six hurdlers”
model. All the elements have considerable medians approaching to 3.00, which mean that
physicians are facing all six hurdlers. Among those hurdlers, the most concerned elements
are “health apps lack surveillance by governments and regulators”, “apps lack
integration with clinical data” and “poor protection of patients’ health data”.
Additionally, physicians also ranked three most concerned weaknesses regarding the
design of health apps, which are “health apps lack evidence-based content”, “unclear
instruction/information”, and “lacks of personalized design”. Moreover, both countries’
physicians consider themselves lack awareness of recommending apps to patients. A
solution can be found in the follow-up interviews is that an incentive system of health
apps may encourage more physicians to recommend health apps to patients, but those
apps should be regulated and quality-guaranteed.
Quality: Double-edged sword to facilitate/encumber the diffusion of health apps
Quality factors are performed as double-edged swords: good quality of health app can
facilitate the diffusion of health apps; conversely, bad quality can encumber the diffusion.
However, the definitions of “good quality” are ambiguity and varied. For this reason, we
extracted all the relevant elements (quality factors) from literatures regarding quality of
health apps and asked physicians to choose which factors can ensure the quality of health
apps in their minds. Interestingly, we can see the results in section 3.1.9, “clinical
proven”, “peer-review”, “user feedback”, “protection of patient health data”, and
“customer services” are considered as the most important factor influencing the quality of
health apps. In other words, the performance of those factors on a health app may
influence physicians’ decision making about recommending it to their patients. Through
the follow-up interviews, clinical tests and randomized controlled trial (RCT) for a
health app are mentioned frequently as the most important solutions for ensuring the
quality of health apps. If a health app can get clinical proven, it definitely enhance its
relative advantage. Moreover, considering about “peer-review” and “user feedback”
approaches, many health app aggregator websites have adopted those methods to evaluate
uploaded health apps in order to provide qualified health apps; however, those websites
are not considered as the main channel according to questionnaire results. For this issue,
on one hand, physicians need to increase their awareness and knowledge about those
websites; on the other hand, as app stores is the main channel for searching health apps,
health app websites construct collaborations with app stores to provide qualified health
41
apps. Additionally, as an incentive system can motivate physicians, health app websites
can be also involved in an incentive system by providing database of qualified apps. As a
result, good customer services/technical supports are even more important to keep updates
of health apps and also ensure apps are compatible with HL7, ISO and other standards to
protect patients’ information and health data, and meanwhile, it increase the possibilities
to integrate health apps with clinical systems, and further to achieve fully integration as
shown in last step of the Health Apps maturity model (Figure 3).
4.2 Different findings from two countries
4.2.1 Main findings from Swedish physicians
4.2.1.1 The role of early adopters
According to the results shown in section 3.1.2, the distribution of early adopters and
innovators spread almost among all age groups, especially in older age groups, which
highlight the aforementioned statement that “Sweden has many early adopters and IT
literacy is high in older age”. The reason behind that could be traced back to the
definition of adopters in Diffusion of Innovation. As early adopters have a higher social
status and good financial liquidity, older age population can meet those features [35].
Moreover, according to the results in section 3.1.3, those physicians who recommended
health apps to their patients are compose of a few groups of progressive physicians
including mostly early adopters and few innovators. Additionally, most physicians tend to
keep “wait-and-see” attitudes that they consider recommend health apps to their patients
in the future, in which also involve a larger number of early adopters. As early adopters
perform as “opinion leaders”, their decisions will definitely influence others. Moreover,
“patients” and “colleagues” are the major channels for Swedish physicians getting
information about health apps, but also the latter is the trustiest channel for physicians to
receive recommendations and collect feedbacks. As “Patients” and “colleagues” may
contain a huge number of innovators and early adopters, our results suggest that
recommendation of health apps requires obtaining supports from opinion leaders in order
to reduce the uncertainties raised by other adopters. However, there is a challenge for us
to identify opinion leaders and even harder to know which opinion leader exerts a positive
influence.
4.2.1.2 Factors influenced the health app diffusion
DOI Driving factors Barriers Quality Factors
Observability Efficient patient encounter; Improve treatment plan
Lack of integration with clinical data
Awareness of Clinical trial; Need trustable Peer-review & User feedback
Trialability Good Trialability, (self-test, patients test, colleagues test )
Cannot find specific one fit patient Unaware of existing apps Lack of multi-language support
Relative low awareness of existing aggregator websites; Require timely Updates
42
Simplicity Good mobility; Easy to run on mobile devices; Easy to download; Replace paper lists
Unclear instruction/information; Lack of personalized design;
Require good customer services/technical support;
Compatibility Personal judgments; Good user experiences from self-use, colleagues, patients
Lack of surveillance by government/regulators
awareness of FDA, ISO; High awareness of Protect patient information/health data; Require to treat health apps as medical devices
Relative advantage
Improve patients’ self-management abilities
Lack of evidence-based content Lack of integration with clinical data
High awareness of Clinical proven (passed clinical trial)
Table 31 Contributive factors of health apps diffusion from Swedish physicians’ perspective
As elements summarized in Table 31, even though few Swedish physicians tend to have
negative attitudes, they still perceive the most important benefit of health apps (i.e.
