EDITORIAL Improved risk assessment in upper GI bleeding Acute upper GI bleeding is the most common GI emergency, with a reported incidence in various epidemiological studies ranging from 50 to over 100 per 100,000 inhabitants per year. Treatment primarily consists of resuscitation, endoscopy with therapeutic intervention in case of high-risk stigmata, and profound acid suppressive therapy, with later diagnosis and treatment of the underlying cause. 1 Resuscitation includes intravenous administration of fluids, including packed cells if needed. Other measures in selected patients may include oxygen administration or mechanical ventilation and correction of the effects of anticoagulant therapy. First, endoscopy is needed for diagnosis of the bleeding source and, second, it offers the opportunity for therapeutic intervention in patients with signs of active bleeding as well as in those with a visible vessel. In case of an adherent clot, attempts at clot removal with endoscopic treatment of any underlying lesion should be considered. The risk that this will induce a spurting bleed is rather limited, whereas clot removal reveals an underlying vessel in a significant proportion of patients. Profound acid suppressive therapy by means of a proton pump inhibitor forms the final mainstay of initial treatment. 2 Despite all these measures, upper GI bleeding remains associated with significant morbidity and mortality. The high incidence of the condition and the associated morbidity and mortality make adequate risk assessment of patients with upper GI bleeding mandatory. The international consensus on management of nonvariceal upper GI bleeding therefore recommends the use of prognostic scales for early stratification of patients into categories of low-risk and high-risk for rebleeding and mortality. 1 This allows, on one hand, early, targeted treatment of highrisk cases, while low-risk cases can be discharged early. Together, this aims to minimize the risk of rebleeding, complications, and mortality while limiting the use of resources. For these purposes, several risk assessment systems have been published over the last 2 decades, by using a limited variety of clinical as well as laboratory and endoscopic criteria (Table 1). They all were proposed on the basis of analysis of a cohort of patients, either single-center or multicenter, with later validation in other series. Some studies directly compared different risk assessment systems within the same patient population. The systems differ in their focus Copyright 2011 by the American Society for Gastrointestinal Endoscopy 0016-5107/$36.00 doi:10.1016/j.gie.2011.07.040 on the situation before, during, or after endoscopy. Furthermore, some are more appropriate for predicting the need for intervention and the risk of rebleeding, whereas others focus on the risk of mortality. Dr. Saltzman and colleagues, in this issue of the Journal, present a carefully derived and validated new prediction rule for mortality of patients with upper GI bleeding. RISK ASSESSMENT BEFORE ENDOSCOPY Several systems are based on clinical and laboratory parameters that can be assessed during the first presentation at an emergency unit. This is a major advantage, because it allows a very rapid assessment at first presentation, which may help determine decisions on admission, timing of endoscopy, and other measures. These preendoscopy systems include the Rockall 3 and the Blatchford systems. 4 The pre-endoscopy Rockall system is based on a patient s age, comorbidities, and blood pressure. The Blatchford system does not take age into account, but instead includes urea and hemoglobin levels as basic laboratory tests. It, in particular, aims to predict the need for any intervention. In a large United Kingdom (U.K.) trial including 676 patients presenting with upper GI bleeding, 16% had a Blatchford score of 0. 4 None of these patients required any intervention or died, and none of those undergoing later endoscopy had any major bleeding focus. Seventeen percent of the same population had a pre-endoscopy Rockall score of 0, but these patients required 44 interventions (ie, endoscopic therapy, surgery, and/or transfusion), and one of them died. As such, the Blatchford system outperformed the pre-endoscopy Rockall system. This was confirmed by other studies. 5-7 The Blatchford system can thus be used as a tool to select patients for early discharge and later endoscopy during office hours. In this respect, one limitation is that the proportion of patients with a Blatchford score of 0 was, in various series, considerably lower than the 16% in the U.K. study. 5,7-9 This variation may be due to differences in local populations and referral indications, but it also may be related to different approaches for endoscopic intervention. This obviously affects the discriminative power for www.giejournal.org Volume 74, No. 6 : 2011 GASTROINTESTINAL ENDOSCOPY 1225
