Heart transplant recipients (HTxR) require lifelong immunosuppressive therapy to prevent rejection episodes. The estimated percentage of medication nonadherence to immunosuppressive treatment after heart transplant (HTx) ranges from 15% to 30% [1]. These rates are worrisome as medication nonadherence impairs quality of life, increases health costs, and is a direct cause of graft loss and death after HTx [1-6].

Medication nonadherence in the HTx population is a dynamic behavior influenced by multilevel patient, provider, and health system factors [7]. To improve medication adherence, it is essential to frequently monitor medication nonadherence and identify modifiable risk factors for medication nonadherence, such as high therapeutic complexity, weak professional-patient relationship, and lack of patient motivation [4,5,7,8].

Subjective methods to evaluate medication adherence, such as self-reporting, are widely used [9] and well correlated with objective methods (eg, immunosuppressive drug level assay or electronic monitoring systems) [5,10]. Indeed, self-reporting is considered the best method to capture patient experiences and individual risk factors for medication nonadherence, such as patients’ medication beliefs [11]. However, this method involves in-clinic facilities and requires patients to travel to the clinic [5].

Emerging research indicates that patient-reported adherence through mobile devices produces data of similar quality to those provided by traditional in-clinic methods [12]. Therefore, the use of electronic patient-reported measures (ePROMs) to detect nonadherent HTxR could help increase the feasibility of self-reporting and overcome current in-clinic limitations [13,14].

Medication adherence ePROMs could also provide valuable information to care providers to implement early and personalized interventions through mobile health technology. Indeed, internet interventions (ie, “treatments, typically behaviorally based, that are operationalized and transformed for delivery via the Internet”) [13,15-17] show a promising impact on prompting changes in health behaviors such as medication adherence [15,16,18-20].

Behavior-based theories with demonstrated effectiveness in reducing medication nonadherence are recommended to be combined when a new intervention program is designed [21,22]. Moreover, motivational interviewing is a useful tool for professionals to deliver such theory-based interventions in the transplant population [23,24]. Although these behavior change techniques are increasingly used, there is a lack of studies applying them to internet interventions in the HTx population [2,25,26].

The mHeart medical device is a mobile app that is primarily intended to measure and manage medication adherence in outpatient HTxR. mHeart was designed to use ePROMs to identify medication nonadherence in the home setting and facilitate behavior change interventions to improve medication nonadherence rates. According to the directions for the International Society for Research on Internet Interventions (ISRII) [13], a prerequisite before recommending the widespread use of internet delivery is to demonstrate the accuracy of ePROM scores and their relationship with traditional in-clinic methods [27]. Moreover, the workflow of any new electronic behavior-based intervention designed to manage medication nonadherence should be tested before scaling it to larger research [28].

The primary aim of this study was to validate mHeart to measure medication nonadherence in early stage HTxR in the home setting. To do this, we sought to identify the quality of the psychometric properties of the ePROMs reported as being critical in electronic health behavior change instruments [29,30], that is, validity, reliability, responsiveness, interpretability, and burden [31].

The secondary aims were (1) to measure patient satisfaction with mHeart and its usability and (2) to explore the impact of theory-based interventions on medication nonadherence rates among HTxR to determine the hypothetical scalability of the treatment within the context of a larger research study.

This prospective research study was conducted in the ambulatory setting of a Heart Failure and Transplant Unit of a tertiary university hospital from July 15, 2016, to December 1, 2016. The study was approved by the institutional review board of the hospital (IIBSP-MHE-2014-55). The participants were informed of the study purposes, the length of the follow-up, all the procedures, and the research team behind the study. Written informed consent was obtained from all participants.

The psychometric quality of the ePROMs was based on the Scientific Advisory Committee of the Medical Outcomes Trust (SAC-MOS) [31] and the COnsensus-based Standards for the selection of health Measurement INstruments (COSMIN) consensus guideline [30]. The quality of the results obtained was compared with the International Society for Quality of Life Research (ISOQOL) standards [11].

