S. W. Ross1, C. E. Reinke1, K. W. Cunningham1, M. M. Jordan1, J. Heaney1, M. Abraham1, F. Farnejad1, P. P. Palmer1, M. Kowalkowski1, M. A. Houston1, A. K. May1 1Carolinas Medical Center,Atrium Health Acute Care Surgery Network,Charlotte, NC, USA
Introduction:
In 2012 the American Association for the Surgery of Trauma (AAST) identified a core set of >450 International Disease Classification (ICD)-9 billing codes defining the scope of EGS. Numerous studies have utilized these codes to examine EGS patient outcomes. In October 2015, the United States converted to ICD-10, rendering ICD-9 definitions obsolete. To date, no published studies examine EGS outcomes utilizing ICD-10 codes or provide a conversion from ICD-9 to ICD-10. The purpose of this effort was to convert the AAST defined EGS codes to ICD-10 and enable future public use for quality, safety, and outcomes research.
Methods:
The AAST EGS codes, defined by Shafi et al, were mapped to ICD-10 codes in an automated fashion utilizing publicly-available billing software from HCUP. This was performed in a forward and then reverse mapping processes two times to incorporate adjacent ICD-10 codes that may be related. Next, code groups were manually reviewed by two authors to verify groupings and identify individual codes in similar groups not mapped in the automated process. A three round Delphi process was performed by 9 General and Acute Care Surgeons from 4 hospitals within our Acute Care Surgery Network to achieve final consensus for included codes.
Results:
Automated mapping identified 935 ICD-10 codes from the 472 AAST ICD-9 codes. Manual review identified another adjacent 1907 codes for a total of 2842. Automated software identified only 33% of codes included for review. Figure 1 displays the results of the Delphi Process. In round one, a unanimous consensus threshold was met for 1,196 (42.1%), with the majority of these codes being associated medical diagnoses that were excluded. For round 2, a threshold of 85% consensus for inclusion/exclusion was met for 1,192 codes (41.9%), leaving only 454 (16.0%) to adjudicate. A conference call was performed by a core group of the authors to discuss the final indeterminant codes and agreement was met among all surgeons through rational discussion and comparison to the original ICD-9 code set. In total, 1,580 (55.6%) of codes that were reviewed were included, and 44.4% were excluded. 880 (55.7%) of the codes did not map automatically through the software.
Conclusion:
Utilizing Delphi methodology, this is the first study to convert the ICD-9 AAST EGS codes to their ICD-10 equivalents. Automated mapping missed 67% of the codes selected for review, and 55.7% of the final ICD-10 codes. The resulting code set will be disseminated for review and validation for defining the EGS population. The resulting ICD-10 code set will enable current and future analysis of the public health burden of EGS diseases, more accurate patient risk assessment and outcomes evaluation, and facilitate system resource development and allocation.
