Wei Xianggang gong ying yi liao xi tong jian li you xiao, ke kao ji shi shi de shou shu cheng xiao guan li yu jing xi tong

Ph.D.Background and Objectives:It is widely acknowledged that effective health systems require evaluation; thus, quality improvement programmes with evaluation and measurement indicators have been developed for surgical operations management in many jurisdictions. However, there is a lack of timely, contextual adjustment and robust measurement indicators in the surgical operation service in Hong Kong.This thesis aimed to develop a close-to-real-time warning signalling system to improve surgical quality and patient care. To achieve the thesis aim, potential risk factors were identified by means of healthcare administrative data to make risk adjustments for emergency surgical operations; diagnosis-related group (DRG) risk-adjusted models were developed and validated for different operating magnitudes; and the quality of surgical performance in Hong Kong public healthcare provision was measured using validated DRG risk-adjusted models.Methods:Multiple methodologies were adopted with systematic literature review, psychometric validation using secondary data and application of validated model using real-time data. Retrospective analysis and hierarchical logistic regression were performed to identify significant risk factors associated with 30-day mortality. Four approaches were used (data splitting approach; receiver operating characteristic (ROC) curve analysis; Hosmer-Lemeshow (H&L) goodness-of-fit test; and cross-validation analysis) to validate the DRG risk-adjusted models from July 1, 2014 to June 30, 2015. Variable life-adjusted display (VLAD) was employed to be a timely reporting surgical performance tool by means of validated DRG risk-adjusted models.Results:A total of 43 potential risk factors under three categories (demographics, laboratory results before operations, and DRG data) were selected from the literature review. Using the multivariable logistic regression analysis, there were 9 and 10 significant risk factors identified from the DRG risk-adjusted models for complex emergency operations and all emergency operations, respectively. These models showed a high correlation in areas under the ROC curve, with excellent discrimination ability and accuracy (C-statistic = 0.9153; H&L p=0.244 for complex operations, C-statistic = 0.9338; H&L p=0.276 for all operations). The VLAD curves were plot for HA overall, as well as for individual hospitals by the validated DRG risk-adjusted models to measure the quality of surgical performance in Hong Kong’s public healthcare system and provide a continuous monitoring mechanism. A significant downward trend was observed for the HA Overall around December 2014, with the “net loss” of around 9 lives, which might be partly due to a lack of senior medical staff available for supervision to housemanship, as well as the winter surge which limits the available medical resources.Conclusions:This was the first study to develop warning signalling system for surgical operations management by incorporating comprehensive contributory factors on the basis of the literature review and practical experience, using administrative data that integrated DRG elements into risk-adjusted models. The validated warning signalling system would identify possible areas for improvement action in surgical outcome management with minimal labour to achieve continuous monitoring and improve the quality of surgical operations and patient outcomes. Such models enabled cost-effectiveness, comprehensive case coverage, and close-to real-time monitoring of surgical patient outcomes that enable subsequent remedial actions to be taken to improve surgical quality and patient care.介紹和目標：現時行之有效的醫療系統均需要評估；因此許多司法管轄區都已為外科手術管理建立具評估及量度指標的質素改善計劃。然而，香港欠缺適時、可按背景調整及穩健的手術服務量度指標。本博士論文旨在建立一個接近實時的預警系統以改善手術與護理質素。本文先從醫療行政數據識別潛在風險因素，為緊急外科手術作風險調整；利用現有行政數據建立涵蓋各類型緊急手術的診斷相關群組 (DRG) 風險調整模型並予以驗證；再以該等模型量度香港公營醫療系統的手術表現質素。方法：研究方法包括系統性文獻回顧、現有數據的心理測量學驗證、以及應用實時數據於經驗證的模型。以回顧性分析及邏輯迴歸識別與30日死亡率相關的重大風險因素。以四種科研方法 (數據分割、ROC曲線分析、H&L適合度測試及交叉驗證)驗證自2014年7月1日至2015年6月30日的DRG風險調整模型。使用VLAD作為該等模型的適時報告工具。結果：從文獻回顧選出43 項潛在風險因素，分為人口因素、手術前化驗結果及DRG 數據三大類別。經過多元邏輯迴歸分析，從複雜及所有緊急手術的DRG 風險調整模型分別識別出9 項及10 項重大風險因素。該兩類模型的ROC 曲線下面積高度相關，擁有良好分辨能力及準確度 (複雜緊急手術C-statistic = 0.9153，H&L p=0.244；所有緊急手術C-statistic = 0.9338，H&L p=0.276)。使用經驗證DRG 風險調整模型繪製醫管局整體及個別醫院的VLAD 曲線，以量度香港公營醫療系統的手術表現質素並提供持續監察機制。醫管局整體VLAD 曲線在2014 年12 月期間呈顯著向下趨勢，手術死亡人數比預期多約9 人，部份原因可能由於駐院實習而缺乏高級醫務人員監督，及冬季服務高峰期令資源緊絀。結論：本論文為首項研究將行政數據將DRG元素納入風險調整模型，結合文獻回顧及實踐經驗所得的主要因素，以建立手術成效管理預警系統。經驗證的預警系統將可以最少人手識別手術成效管理中可改善的範疇，以達到持續監察及改善手術質素及成效。該等模型具成本效益，個案涵蓋全面，接近實時監察外科手術成效，以便制訂後續補救措施改善手術及護理質素。Wong, Kian.Thesis Ph.D. Chinese University of Hong Kong 2020.Includes bibliographical references (leaves 131-153).Abstracts also in Chinese.Title from PDF title page (viewed on November 9, 2021).Wong, Kian