Accepted Manuscript Title: A Personalised Mobile-based Home Monitoring System for Heart Failure: The SUPPORT-HF Study Author: Andreas Triantafyllidis Carmelo Velardo Tracey Chantler Syed Ahmar Shah Chris Paton Reza Khorshidi Lionel Tarassenko Kazem Rahimi PII: DOI: Reference:
S1386-5056(15)00096-9 http://dx.doi.org/doi:10.1016/j.ijmedinf.2015.05.003 IJB 3192
To appear in:
International Journal of Medical Informatics
Received date: Revised date: Accepted date:
3-11-2014 27-12-2014 13-5-2015
Please cite this article as: A. Triantafyllidis, C. Velardo, T. Chantler, S.A. Shah, C. Paton, R. Khorshidi, L. Tarassenko, K. Rahimi, on behalf of the SUPPORT-HF Investigators, ¨ 1 A Personalised Mobile-based Home Monitoring System for Heart Failure: The SUPPORT-HF Study, International Journal of Medical Informatics (2015), http://dx.doi.org/10.1016/j.ijmedinf.2015.05.003 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
A
Personalised
Mobile-based
Home
Monitoring
ip t
System for Heart Failure: The SUPPORT-HF Study
cr
Andreas Triantafyllidisa, Carmelo Velardoa, Tracey Chantlerb, Syed
us
Ahmar Shaha, Chris Patonb, Reza Khorshidib, Lionel Tarassenkoa,
a
an
Kazem Rahimib,c, on behalf of the SUPPORT-HF Investigators*
Institute of Biomedical Engineering, Department of Engineering Science, University of Oxford, Oxford,
b
M
UK
George Institute for Global Health, Nuffield Department of Population Health, University of Oxford, Oxford, UK
d
The full list of SUPPORT-HF Investigators and Collaborators is provided at the end of the manuscript.
Ac ce p
*
Division of Cardiovascular Medicine, Radcliffe Department of Medicine, University of Oxford, Oxford, UK
te
c
Address for Correspondence: Dr Andreas Triantafyllidis
Institute of Biomedical Engineering
Department of Engineering Science University of Oxford Oxford OX3 7DQ, UK. Tel.: +44(0)1865617679
E-mail:
[email protected]
1
Page 1 of 27
Abstract
ip t
Background Despite their potential for improving health outcomes, mobile-based home monitoring systems for heart failure have not yet been taken up widely by the patients and providers.
cr
Objectives To design and iteratively move towards a personalised mobile health monitoring system for patients living with heart failure, according to their health care and usability needs.
us
Methods We present an iterative approach to refining a remote health monitoring system that is based on interactions between different actors (patients, clinicians, social scientists and engineers)
an
and supports the collection of quantitative and qualitative information about user experience and engagement. Patients were provided with tablet computers and commercially available sensing
M
devices (a blood pressure monitor, a set of weighing scales, and a pulse oximeter) in order to complete physiological measurements at home, answer symptom-specific questionnaires, review
d
their personal readings, view educational material on heart failure self-management, and
te
communicate with their health professionals. The system supported unobtrusive remote software
Ac ce p
upgrades via an application distribution channel and the activation or deactivation of functional components by health professionals during run-time operation. We report early findings from the application of this approach in a cohort of 26 heart failure patients (mean age 72 ± 15 years), their caregivers and healthcare professionals who participated in the SUPPORT-HF (Seamless Usercentred Proactive Provision Of Risk-stratified Treatment for Heart Failure) study over a one-year study period (mean patient follow-up duration=270 ± 62 days). Results The approach employed in this study led to several system upgrades dealing in particular with patient requirements for better communication with the development team and personalised self-monitoring interfaces. Engagement with the system was constantly high throughout the study and during the last week of the evaluation, 23 patients (88%) used the system at least once and 16 patients (62%) at least three times.
2
Page 2 of 27
Conclusions Designers of future mobile-based home monitoring systems for heart failure and other chronic conditions could leverage the described approach as a means of meeting patients’ needs
ip t
during system use within the home environment and facilitating successful uptake.
Keywords: Mobile Health, Remote Health Monitoring, Personalised Healthcare, Usability,
cr
Telemedicine, Heart Failure
us
1. Introduction
an
Provision of care for chronic diseases outside hospitals and GP practices is often fragmented and inadequate [1]. For this reason, several telemonitoring systems have recently been introduced
M
for close monitoring of patient health status, clinical assessment, and early intervention in out-ofhospital settings [2]. Such systems typically integrate devices that can monitor various physiological
d
parameters such as blood pressure, heart rate, and temperature. More recently, a plethora of
te
remote health monitoring systems have utilized mobile computing technologies in order to enable
Ac ce p
the remote capture, transmission and processing of health information. It is widely assumed that the adoption of such systems can lead to positive impacts on patient self-efficacy, self-management behaviour and quality of life, which in turn could lead to reduction in hospital readmission and mortality rates for a number of chronic diseases [3], [4]. Heart failure is a long-term condition characterized by periods of clinical deterioration which
requires close monitoring, assessment, and treatment. Therefore, telemonitoring technologies may be particularly suited to assist heart failure patients because they allow health care professionals to assess a patient’s condition remotely, review their care and be more alert to physiological changes that may require medical intervention [5]. At the same time, patients can take a more active role in their care by gaining skills in self-monitoring, which can help them to recognise significant changes in their health status and communicate these with their care providers [6].
