American Journal of Emergency Medicine 32 (2014) 417–420
Contents lists available at ScienceDirect
American Journal of Emergency Medicine journal homepage: www.elsevier.com/locate/ajem
Original Contribution
Comparison of 3 scoring systems to predict mortality from unstable upper gastrointestinal bleeding in cirrhotic patients Shou-Chien Hsu, MD 1, Chih-Yu Chen, MD 1, Yi-Ming Weng, MD ⁎, Shou-Yen Chen, MD, Chi-Chun Lin, MD, Jih-Chang Chen, MD Department of Emergency Medicine, Chang Gung Memorial Hospital and Chang Gung University College of Medicine, Linko, Taiwan
a r t i c l e
i n f o
Article history: Received 8 December 2013 Received in revised form 9 January 2014 Accepted 11 January 2014
a b s t r a c t Objective: We aimed to compare the performance of Glasgow-Blatchford, preendoscopic Rockall, and model for end-stage liver disease (MELD) scores in cirrhotic patients with unstable upper gastrointestinal bleeding (UGIB) in the emergency department (ED). Methods: This was a retrospective cohort study conducted at a university-affiliated teaching hospital. Adult cirrhotic patients who presented with acute UGIB and unstable vital signs (heart rate N100 beats/min or systolic blood pressure b100 mm Hg) between January 2009 and February 2011 were included. Patients who were transferred from another hospital, received no emergency endoscopy study, or had incomplete medical records were excluded. Data were retrieved from the admission list of the ED critical zone using international classification of disease code via computer registration. Results: Among enrolled visits, the initial median hemoglobin level was 8.6 (interquartile range, 7.2-10.1) mg/dL in the ED. The median heart rate and systolic blood pressure were 111.0 beats/min and 94.0 mm Hg, respectively. The endoscopic diagnosis of variceal bleeding accounted for 86.6% of the events. The mortality rate was 16.0% (19/119). Model for end-stage liver disease score performed better with an area under the curve (AUC) of 0.736 (95% confidence interval [CI], 0.629-0.842; P = .001) compared with other scoring systems (Glasgow-Blatchford score: AUC, 0.527; 95% CI, 0.393-0.661; P = .709; preendoscopic Rockall score: AUC, 0.591; 95% CI, 0.465-0.717; P = .208). Conclusion: Model for end-stage liver disease score performed better in terms of predicting mortality of unstable UGIB in cirrhotic patients compared with Glasgow-Blatchford and preendoscopic Rockall scores in the ED. © 2014 Elsevier Inc. All rights reserved.
1. Introduction 1.1. Background Acute upper gastrointestinal bleeding (UGIB) is the most common lethal complication in cirrhotic patients. The initial treatment of unstable UGIB in the emergency department (ED) includes resuscitation with crystalloid and blood transfusion, intravenous vasopressin, and prompt specialist consultation. Persistent or repeated bleeding with failure to respond to treatment results in mortality and morbidity. A previous report demonstrated a mortality rate of 30% in these patients, and only one-third survived for 1 year [1,2]. Although survival has
⁎ Corresponding author. Department of Emergency Medicine, Chang Gung Memorial Hospital and Chang Gung University College of Medicine, Gueishan Shiang, Taoyuan, Taiwan. Tel.: +886 3 3281 200x2505; fax: +886 3 3287 715. E-mail addresses: [email protected], [email protected] (Y.-M. Weng). 1 These authors contributed to this article equally. 0735-6757/$ – see front matter © 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ajem.2014.01.009
improved with modern techniques for controlling variceal hemorrhage, mortality rates remain high. It is a challenge for the emergency physician to identify patients at highest risk for rebleeding or even mortality. The development of a preendoscopy stratification system in an ED setting is particularly useful. Glasgow-Blatchford [3] and preendoscopic Rockall scores [4-9] have been validated in the ED. These scoring systems could be applied prior to endoscopy results and are particularly useful in the ED. Other studies have identified predictors of adverse outcomes of acute variceal hemorrhage in cirrhotic patients, including Child-Pugh class and model for end-stage liver disease (MELD) score [10-14]. There is limited evidence to compare the predictability of different scoring systems in such conditions.
1.2. Importance Knowledge of the performance of different scoring systems helps physicians to choose the proper tool to identify patients at high risk.
418
S.-C. Hsu et al. / American Journal of Emergency Medicine 32 (2014) 417–420
1.3. Goals of this investigation
2.4. Statistical tests
We aimed to compare the performance of Glasgow-Blatchford, preendoscopic Rockall, and MELD scores in cirrhotic patients with unstable UGIB in the ED.
The data were analyzed using SPSS 13.0 for Windows (SPSS, Chicago, IL). The demographic characteristics, scores for the different scoring systems, and outcomes are presented. Categorical variables are presented as numbers and percentages. Nonnormally distributed continuous variables are presented as medians and interquartile range (IQR). The areas under the curve (AUCs) that were abstracted via the receiver operating characteristic curves of the 3 score systems, which predict mortality, were compared. A cutoff optimum sensitivity and specificity were abstracted of each scale. The Hosmer-Lemeshow goodness-of-fit test was used for all 3 scoring systems. A P value less than .05 was considered significant.
