Within the European Union (EU), an estimated 50 million people live with multimorbidity, defined as two or more chronic health conditions [1]. For individuals living with multimorbidity, the self-management of multiple conditions can impose a significant burden [2], with activities that include managing multiple symptoms, medications, information on their conditions, and clinical appointments. In addition, health care services for individuals with multimorbidity are often repetitive (multiple appointments), inconvenient, inefficient (individuals may see different clinicians who give conflicting advice), burdensome, and potentially unsafe due to poorly integrated and coordinated care [3,4]. The outcome for individuals is reduced quality of life, as time and energy are spent managing multiple conditions, limiting their opportunity for social or personal activities [5].

The risk of multimorbidity increases with advancing age, with prevalence rates estimated at 65% in people aged ≥65 years, 85% in people aged ≥85 years, and rising [6]. The rapid aging of the global population brings significant concerns over the sustainability of health services, due to associated increases in health care expenditure, and disparities in the number of practicing health professionals. It is therefore important that efforts are made to explore sustainable digital approaches to support home-based self-management of chronic diseases and multimorbidity. Self-management (or self-care) can be described as an individual’s ability to manage symptoms, treatment, emotions, and lifestyle changes as part of living with a chronic condition [7]. Improving best practices around the provision of person-centered care for a person with multimorbidity requires empowering the Person with multimorbidity to self-manage, actively supported by their care network (CN), which primarily involves informal carers (ICs), formal carers (FCs), and health care professionals (HCPs). The CN of a person with multimorbidity plays an important role in diminishing the impact of disease management, which may subsequently improve health outcomes and quality of life [8].

Digital health technologies have the potential to improve and advance home-based self-management for older persons with multimorbidity, yet most digital solutions focus on single-disease management (eg, diabetes) [9,10]. Therefore, digital solutions that address complex disease management and multimorbidity, taking into account the role, views, and needs of the person with multimorbidity and their CN, are also required.

To date, there has been limited research examining the potential of digital health support for multimorbidity management. This includes understanding the challenges faced by people managing multimorbidity, as well as design requirements for digital technologies to address these challenges [11-15]. Although such research is necessary, to the best of our knowledge, research on digital platforms and systems to support multimorbidity has not progressed beyond examining requirements and suggesting design recommendations.

Within the EU, the ICARE4EU program provides the most robust examination of digital or eHealth use to address multimorbidity management within the context of integrated care [16]. Managers of 101 integrated care programs in Europe were surveyed to understand if they had used eHealth (or digital) solutions and, if so, what were the benefits of and barriers for the solutions in relation to multimorbidity care. Of these programs, 85 adopted eHealth solutions, and 42 of these were targeted specifically at older adults. The types of eHealth technologies implemented within these programs included remote consultation and monitoring, self-management tools (including electronic reminders and web-based decision support), health care management technology such as patient databases and e-referral systems, and electronic health records. However, neither detailed descriptions of these technologies nor their evaluations were presented. Furthermore, the authors noted limitations in that HCPs, patients, and their caregivers were not consulted in terms of the availability of eHealth supports within these programs.

With such limited research in the areas of digital health, integrated care, and multimorbidity management, there is a need for large-scale, longitudinal programs or projects to better understand both the complexities of multimorbidity and how digital technologies can be designed, developed, and implemented to support the person with multimorbidity and their CN. The ProACT project, funded by the European Commission Horizon 2020 programme, brings together a multidisciplinary consortium of 13 European partners for the purpose of developing and evaluating a digital integrated care system to empower home-based, patient-centric care and proactive self-management of chronic conditions for Europe’s 50 million persons with multimorbidity.

This paper reports the protocol for the ProACT Horizon H2020 project main proof-of-concept (PoC) trial conducted in Ireland (by the Trinity Centre for Practice and Healthcare Innovation, Trinity College Dublin, NetwellCASALA at Dundalk Institute of Technology and Home Instead Senior Care) and Belgium (by imec at the Studies in Media, Innovation and Technology in the Vrije Universiteit Brussel) between April 2018 and June 2019. Before the PoC trial, the ProACT platform was designed and developed between 2016 and 2018 through an iterative user-centered process involving input from 166 key stakeholders (older people with multiple chronic conditions, carers, and HCPs) across Ireland, Belgium, and Italy [17-21].

The study evaluated, at a PoC level, a digital health platform (called ProACT) for older persons with multimorbidity to self-manage their conditions with support from their CN. ProACT was implemented at 2 EU trial sites (Belgium and Ireland). The specific aims of the trial were (1) to explore the potential benefits of the ProACT platform for persons with multimorbidity and (2) to obtain feedback from all relevant participants on their experiences using the ProACT platform and on the potential of the platform to improve integration of care and support for multimorbidity disease management.

