Presentation Authors: Hannah Schaps, Brittany Adamic*, Alexander Cope, Craig Labbate, Sangtae Park, Chicago, IL
Introduction: Two recently enacted Federal mandates dramatically impacted medical care in USA. In 2014, the adoption of the electronic medical record (EMR), and in 2015, diagnostic coding using ICD-10 were mandated. While the implementation of ICD-10 aimed to increase diagnostic detail and granularity, the quintupling of diagnostic codes has led to increased documentation burden, burnout and dissatisfaction - leading physicians to use EMR shortcuts such as "favorite" buttons for his/her most commonly encountered diagnoses. We determined whether the accuracy of the ICD-10 bladder cancer (BT) coding is compromised by the use of "favorite" shortcut buttons in the EMR era.
Methods: We queried the Enterprise Data Warehouse (EDW) of our four hospital health system, which has used the EPIC EMR system 17 years, to care for 1.5 million annual outpatient and inpatient encounters. We included all patients diagnosed with bladder cancer (ICD-10 codes C67.0-C67.9) after undergoing TURBT or bladder biopsy (CPT procedural codes 52204, 52224, 52234, 52235, 52240, and 53314) from January 2016 to April 2018. The surgeons' operative reports were referenced to determine the accuracy of their ICD-10 BT coding in the EMR.
Results: Of 131 patients, 15 underwent bladder biopsy in the office. On average, 43.1% of BT patients were given incorrect ICD-10 codes, when their tumor was compared to the findings in the operative report. Incorrect coding was widespread among our institution's urologists, with the incidence of incorrect coding ranging from 34 to 62.5%. While no statistical difference was found when comparing the least accurate to most accurately coding urologist, when the least accurately coding physician was compared to the rest of his urological colleagues, he had significantly more frequent inaccurate ICD-10 diagnoses (p < 0.05) due to his common use of a "favorite" button to speed up documentation. To extrapolate our findings to a national scale, we identified bladder cancer patients in the NCDB (2014) and SEER (2014-2015) datasets, and found that 40% of the 49,146 and 43.4% of 37,003 bladder cancer patients respectively, had been coded as C67.9 "not otherwise specified," suggesting a similar trend nationwide.
Conclusions: The EMR documentation burden and undue complexity/granularity of ICD-10 coding have coincided to more diagnostic inaccuracy because physicians are using EMR short cuts to decrease charting time. In our large group practice, ICD-10 diagnosis was incorrectly entered 43.1% of the time when compared to operative note findings due to "favorited" codes.