2. Materials and methods 2.1. Study design and setting This was a retrospective cohort study conducted at a universityaffiliated teaching hospital with ~ 170 000 visits annually. Emergency physicians care for these patients. This study was approved by the Hospital Ethics Committee on Human Research. The study protocol was reviewed and qualified as exempt from the requirement to obtain informed consent. According to the standard protocol of our ED, patients who develop acute UGIB and an unstable hemodynamic state that warrants emergency endoscopy should be triaged and admitted to the critical care zone. All patients who were admitted to the critical care zone were resuscitated at the discretion of the physician and possibly included the use of endotracheal intubation to protect the airway, 2 peripheral lines for fluid challenge, blood transfusions, a nasogastric tube insertion with aspiration and medications applied, including vasoactive agents such as terlipressin infusion, and prophylactic antibiotics. Hemograms, coagulation profiles, and liver and renal function were checked. The physician in charge decided whether to start a scope team for emergent endoscopy examination according to his or her clinical judgment. An on-duty gastrointestinal specialist made the final decision as to whether an emergency endoscopy should be performed after evaluating the patients. Surgical management was considered for patients who developed persistent or repeated bleeding after failed endoscopy. Patients were admitted to the intensive care unit (ICU) or ward based on their response to initial management. Patients without evidence of active bleeding and in stable condition were transferred to a medical ward and might be discharged if there was no more gastrointestinal bleeding. 2.2. Patient selection Cirrhotic patients older than 18 years who presented with acute UGIB and unstable vital signs (heart rate N100 beats/min or systolic blood pressure b 100 mm Hg) and were admitted to the critical zone in the ED between January 1, 2009, and February 28, 2011, made up the study cohort. Patients who had no liver cirrhosis history, received no emergent endoscopy study, were transferred from another hospital, or had incomplete records were excluded. 2.3. Methods and measurements All patients who were admitted to the critical zone with a diagnosis of acute UGIB of International Classification of Disease, Ninth Edition codes 578.9, 578.0, or 578.2 were enrolled. Data were retrieved from the admission list of the ED critical zone via computer registration. A trained study nurse who was blinded to the study purpose performed the medical record review and data abstraction using a standardized template with a clear definition and code. The first author performed a quality improvement feedback after the data analysis. Demographic data were collected. The primary outcome was mortality during admission to the hospital. Other outcome measurements included repeat endoscopy; need for surgical intervention; amount of blood transfusion in 48 hours, including packed red blood cell and whole blood transfusion; length of ICU stay; and total hospital stay. The Glasgow-Blatchford [15], preendoscopic Rockall [16], and MELD [17] scores were calculated for each individual.
3. Results Three hundred forty visits of adult cirrhotic patients during the study period with a diagnosis of acute UGIB and International Classification of Disease, Ninth Edition code of 578.0, 578.1, or 578.9 were abstracted via admission lists of the ED critical zone. Of these visits, 221 were excluded because of the following criteria: 74 were not cirrhotic, 68 were transferred from other hospitals, 56 did not receive emergent endoscopy, 9 refused endoscopy, and 14 had incomplete records (Fig. 1). Patient characteristics are shown in Table 1. Among 119 visits (106 patients), 64 patients had a history of alcoholism or alcohol abuse, 32 were hepatitis B virus carriers, and 26 were hepatitis C virus carriers. More than half (85.7%) of enrolled patients were male with a median age of 51.0 (IQR, 43.0-58.0) years, and 78.2% of patients presented with hematemesis. The initial median hemoglobin level was 8.6 (IQR, 7.2-10.1) mg/dL in the ED. The median heart rate and systolic blood pressure were 111.0 beats/min and 94.0 mm Hg, respectively. The endoscopy diagnosis of variceal bleeding accounted for 86.6% of the events. The median MELD, GlasgowBlatchford, and preendoscopic Rockall scores were 16.0 (IQR, 13.021.0), 12.0 (IQR, 10.0-14.0), and 4.0 (IQR, 4.0-5.0), respectively.
340 visits with diagnosis of acute upper gastrointestinal bleeding and unstable vital sign 1
74 visits were noncirrhosis patient
56 visits did not receive emergent endoscopy study
68 visits were transferred from other hospitals
9 visits refused endoscopy study
14 visits had incomplete records 119 visits were enrolled.
Fig. 1. Patients enrolled in the study. Number of patient visits included and excluded from the study. Cirrhotic patients older than 18 years who presented with active UGIB and unstable vital signs and were admitted to critical zone in ED, between January 1, 2009, and February 28, 2011, made up the study cohort. Patients who were transferred from other hospital, had no emergent endoscopy study, refused endoscopy study, and had incomplete record were excluded.
S.-C. Hsu et al. / American Journal of Emergency Medicine 32 (2014) 417–420
419
Table 1 Baseline characteristics of enrolled patients n = 119 Patient characteristics Male, n (%) Age (y), median (IQR) Hepatocellular carcinoma, n (%) Presented with hematemesis, n (%) Presented with melena, n (%) Presented with syncope, n (%) Vital signs, median (IQR) Heart rate (beats/min) SBP (mm Hg) DBP (mm Hg) Laboratory tests, median (IQR) Hb (mg/dL) Platelete count (/dL) BUN (mg/dL) Cr (mg/dL) INR ALT (mg/dL) Total bilirubin (mg/dL) Terlipressin infusion, n (%) Prophylaxis antibiotics, n (%) Endoscopy diagnosis Esophageal varices, n (%) Gastric varices, n (%) Endoscopy hemostasis Band ligation, n (%) Sclerotherapy, n (%) Door to endoscopy (min), median (IQR) MELD score, median (IQR) Glasgow-Blatchford score, median (IQR) Preendoscopic Rockall score, median (IQR)
102 (85.7) 51.0 (43.0-58.0) 20 (16.8) 93 (78.2) 18 (15.1) 1 (0.8) 111.0 (97.0-125.0) 94.0 (77.0-121.0) 57.0 (48.0-74.0) 8.6 (7.2-10.1) 96.0 (65.0-136.0) 22.0 (13.7-39.0) 1.0 (0.8-1.5) 1.5 (1.3-1.8) 36.0 (25.0-48.5) 2.0 (1.4-4.2) 108 (90.8) 70 (58.8) 77 (64.7) 26 (21.8) 50 (42.0) 15 (12.6) 258.0 (167.0-493.0) 16.0 (13.0-21.0) 12.0 (10.0-14.0) 4.0 (4.0-5.0)
Fig. 2. The AUC of score systems (MELD: AUC, 0.736; 95% CI, 0.629-0.842; Blatchford score: AUC, 0.527; 95% CI, 0.393-0.661; preendoscopic Rockall score: AUC, 0.591; 95% CI, 0.465-0.717).
Abbreviations: SBP, systolic blood pressure; DBP, diastolic blood pressure; Hb, hemoglobin; BUN, blood urine nitrogen; Cr, creatinine; ALT, alanine aminotransferase.