3
Page 3 of 27
Despite their potential to improve health outcomes, remote health monitoring systems for heart failure have not yet been widely adopted by patients, which limits their role in current clinical practice [7]. Chaudhry et al [8] found that only 55% of 826 patients continued to use a telephone-
ip t
based system at least three times a week, by the end of a 6-month study. This illustrates the challenges faced when trying to ensure the long-term engagement of patients in the application of
cr
telemonitoring interventions. Similarly Domingo et al [9], found that less than half (n=97) of 211 patients who were initially screened, participated in a one-year study involving the use of a
us
television-based platform for education and self-monitoring, and 68 patients (70%) were able to complete the study. Other studies utilizing mobile computing technologies have also reported
an
limited adoption and low long-term adherence or user engagement rates [10]–[12]. Reasons for the limited uptake and practical use of home monitoring systems over the long-
M
term include usage difficulties and low levels of patient satisfaction. These at least partly relate to the limited involvement of users and healthcare professionals in the design of the systems as well as
d
the restricted tailoring options to adapt the system to changing requirements [13]–[15]. In this
te
context, we believe that a vital element for the successful implementation of these interventions is
Ac ce p
being ignored: the iterative personalisation of health monitoring services according to the patient’s ongoing healthcare and usability needs. Given that telemonitoring and telemedicine interventions are described as “complex and ongoing collaborative achievements in unpredictable processes” [16], patient-centred approaches for sustainable delivery of remote health monitoring services are required [17].
We used an iterative approach to refining a tablet computer-based home monitoring system
for heart failure patients based on their experience while interacting with it. The mobile system enables patients to complete physiological measurements at home through commercially available sensing devices, answer symptom-specific questionnaires, review their personal readings, view educational material, and initiate communication with their healthcare professionals. This approach is based on features that are uniquely combined. First, in contrast with other research focusing on
4
Page 4 of 27
system design prior to system use and deployment [18]–[20], we adopt an iterative approach which shifts the focus of system design onto patient needs after system deployment and during system use in a real-life setting. This requires synergy between patients, clinicians, social scientists and
ip t
engineers, as well as the collection of both quantitative and qualitative data regarding user experience while interacting with the system. More specifically, patient monitoring requirements
cr
and usability needs are identified through analysis of observations from home visits, automated collection of user interactions, and telephone interviews. Secondly, the delivery of refinements to
us
the patient’s tablet computer takes place remotely and unobtrusively, through a private distribution channel. In addition, functional features can be activated or deactivated by clinicians during system
an
run-time operation.
System Refinement Methodology
d
2.1
M
2. Methods
te
The iterative refinement approach was informed by “action research” [21] and “agile
Ac ce p
software development” [22] principles, aiming for continual patient satisfaction, and requiring collaboration. According to the action research paradigm, a process of participatory inquiry by both patients and providers is followed in order to help understand the problems and provoke change [23]. In contrast to traditional software development approaches which are based on initial extensive planning, in agile development methodologies, design and implementation are inseparable and evolve iteratively, encouraging rapid response to requirements [24]. In our approach, qualitative and quantitative methods are combined. This leverages the value of qualitative methods to explain and generate hypotheses, and the ability of quantitative methods to test hypotheses [25], enabling the identification of usability requirements and their subsequent validation [26], [27].
5
Page 5 of 27
As seen in Table 1, the refinement of the system during its use within the home environment Table 1. Instruments, goals, and actors within the system’s refinement cycles
Patient routine visits to doctor & remote capture of measurements Face-to-face communication in the community Instruments setting (e.g., semi-structured interviews with patients in their home) or telephone interviews
Actors
Patients, Clinicians, Social Scientists
Private application distribution channel (e.g., “application store”) Remote activation /deactivation of application features
an M
Evaluation Telephone interviews (initiated either by the patient or the development team) Face-to-face communication in the community setting (e.g., semistructured interviews with patients in their homes) Automatic collections of user interactions
Assessment of user engagement, system effectiveness, efficiency, and user satisfaction
Engineers
Engineers, Clinicians
Clinicians, Engineers, Social Scientists, Patients
d
Development of application features according to identified patient needs
System refinements after development of new or modified features System refinements during runtime operation
te
Ac ce p
Goals
Assessment of monitoring requirements (e.g., sensors used, measurement frequency) Identification of patient concerns and usability issues
ICT tools & mobile health equipment
Remote delivery of system refinements
ip t
Development
cr
Understanding patients’ monitoring needs
us
Cycle
is the result of collaboration between clinicians, social scientists and software engineers (the “development team”) and patients. Patients are the primary users of the system. Typically, clinicians utilize the system in order to identify patient needs and health problems. However, in this research, the investigation of problems in the usability context also constitutes part of their tasks. This assessment is covered by social scientists interacting with the patients and systematically observing the system usefulness, ease of use and patient satisfaction, enabling to acquire a deep
6
Page 6 of 27
understanding of personalised requirements. The engineers develop, integrate, test and deliver the software components of the system. Four overlapping and complementary cycles can be distinguished, executed iteratively within the system’s life cycle, as described below (Table 1):
ip t
(i) Understanding patients’ monitoring needs: This cycle includes iterative qualitative and quantitative assessments of the patient’s monitoring requirements by the development team. The
cr
iterative approach is necessary because the patient’s monitoring requirements may change over time (e.g., monitoring devices used, frequency of system use, requested service features) during
us
system run-time operation. In scheduled clinical appointments, clinicians are able to obtain patients’ physiological data and perform examinations, assess their health condition and develop an
an
understanding of the patient’s monitoring needs (e.g., the type and frequency of remote measurements required). This process is also facilitated by the clinician’s remote review of
M
physiological measurements within the system [16]. At the same time, patients can communicate their concerns to the development team, during their routine visits to the clinician, during home
d
visits when a social scientist observes the use of the equipment and interviews patients about their
te
experience, or during subsequent follow-up telephone interviews. In informal face-to-face
Ac ce p
communication, feedback for tailoring and improving the system or correcting possible design flaws can be obtained. More importantly, particular usability problems that patients may face can be identified given the fact that not all of them have the same level of technological skill and some may have a slight visual or motor deficit [27]. (ii) Development: After collaboratively identifying the changing monitoring requirements, the development of the mobile health monitoring service takes place. According to Cao and Ramesh [28], there are two types of requirement changes: adding or dropping features, and modifying already implemented features. Thus during this cycle, new functional features are developed and added by the engineers, while existing features are improved or removed to attain patient satisfaction. The refinement of non-functional features which are not directly related to system functions, encompassing e.g., security, performance, and data quality aspects is also equally
7
Page 7 of 27
important for the efficient and sustainable use of the system [29]. Design, coding, encapsulation into software applications and unit testing, are all included in this cycle. (iii) Delivery: The delivery of the health monitoring equipment (i.e., mobile and sensing devices)
ip t
takes place during the clinician’s (or nurse’s) visit to the patient’s home (home screening visit). At this visit there is also a walk-through presentation of the system’s features. Following the home
cr
screening visit, remote refinements of the system can be initiated, thereby saving valuable human resources. The delivery of the mobile health monitoring refinements takes place unobtrusively, and
us
includes:
1) Delivery of the developed refinements encapsulated in the mobile application through an
an
Internet-linked distribution channel, i.e., a well-structured repository of applications as currently provided in the form of an “application store” by major companies in the mobile industry (e.g.,
ongoing mobile health service delivery [30].
