Health provider perspectives of electronic medication monitoring in outpatient asthma care: a qualitative investigation using the consolidated framework for implementation research

Abstract

Objective:

Little is known about the implementation challenges health providers might face with the use of digital health in outpatient asthma care. To qualitatively explore the experience of health providers with electronic medication monitoring (EMM) using an implementation science framework.

Methods:

Using the Consolidated Framework of Implementation Research (CFIR), we conducted interviews (n = 10) exploring health providers’ experience with EMM with asthma patients from 5 primary care or specialty clinics. The EMM tracked albuterol and inhaled corticosteroid (ICS) use, and health providers called parents whenever ICS adherence waned, or albuterol use increased. Interviews were audio-recorded, transcribed, and deductively analyzed using directed content analysis.

Results:

Health providers reported the intervention’s primary advantage, compared with current asthma care, was the ability to monitor medication use at-home. Most felt the intervention improved care delivery. Nurses and medical assistants described a process of phone calls and checking alerts, that had varying levels of administrative burden and complexity. Health providers felt that sustained implementation of the intervention model would require additional employees to handle the administrative and clinical workload. Half of the interviewed providers were unsure if patient needs were met by the intervention, while some cited technology syncing issues, others liked the enhanced interactions for asthma education.

Conclusion:

Health providers reported positive experiences supporting parents and children with asthma using EMM but also highlighted intervention components that needed improvement or refinement to yield successful implementation in outpatient pediatric clinics. Recommendations for enhancing the intervention for a scaled-up implementation were discussed.

Introduction

To care for an estimated six million U.S. children with asthma, a mainstay of pediatric asthma care is supporting asthma self-management (1,2). Nearly 50% of children with asthma have at least one asthma attack per year and miss one or more school days due to symptoms (3). However, controlling symptoms with improved asthma self-management, specifically adherence to medications, is challenging. Children’s caregivers and health providers must communicate clearly about when symptoms require escalation to a health provider’s attention, based on the caregiver’s assessment, and whether children with asthma took medications as prescribed. Nevertheless, caregivers have consistently reported difficulty in recognizing and monitoring asthma symptoms over time and adhering to medication regimens (4,5). The reliance on patient report of at-home therapy and symptoms can complicate asthma management (6,7).

An emerging tool that may address the inherent limitations of asthma self-management is electronic medication monitoring (EMM). EMM is used increasingly to support chronic disease self-management in adult populations, including asthma and diabetes, and EMM technologies offer an opportunity to support caregivers by allowing health providers to remotely monitor medications (8–10). Asthma EMM involves placing sensors on inhalers to track the date and time of inhaler actuations from daily inhaled corticosteroid (ICS) and rescue albuterol use, and this use can be accessed by providers through an associated web portal. Asthma studies using EMM suggest that this technology can be increasingly utilized in outpatient asthma care. Researchers in Auckland, New Zealand found that using an electronic monitoring device with an audiovisual reminder improved adherence in school-aged children, with median percentage adherence at 85% in the intervention group compared with control group at 30% (11). A randomized trial in the United Kingdom found that children, ages 6–16 years old, using the same device as the Auckland trial, had 70% adherence to ICS, compare to 49% among controls (11,12).

Like experiences with other technology-enhanced interventions, implementation of EMM in outpatient settings could present challenges to existing work processes in clinics and have variable success in outpatient care delivery. The introduction of new technology disrupts established work practices and information flows for health providers (13), which can get overlooked in the intervention design. A critical need exists to understand the human support surrounding technology to achieve reliable and substantive outcomes from digital health interventions, like EMM (14).

We used the Consolidated Framework for Implementation Research (CFIR) to study pediatric health providers’ barrier and facilitators to using EMM with asthma inhaler medications in clinical care delivery (15). CFIR is a well-recognized conceptual framework to guide systematic evaluations of multi-level factors that shape the implementation and effectiveness of interventions (16). Leveraging the Improving Technology-Assisted Recording of Asthma Control in Children (iTRACC) trial in clinics, we aimed to qualitatively explore provider’s experiences, using an established implementation framework as a lens for inquiry, to understand their engagement with the intervention and guide future intervention design. The findings will help health providers, health system administrators, and other stakeholders understand how to better implement EMM in outpatient clinical delivery.

