Digital Health Technology in Asthma: A Comprehensive Scoping Review

BACKGROUND: A variety of digital intervention approaches have been investigated for asthma therapy during the past decade, with different levels of interactivity and personalization and a range of impacts on different outcome measurements. OBJECTIVE: To assess the effectiveness of digital interventions in asthma with regard to acceptability and outcomes and evaluate the potential of digital initiatives for monitoring or treating patients with asthma. METHODS: We evaluated digital interventions using a scoping review methodology through a literature search and review. Of

Abbreviations used ACT-Asthma Control Test HCP-Health care professional ICS-Inhaled corticosteroids SABA-Short-acting b-agonist SMS-Short message service burden. Evidence of improvement in both adherence and patients' impairment due to asthma were seen only with interactive interventions involving two-way responsive patient communication. Digital interventions were generally positively perceived by patients and physicians. Implementation was considered feasible, with certain preferences for design and features important to drive use.

INTRODUCTION Rationale
Asthma continues to be a significant chronic disease burden, with estimated patient numbers approaching 350 million globally. 1 Asthma is a major source of direct and indirect economic cost in addition to reduced quality of life and premature death in patients of all ages, including children. 2 Several areas of unmet clinical need are associated with conventional asthma management; while medications such as inhaled short-acting b-agonists (SABA) and inhaled corticosteroids (ICS) are efficacious, considerable long-term commitment to asthma self-management on behalf of the patient is required for therapeutic success.
Asthma self-management includes advice and education on self-monitoring and a defined asthma management plan supported by regular professional review. Self-management is recognized as an effective method to improve asthma control and quality of life and reduce unscheduled consultations and hospitalizations across diverse demographic groups. 3 For the patient, adherence to medication regimens is a key element of successful asthma management, in conjunction with wide-ranging factors including disease perception, patient education and understanding of symptoms, comorbidities, inhaler technique, access to medications, and health care professional (HCP) support. 4 On the part of clinicians, implementation of evidence-supported management guidelines for assessment, monitoring, patient education, control of environmental factors, and pharmacologic treatment is a key factor, although adherence to such guidelines varies among different physician groups. 5,6 Several areas of asthma self-management could be improved using digital technology, including disease disparity, medication adherence, patienteclinician communication, personalized patient education, and passive monitoring of patient characteristics and behaviors that could prompt timely intervention. Technological innovations have the potential to support patients with disease self-management, allow for remote management by health care providers, and make future asthma management more proactive. 7 Researchers have explored a variety of different digital asthma approaches including passive education, interactive websites, and electronic medication monitoring devices. Given the increasing prevalence of asthma worldwide and patient dependence on traditional inhaler devices to manage the disease, the inclusion of more advanced technology as a part of asthma self-management will be an essential next step to improving therapy for respiratory diseases. Lockdown and social distancing as a result of the COVID-19 pandemic have caused increased uptake and reliance on videoconferencing and telemedicine in various therapy areas. 8 Increased familiarity with telemedicine provides an opportunity for remote monitoring and disease management, particularly in areas where access to HCPs is challenging.
Given the wealth of available literature and heterogeneity of digital technologies used to varying degrees of success, we performed a scoping review to assess the available research in this field. Scoping reviews provide a preliminary assessment of the size and scope of available literature on a topic and characterize the nature of published studies to provide an overview of large, heterogeneous bodies of literature. Furthermore, scoping reviews provide a platform for more precise systematic reviews to look at specific research questions. This scoping review therefore seeks to evaluate the different options that have been explored and assess the future utility of digital health technology in asthma.

Objective
This study aimed to evaluate the current and potential future usability, acceptability, uptake, effectiveness, and adoption of digital health technology for treating or monitoring asthma using a scoping review methodology. The potential future reach and durability of these technologies will also be considered.

Scoping review approach
We conducted a systematic scoping review based on the methodology described 9 and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews guidelines, 10 including identifying the research question, identifying relevant studies, selecting the study, charting data, collating, and summarizing and reporting the results. The purpose of a scoping review is to give a broad overview of the available literature without being directed toward a single discrete research question.