M
Google, Apple, Microsoft, etc.). Such repositories can provide the means for private, secure, and
d
2) Delivery of refinements during system run-time operation. When appropriate, functional features
te
can be activated or deactivated by the clinicians according to specific monitoring requirements. As
Ac ce p
an example, the functional features associated with a technological device which may not be suitable for a specific patient after a time period can be deactivated remotely by the clinician during system run-time operation.
The users should also be notified of any new functionality coming with service refinements, as well as any possible user interface changes, in order to avoid confusion. (iv) Evaluation: The system’s evaluation corresponds to the systematic exploration of usability. According to Scandurra et al [26], usability is determined by identifying system effectiveness, system efficiency, and user satisfaction. In our work, we also consider user engagement with the system as an additional important factor for usability. This can be quantified by recording how often specific system features are used, indicating the extent to which the system is being adopted [31], [32]. Usability evaluation is achieved through qualitative and quantitative information regarding user
8
Page 8 of 27
experience in three dimensions: (1) non-scheduled patient-initiated feedback, (2) scheduled home visits, and (3) system usage logs. Non-scheduled patient-initiated feedback occurs when a patient requests the development team to contact him/her (usually a telephone call) to provide technical
ip t
support. Scheduled home visits involve a social scientist observing the patients’ use of the system in their home environment, and interviewing them about their experience of adopting new technology
cr
for monitoring their health condition. These types of qualitative methods have demonstrated their potential in identifying usability problems and implementing helpful changes [33]. System usage logs
us
capture user-interactions of the different features (e.g., logs of accesses to a specific functional component, duration of user tasks, critical incidents, etc.), aiming at quantifying patient engagement
an
with the system and guiding improved system design and development [34]. By combining all of these methods, the degree of patient satisfaction as well as patient engagement with the system can
SUPPORT-HF System Overview
d
2.2
M
be thoroughly explored.
te
2.2.1 Initial Design & System Architecture
Ac ce p
The methodology described above was used in the SUPPORT-HF (Seamless User-centred
Proactive Provision Of Risk-stratified Treatment for Heart Failure) cohort study with the aim of developing a remote health monitoring and non-pharmacological self-monitoring system for heart failure patients. The initial design of the system was based on: (1) initial requirement specification and comprehensive review by two clinicians, one social scientist, and three engineers; and (2) a codesign workshop with patients demonstrating multiple differentiated functional mock-up screens. The system integrates a tablet computer, used as a front-end and communication gateway for the patients, and various sensing devices including a blood pressure monitor, a set of weighing scales, and a pulse oximeter (Fig. 1). Bluetooth is used for delivering monitoring data from the sensors to a tablet computer, which in turn transmits the data through the Internet to a back-end server
9
Page 9 of 27
Blood Pressure & Heart Rate
ip t
Internet Bluetooth
cr
Weight
Patient Front-end Device & Communication Gateway
Ac ce p
(a)
te
d
M
an
Figure 1. System overview
Back-end
us
SpO2 & Heart Rate
(c)
(b)
(d)
Figure 2. Distinct functional components on tablet computer: (a) self-monitoring (feature of blood pressure selfmonitoring), (b) self-management education through videos, (c) review of personal readings for self-tracking, (d) communication with health professionals (feature of patient-initiated communication with health professionals)
infrastructure for storage, processing, and display to the clinicians. 2.2.2 Patient Interface
10
Page 10 of 27
Patients were enabled to complete symptom-specific questionnaires and take measurements, view educational material (e.g., videos and documents), review their personal recordings in graphical displays for self-tracking purposes, and communicate with their clinicians, by
ip t
accessing distinct functional components through touch-screen interfaces (Fig. 2). Each component is comprised of one or more features. For example, the self-monitoring component includes six
cr
features; the New York Heart Association (NYHA) [35], the EuroQol (EQ-5D) [36], and the Minnesota Living With Heart Failure (MLWHF) [37] questionnaires, as well as the blood pressure, weight, and
us
pulse oximeter self-monitoring features. The components were developed on the Android programming platform [38] and then integrated in the software application which was remotely acting as a private and unobtrusive
an
delivered to the patients through “Google Play Store”
M
application distribution channel. 2.2.3 Clinician Interface
d
A back-end web-based application (Fig. 3) authorised clinicians and nurses to do the
te
following: (1) view the patient data graphically, (2) add notes for reporting findings on both the
Ac ce p
medical status and usability of the system, after communication with patients, and send them via email to other members of the development team, (3) send personalised messages to the patients, (4) activate or deactivate features of the self-monitoring component for a specific patient during system run-time operation, (5) add the required patient demographic, clinical, or technical information. Clinicians and nurses were also provided with a web-based interface to assess the recorded patient interactions with the system, so as to detect the feature usage and identify possible usability problems during system adoption. Such recorded interactions included access logs of specific features (e.g., symptom-specific diary/questionnaire, blood pressure self-monitoring, etc.), incomplete self-monitoring sessions, tap of specific buttons (e.g., “help”, “back” or “exit”
11
Page 11 of 27
ip t cr us an M
d
Figure 3. Back-end system: Display of notes for patient status, and measurement graphs
te
button), and dismissal of displayed messages.
