Open-access programs for injury categorization using ICD-9 or ICD-10.

Area of Science:

• Traumatology
• Public Health
• Health Informatics

Background:

• The International Classification of Diseases (ICD) is crucial for injury coding and severity assessment.
• Established methods like the Abbreviated Injury Scale (AIS) and Injury Severity Score (ISS) have been used for decades.
• The original ICDPIC software, developed in Stata, facilitated injury severity scoring, with ICDPIC-R serving as an updated, open-access version compatible with ICD-9 and ICD-10.

Purpose of the Study:

• To introduce ICDPIC-R, an open-access software for injury categorization and severity scoring.
• To evaluate the performance of ICDPIC-R in deriving an approximate Injury Severity Score (RISS) from ICD-9 and ICD-10 codes.
• To compare the accuracy of RISS derived through different methods (GEM and ROCmax) against established ISS measures and mortality prediction.

Main Methods:

• Utilized data from the National Trauma Data Bank (NTDB) and National Inpatient Survey (NIS) for 2015.
• Employed ICDPIC-R to derive CDC injury mechanism categories and RISS from ICD-9 and ICD-10 codes.
• Applied General Equivalence Mapping (GEM) methods to convert ICD-10 to ICD-9, and a ROCmax method for direct ICD-10 to RISS calculation, comparing against registrar-submitted ISS (ISSAIS).

Main Results:

• ICDPIC-R successfully derived RISS from both ICD-9 and ICD-10 codes.
• For ICD-10 codes, the ROCmax method demonstrated superior mortality prediction (C-statistic 0.85 for NTDB, 0.78 for NIS) compared to GEM methods (0.75-0.76).
• Increasing RISS scores consistently correlated with increased mortality in both NTDB and NIS datasets, validating the score's utility.

Conclusions:

• ICDPIC-R provides a valuable tool for injury severity scoring when direct data is limited.
• The ROCmax method offers a robust approach for calculating RISS directly from ICD-10 codes.
• ICDPIC-R is a practical, open-access solution for injury categorization and severity assessment in public health research.