A governance framework for medical code standardization to enhance multi-institutional data quality.

backgroundData quality management is crucial for performing integrated analyses of medical data across multiple institutions, and mapping facility-specific local codes to standardized codes is a critical component of this process. This study aimed to improve the medical data quality of Medical Information Database Network (MID-NET®)-cooperating institutions by developing and implementing a governance framework for medical code standardization.

methodsA governance center was established at Kyushu University Hospital, which developed a differential output tool for detecting change logs in local and standardized codes. This tool was introduced to 18 MID-NET institutions to extract differences between updates and securely transfer them to the governance center. The governance procedures involved collecting and verifying mapping tables, assigning standard codes (HOT, JLAC-10, or ICD-10), and distributing updates to cooperating institutions. The full-scale operation of the governance process began in July 2020, facilitating continuous improvement in mapping accuracy and efficiency. The most optimal standardized code was proposed by medical professionals, and feedback was provided monthly to each institution.

resultsAfter approximately 1.5 years of governance, the correct standardized code assignment rates across all cooperating institutions were 36% for drugs, 29% for laboratory tests, and 67% for diseases. These values reflected the real-world baseline of standard code utilization in MID-NET institutions, where standardized codes had not been systematically assigned prior to governance implementation. Despite the monthly proposals provided by the governance center, the increase in registrations remained modest, particularly for laboratory tests, where the JLAC-10 codes were complex, highlighting the difficulty of achieving high coverage. However, the accumulation of differential data allowed for continuous monitoring of registration status and provided insights into problems and solutions at each institution. Mechanisms for semi-automatic registration and expansion of the governance system across multiple institutions and vendors were considered to further improve registration rates.

conclusionMaintaining high-quality data is crucial for ensuring reliable clinical collaboration and establishing a foundation for the secondary use of real-world data. This governance model provides a practical framework for data-driven projects that integrate centralized repositories with local electronic medical records, not only within MID-NET but also for other clinical research database initiatives.