Sources and searches
We searched PubMed, Embase, and the Cochrane Library for January 2010 to June 2019 using the search terms and criteria in Table E1 (in this article's Online Repository at www.jaci-inpractice. org), adapted for each database, to identify asthma studies that assessed the effect of digital health technology interventions on patient outcomes, including symptoms, adherence, and medication use or therapy regimen. We also included studies that incorporated the acceptability of digital interventions to patients and HCPs and the feasibility of their wider uptake. Congress abstracts were excluded from the results. Database searches were supplemented by manual searches and references as appropriate.
Nonrelevant articles and duplicates were excluded; the titles and abstracts of remaining citations were reviewed by two authors (G.S. and G.M.) in parallel; a third author (R.K.M.) adjudicated discrepancies. Full-text articles were then obtained and a second round of screening was conducted (two reviewers worked in parallel and adjudication was performed by a third reviewer) and further articles were excluded as necessary. Articles were assessed for the type of digital health technology employed, study design, size of the study, and outcomes assessed.

Data extraction and synthesis
The full text of the selected articles was used to identify the types of technology used. Interventions were grouped (Table I)    basis (eg, short message service (SMS)-based dose reminders and automated pharmacy prescription refill calls, or access to noninteractive, nonepatient specific digital educational material) were termed 'generalized studies'. Interventions in which individualized information was gathered from or sent to patients (eg, parameter monitoring by a device with no feedback to the patient, or provision of patient-specific education) were termed 'patient-specific, noninteractive studies'. Interventions composed of an interactive relationship with the patient, in which insights based on data collected from the patient or resulting from patient behavior were fed back to the patient to assist with asthma management on a personalized basis (eg, an asthma management platform responding to adherence data from a digital inhaler or questionnaire feedback from the patient) were termed 'patient-specific interactive studies'. Within these broad intervention categories, studies were further grouped according to the type of technology involved ( Table I).
Details of the studies performed (eg, subject characteristics, sample size, randomization) and outcomes, including adherence, asthma control or impairment and health care use were captured in a standardized table format as part of the data charting process. Adherence outcomes were assessed for being subjective (ie, reliant on questionnaire data or patient recollection) or objective (ie, monitored automatically as a result of medication use, including one study that calculated adherence from electronic medical records).
To assess the impact of the different intervention categories, we identified studies in which a positive effect (vs no effect or a negative effect) was reported with the intervention in three domains: adherence to ICS (and long-acting b 2 -agonist) therapy, impairment caused by asthma (most commonly assessed by Asthma Control Test [ACT] score, but including changes in symptoms and quality of life when reported), and health care resource use (urgent consultations, emergency department visits, or hospitalizations). Because of heterogeneity in data reporting, to include as many studies as possible in our broad overview, we considered statistically significant, numerical, and anecdotal reports of benefit as evidence of positive effects in each of these three domains. Counts for studies reporting positive versus no effect or negative effects were then summarized.
Articles that reported patient and HCP perceptions of digital technologies were assessed separately, and relevant observations were extracted.

Selection of sources of evidence
Our selection process identified 121 articles for inclusion in this scoping review ( Figure 1); 17 articles described generalized studies, 23 described patient-specific, noninteractive interventions, and 66 described patient-specific interactive interventions. Figure 2 shows a bubble plot of the number of articles for each intervention category published each year. During the past 5 years, interactive interventions were investigated more frequently than noninteractive initiatives, possibly because of the increased availability of mobile Internet access and personal smart devices. Fourteen articles were identified that investigated patients' and clinicians' perceptions of digital interventions. Table II lists the number of studies reporting positive effects on outcomes for different intervention types.

Generalized studies
There were 17 generalized interventions, eight of which employed reminders or automated phone calls for medication doses or pharmacy fills, and nine that provided asthma education or asthma management motivation (n ¼ 9). Most were randomized (n ¼ 12) and most had an intervention period of 3 or more months (n ¼ 15) ( Table III).
Six of the studies described a change in adherence with the intervention; all of these reported a positive effect (6 positive vs 0 no effect or negative). Adherence in these studies was measured using both subjective (n ¼ 3) and objective (n ¼ 3) approaches. Few studies of generalized interventions reported a positive effect on asthma impairment (3 positive vs 7 no effect or negative), and none reported meaningful improvements in health care use (0 positive vs 5 no effect or negative) (Table II and Figure 3).