Specific objectives for all participants were to evaluate the usability, accessibility, and acceptability of the ProACT platform, user adoption and satisfaction with the technology and services, and experiences of participants using ProACT. Additional objectives for persons with multimorbidity were to evaluate the potential impact of the ProACT platform on a range of health, well-being, psychological, and psychosocial outcomes and evaluate the efficacy of ProACT as a behavior change intervention that aims to improve self-management skills for the person with multimorbidity. Additional objectives for IC and FC participants were to evaluate the potential impact of the ProACT platform on their psychological and psychosocial outcomes.

ProACT is a citizen-driven, self-management orientated, digital integrated care platform capable of supporting multiple disease management and well-being parameters (eg, mobility and sleep) on a single user app. The overall platform (Figure 1) consists of the following:

From the person with multimorbidity perspective, the measurement and sensing devices and CareApps are the only platform technologies that they interact with. For CN users, CareApps tailored to their requirements are their point of interaction with the platform. The full list of devices used by the person with multimorbidity is included in Textbox 1.

Within the overall intervention, the primary point of information exchange with the end user (person with multimorbidity or CN support actor) is their CareApp. Textbox 2 outlines the structure and use of each CareApp. Figure 3 provides an overview of the person with multimorbidity home screen co-designed with users. The petal-based interface presents a brief summary of health and well-being data tailored to each person with multimorbidity’s condition and self-management preferences. Using a color-coded traffic light system, persons with multimorbidity are alerted if their data are below or above their personal thresholds (pink), when they have not taken a reading for five days or more (orange), or when all is deemed normal for the person with multimorbidity (blue).

The study was a longitudinal (12-month) PoC trial using an action research design and mixed methods approach. Action research is a period of investigation that describes, interprets, and explains social situations while executing a change intervention aimed at improvement and involvement [22]. The strength of this approach is the capability to generate solutions to practical problems, while garnering methods to understand the context of care, needs, and experiences of the person with multimorbidity group, drawing upon a range of research methods (eg, participant observation and in-depth interviews), to involve and build relationships with persons with multimorbidity and associated CN support actors. Within the PoC trial, this allowed for modifications to the technology based on quantitative and qualitative data collected from platform use statistics (eg, how often participants engage with the platform), platform data (ie, data coming from sensors and technologies), observational and usability testing methods to understand participant interaction with CareApps; person with multimorbidity and CN responses to interviews, questionnaires, and standardized assessments (eg, to evaluate quality of life, device proficiency, and usability).

Persons with multimorbidity and their CNs (consisting of ICs, FCs, and HCPs) were eligible for this study. For inclusion, participants with multimorbidity were aged ≥65 years and had at least two of the following conditions: diabetes, chronic obstructive pulmonary disease (COPD), chronic heart failure (CHF), or chronic heart disease or coronary artery disease including hypertension, atherosclerosis, angina, and arrhythmia; were capable of giving written informed consent; had access to broadband services (this refers to regional infrastructure); or lived in an area with sufficient coverage for mobile broadband or internet. The implemented service costs were covered as part of the trial.

CN participants were invited only on the permission of the person with multimorbidity and were required to be aged ≥18 years; to be providing care or support to a participant with multimorbidity; have access to a computer, tablet, or smartphone with an internet connection; and to be capable of giving written informed consent.

A purposive sample of 120 persons with multimorbidity (60 persons with multimorbidity per trial site in Ireland and Belgium) were recruited to participate in the PoC trial. Although sample size is often cited as a key factor in determining the potential success of a study, this is more relevant for randomized controlled trial studies that seek to answer specific questions regarding the efficacy of interventions (does it work?) and is less relevant for studies related to care and service improvement (how does it work?) [23]. Thus, to determine the PoC sample size, we took a pragmatic approach and reviewed two important factors: (1) it is large enough to provide a reliable analysis of the ecosystem and (2) small enough to be financially feasible. An analysis of the literature suggests that the overall sample size in a PoC, telemedicine and health focused information communications technology trial is low. A review of 1030 studies on telemedicine-based technological interventions for chronic disease management, looking at CHF (436 studies), stroke (422 studies), and COPD (172 studies) between 2005 and 2013 (including 35 systematic reviews and one review of the reviews), suggested that methodologically robust sample sizes for each condition were 17 participants (COPD), 21 participants (stroke), and 19 participants (CHF) [24]. The selected studies were conducted primarily in the United States and Europe.

