First, we identified and described existing clinical workflows. Then, we conducted semistructured interviews with patients and providers to identify the perceptions, needs, and challenges of implementing an EHR-integrated RPM program within these workflows to prepare for a future intervention. No clinical intervention was performed in this study. This study followed applicable Standards for Reporting Implementation Studies (StaRI) guidelines (Multimedia Appendix 1).

The study was approved by the University of Massachusetts (UMass) Chan Institutional Review Board (protocol #00000781) on January 17, 2023. Provider participants were provided a fact sheet and gave verbal consent before interviews. Patient participants gave written informed consent before interviews. All participants received compensation in the form of a US $50 gift card after completion of interviews. All study data were deidentified.

First, our team explored available options to upload commercial wearable device data for PA to the EHR. We discovered a “fitness device flowsheet” order is available within the EHR. This order allowed providers to request patient fitness device metrics such as steps to sync with the EHR as a flowsheet, facilitating a display of device metrics within the EHR, similar to the display of vital sign data (ie, blood pressure). We also found a Fitbit integration available to sync patient PA data from the Fitbit app to the patient portal (myChart) after the flowsheet was ordered by a provider. The ordering provider is then able to review patient data through the EHR through basket message notification. In addition, any provider can view the patient’s PA data in the patient chart. We identified an existing EHR job aid that could be used to educate providers on ordering the fitness device flowsheet (Multimedia Appendix 2). The EHR job aid included details on types of patient data (ie, steps) that could be integrated into the EHR and information on the provider clinical workflow process to set up and receive this data. We also found instructions showing patients how to connect mobile and wearable devices to the patient portal (Multimedia Appendix 3). Although we identified multiple workflow steps necessary for an RPM program within our clinical system, no clinicians were using any of the steps regularly and in a cohesive workflow manner. For example, our review indicated that within the 1-year period before this study (from March 31, 2022, to February 28, 2023), the fitness device flowsheet was ordered a total of 4 times by all providers employed by UMass Memorial Health.

In April 2023, we recruited 10 PCPs from the UMass Chan Division of General Internal Medicine to participate in the study. Our recruited sample size was selected to reach saturation point and consistent with other qualitative studies noting that saturation can be reached with a relatively low sample size [20]. We aimed to recruit at least 1 attending physician, resident physician, and advanced practitioner provider. Providers were recruited by email and at a study presentation at a monthly clinic meeting. Providers were eligible if they were (1) an attending physician, resident physician, or advanced practitioner provider practicing in the UMass Memorial Medical Group Primary Care Clinic; (2) older than 18 years of age; and (3) spoke English.

Providers took part in a 30-minute semistructured interview led by a study team member. They were interviewed by a study team member to assess perceptions, needs, and challenges of implementation of an EHR-integrated RPM program. Participant interview questions are available in Multimedia Appendix 4. During the interview, we shared the pre-existing EHR job aid used to train providers on ordering the fitness device flowsheet (Multimedia Appendix 1) and obtained feedback on the existing clinical workflow.

From May to September 2023, we recruited 8 patients from the UMass Memorial Medical Group Primary Care Clinic to participate in the study. As with provider interviews, our sample size was selected to reach saturation point and consistent with other qualitative studies [20]. We aimed to recruit at least 1 patient identifying as African American and 1 patient identifying as Hispanic. Patients were recruited from the clinical practice of JS during routine clinical visits. Patients who expressed interest in learning more about the study were referred to meet with a study research coordinator. Patients were eligible to participate if they (1) received care at a UMass Memorial Primary Care Clinic and (2) had a BMI of 30 kg/m² or greater (also verified through self-report). Patients were ineligible if they were (1) unable to provide informed consent, (2) incarcerated, (3) were non-English speaking, or (4) were younger than 18 years of age.

Patients completed semistructured interviews regarding their preferences for remote monitoring, PA counseling, data-sharing with providers, and enrollment processes. Participant interview questions are found in Multimedia Appendix 5. We also showed patients screenshots from myChart to obtain feedback on the existing patient-facing workflow.

All interviews were transcribed and reviewed for clarity. We used rapid qualitative analysis during and after the interviews [21,22]. We used an Excel spreadsheet (Microsoft Corporation) to capture data, with each participant as a row, and domains of interest as columns. The coding team (JF, JN, JS, and VA) met initially to develop key domains that addressed our research question and were derived from the interview guides. The team then conducted 2 rounds of coding together, before independently coding an additional 2 transcripts per coding team member. Once all the transcripts were coded, JF and VA independently reviewed the results of each column to identify themes within each domain of interest. They then reviewed across columns to identify any overlap in themes across the domains. Both coders had a high agreement on themes after this process, determined saturation had been met with codes, and reached a consensus on the final list of themes. The team used a validated checklist to ensure interrater reliability [23].

