Diabetes is a leading cause of several highly morbid and costly conditions, including chronic kidney disease, cardiovascular disease, and visual impairment [1]. Attention to diabetes self-management behaviors can help patients avoid or delay many diabetes-related complications; however, consistent engagement in self-care behaviors is challenging for many patients [1,2]. Patient activation (ie, knowledge, skills, and confidence in managing one’s own health) is vital to achieving optimal diabetes self-management and is associated with lower health care costs [3-5].

Patient portals are computerized tools that can connect patients with electronic health data maintained by their health care system. Patient portals can provide an engaging and convenient means for patients to track and visualize health data, obtain education and guidance, and connect patients and doctors [6]. Research has shown that patient portals offer a promising platform to increase patient activation, enhance care, and promote self-management while overcoming the limitations of costly and difficult-to-scale face-to-face interventions [7,8]. We recently applied a user-centered design sprint methodology and key strategies for patient engagement to develop a patient portal intervention called My Diabetes Care (MDC; formerly Diabetes Dashboard) [9].

MDC is embedded within an established patient portal, My Health at Vanderbilt (MHAV), at Vanderbilt University Medical Center (VUMC) [10]. MDC is a multi-faceted intervention designed to help patients better understand their diabetes health data and support self-management [9]. MDC uses infographics to visualize and summarize patients’ diabetes health data; incorporates motivational strategies (eg, social comparisons); provides literacy-level appropriate educational resources; contains secure messaging capability; and links to a diabetes online patient support community and diabetes news feeds, highlighting new discoveries, medicines, and recipes. MDC was founded on the well-established Chronic Care Model adapted for eHealth—eHealth Enhanced Chronic Care Model (eCCM) [11]. By leveraging elements within the model’s 5 domains (self-management support, delivery system design, decision support, clinical information systems, and eHealth education), MDC has the potential to create more informed and activated patients, leading to improved outcomes (Figure 1).

A 1-month usability study of MDC among 60 patients found that participants, including those with limited health literacy, highly rated MDC’s usability and acceptability [12]. User experience data showed that most participants found that the infographics and links to literacy-level appropriate diabetes health information helped them better understand their diabetes health data. Participant feedback also identified areas for improvement, including adding information about diabetes medications and links to diabetes news feeds, highlighting new discoveries, medicines, and recipes. Consistent with the eCCM, the study also found a significant pre-post increase in patient activation scores among the study participants. However, a larger randomized controlled trial (RCT) is needed to assess MDC’s impact on patient activation more definitively.

This study aims to evaluate the effects of MDC on patient activation in adult patients with type 2 diabetes mellitus (T2DM). In addition, we plan to explore secondary outcomes, including system use and usability, and the effects of MDC on diabetes self-efficacy, knowledge, self-care, medication adherence, distress, and clinical endpoints. The study will serve as a pilot for a larger definitive trial evaluating the effect of MDC on clinical endpoints.

To achieve this objective, we are conducting a 6-month, 2-arm, parallel-design, pragmatic pilot RCT of MDC. Participants in both arms are told the purpose of the study is to determine satisfaction with 2 versions of MHAV among patients with diabetes. One version is the currently available version of MHAV. The second version of MHAV contains the MDC intervention. Participants in both arms complete the study questionnaires at 3-time points: T₀=baseline, T₁=3 months, and T₂=6 months.

The study protocol is registered with ClinicalTrials.gov (ID NCT03947333) and is being conducted in accordance with the principles outlined in the CONSORT (Consolidated Standards of Reporting Trials) Statement, extension for pragmatic trials [13]. Pragmatic trials are designed to evaluate the real-world effectiveness of interventions in routine practice environments [14]. Unlike a strictly controlled trial, participants in our study are not constrained to receive a controlled dose of the intervention [14]. Due to necessity, participants are not blinded to the intervention, and we do not attempt to control participants’ communication or information-seeking behaviors beyond the 2-arm randomization described here. The Vanderbilt University institutional review board approved this study.