We followed the European Society for Patient Adherence, COMpliance, and Persistence Medication Adherence Reporting Guideline (EMERGE) [28] recommended criteria for transparent and accurate medication adherence reporting data. The directions for the ISRII [13] and the Consolidated Standards of Reporting Trials of Electronic and Mobile Health Applications and OnLine TeleHealth (CONSORT-EHEALTH) guidelines (section 5) [17] were followed to report the internet-based intervention program. The Theory Coding Scheme (TCS) [32] provided a reliable method to describe the theory underpinning the interventions.

In addition, the Checklist for Reporting Results of Internet E-Surveys [33] was applied to ensure the quality of reporting of the Web-based satisfaction survey.

Enrollment was conducted from July 21, 2016, to October 26, 2016, in the Cardiology Outpatient Clinic by transplant physicians during routine in-clinic appointments. All consecutive adult, early stage HTxR (less than 1.5 years from HTx) owning a smartphone and with no cognitive impairment were included. Cognitive impairment was defined as any condition limiting patients’ ability, including memory and thinking skills, to use the mHeart system and complete the questionnaires. No previous computer or smartphone knowledge was required. HTxR did not receive any financial compensation, a phone, or wearables for their participation. The patient flowchart is shown in Figure 1.

The algorithm summarizing the procedures is shown in Figure 2. After signing the informed consent form, all patients were assessed for eligibility (ie, the same day as enrollment by the physicians), and they completed a baseline face-to-face visit with the transplant pharmacist, followed by an initial mHeart training session.

The interview with the pharmacist lasted approximately 45 min. Sociodemographic and clinical data were extracted from patients’ electronic health records. The data were collected in a form provided in Multimedia Appendices 1 and 2. At the end of the visit, the pharmacist registered the new patient’s profile in the mHeart system. Patient access was facilitated by an automated message sent to the patient’s phone with a username and password.

Thereafter, a technical mHeart initial setup was provided by the mHeart Help Center of the private firm developing the technology. This session was conducted by telephone and lasted at least 15 min to enable at-home monitoring, that is, (1) downloading the app from the app store, (2) guiding the first access, and (3) providing training on the functionalities of the mHeart platform. This service was also responsible for query resolution and user assistance throughout the study.

As soon as the HTxR had received training, the transplant pharmacist sent them a welcome message through mHeart, requesting the patients’ response to confirm their activation in the mHeart follow-up. Once the patients had responded to this message, three consecutive assessments were scheduled. The assessment procedures are described below, and these were conducted to measure the validity properties of the ePROMs.

The mHeart medical device is a home-based mobile phone app complemented by a website [38]. From a technical point of view, access to the tool is multiplatform (ie, smartphone, tablet, or computer), and it can be downloaded for free from app stores [39,40]. mHeart is bidirectionally integrated with the hospital information system (HIS) using encrypted data. This integration between the two systems allows mHeart to directly obtain sociodemographic data from the HIS. In addition, mHeart uploads a weekly clinical report to the HIS, including all the data reported by the patients on the platform. The general layout is represented in Figure 3. An in-depth description of the technical specification of the system and the source code are provided in the online Mendely (Dataset) [41]. The version number of the app used was 3.0.2, and the content was frozen during the study.

From a clinical point of view, the mHeart tool was designed to primarily manage medication nonadherence using several features (Table 1). In addition, three of the subfunctionalities of the platform were to (1) resolve patients’ queries about their treatment and health condition, (2) empower patients in terms of self-care, and (3) facilitate professionals’ interventions based on patient-reported outcomes (ie, symptoms and adverse effects of drugs, heart rate, glycemia, weight, and blood pressure). A detailed demonstration of the clinical use of mHeart in HTxR can be found in a video format in Multimedia Appendix 4. More details about functionalities are also provided in the online Mendely (Dataset) [41].

A total of 31 early stage HTxR were included (age: mean 54 years, SD 12 years) and analyzed, and no attrition was observed (Figure 1). In all, 71% (22/31) of participants were men. The mean follow-up was 2.3 (SD 0.9) months. The mean time between HTx and the study was 1.2 years (SD 0.8 years). The patients’ demographic and clinical characteristics are detailed in Multimedia Appendix 8.

At baseline, 71% (22/31) of patients used technologies frequently. Most of the patients reported that mHeart could be useful (22/31, 71%) or very useful (4/31, 13%). One-third of the patients (9/31, 29%) reported that they needed personal assistance to get started with using the mHeart platform.

