Asthma is a common chronic respiratory disease contributing to a significant health care burden on both the individual and society [1,2]. It is defined as an inflammatory condition of the airways, characterized by episodic or persistent symptoms, such as dyspnea, wheezing, and cough, and is associated with variable airflow limitation and airway hyperresponsiveness [3]. Recommendations for best-practice asthma diagnosis and management are disseminated to primary care providers through the publication of clinical guidelines. In Canada, the Canadian Thoracic Society (CTS) evidence-based guidelines state that best-practice diagnosis is based on a compatible clinical history and objective evidence of asthma or a specialist diagnosis [3]. Effective management includes assessment of asthma control, patient self-management education, identifying triggers and environmental controls, and appropriate pharmacological therapy [3].

However, difficulties integrating clinical guidelines into the process of care have contributed to a number of key asthma care gaps in the primary care setting, causing a separation between best-practice and current clinical practice [3-7]. Several key asthma care gaps have been identified, primarily concerning objective diagnosis, continued monitoring with pulmonary function tests (PFTs), addressing asthma control, and delivering patient self-management education [5,8-11]. Furthermore, individuals with severe asthma (SA; 5%‐10% of patients with asthma) account for a disproportionate 50% of asthma-related health care costs [6]. Hence, in a recent position statement, the CTS highlighted a greater need for adequate recognition and management of individuals with SA and/or uncontrolled asthma, including specialist referral and consideration of biologic therapies when appropriate [6].

Consequently, the introduction of Knowledge Translation (KT) initiatives and tools to aid in adopting clinical guidelines in primary care practice is warranted [4]. Electronic medical records (EMRs) provide a unique opportunity at the point-of-care to integrate novel electronic tools (e-Tools), as they facilitate sentinel surveillance, performance benchmarking, and quality improvement [12]. Several validated and beneficial e-Tools have been researched. However, limitations exist with their use, including limited user uptake beyond the scope of research, the absence of standardized data elements, and limited involvement of end users throughout the design process [12-16].

An e-Tool designed for primary care EMRs that addresses the limitations of current e-Tools may have a significant impact on primary care asthma diagnosis and management and prompt adherence to best-practice guidelines. The Provider Asthma Assessment Form (PAAF) is a novel, point-of-care asthma management tool with an embedded SA algorithm and clinical decision support [17]. It was adapted from the Lung Health Foundation’s Asthma Care Map for primary care (paper tool), is congruent with current CTS guidelines, and incorporates Primary Care—Asthma Performance Indicators (PC-APIs) and Pan-Canadian Respiratory Standards Initiative for Electronic Health Records (PRESTINE) standardized asthma data elements [3,12,17,18]. It was developed by the Asthma Research Unit at Queen’s University in collaboration with primary care providers and has been integrated into the OSCAR primary care EMR at an academic Family Health Team (FHT) in Kingston, Ontario. The ultimate aim of this initiative is to address the major key asthma care gaps in primary care and promote adherence to best-practice guidelines through the introduction of a novel e-Tool to aid primary care providers in their practice.

We conducted a comprehensive evaluation of the integration of the PAAF into a primary care EMR to justify the widespread implementation of this form in primary care EMRs and other practice settings. Principally, we aimed to assess the ability of the PAAF’s implementation to address several key asthma care gaps in asthma diagnosis, management, and surveillance: underuse of objective measures of lung function to diagnose asthma [19-21], inconsistent documentation of asthma diagnosis in primary care EMRs [22], and underrecognition and suboptimal management of SA and/or uncontrolled asthma [6]. The primary objective was to evaluate the impact of the PAAF’s implementation on the use of objective lung function measures to confirm asthma diagnoses, documentation in the EMR of asthma diagnosis status (suspected, confirmed, or excluded), and referral of individuals with suspected or confirmed SA to a specialist.

The study received ethics approval from the Queen’s University Health Sciences & Affiliated Teaching Hospitals’ Research Ethics Board (HSREB #6036789). This study involves using existing data from the Canadian Primary Care Sentinel Surveillance Network (CPCSSN) project (HSREB# 19809). CPCSSN has a data-sharing agreement with the center where this study was conducted. Passive consent was used, with all patients having an opportunity to explicitly opt out of the study. All data were deidentified.