Ac ce p
2.2.4 Non-functional Issues
The system dealt with several non-functional requirements in order to handle network
errors, retain data accuracy, and preserve patient privacy. Specifically, the technical issue of intermittent Internet connections observed in previous studies [39], [40] was solved by applying a data synchronisation mechanism activated every 30 minutes. This mechanism ensured that unsent data from the tablet computer to the back-end infrastructure due to Internet connection failures could be transmitted again whenever synchronisation was possible. In order to maximise the quality of the data captured, recorded patient information was checked during self-monitoring sessions and values outside valid ranges were filtered out. Furthermore, the security protocol Transport Layer Security (TLS) [41] and key-based encryption to patient identifiable data were utilised, in order to
12
Page 12 of 27
secure sensitive personal data communicated over the Internet to the back-end system and prevent possible eavesdropping and tampering.
ip t
2.2.5 System Usage & Patients’ Eligibility
The methodology was used in the SUPPORT-HF observational study for which 26 patients, 17
cr
male and 9 female, with mean age 72 ± 15 years, have used the system at home for 270 ± 62 days. According to our inclusion criteria, patients were to be aged 18 years and above, diagnosed with
us
heart failure, and willing to give informed consent for participation in the study. Potentially eligible patients were recruited by their direct care team during scheduled clinic visits. Patients were
an
informed about the purpose of the study and upon receiving their written consent for participation, a home screening visit by a clinician was followed. Ethics approval was obtained from the NHS
M
Health Research Authority (NRES Committee South Central – Oxford, Reference: B13/SC/0125). The
Ac ce p
3. Results
te
patient’s home.
d
study endpoint for every patient was based on the day a final interview was conducted at the
3.1
System Refinement Examples
Better communication with the development team In month 4 of the study, and after several scheduled visits to patients’ homes for exploring their views on interaction with the system (during the “Understanding patients’ monitoring needs” cycle), the development team realised the necessity to provide a mechanism for rapid support and communication whenever patients faced technical difficulties or usability issues. During the face-toface interviews, patients often reported a number of difficulties when using the system. The development team realised that many of the issues raised could be dealt more effectively, if users had the option of reporting their problems as soon as they occurred. To overcome this, a functional
13
Page 13 of 27
feature was added to the communication component (Fig. 2d), enabling patients to initiate asynchronous communication with the members of the development team, simply by pressing a button (“Development” cycle). The development team was then notified of the occurrence of this
ip t
request via both email and SMS (Short Messaging Service), and it was able to get in touch with the patient, usually by phone. This functional feature was developed by the software engineers,
cr
integrated within the software application, and finally delivered remotely to the patients’ tablet computers through the application distribution channel, “Google play store” (“Remote delivery of
us
health monitoring service” cycle). In total, 29 communication requests were initiated by 12 patients, in order to report issues mainly related to device connectivity (“Evaluation” cycle). The
an
communication request feature enabled the development team to respond quickly to usability problems, while this fast-learning process led also to the development of a troubleshooting guide
M
with step-by-step instructions on how to resolve Bluetooth and Internet connectivity issues.
d
Personalised self-monitoring interface with sensors
te
Early on during the course of the study, it became apparent that the pulse oximeter was not suitable
Ac ce p
for one patient, due to finger tremor that prevented successful completion of the self-monitoring sessions. The clinician was informed of this issue through a telephone conversation following a patient communication request with the development team (“Understanding patients’ monitoring needs” cycle). The clinician added a note regarding the problem using the back-end web-based application and sent it via email to the engineering team who verified that several oximetry selfmonitoring sessions were flagged as unsuccessful in usage logs for more than a week. Soon afterwards, the remote deactivation of the pulse oximeter self-monitoring feature was initiated by the clinician for that specific patient (“Remote delivery of health monitoring service” cycle). The patient was also notified with a personalised message reporting the new change in the user interface. After ethnographic observations made by the social scientist during her visit to the
14
Page 14 of 27
patient’s home, the cause of the problem (finger tremor) was identified and the patient expressed his satisfaction with the new change (“Evaluation” cycle).
ip t
Self-monitoring preferences for questionnaires During an interview, one patient, aged 94, expressed his frustration when completing the MLWHF
cr
questionnaire - activated periodically every 3 months -, because this was rather time consuming and required answering a series of 21 questions in total, as opposed to the daily NYHA questionnaire
us
which consists of 5 questions maximum (“Understanding patients’ monitoring needs” cycle). This distracted the patient from proceeding with normal usage of the system, since the process of
an
responding to the MLWHF questions could take up to 20 minutes (“Evaluation” cycle). To overcome this problem, an initial screen for the questionnaire was introduced, explaining to the users that
M
answering the MLWHF questionnaire was a long process requiring special attention, and providing them with the option to postpone it until the next time they used the system (“Development” cycle).
d
The changes to this feature were then delivered remotely through the application distribution
Ac ce p
te
channel (“Remote delivery of health monitoring service” cycle).
Non-functional requirements
Non-functional requirements relating to the quality of data from self-monitoring and the communication efficiency with the back-end infrastructure were also addressed after system deployment. The data synchronisation mechanism for communicating with the back-end server was improved to deal with the fluctuation of the strength of the Internet connection in many homes. The synchronisation process was designed to run in the background, automatically checking the presence of Internet connectivity and the availability of data to be synchronised, and finally seamlessly connecting to the back-end server to transfer the patient data.