Participants are recruited from 14 VUMC-affiliated adult primary care clinics located throughout Middle Tennessee (4 urban and 10 suburban clinics). An electronic health record (EHR; Epic Systems Corp) stores all clinical data. Patients receive access to their clinical data via an integrated and highly adopted patient portal, MHAV, that is accessible on desktops and via a native mobile app for iOS and Android mobile operating systems.

Participants are eligible for the study if they are a patient at a participating primary care clinic and have T2DM, are currently being treated with at least one diabetes medication, are able to speak and read in English, are 21 years or older, have an existing MHAV account, and have reliable access to a desktop or laptop computer with internet access. We exclude patients living in long-term care facilities, patients with known cognitive deficits, patients with a severe visual impairment, and patients currently participating in another diabetes-related research study.

On a rolling basis, potentially eligible patients are selected from a randomly ordered list of established adult patients with diabetes from participating clinic sites and are sent a recruitment letter describing the study. In addition, we also use My Research at Vanderbilt to send the recruitment letter to current patient portal users who elected to allow investigators to contact them about research opportunities via email. Interested patients contact a research assistant to learn more about the study. To enroll, participants complete a web-based study eligibility screener and electronic consent form on the web via REDCap (Research Electronic Data Capture) version 5.0.8. [15].

A study coordinator contacts all enrolled participants to review study procedures, answer the remaining questions, and confirm eligibility criteria. The enrolled participants are sent a baseline questionnaire via REDCap. After receiving the completed baseline questionnaire, the study coordinator randomly assigns participants to 1 of 2 groups: (1) intervention or (2) usual care. The randomization sequence was generated by the research team biostatistician using a permuted block randomization scheme stratified by clinic site and participants’ age group (65 years and older vs younger than 65 years) to obtain balance across treatment groups on key variables. The randomized assignment for eligible participants is accessible only to the study coordinator and biostatistician using the REDCap randomization module; the other investigators are blinded. Once a randomization assignment is finalized, participants in both arms receive an email with their treatment assignment and an explanation of how to navigate to features of MHAV specific to their treatment group. Participants are asked to reply to the email affirming that they can access MHAV and/or MDC in accordance with their group assignment. Monthly quality assurance checks are used to ensure that MDC is functioning correctly (eg, displaying data correctly) and to ensure the fidelity of the intervention.

A participant may withdraw from the study at any time by notifying the study team. In addition, participants are withdrawn from the study by the investigators if they do not complete the baseline questionnaire needed for randomization. If a participant is withdrawn from the study for any reason, they are notified, and a reason is provided.

Participants randomized to the intervention arm are provided access to a version of MHAV embedded with MDC, as described in the Introduction section. Participants randomized to the intervention are advised to view MDC on a desktop or laptop device because the present version of MDC is not mobile friendly. Figure 2 shows a screenshot of MDC and illustrates its features [16,17]. Participants randomized to the usual care arm have access to the currently available version of MHAV, which includes the ability for patients to review pertinent health data, review medical information about their conditions, and communicate with their health care team.

Enrolled participants receive 3 study questionnaires via REDCap at the associated time points (T₀-T₂): baseline questionnaire, 3-month follow-up questionnaire, and 6-month follow-up questionnaire. On the basis of pilot testing, we estimate the time to completion for the baseline questionnaire to be about 25 minutes and 20 minutes each for the 3-month and 6-month follow-up questionnaires. Participants are compensated US $40 for completing the enrollment questionnaire and US $35 each for completing the 3-month and 6-month follow-up questionnaires.

To describe the study population at baseline, we collect the following sociodemographic and clinical variables (Table 1). Health literacy is assessed by a validated 1-item screener asking respondents to rate their confidence independently filling out medical forms [18,19]. Consistent with previous studies, participants noting any lack of confidence are classified as having limited health literacy [20,21]. eHealth literacy is assessed by the 8-item eHealth Literacy Scale (eHEALS) [22]. The eHEALS uses a 5-point Likert scale ranging from strongly disagree to strongly agree. Total scores range from 8 (worst) to 40 (best). The presence of comorbidities (ie, hypertension and hyperlipidemia) is assessed by 2 clinicians who independently review patients’ problem lists and medications abstracted from the EHR, and disagreements are resolved by consensus.