Methods

Description of trial

The iTRACC study was a 12-month randomized clinical trial conducted from 2016 to 2018 in Chicago, IL (17). The trial enrolled 252 parent-child dyads of children with persistent asthma. Dyads were randomized to two conditions: 1) usual care; or 2) an intervention with Bluetooth-enabled inhaler sensors developed by Propeller Health (Madison, WI) (Appendix Figure 1, Supplementary material). Dyads were eligible for the trial if their child met the following criteria: 1) age 4 to 17 years old; 2) at least one asthma exacerbation requiring oral corticosteroids the year prior to enrollment; and 3) parent reported active prescription of an ICS for at least one year prior to enrollment. Dyads were recruited from five clinic sites (i.e. 3 pediatric primary care clinics, 1 pediatric pulmonary clinic, and 1 family allergy clinic). The study team approached these five sites because the clinics have a diverse racial and socioeconomic mix of patients with both publicly and privately insured patients. Every site was interested in collaborating for the study, and the study team explained the trial through in-person meetings at clinic. All clinics and champions agreed to participate, and no further recruitment was needed. Children were excluded if they: 1) were non-English speaking; 2) had a co-morbid condition that might interfere with asthma symptom assessment (Appendix Table 1, Supplementary material); or 3) were involvement in other interventions that would interfere with iTRACC’s EMM (i.e. using alternate inhaler sensor devices). The EMM technology allowed parents and health providers to track albuterol and ICS use on a cellphone application (app) or a web portal, respectively (Appendix Figure 2, Supplementary material). Alerts by email and through the portal notified health providers if their patients had increased albuterol use (i.e. >4 uses in a 24-h period) or decreased ICS use (i.e. no detected ICS doses in 4 days). Upon receiving the alerts, health providers called parents to triage how to improve adherence or discern the cause of increased albuterol use. Each clinic’s providers determined their timing for calling in response to alerts and how many attempted calls to reach patients. Figure 1 depicts the general intervention schematic. The trial was registered (NCT02994238).

Figure 1.

Intervention schematic of electronic medication monitoring.

Sample and data collection

We conducted semi-structured interviews (n = 10) with champions from each of the 5 trial sites to explore their experiences with implementation of the EMM intervention 6 months following trial completion. Champions were physicians, nurses, or medical assistants at each of the clinics who agreed to oversee the EMM at their respective clinic. Champions were sent an email about the qualitative study and those interested were scheduled for an in-person or telephone interview depending on preference; only one champion was not interviewed because they had changed employment. Interviews were conducted between June and July 2019 by one trained facilitator (KK) and interviews averaged 48 min (range: 38 min-61 min). Participants were compensated $100 for their time. The study was approved by the hospital Institutional Review Board (IRB 2016–698) and informed consent was obtained for all participants.

Interviews

A semi-structured interview guide designed to address the CFIR domains of implementation was drafted and reviewed by the trial’s research team members, a clinical research manager at Propeller, and a qualitative methodologist (SS). The guide was revised to incorporate feedback. Interview questions are shown in Table 1. CFIR provides a structure for which implementation factors are interpretable and generalizable across programs, and the framework has constructs embedded into five domains to encompass a wide range of factors that influence implementation outcomes (16,18). Application of CFIR is flexible so that the framework can be tailored to the needs of the evaluation, and not all constructs or domains are relevant or must be used (16,19). Our interview guide was designed to address three CFIR domain: 1) characteristics of the intervention; 2) inner setting; and 3) characteristics of the individual. Selection of domains and constructs were based on prior knowledge by investigators of what were the most relevant issues to the intervention design during the study, issues during the trial, and minimizing respondent burden. Questions from the domains of outer setting and processes were not included as the intervention did not involve organizational stakeholders nor did the original trial involve implementation planning and engagement that can be assessed qualitatively.

Table 1.

Summary of domains and constructs interview guide questions.

