Abstract
BACKGROUND: The importance of codes for documenting healthcare concepts has expanded from a historical need to collect statistics to today's wide array of clinical and other secondary uses. The most deeply embedded system for coding diagnoses is the World Health Organization's (WHO) International Classification of Diseases (ICD). WHO's eleventh revision, ICD-11, was evaluated to determine how best to meet the needs of stakeholders concerned about clinical quality, healthcare decision-making, and other secondary use cases. METHODS: This study explored the capability of ICD-10-CM, the standard ICD-11 for Mortality and Morbidity Statistics (ICD-11-MMS), and an enhanced ICD-11-MMS prototype (ICD-11-CCL) to represent clinical patient profiles for two natural clinical language scenarios. Systems were evaluated for granularity, comprehensiveness, accuracy, precision, relevancy, and their capability to provide user-friendly coding support. RESULTS: Both the standard and enhanced ICD-11-MMS produced code clusters that captured more clinical details and insight than ICD-10-CM's codes. However, ICD-11-MMS was less comprehensive than ICD-11-CCL and introduced occasional new inaccuracies. Unlike ICD-10-CM and ICD-11-MMS, ICD-11-CCL enabled clinicians to designate their clinical perspectives directly. While ICD-11-MMS produced numerous synonymous code clusters, the enhanced syntax of ICD-11-CCL resulted in a unique code cluster for each scenario. CONCLUSIONS: Adoption of an enhanced ICD-11-MMS such as ICD-11-CCL could provide clinically nuanced data to support important secondary data applications more effectively than standard ICD-11-MMS, while retaining ICD-11-MMS's capability to support international comparative analyses such as mortality and morbidity statistics. With automated coding tools, enhanced ICD-11-MMS could also reduce the burden on humans to select, review, memorize or decipher codes. Enhanced ICD-11-MMS represents a substantial improvement over customized modifications of ICD-10 and, with further development and testing, could be a viable option for future adoption.