Patient-specific, noninteractive, unidirectional interventions
Of the 24 studies using patient-specific, noninteractive, unidirectional interventions, most focused on either digital questionnaire-based studies (n ¼ 9) or device monitoring studies (without feedback to the patient; n ¼ 8). Thirteen studies were randomized and 11 were single-arm. Nine interventions lasted for less than 3 months (Table IV).
Overall, few studies reported positive effects; only two studies found benefit in each of the adherence and asthma impairment domains and none reporting improvements in health care use (Table II and Figure 3). Positive effects on adherence were noted in one study of education provision and one involving adherence tracking with an automated reminder call; moreover, positive effects on asthma impairment resulted from one study involving a smartphone symptom and ACT diary with additional education and motivation, and one study in which a decision support algorithm was used to enter patients into different randomized controlled trials.

Patient-specific interactive, bidirectional interventions
Patient-specific, bidirectional interactive interventions were the largest group of interventions (n ¼ 66). Most often, these entailed an asthma management platform (n ¼ 32), a digital inhaler combined with an asthma management platform (n ¼ 14), or a telemedicine-based intervention (n ¼ 14). Most studies (n ¼ 45) were randomized, and 15 of the interventions lasted less than 3 months (including two with the study period not stated) (Table V).
Reported positive effects outweighed no effect for both adherence (17 vs 8 studies, respectively) and asthma impairment (31 vs 15 studies, respectively) (Table II and Figure 3). Adherence was measured subjectively (n ¼ 18) and objectively (n ¼ 14) in a similar number of studies. Twelve studies found reductions in health care use, although a similar number (n ¼ 14) found no effect.
This was the only intervention category within which improvements in both adherence and asthma impairment were reported more often than not and were seen consistently across the three most-studied intervention types.

Acceptability and feasibility of interventions
When reported, all types of digital interventions were considered to be acceptable or favorable to patients or their caregivers and feasible to implement, although this aspect was not rigorously investigated in articles reporting intervention results. 13,14,[16][17][18][19][22][23][24]26,28,43,56,87,105,108,109,111,115,117 Digital inhaler systems were described as feasible to implement 31,61,62,65,[69][70][71] and acceptable to most patients. 11,31,61,70,118 Similarly, digital asthma management platforms were reported to be both feasible 52,57,82,84,93,119 and acceptable. 75,76,92,96,[100][101][102]107 One study reported low use of a Web portal that was not specifically targeted to symptomatic asthma patients. 81 Although patients were generally satisfied with digital devices and were willing to use them, alerts were sometimes perceived as unwelcome if they were received at inopportune times. 35 Noninhaler devices (including Fitbits (Google LLC, San Francisco, CA) to track activity levels and sleep, and devices designed to assess lung function) were also considered acceptable by patients. 35

Cost-effectiveness
Only one 112 of the five studies to comment on costeffectiveness 13,22,109,112,116 found that the digital intervention was not cost-effective, although formal investigations of costeffectiveness were not performed. A smart nebulizer system was associated with decreased costs of additional treatments associated with asthma, such as antibiotics and corticosteroids. 64 Web-based monitoring of asthma control was deemed likely to be cost-effective after an analysis of an economic model. 39