Tables 2 and 3 show the outcomes and the scores of each scale in different outcome groups, respectively. The AUC of the 3 score systems is shown in Fig. 2 (MELD: AUC, 0.736; 95% confidence interval [CI], 0.629-0.842; P = .001; Glasgow-Blatchford score: AUC, 0.527; 95% CI, 0.393-0.661; P = .709; preendoscopic Rockall score: AUC, 0.591; 95% CI, 0.465-0.717; P = .208). A cutoff value of 16, 9, and 4 were abstracted, which yields optimum sensitivity and specificity of MELD, Glasgow-Blatchford, and preendoscopic Rockall score, respectively (Table 4). The 3 scoring systems did not achieve statistical significance for goodness of fit for regression model to predict mortality (Glasgow-Blatchford score, P = .967; preendoscopic Rockall score, P = .334; MELD, P = .129). 4. Discussion The mortality of unstable UGIB in cirrhotic patients was high, with 1 death per 6.3 events in our study. According to our study results, MELD score performed better in terms of predicting mortality from unstable UGIB in cirrhotic patients compared with Glasgow-Blatchford or preendoscopic Rockall scores. The 3 scoring systems could be applied before endoscopic examination in the ED. Thus, the MELD score might be the most appropriate tool for risk stratification and work as an
indicator of close monitoring and early consultation for emergent intervention in the ED. Further prospective validation is warranted. There were several explanations for our study results. First, our study enrolled patients who had a diagnosis of liver cirrhosis. The pressure of varices, which correlated with portal hypertension and the grade of liver disease, might be the pivot of the prognosis. A previous study had demonstrated that mortality and rebleeding were influenced by the severity of liver failure, with most deaths occurring in Child-Pugh grade C patients [4]. In contrast, a history of hepatic disease or liver failure accounted for 2 of 29 and 3 of 9 points in the Glasgow-Blatchford and preendoscopic Rockall scores, respectively. Limited parameters of the severity of liver disease or residual liver function were measured in both scoring systems. Other studies had shown that MELD score predicted the mortality of cirrhotic patients with UGIB [14] and was better than the ChildPugh score [13,18,19]. In addition, all measurements in the MELD score were objective and available in the ED. Several categories of Child-Pugh score are subjective and may not be available during patients' first visit, including the amount of ascites and grade of hepatic encephalopathy. In short, MELD score serves as a good assessment tool for risk stratification, and our study demonstrates its usefulness in such conditions. Second, we included patients who presented with unstable vital signs (heart rate N 100 beats/min or systolic blood pressure b100 mm Hg) at triage. Patients with an unstable hemodynamic state might yield high Glasgow-Blatchford and preendoscopic Rockall scores.
Table 2 Outcomes n = 119 Repeat endoscopy, n (%) Surgical intervention, n (%) Blood transfusion required in 48 h (unit), median (IQR) Length of ICU stay (d), median (IQR) Length of hospital stay (d), median (IQR) Mortality, n (%)
48 6 8.0 4.0 10.0 19
(40.3) (5.0) (6.0-12.0) (3.0-11.0) (5.0-18.0) (16.0)
Table 3 Scores of each scale in different outcome groups
MELD score, median (IQR) Glasgow-Blatchford score, median (IQR) Preendoscopic Rockall score, median (IQR)
Mortality (n = 19)
Survival (n = 100)
20.0 (17.0-24.0) 12.0 (10.0-14.0) 5.0 (4.0-5.0)
15.0 (12.0-20.0) 12.0 (10.0-14.0) 4.0 (3.0-5.0)
420
S.-C. Hsu et al. / American Journal of Emergency Medicine 32 (2014) 417–420
References
Table 4 A cutoff optimum sensitivity and specificity of each scale
Sensitivity, % (95% CI) Specificity, % (95% CI)
MELD score N16
Glasgow-Blatchford score N9
Pre-endoscopic Rockall score N4
94.7 (75.4-99.1) 54.0 (44.3-63.4)
84.2 (62.4-94.5) 24.0 (16.7-33.2)
57.9 (36.3-76.9) 59.0 (49.2-68.1)
Therefore, the difference of performance between these 2 scoring systems might be trivial and had no statistical significance in our study population. In another aspect, Glasgow-Blatchford [6,20-22] and preendoscopic Rockall scores [5] were useful to identify patients at low risk who might not need an urgent endoscopic examination and may be suitable for outpatient management. However, both scoring systems seem to be poor predictors of mortality [23,24]. Our study results echo previous findings and showed poor performance of both scoring systems as predictors of mortality. Many other scoring systems have been developed for this patient group in different settings. A 4-item Bleeding Risk Score, including cholinesterase, international normalized ratio (INR), viral or alcoholic etiology, and presence of varices, was used to identify patients at high risk for bleeding [25]. Another study provides a simple algorithm, which was based on the Child-Pugh score, creatinine level, and bacterial infection, allowing accurate early discrimination of the risk of mortality [26]. Cholongitas et al [27] and Tu et al [28] examined the general ICU models and concluded that Sequential Organ Failure Assessment, Acute Kidney Injury Network criteria, and MELD score had good predictive ability for in-hospital mortality in an ICU setting. Model for end-stage liver disease score had the advantage of minimal requirements with only 4 items: serum creatinine, bilirubin, INR level, and history of dialysis. All these requirements are objective, easy to obtain, and readily applied in the ED setting. 4.1. Limitations The present study should be interpreted in the context of the following limitations. First, our study was retrospective, and the data were collected from a computer database and medical record review. The retrospective design of our study may raise concerns about selection bias due to the large proportion of excluded patients who were transferred from other hospitals or received no emergent endoscopic study. Nevertheless, all enrolled patients had received no treatment prior to admission to our ED and had definite UGIB identified by scope examination. Therefore, the initial vital signs and laboratory results might correlate with the severity of disease and reduce interference of different scoring systems. Second, there might be confounding factors in our study, such as use of different treatment strategies according to physician judgment. In another aspect, all parameters of the 3 scoring systems of each enrolled patient were collected. Patients with incomplete records were excluded. Third, this study was conducted in a university-affiliated teaching hospital in Taoyuan, Taiwan, which may limit the general applicability of our findings. A comparative study in a different setting would be of interest. Fourth, the 3 scoring systems could not achieve statistical significance for goodness of fit for regression model to predict mortality. Although MELD score seems to be superior to the other 2 systems in our study, further research to developed a modified regression model based on MELD score to predict mortality in these patients might be warranted. 5. Conclusion Model for end-stage liver disease score performed better in terms of predicting mortality of unstable UGIB in cirrhotic patients compared with Glasgow-Blatchford and preendoscopic Rockall scores in the ED.