Table 2 shows the primary and secondary outcomes and related measures contained within the study questionnaires and their associated time points. The same study measures are administered to all participants in both arms, except for the system use and user experience, which contain items unique to the participants’ assigned condition (ie, intervention vs control).

Figure 3 shows the flowchart of the recruitment process. Recruitment began in March 2020 and ended in May 2020. Throughout the recruitment period, 4388 unique letters were sent to patients identified as potentially eligible. Separately, 2609 unique emails were sent to patients who use MHAV and previously agreed to be contacted by email about research studies for which they might be eligible. As it was not possible for the study team to cross reference the list of those who were sent letters against the list of those who were sent emails, some overlap is possible. The letters and emails generated 702 visits to the web-based REDCap eligibility screener, resulting in 576 completed screeners. Of the 576 complete screeners, 163 (28.3%) were ineligible and 413 (71.7%) were eligible. Of the 413 eligible screeners, 113 (27.4%) declined to participate and 300 (72.6%) were enrolled. We administratively withdrew 10% (30/300) of those enrolled, and the remaining 270 participants were randomized.

Of those randomized, most (214/267, 80.1%) were non-Hispanic White; 13.1% (35/267) were non-Hispanic Black; and 6.7% (18/267) reported being of another race, including American Indian or Alaska Native, Asian, Native Hawaiian or other Pacific Islander, more than one race, Hispanic Black, and Hispanic White. In addition, 43.7% (118/270) reported being 65 years or older. Furthermore, 10.1% (27/268) reported educational attainment of a high school degree or less, 33.6% (90/268) had limited health literacy, and 39.6% (106/268) had only a US governmental health plan (eg, Military, Civilian Health and Medical Program of the Uniformed Services, Veterans Affairs, Medicaid, and Medicare). Approximately one-third (82/270, 30.4%) were taking insulin, the mean duration of diabetes was 12.5 (SD 8.6) years, and the mean HbA_1c level at baseline was 7.1 (SD 1.3). The overall clinical population with diabetes has a somewhat different demographic distribution: 66.49% (5317/7997) are non-Hispanic White, 21.83%% (1754/7997) are non-Hispanic Black, and 42.34% (3386/7997) are 65 years or older. The mean HbA_1c of the 7997 patients within the overall clinical population and a laboratory value in the past 12 months was 7.40. As of October 2020, we have at least 95.6% (258/270) completion among participants through the 3-month follow-up assessment.

Our study will be one of the very few RCTs to evaluate a patient-facing diabetes digital health intervention delivered via a patient portal. Although diabetes digital health interventions have great potential, their impact has been limited due to the difficulty in integrating the interventions into routine care [32]. By embedding MDC into Epic’s MyChart platform with more than 127 million patient health records [33], our intervention is directly integrated into patients’ health care systems and is highly scalable and sustainable. Unlike independent health apps, patient portals, by their very nature, are integrated into routine care and therefore offer greater potential for uptake and sustained use [7]. The challenge for health systems and investigators is how to make the most of patient portals to improve care.

We designed MDC to enhance and expand on existing aspects of patient portals, including access to personal health data and education, to better support diabetes self-management. Although patient portals offer easy access to personal health data, previous research suggests that complex data displays—showing many tests in small format on a single page without any indication of their clinical significance—make it difficult for patients to find and correctly interpret a particular test result [7]. MDC uses a simplified infographic to indicate normal, modestly abnormal, and more severely abnormal results and literacy-level appropriate materials to help patients better understand their diabetes health data. Thus, this research will inform how different data displays and user interface designs impact patients’ ability to understand their personal health data.