3.2
Evaluation of System Usage
15
Page 15 of 27
The overall evaluation of the system usage corresponded to the analysis of the recorded usage logs in the SUPPORT-HF study and the quantification of (1) patient adherence (i.e., engagement with the system), (2) system effectiveness, and (3) system efficiency. The systematic
ip t
qualitative assessment of the patient satisfaction with the SUPPORT-HF system, according to the “Evaluation” cycle of our development approach, will be presented in the final study paper.
cr
During the last week of system usage (based on the study endpoint for every patient), it was
Ac ce p
te
d
M
an
us
found that 23 patients (88%) used it at least once, while 16 patients (62%) used it at least 3 times. In
Figure 4. Mean adherence in using the system as the number of days per week, for which patients completed at least one self-monitoring session.
order to identify adherence throughout the whole study duration, we calculated the mean number of days during a week, in which patients completed at least one self-monitoring session (either questionnaire-based or sensor-based) during a single day. As presented in Fig. 4, the mean adherence in using the system was constantly high (mean 4.96 ± 0.50). System effectiveness was considered as the percentage of successfully completed selfmonitoring sessions involving three steps as recorded within the usage logs: (1) initiation of the session by tapping the relevant button on the tablet computer, (2) message displayed on the tablet computer screen providing the patients with their measurement results, and (3) results successfully
16
Page 16 of 27
transmitted to the back-end system. These percentages were 98%, 99%, and 99% for the blood pressure, weight, and oximetry sessions respectively. The system efficiency was considered as the percentage of successfully completed sessions at the first attempt, to illustrate the efficiency in
ip t
making use of resources such as the user’s time and effort (a task not successfully completed at the first attempt obviously requires further time and effort from the user before successful completion).
cr
These percentages were 96%, 98%, and 97% for the blood pressure, weight, and oximetry sessions
us
respectively.
4. Discussion
an
The current design of health information systems typically includes an initial stage of requirement definition carried out by healthcare professionals and health informatics specialists
M
[26]. User involvement in the design and development of applications in medical informatics has also been emphasized in the literature [42][43]. This can bring benefits such as improved
d
communication between users and technology providers, increase of the likelihood of meeting
te
users’ needs, and decrease of system failures [44].
Ac ce p
Despite the research efforts in participatory design, co-design, and co-realisation of medical informatics especially from a sociotechnical perspective [45], little has been done to realise user involvement, beyond the construction of early prototypes [46][47]. In this context, prototype health information systems have been developed for a number of diseases or conditions such as diabetes [48], obesity [49] and heart failure [50]. A variety of established methods such as usability tests, focus groups, and in-depth interviews [27] can be used to involve users and solicit feedback in order to assess system usefulness and ease of use. However, user-centred and personalised approaches aiming to improve system features according to user requirements identified during system use in real-life settings have not been applied systematically. This is surprising, given that the healthcare domain is complex, technology is evolving rapidly, and the requirements and/or preferences of both
17
Page 17 of 27
patients and healthcare professionals can change substantially over time. Regular adaptation to changing circumstances is therefore necessary for widespread and sustained adoption. The unmet need for personalised systems for remote health monitoring has motivated our
ip t
approach, in which a tablet computer system for monitoring heart failure was refined iteratively by a multi-disciplinary team with the active involvement of patients and health professionals.
cr
Refinements were gradually introduced, mainly based on observing the patients’ use of the system and the interactions between the development team and the patients. The SUPPORT-HF study
us
showed the importance and value of involving patients in identifying factors which can potentially influence system usability over time.
an
The examples of system refinements given in the paper show the virtue of the iterative refinement methodology for system personalisation. Features such as communication with the
M
development team and customised user interfaces for personalised self-monitoring were introduced after receiving user feedback. Qualitative information retrieved e.g., from scheduled semi-structured
d
interviews or phone conversations, and quantitative information retrieved from automatic
te
collections of user interactions, helped to understand patients’ usability needs.
Ac ce p
The remote delivery of health monitoring service refinements to patients’ tablet computers was found to be robust and had no negative effect on user engagement with the system, and system effectiveness and efficiency. A commercially available “application store” was used as an enabling technology to deliver system upgrades unobtrusively. Activation or deactivation of the system’s functional features by healthcare professionals also enabled the personalisation of the selfmonitoring experience, especially when usability problems with specific devices were identified. Patients’ engagement with the system was found to be high throughout the study. It is our belief that the iterative refinement approach during system use contributed significantly to the high levels of adherence from the patients. Other factors with an impact on patient satisfaction and engagement such as the usability of the sensing devices which were utilised in the study and patient self-efficacy, were not analysed in this study [3].
18
Page 18 of 27
Further studies are needed to assess the efficacy and cost-effectiveness of iterative development approaches. In particular, our approach was based on frequent interactions between patients and the development team, realised through a variety of means. For systems which cannot
ip t
benefit from such interactions, this approach could “pose risks such as requirements that are inadequately developed or, worse still, wrong” [28]. Usability testing studies in controlled
cr
environments after system deployment could facilitate to better understand the usability problems that might be encountered by patients and providers [51]. Conducting such studies with a selected
us
number of participants is necessary before applying major system refinements that require thorough investigation of their usefulness and ease of use.
an
In conclusion, a systematic approach for the refinement and delivery of a health monitoring system for heart failure patients during its use within the home environment was presented. The
M
approach involves interactions between patients, clinicians, social scientists and engineers in order to better understand the patients’ experience while using the system, and enable useful refinements
d
to system features. Designers of future mobile-based home monitoring systems for chronic
te
conditions could leverage the described approach as a possible means of achieving successful uptake
Ac ce p
by the patients, thereby meeting their needs on a long-term basis.
Authors’ Contributions
Authors LT and KR were responsible for the SUPPORT-HF study conception; All authors participated in system design; AT, CV, and SA developed the system and constantly refined its features according to the described methodology; all authors analysed the data; AT wrote a first draft of the manuscript and all other authors contributed to the final version. All authors have read and agreed to the paper being submitted as it is.