improving patient’s self-management ability). By analyzing the interview results, we
understand that physicians have observed some remarkable changes when their patients
use recommended apps, which improve the efficiency of patient encounters, and further
improve treatment plans. Moreover, most physicians test health apps before they
recommend to patients, their knowledge combined with own user experiences can judge
the performance of adopted health apps. Those observable benefits demonstrate
significant evidences to enhance the relative advantages of health apps. Combined with
good mobility and trialability, those driving factors motivate the diffusion from
innovators and early adopters to their colleagues and patients.
However, in the current situation, physicians are still facing many hurdlers. Even though
some physicians do recommend health apps, they are still facing many challenges as other
physicians have. Firstly, our results have confirmed the influences of the elements
presented in IMS’s “six hurdlers” model. All the elements have considerable median
approached to 3.00, which mean that most physicians agree with those issues are their
facing barriers. According to the results, we found out that the most concerned elements
of them are “lack of surveillance by governments and regulators” and “lack of choices”.
Through the interviews, all Swedish interviewees have expressed their opinions on those
two issues. For the former, a suggestion is to treat health apps like medical devices. A
rigorous standard as “medical device directive” may hinder the progress of innovations,
but it may effectively fix the weaknesses such as “Lack of evidence-based content”,
“poor protection of patient health data”, and “unclear instruction/information” and so
on. The standardization of health apps may also encourage the fully integration between
health apps and clinical systems. As a result, it may enhance the relative advantages of
health apps and also be compatible with existing values in order to gain endorsements
from healthcare organizations. On the other hand, the other consideration from interview
states that currently is difficult to push the government or regulators to oversee the
qualities of health apps like medical devices, which suggests that regulators should keep
43
an eye on the intentions of health apps to forbid advertising and criminal purposes.
Actually, FDA has cleared more than 300 health apps and published them for publics [52].
FDA’s regulation regarding health apps could meet their requirements to some extent.
However, based on the results from questionnaires, most physicians have not yet
considered this solution as the most important factors that ensure the quality of health
apps. It is hard to know the reasons behind it, but one respondent has given an answer that
is “I don’t know FDA has the responsibility for regulating health apps”. Therefore,
unawareness of current relevant regulations could be a reason. No matter what regulations
they knew, to implement randomized clinical trial (RCT) for a health app is always
considered as the most important solution for ensuring the quality of health apps. Some
hospital-branded health apps could be released with RCT [53], but the comments
regarding “hospital-branded health apps” are quite different. On one hand, some
physicians consider it as a “market gimmick”. On the other hand, fewer physicians agree
it could be an important factor for quality assurance. As one comment wrote in the
questionnaire, “at least I knew how the app developed by my hospital, I would be
confident to recommend it”, hospital-branded apps could have an opportunity to perform
as a relative advantage to show evidence to all adopter groups.
Considering about the other aforementioned hurdle “lack of choices”, our results show
that most physicians only seek health apps on mobile application stores. However, there
are numerous apps in mobile application stores that may not qualified as a medically
relevant apps even they are showing in the “Medical” or “Health & Fitness” categories
[24]. The anonymous user reviews in apps stores would involve too much commercial
purpose that limited their utility. Beside app stores, there are many health app aggregator
websites providing qualified health apps, however, they are not considered as the main
channel according to questionnaire results. On the other hand, one interviewee (innovator)
states that he/she uses some health app websites, and the other interviewee (early adopter)
emphasizes the importance of “peer-review”. This point of view about “peer-review”
caters the trend rated by physicians in the questionnaire. Therefore, a health app
aggregator website with “peer-review” evaluations could be a solution ensures the quality
of health apps, but it needs to be generalized in a larger population. However, there a
potential risk presented by one of interviewee that some “peer-reviewed” health apps may
be out-of-date by the time of release, thus, physicians themselves should be equipped with
relevant knowledge and pay attention to any potential risks [24]. Therefore, it could be a
reason that most physicians considered “Good customer service/technical support” as one
of the most important factor ensure the updates of health apps. Another issues regarding
“lack of choices” of health apps are “lack of multi-language support” and “lack of
personalized design”. The former one is also raised by one of the interviewees that “Some
migrations and elderly patients had difficulties on foreign languages, but most health
apps only ran in English”. This weakness would impact the Trialability of health apps
and also weakens user experiences. On the other hand, “lack of the personalized design”
is not a novelty for any IT innovations. It weakens user experiences from both physicians
and patients and further stop the diffusion of a health app. If patient can be involved in
44
health apps development process, it will improve user experiences and add more value to
the utility of health apps. The questionnaire results show that most physicians think
patient engagement is important quality factor. Meanwhile, interviewees also suggest that
patient engagement in app development process would be very important, because
patients are the end-users know what they need and what can help them manage their
diseases.