The sample included 10 provider participants, 50% (5/10) of which were female and the rest 50% (5/10) were male. In all, 40% (4/10) were residents, 20% (2/10) were providers, and 40% (4/10) were attendings, with 40% (4/10) having 0-3 years of experience, 20% (2/10) having 3-9 years of experience, and 40% (4/10) having 10+ years of experience (Table 1). We also included 8 patient participants: 88% (7/8) were female and 12% (1/8) were male, with a mean age of 58.5 (SD 13) years. Of these participants, 25% (2/8) were non-White, 75% (6/8) were White, 88% (7/8) were non-Hispanic, and 12% (1/8) were Hispanic.

Provider feedback regarding the implementation of an RPM program around PA using commercial wearables was positive. Providers appreciated the importance of PA guidance for healthy weight maintenance. They noted the proposed clinical workflow appeared simple to implement in practice. However, potential challenges were noted related to provider burden, patient and provider resource support, and health equity. Overall, 3 themes were identified from the interviews: (1) providers’ knowledge of PA prescription is focused on general guidelines with limited knowledge on how to tailor guidance to patients, (2) providers were open to receiving PA data but were worried about being overburdened in their workflow, and (3) providers were concerned about patients being able to equitably access and participate in digital health interventions. Themes and representative quotes are included in Textbox 1.

Patient participant feedback from interviews regarding RPM of PA was enthusiastic. The enthusiasm stemmed from their frustration about perceived inadequate PA guidance from their health care team. The patient noted previous positive experiences with wearable devices but noted inconsistent adherence without external reinforcement. They were in favor of sharing exercise metrics with providers to facilitate improved counseling from the health care team. They did not foresee significant difficulty in setting up an RPM program from an IT perspective if specific support was provided. A total of 3 themes were identified from the analysis, including (1) patients received limited or general guidance regarding PA from providers and other resources, (2) patients want to share exercise metrics with the health care team and receive tailored PA guidance at regular intervals, and (3) patients need written resources to support the setting up of a wearable program with access to live supports on an as-needed basis. Themes and representative quotes are included in Textbox 2.

Our preliminary study assessed the perceptions of, need for, and the challenges of implementing an EHR-integrated RPM program to facilitate PA counseling of primary care patients through patient and provider interviews. We found that providers had limited confidence in their abilities to tailor PA recommendations to patient-specific clinical factors. Similarly, patients felt PA counseling received from their health care team was not individualized to factors such as age and medical conditions. However, patients were eager to share PA data through an EHR integration in the hopes it could facilitate more specific PA counseling from their health care team. Providers were open to implementing a PA-focused RPM solution that efficiently integrated into their workflow and aligned with the principles of health equity and accessibility. As noted previously [19], we found that implementation of an EHR-integrated RPM program into the primary care clinical workflow presents multiple challenges, including limited provider knowledge about exercise counseling, limited provider time, and the need to assist patients in properly setting up and managing wearable devices to allow data to integrate into the EHR. The results of this formative work will inform the development of our future pilot study to establish the feasibility and acceptability of this approach.

Providers expressed strong familiarity with well-known medical guidelines for PA and the ability to provide general counseling to patients. For example, providers appeared familiar with the Centers for Disease Control and Prevention PA guidelines suggesting that adults engage in at least 150-300 minutes of moderately intense aerobic PA weekly [24]. However, while individuals with medical comorbidities who cannot achieve these targets are advised to engage in the maximum level of PA possible to improve health outcomes [24], providers expressed low confidence in their ability to tailor exercise recommendations based on patient-specific factors. Patients in this study also agreed with this sentiment. Furthermore, providers believed they did not have the training or expertise to adequately counsel patients on exercise goals. Yet, evidence and expert guidance indicate patient guidance and titration schedules for both sedentary patients and those with medical comorbidities are available [25]. To address this challenge, strategies to provide succinct evidence-based educational materials with guidance on implementing tailored PA recommendations using remote data should be explored. As providers may be hesitant to increase activity too quickly for the patient with certain medical conditions, these materials can include tailored titration schedules using daily step counts that address this patient’s activity (ie, sedentary lifestyle) and major medical conditions (ie, musculoskeletal conditions, chronic pain, and cardiopulmonary disease) that impact activity guidance. These materials should be made accessible to providers before the implementation of any PA intervention in an electronic format. In addition, patient message templates can be created that include these guidelines that providers can easily elect to send to selected patients through the EHR or standard mail. Future studies should assess the format, delivery mode, and effectiveness of such educational materials in increasing providers’ PA knowledge and tailored PA counseling.