Domain	Construct	Definition	Interview Guide Questions
Intervention characteristics	Adaptability	The degree to which an intervention can be adapted, tailored, or refined to meet local needs.	• Please list the top 3 iTRACC intervention components that you feel should not be altered and why.
			• Please list the top 3 most important changes you think need to be made to the iTRACC intervention and why.
			Please list the top 3 ways that the intervention influenced the care of your pediatric asthma patients and why.
			• How well do you think the intervention met the needs of patient and families served by your clinic?
			• Satisfaction using the Web portal, receiving emails and calling patients when receiving asthma notifications
	Relative Advantage	Stakeholders’ perception of the advantage of implementing the intervention versus an alternative solution.	• What are other programs or changes have your clinic made to improve pediatric asthma care and how does iTRACC compare to those?
	Complexity	Perceived difficulty of the intervention, such as scope, disruptiveness, and number of steps required to implement.	• How does iTRACC fit with your existing work processes and practices in your clinic?
Inner setting	Readiness for Implementation	Indicators of organizational commitment to its decision to implement an intervention, such as access to information and available resources.	• What kind of clinic-level changes will need to happen for an intervention like iTRACC to become routine care?
			• If the intervention were routine in your clinic, who on the clinical team would be most ideal for responding to notifications about their adherence to ICS?
			• And for notifications about albuterol use?
Characteristics of individuals	Knowledge and beliefs about the intervention.	Individuals’ attitudes toward and value placed on the intervention.	• Please share up to 5 expectations you had for the intervention and whether your expectation was met and why
			• Satisfaction using the Web portal, receiving emails and calling patients when receiving asthma notifications
	Self-efficacy	Individual belief in their own capabilities to achieve implementation goals.	• Rate how confident you were following components (e.g. using the Web portal and calling patients when receiving asthma notifications) of iTRACC from 0 to 5, with 0 being “very unconfident,” and 5 “very confident.” & why you gave that score.

Data analysis

Interviews were audio-recorded and transcribed verbatim by a hospital-approved transcription consultant (Cheryl Westercamp, Chicago, IL), and de-identified for analysis. We conducted a deductive, directed content analysis approach using CFIR constructs from the three domains and definitions as an a priori analysis framework (20). Our directed content analysis process involved five steps. First, we created Excel spreadsheets representing the three CFIR constructs to capture the corresponding interview questions and participant responses to each. Second, participant responses (i.e. transcript excerpts and detailed field notes) to each interview question were entered the CFIR construct spreadsheets for analysis. Third, questions were assigned to each member of the analysis team (KK, SS, MK, LM), consisting of a pediatrician, qualitative methodologist, a research staff member involved in recruitment and sustainment of the iTRACC trial, and another pediatric and adolescent health research staff member. To reduce bias, two members of the analytic team were not involved in the trial. The team reviewed all responses and identified concepts pertaining to the assigned CFIR constructs and drafted detailed summaries of the key concepts for each construct and level of support. Fourth, because the software used for analysis (Excel) precluded calculation of inter-rater reliability, we evaluated reliability through a process in which the raw data for each CFIR construct were reviewed and summarized by another team member with differing background or expertise. For example, responses originally reviewed and summarized by member involved with the iTRACC intervention, were reviewed and summarized separately by a member not involved in the intervention. Fifth, summaries of key concepts from both analysts were discussed in a series of peer debriefing meetings wherein discrepancies across the two summaries for each construct were identified and resolved through team discussion, prior to finalizing themes and interpretations for each CFIR construct (21).

Results

We interviewed a physician and a medical assistant or nurse from each clinic, totaling 10 health providers. Most (70%) identified as non-Hispanic White. Half (50%) worked in clinics with predominantly, publicly-insured patients.(Table 2) In routine clinical duties, nurses and medical assistants (Mean = 95 calls per week) self-reported more patient calls per week than physicians (Mean = 13 call per week). Results of the analysis are presented below according to CFIR domains and their respective constructs (Table 3).

Table 2.

Health Provider Characteristics (n = 10).