[1] Graham DY, Smith JL. The course of patients after variceal hemorrhage. Gastroenterology 1981;80(4):800–9. [2] Smith JL, Graham DY. Variceal hemorrhage: a critical evaluation of survival analysis. Gastroenterology 1982;82(5 Pt 1):968–73. [3] Blatchford O, Murray WR, Blatchford M. A risk score to predict need for treatment for upper-gastrointestinal haemorrhage. Lancet 2000;356(9238):1318–21. [4] Rockall TA, Logan RF, Devlin HB, Northfield TC. Risk assessment after acute upper gastrointestinal haemorrhage. Gut 1996;38(3):316–21. [5] Tham TC, James C, Kelly M. Predicting outcome of acute non-variceal upper gastrointestinal haemorrhage without endoscopy using the clinical Rockall Score. Postgraduate Med J 2006;82(973):757–9. [6] Stanley AJ, Dalton HR, Blatchford O, et al. Multicentre comparison of the Glasgow Blatchford and Rockall Scores in the prediction of clinical end-points after upper gastrointestinal haemorrhage. Alimentary pharmacology & therapeutics 2011;34(4): 470–5. [7] Sanders DS, Carter MJ, Goodchap RJ, Cross SS, Gleeson DC, Lobo AJ. Prospective validation of the Rockall risk scoring system for upper GI hemorrhage in subgroups of patients with varices and peptic ulcers. The American journal of gastroenterology 2002;97(3):630–5. [8] Wang CH, Chen YW, Young YR. A prospective comparison of 3 scoring systems for UGIB. Am J Emerg 2013;31:775–8. [9] Bryant RV, Kuo P, Williamson K, et al. Performance of the Glasgow-Blatchford score in predicting clinical outcomes and intervention in hospitalized patients with UGIB. Gastrointest Endoscopy 2013;78:576–83. [10] North Italian Endoscopic Club for the S, Treatment of Esophageal V. Prediction of the first variceal hemorrhage in patients with cirrhosis of the liver and esophageal varices. A prospective multicenter study. The N Engl J Med 1988;319(15):983–989. [11] Butt AK, Khan AA, Alam A, Shah SW, Shafqat F, Naqvi AB. Predicting hospital mortality in cirrhotic patients: comparison of Child-Pugh and Acute Physiology, Age and Chronic Health Evaluation (APACHE III) scoring systems. Am J Gastroenterol 1998;93(12):2469–75. [12] Bae WK, Lee JS, Kim NH, Kim KA, Moon YS, Oh MK. Usefulness of DeltaMELD/ month for prediction of the mortality in the first episode of variceal bleeding patients with liver cirrhosis: comparison with CTP, MELD score and DeltaCTP/ month. Korean J Hepatol 2007;13(1):51–60. [13] Benedeto-Stojanov D, Nagorni A, Bjelakovic G, Stojanov D, Mladenovic B, Djenic N. The model for the end-stage liver disease and Child-Pugh score in predicting prognosis in patients with liver cirrhosis and esophageal variceal bleeding. Vojnosanitetski pregled Military-medical and pharmaceutical review 2009;66(9): 724–8. [14] Hunter SS, Hamdy S. Predictors of early re-bleeding and mortality after acute variceal haemorrhage. Arab journal of gastroenterology: the official publication of the Pan-Arab Association of Gastroenterology 2013;14(2):63–7. [15] http://www.mdcalc.com/glasgow-blatchford-bleeding-score-gbs/. [16] http://en.wikipedia.org/wiki/Rockall_score. [17] http://www.mdcalc.com/meld-score-model-for-end-stage-liver-disease-12-andolder/. [18] Chen WT, Lin CY, Sheen IS, et al. MELD score can predict early mortality in patients with rebleeding after band ligation for variceal bleeding. World journal of gastroenterology: WJG 2011;17(16):2120–5. [19] Primignani M, Dell'Era A, Bucciarelli P, et al. High–D-dimer plasma levels predict poor outcome in esophageal variceal bleeding. Digest Liver Dis 2008;40(11): 874–81. [20] Pang SH, Ching JY, Lau JY, Sung JJ, Graham DY, Chan FK. Comparing the Blatchford and pre-endoscopic Rockall score in predicting the need for endoscopic therapy in patients with upper GI hemorrhage. Gastrointestinal Endoscopy 2010;71(7): 1134–40. [21] Giese A, Grunwald C, Zieren J, Buchner NJ, Henning BF. Pre-endoscopic assessment to predict outcome of acute upper gastrointestinal bleeding—a retrospective study in patients subject to emergency endoscopy outside regular working hours. Deutsche medizinische Wochenschrift 2013;138(21):1096–1101. [22] Srygley FD, Gerardo CJ, Tran T, Fisher DA. Does this patient have a severe upper gastrointestinal bleed? JAMA: the journal of the American Medical Association 2012;307(10):1072–9. [23] Koch A, Buendgens L, Duckers H, et al. Bleeding origin, patient-related risk factors, and prognostic indicators in patients with acute gastrointestinal hemorrhages requiring intensive care treatment. A retrospective analysis from 1999 to 2010. Medizinische Klinik Intensivmedizin und Notfallmedizin 2013;108(3):214–22. [24] Amitrano L, Guardascione MA, Manguso F, et al. The effectiveness of current acute variceal bleed treatments in unselected cirrhotic patients: refining short-term prognosis and risk factors. The American journal of gastroenterology 2012;107(12): 1872–8. [25] Tacke F, Fiedler K, Trautwein C. A simple clinical score predicts high risk for upper gastrointestinal hemorrhages from varices in patients with chronic liver disease. Scandinavian journal of gastroenterology 2007;42(3):374–82. [26] Augustin S, Muntaner L, Altamirano JT, et al. Predicting early mortality after acute variceal hemorrhage based on classification and regression tree analysis. Clin Gastroenterol Hepatol 2009;7(12):1347–54. [27] Cholongitas E, Senzolo M, Patch D, Shaw S, Hui C, Burroughs AK. Review article: scoring systems for assessing prognosis in critically ill adult cirrhotics. Alimentary pharmacology & therapeutics 2006;24(3):453–64. [28] Tu KH, Jenq CC, Tsai MH, et al. Outcome scoring systems for short-term prognosis in critically ill cirrhotic patients. Shock 2011;36(5):445–50.
