Adoption of a Portal for the Primary Care Management of Pediatric Asthma: A Mixed-Methods Implementation Study

Introduction

Patient portals, Web-based health care applications that enable patients to interact and communicate with their health care providers from outside the office [1], offer a resource to improve communication between patients and clinicians between visits. Patient portal use has increased recently [2]; however, adoption has not been rapid [3], and overall rates of sustained use remain low [4]. Recent research suggests that to effectively engage patients as portal users, several barriers may need to be overcome. For organizations, leadership challenges, marketing problems, and limited staff commitment have constrained portal adoption [5]. Studies have also found that patients who are white and have more health problems are more likely to use portals than others [4,6-8]. In addition, portals have not been as widely used in pediatrics as in the adult setting.

Pediatric asthma is an ideal condition for evaluating the feasibility of implementing portals to facilitate the management of chronic disease in practice. More than 7 million children in the United States have asthma [9], the most common pediatric chronic illness. Asthma is associated with lower quality of life [10,11], more missed days of school for children and work for parents [11-13], higher rates of hospitalization, emergency department visits [14], and death [15]. As appointment follow-up varies in pediatric primary care [16,17], and time constraints of office visits limit discussion, portals may facilitate decision making between families at home and primary care practices. However, the feasibility of using portals to collect patient-generated health information and portals’ impact on clinical care across diverse pediatric settings has not been established.

This study evaluated the determinants of implementation success for a portal in pediatric primary care to facilitate communication between families and clinicians regarding treatment concerns and goals, asthma symptoms, medication use, and side effects. In a subset of children with poorly controlled asthma, we further assessed the impact of portal use on asthma management, as clinical impact justifies implementation efforts. Finally, we qualitatively evaluated barriers to and facilitators of portal use experienced by families and primary care practices.

Methods

Setting

Twenty primary care practices were enrolled from 2 practice-based research networks: Pediatric Research in Office Settings (PROS) of the American Academy of Pediatrics and the Pediatric Research Consortium (PeRC) of the Children’s Hospital of Philadelphia (CHOP). PeRC is a hospital-owned primary care practice-based research network with 31 primary care practices and 231 clinicians in Pennsylvania and New Jersey [18]; PROS includes 728 practices and 1831 clinicians across the United States and Canada. A convenience sample of 9 PROS and 11 PeRC practices was enrolled.

Study Population

Eligible participants included English-speaking parents or guardians (subsequently “parents”) of children aged 6-12 years, who received treatment at a participating practice, had an asthma diagnosis at the time of recruitment, and had an office visit during the past 12 months.

Recruitment

To ensure that low-income children were represented, study practices were required to have ≥20% of children insured by Medicaid or the Children’s Health Insurance Program. Each practice was contacted by the investigators (AF, SF), invited to participate, and received an in-person or remote presentation of study procedures. After enrollment, rosters of all eligible children were generated from the electronic health record (EHR) to identify families for recruitment. Because of technical differences in the portals between PeRC and PROS, we tailored recruitment by setting. In PeRC, the study team mailed up to 2 letters to all eligible families, inviting them to call the study team to enroll. Parents provided verbal consent over the phone and enrolled in the MyAsthma portal. In PROS, families were mailed letters with a link to the portal website where families could enroll and consent. Telephone recruitment was used for a random sample of 50 families at each practice who did not respond to the letters. These phone calls were completed by the study team in PeRC and by the primary care practice clinicians/staff in PROS. Informational cards were available and posters were on display in participating offices. Enrolled parents received a $10 incentive for using the portal. PROS practices received $1000 for participating in the study, recognizing the additional work required for data extraction from independent practices.

The MyAsthma Portal

The MyAsthma portal was developed and tested at CHOP to facilitate shared decision making and improve asthma outcomes [19,20]. The portal was designed through a user-centered process including interviews and focus groups with 7 parents of children with asthma and 51 clinical team members, including doctors, nurse practitioners, and nurses. Functions of the portal were designed to reflect features families and clinicians prioritized, and iterative usability testing with parents and clinicians refined the portal system [19]. MyAsthma provides educational material; enables sharing of families’ treatment concerns, goals, asthma symptoms, medication adherence, and side effects with the primary care clinical team; tracks asthma control over time for families through the portal and clinicians through the EHR; and provides decision support to both families and clinicians regarding asthma control and side effects. On enrollment, families entered information about their treatment concerns and goals and completed an asthma control survey that assessed symptoms, medication adherence, and side effects. We used a version of the Asthma Control Test [21] that had been modified slightly to allow for parent proxy report of child symptoms. Subsequently, families were prompted by email each month to complete the asthma control survey. At CHOP, MyAsthma is embedded within an existing patient portal (MyChart, Epic, Verona, WI, USA; Figure 1). In PROS, families interacted with the portal through a Web interface (Figure 2), and decision support was provided on screen to families and via fax to practices based on asthma control survey results.