4.2.2 Main findings from Singaporean physicians
4.2.2.1 The role of Innovators
Compared to the population distribution of Swedish samples, the results show that the
average age of Singapore sample trend to be younger. The bar chart (figure 9) in section
3.1.4 shows that Singaporean samples involve a larger population of innovators and early
majorities. According to the results in section 3.1.3, a few groups of progressive
Singaporean physicians are composed of mostly innovators recommended health apps to
their patient. Most physicians have “wait-and-see” attitudes as same as the Swedish
physicians have. But this time, the major adopters are innovators. Moreover, Singaporean
physicians tend to gain information of health apps via mobile application stores, search
engine, and health app websites, which are counted as non-human resources compared to
“patients” and “colleagues” for Swedish physicians. As the Diffusion of Innovation
theory defines that innovators usually the first one adopting new innovations, have the
highest risk tolerance, highest social status, and closest contacts with scientific source,
thus they do not worried about risks and have no need to find evidences from other
adopter groups [35], which can explain the differences of information channels between
two counties. As Singaporean respect elitism, innovators’ opinions may have significant
effects on the others’ decision making due to their highest social status. In addition, based
on the questionnaire results, there is a considerable ratio of early majorities which is
larger than early adopters but smaller than the innovators. Roger et al. [32] state that the
majorities are willing to see evidences from early adopters rather than other adopters, a
few number or a lack of early adopters can slow down the speed of diffusion. Moreover,
by analyzing the results of NRA groups (physicians who don’t recommend health apps),
the major populations are late majorities, thus they are more skeptical and would like to
take a long time to wait more successful stories from others. Our results also reflect their
cautious attitudes. For instance, only one of them used health app himself/herself and they
have negative attitudes on all the benefits of health apps (see section 3.1.7). Therefore,
our results also suggest that the recommendation of health apps requires obtaining
supports from innovators, which should take more responsibilities to perform as “opinion
leaders”. As the results show that the Singaporean physicians who have recommended
health apps also test each health apps and ask their patients for feedback, they would have
better understanding about the benefits and risks of health apps perceived by themselves
and patients. As a result, they should convey all the information they know about the
health apps if it is to make sense to other potential adopters including physicians and
patients, and further help them reduce perturbations.
45
4.2.2.2 Factors influenced the health app diffusion
DOI Driving factors Barriers Quality Factors
Observability Efficient patient encounter; Health promotion Remote medical support
Lack of integration with clinical data
Higher awareness of clinical trial; Higher awareness of trustable Peer-review & User feedback
Trialability Good Trialability, (self-test, patients test, colleagues test )
Unaware of existing apps Lack of reimbursement standards;
Dedicated Aggregator websites; Updates
Simplicity Good mobility; Easy to run on mobile devices; Easy to download; Easy for Youth
Unclear instruction/information; Lack of personalized design;
Higher awareness of Customer services/technical support;
Compatibility Personal judgments; Good user experiences from mainly self-use
Lack of surveillance by government/regulators; Data overload if integrated; Lack of endorsement; Physicians lacks incentives;
Higher awareness of FDA, ISO; High awareness of protection of patient information/health data;
Relative advantage
Improve patients’ self-management abilities Free/low charge
Lack of evidence-based content Lack of integration with clinical data
Higher awareness of Clinical proven (passed clinical trial)
Table 32 Contributive factors of health apps diffusion from Singaporean physicians’ perspective
Compared to the factors aforementioned from Swedish side (section 4.2.1.2), differences
are highlighted in Table 32. The analyses show that Singaporean physicians not only
agree with “improving patients’ self-management abilities” as the major benefits, but also
perceive “patients can use apps for free/low charge” as one of the major benefits that
performs as relative advantages for health apps. As mentioned, Singaporean citizens need
to pay the most part of their health expenditure, thus, it could be a reason to explain the
differences. Besides that, from the interview results, we understand that Singaporean
physicians have perceived some remarkable changes on health promotion while their
patients using health apps. Their observed changes combine with their knowledge can
judge the performance of the adopted health app. Those observable benefits can
demonstrate significant evidences to enhance health apps’ relative advantages. However,
Singaporean physicians does not have strong believes that health apps could improve
treatment plan, but they do agree that health apps could improve remote medical support.