Studies of other technology-enabled diabetes self-management solutions suggest additional strategies that may benefit patients [32,34]. These include analysis of patient-generated health data and tailored education and feedback [32,34]. To maximize scalability and sustainability, functionality enabling these strategies is best built directly into the EHR vendor’s patient portal platform (eg, Epic’s My Chart). Solutions built external to the platform may be more challenging to integrate into routine practice and run the risk of quickly becoming out of date and requiring reprogramming when vendors release platform updates. For this reason, we did not include these strategies in our intervention. However, EHR vendor solutions are emerging and may allow us to incorporate these additional patient engagement strategies into future iterations of MDC [35].

We designed MDC to be usable by the greatest number of patients, including those with limited health literacy [12]. Limited health literacy is typically associated with worse outcomes among patients with diabetes and can be a barrier to patient portal use [36,37]. Previous research has shown that patients with limited health literacy struggle to use patient portals because of complex medical terminology and a lack of literacy-level appropriate health information [20,21]. Although patient portals have the potential to worsen health inequities by further advantaging well-educated patients with greater resources, if designed and implemented appropriately, patient portals also have the potential to lower health literacy demands by ensuring that patients are presented with the health information and resources in a format that is convenient and easy to navigate and understand [38].

Our study population has a somewhat smaller proportion of racial or ethnic minorities than the overall clinical population, suggesting that additional strategies may be needed to increase adoption among these groups. Digital navigators—trained staff or volunteers who assist patients in accessing and learning how to use technology to meet their needs—have been used to increase patient portal adoption among vulnerable populations [20,39]. Smartphone use is increasingly common across different socioeconomic and racial or ethnic backgrounds, and for patients that lack broadband home internet connections, smartphones may be their only way to access the internet [40]. Thus, developing interventions suitable for mobile platforms may reduce barriers to adoption. Since the initiation of this trial, we have begun the development of a mobile-friendly version MDC that we hope will further increase its utility and accessibility.

Finally, given that racial and ethnic minorities are disproportionately affected by T2DM, future studies of MDC and other technology-delivered diabetes self-care interventions should consider using oversampling techniques, as demonstrated by Nelson et al [41,42], to recruit study populations that closely represent the overall population of patients with T2DM. Doing so will help ensure that technology-delivered diabetes self-care interventions are effective in the populations with the greatest need and inform any revisions to those interventions and/or their implementation needed to address disparities.

This study has important limitations. It relies on self-reported measures of patient activation and several secondary outcomes that are subject to social desirability and recall bias. However, the chosen measures are validated, widely used, and accepted, offering the advantage of being brief, inexpensive, and unobtrusive compared with more objective measures. Our study is powered to examine the effects of MDC on patient activation; therefore, analyses examining the effects of other outcomes (eg, self-care behaviors and HbA_1c) and comparing the effects among subgroups (eg, patients with limited health literacy or poorly controlled diabetes) may be very informative but may also be underpowered. We hope that this study will serve as a pilot for a larger definitive trial evaluating the effect of MDC on clinical endpoints. Should MDC prove effective at increasing patient activation, the 6-month trial duration will not allow us to determine if the effect is temporary or sustained. A longer trial of a year or more in duration is needed to examine sustained effects. MDC is currently available only in English. This was necessary to increase the feasibility of designing the intervention and successfully completing this initial trial. However, diabetes disproportionately affects Spanish-speaking groups, so translation into Spanish will be an important goal, if MDC should prove beneficial. Finally, although patient portal interventions offer the advantages of direct integration into routine care, scalability, and sustainability, they are subject to inequities in patient portal adoption [43] and may appeal to more activated patients [44]. However, research shows that patient portal adoption is increasing [10,45], and if designed appropriately, patient portals could reduce health disparities [38,46]. Moreover, recent research finds that patient portal users have similar levels of patient activation as nonusers, although portal users are more likely to have internet access and a higher level of education [47].

We expect that this study will help determine the effectiveness of MDC in increasing patient activation among patients with diabetes. Beyond this primary objective, we will also be able to examine data on secondary cognitive, behavioral, and clinical outcomes and users’ perceptions of and satisfaction with the intervention. Our findings and evolving patient portal functionality will inform the continued development of the intervention to best meet users’ needs and a larger trial focused on the impact of MDC on clinical endpoints.