‎View this figure

‎View this figure

Outcomes

The primary outcomes included adoption (completion of at least one portal survey during the study period) and sustained use (completion of at least two surveys) of MyAsthma; outcomes were informed by an evidence-based conceptual model of factors influencing implementation success (Figure 3) [22]. We assessed additional outcomes (asthma office visit or asthma medication refill/change within 30 days of survey completion) in a subgroup of children who had uncontrolled asthma according to the results of their first asthma control survey. We focused on these actions because they are appropriate measures to take in response to poor asthma control. These data were extracted from each child’s EHR. In addition, parents and guardians reported whether they were more or less likely to (1) speak to their child’s doctor, (2) make a change to their child’s medication dosage, or (3) make a change to their home environment after using the portal using a 5-point Likert scale.

‎View this figure

Covariates

We extracted the following covariates from the EHR: patient age, sex, race and ethnicity, asthma severity (mild intermittent, mild persistent, moderate or severe persistent), insurance status (public vs private), and asthma controller medication use at study start (including inhaled steroids, montelukast, combination of inhaled steroid or long-acting β-agonists, and oral steroids). Parent-level covariates were collected via survey from enrolled participants and included age, race and ethnicity, educational attainment, employment status, and relationship to the child. Practice-level covariates included urbanicity (rural, suburban, or urban) and US census region (Northeast, South, Midwest, West).

Statistical Analysis

The study population was described using proportions, means, and standard deviations. Characteristics of children whose parents/guardians completed the portal survey were compared with those of children whose parents and/or guardians did not, using chi-square and t tests. Fisher exact tests were used for categorical data with sparse cell counts, and Mann–Whitney U tests were used for skewed continuous variables. Characteristics of children with sustained use were compared with those of children whose parents or guardians only completed the portal survey once. Multivariable logistic regression was used to model the association of patient characteristics and practice site with portal adoption to identify factors associated with adoption. The proportion of families who enrolled in the portal in response to a mailed letter versus a telephone call was also compared descriptively.

In the subgroup of patients with uncontrolled asthma, we described the proportion of children with an asthma office visit or medication refill or change within 30 days of survey completion. In a sensitivity analysis, we repeated these analyses with a period of 14 days. Furthermore, for each child, we compared these results to the same 1-month period a year earlier to assess whether rates of office visits and medication adjustments changed. We calculated 95% CIs around the change in proportions between years using logistic regression with the margins command in Stata (StataCorp, College Station, TX, USA). We also described parent responses to survey questions regarding the impact of portal use.

All analyses were completed using Stata, version 13.1. The Institutional Review Boards at the American Academy of Pediatrics (reference number: 13 FI 01) and CHOP (reference number: 13-010285) approved this study. All parents provided informed consent and child assent was waived as all information was collected from parents only and because children would not necessarily be readily available when parents were consented by telephone.

Qualitative Study

To evaluate implementation success and identify barriers to and facilitators of portal adoption (Figure 3) [22], trained research assistants on the study team used an interview guide based on our conceptual model to conduct 22 semistructured interviews by phone with parents, purposively sampled to include enrolled (14) and unenrolled (8) from both PROS (7) and PeRC (15), and 10 focus groups (PeRC in-person, PROS by phone, purposively sampled to include diverse representation from both networks) with 46 clinicians. All interviews were recorded then transcribed and coded using NVivo10 (QSR, Cambridge, MA, USA) and interpreted in the context of the conceptual model. Differences in coding were resolved by team consensus.

Results

Adoption and Sustained Use

Few invited families adopted the portal. Out of 9133 eligible patients, 237 (2.59%) completed the portal asthma control survey at least once (adoption). A total of 156 (65.8 % of portal adopters, 1.71% of eligible parents) completed the portal survey more than once (sustained use). Adoption varied widely across practices (0.6%-13.6%; Figure 4). Similarly, sustained use ranged from 0.0% to 13.6%. Reflecting a high level of quality of care, 93.42% of children at PeRC practices with persistent asthma were on a controller medication at baseline. Data on asthma severity were not available in PROS.

