護理品管壓傷資料庫與複合式查詢功能之建立與操作

Implementation and Operation of Pressure Injury Database and Compound Query in Nursing Quality Control

鍾明惠(Min-Huey Chung)

指導教授 ： 楊立偉

國立臺灣大學/進修推廣學院/事業經營碩士在職學位學程/碩士(2019年)

https://doi.org/10.6342/NTU201902641

摘要

透過管理醫療照護中的異常事件可以幫助管理階層檢測錯誤、分析其性質和原因、並建立預防錯誤的機制。然而大多數異常事件資訊管理系統未提供事件預測功能，而異常事件相關研究也多使用小樣本。因此本研究開發一個異常事件資料庫與管理資訊系統，提供臨床人員應用於異常事件的管理與預防。本文以壓傷為例，研究目的與方法包括：(1)設計異常事件關聯式資料庫，包括藥物、跌倒、管路、壓傷及針扎事件，本文以壓傷為例；(2)建立應用程式網頁服務，供使用者資料管理、查詢、及產生異常事件統計圖表；(3)提供使用者於網頁中以決策樹模型找出異常事件傷害程度的重要因子。本研究以機器學習中的隨機森林模性預測醫院壓傷事件嚴重程度，結果之正確率達54.5~64.4%。影響壓傷事件傷害程度最重要的因素是發生的單位，其餘則包括發生部位、骨突處反覆摩擦、使用輔具、長時間固定姿位、Barden Scale分數等。本系統除了方便醫療人員對院內的異常事件進行報表檢核，決策樹模型結果亦可供單位人員參考，對於異常事件的預防與管理有重要的貢獻，以提升病人安全。

關鍵字

異常事件 ； 雲端資料庫 ； 資訊系統 ； 決策樹

並列摘要

By managing adverse events in medical care, administrator can detect errors, analyze nature and causes of adverse events, and establish mechanisms to prevent them. However, most adverse event information management systems do not provide event prediction function, and most adverse event related research uses a small sample. Therefore, the present research develops an adverse event cloud database and management information system to provide clinical personnel for the management and prevention of adverse events. The specific methods and objectives include: (1) Designing and establishing an adverse event database with MySQL, including drugs events, falls events, tubing events, pressure injury and needle stick injury; (2) Building an application web service with PHP for user to manage data, search events, and generate statistical charts of adverse events; (3) Providing users explore important factors of damage of adverse events with decision tree model function in the web page. In this study, the random forest model was used to predict the severity of hospital pressure injuries, and the correct rate was 54.5~64.4%. The most important factors affecting the degree of pressure injury in the adverse events were the occurred units. The others include the site of occurrence, repeated friction at the apophysis, the use of assistive devices, long-term fixed posture, and Barden Scale score. In addition to facilitating medical personnel to report on adverse events in the hospital, this system also provides decision tree model results. These can be reference for clinical and have important contributions to the prevention and management of adverse events, and finally to enhance patient safety.

並列關鍵字

adverse events ； cloud database ； information system ； decision tree

參考文獻

Alderden, J., Pepper, G. A., Wilson, A., Whitney, J. D., Richardson, S., Butcher, R., . . . Cummins, M. R. (2018). Predicting Pressure Injury in Critical Care Patients: A Machine-Learning Model. American Journal Critical Care, 27(6), 461-468. doi:10.4037/ajcc2018525

Alderden, J., Whitney, J. D., Taylor, S. M., & Zaratkiewicz, S. (2011). Risk profile characteristics associated with outcomes of hospital-acquired pressure ulcers: a retrospective review. Critical Care Nurse, 31(4), 30-43.

Amir, Y., Lohrmann, C., Halfens, R. J., & Schols, J. M. (2017). Pressure ulcers in four Indonesian hospitals: prevalence, patient characteristics, ulcer characteristics, prevention and treatment. International Wound Journal, 14(1), 184-193.

Arnold-Long, M., Ayer, M., & Borchert, K. (2017). Medical device–related pressure injuries in long-term acute care hospital setting. Journal of Wound, Ostomy and Continence Nursing, 44(4), 325-330.

Børsting, T. E., Tvedt, C. R., Skogestad, I. J., Granheim, T. I., Gay, C. L., & Lerdal, A. (2018). Prevalence of pressure ulcer and associated risk factors in middle‐and older‐aged medical inpatients in Norway. Journal of Clinical Nursing, 27(3-4), e535-e543.