Ethical approval was granted from participating health service organizations where recruitment took place and from academic partners. Informed consent was obtained on an individual basis in accordance with legal and ethical guidelines at each trial site region, following careful explanation of the study and provision of participant information and informed consent forms for the person with multimorbidity and participating members of their CN. All participants had the right to withdraw from the study at any time without any questions. Following a review of recruitment procedures by ethical committees in Ireland and Belgium, it was agreed that researchers should only contact a person’s HCP if they had provided this consent.

In both Ireland and Belgium, participants were selected by several different methods, depending on which recruitment source they were accessed through, as outlined below:

Invited person with multimorbidity participants had at least 7 days to review the participant information leaflet and have queries answered before technology deployment, which occurred over 2 visits to the person with multimorbidity’s home. All researchers ensured that ProACT technology was deployed correctly and in a consistent manner across trial sites, following a strict deployment plan.

During the first visit, members of the research team obtained written consent from the participants. Each participant received devices depending on their condition profile. Participants also had the option to use any existing device (that they currently use at home) to measure an included parameter (eg, blood glucose monitor) by manually entering readings from the device into the person with multimorbidity CareApp. ProACT sensor devices were connected by Wi-Fi or Bluetooth, and a broadband internet connection was provided for the duration of the trial for any participants who did not have existing broadband in their homes. Participants were trained on how to use their ProACT devices during their initial visit. This included a brief introduction on how to use the ProACT CareApp and associated third party apps (eg, using the Withings HealthMate [25] app to take a blood pressure reading), as it was important that the person with multimorbidity was not overloaded with information on all ProACT technology features during the first visit. Participants were also provided with a paper-based manual, containing detailed instructions for using each device, along with common troubleshooting instructions.

Approximately 1 week after the first visit, the researchers conducted a second deployment visit. Detailed training on the CareApp took place with additional web-based training materials and videos made available through the ProACT CareApp. A study helpdesk, staffed by respective research team members in Ireland and Belgium, was available (from 9:30 AM- 4:30 PM, Monday to Friday) to assist participants with queries and technical difficulties. In both Ireland and Belgium, a dedicated clinical triage service for monitoring vital signs was also available (9 AM-5 PM, Monday to Friday). Triage personnel (clinical nursing staff) had access to data from all persons with multimorbidities participating in the trial via the SIMS. A protocol for dealing with potential adverse events was developed using triage personnel. This included defining thresholds for abnormal vital sign values for each parameter being monitored. For example, thresholds for high and low blood glucose values were set in the SIMS for participants with diabetes. At the outset of the trial, global threshold values were set for all participants. However, over the course of the trial, such thresholds were often adjusted for individuals based on their normal values. If a participant’s vital sign reading is outside the normal threshold, an alert is triggered on the SIMS triage interface, and as noted above, the participant will see a pink petal on their CareApp dashboard (Figure 3). In such instances, the triage nurse calls the participant to discuss the reading and determine whether an escalation is required. In both trial regions, clinical triage was not provided for nonvital sign data (eg, sleep or activity). Participants were reminded that this was a research study and that the triage service would not be considered as a replacement for normal care. In the event that a person with multimorbidity felt ill, they were recommended to seek medical advice or care as they normally would. Persons with multimorbidity were also reminded of this at regular intervals through a pop-up message on their CareApp, as requested by the ethics committees. Following completion of the second deployment visit, the participants began their trial period.

Invited members of the person with multimorbidity’s CN were provided with access to their relevant CareApp that they could use on their own devices (smartphones, tablets, or computers). These customized CareApps allowed those in the CN to view relevant data from the participant with multimorbidity and educational materials related to condition management, well-being, and technology use. Participants with multimorbidity chose what data to share with each CN participant. The data viewed by HCP participants via their CareApp were not used to make clinical decisions. This was clearly outlined in the participant consent forms and information leaflets for all trial participants.

As a PoC trial, a key outcome is to understand whether a larger trial that makes a definitive assessment of benefit is warranted. Pilot and PoC studies are more about learning than confirming or formally assessing evidence of the impact or benefit associated with an intervention. Therefore, analyses should focus on providing descriptive evidence and indications of the range of possible responses rather than on formal hypothesis testing [45]. Analyses were therefore mainly descriptive and aimed at understanding user experiences in relation to the use of the ProACT platform. Qualitative methods encouraged participants to speak about their experiences of living with and managing multimorbidity and their experience of using ProACT technologies. Quantitative data analysis ensured comparability and consistency of questions across participants and time points.