19
Page 19 of 27
SUPPORT-HF Collaborators and Investigators
Steering Committee: Felicity Emptage (patient representative), Prof John Cleland (Imperial College
ip t
London), Dr Tracey Chantler (George Institute for Global Health, University of Oxford), Prof Andrew Farmer (Department of Primary Care Health Sciences, University of Oxford), Prof Ray Fitzpatrick
cr
(Department of Public Health, University of Oxford), Prof Richard Hobbs (Department of Primary
Care Health Sciences, University of Oxford), Prof Stephen MacMahon (George Institute for Global
us
Health, University of Oxford), Dr Chris Paton (George Institute for Global Health, University of
an
Oxford), Alan Perkins (patient representative), Prof Kazem Rahimi (Chief Investigator, George Institute for Global Health, University of Oxford), Dr Syed Ahmar Shah (Institute of Biomedical Engineering, University of Oxford), Prof Lionel Tarassenko (Institute of Biomedical Engineering,
M
University of Oxford), Dr Andreas Triantafyllidis (Institute of Biomedical Engineering, University of Oxford), Dr Carmelo Velardo (Institute of Biomedical Engineering, University of Oxford), Prof Mark
te
d
Woodward (George Institute for Global Health, University of Oxford).
Ac ce p
Other investigators: Dr Paul Altmann (Oxford University Hospitals NHS Foundation Trust), Dr Badri Chandrasekaran (Great Western Hospitals NHS Foundation Trust, Swindon), Nathalie Conrad (George Institute for Global Health), Dr Jeremy Dwight (Oxford University Hospitals NHS Foundation Trust), Johanna Ernst (Institute of Biomedical Engineering), Dr Paul Foley (Great Western Hospitals NHS Foundation Trust, Swindon), Prof Alastair Gray (Health Economics Research Centre, University of Oxford), Joanne Noble (Heart Failure Clinical Lead, Heart Failure Community Nursing Service, Oxford Health NHS Foundation Trust).
Acknowledgements
20
Page 20 of 27
The work described in this paper was supported by the NIHR Biomedical Research Centre programme, Oxford and the George Institute for Global Health. Dr Kazem Rahimi is funded by an NIHR Career Development Fellowship.
ip t
Educational material within the described system was extracted from heartfailurematters.org and was reproduced with kind permission of the European Society of Cardiology and the British Heart
cr
Foundation. We are also grateful to the Health Experiences Research Group, Department of Primary Care Health Sciences, University of Oxford for sharing interview extracts with us from their study of
us
heart failure, which are also available on the www.healthtalkonline.org website.
an
We would like to thank our study participants for their support of our research.
M
Conflicts of Interest
te
Ac ce p
Summary Points
d
The authors of this manuscript declare no conflicts of interest.
What was already known on the topic?
Telemonitoring technologies may assist heart failure patients because they allow health care professionals to assess a patient’s condition remotely, and enable patients to take a more active role in their care.
Although several studies have demonstrated that home monitoring systems for chronic conditions including heart failure may have positive clinical outcomes for the patients, these systems have not yet enjoyed widespread uptake by patients.
21
Page 21 of 27
What this study added to our knowledge? To the best of our knowledge, there has been no systematic approach to the iterative refinement and delivery of health monitoring systems for heart failure, tailored to patients’
ip t
changing needs during system use within the home environment. The provision of a personalised mobile-based remote health system for heart failure is
cr
feasible through applying an iterative refinement approach during system use, which involves interactions between patients, clinicians, social scientists and engineers and the
us
collection of both quantitative and qualitative data regarding patient experience and usability needs.
an
The use of a private on-line “application store” as well as the activation or deactivation of system functional features by clinicians during run-time operation may be considered as an
M
effective means of delivering health monitoring system upgrades remotely. Designers of mobile health monitoring systems in future studies could leverage the iterative
te
Ac ce p
long-term basis.
d
approach proposed here, as a possible means of achieving successful uptake by patients on a
References
[1] [2]
O. Gröne and M. Garcia-Barbero, “Integrated care: a position paper of the WHO European Office for Integrated Health Care Services.,” Int. J. Integr. Care, vol. 1, p. e21, Jan. 2001.
S. Koch, “Home telehealth--current state and future trends,” Int. J. Med. Inform., vol. 75, no. 8, pp. 565–576, 2006.
[3]
C. K. L. Or and B.-T. Karsh, “A systematic review of patient acceptance of consumer health information technology.,” J. Am. Med. Inform. Assoc., vol. 16, no. 4, pp. 550–60, Jan. 2009.
[4]
M. E. Wilcox and N. K. Adhikari, “The effect of telemedicine in critically ill patients: systematic review and meta-analysis.,” Crit. Care, vol. 16, no. 4, p. R127, Jul. 2012.
22
Page 22 of 27
I. H. Kraai, M. L. A. Luttik, R. M. de Jong, T. Jaarsma, and H. L. Hillege, “Heart Failure Patients Monitored With Telemedicine: Patient Satisfaction, a Review of the Literature,” J. Card. Fail., vol. 17, no. 8, pp. 684–690, 2011.
[6]
T. Andersen, P. Bjørn, F. Kensing, and J. Moll, “Designing for collaborative interpretation in telemonitoring: re-introducing patients as diagnostic agents.,” Int. J. Med. Inform., vol. 80, no. 8, pp. e112–26, Aug. 2011.
[7]
C. Sanders, A. Rogers, R. Bowen, P. Bower, S. Hirani, M. Cartwright, R. Fitzpatrick, M. Knapp, J. Barlow, J. Hendy, T. Chrysanthaki, M. Bardsley, and S. P. Newman, “Exploring barriers to participation and adoption of telehealth and telecare within the Whole System Demonstrator trial: a qualitative study.,” BMC Health Serv. Res., vol. 12, no. 1, p. 220, Jan. 2012.
