Clinically meaningful categorisation of ICD-10-AM (Australian modification).

Area of Science:

• Health Informatics
• Medical Classification Systems
• Public Health Data Analysis

Background:

• Current International Statistical Classification of Diseases and Related Health Problems (ICD) categorization methods present limitations in administrative data analysis and healthcare outcome monitoring.
• Issues include many-to-one code relationships, non-linear sequencing, collinearity, and ambiguous miscellaneous codes, hindering accurate data interpretation.

Purpose of the Study:

• To present a novel methodology for clinically meaningful categorization of the 12th Edition of ICD Version 10 Australian Modification (ICD-10-AM).
• To develop a system that addresses the limitations of existing ICD-10-AM categorization for administrative data.

Main Methods:

• Diagnosis codes from ICD-10-AM were aggregated into Clinical Diagnosis Group (CDG) sets based on clinical features and associated risks (in-hospital death, complications).
• Residual codes, reasons for encounter, and ill-defined symptom codes were excluded from the CDG sets.
• Administrative data from Victoria, Australia (July 2020-June 2023) were analyzed to determine the frequency of diagnoses within CDG sets.

Main Results:

• 87.9% of total ICD-10-AM codes (12,716) were successfully aggregated into 406 CDG sets.
• 12.1% of codes (1,753) were excluded, including residual, reason for encounter, and ill-defined symptom codes.
• Over 36 months, 90.1% of unique ICD-10-AM diagnoses from 11.8 million separations were included in CDG sets, with 91.2% of codes associated with death or complications present in a CDG set.

Conclusions:

• The developed Clinical Diagnosis Group (CDG) list offers a clinically meaningful approach to categorizing and interrogating ICD-10-AM datasets.
• This methodology complements existing classification systems, improving the analysis of healthcare administrative data and outcomes.