The reason behind is unknown, which may be related to the successful stories they have
seen or their implemented national healthcare strategies.
Singaporean physicians are also facing many barriers. Firstly, most physicians agree with
the effects of elements presented in “six hurdles” model. The most concerned barriers are
“lack of surveillance by governments and regulators”, “lack of choices” and “lack of
endorsement from healthcare institutions with legal agreements”. Moreover, all barriers
gain higher medians compared to the results from Swedish side, especially on surveillance
46
and endorsement related issues. One comments from questionnaire could explain the
reason as “I have to know all the detail about liability…If something happened caused by
the apps, am I liable?” Therefore, Singaporean physicians have a higher awareness on
legal aspects. In addition, they also have higher awareness on the role of clinical trial and
“peer-review”. Through the follow-up interviews, interviewees have emphasized the
importance of Randomized Controlled Trial in clinic, which can provide significant
evidence for showing the effects of a health app. Moreover, Singaporean physicians have
higher awareness on the role of FDA and ISO. One interviewee emphasized that “we are
facing a dangerous situation since most health apps are not regulated by FDA”.
Therefore, Singaporean physicians tend to gain certifications or guarantees from official
organizations, which may help them reduce the uncertainties or provide a legal basis for
both patients and physicians. Considering about the “lack of choices” issues, Singaporean
physicians are considered that have more difficulties than Swedish physicians due to their
higher median of results in this regard. On one hand, most Singaporean physicians were
searching health apps in app stores. We have discussed the weakness of mobile app stores
in section 4.2.1.2, which can be a reason. On the other hand, Health app websites such as
NHS was rated as the third major channel, but only 10 respondents know them, in which
mostly consisted of innovators. As aforementioned feature of innovators, they had few
communications with heterogeneous groups. As a result, it potentially restrains the
information flow from innovators to potential adopter groups.
4.3 Strength and Limitations
On one hand, the strengths of this study mainly lie in six aspects:
This study is a comparative study focuses on a high relevant and timely topic;
It has a strong theoretical basis and designates precisely how theories was used in
order to ensure the stability reliability and face validity;
This study has strong relations built between theories and data collection according to
Data Requirements Table (Table 5) to ensure the internal consistency and construct
validity;
It precisely defines the measurements scales and methods for data analysis to ensure
the equivalency reliably and criterion-related validity and to avoid measurement bias;
It uses two sets of data: a large amount of quantitative data from questionnaire, which
map the diffusion of health app from physicians’ perspectives, the second data set
comes from follow-up interviews, which complements the results and analysis to
ensure the diversity of data;
It precisely combines theories in results analysis and discussion to ensure the
consistency.
Therefore, this study is well-designed and has high reliability and validity that allow its
replicability. This study also examines the elements of the “six hurdles” model and other
elements extracted from literature review regarding driving factors and barriers, which are
already tested once in their original studies, thus, those elements can be generalized.
Moreover, the survey questionnaire works as feed-forward tool for interviewees that not
47
only saves amount of time for introducing the research problem to them, but also
generates overall impressions of each interviewee’s attitudes and preferences. According
to the feed-forward information, it is easy to generate relevant questions for asking each
interviewee for further information. Moreover, as email interview has no time-limitation,
interviewees can think about each question in depth and answer it in a well-structured text
form, which also save the time for transcribing data.
On the other hand, this study contains some limitations that causes some bias can weaken
its generalizability. The limitations lie in seven aspects:
Sweden’s sample (n=44) and Singapore’s sample (n=27) giving a total sample size
(n=71) is too small to allow the statistics to be safely inferred to entire population,
especially to the Singaporean physician population, which also limits the analysis of
diversity and heterogeneity;
Most Swedish participants are primary care physicians mainly from SFAM but few
participants from hospitals, which cause inclusive bias and omission bias, thus the
sample group is not truly representative of the whole population of Swedish
physicians;
Two countries were adopted different methods for data collection, which causes
inclusive bias that weaken reliability for making comparisons;
Only use email interview technique is adopted for collection qualitative data that may
lose some possible potential information; interviewees also can skip some questions;
Limited number of follow-up interviews (n=4) that may limited the transferability of
the findings from qualitative data;
All 4 interviewees are from RA groups (physicians who recommended apps), thus
there is no interviewees from other groups;
Coding system is not fully adopted in qualitative data analysis which weakens the
preciseness of chosen research method.