Contents lists available at ScienceDirect
American Journal of Emergency Medicine journal homepage: www.elsevier.com/locate/ajem
Original Contribution
Comparison of 3 scoring systems to predict mortality from unstable upper gastrointestinal bleeding in cirrhotic patients Shou-Chien Hsu, MD 1, Chih-Yu Chen, MD 1, Yi-Ming Weng, MD ⁎, Shou-Yen Chen, MD, Chi-Chun Lin, MD, Jih-Chang Chen, MD Department of Emergency Medicine, Chang Gung Memorial Hospital and Chang Gung University College of Medicine, Linko, Taiwan
a r t i c l e
i n f o
Article history: Received 8 December 2013 Received in revised form 9 January 2014 Accepted 11 January 2014
a b s t r a c t Objective: We aimed to compare the performance of Glasgow-Blatchford, preendoscopic Rockall, and model for end-stage liver disease (MELD) scores in cirrhotic patients with unstable upper gastrointestinal bleeding (UGIB) in the emergency department (ED). Methods: This was a retrospective cohort study conducted at a university-affiliated teaching hospital. Adult cirrhotic patients who presented with acute UGIB and unstable vital signs (heart rate N100 beats/min or systolic blood pressure b100 mm Hg) between January 2009 and February 2011 were included. Patients who were transferred from another hospital, received no emergency endoscopy study, or had incomplete medical records were excluded. Data were retrieved from the admission list of the ED critical zone using international classification of disease code via computer registration. Results: Among enrolled visits, the initial median hemoglobin level was 8.6 (interquartile range, 7.2-10.1) mg/dL in the ED. The median heart rate and systolic blood pressure were 111.0 beats/min and 94.0 mm Hg, respectively. The endoscopic diagnosis of variceal bleeding accounted for 86.6% of the events. The mortality rate was 16.0% (19/119). Model for end-stage liver disease score performed better with an area under the curve (AUC) of 0.736 (95% confidence interval [CI], 0.629-0.842; P = .001) compared with other scoring systems (Glasgow-Blatchford score: AUC, 0.527; 95% CI, 0.393-0.661; P = .709; preendoscopic Rockall score: AUC, 0.591; 95% CI, 0.465-0.717; P = .208). Conclusion: Model for end-stage liver disease score performed better in terms of predicting mortality of unstable UGIB in cirrhotic patients compared with Glasgow-Blatchford and preendoscopic Rockall scores in the ED. © 2014 Elsevier Inc. All rights reserved.
1. Introduction 1.1. Background Acute upper gastrointestinal bleeding (UGIB) is the most common lethal complication in cirrhotic patients. The initial treatment of unstable UGIB in the emergency department (ED) includes resuscitation with crystalloid and blood transfusion, intravenous vasopressin, and prompt specialist consultation. Persistent or repeated bleeding with failure to respond to treatment results in mortality and morbidity. A previous report demonstrated a mortality rate of 30% in these patients, and only one-third survived for 1 year [1,2]. Although survival has
⁎ Corresponding author. Department of Emergency Medicine, Chang Gung Memorial Hospital and Chang Gung University College of Medicine, Gueishan Shiang, Taoyuan, Taiwan. Tel.: +886 3 3281 200x2505; fax: +886 3 3287 715. E-mail addresses: [email protected], [email protected] (Y.-M. Weng). 1 These authors contributed to this article equally. 0735-6757/$ – see front matter © 2014 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.ajem.2014.01.009
improved with modern techniques for controlling variceal hemorrhage, mortality rates remain high. It is a challenge for the emergency physician to identify patients at highest risk for rebleeding or even mortality. The development of a preendoscopy stratification system in an ED setting is particularly useful. Glasgow-Blatchford [3] and preendoscopic Rockall scores [4-9] have been validated in the ED. These scoring systems could be applied prior to endoscopy results and are particularly useful in the ED. Other studies have identified predictors of adverse outcomes of acute variceal hemorrhage in cirrhotic patients, including Child-Pugh class and model for end-stage liver disease (MELD) score [10-14]. There is limited evidence to compare the predictability of different scoring systems in such conditions.
1.2. Importance Knowledge of the performance of different scoring systems helps physicians to choose the proper tool to identify patients at high risk.
418
S.-C. Hsu et al. / American Journal of Emergency Medicine 32 (2014) 417–420
1.3. Goals of this investigation
2.4. Statistical tests
We aimed to compare the performance of Glasgow-Blatchford, preendoscopic Rockall, and MELD scores in cirrhotic patients with unstable UGIB in the ED.
The data were analyzed using SPSS 13.0 for Windows (SPSS, Chicago, IL). The demographic characteristics, scores for the different scoring systems, and outcomes are presented. Categorical variables are presented as numbers and percentages. Nonnormally distributed continuous variables are presented as medians and interquartile range (IQR). The areas under the curve (AUCs) that were abstracted via the receiver operating characteristic curves of the 3 score systems, which predict mortality, were compared. A cutoff optimum sensitivity and specificity were abstracted of each scale. The Hosmer-Lemeshow goodness-of-fit test was used for all 3 scoring systems. A P value less than .05 was considered significant.
