The Path Toward Harmonized Codesets for Emergency General Surgery

INTRODUCTION In 2012, the American Association for the Surgery of Trauma (AAST) established a list of International Classification of Diseases, Ninth Edition, Clinical Modification (ICD-9-CM) codes that identify primary emergency general surgery (EGS) diagnoses through a Delphi process.1 In 2015, however, ICD-9-CM was replaced with the International Classification of Diseases, Tenth Edition, Clinical Modification (ICD-10-CM). Since then, EGS researchers have had to manually develop inclusion and exclusion logic to generate EGS datasets without a standardized approach. The manuscript, “Lost in translation: evaluation of an automated software method versus a Delphi process to translate emergency general surgery ICD diagnosis codes,”2 addresses this problem by first evaluating MapIT, an automated software that crosswalks ICD-9-CM codes to ICD-10-CM equivalents. The authors then performed a Delphi analysis and found that MapIT grossly underestimated the appropriate code inclusion set. After refining the codes through the Delphi process, they validated accuracy by performing a trend analysis using the healthcare cost and utilization project (HCUP) National Inpatient Sample before and after ICD-10-CM implementation. The study’s design intentionally couples computational efficiency (MapIT crosswalks) with clinical adjudication (adjacent-code review and Delphi), an essential step, given that a single International Classification of Diseases, Ninth Edition (ICD-9) code often fragments into multiple International Classification of Diseases, Tenth Edition (ICD-10) variants encoding critical nuance. Notably, more than half of the final ICD-10 codes would have been missed if relying on MapIT alone, explaining why their National Inpatient Sample analyses based on MapIT systematically underestimate EGS burden compared with the Delphi list. These discrepancies are not trivial: at the national level, they translate into millions of encounters and can dramatically alter conclusions about disease burden, resource allocation, and quality gaps. This work builds on prior efforts. Ingraham et al3 performed an AAST-sponsored study refining an ICD-10 EGS list using general equivalence mappings with physician review, finalizing 985 ICD-10-CM codes, and demonstrating the national burden of adult EGS encounters in 2016 to 2017.3 Dalton et al4 compared the GEM crosswalk to historical ICD-9 volumes in the National Inpatient Sample, mapping the AAST ICD-9 set to 1206 unique ICD-10 codes. They identified five recurrent failure modes—changes in admission volumes, loss of necessary modifiers, lack of specific ICD-10 codes, mapping to unrelated conditions, and nomenclature changes—all of which required manual clinician correction similar to this study’s Delphi process. LIMITATIONS AND FUTURE DIRECTIONS The primary limitation of this and other studies to date is the subjectivity inherent to the Delphi consensus. This study suffers from an additional limitation because the panel was drawn from a single health system. That said, the diversity of practice settings represented enhances external validity, and validation against national data adds further credibility. A broader concern is that each group performing this analysis has generated different codesets to define the EGS population, and none have been prospectively validated against manually curated and multi-institutional EGS registries and electronic health record service lists, leaving researchers without a clear standard for accuracy and comparability. Here, the researchers identified almost 600 additional ICD-10-CM codes than the AAST-supported study led by Ingraham. As EGS-specific registries (eg, Emergent general surgery quality improvement program (EGSQIP)) gain traction, establishing a nationally accepted consensus set of inclusion codes becomes urgent. Without this, automated registry inclusion will remain fragmented and inconsistent. IMPACT AND NEXT STEPS Research based on incomplete cohorts risks underestimating system demand and mischaracterizing quality. One potential next step is a head-to-head validation of existing EGS codesets. This could be prospectively tested across multiple centers by comparing identified patients with actual EGS consult, operation, and admission logs, measuring who was captured, who was missed, and with what degree of accuracy. Such evidence can support an AAST Delphi refinement and ultimate endorsement, creating the necessary foundation for standardization and stability in the field. CONCLUSION This article delivers a validated, clinician-curated ICD-10 codeset that preserves fidelity to the ICD-9 era while leveraging ICD-10’s diagnostic granularity. However, the coexistence of multiple competing EGS codesets remains problematic. Establishing best practices now will also lay the groundwork for future transitions, including the adoption of ICD-11, which many countries have already started to implement.