Portal users were more likely to have children aged 6-9 years (P=.009), to be white (P<.001), to be privately insured (P<.001), to have mild persistent or moderate or severe persistent asthma (P=.002), to be on an asthma controller medication (P<.001), and to be receiving a greater number of asthma medications at baseline on average than those who did not use the portal (P<.001; Table 1). In addition, those with persistent asthma were twice as likely to use the portal versus those with intermittent asthma (2.37% vs 1.25% at CHOP practices where these data were available, P<.001). Sustained portal users were more likely than one-time users to have children who were Hispanic (P=.02), have private insurance (P=.02), and be from the Northeast (Table 2, P=.001). Parents who had sustained use of the portal also had higher educational levels (P=.002).

In multivariable logistic regression, the following characteristics were positively associated with portal adoption: receipt of a controller medication at baseline (odds ratio, OR, 2.0, [95% CI 1.5, 2.7]), private insurance (2.0 [1.3, 3.1]), lower child age (1.4 [1.1, 1.9]), and greater asthma severity (1.9 [1.2, 3.0] mild and 1.9 [1.0, 3.5] for moderate or severe persistent versus intermittent; Table 3).

‎View this figure

Effect of Phone Versus Letter Recruitment 

Portal adopters reported how they learned about the portal. Letters to families resulted in the greatest number of enrolled families. Overall, 208 of 237 enrolled received a letter, 17 received a phone call, 35 heard about it from their child’s doctor, nurse practitioner, or nurse, and 3 from an informational card at the practice (25 reported multiple methods). Overall, 2.6% of children contacted by mail only enrolled, whereas 2.7% of those randomized to receive phone calls (they previously received letters) enrolled.

Effect of Portal Use on Asthma Management

Those with uncontrolled asthma commonly planned changes in management after portal use. After completing the first survey, 16% reported an intention to change their child’s asthma medication, 27% to contact their child’s doctor, and 20% to make a change to their child’s environment, with more than one-third (27 parents, 36%) reporting an intention to take at least one action. On follow-up surveys, 22% reported a medication change, 41% reported contacting their child’s doctor, and 16% reported making a change to their child’s environment (Table 4).

Health records confirmed that portal completion was associated with changes in asthma care. Of the 76 children with uncontrolled asthma after the first survey, 20 (26%) had a medication change or refill within 30 days of survey completion, and 21 (28%) had an asthma-related primary care visit within 30 days (Table 5). These numbers represent a significant increase in medication changes or refills and asthma-related visits when compared with the same period the year prior for each child (14% increase in medication changes [95% CI, 2%, 27%] and 16% increase in visits [95% CI, 3%, 28%]). Results were similar in a sensitivity analysis that examined the 14-day period after portal use.

Qualitative Results

Qualitative results revealed the importance of practice organization, family, and innovation characteristics to portal adoption (Table 6). Few health system factors were discussed and, when mentioned, clinicians disagreed about the value of incentives to promote portal adoption. Only 1 parent interviewed mentioned that an incentive (in this case from the research team) encouraged her to try the portal. For practices or clinicians, 3 primary themes emerged: the need for well-defined and coordinated workflows, the importance of practice responsiveness to portal surveys, and challenges related to identifying children with asthma through the EHR, which resulted in the recruitment of children without recent symptoms. In terms of workflow, clinicians and parents described that portal implementation was facilitated at practices that designated a specific person to coordinate the portal surveys and hampered when workflows were not well defined. Specifically, clinicians in 2 large urban practices reported being short staffed, lacking infrastructure in terms of care coordinators, and uncertainty about the ideal workflow for managing portal surveys. In addition, a perceived need among clinicians for more training diminished enthusiasm at some sites. Among those interviewed, parents of children with well-controlled asthma found MyAsthma less useful if they did enroll. Clinicians, especially in less affluent settings, perceived a lack of computer access as a barrier for parents. At the innovation level, features of MyAsthma that families and clinicians valued included facilitation of communication, increasing family awareness of and responsiveness to uncontrolled asthma, and ease of portal use.

