| Summary: | 碩士 === 國立臺北護理健康大學 === 護理研究所 === 103 === Nursing records are critical legal documents representing the process of nursing care. Recently, most nursing records in Taiwan have been computerized, but the quality of electronic nursing records has rarely been discussed. This research involved employing a data mining approach to evaluate the quality of nursing records by exploring the application of nursing staff recording system-formulated and nurse-formulated nursing events. This cross-sectional retrospective study was conducted at a 2,300-bed medical center in Taiwan. A total of 235,798 electronic nursing records for 6,277 patients discharged from the Department of Internal Medicine between January and June in 2014 were selected through purposive sampling as the research sample. SAS Enterprise Guide 6.1 was employed to analyze the structural data of system-formulated nursing events, and SAS Text Miner was employed to analyze the unstructured data of nurse-formulated nursing events. The efficiency of text mining based on a benchmark established by nursing experts was compared with that of text mining performed using SAS Text Miner in order to establish the sensitivity, specificity, and accuracy of SAS. In addition, we analyzed the relationship between nurse-formulated nursing events and system-formulated nursing events to provide a reference for evaluating nursing process record systems in the future. The results of this research show that (a) more nurses reported system-formulated nursing events (88.40%) than nurse-formulated nursing events (11.60%). (b) “Routine nursing round,” “follow-up events,” “admission nursing,” “fever,” and “start of chemotherapy” were used most frequently in system-formulated nursing events, whereas “wound care,” “discharge nursing,” “pain,” “catheter care,” and “blood transfusion” were used the most frequently in nurse-formulated nursing events. (c) In the text data mining of nurse-formulated nursing events, the sensitivity of SAS Text Miner in the training (testing) data set was approximately 0.96 (0.94), and the specificity and accuracy were 0.99 (0.99). (d) There was an 8.08% similarity between the nurse-formulated nursing events and system-formulated nursing events, and 29.72% of appropriately worded nurse-formulated nursing events were considered to have been added to the system-formulated nursing events. Therefore, the data-mining results could be considered a reference for updating nursing process record systems. The results of this research can be applied as teaching materials for writing nursing records and as a model for auditing the quality of nursing records at the research hospitals. We recommend that hospitals apply SAS Text Miner as a tool to facilitate auditing nursing record quality in the future. In addition, applying the proposed research data-mining model to auditing unstructured electronic nursing records may enhance the quality of electronic nursing records and facilitate implementing nursing records information systems.
|
|---|
