Adults with asthma experience no increase in asthma related exacerbations when digital communication tools are employed to offset provider workload: a pragmatic randomized trial

Abstract

Background.

Challenges to healthcare efficiency are increasingly addressed with the help of digital communication tools (DCT).

Objectives.

To test whether DCT, compared to usual care, can reduce healthcare clinician burden without increasing asthma-related exacerbations among patients with asthma in a large integrated healthcare system.

Research Design.

The Breathwell program was a pragmatic, randomized trial at Kaiser Permanente Colorado, where asthma nurses screen patients for poor symptom control when beta₂-agonist refill requests came within 60 days of previous fill or in the absence of a controller medication fill within 4 months (beta agonist overfill; BAO). 14,978 adults with asthma were randomized to Usual Care or one of two DCT intervention groups (Text/phone call or Email).

Subjects.

Participants included adults 18 and older with an asthma diagnosis at the time of randomization and no history of chronic obstructive pulmonary disease.

Measures.

Primary outcome measures included asthma-related healthcare resource utilization (e.g., asthma nurse contacts), medication use, and exacerbations.

Results.

A total of 1,933 patients had 4,337 events which met BAO criteria. Of the 2,874 events in the intervention arm, 1,188 (41%) were resolved by DCT contact and did not require additional clinician contact. Asthma medication use and exacerbations over 12 months did not differ among the three groups.

Conclusions.

DCT tools can successfully contact adult asthma patients to screen for symptoms and facilitate intervention. The absence of differences in medication fills and healthcare utilization indicates that the strategic replacement of nursing interventions by digital outreach did not reduce treatment adherence or compromise healthcare outcomes.

INTRODUCTION

Healthcare delivery in the United States faces significant challenges. Americans spend twice as much as 10 other high-income countries while remaining less healthy.(1) As early as 2007 in the US, direct medical expenses (which excludes lost wages and productivity) attributable to asthma were estimated to be more than $56 billion.(2) Current challenges to asthma healthcare include inadequate access, inefficiency, overburdened clinicians, and gaps in patient education including self-management skills and medication adherence.(3–6) The introduction of digital communication technology (DCT) has produced new tools that may address some of these barriers. In particular, DCT tools have the potential to reduce clinician burden while helping to effectively engage patients.

Employing DCT to connect patients and clinicians.

Reduction in barriers and improvement in healthcare efficiency may be more readily accomplished with DCT that can reach out to patients, integrate patient- and electronic health record (EHR)-generated information, suggest solutions to barriers, and facilitate timely patient-clinician discussion. Digital connectivity is increasingly accepted by patients as a means of communicating with their healthcare team.(7) The combination of digital and human touch, in the right proportions, can improve the overall experience for patients while increasing efficiency of healthcare delivery.(8) It is important to establish the optimal synergy between human and DCT care using strong research designs to determine when, how, and in what combinations these tools improve health and disease management.

We previously conducted a pragmatic, randomized controlled trial (RCT) with pediatric asthma patients within Kaiser Permanente Colorado, using interactive voice response (IVR) to call parents and encourage refilling of their child’s asthma controller medication. The automated intervention both improved asthma controller medication adherence and linked patients to the clinical asthma care manager (ACM) staff.(9) The present pragmatic RCT, Breathewell, takes a similar approach, combining technology and human interaction to more efficiently provide asthma care to nearly 15,000 adults with asthma. Breathewell was designed to test the hypothesis that a DCT intervention designed to screen for current asthma symptoms can reduce demand on healthcare staff resources without compromising patient asthma-related outcomes.

METHODS

Design

The Breathewell study was a pragmatic RCT designed to test the implementation effectiveness of a DCT intervention. Pragmatic trials are typically large trials assessing healthcare delivery with interventions conducted in multiple clinical settings.(10, 11) Study enrollment began in February 2017 and continued for 12 months; study follow-up continued for a minimum of 6 months (for those who first met study enrollment criteria in February 2018) and up to 18 months (for those who met study criteria in February 2017). The intervention was delivered through a program utilizing information from the Kaiser Permanente Colorado EHR database. This study was approved by the Institutional Review Boards of National Jewish Health and Kaiser Permanente Colorado.