Polypharmacy was common; the mean total medication count was 13 (SD 4; range 7-18), exceeding 14 drugs per day in 36% (11/31) of patients. Patients reported a mean number of 6 (SD 4) adverse effects. As many as 61% (19/31) of them reported being self-reliant for medication management.

Medication-related inconvenience was moderate to high (>6 of 10) in 25% (8/31) of HTxR. As many as 74% (23/31) of them believed they were taking excessive medication. The danger of sometimes not taking immunosuppressive drugs was understood by 42% (13/31) of recipients. Furthermore, 32% (10/31) of recipients were unaware of the consequences of completely abandoning antirejection therapy. More details are provided in Multimedia Appendix 9.

The completion rate was 93% (29/31 patients). The reasons for no response to the survey are detailed in Figure 1. HTxR reported no inconvenience because of the mHeart intervention approach employed. The mean ePROM appropriateness score was 8 (SD 2; score range: 0-10). The mean score for overall satisfaction with the mHeart approach was 9 (SD 2; score range: 0-10). All 100% (29/29) of the patients would recommend the mHeart platform to other recipients. Regarding patient suggestions for improving the platform, 24% (7/29) of HTxR made eight suggestions and 76% (22/29) of them responded, “No, I like it just as it is.”

Improvements were implemented based on patient feedback, for example, (1) to avoid patient recall bias, the order of the ePROM items was designed to automatically change whenever the test is completed; (2) patients could graphically consult any values they recorded in mHeart (eg, blood pressure); (3) pop-up alerts were established to let patients know that a new text message from the provider had arrived; and (4) diverse actions were implemented to decrease telephone use by patients to inquire about the compatibility of new therapies: the usability of the mHeart function to inquire about new therapies was improved, and text messages were sent to the patients, explaining how to use this function.

The details of each survey item score, patient suggestions, and the subsequent improvements are provided in Multimedia Appendix 11.

In this study, the first challenge was to validate mHeart ePROMs to detect early stage HTxR at risk for medication nonadherence in the home setting. With this aim in mind, the COSMIN [30] and SAC-MOS [31] standards provided a solid framework to support the quality of the assessed ePROM psychometric variables. As the mHeart ePROMs meet the minimum standards set by the ISOQOL [11], they can be used in clinical practice and comparative effectiveness research.

The excellent agreement observed between ePROMs and on-site PROMs confirmed that the mHeart electronic approach was as effective as the traditional on-site method in identifying medication nonadherence. The ePROMs used in mHeart showed multiple advantages over on-site PROMS, such as eliminating potential professional interpretation of ambiguous responses that could affect medication adherence rates. Furthermore, the electronic approach required fewer in-clinic facilities than the traditional method of assessing medication nonadherence. The integration of the mHeart data with the HIS also reduced the time required to record ePROM responses in patients’ medical records. These advantages reduced burden and enabled the pharmacists to focus on clinical tasks. This is clinically significant, as a pharmacist intervention is associated with better use of evidence-based therapies, reducing medication errors and emergency department visits while increasing patient satisfaction [58].

Regarding the baseline medication nonadherence values, the percentage of HTxR nonadherent to overall medication in this study was worrisome according to the Morisky-Green-Levine PROMs (32%) but similar to that observed in another series (33%) [59]. According to the SMAQ, the percentage (39%) of medication nonadherence to immunosuppressive treatment was slightly higher than the overall percentage reported in the HTx population (34%) [7] and considerably higher than that reported in a meta-analysis of the solid organ transplant population (25%) [4,25]. These results suggest that the ePROMs used in our study have a synergistic effect in identifying nonadherent recipients. However, they also highlight the need for intervention programs to improve medication nonadherence, as almost half of the HTxR were unaware of the consequences of medication nonadherence. Nevertheless, comparisons among medication nonadherence rates in the field of transplantation should be interpreted with caution as studies use different populations and methodologies [4].