Furthermore, this study is conducted with an abduction approach that combines both
induction and deduction, thus the chosen methodology comprises the advantages and
disadvantages of both approaches. The deduction side emphasizes on defining a highly
structured theoretical background and tends to collect large amount of quantitative data
through for example survey strategy in order to test theories and hypothesis and
generalize the conclusions. However, it has limited flexibility of changes in method due to
its rigorous structure. With the research into a further stage, new findings emerge which
that cannot be explained only according to deductive results, an inductive is more
appropriate to use to collect qualitative data through for example interview strategy in this
study. However, the inductive provides more flexible structure and permits changes of
research emphasis, thus it has less feasibility regarding generalizing its results.
Moreover, this study mainly adopts survey strategy, thus the results are predominated by
quantitative data from survey questionnaire. There is a lack of enough qualitative data
from follow-up interview (only email interview) to explain cause-effect relations
48
comprehensively. An alternative qualitative data collection technique such as semi-
structured interview would be adopted. Compared to email interview, a semi-structured
interview is more open and allows new ideas to be brought out during the interview in a
real time. As more interviews we have, more qualitative data we get. Those data can be
coded systematically to develop an existing theory or form a new theory. Therefore, it is
important to find enough number of interviewees to participate the research. A Multiple
case study strategy can help researcher access several healthcare settings to approach
more potential interviewees (physicians). A larger population provides more opportunities
to categorize them into different adopter groups according to diffusion of innovation
Theory [32], in order to understand different opinions regarding diffusion of health apps.
However, as this research is a comparative study that takes Sweden and Singapore into
account, there is a challenge that to gain supports from both countries’ healthcare
organizations.
Another alternative research strategy is Ethnography which is rooted firmly in the
inductive approach, which particularly helps researchers to describe and explain any
phenomenon in social world graphically and culturally [37]. It usually involves huge
amount of qualitative data through interview and observations. Ethnography strategy
would have advantages for this comparative study that the analysis would present
comprehensive comparisons including cultural studies, economics, education, political
aspects, psychology, and other different aspects. Those analysis would help researcher
have multi-dimensional understandings about what involved in health apps diffusion
under a culture, how it works and why it works like that [37]. However, Ethnography
strategy is very time-consuming and requires researchers to fully participate in the social
activities and also build long-term trust relationships with research participants. Therefore,
it is difficult to conduct an Ethnography strategy in a research within a short period.
Additionally, another limitation of this study is that physicians are the research objects
rather than patients who are the intended users of health apps. As this study is motivated
by previous studies [5,8,14,30-31] that consider physicians as potential diffusion channel
of health apps, it is conducted to exam the assumptions and also explore other relevant
issues. However, physicians are not app experts, their knowledge of health apps, attitudes,
and suggestions may be only based on their experiences and from medical perspectives,
which may cause bias that this study only depicts a small part of the diffusion process of
health apps in this pre-defined social system rather than depicts the whole picture
including health care providers, patients, and many other elements. Moreover, this study
treats health apps as a whole rather than focuses on a specific category. As health apps are
numerous and various today, physicians may have different views on each kind of health
apps. Therefore, those limitations weaken the generalization of this study.
49
4.4 Future research
The future research will enlarge the sample sizes and enhance the representativeness and
diversity of sample groups in order to ensure the findings can be generalized to the whole
population. As some respondents state that nurse and other health care providers also
significantly contribute to the diffusion of health apps to patients, the future research may
take them into account. This research not only has confirmed the existence of
recommendation behavior regarding health apps in real world, but also has highlighted
several contributive factors that influence the diffusion of health apps according to
physician’s perspectives. The future research will introduce more qualitative data
collection methods and analysis techniques focus on those factors to explore proper
solutions, in order to make contributions to this research area.
5 Conclusion
This comparative study investigates physicians’ attitudes towards recommending health
apps to patients and describes contributive factors that influence physicians to recommend
apps or not, taking the specifics of two early adopter countries, Singapore and Sweden,
into account. It confirms the existence of diffusion between physicians and patients in the
real world. Figure 11 has answered the first research question by illustrated the current
situation about the diffusion of health apps from physicians’ perspectives for both
countries. Only a small group of progressive physicians have recommended apps to
patients. Most respondents keep “wait-and-see” attitudes. The results highlight the
importance of innovators and early adopters which have positive effects that facilitate the
diffusion. For answering the second research question, Table 30 has summarized main
driving factors and barriers according to the similarities between Swedish and
Singaporean physicians’ results, and further corresponded to five characteristics presented
by Diffusion of Innovation theory. For answering the other part of the third research
questions about the differences between two countries’ results, section 4.2 has deeply
explained the differences of physicians’ knowledge of and attitudes towards health apps,
and also summarized the contributive factors in Table 31 for Sweden’s results and Table
32 for Singapore’s results. The main differences lie in five aspects: 1) Swedish early
adopters perform as opinion leader compared to Singaporean innovators had more
positive effects; 2) Swedish physicians have higher awareness of adopting health apps; 3)
Singaporean physicians consider health apps can provide more benefits on remote
medical support and free/low charges to use; 4) Singaporean physicians have more
considerations about legal issues and higher demands of incentive system; 5) Swedish
physicians have less demands regarding official certification of health apps especially
hospital-branded apps, but Singaporean physicians would like to gain more supports from
official organization such as FDA, ISO, etc.