Setting

This study was conducted within Kaiser Permanente Colorado, an integrated healthcare organization serving approximately 600,000 members in the Denver Metro area.

Subjects

Consistent with pragmatic implementation trials, participants included adults 18 and older who were current members of Kaiser Permanente Colorado and had a diagnosis of intermittent or persistent asthma at the time of randomization. Individuals whose most recent diagnosis was intermittent asthma had a further requirement that they had a diagnosis of persistent asthma at some point in the past year. Patients were excluded only if they had a diagnosis of chronic obstructive pulmonary disease or had previously documented that they did not want to be involved in research studies. Kaiser Permanente Colorado members are demographically representative of Denver, Colorado Metropolitan Area; 2% are Spanish speakers with limited English proficiency.

Approach

In Breathewell, standard care includes a group of ACMs, nurses who contact patients requesting a refill of their asthma reliever, usually a beta₂-agonist (BA) if: 1) the request occurs more frequently than every 60 days or 2) the patient attempts to fill a BA without filling an asthma controller medication (typically an inhaled corticosteroid [ICS]) in the last 4 months. Following the standard clinical protocol, the ACM reviews the patient’s EHR to determine if further outreach is needed. The underlying rationale for this process is that overuse of a BA is a predictor of asthma exacerbations,(12) particularly in the face of underutilized asthma controller medication.(13) In turn, outreach to asthma patients at the time of a BA request, including the offer of an asthma specialist appointment, significantly reduces excessive BA overuse.(14) However, this process is resource intensive for healthcare staff. Thus, Breathewell was designed to assess whether a DCT intervention can reduce asthma clinician workload without increasing asthma-related exacerbations.

Randomization

At baseline, all qualifying Kaiser Permanente Colorado adults with asthma who had not previously opted out of research were randomized to one of three groups: Text/phone call intervention, Email intervention, or Usual Care (Figure 1). Those who subsequently met BA overfill criteria received Text/phone call intervention, Email intervention, or Usual Care consistent with their baseline randomization. Patients who did not have a BA overfill (BAO) received no further study communication and were not included in the data analyses.

Figure 1.

Asthma Patient Randomization and Beta₂-Agonist Overfill Outreach

Intervention

Patients who met BAO criteria received Text/phone call intervention, Email intervention or Usual Care concordant with their previously-randomized group. Those in the Text/phone call group were contacted by text if their phone was text-enabled. Where phones were not text enabled, patients were contacted instead by an IVR phone call. The IVR system was already existent in Kaiser Permanente Colorado and used speech recognition software able to conduct a simulated human conversation related to a patient’s recent symptoms. The program utilized information in the EHR such as the name of the patient and the medication to tailor the conversation, asks questions, and gather responses. It distinguished cellular phones from land lines to determine whether the phone was text enabled, thus allowing the system to determine when to send a text versus a phone call.(15, 16) Patients in the Email group received an email. The text, phone call, and email scripts contained similar content (see below) and were programmed for automated delivery. Patients with a recorded preference for Spanish were sent the communications in Spanish. Regardless of the assigned study group and subsequent outreach, patients’ BA refill requests were filled as part of standard operating procedure.

