Chronic disease surveillance networks

Chronic disease surveillance networks are a category of asthma KT eTool that exist at international, national, and regional levels [34]. An example of a chronic disease network is the Canadian Primary Care Sentinel Surveillance Network (CPCSSN) [35]. CPCSSN gathers information from physician billing codes, emergency department visits, and administrative data to facilitate standardized estimates of the incidence, prevalence, and outcomes related to various chronic diseases and are used by governments and researchers. Although CPCSSN collects data on a variety of chronic diseases, it does not currently collect data on adult asthma [35]. An example of a chronic disease surveillance network for asthma is the Ontario Asthma Surveillance Information System (OASIS) [36]. OASIS uses a previously validated asthma case definition derived from hospital administrative data in Ontario and provides a population-based longitudinal surveillance system for asthma. OASIS data is used to provide estimates of asthma incidence and prevalence, measures of asthma-related morbidity, mortality, health services use, and provider practice patterns using hospital administrative data to track quality of care over time [36]. Chronic disease surveillance networks are often limited by their source of data and restricted criteria for defining a case of the condition [37]. While limited criteria from health administrative data such as billing codes are sufficient for chronic diseases such as diabetes, for more complex and heterogenous conditions such as asthma, health administrative data often do not accurately reflect when and how diagnoses were made [38]. These chronic disease surveillance networks that are based on health administrative data have limitations on the amount and quality of data they can leverage to improve asthma diagnosis and surveillance.

Asthma registries

Registries have been effective tools to collect uniform data to evaluate specific outcomes for a population defined by a particular disease. Registries differ from surveillance networks in that registries use data that is voluntarily provided and entered while surveillance networks make use of pre-existing data. The scope of asthma registries has been limited to severe asthma, which represents a minority of total asthma cases in Canada and worldwide [39, 40]. Severe asthma registries gather anonymous, longitudinal, real-life data for patients with severe asthma. To date, there are over 25 severe asthma registries, the majority of which operate at a national level and contribute information to the International Severe Asthma Registry (ISAR), the Severe Heterogenous Asthma Research collaboration, Patient-centered (SHARP), or the Severe Asthma Research Program (SARP) [41–43].The availability of health information varies greatly between databases [41]. In addition to ISAR, SHARP, and SARP there are several local and regional initiatives aimed at creating registries for cases other than severe asthma however, to date there remains no asthma registry for general asthma patients, only severe asthma patients [44]. The result of this is a lack of centralized information worldwide for general asthma patients, which creates a significant challenge to monitor the state of asthma at a national or international level.

Asthma quality of care monitoring systems

Quality of care monitoring systems for asthma are emerging as useful eTools for KT. Audit and feedback systems for quality improvement of chronic disease can be effective in creating behaviour changes in health care providers when successfully implemented and can monitor quality of care over time [45]. There are currently several initiatives centered around creating quality of care monitoring systems for asthma in different forms. The Asthma Care Map (ACM) developed by the Lung Health Foundation (formerly the Ontario Lung Association) and used in the Primary Care Asthma Program (PCAP) is a paper KT tool for asthma quality of care [46] that is currently being adapted to integrate within primary care EMRs. Advancements in computing and data quality within EMRs have given rise to the potential for electronic systems to aid in the surveillance of chronic diseases, with the end goal of improving patient outcomes [47]. One example of a management and monitoring system that can be used for quality improvement is the Asthma Management and Outcomes Monitoring System (AMOMS) [48]. AMOMS is a point-of-care charting tool the prompts providers to document care in accordance with best practice guidelines. In doing so, AMOMS collects data from the patient and physician that can be extracted and used to support performance measurement, benchmarking, and quality improvement. Another quality of care monitoring system for asthma developed by Tomasallo et al. (49) found their EMR-based asthma surveillance system can be used to estimate the prevalence of asthma in both adults and children across time and was able to identify patients at risk of asthma in 50% more cases than traditional telephone surveys [49]. Despite the potential for quality of care monitoring systems to improve adherence to best-practice guidelines and patient outcomes, this literature review did not identify any quality of care monitoring systems for asthma that have achieved scale at a national or international level.

Work-related asthma surveillance systems

There are several surveillance systems dedicated to work-related asthma (WRA). WRA surveillance programs have been active for over 20 years in jurisdictions ranging from local to national in scale [50, 51]. Similar to chronic disease surveillance networks, there are also occupational disease surveillance systems that record the incidence, prevalence, and outcomes of WRA [52]. WRA surveillance systems have been effective in supporting individuals with WRA; however, their scale is often limited by the source of the data and represents only a fraction of total asthma cases. In addition, previous efforts to develop workplace asthma surveillance systems have relied on practitioner reporting and have reported low uptake rates.[53] Due to their small scale and limited sources of data, WRA surveillance systems are unlikely to scale to the national or international level.

Lack of confirmation of asthma diagnosis

KT eTools for surveillance rely first on ensuring patients have an accurate diagnosis of their condition by a health care provider. The gold-standard definition of asthma, as outlined by the Canadian Thoracic Society and Global Initiative for Asthma, requires objective measurement of lung function or airway responsiveness using pulmonary function tests [57, 62, 63]. Reliance on clinical history without the support of objective measurements leads to misdiagnosis of asthma in 33% of cases [64]. While rates of use of pulmonary function tests vary widely based on jurisdiction, objective measurements are not widely utilized by primary care providers in asthma diagnosis [7]. Underutilization of objective measurements to diagnose asthma in primary care sites can lead to both underdiagnosis and overdiagnosis of asthma [65]. Individuals who suffer from asthma but have not received a diagnosis continue to struggle with symptom management and further contribute to the burden of asthma on the health care system through underdiagnosis [66]. Likewise, overdiagnosis of asthma also compounds the effects on the health care system through unnecessary patient visits and medication prescriptions [65].