Characteristics	% (n)
Provider Race
White	70 (7)
African-American or Black	10 (1)
Asian	10 (1)
Other	10 (1)
Provider Ethnicity, Hispanic	10 (1)
Provider Role in Clinic
Nurse	40 (4)
Physician	40 (4)
Medical assistant	20 (2)
Years in Practice, mean (SD)	17.5 (8.5)
Clinic Insurance mix, ≥ 80% publicly-insured	50 (5)
Number of Patient Phone Calls per week, mean (SD)
By nurses/medical assistants	95 (69.2)
By physicians	13 (12.1)

Table 3.

Summary of health providers’ qualitative findings, using the consolidated framework of implementation research.

Domain	Construct	Summary	Exemplar Quote
Intervention Characteristics	Adaptability	Health providers found the intervention adaptable to address patient needs, such as identifying needed refills, but other intervention changes were needed for implementation success.	“Iťs just figuring out how to use it is the question. Figuring out where it fits in. Figuring out how to make it so iťs not overwhelming to doctors, staff.”- PR4 Physician
	Relative Advantage	Clinics had no comparable supports that could improve detection of ICS adherence.	“I don’t think our clinic has ever really done another program [like]… what iTRACC is doing.”- PR7 Medical Assistant
	Complexity	Health providers did not find the intervention complex and found ways to fit the intervention into their existing clinic processes, but a few identified it as burdensome, requiring many extra steps.	“An additional like 15 to 20min of work, so. looking up, going into the portal, trying to remember my password, looking at. the information, going into EPIC, finding that patient and then. looking at their chart, deciding whether or not I should call the family myself or if someone has … recently talked to them …” -PR5 Physician
Inner Setting	Readiness for Implementation	Health providers felt that if this intervention was routine care that they would need to hire additional staff to support the increased clinical and administrative needs.	“I’d have to have a separate employee or resource to be able to be that touch person for installation and follow-up of device care, so … the patient [will] probably have one increased visit to come back and do that… I’m [going to] have to take away at least an employee resource to take the time to educate them over the phone and follow up with them over the phone.” -PR2 Physician
Characteristics of Individuals	Knowledge and beliefs about the	Health providers reported mixed feelings about the intervention’s success in monitoring patients, tailoring education, and enhancing communication.	“I do feel like we did … a lot of sick calls more than what I expected. A lot of our families, like prior to being involved … knew asthma enough so they knew what to do so they wouldn’t … necessarily call us and let us know when they were sick and so it was kind of nice to have us be notified.”-PR3 Nurse
	Self-Efficacy	Most providers felt confident checking a web-based portal to monitor medications and making calls to patients in response to the alerts.	“Very confident … because you know it wasn’t ever, it never came across where a family was like oh absolutely not, we have not been using the albuterol.” -PR6 Nurse

Intervention characteristics

Adaptability

According to CFIR, the adaptability construct represents the extent to which the intervention can be tailored or refined to meet the clinics’ or patients’ needs. In responding, health providers focused primarily on the intervention’s adaptability to the clinic’s goals of supporting patients, citing how the alerts and phone calls could be tailored to address patients’ needs. Most health providers reported that having an alert system to the clinic teams, and the general parameters for the alerts (i.e. decline in ICS adherence or increase in albuterol use) were parts of the intervention that worked well for their clinics. They also felt that the phone calls to families, triggered by an alert, could be tailored to meet the families’ needs because health providers could assess the context (e.g. forgetting to pick up refills) of the alerts and provided appropriate education around the barrier to ICS adherence. For example, one nurse commented:

There were quite a few times where I would call … because the family had not been using [inhaled corticosteroids] and they would be like oh yeah, I’ve been meaning to call you, we do need a refill. There were a number of times that I brought them in for visits and they ended up [needing a course of … steroids. -PR6 Nurse

Health providers also reported that the alerts on albuterol gave the clinical team an opportunity to detect when families might not have realized asthma symptoms were severe and would recommend scheduling an acute clinical visit or going to the ED. Health providers, however, also noted the intervention required refinements in other ways prior to wide-scale implementation, such as enhanced integration into the electronic health record (20%), increased portal access and functionality to enable documentation and status updates (20%), and standardized onboarding and continued education to orient staff to the intervention and processes (20%). A few health providers (40%) felt that a different process was needed for following up with patients about alerts, including a text message to families prior to calling.