Editorial Kuipers TABLE 1. Parameters of the Baylor, 10 Cedars-Sinai, 11 Rockall, 3 Blatchford, 4 and AIMS65 13 scoring systems* Baylor Cedars-Sinai Rockall Blatchford AIMS65** Category/parameter Parameter Score Parameter Score Parameter Score Parameter Score Parameter Score Age, y 30 0 N/A 60-79 1 N/A 65 1 30-49 1 80 2 50-59 2 60-69 3 70 5 Clinical status Systolic blood pressure N/A Stable 0 100 2 100-109 1 90 1 Intermediate 1 90-99 2 Unstable 2 90 3 Pulse N/A N/A 100 1 100 1 N/A Melena N/A N/A N/A Present 1 N/A Syncope N/A N/A N/A Present 1 N/A Altered mental status N/A N/A N/A N/A Present 1 Comorbidities 1-2 1 1 0 Any major comorbidity 3-4 4 2 1 Renal/liver failure, or disseminated malignancy 2 Hepatic disease 2 N/A 3 Cardiac failure 2 5 5 3 2 Chronic life-threatening disease Acute life-threatening disease 4 4 3 5 Time since symptom onset N/A 48 h 0 N/A N/A N/A 48 h 1 In hospital 2 Laboratory parameters Blood urea, mmol/l N/A N/A N/A 6.5-7.9 2 N/A 8.0-9.9 3 10-24.9 4 25.0 6 Hemoglobin, g/l N/A N/A N/A Men, 120-130 1 N/A Women, 100-120 1 Men, 100-120 3 Men and women, 100 6 INR N/A N/A N/A N/A 1.5 1 Albumin N/A N/A N/A N/A 3.0 g/dl 1 Endoscopy Endoscopic diagnosis Bleeding from posterior wall bulb 4 No focus or Mallory-Weiss, PUD no SRH 0 No focus or Mallory- Weiss 0 N/A N/A Spot or clot 1 Upper GI malignancy PUD SRH 2 All other diagnoses 2 1226 GASTROINTESTINAL ENDOSCOPY Volume 74, No. 6 : 2011 www.giejournal.org
Kuipers Editorial TABLE 1. Continued Baylor Cedars-Sinai Rockall Blatchford AIMS65** Category/parameter Parameter Score Parameter Score Parameter Score Parameter Score Parameter Score Persistent UGIB, varices, cancer 3 4 1 Endoscopic SRH Clot 1 N/A None/dark spot only 0 N/A N/A Visible vessel 3 Blood/clot/vessel Active bleeding 5 2 N/A, Not applicable; INR, international normalized ratio; PUD, peptic ulcer disease; SRH, stigmata of recent hemorrhage; UGIH, upper GI bleeding. *The Blatchford and AIMS65 are pre-endoscopy scoring systems, the Rockall system has pre-endoscopy and post-endoscopy components, whereas the Baylor and Cedars-Sinai systems are post-endoscopy systems. Low risk defined as score of 6. Low risk defined as score of 2. Scores before and after endoscopy. Low risk defined as scores of 2, high risk as scores 6. Low risk defined as score of 0. **Low risk defined as score of 1, high risk as score 2. patients at very low risk. Although one may consider for local use to increase the low-risk cutoff to a score of 2, this no longer excludes the need for intervention and thus will put some patients at risk. 4 RISK ASSESSMENT DURING ENDOSCOPY Endoscopic predictors of an increased risk of rebleeding are signs of active bleeding, and the presence of a non-bleeding, visible vessel or adherent clot, with rebleeding rates during medical therapy ranging from 0% to 10% for ulcers with a clean base to 80% to 90% for ulcers showing a spurting bleed during endoscopy. This is captured by the Forrest classification, with Forrest stage I reflecting active bleeding, stage II reflecting a visible vessel and adherent clot, and stage III reflecting a clean-base ulcer. The classification is widely used yet has two limitations. The first lies within category IB, referring to oozing bleeds. Although the Forrest classification lists oozing bleeds as one category, oozing from a visible vessel is associated with a high risk of recurrent significant bleeding, whereas oozing from an ulcer margin represents a low risk for significant rebleeding. The second limitation of the Forrest classification is that it does not take ulcer size and location into account, whereas larger ulcers and those at the posterior duodenal bulb represent a higher risk for significant rebleeding than other lesions. The same pertains to varices and upper GI cancers. Together, these endoscopic