The Text/phone call and Email communications explained that the patient was being contacted as a result of their request for BA medication refill and asked this question: “Other than when you’re getting ready to exercise, during the past 4 weeks have you used your quick reliever inhaler 2 or more times a week?” Previous research has established the capacity of that single question from the Asthma Control Test to identify patients at risk of an asthma exacerbation.(17) If the patient answered “No,” they were thanked, the encounter closed and a notation added to the EHR. If they answered “Yes,” a request for patient contact was forwarded to the ACM group through the EHR. Key information automatically drawn from the EHR chart regarding exacerbations and previous ACM contacts was provided to the ACM. The ACM then contacted the patient for a shared decision-making discussion about possible increasing symptoms, therapy adjustment, or a clinic appointment. In the case where the patient did not respond, the encounter was either closed or forwarded to the ACM based on asthma events in the prior year relative to the BA request date (Figure 2). For example, if the patient had been contacted by an ACM within the previous 90 days, the event was closed regardless of asthma symptoms. If patients with a clinical diagnosis of persistent asthma had not received an ACM call within the previous 90 days but had evidence of a clinical asthma exacerbation in the past year, defined as an emergency room, urgent care visit, hospitalization stay, or oral corticosteroid bursts with an asthma diagnosis, and had not filled their ICS within the last four months, they were forwarded to the ACMs for review. If during the study interval a patient met BAO overfill criteria more than once, they again receive the intervention. All standard clinical services, including telephone nursing and pharmacy consultation, clinic appointments, and access to educational information, were available to patients in both the intervention and usual care groups.

Figure 2:

Algorithm for determining Asthma Care Management outreach for beta₂-agonist overfills

Data analysis

The study population was characterized overall and by study group. Demographic variables such as age, gender, race, ethnicity, and asthma diagnoses were captured at the date of randomization. Census tract variables (income and education derived from the patient’s address and used as proxies for measures of socioeconomic status), body mass index, and tobacco and/or marijuana use were captured at the time of the initial qualifying BA refill. In descriptive analyses, the groups were compared using Kruskal-Wallis tests for continuous and ordinal variables. Fisher’s exact test was used for dichotomous variables and Pearson Chi-square for categorical variables with more than two categories.

To maximize capacity to detect differences in outcomes between intervention and control, the primary analyses were comparisons of outcomes between the combined intervention group (Text/phone call and Email intervention) and the Usual Care group. The secondary analysis were comparisons of outcomes among the two intervention groups, Text/Phone call versus Email.

Primary outcomes included healthcare resource utilization, asthma medication use, and asthma exacerbations. These were compared between the Usual Care group and the combined Text/phone call and Email intervention groups. Definitions, measurements, and interpretations of these outcomes are detailed in Table 1. Patients were followed from the time of the initial BAO refill (the one year intervention period was February 2017 to February 2018) until the end of the study (August 2018), disenrollment, or death. All patients had at least 6 months follow-up, while the maximum follow-up period was 18 months. Baseline health care utilization and asthma medication patterns were captured one year prior to the date of the first BAO refill; post period outcomes were obtained during the follow-up period.

Table 1.

Healthcare Resource Utilization, Asthma Medication Use, and Asthma Exacerbation Outcome Metrics, Measurements, and Interpretations

Outcome Metric	Components of Definition	Measurement	Interpretation
Primary Outcomes (Combined text/phone call and email intervention groups vs. usual care group)
Healthcare resource utilization	Beta2-agonist overfill-related contacts by asthma care managers	1) Number of contacts 2) Number of patients contacted	Difference/no difference between combined intervention groups and usual care group; desired outcome is fewer contacts by asthma care managers in combined intervention group
Asthma medication use	Beta2-agonist canisters (reliever medication) and inhaled corticosteroid (controller medication) canisters	1) Number of reliever canisters 2) Number of controller canisters 3) Asthma Medication Ratio (AMR), calculated as the ratio of the number of controller canisters dispensed divided by the sum of reliever plus controller canisters dispensed. AMR ≥ 0.5 associated with improved asthma outcomes (i.e., fewer asthma-exacerbations)	Difference/no difference between combined intervention groups and usual care group; desired outcome is no difference between groups
Asthma exacerbations	1) Oral/injectable corticosteroid bursts and 2) Asthma-related after-hours/urgent care visits, emergency visits, and hospitalizations	1) Number of oral/injectable corticosteroid bursts with an accompanying asthma diagnosis and 2) Number of asthma-related after-hours/urgent care visits, emergency visits, and hospitalizations	Difference/no difference between combined intervention groups and usual care group; desired outcome is no difference between groups
Secondary Outcomes (Text/phone call intervention group vs. email intervention group)
Healthcare resource utilization	Beta2-agonist overfill-related contacts by asthma care managers	1) Number of contacts 2) Number of patients contacted	Difference/no difference between text/phone call group and email group; desired outcome is no difference in contacts by asthma care managers between intervention groups
Asthma medication use	Beta2-agonist canisters and inhaled corticosteroid canisters	1) Number of reliever canisters 2) Number of controller canisters 3) AMR	Difference/no difference between text/phone call group and email group; desired outcome is no difference between groups
Asthma exacerbations	1) Oral/injectable corticosteroid bursts and 2) Asthma-related after-hours/urgent care visits, emergency visits, and hospitalizations	1) Number of oral/injectable corticosteroid bursts with an accompanying asthma diagnosis and 2) Number of asthma-related after-hours/urgent care visits, emergency visits, and hospitalizations	Difference/no difference between text/phone call group and email group; desired outcome is no difference between groups