This study contains some limitations that weaken the generalization of results, but it still
demonstrates many interesting findings in this high relevant and timely topic research
50
area. Physicians have a significant role involved in the diffusion of health care
innovations like health apps, their knowledge of health apps and attitudes from medical
perspective also influence the speed of spreading health apps in the social system between
physicians and patients. As more and more mobile devices are adopted by patients for
health care reasons, more and more problems are disclosing to publics. This study not
only reveals physicians attitudes towards recommending apps to patients, but also exposes
the driving factors and some barriers which need to be solved urgently. Additionally, this
study highlights the importance of quality factors that can increase physicians’
confidences to recommend health apps. The future studies will focus on investigating
proper solutions to barrier and also enlarge the study samples that nurse and other health
care providers will be also taken in account.
51
References
[1] Royal Tropical Institute, “What is mHealth?” [Online]. Available: http://www.mhealthinfo.org/what-
mhealth. [Accessed: 04-Apr-2014].
[2] FDA, “Mobile Medical Applications,” FDA, 2013. [Online]. Available:
http://www.fda.gov/MedicalDevices/ProductsandMedicalProcedures/ConnectedHealth/MobileMedical
Applications/default.htm. [Accessed: 28-Feb-2014].
[3] “Mobile Application Software,” Wikipedia, the free encyclopedia. [Online]. Available:
http://en.wikipedia.org/wiki/Mobile_application_software. [Accessed: 09-Feb-2014].
[4] S. Lim, L. Xue, C. C. Yen, L. Chang, H. C. Chan, B. C. Tai, H. B. L. Duh, and M. Choolani, “A study
on Singaporean women’s acceptance of using mobile phones to seek health information.,” Int. J. Med.
Inform., vol. 80, no. 12, pp. e189–202, Dec. 2011.
[5] IMS Institute for Healthcare Informatics, “Patient Apps for Improved Healthcare From Novelty to
Mainstream,” New Jersey, 2013.
[6] D. J. Stevens, J. a Jackson, N. Howes, and J. Morgan, “Obesity surgery smartphone apps: a review.,”
Obes. Surg., vol. 24, no. 1, pp. 32–6, Jan. 2014.
[7] N. Ramanathan, D. Swendeman, W. S. Comulada, D. Estrin, and M. J. Rotheram-Borus, “Identifying
preferences for mobile health applications for self-monitoring and self-management: focus group
findings from HIV-positive persons and young mothers.,” Int. J. Med. Inform., vol. 82, no. 4, pp. e38–
46, Apr. 2013.
[8] A. S. M. Mosa, I. Yoo, and L. Sheets, “A systematic review of healthcare applications for
smartphones.,” BMC Med. Inform. Decis. Mak., vol. 12, no. 1, p. 67, Jan. 2012.
[9] Jenny Gold, “FDA Seeks To Tame Exploding Medical App Market,” KNH, kaiser health news, 2012. .
[10] F. Haffey, R. R. W. Brady, and S. Maxwell, “A comparison of the reliability of smartphone apps for
opioid conversion.,” Drug Saf., vol. 36, no. 2, pp. 111–7, Feb. 2013.
[11] E. B. Tate, D. Spruijt-Metz, G. O’Reilly, M. Jordan-Marsh, M. Gotsis, M. A. Pentz, and G. F. Dunton,
“mHealth approaches to child obesity prevention: successes, unique challenges, and next directions.,”
Transl. Behav. Med., vol. 3, no. 4, pp. 406–15, Dec. 2013.
[12] J. G. Klann, P. Szolovits, S. M. Downs, and G. Schadow, “Decision support from local data: Creating
adaptive order menus from past clinician behavior.,” J. Biomed. Inform., Dec. 2013.
[13] Jenny Gold, “FDA Seeks To Tame Exploding Medical App Market,” KNH, kaiser health news, 2012.
[Online]. Available: http://www.kaiserhealthnews.org/stories/2012/june/27/fda-medical-app-
market.aspx. [Accessed: 03-Feb-2014].