2. Materials and methods 2.1. Study design and setting This was a retrospective cohort study conducted at a universityaffiliated teaching hospital with ~ 170 000 visits annually. Emergency physicians care for these patients. This study was approved by the Hospital Ethics Committee on Human Research. The study protocol was reviewed and qualified as exempt from the requirement to obtain informed consent. According to the standard protocol of our ED, patients who develop acute UGIB and an unstable hemodynamic state that warrants emergency endoscopy should be triaged and admitted to the critical care zone. All patients who were admitted to the critical care zone were resuscitated at the discretion of the physician and possibly included the use of endotracheal intubation to protect the airway, 2 peripheral lines for fluid challenge, blood transfusions, a nasogastric tube insertion with aspiration and medications applied, including vasoactive agents such as terlipressin infusion, and prophylactic antibiotics. Hemograms, coagulation profiles, and liver and renal function were checked. The physician in charge decided whether to start a scope team for emergent endoscopy examination according to his or her clinical judgment. An on-duty gastrointestinal specialist made the final decision as to whether an emergency endoscopy should be performed after evaluating the patients. Surgical management was considered for patients who developed persistent or repeated bleeding after failed endoscopy. Patients were admitted to the intensive care unit (ICU) or ward based on their response to initial management. Patients without evidence of active bleeding and in stable condition were transferred to a medical ward and might be discharged if there was no more gastrointestinal bleeding. 2.2. Patient selection Cirrhotic patients older than 18 years who presented with acute UGIB and unstable vital signs (heart rate N100 beats/min or systolic blood pressure b 100 mm Hg) and were admitted to the critical zone in the ED between January 1, 2009, and February 28, 2011, made up the study cohort. Patients who had no liver cirrhosis history, received no emergent endoscopy study, were transferred from another hospital, or had incomplete records were excluded. 2.3. Methods and measurements All patients who were admitted to the critical zone with a diagnosis of acute UGIB of International Classification of Disease, Ninth Edition codes 578.9, 578.0, or 578.2 were enrolled. Data were retrieved from the admission list of the ED critical zone via computer registration. A trained study nurse who was blinded to the study purpose performed the medical record review and data abstraction using a standardized template with a clear definition and code. The first author performed a quality improvement feedback after the data analysis. Demographic data were collected. The primary outcome was mortality during admission to the hospital. Other outcome measurements included repeat endoscopy; need for surgical intervention; amount of blood transfusion in 48 hours, including packed red blood cell and whole blood transfusion; length of ICU stay; and total hospital stay. The Glasgow-Blatchford [15], preendoscopic Rockall [16], and MELD [17] scores were calculated for each individual.
3. Results Three hundred forty visits of adult cirrhotic patients during the study period with a diagnosis of acute UGIB and International Classification of Disease, Ninth Edition code of 578.0, 578.1, or 578.9 were abstracted via admission lists of the ED critical zone. Of these visits, 221 were excluded because of the following criteria: 74 were not cirrhotic, 68 were transferred from other hospitals, 56 did not receive emergent endoscopy, 9 refused endoscopy, and 14 had incomplete records (Fig. 1). Patient characteristics are shown in Table 1. Among 119 visits (106 patients), 64 patients had a history of alcoholism or alcohol abuse, 32 were hepatitis B virus carriers, and 26 were hepatitis C virus carriers. More than half (85.7%) of enrolled patients were male with a median age of 51.0 (IQR, 43.0-58.0) years, and 78.2% of patients presented with hematemesis. The initial median hemoglobin level was 8.6 (IQR, 7.2-10.1) mg/dL in the ED. The median heart rate and systolic blood pressure were 111.0 beats/min and 94.0 mm Hg, respectively. The endoscopy diagnosis of variceal bleeding accounted for 86.6% of the events. The median MELD, GlasgowBlatchford, and preendoscopic Rockall scores were 16.0 (IQR, 13.021.0), 12.0 (IQR, 10.0-14.0), and 4.0 (IQR, 4.0-5.0), respectively.
340 visits with diagnosis of acute upper gastrointestinal bleeding and unstable vital sign 1
74 visits were noncirrhosis patient
56 visits did not receive emergent endoscopy study
68 visits were transferred from other hospitals
9 visits refused endoscopy study
14 visits had incomplete records 119 visits were enrolled.
Fig. 1. Patients enrolled in the study. Number of patient visits included and excluded from the study. Cirrhotic patients older than 18 years who presented with active UGIB and unstable vital signs and were admitted to critical zone in ED, between January 1, 2009, and February 28, 2011, made up the study cohort. Patients who were transferred from other hospital, had no emergent endoscopy study, refused endoscopy study, and had incomplete record were excluded.
S.-C. Hsu et al. / American Journal of Emergency Medicine 32 (2014) 417–420
419
Table 1 Baseline characteristics of enrolled patients n = 119 Patient characteristics Male, n (%) Age (y), median (IQR) Hepatocellular carcinoma, n (%) Presented with hematemesis, n (%) Presented with melena, n (%) Presented with syncope, n (%) Vital signs, median (IQR) Heart rate (beats/min) SBP (mm Hg) DBP (mm Hg) Laboratory tests, median (IQR) Hb (mg/dL) Platelete count (/dL) BUN (mg/dL) Cr (mg/dL) INR ALT (mg/dL) Total bilirubin (mg/dL) Terlipressin infusion, n (%) Prophylaxis antibiotics, n (%) Endoscopy diagnosis Esophageal varices, n (%) Gastric varices, n (%) Endoscopy hemostasis Band ligation, n (%) Sclerotherapy, n (%) Door to endoscopy (min), median (IQR) MELD score, median (IQR) Glasgow-Blatchford score, median (IQR) Preendoscopic Rockall score, median (IQR)
102 (85.7) 51.0 (43.0-58.0) 20 (16.8) 93 (78.2) 18 (15.1) 1 (0.8) 111.0 (97.0-125.0) 94.0 (77.0-121.0) 57.0 (48.0-74.0) 8.6 (7.2-10.1) 96.0 (65.0-136.0) 22.0 (13.7-39.0) 1.0 (0.8-1.5) 1.5 (1.3-1.8) 36.0 (25.0-48.5) 2.0 (1.4-4.2) 108 (90.8) 70 (58.8) 77 (64.7) 26 (21.8) 50 (42.0) 15 (12.6) 258.0 (167.0-493.0) 16.0 (13.0-21.0) 12.0 (10.0-14.0) 4.0 (4.0-5.0)
Fig. 2. The AUC of score systems (MELD: AUC, 0.736; 95% CI, 0.629-0.842; Blatchford score: AUC, 0.527; 95% CI, 0.393-0.661; preendoscopic Rockall score: AUC, 0.591; 95% CI, 0.465-0.717).
Abbreviations: SBP, systolic blood pressure; DBP, diastolic blood pressure; Hb, hemoglobin; BUN, blood urine nitrogen; Cr, creatinine; ALT, alanine aminotransferase.
