Jonathan Arias, Ed O'Beirne, Luke S Spencer Gardner, Paul A Rizk, Keith T Aziz, Courtney E Sherman, Bryan D Springer
Ambient listening-generated documentation meaningfully influences E/M code determination in orthopaedic outpatient practice, supporting higher and more appropriate E/M level selection compared with dictation. Although distinct error patterns persist, ambient listening was associated with reduced undercoding and measurable financial impact. These findings position ambient listening technology as a clinically and financially consequential documentation modality, warranting thoughtful implementation and continued coding oversight.
BACKGROUND: Accurate Evaluation and Management (E/M) coding depends on the precise documentation of medical decision making. Advances in artificial intelligence have introduced ambient listening technology as a novel documentation tool. Despite its rapid adoption, the effect of ambient listening technology on E/M code selection, coding accuracy, and downstream financial implications remains poorly defined. This study aimed to evaluate the effect of ambient listening technology on E/M code selection, coding concordance, and associated revenue implications within an orthopaedic surgery practice.
METHODS: A comparative cross-sectional analysis was done of all outpatient E/M services at a single academic medical center. Providers were required to have a minimum of 20 encounters documented using ambient listening technology and traditional dictation for the same encounter type. The final analytic cohort included 7,236 encounters. A randomized subset of 200 encounters (100 ambient listening and 100 traditional) underwent blinded, independent audit by an experienced orthopaedic coding auditor. Outcomes included concordance between provider-selected and auditor-determined E/M levels, weighted coding discrepancies, accuracy rates, and modeled revenue effect.
RESULTS: Encounters documented with ambient listening technology were associated with higher E/M levels than those documented with traditional dictation, as assessed by both auditors (3.24 vs 2.96) and providers (3.11 vs 2.82). Overall coding accuracy was 68% for ambient listening and 74% for dictation, while focused review of higher-level ambient listening encounters demonstrated 81.3% accuracy. Revenue modeling projected a potential annual increase of $232,313 if dictation-based encounters achieved E/M distributions observed with ambient listening.
CONCLUSION: Ambient listening-generated documentation meaningfully influences E/M code determination in orthopaedic outpatient practice, supporting higher and more appropriate E/M level selection compared with dictation. Although distinct error patterns persist, ambient listening was associated with reduced undercoding and measurable financial impact. These findings position ambient listening technology as a clinically and financially consequential documentation modality, warranting thoughtful implementation and continued coding oversight.