[14] C. Gerada, “10 health apps to help you stay fitter and feel better,” Theguardian, 2014. [Online].
Available: http://www.theguardian.com/technology/2014/feb/08/health-apps-smartphones-top-10-gp-
recommends. [Accessed: 20-Feb-2014].
[15] Research2Guidance, “mHealth market 2015:500m people will be using healthcare smartphone
applications,” research2Guidance, 2010. [Online]. Available: http://www.research2guidance.com/wp-
content/uploads/2010/11/mHealth-market-2015-blog-image.jpg. [Accessed: 20-Feb-2014].
[16] Gabriel Perna, “Research: Mobile Medical App Market $150 Million and Rapidly Growing.” [Online].
Available: http://www.healthcare-informatics.com/news-item/research-mobile-medical-app-market-
150-million-and-rapidly-growing. [Accessed: 03-Feb-2014].
52
[17] “Mobile Health 2012,” PewResearch Internet Project, 2012. [Online]. Available:
http://www.pewinternet.org/2012/11/08/mobile-health-2012/. [Accessed: 01-Mar-2014].
[18] PWC, “Emerging mHealth:Paths for growth,” 2012.
[19] GSMA, “Mobile Economy: Europe 2013,” PwC, 2013. [Online]. Available:
http://www.pwc.com/en_GX/gx/healthcare/mhealth/assets/pwc-emerging-mhealth-exec-summary.pdf.
[Accessed: 01-Mar-2014].
[20] D. Collins, “Singapore and mHealth: A preview of the week ahead,” mHealthNews, 2013. [Online].
Available: http://www.mhealthnews.com/blog/singapore-and-mhealth-preview-week-ahead?single-
page=true. [Accessed: 30-Mar-2014].
[21] B. Sterling, “2 MAJOR COUNTRIES RANKED BY SMARTPHONE PENETRATION RATES,”
Wired.com, 2011. [Online]. Available: http://www.wired.com/beyond_the_beyond/2011/12/42-major-
countries-ranked-by-smartphone-penetration-rates/. [Accessed: 01-Mar-2014].
[22] I. Olsson, P. Millet, S. Vinberg, G. Olsson, A. Bergroth, and B. J. Landstad, “Social Welfare in
Singapore and Sweden: Differences in Organisational Systems of Health Care, Social Security and
Rehabilitation,” Int. J. Disabil. Manag., vol. 3, no. 01, pp. 30–38, Feb. 2012.
[23] The Swedish Wire, “Stockholm a frontrunner in m-health,” The Swedish Wire, 2011. [Online].
Available: http://www.swedishwire.com/business/9098-stockholm-a-frontrunner-in-m-health.
[24] T. D. Aungst, K. a Clauson, S. Misra, T. L. Lewis, and I. Husain, “How to identify, assess and utilise
mobile medical applications in clinical practice.,” Int. J. Clin. Pract., vol. 68, no. 2, pp. 155–62, Feb.
2014.
[25] happtique, “Happtique - Health and medical app certification program.” [Online]. Available:
http://www.happtique.com/category/press-releases/. [Accessed: 30-Jan-2014].
[26] NHS, “Health Apps Library - NHS Choices,” NHS. [Online]. Available: http://apps.nhs.uk/. [Accessed:
30-Jan-2014].
[27] D. L. SCHER, “Why health and medical apps should be certified.” [Online]. Available:
http://www.kevinmd.com/blog/2012/07/health-medical-apps-certified.html. [Accessed: 29-Jan-2014].
[28] M. Mitka, “FDA lays out rules for regulating mobile medical apps.,” JAMA, vol. 310, no. 17, pp.
1783–4, Nov. 2013.
[29] E. R. Breton, B. F. Fuemmeler, and L. C. Abroms, “Weight loss-there is an app for that! But does it
adhere to evidence-informed practices?,” Transl. Behav. Med., vol. 1, no. 4, pp. 523–9, Dec. 2011.
[30] Research2Guidance, “Research2Guidance’s 4th Global mHealth App developer survey 2014,” 2014.
[Online]. Available: http://r2g.mhealth-app-developer-economics-survey-2014-blog.sgizmo.com/s3/.
[31] K. F. B. Payne, H. Wharrad, and K. Watts, “Smartphone and medical related App use among medical
students and junior doctors in the United Kingdom (UK): a regional survey.,” BMC Med. Inform. Decis.
Mak., vol. 12, no. 1, p. 121, Jan. 2012.
[32] E. M. Rogers, Diffusion of Innovations, 5th ed. New York, New York, USA: The Free Press, 2003.
[33] M. Cain and R. Mittman, “Diffusion of Innovation in Health Care,” Oakland, USA, 2002.