Non-linear mixed models were used to analyze non-normal outcome variables that were measured pre and post of the BAO index event. A time indicator was used to distinguish the records for pre and post period outcomes. Intervention effects were estimated by an interaction between the time indicator and intervention versus usual care group indicator. Specifically, Poisson regression models were used to analyze count data (number of BAO refills, reliever canisters, controller canisters, and corticosteroid bursts with an asthma diagnosis) using the Poisson distribution and the log link function.(18) In the Poisson regression models, log person-years was included as an offset and overdispersion was accounted for due to the heavy right-tailed distributions of the count variables. Since the Asthma Medication Ratio (AMR; the ratio of asthma controller medication canisters to the total of asthma controller canisters plus asthma reliever medication canisters filled) was a fraction (counts in both the numerator and denominator) ranging from 0 to 1 inclusive, AMR was analyzed using the fractional logit model with the logit link function.(18) Finally, logistic regression was used for the composite healthcare utilization outcome of any asthma-related urgent care visit, emergency visit, or hospitalization since this was a dichotomous variable. We did not account for multiple hypothesis testing in all analyses.

Secondary outcomes included comparisons of healthcare resource utilization, asthma medication use, and asthma exacerbations between the Text/phone call vs. Email intervention groups separately. Analytic approaches were similar to those conducted for the primary analyses.

Sensitivity analyses were conducted for descriptive, primary, and secondary analyses removing patients that opted out, had incorrect or no contact information, or did not speak English or Spanish.

RESULTS

14,978 adults with asthma met inclusion criteria and were randomized to receive Text/phone call (4,953), Email (5,046), or Usual Care (4,979) outreach. As expected, given randomization, most baseline variables were balanced among the three study groups (Table 2). The only baseline variable with a significant difference among groups was “marijuana use ever”, which was slightly higher in the Text/phone group than in the Usual Care group (p = 0.04). Overall, patients’ average age was 49 years, 60% were female, 67% identified themselves as White, and 9% were current smokers. At baseline, 78% of patients were overweight or obese.(19)

Table 2.

