A Primary Care-Based Quality Improvement Project to Reduce Asthma Emergency Department Visits Free

NCH made improvement in asthma care, a key element of its strategic plan in 2010. Our initiative was the primary care component of a hospital-wide QI program to reduce asthma healthcare utilization.¹⁵  In January 2010, at project onset, there were 21.7 asthma-related ED visits per 1000 network patients aged 2 to 18 years per year. Our overall QI goal was to reduce asthma ED visits by network patients ages 2 to 18 years by 3% each year. We chose a 3% annual reduction as our goal because we believed this was an attainable objective that, when sustained over time, would have a meaningful impact on ED utilization rates. We display our outcome data only through December 2019. We do not include data past that date because the coronavirus disease 19 pandemic that began in March 2020 is associated with a sharp decrease in asthma ED visits.¹⁶ 

The network consists of primary care clinics in predominantly urban, lower socioeconomic and medically underserved communities in Columbus, Ohio. In 2010, nearly 100 providers cared for over 88 000 unique pediatric patients at the 11 sites. A demographic description of the total network population and those children with asthma is found in Table 1. All racial and ethnic identities were self-reported. In 2019, the network cared for over 8500 unique patients with asthma, 10% of the total network population. Almost 90% of patients were Medicaid-enrolled or uninsured, and 20% of patients and families had limited English proficiency. Male patients and Black patients were over-represented among asthma patients as shown in Table 1, corresponding to national data on increased prevalence of asthma among these groups.¹  However, we did not focus on improvement in any specific group because large segments of our population did not fall into either category.

We used the Model for Improvement to implement QI process change interventions.¹⁷  We developed the principle that all encounters with asthma patients should address asthma control, regardless of the chief complaint. We focused on 5 core QI interventions: (1) increase the use of asthma action plans, (2) increase the number of visits in our primary care-based asthma specialty clinics (ASCs),¹⁸  (3) increase the use of a standard asthma note at any visit for a patient with asthma, (4) increase the documentation of the ACT, and (5) increase the use of step-up therapy for poorly controlled asthma patients. Figure 1 represents the key driver diagram that shows these interventions. Metrics were collected using available reporting sources within our institution.

The size and complexity of the network posed a challenge for QI implementation. The issues of communication of practice change and subsequent buy-in by clinicians and staff were significant. Therefore, we introduced interventions sequentially to avoid overwhelming clinicians and staff with practice changes, give them time to adapt to each new intervention, and achieve change acceptance to prioritize sustainability.

Improving asthma care was added to the hospital’s strategic plan in 2010. Substantial resources were allocated for this project. We created a primary care asthma QI team that included a primary care pediatrician, a pulmonologist, an asthma nurse, and a QI data specialist. Initially in 2010, the primary care pediatrician was designated 0.1 full time equivalent (FTE) salary support for this project. Because of the size of the network and the scope of work, this role was increased to 0.6 FTE in 2013 divided among 3 primary care pediatricians. The QI specialist was designated 0.1 FTE. This leadership expansion was instrumental to increasing intervention development, improving efficiency, and enhancing communication across the network.

We implemented catered quarterly educational lunches led by the 3 primary care asthma QI physicians in 2013 to provide direct communication between asthma QI leadership and the entire clinic team. Clinicians and staff received updates regarding existing and new processes. Attendees were able to share their experiences and concerns and we received feedback from the clinics to gain buy-in and learn about successes and barriers.

Clinician practice change was supported by EMR strategies, including best practice alerts and guideline-based documentation and order templates. The team created EMR support in a manner that allowed later data query and performance feedback.

Table 2 shows each asthma QI intervention, dates of introduction and shift to maintenance phase, definition of measure, baseline data, and goal. These interventions were supported by techniques including a division-wide education session on asthma at the beginning of the project, educational lunches, occasional use of monetary incentives for providers and nursing staff, and quarterly clinic-level and individual clinician feedback. Monetary incentives encouraged adherence to interventions, varied throughout the timeline, and ranged from $25 to $100 per quarter for nurses and $500 to $1000 per year for providers.

Asthma action plans are effective components of asthma management. In 2011, we used an NCH institution-wide EMR-embedded asthma action plan initially developed to comply with the 2003 Joint Commission Children’s Asthma Care metric (CAC-3).¹⁹  The asthma action plan³  was available in English, Spanish, and Somali languages and was revised over time to address health literacy. Our goal was 80% completion.

The project team established ASCs within the network in 2011 and initially tested them at 3 locations. These clinics were staffed by a primary care pediatrician within the child’s main primary care site but were specifically designated for the care of children with more complex or symptomatic asthma. The ASCs allowed extended-length visits for higher-risk asthma patients and provided comprehensive asthma encounters that included multidisciplinary education by a respiratory therapist and/or an asthma nurse, same-day spirometry, and, in some clinics, tablet-based educational modules. Early evaluation showed a positive impact on ED visit rates for patients seen in the ASCs.¹⁸  Consequently, we added ASC sessions at other network clinics over time. We established an initial goal to increase the number of monthly ASC visits across the network to 120 in January 2013 and increased this to 250 per month in July 2014, giving more patients access to enhanced asthma care. The project team tracked completed ASC visits in the network monthly.