Qualitative data were analyzed using thematic analysis (TA) to identify and understand emerging themes. An inductive approach was adopted to identify themes at a latent level. An inductive TA is data-driven, as opposed to analyst-driven TA [46]. This approach helps generate novel insights from interview data that may have differed greatly from pre-existing research in the area pertaining to the research questions. This is essential to the action research design of the trial to analyze differences in responses across time points. Furthermore, identifying themes at a latent or interpretative level goes beyond the semantic meaning of the presented data, encouraging interpretative analysis by the researchers. Across the PoC trial locations, a protocol (including in-person and web-based training) was put in place to ensure that the TA followed a strict analytical process, with researchers ensuring transparency and consensus across each step. Individual researchers coded the transcripts according to an established analysis protocol. Pairs of researchers collapsed and categorized codes into themes. Discussions and recoding workshops were conducted to ensure agreement on theme and subtheme names were reached among the wider trial site teams. In Ireland, NVivo for Mac (version 11; QSR International) [47] was used to conduct the coding part of the analysis, while in Belgium, MAXQDA Analytics Pro (VERBI GmbH; [48]) was used. Using different software did not impact the analysis, as the same methodological approach was used at both sites.

Quantitative questionnaire data were analyzed at both trial sites using SPSS statistical software (version 25; IBM SPSS Statistics [49]). The primary analysis was to evaluate changes in scores between the assessment points. Descriptive statistics were used to summarize participant demographic data and general outcomes from the questionnaire data. Sensitivity analysis was performed to treat missing data. Missing data were imputed based on the methods suggested for each questionnaire. In case a standardized method was not reported in the literature, mean substitution, using similar imputations for all questionnaires, was used for all time points, if less than 20% of the data were missing. Initial analyses were conducted to assess the distribution of all variables and check for relevant assumptions, including normality. Given the small sample size at each trial site, the majority of variables violated normality. Therefore, to maintain the intrinsic value of the quantitative data in this circumstance, no transformations were performed, and for further inferential analysis, nonparametric (Friedman and Wilcoxon signed-rank) tests were implemented.

The SIMS component of the ProACT platform supported the analysis of additional data (including sensor data from the devices and engagement with the devices and ProACT CareApps). Metrics of interest for analysis included symptom (eg, blood pressure, blood glucose, SpO₂, and weight) trends or patterns over time; the ratio of alerts to symptom readings over time; trends or patterns in activity and sleep data over time and engagement with various parts or features of ProACT and the CareApp; and responses to self-report questions on health and well-being.

Across the EU, there is a growing drive to meet the complex care needs of older people with multimorbidity. eHealth or digital health options are now recognized as potential support [22]. However, EU health care systems are not yet equipped to address the comprehensive care needs of people with multimorbidity [50]. The use of innovative person-centered digital health technologies are increasingly viewed as a means to address the challenge of multimorbid care (eg, tools to support patients’ self-management and multidisciplinary collaboration between professionals [51] may play a key role in advancing the integration of health and social care needs). Despite this, research into the design and development of digital health systems, focusing on multimorbidity management, particularly for older adults, is in its infancy.

It is important to re-emphasize that the focus of this research is on multimorbidity (multiple co-occurring chronic conditions, but with a focus on multiple conditions) as opposed to comorbidity (multiple co-occurring chronic conditions, but with a focus on a singular condition) [52], which seeks to advance a multi-country understanding of the challenges for defining, designing, implementing, and evaluating a digital intervention, focused primarily on multimorbidity management across diverse populations. To our knowledge, ProACT is also the first digital intervention to systematically incorporate (and evaluate) behavioral change and human-computer interaction methods to advance persons with multimorbidity’s self-management practices in relation to multimorbidity.

With the mixed methods, action research PoC study of the ProACT platform, we are further addressing the need for increased longitudinal and applied research in the areas of digital health, integrated care, and multimorbidity management. The two primary aims of ProACT are as follows:

Outcomes from trials [53] are positive in terms of user engagement with ProACT and a shift in behavior to adopt this digital intervention. These outcomes will help advance both the state of the art on how to design and conduct research with older persons with multimorbidity and their CN and deliver a new digital health solution to address the challenge of multimorbidity management and care.

Although substantial research has been conducted on the implementation and use of digital health technologies to address single-disease management, a clear gap exists in understanding the requirements for managing multimorbidity from the perspective of older persons with multimorbidity and their CN and how supported self-management happens in practice. The findings from the ProACT PoC trials will contribute significantly to the research in this field. With 120 older persons with multimorbidity and 73 CN participants, the trials have provided a novel multi-stakeholder, multi-country perspective on multimorbidity self-management and integrated care. With a primary focus on qualitative outcomes, the PoC trials have provided detailed insight into the person with multimorbidity’s self-management journey facilitated by a digital health platform, longitudinally over 12 months. Outcomes will evaluate the impact of ProACT at a PoC level to determine whether a larger trial, which makes a definitive assessment of benefit, is warranted.