offering a real-time gauge of both individual- and institutional-level quality indicators, according to the developers.
The tool boasts an accuracy level exceeding 96%, integrates with multiple electronic health records, and requires minimal additional work time, reported Anmol Singh, MD, of TriStar Centennial Medical Center, Nashville, Tennessee, and colleagues.
“Implementation of quality indicator tracking remains difficult due to the complexity of ERCP as compared with other endoscopic procedures, resulting in significant limitations in the extraction and synthesis of these data,” the investigators wrote in Techniques and Innovations in Gastrointestinal Endoscopy. “Manual extraction methods such as self-assessment forms and chart reviews are both time intensive and error prone, and current automated extraction methods, such as natural language processing, can require substantial resources to implement and undesirably complicate the endoscopy work flow.”
To overcome these challenges, Dr. Singh and colleagues designed an analytics tool that automatically collects ERCP quality indicators from endoscopy reports with “minimal input” from the endoscopist, and is compatible with “any electronic reporting system.”
Development relied upon endoscopy records from 2,146 ERCPs performed by 12 endoscopists at four facilities. The most common reason for ERCP was choledocholithiasis, followed by malignant and benign biliary stricture. Most common procedures were stent placement and sphincterotomy.
Data were aggregated in a Health Level–7 (HL-7) interface, an international standard system that enables compatibility between different types of electronic health records. Some inputs were entered by the performing endoscopist via drop-down menus.
Next, data were shifted into an analytics suite, which evaluated quality indicators, including cannulation difficulty and success rate, and administration of post-ERCP pancreatitis prophylaxis.
Manual review showed that this approach yielded an accuracy of 96.5%-100%.
Beyond this high level of accuracy, Dr. Singh and colleagues described several reasons why their tool may be superior to previous attempts at an automated ERCP report card.
“Our HL-7–based tool offers several advantages, including versatility via compatibility with multiple types of commercial reporting software and flexibility in customizing the type and aesthetic of the data displayed,” they wrote. “These features improve the user interface, keep costs down, and allow for integration into smaller or nonacademic practice settings.”
They also highlighted how the tool measures quality in relation to procedure indication and difficulty at the provider level.
“Unlike in colonoscopy, where metrics such as adenoma detection rate can be ubiquitously applied to all screening procedures, the difficulty and risk profile of ERCP is inextricably dependent on patient and procedural factors such as indication of the procedure, history of interventions, or history of altered anatomy,” Dr. Singh and colleagues wrote. “Prior studies have shown that both the cost-effectiveness and complication rates of procedures are influenced by procedural indication and complexity. As such, benchmarking an individual provider’s performance necessarily requires the correct procedural context.”
With further optimization, this tool can be integrated into various types of existing endoscopy reporting software at a reasonable cost, and with minimal impact on routine work flow, the investigators concluded.
The investigators disclosed relationships with AbbVie, Boston Scientific, Organon, and others.