[8]
S. I. Chaudhry, J. A. Mattera, J. P. Curtis, J. A. Spertus, J. Herrin, Z. Lin, C. O. Phillips, B. V Hodshon, L. S. Cooper, and H. M. Krumholz, “Telemonitoring in patients with heart failure.,” N. Engl. J. Med., vol. 363, no. 24, pp. 2301–9, Dec. 2010.
[9]
M. Domingo, J. Lupón, B. González, E. Crespo, R. López, A. Ramos, A. Urrutia, G. Pera, J. M. Verdú, and A. Bayes-Genis, “Evaluation of a telemedicine system for heart failure patients: feasibility, acceptance rate, satisfaction and changes in patient behavior: results from the CARME (CAtalan Remote Management Evaluation) study.,” Eur. J. Cardiovasc. Nurs., vol. 11, no. 4, pp. 410–8, Dec. 2012.
[10]
N. Tatara, “Long-Term Engagement With a Mobile Self-Management System for People With Type 2 Diabetes,” JMIR mhealth and uhealth, vol. 1, no. 1. JMIR mhealth and uhealth, p. e1, 27-Mar-2013.
[11]
Z. Faridi, L. Liberti, K. Shuval, V. Northrup, A. Ali, and D. L. Katz, “Evaluating the impact of mobile telephone technology on type 2 diabetic patients’ self-management: the NICHE pilot study.,” J. Eval. Clin. Pract., vol. 14, no. 3, pp. 465–9, Jun. 2008.
[12]
D. A. Hanauer, K. Wentzell, N. Laffel, and L. M. Laffel, “Computerized Automated Reminder Diabetes System (CARDS): e-mail and SMS cell phone text messaging reminders to support diabetes management.,” Diabetes Technol. Ther., vol. 11, no. 2, pp. 99–106, Feb. 2009.
[13]
B. Maric, A. Kaan, A. Ignaszewski, and S. A. Lear, “A systematic review of telemonitoring technologies in heart failure.,” Eur. J. Heart Fail., vol. 11, no. 5, pp. 506–17, May 2009.
[14]
K. Wac, “Smartphone as a personal, pervasive health informatics services platform: literature review,” Yearb. Med. Inform., vol. 7, no. 1, pp. 83–93, 2012.
[15]
S. A. Mulvaney, L. M. Ritterband, and L. Bosslet, “Mobile intervention design in diabetes: review and recommendations.,” Curr. Diab. Rep., vol. 11, no. 6, pp. 486–93, Dec. 2011.
[16]
A. G. Ekeland, A. Bowes, and S. Flottorp, “Effectiveness of telemedicine: A systematic review of reviews,” Int. J. Med. Inform., vol. 79, no. 11, pp. 736–771, 2010.
[17]
M. M. Davis, M. Freeman, J. Kaye, N. Vuckovic, and D. I. Buckley, “A systematic review of clinician and staff views on the acceptability of incorporating remote monitoring technology into primary care.,” Telemed. J. E. Health., vol. 20, no. 5, pp. 428–38, May 2014.
Ac ce p
te
d
M
an
us
cr
ip t
[5]
23
Page 23 of 27
J. A. Cafazzo, M. Casselman, N. Hamming, D. K. Katzman, and M. R. Palmert, “Design of an mHealth app for the self-management of adolescent type 1 diabetes: a pilot study.,” J. Med. Internet Res., vol. 14, no. 3, p. e70, Jan. 2012.
[19]
D. Sánchez-Morillo, M. Crespo, A. León, and L. F. Crespo Foix, “A novel multimodal tool for telemonitoring patients with COPD.,” Inform. Health Soc. Care, Dec. 2013.
[20]
J. Vermeulen, J. C. Neyens, M. D. Spreeuwenberg, E. van Rossum, W. Sipers, H. Habets, D. J. Hewson, and L. P. de Witte, “User-centered development and testing of a monitoring system that provides feedback regarding physical functioning to elderly people.,” Patient Prefer. Adherence, vol. 7, pp. 843–54, Jan. 2013.
[21]
C. Munn-Giddings and R. Winter, A Handbook for Action Research in Health and Social Care. 2013.
[22]
P. MESO and R. JAIN, “Agile software development : Adaptive systems principles and best practices,” Inf. Syst. Manag., vol. 23, no. 3, pp. 19–30.
[23]
G. N. Hearn and M. Foth, “Action Research in the Design of New Media and {ICT} Systems,” in Topical Issues in Communications and Media Research, K. Kwansah-Aidoo, Ed. New York, {NY}: Nova Science, 2005, pp. 79–94.
[24]
T. Dyba and T. Dingsoyr, “What Do We Know about Agile Software Development?,” IEEE Softw., vol. 26, no. 5, pp. 6–9, Sep. 2009.
[25]
R. B. Johnson and A. J. Onwuegbuzie, “Mixed Methods Research: A Research Paradigm Whose Time Has Come,” Educ. Res., vol. 33, no. 7, pp. 14–26, Oct. 2004.
[26]
I. Scandurra, M. Hägglund, and S. Koch, “From user needs to system specifications: multidisciplinary thematic seminars as a collaborative design method for development of health information systems.,” J. Biomed. Inform., vol. 41, no. 4, pp. 557–69, Aug. 2008.
Ac ce p
te
d
M
an
us
cr
ip t
[18]
[27]
J. Nielsen, Usability Engineering. Morgan Kaufmann, 1993.
[28]
L. Cao and B. Ramesh, “Agile Requirements Engineering Practices: An Empirical Study,” IEEE Softw., vol. 25, no. 1, pp. 60–67, Jan. 2008.
[29]
L. Chung, B. A. Nixon, E. Yu, and J. Mylopoulos, Non-Functional Requirements in Software Engineering. Boston, MA: Springer US, 2000.
