建構呼吸照護中心長期機械通氣病人死亡率之諾莫圖

A Nomogram for Predicting Hospital Mortality in Respiratory Care Center Patients with Prolonged Mechanical Ventilation

病人因呼吸衰竭（respiratory failure）而致需要長期機械通氣（prolonged mechanical ventilation, PMV）使用，其原因錯綜複雜，PMV病人相關預後及死亡率高，及早發現死亡因子，可輔助相關醫療決策。目前國內尚未有對於PMV病人量化的死亡風險預測，本研究目的是開發一個可評估及量化PMV病人住院死亡率的諾莫圖。臨床醫師可以用它來估計、及早預測PMV病人的院內死亡風險，並將模型預測的準確性與急性生理和慢性健康評分（Acute Physiology and Chronic Health Evaluation, APACHE II score）另進行比較。本研究採回溯性，研究對象開發組自2009年4月至2014年12月在加護病房因使用呼吸器困難脫離而轉入亞急性照護病房（respiratory care center, RCC）的病人，進行Lasso迴歸分析篩選死亡風險重要變項，而驗證組由2015年1月至2021年12月期間符合相同收案條件的病人所組成。預後結果定義為追蹤病人於當次住院期間是否死亡。研究結果，開發組轉入RCC之PMV病人（n=1024）、驗證組（n=776）。開發組的諾莫圖曲線下面積AUC（95%CI）：0.742（0.708-0.776）和驗證組AUC（95%CI）：0719（0.680-0.758），二組並無顯著差異（p=0.387），亦優於傳統APACHE II之ACU（95% CI）：0.685（0.648-0.722）達顯著差異（p=0.026）。本研究所建構的諾莫圖，可提供臨床醫師作為評估PMV病人死亡風險的實用且有效工具，具有優於APACHE II評分的預測能力，並可輔助醫療決策，藉由早期發現危險因子，及早介入適當醫療處置或臨終決定，以提升醫療照護品質。

The patient requires prolonged mechanical ventilation (PMV) due to respiratory failure, which has complex reasons. Patients requiring prolonged mechanical ventilation tend to have a poor prognosis and high mortality, but early diagnosis and treatment can reduce mortality. There is no quantitative prediction model for the mortality risk of patients with PMV, so the purpose of this study is to develop a nomogram that can assess and quantify the mortality rate of patients with PMV. It can used by clinicians to estimate and timely predict the hospital mortality risk of patients with PMV, and compare the prediction accuracy of the model with Acute Physiology and Chronic Health Evaluation (APACHE II) score. A retrospective study was conducted. The development group included patients who were treated with mechanical ventilation in the Intensive Care Unit between April 2009 and December 2014 and were transferred to the Respiratory Care Center (RCC) due to difficult weaning from mechanical ventilation. Lasso regression analysis was used to identify the important variables associated with the risk of mortality. The validation group consisted of patients who met the same inclusion criteria from January 2015 to December 2021. The study results showed that the development group consisted of 1,024 patients with PMV transferred to the RCC, while the validation group included 776 patients. The AUC (95% CI) for the development group was 0.742 (0.708-0.776), while for the validation group, it was 0.719 (0.680-0.758). There was no significant difference between the two groups (p=0.387). Furthermore, the AUC value was better than that of the traditional APACHE II (95% CI), which was 0.685 (0.648-0.722), with a significant statistical difference (p=0.026). The nomogram constructed in this study can provide clinicians with a practical and effective tool for assessing the mortality risk of patients with PMV. It has better predictive ability than the APACHE II score and can help identify risk factors early and intervene with appropriate medical treatment, thereby reducing and improving the mortality rate of patients with PMV.