The exploratory intervention program established to deal with this problem showed promising results. Immunosuppressive treatment adherence rates significantly improved in one-third of the recipients, according to the SMAQ. This figure is higher than that reported by most studies showing low or medium effect sizes of around 10% to 20% in medication adherence improvement [10,45]. In contrast to these studies, our program was designed to deliver personalized, internet-based multilevel interventions based on behavioral theories [7,21,22]. Indeed, human support and tailored interventions have been shown to be a requisite to improve medication nonadherence rates throughout eHealth [10,18]. Moreover, our exploratory study meets 72% of the TCS criteria (ie, items 1-11), indicating that the interventional study design complies with the theoretical basis of the intervention [32]. This is important as interventions meeting a minimum of 60% of the TCS criteria have been found to be highly effective [60].

Nonelectronic theory-based interventions have been considered highly effective when improvement was >20% [45]. Therefore, as the improvement in medication nonadherence in our exploratory study was higher (30%), the strategies applied proved to be synergistic and to enhance the effectiveness of the program [21,32]. Equally important, patients adhered well to the study protocol and provided excellent feedback. Among the benefits of the mHeart approach, patients highlighted personalized communication, support from professionals, and self-empowerment, which were the most relevant criteria used to design the mHeart intervention program.

Our study includes a limited but representative sample comprising 86% of all early stage HTxR in our center. This characteristic is common in transplant population, as the prevalence is limited [61]. We did not enroll chronic-stage recipients for the following reasons: (1) early posttransplant medication nonadherence is a high-risk behavior with a huge impact on survival [1], (2) we wanted to avoid wide heterogeneity in chronic-stage providers and treatments, and (3) we wanted to avoid chronic-stage recipients having to travel to the clinic for the study. In addition, although early stage recipients are typically better adherers [3,62], this did not prevent us from observing an effect in the highest risk period after transplant.

The interval between medication nonadherence assessments may have led to recall bias. Although this bias could have influenced the electronic score, this limitation is intrinsically related to the validation methodology to ensure that the electronic and traditional methods are performed in similar conditions and in patients with similar psychological and functional status. Moreover, the short study periods used were methodologically grounded according to the main study aim of validating the ePROMs. In sensitivity to change measures, a 1-month interval is considered an adequate interval to measure the validity of an indirect smartphone health measure [51,52]. Moreover, fortnightly assessments are sufficient to identify additional medication nonadherence in the transplant population [46]. In reproducibility measures, intervals of 1 to 2 weeks are common [63]. Therefore, a 7-day interval was selected to minimize the effect of possible confounding variables [31] related to the multifaceted factors affecting posttransplant medication nonadherence [10,46].

Adherence monitoring is recognized as a standard quality practice in transplant centers [7]. In the past few decades, there has been a growing interest in improving the screening opportunities of medication nonadherence without increasing the in-clinic burden. The quality of the ePROM psychometric variables and the patient satisfaction reported in this study support the scalability of the mHeart ePROM for use in clinical practice and research [11]. The results obtained indicate that the electronic self-reporting approach provides a highly sensitive medication nonadherence measure in the transplant population to complement traditional and more time-consuming methods, such as blood tests or medication refills [25,46].

Furthermore, given that medication nonadherence behavior in the transplant population is influenced by several factors [7], optimal daily adherence is a real challenge for recipients [64]. Consequently, feasible holistic strategies are needed to help recipients reduce the negative impact of medication nonadherence on health outcomes [20]. The exploratory results of this behavioral theory–based intervention on medication nonadherence rates are encouraging. Future studies will determine the intervention’s effectiveness on clinical outcomes [13]. With this aim in mind, the EMERGE [28], TCS [32], and CONSORT-EHEALTH reporting criteria [17] standards were followed to support the scalability of the intervention methodology used in larger research. For now, the feasibility and effectiveness found in this study encourage following this path to curb the widespread problem of medication nonadherence.

The electronic method implemented in the mHeart medical device successfully identified medication nonadherence in the HTx population. ePROMs demonstrated their potential to overcome the limitations of traditional on-site methods. The ePROMs’ quality properties supported their widespread use in research and clinical practice. The theory-based intervention program showed an encouraging improvement in medication adherence rates, with excellent patient satisfaction and usability scores. Therefore, the mHeart program resulted in significant benefits in estimating medication nonadherence in the HTx population and showed promise in guiding professionals’ interventions with the potential to optimize HTx outcomes.