Tables 2 and 3 show the outcomes and the scores of each scale in different outcome groups, respectively. The AUC of the 3 score systems is shown in Fig. 2 (MELD: AUC, 0.736; 95% confidence interval [CI], 0.629-0.842; P = .001; Glasgow-Blatchford score: AUC, 0.527; 95% CI, 0.393-0.661; P = .709; preendoscopic Rockall score: AUC, 0.591; 95% CI, 0.465-0.717; P = .208). A cutoff value of 16, 9, and 4 were abstracted, which yields optimum sensitivity and specificity of MELD, Glasgow-Blatchford, and preendoscopic Rockall score, respectively (Table 4). The 3 scoring systems did not achieve statistical significance for goodness of fit for regression model to predict mortality (Glasgow-Blatchford score, P = .967; preendoscopic Rockall score, P = .334; MELD, P = .129). 4. Discussion The mortality of unstable UGIB in cirrhotic patients was high, with 1 death per 6.3 events in our study. According to our study results, MELD score performed better in terms of predicting mortality from unstable UGIB in cirrhotic patients compared with Glasgow-Blatchford or preendoscopic Rockall scores. The 3 scoring systems could be applied before endoscopic examination in the ED. Thus, the MELD score might be the most appropriate tool for risk stratification and work as an
indicator of close monitoring and early consultation for emergent intervention in the ED. Further prospective validation is warranted. There were several explanations for our study results. First, our study enrolled patients who had a diagnosis of liver cirrhosis. The pressure of varices, which correlated with portal hypertension and the grade of liver disease, might be the pivot of the prognosis. A previous study had demonstrated that mortality and rebleeding were influenced by the severity of liver failure, with most deaths occurring in Child-Pugh grade C patients [4]. In contrast, a history of hepatic disease or liver failure accounted for 2 of 29 and 3 of 9 points in the Glasgow-Blatchford and preendoscopic Rockall scores, respectively. Limited parameters of the severity of liver disease or residual liver function were measured in both scoring systems. Other studies had shown that MELD score predicted the mortality of cirrhotic patients with UGIB [14] and was better than the ChildPugh score [13,18,19]. In addition, all measurements in the MELD score were objective and available in the ED. Several categories of Child-Pugh score are subjective and may not be available during patients' first visit, including the amount of ascites and grade of hepatic encephalopathy. In short, MELD score serves as a good assessment tool for risk stratification, and our study demonstrates its usefulness in such conditions. Second, we included patients who presented with unstable vital signs (heart rate N 100 beats/min or systolic blood pressure b100 mm Hg) at triage. Patients with an unstable hemodynamic state might yield high Glasgow-Blatchford and preendoscopic Rockall scores.
Table 2 Outcomes n = 119 Repeat endoscopy, n (%) Surgical intervention, n (%) Blood transfusion required in 48 h (unit), median (IQR) Length of ICU stay (d), median (IQR) Length of hospital stay (d), median (IQR) Mortality, n (%)
48 6 8.0 4.0 10.0 19
(40.3) (5.0) (6.0-12.0) (3.0-11.0) (5.0-18.0) (16.0)
Table 3 Scores of each scale in different outcome groups
MELD score, median (IQR) Glasgow-Blatchford score, median (IQR) Preendoscopic Rockall score, median (IQR)
Mortality (n = 19)
Survival (n = 100)
20.0 (17.0-24.0) 12.0 (10.0-14.0) 5.0 (4.0-5.0)
15.0 (12.0-20.0) 12.0 (10.0-14.0) 4.0 (3.0-5.0)
420
S.-C. Hsu et al. / American Journal of Emergency Medicine 32 (2014) 417–420
References
Table 4 A cutoff optimum sensitivity and specificity of each scale
Sensitivity, % (95% CI) Specificity, % (95% CI)
MELD score N16
Glasgow-Blatchford score N9
Pre-endoscopic Rockall score N4
94.7 (75.4-99.1) 54.0 (44.3-63.4)
84.2 (62.4-94.5) 24.0 (16.7-33.2)
57.9 (36.3-76.9) 59.0 (49.2-68.1)
Therefore, the difference of performance between these 2 scoring systems might be trivial and had no statistical significance in our study population. In another aspect, Glasgow-Blatchford [6,20-22] and preendoscopic Rockall scores [5] were useful to identify patients at low risk who might not need an urgent endoscopic examination and may be suitable for outpatient management. However, both scoring systems seem to be poor predictors of mortality [23,24]. Our study results echo previous findings and showed poor performance of both scoring systems as predictors of mortality. Many other scoring systems have been developed for this patient group in different settings. A 4-item Bleeding Risk Score, including cholinesterase, international normalized ratio (INR), viral or alcoholic etiology, and presence of varices, was used to identify patients at high risk for bleeding [25]. Another study provides a simple algorithm, which was based on the Child-Pugh score, creatinine level, and bacterial infection, allowing accurate early discrimination of the risk of mortality [26]. Cholongitas et al [27] and Tu et al [28] examined the general ICU models and concluded that Sequential Organ Failure Assessment, Acute Kidney Injury Network criteria, and MELD score had good predictive ability for in-hospital mortality in an ICU setting. Model for end-stage liver disease score had the advantage of minimal requirements with only 4 items: serum creatinine, bilirubin, INR level, and history of dialysis. All these requirements are objective, easy to obtain, and readily applied in the ED setting. 4.1. Limitations The present study should be interpreted in the context of the following limitations. First, our study was retrospective, and the data were collected from a computer database and medical record review. The retrospective design of our study may raise concerns about selection bias due to the large proportion of excluded patients who were transferred from other hospitals or received no emergent endoscopic study. Nevertheless, all enrolled patients had received no treatment prior to admission to our ED and had definite UGIB identified by scope examination. Therefore, the initial vital signs and laboratory results might correlate with the severity of disease and reduce interference of different scoring systems. Second, there might be confounding factors in our study, such as use of different treatment strategies according to physician judgment. In another aspect, all parameters of the 3 scoring systems of each enrolled patient were collected. Patients with incomplete records were excluded. Third, this study was conducted in a university-affiliated teaching hospital in Taoyuan, Taiwan, which may limit the general applicability of our findings. A comparative study in a different setting would be of interest. Fourth, the 3 scoring systems could not achieve statistical significance for goodness of fit for regression model to predict mortality. Although MELD score seems to be superior to the other 2 systems in our study, further research to developed a modified regression model based on MELD score to predict mortality in these patients might be warranted. 5. Conclusion Model for end-stage liver disease score performed better in terms of predicting mortality of unstable UGIB in cirrhotic patients compared with Glasgow-Blatchford and preendoscopic Rockall scores in the ED.