Patients expressed a positive sentiment toward sharing RPM exercise metrics with their provider if this would lead to more specific PA counseling from their health care team. In addition, a subset of our interview sample reported previous experiences with wearable devices that were generally positive but were limited by inconsistent long-term adherence. The effectiveness of mobile health (mHealth) interventions is commonly limited by low long-term adherence rates. For instance, a review of mHealth interventions for anxiety and depression found attrition rates between 1% and 50% [26]. The supportive accountability model proposes that human support increases adherence from an individual who is trustworthy, is benevolent, and has expertise [27]. Our proposed integration of an RPM program within primary care suggests the ability to leverage the provider-patient therapeutic relationship to increase patient adherence.

Providers also expressed concerns about their EHR workflow being overburdened by notifications related to an RPM program. Information overload in the era of EHRs is an important concern. While a small study found physicians are spending about 20% of their time solely on EHR-related tasks [28], studies examining the amount of time PCPs spend reviewing remote data were not readily identified. Information overload specifically related to clinical decision-making systems has been well studied. Repeated similar alerts for the same patient are known to lead to alert fatigue and may lead to lower attention to the alert for its intended function in the primary care setting [29]. While notification fatigue in this practice of primary care is unlikely to be eliminated, we propose measures to limit provider burden using focused in-basket notifications. This may include a provider-initiated limit on notification frequency and an option to receive notifications when data falls outside of prespecified boundaries.

Ultimately, as provider time continues to be burdened by administrative tasks, future studies may need to assess the inclusion of a dedicated team member with specific expertise in exercise counseling for RPM management. A recent study found primary care teams using mHealth strategies can be effective in the care of chronic diseases, such as diabetes [30], suggesting a team-based approach could potentially be effective for lifestyle counseling in primary care. Nurse practitioners have also been suggested as being critical in training, delivering, and monitoring digital PA programs [31]. A dedicated team member for exercise counseling would facilitate increased review of RPM data and allow for dedicated time for communication with patients directly about any challenges with exercise titration potentially due to existing health conditions (ie, chronic pain exacerbation).

Patients express varying comfort with digital technologies, and we identified a major need for patient assistance with device set up and maintenance. Furthermore, this challenge may be intertwined with provider concerns related to health equity and accessibility. Previous studies have noted that access to electronic health portals may be affected by patient factors including age, race, and ethnicity [32]. In our study, patients expressed comfort with written materials as a primary resource for instruction, while also noting access to live support would be helpful. We propose that any RPM protocol should create comprehensive written material for device set up and connection to the medical record. Furthermore, a team member should be available for support on an as-needed basis to both minimize provider burden and assist a diverse array of patients. For the deployment of an RPM program in the real world, we propose a health care team member, such as a medical assistant, should be trained in supporting patients with digital technologies.

Finally, to facilitate the scalability of any RPM program within the clinical setting, insurance reimbursement for the time and decision-making related to RPM is vital. In 2019, the Centers for Medicare and Medicaid Services implemented new billing codes that cover remote monitoring of physiological data, which have also been adopted by some commercial payers. Specific current procedural terminology code examples include billing for services such as device setup, collection or interpretation of physiological data by health care professionals, time-based billing for physicians for communication with patients, and delivery of results by practice staff to the physician [33]. These developing reimbursement mechanisms suggest a potential billing structure to facilitate RPM in the primary care setting.

Our study is limited by the fact that we recruited patients and providers from a single, large, urban, academic primary care practice. A relatively small number of participants were interviewed. However, thematic analysis did show we reached a saturation point with the sample size used, consistent with similar qualitative studies [20]. Our sample was comprised largely of female patient participants, although these demographics are consistent with differences in attitudes and behaviors regarding weight [34]. Furthermore, a diversity of viewpoints from provider participants was obtained by provider type and experience level. The EHR used in the study was developed by Epic Systems and we did not attempt RPM integration with any other EHR vendors. Alternate informatics approaches might be needed with other EHRs. However, the principles obtained from participant interviews in this study will still apply broadly to many primary care practices in the development of specific clinical workflows. Finally, while our proposed workflow incorporates RPM data metrics, patients’ subjective impressions (ie, pain with exercise) are not captured using RPM and alternative strategies may need to be developed to elicit this information.

Our formative assessment found that a potential implementation of an EHR-based RPM program using commercially available wearables for tailored PA counseling by the primary care team was a well-received intervention by both patients and providers. Consistent with our primary objective, we identified key perceptions, needs, and challenges to the implementation of this protocol and proposed potential solutions. Our ongoing work will use these findings in a pilot study to evaluate the feasibility and acceptability of tailored PA counseling leveraging RPM within a primary care practice.