A pre- and postobservational study was conducted at an FHT in Kingston, Ontario. The study design consisted of 2 cohorts of patients: a retrospective baseline cohort, consisting of patients who had an asthma visit before implementation of the PAAF, and a postimplementation cohort, consisting of patients with a visit for asthma after integration of the PAAF into the OSCAR EMR at an FHT. An asthma visit was defined as an FHT encounter where asthma was suspected, diagnosed, managed, or treated either at an in-person or telephone appointment. Case identification was accomplished by applying the newly validated adult asthma case definition for primary care EMRs to CPCSSN-deidentified EMR data for the FHT [22]. This case definition identified adult patients (≥18 years of age) with either suspected or confirmed asthma and a documented encounter during the defined study period. This validated case definition has a sensitivity of 81% and specificity of 96% and is based on International Classification of Diseases, Ninth Revision (ICD-9) codes for asthma billing [22]. A single abstractor completed detailed manual chart abstraction on identified cases. A programmed data extraction from the FHT EMR was subsequently performed for all PAAFs that were used within the study period.

The retrospective baseline cohort included all asthma visits in a 2-year period between January 2018 and December 2019 prior to the COVID-19 pandemic to ensure that baseline data were representative of usual care. The postimplementation period was set as January 2023 to December 2023 (1 calendar year) for manual chart abstractions. Due to barriers faced with implementation (technological difficulties with the form’s functionality in the EMR from May to July 2023) and challenges with user uptake, the period for completed PAAF data extraction was delayed until October 2022 to July 2024. The sample derivation methodology is summarized in Figure 1.

The PAAF is a web-based electronic form with direct access from the patient dashboard in the OSCAR EMR at the FHT. Patient demographics such as date of birth and chart number are automatically populated when the electronic form is opened in a patient chart. It contains 12 sections pertaining to the diagnosis, family history, smoking history, asthma severity, occupational history, respiratory medications, asthma control, care, management and referrals, asthma action plan, asthma control zone, pulmonary function tests, and assessment tools. Select sections include interactive features, such as a calculation of smoking pack-years and a determination of “yes” or “no” if the patient’s asthma is controlled, based on the CTS control criteria. The SA algorithm feature consists of a decision-support prompt appearing at the top of the form once relevant sections have been completed. Decision support prompts include suggesting confirmation of an asthma diagnosis through objective measures, tapering inhaled corticosteroid (ICS) medication if the patient’s asthma is in control, addressing reasons for poorly controlled asthma, and considering a referral to a specialist for SA and/or uncontrolled asthma.

Various methods of implementation were used throughout the PAAF study period from October 2022 to July 2024. Initial implementation was conducted by a site champion, a first-year family medicine resident at the FHT. The PAAF was initially introduced by the site champion to attending and resident physicians through a presentation and via email. Subsequent implementation efforts throughout the course of the study period included a reminder poster in the FHT clinic team rooms about the PAAF and how to access it in OSCAR.

A summary email and an attached 1-page handout detailing the purpose of the PAAF and how to use it were also sent to all FHT providers via email twice throughout the study period. Additionally, the study team collaborated with a nursing student community-based project group. The nursing students were given a summary presentation about asthma, the key asthma care gaps, and how the PAAF can be used in clinical practice. They participated in a tutorial and live demonstration of the PAAF within the OSCAR EMR. The students then used the form during telephone appointments of patients with asthma with supervision from an instructor.

Finally, implementation strategies for the PAAF included educational presentations at Family Medicine Grand Rounds. In both July 2023 and July 2024, the study team gave presentations to incoming first-year family medicine residents during their orientation period. Additional presentations were given throughout the year to upper-year residents on rotation at the FHT.

A total of 373 patients were deemed eligible in the final analysis of this study. The sample derivation is displayed in Figure 2. In the retrospective baseline patient cohort, 230 patients were included. In the postimplementation patient cohort, a total of 143 patients were included, of whom the PAAF was used in 12. A total of 27 patients were excluded from manual chart abstraction (retrospective baseline: n=19 and postimplementation: n=8), as they did not meet eligibility for having an asthma visit within the study period. A total of 6 patients (6 PAAFs) were excluded, as they were identified as either a pediatric patient or a test patient chart (ie, a chart generated for training and programming purposes). There were 7 new patients from the PAAF data extraction, and 5 patients overlapped with the postimplementation chart abstraction.

Table 1 displays the patient characteristics of included patients from the retrospective baseline and postimplementation. No significant differences were found between the 2 groups, except for a significantly higher percentage of patients diagnosed with diabetes in the postimplementation cohort (retrospective baseline: n=14, 6.1% vs postimplementation: n=18, 12.6%; P=.04). Both the mean age (retrospective baseline: mean 46.9, SD 17.4; postimplementation: mean 49.7, SD 17.5; P=.14) and BMI (kg/m²) were similar between groups (retrospective baseline: mean 30.0, SD 8.4; postimplementation: mean 31.3, SD 7.7; P=.14). There was a high percentage of female patients compared to male patients (sex assigned at birth) in both cohorts (retrospective baseline: n=135, 58.7% female vs postimplementation: n=90, 62.9% female; P=.45). There was an observed high percentage of comorbidities with both groups having over 45% of patients with ≥4 comorbidities (retrospective baseline: n=107, 46.5% vs postimplementation: n=65, 45.5%; P=.60). The most common comorbidity among both groups was allergies (retrospective baseline: n=187, 81.3% vs postimplementation: n=113, 79%; P=.59).