[1] Graham DY, Smith JL. The course of patients after variceal hemorrhage. Gastroenterology 1981;80(4):800–9. [2] Smith JL, Graham DY. Variceal hemorrhage: a critical evaluation of survival analysis. Gastroenterology 1982;82(5 Pt 1):968–73. [3] Blatchford O, Murray WR, Blatchford M. A risk score to predict need for treatment for upper-gastrointestinal haemorrhage. Lancet 2000;356(9238):1318–21. [4] Rockall TA, Logan RF, Devlin HB, Northfield TC. Risk assessment after acute upper gastrointestinal haemorrhage. Gut 1996;38(3):316–21. [5] Tham TC, James C, Kelly M. Predicting outcome of acute non-variceal upper gastrointestinal haemorrhage without endoscopy using the clinical Rockall Score. Postgraduate Med J 2006;82(973):757–9. [6] Stanley AJ, Dalton HR, Blatchford O, et al. Multicentre comparison of the Glasgow Blatchford and Rockall Scores in the prediction of clinical end-points after upper gastrointestinal haemorrhage. Alimentary pharmacology & therapeutics 2011;34(4): 470–5. [7] Sanders DS, Carter MJ, Goodchap RJ, Cross SS, Gleeson DC, Lobo AJ. Prospective validation of the Rockall risk scoring system for upper GI hemorrhage in subgroups of patients with varices and peptic ulcers. The American journal of gastroenterology 2002;97(3):630–5. [8] Wang CH, Chen YW, Young YR. A prospective comparison of 3 scoring systems for UGIB. Am J Emerg 2013;31:775–8. [9] Bryant RV, Kuo P, Williamson K, et al. Performance of the Glasgow-Blatchford score in predicting clinical outcomes and intervention in hospitalized patients with UGIB. Gastrointest Endoscopy 2013;78:576–83. [10] North Italian Endoscopic Club for the S, Treatment of Esophageal V. Prediction of the first variceal hemorrhage in patients with cirrhosis of the liver and esophageal varices. A prospective multicenter study. The N Engl J Med 1988;319(15):983–989. [11] Butt AK, Khan AA, Alam A, Shah SW, Shafqat F, Naqvi AB. Predicting hospital mortality in cirrhotic patients: comparison of Child-Pugh and Acute Physiology, Age and Chronic Health Evaluation (APACHE III) scoring systems. Am J Gastroenterol 1998;93(12):2469–75. [12] Bae WK, Lee JS, Kim NH, Kim KA, Moon YS, Oh MK. Usefulness of DeltaMELD/ month for prediction of the mortality in the first episode of variceal bleeding patients with liver cirrhosis: comparison with CTP, MELD score and DeltaCTP/ month. Korean J Hepatol 2007;13(1):51–60. [13] Benedeto-Stojanov D, Nagorni A, Bjelakovic G, Stojanov D, Mladenovic B, Djenic N. The model for the end-stage liver disease and Child-Pugh score in predicting prognosis in patients with liver cirrhosis and esophageal variceal bleeding. Vojnosanitetski pregled Military-medical and pharmaceutical review 2009;66(9): 724–8. [14] Hunter SS, Hamdy S. Predictors of early re-bleeding and mortality after acute variceal haemorrhage. Arab journal of gastroenterology: the official publication of the Pan-Arab Association of Gastroenterology 2013;14(2):63–7. [15] http://www.mdcalc.com/glasgow-blatchford-bleeding-score-gbs/. [16] http://en.wikipedia.org/wiki/Rockall_score. [17] http://www.mdcalc.com/meld-score-model-for-end-stage-liver-disease-12-andolder/. [18] Chen WT, Lin CY, Sheen IS, et al. MELD score can predict early mortality in patients with rebleeding after band ligation for variceal bleeding. World journal of gastroenterology: WJG 2011;17(16):2120–5. [19] Primignani M, Dell'Era A, Bucciarelli P, et al. High–D-dimer plasma levels predict poor outcome in esophageal variceal bleeding. Digest Liver Dis 2008;40(11): 874–81. [20] Pang SH, Ching JY, Lau JY, Sung JJ, Graham DY, Chan FK. Comparing the Blatchford and pre-endoscopic Rockall score in predicting the need for endoscopic therapy in patients with upper GI hemorrhage. Gastrointestinal Endoscopy 2010;71(7): 1134–40. [21] Giese A, Grunwald C, Zieren J, Buchner NJ, Henning BF. Pre-endoscopic assessment to predict outcome of acute upper gastrointestinal bleeding—a retrospective study in patients subject to emergency endoscopy outside regular working hours. Deutsche medizinische Wochenschrift 2013;138(21):1096–1101. [22] Srygley FD, Gerardo CJ, Tran T, Fisher DA. Does this patient have a severe upper gastrointestinal bleed? JAMA: the journal of the American Medical Association 2012;307(10):1072–9. [23] Koch A, Buendgens L, Duckers H, et al. Bleeding origin, patient-related risk factors, and prognostic indicators in patients with acute gastrointestinal hemorrhages requiring intensive care treatment. A retrospective analysis from 1999 to 2010. Medizinische Klinik Intensivmedizin und Notfallmedizin 2013;108(3):214–22. [24] Amitrano L, Guardascione MA, Manguso F, et al. The effectiveness of current acute variceal bleed treatments in unselected cirrhotic patients: refining short-term prognosis and risk factors. The American journal of gastroenterology 2012;107(12): 1872–8. [25] Tacke F, Fiedler K, Trautwein C. A simple clinical score predicts high risk for upper gastrointestinal hemorrhages from varices in patients with chronic liver disease. Scandinavian journal of gastroenterology 2007;42(3):374–82. [26] Augustin S, Muntaner L, Altamirano JT, et al. Predicting early mortality after acute variceal hemorrhage based on classification and regression tree analysis. Clin Gastroenterol Hepatol 2009;7(12):1347–54. [27] Cholongitas E, Senzolo M, Patch D, Shaw S, Hui C, Burroughs AK. Review article: scoring systems for assessing prognosis in critically ill adult cirrhotics. Alimentary pharmacology & therapeutics 2006;24(3):453–64. [28] Tu KH, Jenq CC, Tsai MH, et al. Outcome scoring systems for short-term prognosis in critically ill cirrhotic patients. Shock 2011;36(5):445–50.