In primary care, one of the greatest contributors to misdiagnosis stems from this lack of objective pulmonary function measurement in the process of diagnosing a patient with asthma [57, 62]. The use of objective measurements such as spirometry, methacholine challenge tests, and exercise challenge tests are crucial to ensuring the confirmation of an asthma diagnosis. Without completion of objective measures confirming asthma diagnoses, the accuracy of EMR data for patients labelled with asthma may not be valid. This creates barriers for researchers in creating KT eTools for asthma surveillance as many charts billed for asthma are in fact suspected asthma and not confirmed asthma.

Gaps in case definitions of asthma in EMRs

The gold-standard definition for the diagnosis of asthma is well established and provides clear guidance for practitioners to make an accurate diagnosis of asthma for patients with a clinical suspicion of the condition. There have been several attempts to translate the evidence-based clinical definition of asthma into a case definition to incidences of asthma in EMRs, however, no consensus has been reached [67]. Al Sallakh et al. (67) conducted an extensive review of attempts to define asthma using electronic health record data [67]. The review analyzed a total of 76 case definitions to identify asthma in EMRs and found significant heterogeneity in the case definitions proposed. This review found that for case definitions to be effective, they must be tailored to the EMR environment in which they function and consider the charting techniques of the practitioners who use the EMR. In Canada, there have been recent attempts to create a case definition for asthma suitable for Canadian EMR vendors and primary care practitioners. Previous efforts to create and validate case definitions for asthma have come from a single EMR or restricted data environments [68, 69]. Xi et al. (68) proposed a variety of case definitions using similar search fields with the addition of a search for asthma in the free text portion of the EMR and found a case definition of asthma that had a sensitivity of 90.2%, and a specificity of 83.9% [68]. Another recent publication from Cave et al. (69) conducted a study to validate a case definition for asthma using data from the Southern Alberta Primary Care Research Network, a node of the CPCSSN (SAPCReN-CPCSSN) [69]. The authors created a case-finding algorithm using a combination of search fields from the EMR including billing information, recorded encounter diagnosis information, and information inputted into a health condition field within the EMR. Cave et al. (2020) compared their algorithm against expert physician review of patient charts and found a sensitivity of 83.3% and specificity of 99.3%. Case definitions for asthma in EMRs have been proposed however, the limits of their generalizability remain unknown, as attempts have been limited to single EMRs in single jurisdictions and restricted data environments that do not have the ability to use all available data to make an accurate diagnosis.

Gaps in validation of case definitions for asthma

The reporting and validation of proposed case definitions within EMRs are critical to be able to draw reliable conclusions from the results of studies that derive their data from EMRs [70]. Determining the validity of the case definition of diseases such as asthma is more challenging than many other chronic conditions [71]. Asthma case definitions require a complex combination of symptom assessment, pulmonary function tests, and practitioner interpretation of the objective measurements. Meanwhile, other chronic conditions such as diabetes can be confirmed through a single blood test. Previous research has suggested several methods for creating and validating case definitions, including manual validation, comparison to external databases, comparison of rates in similar populations, and machine learning algorithms [72]. The preferred method for ensuring the accuracy of reference diagnoses when establishing cases of asthma is manual chart review, however manual reviews are labour-intensive and often involve additional methods to protect confidentiality that constitute barriers to this methodology [46]. Several studies assessed in this review contained minimal or no information on the methodology used in validating their proposed case definition for asthma. Overall, the information available on the methodology and results of previously published case definition validation studies for asthma is suboptimal. Thus, additional work is required to appropriately validate a case definition for asthma in which the methodology can be replicated.

Limitations of data sources

An important aspect of improving asthma outcomes is high quality datasets from which to derive health information to inform KT eTools for asthma diagnosis and surveillance [73]. A critical barrier to the scalability of asthma eTools are sources of accurate data. Most KT eTools described in this review originate from the local context in which they were created. This can limit the scalability of the eTools to other settings. The majority of KT eTools that have been developed are only functional in one EMR environment or derive information from health administrative data, which is limited in detail [38, 67–69]. While these tools may be effective in a local context, the inability to scale tools across multiple EMRs and the limited information provided by health administrative data create limitations on the ability of these KT eTools to scale to the national and international level. Furthermore, the quality of data available for eTools within the data sources serves as another potential limitation in the development of eTools for asthma. Inconsistencies in practitioner charting behaviours for asthma, particularly in primary care, can have a negative impact on the quality of data that informs eTools [74]. Further studies have demonstrated the high variability and generally low quality of information inputted into various EMR fields, affecting the completeness of data available to these eTools [75]. Adopting PRESTINE data elements in EMRs could improve upon the data sources currently available KT eTools.

Setting of KT eTool implementation

Another barrier to the development of KT eTools for asthma is that proposed KT eTools are often designed for different purposes depending on the health care setting or the purposes of the original study from which the definition was derived [67]. Many studies designing case definitions for asthma to incorporate into eTools have a wide range of sensitivity, specificity, positive predictive value, and negative predictive value. This variation and how it affects the selection of a case definition is important because a case definition with a high sensitivity is key to identifying all cases of asthma within a database, but if excluding non-cases is the area of interest then a high specificity is more important. Likewise, there is a trade-off between PPV and NPV in which one statistic can be more important than the other depending on whether the aim of the study is to determine true positives or true negatives. As a result, testing multiple case definitions to determine the case definition for asthma that is best suited for KT intervention which it will be used for is crucial to ensure the best case definition for the purpose of the intervention is selected.