ICD-10-CM-Based Algorithms to Identify Bacteremia Incorporating Pathogen-Specific Codes: A Validation Study

Area of Science:

• Clinical Informatics
• Infectious Diseases
• Health Services Research

Background:

• Real-world bacteremia studies often rely on microbiologic data.
• When microbiologic data is absent, diagnostic codes are used to classify bacteremia.
• Accurate classification of bacteremia types is crucial for patient management and research.

Purpose of the Study:

• To develop and validate International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM)-based algorithms.
• To identify any bacteremia, Gram-negative bacteremia, and Enterobacteriaceae bacteremia.
• To assess the performance of these algorithms in a real-world clinical setting.

Main Methods:

• Retrospective analysis of adult hospitalizations from January 1, 2021, to May 31, 2024.
• Used blood culture results as the reference standard for bacteremia classification.
• Evaluated 15 candidate algorithms based on ICD-10-CM codes, including bacteremia and pathogen-specific codes, assessing sensitivity, specificity, and predictive values.

Main Results:

• Algorithms incorporating bacteremia and pathogen-specific codes demonstrated high specificity (97.3%-100.0%) but low sensitivity (4.9%-58.1%).
• Algorithms using only bacteremia codes showed higher sensitivity (67.4%-68.6%) with lower specificity (91.1%-91.2%).
• Positive predictive value for any bacteremia was highest in elderly patients discharged on Gram-negative antibiotic coverage.

Conclusions:

• Validated ICD-10-CM-based algorithms for identifying bacteremia, Gram-negative bacteremia, and Enterobacteriaceae bacteremia in adult patients.
• Algorithms including pathogen-specific codes are suitable for studies prioritizing specificity.
• Algorithms requiring only a general bacteremia code are preferable for studies prioritizing sensitivity.