Perceptions of digital health technology
Fourteen articles commented on perceptions regarding digital interventions, 95,119-131 and all intervention types were widely considered to be acceptable and useful across groups of patients, prescribers, and families.
Patients' uptake of smartphone apps was affected by perceived need (ie, favorable if asthma not considered to be wellcontrolled) 125 ; for smartphone app uptake in general, ease of access (ie, downloading) was considered important. Although social media could attract many initial downloads, long-term engagement was more successful with face-to-face recruitment by practices. 124 Adolescents thought that smartphone apps should be designed to be visually simple, easy to use, informative, and customizable, and include familiar components (eg, alerts that appeared and sounded like their usual messaging apps), and they were open to using apps for support and information. 121,122,130,131 They preferred time-based to event-based reminders 44 and favored video and peer-chat functionality in addition to medication reminders. 119 They were also sensitive to design choices that were too childish for their age range, but appreciated gamification of asthma management. [121][122][123] Adolescent patients and pharmacists had a positive attitude toward interactive mobile health interventions, 119 and adults and adolescents were receptive toward educational and directive SMS messaging; adolescents were also in favor of sending messages to a support person (ie, parent or guardian). 125 Tablet-and smartphone-based asthma questionnaires were acceptable to patients and useful in certain situations (eg, when HCP time was limited, or in rural areas). 126 Patients and caregivers were supportive of interactive Web-based education provision based on patient information and preferences, and of augmented reality inhaler technique training. 128 Patients considered that smart devices and inhalers should be easy to use and convenient in terms of size and portability. 95,127 Clinicians valued the concept of intervisit improvement of adherence and lung function data and considered adherence data to be especially important, with a preference for ICS adherence data. Rescue inhaler information was considered valuable to access if patients were overusing them, because this could indicate a need to adjust the ICS regimen. The opportunity to use a patient's data to illustrate how adherence affected symptoms or impairment during consultations was also appreciated. 120 Some interesting concerns for clinicians included liability for acting on real-time data collection between visits when they could not respond, a preference to see data only before patient visits, and validation of data accuracy and security when devices could be shared. 120 outcomes from asthma therapy. Such interventions usually focus on improving patient's adherence to their treatment regimen, through dose and prescription-fill reminder systems, behavior change as a result of education and motivation, or increased access to health care professionals and asthma advice through telemedicine or interactive asthma management platforms. We considered the interventions identified by our literature search based on their interactivity with patients, and whether data communication was unidirectional (data sent either from or to the patient) or bidirectional (data received from the patient prompting a response from the intervention). In terms of study outcomes, we looked for evidence of improvement with the various interventions in terms of patients' adherence to ICS regimens, improvements in impairment owing to asthma, and improvements in health care use (by reductions in unplanned or emergency consultations).   Generalized dose and prescription reminders and nonpersonalized education and motivation were effective in improving adherence, but not impairment owing to asthma. Although subjective measurement of adherence is thought to overestimate adherence compared with objective measurement, 33,36 both types of adherence measurement were employed across these generalized studies; it is possible that adherence improvements were not sustained long enough to make a meaningful impact on asthma impairment. The patient-specific, noninteractive interventions included several different intervention types, but few reported positive impacts on either adherence or asthma impairment. However, the most-studied category of intervention, patient-specific interactive bidirectional interventions, reported the greatest number of improvements in adherence and asthma impairment. These benefits were largely found in studies of the most interactive interventions, those using interactive telemedicine, or those using an interactive management platform, with or without a digital inhaler. Each of these intervention types provides a high level of interactivity and instant feedback for patients through either live HCP consultation (telemedicine) or responses to monitored data or questionnaires (digital platforms with or without digital inhalers). Correlating patients' own asthma experiences (through ACT scores or other measured parameters) with their asthma management behaviors (eg, ICS adherence, trigger avoidance) may be a powerful engagement tool to motivate and educate patients about the need to stay on top of managing their condition.
We were able to identify broad trends in adherence and changes in asthma impairment among different intervention categories. The included studies generally did not find differences in health care use with interventions, either because it was not studied or because the incidence of emergency department visits or other unscheduled consultations was generally low.
All types of digital health interventions were well-received and acceptable to patients, caregivers, and clinicians. It is intuitive to think that digital interventions would be easily scalable to large populations and therefore be cost-effective, but currently there are few formal cost-effectiveness studies to confirm this assumption. In particular, the translation of improved adherence to improved symptom control and reduced health care use is not guaranteed, although several studies noted reductions in reliever or oral steroid use and reductions in unscheduled primary care visits. There may also be difficulties in transitioning coding and billing practices to accommodate digital interventions.
The progression of digital health monitoring and associated interventions in asthma still faces significant challenges. It is clear that a successful intervention must be able to combine objective monitoring with active patient engagement and education or motivation to maintain behaviors helpful to asthma therapy, and we have yet to develop an ideal system in this respect. Although there is further opportunity to incorporate digital interventions into patient care with respect to linking to patients' electronic health records and using data monitoring to trigger further HCP interventions, there is also the possibility of providing regular patient support through education about inhaler technique, avoidance of asthma triggers, and environmental monitoring and advice (eg, warnings related to high pollen count or air pollution), which could be equally important in reducing asthma burden.
