急診病患簽立不施行心肺復甦術的臨床實效研究

背景：台灣高齡人口快數增加，預估2025年老年人口會達到20%以上，正式進入超高齡社會，2000年6月《安寧緩和醫療條例》正式立法通過施行及《病人自主權利法》已於2019年1月6日通過且公告，有鑑於國人面對安寧緩和醫療照護的意識抬頭及需求日益增加，因此急診面對求醫人口越來越多、疾病型態日益多元化、複雜化，面臨死亡的機率也漸增，對於急診安寧緩和醫療照護相關的議題更顯重要。若能在急診評估相關死亡風險因子及預測死亡，有助於確定生命末期病人之安寧緩和醫療需求。研究結果可作為推動急診病患安寧緩和醫療照護之參考。目的：本研究之目的為急診住院病患簽立不施行心肺復甦術的風險評估及死亡率的臨床時效性研究，探討急診住院病患簽立不施行心肺復甦術是否對死亡有影響及依據之前建立的風險評分篩檢模型之結果進行外部驗證，並用於預測4年的死亡率及安寧緩和醫療需求的確立。方法：本研究於臺北市某區域醫院急診進行之回溯性研究，收案條件為15歲以上經急診住院之患者，排除兒科病患。由2015年6月至2016年5月的4627名患者組成發展族群(developing cohort)，由2017年1月至2020年12月的20970名患者組成驗證族群。A-qCPR (Age, qSOFA, Cancer, Performance scales, DNR)模型之風險評分：年齡（每年0.05分）、qSOFA≥2(1分)、日常功能狀態量表≥2(2分)、DNR狀態(3分)和癌症(4分)。描述性統計以α＝0.05為顯著水準，若是連續變項時，以t-test來檢定；類別變項時，則使用卡方檢定(Chi-square test)，通過邏輯迴歸模式計算具有95%信賴區間(CI)的勝算比(OR)，並採用多變量邏輯迴歸模型用於確定四年死亡率的最重要決定因素。驗證的預測概率繪製了ROC曲線，通過驗證族群中的ROC曲線下面積(AUROC)對預測模型進行了外部驗證。Kaplan-Meier方法評估生存率，並應用對數秩檢定(log-rank test)評估組間的生存率差異。進一步嘗試對於不同危險因子進行分析，使用Cox proportional hazard model來研究調整風險因素影響存活率。最後使用AFT model (accelerated failure time model)評估對存活時間影響。統計方法以SAS 9.4版進行分析。結果：我們模型在發展族群之AUROC曲線是0.84（0.83-0.85），基於發展族群在驗證族群為0.707（0.700-0.714）;多變量邏輯迴歸模型，4年死亡率的ROC曲線下面積為0.733(0.727-0.740)。多變量邏輯迴歸模式分析急診住院患者的死亡危險因子，以下變項具有統計顯著意義：年齡(1.02-1.02)、性別(1.16-1.30)、qSOFA≧2(1.57-2.07)、PS≧2(1.80-2.09)、有DNR(1.04-1.20)、有Cancer(2.85-3.31)、有創傷(0.67-0.78)、SQ(1.75-2.11)、Triage 1(2.37-3.12)、Triage 2(1.40-1.77)及Triage 3(0.79-0.97)。低風險（≤4分）、中等風險（4到9分）和高風險（>9分以上）這三個類別的4年死亡率分別為23.2% (22.1%-24.3%)、47.4% (46.5%-48.3%) 和65.5%(64.0%-67.0%)。本篩檢工具相較驚訝問題(SQ)：0.195(0.185-0.205)與PS：0.415(0.403-0.428)有較高敏感度0.949(0.943–0.955)和SQ：0.595(0.588-0.602)與功能狀態量表(PS;performance scales)：0.673(0.629-0.645)有較高陰性預測值0.793(0.776–0.809)。多變項Cox proportional hazard model分析風險危險時，以下變項具有統計顯著意義：年齡(1.008-1.011)、性別(1.088-1.181)、qSOFA≧2(1.269-1.476)、PS≧2(1.360-1.500)、有DNR(1.045-1.153)、有Cancer(1.855-2.033)、有創傷(0.750-0.841)、SQ-N (1.307-1.457)、SQ-D(1.328-1.803)、Triage 1(1.798-2.039)、Triage 2(1.333-1.477)及Triage 3(0.703-0.788)。多變項AFT model分析顯示以下變項具有統計顯著意義：年齡(0.989-0.994)、性別(0.841-0.959)、qSOFA≧2(0.511-0.646)、PS≧2(0.648-0.759)、有DNR(0.791-0.926)、有Cancer(0.290-0.337)、有創傷(1.404-1.697)、SQ-N (0.668-0.805)、SQ-D(0.573-0.821)、Triage 1(0.310-0.411)、Triage 2(0.533-0.687)及Triage 3(1.028-1.297)。其中有簽DNR比沒簽DNR少活15.5%。結論：急診住院病患以死亡風險評分時，可以快速、簡單且客觀找出需要臨終關懷和安寧緩和醫療需求者，有助於凝聚醫病雙方安寧緩和照顧之共識，協助末期病患家屬進行安寧決策，協助後續身心靈的「療癒（healing）」，藉由實現「臨終者善終；失親者善別；在世者善生」，達生死兩相安。