Baseline Characteristics of 1,933 Adult Patients with Persistent Asthma

Characteristic *	All Patients (N=1,933)	Usual Care (N=655)	Intervention Groups (N=1,278)
			Text/Phone (N=657)	Email (N=621)
Age in years, mean (SD)†	48.8 (16.2)	48.2 (16.6)	48.5 (15.9)	49.7 (16.2)
Female, N (%)†	1167 (60.4)	400 (61.1)	389 (59.2)	378 (60.9)
Race, %†
Asian	2.1	2.0	2.7	1.5
African American	6.1	4.9	5.9	7.4
American Indian/Alaskan Native	0.8	0.6	1.1	0.6
White	66.8	68.7	64.8	66.8
Other	3.2	3.1	3.4	3.2
Hispanic ethnicity	16.8	17.0	17.8	15.6
Unknown	4.3	3.8	4.3	4.8
Asthma Diagnoses, %
Persistent only†	87.2	87.9	86.9	86.6
Body Mass Index (BMI) in kg/m2, %
< 24.9	22.1	22.4	22.6	21.3
25 – 29.9	32.0	33.9	32.5	29.3
>=30	45.9	43.7	44.9	49.4
Tobacco use, %
Current smoker	8.7	8.6	9.0	8.5
Former smoker	32.1	31.3	31.2	32.8
Never smoker	58.5	59.4	59.1	56.8
Unknown	0.8	0.8	0.8	0.8
Marijuana Use, yes N (%)‡	91 (4.7)	33 (5.0)	39 (5.9)	19 (3.1)
Less than high school education, % in Census block	9.7%	9.4%	9.8%	9.9%
Family income, median $ (5th, 95th %ile)	77,260 (39,654, 131,103)	77,250 (41,201, 132,092)	80,453 (38,856, 135,054)	76,578 (41,005,130,134)
Total number of ambulatory visits per patient, mean (SD)	5.3 (6.5)	5.0 (5.8)	5.3 (5.9)	5.6 (7.8)
Number (%) of patients with any asthma exacerbation§	127 (6.6)	41 (6.3)	50 (7.6)	36 (5.8)
Ambulatory appointments missed, % of patients
None	65.3	65.5	64.7	65.7
1	18.3	18.6	18.4	17.9
>=2	16.4	15.9	16.9	16.4
Insurance plan type, %
Traditional HMO	96.3	96.6	96.7	95.7
High deductible	7.9	8.5	7.2	7.9
Medicare	21.2	21.2	19.3	23.0
Medicaid	13.1	12.2	14.2	12.9
Other	3.4	2.9	3.2	4.2
Number of oral/injectable corticosteroid bursts per patient, mean (SD)	0.3 (0.8)	0.3 (0.8)	0.3 (1.0)	0.3 (0.7)
Asthma Medication Ratio (AMR), mean (SD)	0.5 (0.3)	0.5 (0.3)	0.5 (0.3)	0.5 (0.3)
Beta2-agonist canisters dispensed, mean (SD)	5.0 (4.7)	5.0 (5.4)	5.3 (4.7)	4.8 (3.9)
Inhaled corticosteroid dispensed, mean (SD)	5.9 (6.2)	6.0 (6.8)	5.7 (5.5)	6.1 (6.0)
Asthma care coordinator outreach within 90 days prior to outreach, N (%) with any	266 (13.8)	83 (12.7)	100 (15.2)	83 (13.4)

^*During the 12 months prior date of first beta₂-agonist (BA) overuse within study period unless otherwise indicated

^†At date of randomization (1/6/2017)

^‡p-value of .0405 using Fisher’s exact test comparing Usual Care, Text/phone, and Email groups. Continuous and ordinal variables evaluated using Kruskal-Wallis test, categorical variables used Pearson Chi-square, and dichotomous variables used Fisher’s exact test. With the exception of marijuana use, no significant differences at baseline exist.

^§After-hour visits, emergency department visits, and/or hospitalizations

From this cohort of patients with asthma, 1,933 (12.9%) met BA overfill criteria (657 in the Text/phone group, 621 in the Email group, and 655 in the Usual Care group). Within that group, 215 were excluded because they opted out at some time before their first BA overfill or were unreachable throughout the study, resulting in a final count of 1,718 patients across groups who met BA overfill criteria (557 in the Text/phone group, 523 in the Email group, and 638 in the Usual Care group) (Figure 1). Some patients had more than one BA overfill and hence required more than one contact. Thus, there were 2,874 BAO intervention events. Of these events,1,188 contacts (41%) that would have required an ACM contact in Usual Care were successfully resolved by DCT outreach.