The extensive implementation of digital interventions is attractive from the perspective of real-time population health monitoring as well as the increased opportunity and ability to    improve individual asthma outcomes through treatment personalization and daily disease management. Data analysis may also assist the understanding of relations and causal mechanisms among patient characteristics, environment, and asthma symptoms and exacerbations, ultimately leading to the development of patient digital phenotypes.
However, beyond the challenge of developing suitable technologies, considerable financial, technological, and political barriers still need to be overcome. The development of artificial intelligence protocols to interpret and apply the huge amount of monitoring data involved will be important. There will also be a considerable emphasis on the infrastructure necessary to incorporate continuous monitoring into patients' overall health care to support physicians. Such a fundamental change in the organization and delivery of health care will also have significant effects on liability for physicians making artificial intelligenceaugmented decisions. 133,134 Large-scale personal health data collection naturally generates concern regarding data security and use, and patients and physicians must be confident about how their data will be used, who will have access to it, and the extent to which individual patients will own their data. 135 Regardless of pharmacological therapy, consultations can already be implemented regularly by videoconference, which is especially important for patients with reduced mobility or reduced free time, or in remote locations; videoconferencing could also be used to provide support between face-to-face appointments. In the context of a pandemic such as COVID-19, reducing face-to-face contact between clinicians and selected patients has been a necessity. The use of telemedicine and video consultation was implemented as an elegant solution to this challenge for patient assessment and education, both of which can be delivered remotely. 136,137 One related effect of the current pandemic situation is a large-scale uptake of videoconference technology for work and social contact in the general population, which could help acclimatize patients and physicians to accessing and providing health care in this way. We consider it likely that many of the benefits of telemedicine realized during the pandemic, initially through necessity, will continue to gain increasing acceptance and form an integral part of respiratory care in the future. 137,138 Limitations Conclusions from this scoping review are limited by heterogeneity in terms of the wide variety of study designs and outcome measures and the different ways of implementing similar interventions for patient-independent and patientdependent initiatives. In addition, study heterogeneity limited our interpretation of different outcomes for the same intervention category (eg, the changes observed in adherence were not necessarily measured in the same studies as changes in ACT). The ACT was the most common measure of asthma burden and is a subjective measure of patients' perception of impairment resulting from asthma, rather than a measure of risk for an asthma exacerbation. Changes in asthma exacerbations were not commonly used as a primary outcome measure, possibly because the expected number of exacerbations in study populations was not sufficiently frequent to make this a useful end point.
Because patients were generally monitored for no more than 12 months (and commonly only 3 or 6 months), the long-term effects of the various interventions are unknown. Studies also typically did not account for seasonality or incorporate environmental information (such as aeroallergen or air pollution levels). Most of the studies analyzed did not specify participants' asthma severity at baseline and lacked stratification by asthma severity. An abundance of studies also reported outcomes only at single specific time points or as a single assessment for the whole study period, rather than as change over time associated with a specific intervention. The relation between asthma controller adherence and health care use is an important issue, but one that is challenging to study. In addition to the acknowledged problems with the accuracy of self-reported adherence, 48,139,140 in a general population of treated asthma patients who are not under close clinical supervision, it would be expected that the frequency of emergency or unplanned clinical consultation would be low, and therefore statistically significant changes would be difficult to detect. The studies identified for this review rarely assessed health care use systematically, and few reported on both adherence and health care use. Furthermore, correlating changes in adherence with changes in use outcomes is confounded by several patientrelated and sociobehavioral factors including symptom severity, ethnicity, level of educational achievement, household income, and attitude toward treatment. 141 Therefore, the fact that reductions in health care use were not reported with digital asthma interventions in the studies discussed in this review does not mean that these interventions are not associated with benefit in terms of health care resource use in the real world. The patient populations, intervention assessments, and data collection methods might not have been capable of detecting health care use changes accurately, even where these may have occurred. Further research into the effects of adherence on health care use is undoubtedly needed, and considerable effort will be required to account for these factors in the design of future studies assessing digital interventions. Furthermore, future research into digital technology as a part of asthma interventions would benefit from a convention on standard approaches to outcomes and their measurement to enhance similarity in the ways in which we assess asthma treatment.

CONCLUSIONS
Digital health interventions are acceptable to asthma patients and clinicians, and the relative ubiquity of mobile Internet-and Bluetooth-connected devices in everyday life makes real-time monitoring of asthma treatment and symptoms extremely feasible. However, an effective digital intervention must fulfill several important design criteria. Both inhaler devices and software should be intuitive and easy to use, and hardware should be unobtrusive with accurate and objective measurement of adherence and other parameters, and be seamlessly integrated with software providing convenient assessment of symptoms and asthma control. Software should be designed to educate and maintain patients' engagement (at a variety of different ages), provide clear communication of hardware-and softwaremeasured parameters for access by patients and clinicians, and be customizable by patients to fit into their lives and routines. Motivating patients to take an active role in managing their condition and maintaining their interest in doing so is of paramount importance in designing a successful future digital intervention.