[34] V. Paudyal, D. Hansford, S. Cunningham, and D. Stewart, “Over-the-counter prescribing and
pharmacists’ adoption of new medicines: diffusion of innovations.,” Res. Social Adm. Pharm., vol. 9,
no. 3, pp. 251–62, 2013.
[35] E. M. Rogers, U. E. Medina, M. A. Rivera, and C. J. Wiley, “COMPLEX ADAPTIVE SYSTEMS
AND THE DIFFUSION OF INNOVATIONS,” Innov. J. public Sect. Innov. J., vol. 10, no. 3.
53
[36] I. Vedel, V. Ghadi, M. De Stampa, C. Routelous, H. Bergman, J. Ankri, and L. Lapointe, “Diffusion of
a collaborative care model in primary care: a longitudinal qualitative study.,” BMC Fam. Pract., vol.
14, no. 1, p. 3, Jan. 2013.
[37] M. Saunders, P. Lewis, and A. Thornhill, Research method for business students, 6th ed. Pearson
Education, Inc., 2012.
[38] C. Robson, Real World Research, 2nd ed. Oxford: Blackwell, 2002.
[39] M. Easterby-Smith, R. Thorpe, P. Jackson, and A. Lowe, Management Research, 3rd ed. London::
Sage, 2008.
[40] P. Susanna, “Läkares attityd till hälsoappar undersöks,” Dagens Medicin, 2014. [Online]. Available:
http://www.dagensmedicin.se/nyheter/lakares-attityd-till-halsoappar-undersoks/. [Accessed: 07-May-
2014].
[41] S. Koch, “Achieving Holistic Health for the Individual through Person-Centered Collaborative Care
Supported by Informatics,” Healthc. Inform. Res., vol. 19, no. 1, pp. 3–8, Mar. 2013.
[42] P. Leijdekkers and V. Gay, “Mobile apps for chronic disease management: lessons learned from
myFitnessCompanion®,” Health Technol. (Berl)., vol. 3, no. 2, pp. 111–118, Feb. 2013.
[43] M. Tomlinson, M. J. Rotheram-Borus, L. Swartz, and A. C. Tsai, “Scaling up mHealth: where is the
evidence?,” PLoS Med., vol. 10, no. 2, p. e1001382, Jan. 2013.
[44] T. Martin, “Assessing mHealth: opportunities and barriers to patient engagement.,” J. Health Care
Poor Underserved, vol. 23, no. 3, pp. 935–41, Aug. 2012.
[45] Kevin D.Blanchet, “Medical Apps for Urologists: Usefulness still evolving for physicians and patients,”
UROSCAN.
[46] A. J. Barton, “The regulation of mobile health applications.,” BMC Med., vol. 10, p. 46, Jan. 2012.
[47] F. Haffey, R. R. W. Brady, and S. Maxwell, “A comparison of the reliability of smartphone apps for
opioid conversion.,” Drug Saf., vol. 36, no. 2, pp. 111–7, Feb. 2013.
[48] L. H. Iwaya, M. a L. Gomes, M. a Simplício, T. C. M. B. Carvalho, C. K. Dominicini, R. R. M.
Sakuragui, M. S. Rebelo, M. a Gutierrez, M. Näslund, and P. Håkansson, “Mobile health in emerging
countries: a survey of research initiatives in Brazil.,” Int. J. Med. Inform., vol. 82, no. 5, pp. 283–98,
May 2013.
[49] JEFF LIVINGSTON, “Will doctors recommend health apps to patients?,” kevinMD. [Online].
Available: http://www.kevinmd.com/blog/2013/01/doctors-recommend-health-apps-patients.html.
[Accessed: 01-Feb-2014].
[50] P. Ghauri and K. Gronhaug, Research Methods in Business Studies: A Practical Guide, 4 edition.
Financial Times/ Prentice Hall, 2010, p. 288.
[51] Online QDA, “What is Qualitative Data Analysis (QDA)?,” Onilne QDA, 2012. [Online]. Available:
http://onlineqda.hud.ac.uk/Intro_QDA/what_is_qda.php. [Accessed: 02-Apr-2014].
[52] A. K. Yetisen, J. L. Martinez-Hurtado, F. da Cruz Vasconcellos, M. C. E. Simsekler, M. S. Akram, and
C. R. Lowe, “The regulation of mobile medical applications.,” Lab Chip, vol. 14, no. 5, pp. 833–40,
Mar. 2014.
[53] J. Y. Breland, V. M. Yeh, and J. Yu, “Adherence to evidence-based guidelines among diabetes self-
management apps.,” Transl. Behav. Med., vol. 3, no. 3, pp. 277–86, Sep. 2013.