Formalisation And Automated Computation Of Diabetes Clinical Quality Indicators With Chinese Hospital Patient Data

Purpose Clinical quality indicators are used for the measurement of healthcare services and hospital performance. They can be classified into three kinds: structure-related indicators, process-related indicators and outcome indicators. The process related indicators measure the real activities done during healthcare services delivery and are very important for quality improvement, whose measurement are most complex. In China, most widely adopted indicators are structure-related and outcome-related, and often measure the whole quality of a hospital with a lack of process-related indicators and disease specific indicators. This study aim to use diabetes indicators as a focus to investigate whether clinical quality indicators(especially the process-related indicators) can be automated computed based on Chinese electronic records data.Methods 38 diabetes indicators were collected for the use of this study. Based on the ideas and theories of knowledge representation, the modified clinical quality indicators formalization(CLIF) tool and SNOMED CT terminology were adopted in this study for the extraction of concepts and relationships of the indicators and the construction of computer executable queries. Then patient data were collected from an electronic medical record system of a diabetes specialty hospital for the computation of the indicators.Results(1) Based on CLIF tool and SNOMED CT, all indicators were formalized successfully to constructed SQL queries. 6 indicators were not successfully computed because of the lack of specific data fields of the patient data. The formalization process of the indicators is in consistency of the 9 steps methods of CLIF and conducted in CLIF.(2) The CLIF tool was connected to the patient database and the constructed queries were run in the CLIF tool to get the computation results of the clinical quality indicators. Each of the 32 indicators can return a computation result.(3) SNOMED CT was used for the extraction of clinical quality indicators concepts and the coding of patient data. 21 concepts were used for the encoding of clinical quality indicators and 363 concepts were used for the coding of patient data.(4) Experts consultations were conducted for the evaluation of the formalization and automated computation results, the results are of high reliability.Conclusion The results indicated that the computation of diabetes quality indicators based on Chinese electronic medical record data is feasible. Besides, some structure and content characteristics of the data also impede the computation of the indicators. The electronic medical records can be improved to better support the computation of clinical quality indicators. The CLIF is a robust tool for the formalization of clinical quality indicators and SNOMED CT is adequate for the concepts representation of diabetes clinical quality indicators and the coding of patient data concepts.