In total, only 31.3% (n=72) of patients in the retrospective baseline and 23.8% (n=34) in the postimplementation cohorts were classified as having confirmed asthma (Table 2). The most common method of diagnosis in both groups was evidence of reversible airflow obstruction on spirometry (retrospective baseline: n=36, 50% vs postimplementation: n=21, 61.8%). The second most common method was a positive or borderline methacholine challenge test (retrospective baseline: n=29, 40.3% vs postimplementation: n=6, 17.6%). A higher proportion of confirmed patients with asthma in the postimplementation cohort had their asthma diagnosed by a specialist (respirologist, allergist, or internal medicine) without documented evidence of objective measures (postimplementation: n=4, 11.8% vs retrospective baseline: n=7, 9.7%). A visual representation of suspected versus confirmed asthma can be appreciated in Multimedia Appendix 1.

Table 3 displays results pertaining to monitoring with PFTs. The most noteworthy finding was that PFTs were requested by physicians in a significantly higher proportion of patients in the postimplementation period (postimplementation: n=70, 49% vs retrospective baseline: n=71, 30.9%; P<.001). A nonsignificant higher proportion of patients in the retrospective baseline cohort had been monitored with PFTs within 12 months before the documented encounter (retrospective baseline: n=67, 29.1% vs postimplementation: n=28, 19.6%; P=.05). However, a nonsignificantly higher proportion of patients had been monitored with PFTs either 12 months before or after the documented encounter (postimplementation: n=50, 35% vs retrospective baseline: n=71, 30.9%; P=.43). A visual representation of PFTs completed versus PFTs requested by providers can be appreciated in Multimedia Appendix 2.

Table 4 summarizes the asthma medications documented in the charts of patients at the time of the asthma visit included in the study. Two primary significant differences were found. In the postimplementation group, 21.7% (n=31) of patients were on single inhaler reliever and controller therapy (single inhaler therapy [SIT]) compared to 0.9% (n=2) on SIT in the retrospective baseline cohort (P<.001). Similarly, a significantly greater percentage of patients (n=36, 25.2%) in the postimplementation groups were on ICS/long-acting β-2 agonist (LABA) controller therapy compared to the retrospective baseline (n=34, 14.8%; P=.01). No other significant differences were found between groups. Nonsignificant trends include a 5.6% decrease in systemic steroid use and a 5% decrease in adherence issues after implementation. A summary of reliever, ICS, second controller, systemic steroid use, and SIT can be found in Multimedia Appendix 3.

At least 1 CTS control parameter was documented in a high percentage of patients in both groups (retrospective baseline: n=226, 98.3% vs postimplementation: n=138, 96.5%; P=.31; Table 5). A significantly higher percentage of postimplementation patients were on an ICS if their asthma was uncontrolled (postimplementation: n=69, 62.2% vs retrospective baseline: n=100, 43.5%; P=.002). Furthermore, a significantly higher number of CTS asthma control parameters were documented after implementation (postimplementation: mean 2.5, SD 1.3 vs retrospective baseline: mean 2.2, SD 1.2; P=.02). A summary of asthma control documentation is represented in Multimedia Appendix 4.

Table 6 summarizes the health services use and management between cohorts. A significantly higher proportion of barriers were addressed during the documented asthma visit in the postimplementation group compared to the baseline (postimplementation: n=24, 16.8% vs retrospective baseline: n=11, 4.8%; P<.001). No other significant differences were found. Although not statistically significant, the proportion of recent emergency department visits in <1 year decreased by 4% in the postimplementation group (postimplementation: n=16, 11.2% vs retrospective baseline: n=35, 15.2%; P=.21). An Asthma Action Plan (AAP) was provided or revised or reviewed more often (4.1% more) in the postimplementation period (P=.15). Several PC-APIs from both cohorts are represented in Table 7.

Overall, 12 PAAFs were used during asthma encounters with 11 patients total in the postimplementation period. When the PAAF was used, diagnosis status was documented 100% (n=12) of the time. When the PAAF was not used, a detailed manual chart review was needed to determine the diagnosis status.

