急診醫師決定急性上消化道出血病患住院的預測模式

The Predictive Model for Emergency Physicians Deciding to Admit Patients of Acute Upper Gastrointestinal Hemorrhage

吳元暉(Yuan-Hui Wu)

指導教授 ： 張睿詒

國立臺灣大學/公共衛生學院/醫療機構管理研究所/碩士(2008年)

https://doi.org/10.6342/NTU.2008.03022

摘要

背景：急性上消化道出血是急診常見疾病，嚴重出血的病患有死亡的潛在危險，應當住院治療，而病情穩定的病患往往在急診治療後不久即可出院。急診醫師是依據什麼樣的條件，決定上消化道出血病患是否應該住院，目前並沒有一致的標準。

目的：建立急診醫師決定急性上消化道出血病患是否住院的預測模式；建立急性上消化道出血病患總留院日數的預測模式。

方法：以醫院資訊系統與病歷審閱收集研究資料進行回溯性研究，選取2006年1月1日到2006年6月30日期間，至台北縣某家規模1000床教學醫院急診就診之所有上消化道出血病患，篩選主診斷為急性上消化道出血治療過程完整就醫前30日沒有出血紀錄、沒有嚴重合併症且毋須栓塞術、手術或加護病房治療者成為研究對象，將研究對象分為在急診治療達穩定後直接出院的急診組和由急診轉住院治療達穩定後出院的住院組，比較兩組病患在人口學、過去病史、藥物史、出血症狀與起始到就診時間、生命徵象、實驗室數據、輸血量、內視鏡診斷與出血痕跡、是否經內視鏡進行止血等各研究變項上的差異，以羅吉斯迴歸來分析住院的預測因子；以線性迴歸來分析總留院日數的預測因子。

結果：研究期間該醫院急診共有急性上消化道出血病患690人次就醫，經篩選進入研究資料庫有343人次，其中資料完整的335人次，包含急診組132人次和住院組203人次，經羅吉斯迴歸分析，校正和排除不顯著的自變項之後，是否有肝病、起始血色素和最嚴重內視鏡診斷成為決定病患住院最重要的預測因子，其中Forrest II潰瘍相對於非潰瘍非靜脈曲張診斷的勝算比是66.738倍；Forrest I潰瘍是61.107倍；靜脈曲張是24.813倍，最需要被安排住院。經線性迴歸分析，校正和排除不顯著的自變項之後，慢性病牽涉系統數、使用非類固醇消炎藥、輸血量、最嚴重內視鏡診斷和是否接受內視鏡止血術成為決定病患總留院日數最重要的預測因子，其中最嚴重診斷為Forrest III潰瘍、Forrest II潰瘍、Forrest I潰瘍與靜脈曲張時，相對於非潰瘍非靜脈曲張診斷，總留院日數分別增加0.915、1.209、1.357與2.106日，如果接受內視鏡止血治療，總留院日數比沒有接受內視鏡止血治療者再增加1.728日。

結論：在本研究進行的醫院，是否有肝病、起始血色素和最嚴重內視鏡診斷是急診醫師決定急性上消化道出血病患是否住院最重要的預測因子。而慢性病牽涉系統數、使用非類固醇消炎藥、輸血量、最嚴重內視鏡診斷和是否接受內視鏡止血術是影響總留院日數最重要的預測因子。

關鍵字

上消化道出血 ； 急診 ； 預測模式 ； 迴歸

並列摘要

Background: Acute upper gastrointestinal hemorrhage is a common disease of emergency medicine. Patients with severe hemorrhage have potential of mortality and admission should be arranged. Patients with stable condition may be discharged after a short-term emergency management. There is no consensus among emergency physicians deciding to admit a patient or not currently.

Objective: I attempted to establish a predictive model for emergency physicians deciding to admit patients of acute upper gastrointestinal hemorrhage and another predictive model of total length of stay among these patients.