Relative advantage

The relative advantage of an intervention refers to the advantage of the intervention versus an alternate solution to improve patient’s asthma management, which was usual asthma care with no sensor technology. Health providers felt that EMM with the iTRACC intervention was an unduplicated innovation. Health providers (100%) reported there were no previous clinic efforts to improve medication adherence detection. For example, one provider shared:

iTRACC tracked usage and that’s what made it different because then we could see them using [their inhalers]. When we do teaching, we teach and then we hope they use it. iTRACC was actually tracking if they used it at all. -PR1 Nurse (emphasis added by authors, based on interviewee’s inflection)

Complexity

In CFIR, the complexity construct represents the perceived difficulty in using the intervention. Health providers (40%) felt that the intervention processes, such as calling patients back after alerts, fit their typical clinical responsibilities. However, they highlighted that additional volumes of calls (30%), due to inability to reach families with the first call, needing to resolve alerts in the portal, and separately documenting calls in the electronic health record (EHR) added to the complexity of the intervention (50%). The extra phone calls for nurses and medical assistants and the separate interface from the EHR for documentation were added barriers to integrating the intervention into their workflow.

Inner setting

Readiness for implementation

Readiness for implementation represents the clinic’s commitment to intervention implementation measured by access to information and knowledge of the intervention or availability of intervention support resources. Most health providers (90%) identified major gaps in readiness for implementation in available personnel and preparedness to work with innovative technology in outpatient care. In iTRACC, health providers reported no major clinic changes were made to accommodate the intervention, such as dedicated hours to responding to alerts or intermittent training on the EMM technology. Thus, most health providers shared that additional staff would be necessary for a non-research supported implementation of the iTRACC study model. Health providers felt more staff was necessary because nurses and medical assistants were often the first respondents to EMM generated alerts, were part-time at the clinic sites, and were unable to respond to alerts 7 days a week. For example, one medical assistant explained the additional resources needed for implementing the intervention:

I mean we [would need] … education on how … everything is used, the device and apps and … I don’t know how long it would take to explain it to a patient but you would have to maybe have somebody more designated to be able to explain the process to everybody so they would have to hire probably somebody else just with the busy flow of what we already do … to take 20 minutes of our time to explain something to a person would really back things up honestly. -PR9 Medical Assistant

Further, health providers (30%) explained that successful implementation of an EMM would require that all staff be aware of the technology and have “buy-in” by providing additional training and educational materials to providers and helping providers interpret and respond to the data. Lastly, a health provider highlighted that implementation of EMM should be “seamless” with the EHR so that the EMM data could be easily integrated in the clinic visit with asthmatic patients.

Characteristics of individuals

Knowledge and beliefs about the intervention

In CFIR, individual-level factors are emphasized, specifically, the knowledge and beliefs construct refer to the individuals’ attitudes toward and value placed on the intervention. When asked about their expectations of the intervention, health providers’ responses revealed themes in three areas: monitoring of medication use, asthma education, and increasing communication with patients. Health providers (100%) reported that overall the intervention enhanced their tracking of asthma medication use and expressed surprise at patients’ low level of adherence to ICS. Provider beliefs were mixed regarding whether the intervention was an effective means of asthma education. A few health providers (30%) explained that the intervention created opportunities for them to discuss how parents should follow the asthma action plan or ask for medication refills as soon as possible. While some parents had already contacted their clinic or gone to the emergency department by the time clinical staff called about the alerts, other providers (40%) felt EMM increased communication by helping to detect which patients needed clinical evaluation prior to parents contacting the clinic. One provider described:

A secondary option for people [who] might have not thought to contact their doctor yet but it gave us like a little ding to let us know that you know maybe this person needs some extra help or like what’s going on with them –PR9 Medical Assistant

Other health providers felt that effective monitoring, education, and communication were complicated by problems with sensor syncing leading to false alerts (50%) and inability to reach families by phone (30%). Some health providers were concerned that the sensor data may reflect over-detection or under-detection of asthma issues. Provider examples of this included increased albuterol frequency may be related to appropriate pre-exercise treatment, while missed ICS doses may be due inability to sync the phone and sensors when parents were out of town. Providers descriptions of the need to contextualize data were tied to provider hesitancy about how to use EMM effectively in outpatient care.

