The ADR is a high-value quality indicator because it is readily measured and is an established surrogate for the rare outcome post-colonoscopy cancer risk. Thus, it is imperative that all endoscopists have the means to know and track their ADR, and tools to facilitate more widespread measurement of ADR have the potential to support the delivery of high-value care. Here, we describe a novel method, leveraging pre-existing infrastructure within a widely adopted EHR, to do just that. In a pilot testing environment, our tool demonstrated a high sensitivity and accuracy compared to a manual chart review, similar to what studies evaluating NLP methods report8,9,10,11,12,13,14. To our knowledge, this is the first description of a method using EHR functionality to accurately capture endoscopic QIs.
There remain weaknesses to the tool we propose. It still requires effort from clinical staff to open a note and utilize the SmartList at the time of pathology result documentation. This requires several extra clicks beyond the normal workflow and in our pilot study resulted in 2/78 adenoma positive results being incorrectly classified. Additionally, our tool in its present form does not account for colonoscopy indication, completeness or prep quality. However, as previously noted tADR is an accurate surrogate for ADR, and others have even proposed that it may be a preferred colonoscopy QI as it simplifies measurement and may prevent gaming the ADR metric by changing colonoscopy indication18,21. Furthermore, additional macro selections (for instance for indication, bowel preparation quality, polyp histology, such as serrated lesions) can certainly be added to future iterations of our tool. These features can be customized based on a specific practices’ priorities in quality metric tracking and reporting (for instance if this information is not tracked elsewhere). The adaptability and customizability of our tool is a great strength, particularly if professional societies add additional lesion detection rates/benchmarks (such as serrated lesion detection rates) to quality metrics that should be measured.
We believe the tool we propose has several benefits. Because our tool relies largely on pre-existing abilities imbedded within an EHR, it does not require access to specialized data management systems often needed to adopt NLP based solutions. In addition, while NLP methods are often successful at individual institutions, adapting those tools across more diverse clinical settings has proven challenging15. Our tool can be scaled for use by anyone using the Epic EHR. Our tool provides real-time feedback within the EHR related to QI performance, allowing endoscopists to confront their own performance in the same interface in which they regularly manage patient care. While our tool does require some minimal effort from clinical staff, this is largely within normal clinical workflow and remains far less than what is required for manual extraction. Finally, though our tool was built using the Epic EHR, multiple EHRs have similar discrete data macro functionalities which could allow a similar tool to be developed in different systems both in the United States and Europe17,22.
Further work is needed to validate this tool among a greater proportion of endoscopists and ideally among multiple centers using the Epic EHR. Reassuringly, Smartlists are already widely used among everyday documentation in Epic. Additionally, prior research has demonstrated excellent adoption of Smartlists in post-colonoscopy EHR documentation23. Adjusting the structure of the Reporting Workbench algorithm by utilizing other macro data tools like additional SmartLists or flowsheets within the EHR may also allow for capture of additional data such as colonoscopy indication, prep quality and even allow for use of a similar tool to capture QIs in other endoscopic arenas.
This pilot study demonstrates the potential to leverage existing EHR functionality to achieve accurate measurement and feedback of tADR, a reliable surrogate for ADR. This tool may present an easily adoptable alternative to complex NLP based systems or time-intensive chart review to facilitate QI measurement and assure delivery of high-value care.