In the primary analysis, no differences emerged between the Usual Care and combined intervention group on any health outcome including healthcare utilization, AMR, or controller medication use (Table 3). In secondary analysis, significant differences emerged between the Text/phone and Email groups for controller medication use (Table 4, supplemental digital content). A significant difference was also noted in the AMR, likely contributed to by the increase in controller medication use, a component of both the AMR numerator and the AMR denominator. Specifically, for the Text/phone group the adjusted number of controller canisters increased from 5.8 at baseline to 6.8, while in the Email group the change was 6.1 to 6.5 canisters (interaction p-value=0.03). Similarly, for the Text/phone group, the AMR increased from 0.52 to 0.57, while the Email group change was 0.56 to 0.58 (interaction p-value=0.05).

Table 3.

Outcomes of Patients with Beta₂-Agonist (BA) Overuse: Usual Care Group vs. Combined Text/phone call and Email Intervention Groups

Outcome*	Usual Care (N=655)		Combined Intervention (N=1,278)		P-Value†
	Baseline	Post‡	Baseline	Post‡
Number of beta-agonist overuse outreaches, mean (SE) §¶	0.63 (0.05)	2.30 (0.08)	0.69 (0.04)	2.30 (0.06)	0.36
Beta-agonist canisters dispensed during follow-up, mean (SE) §¶	5.08 (0.21)	4.71 (0.23)	5.05 (0.12)	4.96 (0.16)	0.19
Short-acting corticosteroid inhalers dispensed during follow-up, mean (SE) §¶	6.04 (0.27)	6.43 (0.26)	5.94 (0.16)	6.70 (0.20)	0.19
Asthma Medication Ratio (AMR) during follow-up, mean (SE) **	0.54 (0.01)	0.58 (0.01)	0.54 (0.01)	0.57 (0.01)	1.00
Number of oral/injectable corticosteroid bursts per patient during follow-up, mean (SE) §¶	0.28 (0.03)	0.23 (0.02)	0.31 (0.02)	0.24 (0.02)	1.00
Percent (SE) of patients with any asthma-related after-hour visits, emergency department visits, and/or hospitalizations††	6.3 (0.01)	7.5 (0.01)	6.7 (0.01)	6.1 (0.01)	0.23

^*All patients had at least 6 months follow-up. Maximum follow-up was 18 months. Follow-up time was defined as time from first beta agonist overuse to study end date or censoring (e.g., disenrollment from health plan).

^†p-values were obtained from the interaction term between the time indicator period and usual care-combined intervention indicator. Models were adjusted for person years except for Asthma Medication Ratio and asthma-related after-hour visits.

^‡By the cohort definition, post period had at least one beta-agonist overuse outreach, so the minimum value was 1.

^§Rates were adjusted for person years.

^¶Poisson regression

^**Fractional logit model.

^††Logistic regression used to evaluate 1 or more events vs. 0 events.

Sensitivity analyses included 1,718 (89% of 1,933 cohort) patients. Again, no differences were found between the Usual Care and combined intervention groups for any health outcome. In sensitivity analysis of the Text/phone vs. Email groups, the controller medication use and AMR differences were no longer significant.

DISCUSSION

This study demonstrated that a DCT implemented in a large integrated healthcare organization can successfully reach out to adult asthma patients to screen for increasing symptoms and facilitate intervention. The absence of differences in healthcare resource utilization and medication adherence across groups indicates that the strategic replacement of nursing intervention by digital outreach did not reduce treatment adherence or compromise healthcare outcomes. Notably, standard care within Kaiser Permanente Colorado follows current, evidence based guidelines and rates of urgent care visits and hospitalizations are relatively low compared to other patient populations.(9, 20) The automated intervention hence has maintained the same low rate of urgent care use. The intervention effectively programmed to work in conjunction with the EHR databases, alerted asthma nurses to contact patients requiring intervention. This in turn conserved nursing time and improved efficiency by directing their actions to more urgent patient concerns. Over 12 months, the DCT intervention successfully delivered 2,874 DCT interventions, reducing by 41% the number of ACM reviews that would have been required in usual care.