Self-efficacy

Another individual-level construct was self-efficacy—that is, the individual’s belief in their own capabilities to use the intervention and achieve implementation goals. When asked about their level of confidence in the two main responsibilities with the iTRACC intervention—calling families about alerts and using the Web portal—most health providers (80%) reported that they felt confident. A nurse stated, “I knew what I was looking for, what kind of questions I needed to ask, and who I needed to send that to.” (PR3) When asked about the use of the Web portal, most health providers described it as easy to use, even those who found it difficult in the beginning reported they were able to rapidly familiarize themselves with it. For example, one nurse explained, “The first couple of times it was kinda like iffy but then once I started using it more it became more familiar and easy to use.” (PR1)

Discussion

Through our qualitative investigation of health providers’ experiences of EMM in asthma care through a CFIR lens, we identified areas of strengths and needed improvements to implement EMM in pediatric outpatient asthma care. The iTRACC intervention was multicomponent with the cornerstone being the EMM technology. In the present study, we focused primarily on the CFIR domains of intervention characteristics, inner setting, and characteristics of the individual to frame our interview questions and gain understanding of provider experiences with the technology-enhanced, asthma intervention.

For intervention characteristics, health providers were nearly unanimous in their agreement that there was no equivalent to EMM in outpatient asthma care. While EMM technology provided opportunities for enhanced adherence monitoring, asthma education, and patient communication in outpatient care, our findings suggest there were also unique challenges that the technology introduced. Two challenges of using EMM in outpatient asthma care were the needs to: 1) integrate sensor data and staff responses to the data into the EHR instead of a web portal; and 2) provide continued education for clinical staff about how the EMM tool worked. The push to have patient-generated data—i.e. data created, recorded, or collected by patients or their families, typically by sensors, activity trackers, or other mobile applications—integrated into the EHR is an emerging area of interest nationally (22,23). Previous studies, including an asthma study collecting patient-reported outcomes through a web portal in pediatric primary care, have demonstrated low adoption when data were collected and stored separately from the EHR due to workflow barriers (24). Integration of sensor data, however, requires commitment from EHR companies or health systems and is typically beyond administrative scope and capital investment capacity for local clinic practices. Thus, to fully implement EMM technology-enhanced asthma care, it is critical to address this data integration barrier.

In addition, while health providers were provided training with the intervention at the beginning of iTRACC, health providers indicated that they were unable to answer parents’ questions about the device and mobile app throughout the trial period. Health providers emphasized that continued support around the devices, web portal, and other intervention components are necessary throughout implementation. Since health providers only interacted with one facet of the intervention, they were unfamiliar with how to answer parent questions about troubleshooting the EMM technology. To enhance health providers’ comfort in responding to questions, they must fully understand the technology and what parents are experiencing in order to support and reap the benefits of the intervention. Conducting initial and repeat training sessions for health providers on how the technology works and how users (i.e. parents) interact with the mobile app could assist health providers in answering future technology-focused questions. Lastly, a few health providers felt that the intervention needed to adapt a different process for following-up with patients about alerts. The call burden and inability to connect with patients were inefficient, and providers suggested a future design should include a text message to patients, when possible, to confirm whether phone call outreach was needed.

In considering clinics’ readiness for implementation, health providers felt existing resources, especially personnel, were insufficient to support scaled-up implementation. At present, the current clinic workflows were not designed to effectively implement EMM at an individual practice level given the additional clinical and administrative effort needed. One alternative to overcome practice-level workflow barriers would be to implement EMM at the health system level. Embedding in a health system through a population health model could provide consistent monitoring, communication with patients and practices, and provide support needed to sustain EMM where data are shared with patients and health providers. This is consistent with recommendations proposed for enhancing effectiveness of a pediatric asthma telephone peer-coaching program designed to provide parent support through telephone calls (25). Specifically, the authors discussed that a return on investment across a low-income population with asthma would likely lend itself to a health systems-level, rather than practice-level, implementation model. The same conceptualization could be applied to use EMM in a centralized way, given the high administrative burden created for individual practices.