We developed a standardized evidence-based EMR asthma note in 2012 to enable efficient documentation of asthma assessment during clinic encounters. Asthma note patient history was based on guidelines from the NHLBI Expert Panel Report 3: Guidelines for the Diagnosis and Management of Asthma.³  In addition, the note provided a standardized way to classify asthma severity, assess control, and make medication management decisions. Feedback relating to note content and design was obtained from network clinicians at several stages of development and incorporated into the note. Over time, the note evolved from primarily prompting documentation to also promoting guideline-based care via management choices built into the note. When we introduced the ACT to our clinics, we added the ACT score to the asthma note.

A stand-alone comprehensive asthma note was created for routine asthma encounters in which asthma was the chief complaint. The note template included the ACT score, impairment questions, risk questions, asthma severity classification, level of control, and plan components. A less detailed note, referred to as an “asthma short note,” had the same components as the regular note without full visit details (eg, omitting review of systems and physical exam). It was incorporated when asthma was not the chief complaint. For patients meeting asthma diagnosis criteria, an asthma short note was added by clinicians to a well child visit template. Consistent use of standardized asthma notes allowed the QI team to begin assessing management decisions by clinicians. Our goal was to use the asthma note at 65% of all clinic encounters.

We introduced the routine use of the ACT²⁰  in 2013 when a patient with asthma who was ≥4 years old was seen in the network for any reason. An EMR Best Practice Alert (BPA) flagged appropriately aged asthma patients, prompting nurses to provide an ACT during the rooming process at every encounter. Nurses were required by the EMR to indicate that an ACT was “given,” “not given,” or “postponed” before moving to the next EMR task. If “postponed” was selected, the BPA would continue to flag every time a nurse opened the chart during that encounter. Nurse compliance with this BPA was tracked. Clinicians also received a separate BPA that identified patients with an asthma diagnosis on their current problem list or asthma medication on their current medications list. The ACT was initially documented separately within the clinician’s documentation; later, the ACT was embedded within the standardized asthma note. As the ACT was given to patients at all encounters regardless of reason for visit, clinicians were expected to document the ACT score even if asthma did not require further discussion. Our goal was a documentation rate of 70%.

In 2015, the team introduced an intervention to improve management of patients with poorly controlled asthma. The concept of stepping up therapy for poorly controlled asthma is embedded in the NHLBI guidelines.³  We measured the percentage of patients assessed by their clinician as having poorly controlled asthma in the standard asthma note who also had a step-up in pharmacotherapy documented. “Step-up” was defined as adding an inhaled corticosteroid (ICS), adding a leukotriene inhibitor, increasing the strength of an ICS, switching from an ICS to an ICS and long-acting β-agonist combination, or increasing the strength of an ICS-long-acting β-agonist. An example includes patients who had suboptimal adherence to their previously prescribed regimen. We provided options for clinicians within the asthma note to indicate other plan elements, including counseling to improve medication adherence, addressing barriers, continuing current therapy, referral to a school-based asthma therapy program,²¹  referral to a home-based asthma education program, referral to pulmonary or allergy specialty clinics, and resuming current therapy that had been discontinued by the patient’s caregiver. We set a goal of 50% to allow for encounters in which a step-up in pharmacotherapy was not indicated.

The outcome measure for this QI work was the change in asthma ED utilization per 1000 unique network patients ages 2 to 18 years. We calculated the ED average asthma visit rate from years 2008 through 2010 and used this as our baseline rate. This rate was measured by calculating the number of asthma-related ED visits in a given year per 1000 network patients and tracked monthly in a seasonally adjusted²²  statistical process control (SPC) chart. For this measurement, eligible patients included anyone 2 to 18 years of age with a visit to a network clinic in the prior 13 months. Patients were classified as having asthma if there were any diagnostic claims for asthma in the prior 24 months at any NCH clinical encounter.

We monitored for 1 potential negative impact, specifically an increase in clinic encounter time because of additional workflow related to asthma assessment and treatment.

The NCH Institutional Review Board deemed this project to be QI and not research and therefore exempt from review.

We used SPC charts to visually display both process and outcome data. Upper and lower control limits were calculated after establishing a baseline, and special cause variation was determined using standard SPC rules.²²  Significant process measure and outcome changes related to intervention implementation or other external influence are represented by a centerline shift, which was computed using standard SPC rules.²²  Our data were adjusted for noted seasonal variation in asthma ED visits. Monthly effects were observed, and data points were standardized by calculating the average deviation from the centerline for all months. These adjusted points were then plotted.²² 

By the end of 2019, asthma-related ED visits fell to 14.5 per 1000 patients per year from a baseline of 21.7 asthma-related ED visits per 1000 patients per year in 2010. This represented a 33% decrease in ED utilization for asthma exacerbations. This significant difference is represented by 2 center line shifts in December 2014 and in October 2018 (Fig 2).