[30]
M. N. K. Boulos, S. Wheeler, C. Tavares, and R. Jones, “How smartphones are changing the face of mobile and participatory healthcare: an overview, with example from eCAALYX.,” Biomed. Eng. Online, vol. 10, no. 1, p. 24, Jan. 2011.
[31]
S. Barello, G. Graffigna, and E. Vegni, “Patient engagement as an emerging challenge for healthcare services: mapping the literature.,” Nurs. Res. Pract., vol. 2012, p. 905934, Jan. 2012.
[32]
A. Coulter, “Patient engagement--what works?,” J. Ambul. Care Manage., vol. 35, no. 2, pp. 80–9.
24
Page 24 of 27
R. Farzanfar, J. Finkelstein, and R. H. Friedman, “Testing the Usability of Two Automated Home-Based Patient-Management Systems,” J. Med. Syst., vol. 28, no. 2, pp. 143–153, Apr. 2004.
[34]
E. S. Chen and J. J. Cimino, “Automated discovery of patient-specific clinician information needs using clinical information system log files.,” AMIA Annu. Symp. Proc., pp. 145–9, Jan. 2003.
[35]
R. Holland, B. Rechel, K. Stepien, I. Harvey, and I. Brooksby, “Patients’ self-assessed functional status in heart failure by New York Heart Association class: a prognostic predictor of hospitalizations, quality of life and death.,” J. Card. Fail., vol. 16, no. 2, pp. 150–6, Feb. 2010.
[36]
B. Schweikert, H. Hahmann, and R. Leidl, “Validation of the EuroQol questionnaire in cardiac rehabilitation.,” Heart, vol. 92, no. 1, pp. 62–7, Jan. 2006.
[37]
T. S. Rector and J. N. Cohn, “Assessment of patient outcome with the Minnesota Living with Heart Failure questionnaire: Reliability and validity during a randomized, double-blind, placebo-controlled trial of pimobendan,” Am. Heart J., vol. 124, no. 4, pp. 1017–1025, Oct. 1992.
[38]
R. Rogers, J. Lombardo, Z. Mednieks, and B. Meike, “Android Application Development: Programming with the Google SDK,” May 2009.
[39]
J. C. A. Trappenburg, A. Niesink, G. H. de Weert-van Oene, H. van der Zeijden, R. van Snippenburg, A. Peters, J.-W. J. Lammers, and A. J. P. Schrijvers, “Effects of telemonitoring in patients with chronic obstructive pulmonary disease.,” Telemed. J. E. Health., vol. 14, no. 2, pp. 138–46, Mar. 2008.
[40]
J. Cleland, J. Caldow, and D. Ryan, “A qualitative study of the attitudes of patients and staff to the use of mobile phone technology for recording and gathering asthma data.,” J. Telemed. Telecare, vol. 13, no. 2, pp. 85–9, Jan. 2007.
Ac ce p
te
d
M
an
us
cr
ip t
[33]
[41]
E. Rescorla, “HTTP Over TLS,” May 2000.
[42]
S. Pilemalm and T. Timpka, “Third generation participatory design in health informatics-making user participation applicable to large-scale information system projects.,” J. Biomed. Inform., vol. 41, no. 2, pp. 327–39, Apr. 2008.
[43]
A. De Vito Dabbs, B. A. Myers, K. R. Mc Curry, J. Dunbar-Jacob, R. P. Hawkins, A. Begey, and M. A. Dew, “User-centered design and interactive health technologies for patients.,” Comput. Inform. Nurs., vol. 27, no. 3, pp. 175–83.
[44]
S. De Rouck, A. Jacobs, and M. Leys, “A methodology for shifting the focus of e-health support design onto user needs,” Int. J. Med. Inform., vol. 77, no. 9, pp. 589–601, 2008.
[45]
M. Berg, “Patient care information systems and health care work: a sociotechnical approach,” Int. J. Med. Inform., vol. 55, no. 2, pp. 87–101, Aug. 1999.
[46]
M. Hartswood, R. Procter, R. Slack, A. Voss, M. Buscher, M. Rouncefield, and P. Rouchy, “Corealization: Toward a principled synthesis of ethnomethodology and participatory design,“ Resources, Co-Evolution and artifacts, pp. 59-94, Springer, London, 2008.
25
Page 25 of 27
A. Dearden and H. Rizvi, “Participatory IT design and participatory development: a comparative review,” pp. 81–91, Oct. 2008.
[48]
J. Clemensen, S. B. Larsen, M. Kyng, and M. Kirkevold, “Participatory design in health sciences: Using cooperative experimental methods in developing health services and computer technology.,” Qual. Health Res., vol. 17, no. 1, pp. 122–30, Jan. 2007.
[49]
A. Das and D. Svanæs, “Human-centred methods in the design of an e-health solution for patients undergoing weight loss treatment.,” Int. J. Med. Inform., vol. 82, no. 11, pp. 1075– 91, Nov. 2013.
[50]
G. R. Casper, B.-T. Karsh, C. K. L. Or, P. Carayon, A.-S. Grenier, and P. F. Brennan, “Designing a technology enhanced practice for home nursing care of patients with congestive heart failure.,” AMIA Annu. Symp. Proc., pp. 116–20, Jan. 2005.
[51]
M. W. M. Jaspers, “A comparison of usability methods for testing interactive health technologies: methodological aspects and empirical evidence.,” Int. J. Med. Inform., vol. 78, no. 5, pp. 340–53, May 2009.
Ac ce p
te
d
M
an
us
cr
ip t
[47]
26
Page 26 of 27
Highlights
ip t
We iteratively developed a mobile health monitoring system for heart failure patients. Quantitative and qualitative information about user experience was collected. The system was used in an observational study over a one-year study period. Our approach led to several system upgrades for personalised self-monitoring. Engagement with the system was constantly high throughout the study.
Ac ce p
te
d
M
an
us
cr
27
Page 27 of 27
