Do coder characteristics influence validity of ICD-10 hospital discharge data?

Area of Science:

• Health Informatics
• Health Services Research
• Medical Coding

Background:

• Administrative data are crucial for health systems research and policy decisions.
• Understanding factors influencing administrative data validity is essential.
• Coder characteristics' impact on data validity remains largely unknown.

Purpose of the Study:

• To examine the relationship between coder characteristics and the validity of hospital discharge data.
• To assess how coding volume, employment status, and hospital type influence data validity indicators.

Main Methods:

• A descriptive study analyzed 6 face validity indicators in ICD-10 coded discharge records from 4 Calgary hospitals (2002-2007).
• Compared mean diagnoses, procedures, complications, Z-codes, and specific code endings by coder volume, employment status, and hospital type.
• Assessed inter-rater reliability using kappa statistics comparing coded data with chart reviews.

Main Results:

• No significant variation in most face validity indicators was found based on coder characteristics.
• The mean number of diagnoses per record decreased over time, while annual coding volume increased.
• Agreement between coded data and chart reviews did not consistently correlate with coder characteristics.

Conclusions:

• Coder characteristics appear to have minimal influence on the validity of hospital discharge data.
• Implementing standardized training, management, and rotation policies may enhance data quality.
• Privacy concerns limited the study to a few coder characteristics.