Methods: To perform the retrospective study, the data were collected from the hospital information system and paper medical record. I collected data of patients visiting emergency department with any cause of upper gastrointestinal hemorrhage at a 1000-acute-bed teaching hospital in Taipei county from Jan. 1, 2006 to Jun. 30, 2006. I enrolled patients with a major problem of upper gastrointestinal hemorrhage, without previous hemorrhage within 30 days, severe acute comorbidities and needs of trans-arterial embolization, surgery, or intensive care unit admission. All the enrolled patient were separated into the following two groups. In ED group, patients received all the management in the emergency department and were discharged by emergency physician finally. In Ward group, patients were admitted to general ward after emergency management and were discharged by gastroenterologist finally. I compared the variables of demography, past history, drug history, symptoms, duration from symptoms onset to arrival, vital signs, lab data, transfused blood units, endoscopic diagnoses and stigmata, with or without any hemostasis procedure and so on between these two groups. Multiple logistic regressions were used to analyze the predictive factors of admission. Linear regressions were used to analyze the predictive factors of total length of stay.

Results: There were 690 cases of acute upper gastrointestinal hemorrhage visiting the emergency department of the hospital during the study period. Three hundred and forty-three cases were enrolled to the data base. Three hundred and thirty-five of them, including 132 cases in ED group and 203 cases in Ward group, had complete data and entered to both logistic and linear regression analysis mentioned above. After adjusting and removing more nonsignificant variables, liver disease, initial hemoglobin and most severe endoscopic diagnosis had most influence to decide admission or not. The odds ratio of Forrest II ulcer compared with non-varice-non-ulcer diagnosis was 66.738; the odds ratio of Forrest I ulcer was 61.107; the odds ratio of varice is 24.813. Admission was highly suggested at these diagnoses. After adjusting and removing more nonsignificant variables, Numbers of systems that chronic diseases involved, NSAID use, transfused blood units, most severe endoscopic diagnosis and endoscopic hemostasis procedure had most influence of total length of stay. The endoscopic diagnosis and hemostasis procedure were the most important factors of total length of stay. Compared with non-varice-non-ulcer diagnosis, it would increase 0.915, 1.209, 1.357 and 2.106 days in total length of stay when the most severe diagnoses were Forrest III ulcer, Forrest II ulcer, Forrest I ulcer and varice respectly. Another 1.728 days would be added to total length of stay if hemostasis procedure were performed.

Conclusions: In the study hospital, liver disease, initial hemoglobin and most severe endoscopic diagnosis were the most important predictive factors for emergency physician deciding to admit patients of acute upper gastrointestinal hemorrhage. Numbers of systems that chronic diseases involved, NSAID use, transfused blood units, most severe endoscopic diagnosis and endoscopic hemostasis procedure were the most important predictive factors for total length of stay.

並列關鍵字

upper gastrointestinal hemorrhage ； emergency department ； predictive model ； regression

參考文獻

36. Hu SC, Chang KS, Lee CH. Analysis of the characteristics of emergency patients in Taipei City. Tzu Chi Med J 1998;10(1): 9-17.

2. Rockall TA, Logan RF, Devlin HB, Northfield TC. Incidence of and mortality from acute upper gastrointestinal haemorrhage in the United Kingdom. BMJ. 1995 Jul 22;311(6999):222-6.

3. Longstreth GF. Epidemiology of patients hospitalization for acute upper gastrointestinal hemorrhage: A population-based study. Am J Gastroenterol 1995;90:206-210.

4. Blatchford O, Davidson LA, Murray WR, Blatchford M, Pell J. Acute upper gastrointestinal haemorrhage in west of Scotland: case ascertainment study. BMJ. 1997 Aug 30;315(7107):510-4.

5. Vreeburg EM, Snel P, de Bruijne JW, Bartelsman JF, Rauws EA, Tytgat GN. Acute upper gastrointestinal bleeding in the Amsterdam area: incidence, diagnosis, and clinical outcome. Am J Gastroenterol. 1997 Feb;92(2):236-43.