Process measure outcomes are displayed in Table 2 and Fig 3. We achieved goals in 4 out of 5 metrics. For example, we achieved the goal of writing an asthma note in 65% of potential cases by mid-2014 and subsequently exceeded this goal. We intermittently achieved the goal of 250 patient visits in our ASCs per month but did not sustain this over time. For our balancing measure, the turnaround time for all primary care appointments decreased from a monthly average of 71 minutes in January 2010 to 63 minutes in December 2019.

Between 2010 and 2019, we achieved a significant 33% reduction in asthma-related ED visits in a large, predominantly Medicaid-eligible patient population that was racially, ethnically, and linguistically diverse. Before this initiative, a standardized method to evaluate patients with asthma was lacking, and asthma was typically addressed only when the patient or caregiver prompted the discussion as a chief complaint. The evidence-based interventions we introduced led to a more routine, comprehensive evaluation of our population’s asthma control. A stepwise, longitudinal QI approach was necessary to implement the most successful interventions and allowed us to progressively incorporate them into sustainable practice. Our consistent and routine outreach to all clinicians and staff reinforced these processes through academic detailing efforts encouraging bidirectional communication about our interventions. Despite the introduction of many new processes in asthma care over time, there was no evidence that turnaround time in the clinical setting increased.

From the onset, our efforts were made possible with strong institutional support that was quickly augmented with allocation of greater resources to increase our team to 3 primary care asthma QI physician leaders. This expansion of leadership was necessary for intervention development, implementation, and communication. Collective use of QI methodology helped us develop, implement, and assess these measurable interventions.

The clinically meaningful improvement demonstrated over 9 years was achieved in a network of nearly 100 clinicians across 13 clinics with an asthma population exceeding 86 000 children, in predominantly medically underserved urban areas. We have shown improvement in a racially and ethnically diverse population. Our strategy, which included the implementation of multiple, serial interventions as part of a longitudinal QI initiative, resulted in subsequent marked compliance (Fig 3) with practice improvement associated with decreased ED utilization for asthma. This data suggests that our strategy of introducing the interventions sequentially was successful, potentially because clinicians were not overwhelmed by multiple clinical care improvement changes simultaneously. Over time, we achieved compliance with the “bundle” of ideal interventions for asthma, but we allowed time for this to develop. These results (Fig 3) also suggest that interventions that require nurse and clinician process change are more achievable than those that rely on caregivers making and keeping appointments, as we found that the only intervention in which we did not achieve our goal was patient attendance at ASCs.

Practice change sustainability is a key element in QI. There are several potential reasons to explain our lasting practice change. Our team met weekly throughout this entire time period and received bimonthly data reports, so if there were problems with metrics, the team became aware quickly and could take action. These meetings were essential in project development, timely response to challenges, and sustainability of improvements. The goals were feasible and based on the QI teams’ experience with primary care patient flow. The 3 asthma QI physician leaders in the Division of Primary Care were all clinicians familiar with clinic processes, flow, and culture. For example, use of the standardized asthma note for any visit was set at an initial goal of 60% and then increased to 65%. The QI leaders recognized that there would be many patient encounters during which use of an asthma note would not be feasible or necessary, such as if a mental health issue was the chief complaint or if asthma was assessed recently and the ACT remained high. Finally, decision support within the EMR “asthma note” that we developed streamlined the process for clinicians to record the ACT or decide to “step-up therapy.”

The increase in the network’s ED visit rate in 2014 was because of special cause variation correlated with the enterovirus D-68 outbreak throughout much of the country.²³  There was no reduction in adherence to interventions at that time. Of note, the rate during 2014 was similar to that of our baseline, showing that previously established interventions still may have positively impacted ED rates during the outbreak.

There are several limitations to this study. First, the absolute impact of each intervention cannot be determined. Although interventions were introduced in a stepwise approach, achievement of consistent adherence by clinicians to multiple interventions may have occurred simultaneously. Second, we only collected data from NCH ED encounters. If patients received medical treatment at unaffiliated local EDs, those were not accounted for in this project, which may lead to an inaccurate interpretation of our outcomes. However, there is no indication that the proportion of patients who used an ED outside of NCH would have changed over the course of this initiative. In addition, we acknowledge that we had generous and continuous institutional financial support for this work, providing salary support for the physician leaders, QI data support, and funds used for educational lunches and financial incentives. We believe that this was instrumental to our success and limits generalizability to pediatric settings that cannot provide such funding. Finally, our population was over 90% Medicaid-enrolled. Generalizability to other populations may need further investigation. However, since this was a population at high risk for asthma exacerbations, our positive and sustained results are notable and similar processes could be applied to other large urban pediatric medical centers.

Evidence-based and innovative interventions implemented using QI science can reduce asthma ED utilization in a large, high-risk pediatric population. This project’s initiatives, implemented and used over time, suggest that robust and sustainable outcomes can be achieved in a large network of primary care clinicians.

The authors would like to thank Dr Richard Brilli for his help in preparing this manuscript.

ACT

asthma control test

ASC

asthma specialty clinic

ED

emergency department

NCH

Nationwide Children’s Hospital

NHLBI

National Heart, Lung, and Brain Institute

QI

quality improvement