Discussion

Principal Findings

We conducted a mixed-methods, multisite implementation study involving practices from 11 states to assess the feasibility for pediatric primary care practices of using a portal to facilitate communication between clinicians and families regarding asthma treatment, to assess the impact of portal use on asthma care for children with poorly controlled asthma, and to assess barriers and facilitators of portal adoption and sustained use. Overall, we found low rates of portal adoption and sustained use that varied from 0.6% to 13.6% across study practices. However, for those children with uncontrolled asthma, parent use of the portal was associated with a significant increase in asthma medication changes/refills and asthma visits to primary care practices. Qualitative methods underscored the importance of coordinated practice workflows, including practice responsiveness to portal surveys to implementation success. Parents, especially those with children with uncontrolled asthma, were motivated to continue using the portal because it facilitated a better understanding and tracking of asthma control.

Researchers and health systems in other settings have described low rates of portal adoption. For example, a study of the adoption of a portal for parents of children with cystic fibrosis, juvenile idiopathic arthritis, or diabetes reported that only 28% of invited families obtained a portal account, and only 48% of those (13% total) actually used the portal [23]. Even lower activation rates were observed in a study of a non–disease-specific portal in pediatric primary care, where rates of adoption have been lower than in adult health care settings [24]. Although prior studies found that patients with chronic diseases were more likely than others to register for or use a portal than others [25,26], the adoption rate in our population of children with asthma was quite low. Our qualitative results revealed that, at least in part, the low participation rate resulted from the inclusion of children that parents perceived had well-controlled asthma. These results are consistent with studies in diabetes that found that patients who believed their disease was well controlled felt that entering information over time was unnecessary [27] and were less likely to enroll [28].

Adoption may also have been limited by practices’ infrastructure and workflow for managing electronic receipt of patient-reported information. In our qualitative study, both clinicians and families highlighted the importance of coordinated and responsive workflows to implementation success. Workflow issues have been described previously as a challenge to portal implementation [5,27,29] and a reason for variability in adoption between practices [5,24]. In a case report of the portal adoption experience at 4 different adult primary care practices, practices with strong leadership and high staff engagement had higher rates of enrollment [5]. Learning collaboratives focused on workflow redesign in family medicine practices resulted in rates of portal use exceeding 25% of patients [26]. Especially relevant to the 32 billion dollar Federal Meaningful Use Program in the United States [30], these findings underscore the importance of integrating the portal into office systems and focusing provider and staff attention on their use.

The value to parents, practices, and the health system of implementing portals depends on their ability to improve communication and, ultimately, outcomes. Although adoption of the portal was low, portal use was associated with increased family and practice engagement in asthma management. These results are consistent with our prior pilot trial of MyAsthma, in which clinical outcomes including frequency of asthma flares and days of work missed by parents improved significantly among enrolled families [20]. Our finding that prescriptions and asthma visits increased among uncontrolled patients after using the portal is also consistent with studies among both children with autism and adults with diabetes that showed more active management after enrollment [31,32]. Mechanistically and as supported by our qualitative interviews, portal use may support disease management by improving patient–provider communication [33-36]. These findings support continuing effort to spur portal adoption and sustained use.

Collectively, the results of this study suggest multiple persistent barriers for the use of portals to support chronic disease management in pediatrics and that achieving high rates of adoption likely depends on the extent of existing practice infrastructure focused on disease management. We found that some families invited to participate did not consider their children to have active (or any) asthma. These results highlight the need to cautiously define the population to establish any metric for implementation success. In the case of asthma, researchers have developed definitions of an “asthma computable phenotype,” an algorithm based on data such as diagnosis, visits, medication, and laboratories, that accurately identifies patients with asthma [37,38]. Tailoring such definitions will be important for directing portals toward those most likely to benefit.

Limitations

This study had several limitations. First, although we enrolled practices from 11 states and practices varied greatly in adoption, slightly more than half of practices were from a single health system, potentially limiting the generalizability of results. In addition, although the asthma portal was implemented within primary care practices, it was implemented within the context of a research study. Findings may not reflect the results that would be observed if practices implemented a portal themselves. Third, this study had a relatively short follow-up period, limiting our ability to assess sustained use over a longer timeframe. Fourth, in our analysis of changes to medication and visits among children with uncontrolled asthma, we were unable to adjust for asthma severity due to limited sample size. Finally, we focused on a single chronic condition; however, asthma is a common chronic condition for which clinical trial evidence supports improved outcomes with portal use [20].

Conclusion

Despite the potential for real benefits to communication and child health outcomes, results of this multisite implementation study suggest that achieving high levels of portal adoption is unlikely in the short term. Many practices will require redesigned and coordinated workflows and will need to develop targeted outreach to families of children with poor asthma control to ultimately support the use of this technology.