From the CFIR constructs of characteristics of individuals, health providers’ belief and knowledge of the intervention were associated with needing more support in contextualizing the sensor data. The need to contextualize patient-generated data is a common concern with technology-enhanced interventions, where data is filled with noise, variability, and missingness making it difficult for clinical interpretation (26). In the present study, the challenge of interpreting the sensor data for health providers was critical to the ability for health providers to use the data to meet their expectations for the intervention—enhanced communication and education around asthma with parents and children. Aligned with our findings on expectations, prior research theorized that similar principles of effective technology-enhanced interventions in chronic diseases, such as diabetes, was necessary for care (10). Specifically, Greenwood, et al.’s conducted a systematic review of technology-enhanced diabetes self-management and care and outlined four elements for effective technology-based, interventions: 1) two-way communication, 2) analysis of patient-generated health data, 3) tailored patient education, and 4) individualized feedback. To achieve these goals with a technology-enhanced, asthma intervention, the solution is multifold, requiring analysis of patient’s data patterns and trends, as well as examination of how best to transmit the information back to the clinical team. In busy outpatient pediatric practices, translating the rich, high frequency patient-generated data in a way that health providers can use for actionable feedback is significant. Addressing these implementation issues is one step to improving interventions in asthma self-management, but adherence remains a complicated issue, involving multiple barriers (27,28). Theoretically-supported interventions that address issues, such as caregiver and patient beliefs about treatment and self-efficacy, in addition to knowledge, is also necessary (29,30). Further investigation in partnership with health providers, technology developers, and patients would be needed to understand how to design a clinically integrated intervention to use EMM effectively.

Certain limitations are relevant to our study. Our qualitative exploratory study sample was small and limited to evaluation of one intervention in five clinics but consistent with qualitative research. Additionally, given the nature and aims of our study and deductive directed content analysis approach, we analyzed the data in Excel which is not uncommon for qualitative research (31–33). However, our analytic methods in Excel did introduce a couple limitations that deserve mention. First, although this was a deductive analysis based on CFIR, it is possible that extracting responses from transcripts may have precluded identification of themes or concepts that could only be observed in context of the full transcript. Second, because we extracted responses prior to analysis, we were unable to conduct a formal assessment of inter-rater reliability which reduces rigor of the research; however, we worked to overcome this limitation by having two analysts representing different backgrounds review and summarize the data independently and then resolved discrepancies through team discussion. We recognize also that health providers participated due to interest in the original trial, and hence that may introduce volunteer bias. We were additionally limited by the turnover in staff, whom may have had varying levels of interaction with the intervention. While we were able to capture negative and positive views of the intervention from champions, future studies should include a wider involvement of the clinical staff to ensure against positive response bias. Further understanding how EMM might not work for all patients, especially those who do not want to be monitored, must also be considered in future intervention design.

In the future, researchers should also consider a small-scale implementation-effectiveness trial to evaluate implementation effectiveness, such as adoption and fidelity; and further examine the other CFIR domains (e.g. outer setting, processes) through both quantitative and qualitative evaluation. Future multi-level evaluation should incorporate a range of stakeholders to ascertain how technology-enhanced, asthma care fits with the wider health system or clinic management’s organizational priorities and experience. Researchers could then also expand EMM implementation to additional outpatient sites, including federally-qualified health centers, where clinic structures may differ, and they predominantly care for low-income families.

Trends in self-management of chronic diseases have favored the emergence of technology-based tools, including the use of sensors and app-based programs (34–36). With hundreds of mobile health apps for asthma self-management available for patients, understanding how health providers engage with these tools and the user experience is crucial (37). Using EMM, as one form of digital health support in asthma care, remains promising; however, our findings revealed areas needed for improvement in order to ultimately facilitate improved asthma care for patients and health providers.