The digital engagement was EHR driven, not patient driven. In contrast to many app-based asthma programs or social media sites, the Breathewell intervention did not require patients to be motivated to download an app, request the intervention, record and submit information to their clinician, or join a social media site. The intervention also did not increase workload for physicians or other healthcare clinicians. The intervention aligned with the Kaiser Permanente Colorado EHR system, allowing digital tracking and outreach to occur automatically based on information obtained from the system database.

DCT interventions, when incorporated with an EHR system, can take over time-consuming tasks previously handled by clinicians. Based on these data, the Text/phone intervention produced slightly more controller medication refilling than did the Email group, possibly reflecting patient-expressed preference for the more rapid communication format.(21) DCT outreach can gather information from patients in real time, determine the need for further assistance, connect the patient with a care clinician, and make information available to facilitate a time-efficient discussion between clinician and patient. Previous DCT research has established that tracking BA (13) or inhaled corticosteroid use (22, 23) can help improve asthma treatment adherence and facilitate self-management,(24) although in most cases these interventions were not delivered in context with an EHR system. Strengths to this study include its implementation design and 12-month intervention in a large healthcare setting with existing clinicians, databases, and the EHR. Using EHR data, the Breathewell program was able to identify and longitudinally follow patients with persistent asthma, track their outcomes (emergency room visits and inpatient stays, prescription refills of asthma medications) and compare rates across randomized arms as well as to published incident rates. The multidisciplinary research team included operations leadership, clinicians, and researchers working collectively to assure successful implementation and early identification of problems. Some limitations to this study are noted. Healthcare systems comprised primarily of low-income, publicly-insured and minority patients may face other challenges that require different digital outreach than does this healthcare system (comprised primarily of privately-insured patients). Patients did not choose which outreach type (text/phone or email) they would receive; allowing patients to choose might increase response rate. A previous report established higher response rates to text messages than IVR phone calls or emails.(21) Whether subgroups of patients at greater risk for an asthma exacerbation, including those who smoke tobacco (25) or cannabis,(26) are overweight,(27) or underutilize ICS,(5) may benefit from additional interventions remains an important question for future investigation.

DCT cannot replace human interaction in healthcare. Further DCT investigation is necessary to more fully develop an understanding of how DCT can be woven into healthcare systems to find the optimal combination of digital and human touch that improves efficiency and health outcomes while maintaining patient satisfaction with their care. New investigations can increase digital touch and determine key points where telephone or in-office discussion with a clinician is most important. In many clinical settings, patients can already make online appointments and view test results. Further development of DCT may include exploration of expanded use of digital outreach to answer or ask questions of patients. Such computerized conversations can also gather additional important patient-reported information, such as symptom frequency or environmental exposures. These data may in turn be used to determine the degree and timing of interventions. DCT can also coach patients toward improved management of chronic illness and adoption of healthy lifestyles.(28) Beyond patient-provided, digitally gathered information, additional biosensor data may help establish risk profiles and direct additional care. For example, smartphone enabled technology, such as Global Positioning System (GPS) tracking of environmental exposures and mobile peak flow or fractional exhaled nitric oxide (FeNO) monitors,(29) may add additional data to allow for more refined predictions about patients at risk for an asthma exacerbation. Importantly, new interventions need to be tested within implementation research models to fully establish their feasibility and cost effectiveness in healthcare systems.(20)

Abbreviations:

DCT

Digital communication tool

BAO

Beta₂agonist overfill

EHR

Electronic health record

RCT

Randomized controlled trial

IVR

Interactive voice response

ACM

Asthma care manager

BA

Beta₂-agonist

ICS

Inhaled corticosteroid

AMR

Asthma medication ratio

FeNO

Fractional exhaled nitric oxide

GPS

Global positioning system