Asthma control documentation was compared between PAAFs used and manual chart abstraction from EMR encounter notes (Table 8). In terms of asthma control, the average number of CTS asthma control parameters documented was significantly higher when the PAAF was used (PAAF: mean 5.4, SD 1.9 vs manual chart abstraction: mean 2.3, SD 1.2; P<.001). Asthma control was documented significantly more often in manual chart abstractions (PAAF: n=7, 58.3% vs manual chart abstraction: n=360, 98.4%; P<.001). Finally, there was a significantly higher proportion of patients whose asthma was uncontrolled at the time of the visit in manual chart abstractions (manual chart abstraction: n=302, 82.5% vs PAAF: n=6, 50%; P=.01).

Following a multifaceted intervention of implementing the PAAF in this primary care practice, there were substantial improvements in asthma-specific documentation and adherence to key evidence-based recommendations for care. There was an 18.1% increase in the number of PFTs requested after the implementation of the PAAF. This did not translate to the number of confirmed asthma cases, presumably due to the impacts of the COVID-19 pandemic on PFT wait times. Furthermore, a highly significant and clinically relevant increase was noted in the proportion of patients prescribed SIT after implementation of the PAAF (postimplementation: n=31, 21.7% vs baseline: n=2, 0.9%), and a significantly greater proportion of patients were on ICS/LABA controller therapy after implementation. In general, a higher percentage of patients were on reliever, ICS, and second controller therapy after implementation, and a smaller proportion had received systemic corticosteroids in the last year. Asthma control was highly documented in both cohorts, but significantly more CTS asthma control parameters were documented after implementation, and a higher proportion of patients with uncontrolled asthma were prescribed ICS therapy. A greater percentage of patients had barriers addressed at the asthma encounter after implementation compared to baseline (postimplementation: n=24, 16.8% vs baseline: n=11, 4.8%). There were no meaningful changes in specialist referrals due to suspected SA. Although it is not possible to discern causality for observed changes based on this study, overall, care as assessed by key PC-APIs showed improvement in the postimplementation cohort.

The main limitation of this study was the limited uptake of the PAAF in the primary care practice. Therefore, the sample size was limited, potentially impacting the power of our study. As can be seen from the CPCSSN case definition search, there were 136 visits that would have been eligible for the PAAF to be used in the postimplementation time frame. However, it was only used 12 times throughout the study duration. The perceived utility, provider satisfaction, and barriers and enablers to PAAF use are the focus of another study.

Furthermore, it is important to note that the results of this study can only be interpreted in the context of the broader multifaceted intervention encompassing not only the PAAF tool but also the various implementation strategies used as part of the training and orientation to the tool. Further studies are needed to differentiate the specific impact of the tool itself, rather than the global strategies used as part of the implementation of the PAAF in a primary care setting.

It is important to note that it is difficult to discern causality for the changes seen in asthma management practice in this study. As this study was conducted at an academic primary care practice, the resident physicians involved with patient care receive additional instruction and education on effective asthma management during their academic instructional time. As well, there was a concurrent quality improvement project focusing on the Choosing Wisely Canada climate change and inhaler recommendations for patients with asthma [37]. This quality improvement initiative aimed to decrease the number of patients on pressurized metered dose inhalers and increase the number of patients on dry-powder inhalers, which may have influenced the proportion of patients on SIT (which is commonly a dry-powder inhaler). Therefore, it is important to recognize that this initiative may have increased the focus on asthma management in the primary care practice, beyond the implementation efforts for the PAAF.

Additionally, the percentage of postimplementation patients with a PFT in the last 12 months was likely impacted due to a persistent backlog in the Pulmonary Function Lab at Kingston Health Sciences Centre (1 of 2 centers for PFT testing in Kingston) due to the COVID-19 pandemic. Hence, this study may not reflect a complete picture of PFT monitoring in baseline compared to postimplementation. In an effort to address this limitation, the number of requested PFTs within 12 months of the documented encounter was included in the analysis.

It is pertinent to recognize the inherent limitations of manual chart abstraction as well. Data elements collected are limited by the information providers document within the chart and encounter notes; thus, the documentation may not fully capture all aspects of care discussed during an asthma visit.

Following the multifaceted intervention of implementing the PAAF in this primary care practice, there were significant improvements in asthma-specific documentation and adherence to key evidence-based recommendations for care. The use of the PAAF increased asthma visit–specific documentation, such as clearly documenting diagnosis status and asthma control parameters. However, asthma care gaps such as the underuse of PFTs for diagnosis and monitoring, asthma education, and addressing reasons for poor asthma control were still common in this study. Future directions should involve evaluating the impact of the PAAF after widespread implementation in primary care settings and investigating methods to increase user uptake of the PAAF.