關鍵字

急診科；緩和醫療；臨終關懷；日常功能狀態量表；器官衰竭評估；不施行心肺復甦術； A-qCPR 模型

並列摘要

Background：The population of the elderly is increasing rapidly in Taiwan. It is estimated that the elderly will reach more than 20% of the total population in 2025, and formally enter a super-aged society. The " Hospice Palliative Care Act " was formally enacted and passed in June 2000, and the "Patient Right to Autonomy Act" announced on January 6, 2019. In view of the rising awareness and increasing demand for hospice and palliative care among patients and their families in ED (emergency department). However, we have to deal with more diversified and complex disease patterns in the ED, and patients and their families have different goals of treatments, such as aggressive medical care for cure or hospice with palliative care. Hence, it is more important to screen hospice needs now. If we can assess the relevant death risk factors and predict death in the emergency department, it will help to communicate with patients or their families for determining whether the hospice/palliative care at the end of life or not. The results of the study can be used as a reference for promoting the palliative medical care of emergency patients. Objective：The study is aimed to evaluate the impact of patients’ signing DNR (do not cardio-pulmonary resuscitation) on the mortality in the ED, and to evaluate the previously hospice need screening model for predicting 4-year mortality and potential hospice need by external validation. Methods：A retrospective cohort study was conducted. Patients with more than 15-year-old who were admitted to the hospital from ED were enrolled in Taipei City Hospital, Ren-Ai branch, and pediatric patients were excluded. The developing cohort consisted of 4,627 patients from June 2015 to May 2016, and was followed-up until May 2017 in the ED. The validation cohort, aimed at confirming the value of mortality prediction, consisted of 20,970 patients in the ED between 2017 and 2020. A-qCPR (Age, qSOFA, Cancer, Performance scales, DNR) model was developed：age (0.05 point per year), qSOFA ≥ 2 (1 point), Eastern Cooperative Oncology Group Performance Status score ≥ 2 (2 points), Do-Not-Resuscitate status (3 points), and Cancer (4 points). In descriptive statistics, Chi-square test for categorical data, and T-test for continuous data were used. A two-sided P value of less than 0.05 indicated statistical significance. Odds ratios (OR) with 95% confidence intervals (CI) were calculated by the logistic regression model. In addition, the multi-variable logistic regression model with a stepwise selection procedure was used for identifying the most important determining factors for mortality. The area under the ROC curve (AUROC) was used for validating our model in the validation cohort. Furthermore, a summary ROC curve was plotted based on the validated predicted probabilities. Survival was estimated by the Kaplan–Meier method, and differences in survival between groups were assessed by the log-rank test. The Cox proportional hazard model was used for identifying risk factors of death. Finally, AFT (accelerated failure time) model was used to evaluate the impact on the survival time. SAS statistical software (Version 9.4; SAS Institute Inc, Cary, NC) was used for these analyses. Results：Our model is developing cohort area under the ROC curves for the A-qCPR model were 0.84 (0.83-0.85) based on the result of cross validation in the developing cohort. However, the area under the ROC curves for the A-qCPR model were 0.707 (0.700-0.714). The multivariate logistic regression model, the area under the ROC curve of the 4- year mortality rate is 0.733 (0.727-0.740). The multivariate logistic regression model was also used for analyzing risk factors of death in the validation cohort. The following variables are statistically significant: age (1.02-1.02) , gender (1.16-1.30) , qSOFA ≧ 2 (1.57-2.07) , PS ≧ 2 (1.80-2.09) , with a DNR (1.04-1.20) , there Cancer (2.85-3.31) , trauma (0.67-0.78) , SQ (1.75-2.11) , the Triage. 1 (2.37-3.12) , the Triage 2 (1.40- 1.77) and Triage 3 (0.79-0.97) . Low risk ( ≤4 points), moderate risk ( 4 Dao 9 points) and high risk ( > 9 points or more) of these three classes of 4 -year mortality rates were 23.2% (22.1% -24.3%) , 47.4% (46.5 %-48.3%) and 65.5% (64.0%-67.0%) .This screening tool compared SQ : 0.195 (0.185-0.205) and PS : 0.415 (0.403-0.428) had a relatively high sensitivity of 0.949 (0.943-0.955) and SQ : 0.595 (0.588-0.602) and PS(performance scales) : 0.673 (0.629- 0.645) had relatively high negative predictive value of 0.793 (0.776-0.809). The multivariable Cox proportional hazard model was used for the analysis of risk factors associated with death. The following variables is statistically significant and significance : Age (1.008-1.011) , gender (1.088-1.181) , qSOFA ≧ 2 (1.269-1.476) , PS ≧ 2 (1.360-1.500 ) , with a DNR (1.045-1.153) , there Cancer (1.855-2.033) , trauma (0.750-0.841) , SQ-N (1.307-1.457) , SQ-D (1.328-1.803) , the Triage. 1 (1.798-2.039 ) , Triage 2 (1.333-1.477) and Triage 3 (0.703-0.788) . The multivariate AFT model showed that age (0.989-0.994) , gender (0.841-0.959) , qSOFA ≧ 2 ( 0.511-0.646 ) , PS ≧ 2 (0.648-0.759) , with DNR ( 0.791-0.926) , Cancer (0.290-0.337) , Trauma (1.404-1.697) , SQ-N (0.668-0.805) , SQ-D (0.573-0.821) , Triage 1 (0.310-0.411) , Triage 2( 0.533-0.687) and Triage 3 ( 1.028-1.297 ) were statistically significant. The survival time of patients with signing DNR is 15.5% less than those without signing DNR. Conclusion：When patients, who were admitted from ED, were screened by our risk scores, we can quickly, simply, and objectively find out those who need hospice care and palliative care. The tools can help to unite the consensus of palliative care between doctors and patients, to assist the family members of terminal patients in making decisions, to assist in the follow-up "healing" of body and mind, and to achieve peace of life and death by realizing "the dying, the bereaved, and the living".