stigmata are by themselves a major predictor for the subsequent clinical course and determine the need for endoscopic intervention. They are thus also used in the post-endoscopic riskassessment systems. RISK ASSESSMENT AFTER ENDOSCOPY The combination of clinical, laboratory, and endoscopy data are used in post-endoscopy risk scores. These scores are more suitable for the prediction of rebleeding and mortality, in comparison with the pre-endoscopy scores. They have the advantage of using information provided by endoscopy but thus the disadvantage that can be determined only with delay, that is, after endoscopy. This category includes the Baylor 10 and Cedars-Sinai Index scores 11 and the more widely used post-endoscopy Rockall score. 3 All these scoring systems are of earlier dates and were developed at a time when combination endoscopic therapy, continuous profound acid suppression, and other measures such as transarterial embolization were not in routine use, and Helicobacter pylori associated ulcer bleeding was far more common than nonsteroidal antiinflammatory drug associated ulcer bleeding. Furthermore, the Baylor and Cedars-Sinai Index scores were based only on single cohorts of respectively 80 and 500 patients with upper GI bleeding, whereas the Rockall score was based on analysis of over 5000 patients. A comparative trial including 343 patients showed that the post-endoscopy, or complete Rockall score, outperformed the Baylor and Cedar Sinai systems in terms of identifying patients at low risk. 12 NEW DEVELOPMENTS Nearly all of these risk assessment tools originate from the U.K., the United States, and Canada. An excellent study in this issue of Gastrointestinal Endoscopy adds to that tradition. Based on an analysis of approximately 60,000 patients presenting over a 4-year period in 187 U.S. hospitals, Saltzman et al 13 developed a pre-endoscopy mortality risk score based on age, systolic blood pressure, mental status, serum albumin level, and international normalized ratio (INR). The resulting AIMS65 system was first derived from a group of 29,222 patients and was then validated on a further 32,504 patients. Extensive data analysis yielded an excellent and remarkably simple predictor based on only 5 parameters with equal weight, leading to a 0 to 5 score. Age and systolic blood pressure also served www.giejournal.org Volume 74, No. 6 : 2011 GASTROINTESTINAL ENDOSCOPY 1227
Editorial Kuipers in most of the previous prediction systems. The other 3 markers, mental status, serum albumin level, and INR, are new. Previous data on INR as a predictor of mortality in upper GI bleeding patients are conflicting. 1 However, a recent systematic review concluded that INR 1.5 is a significant predictor for mortality, but not for rebleeding, supporting the assumption that the INR serves as a marker for hepatic and cardiovascular comorbidity, in particular. 14 Together, the markers served as a good predictor for mortality and for length of hospital stay and costs. In the validation cohort, mortality was 0.3% for patients presenting with a score of 0 and 1.2% for those with a score of 1, compared with 5.3% in those with a score of 2, increasing to 10.3%, 16.5%, and 24.5%, respectively, in those with scores of 3, 4, and 5. 13 This good discerning power with respect to mortality risk identified a large proportion of patients at low risk, being 60.0% of the validation cohort. Additional advantages that may explain its performance are that the system is based on a very large set of data from a far larger number of patients than any of the previous systems validated with an even larger number of patients and that these data were derived from a range of hospitals of varying size, both rural and urban, both teaching and non-teaching. A further advantage is that the data were obtained in recent years (2004-2007), at a time when endoscopic therapy and continuous profound acid suppressive therapy had been widely introduced into clinical practice. This is reflected by the very low overall mortality rate (3.2% and 2.7% for derivation and validation cohorts, respectively). Some potential limitations have to be assessed with further use of the AIMS65 system. First, the system contains one semiqualitative marker that depends on clinical judgment, that is, mental status. The authors defined altered mental status as a Glasgow coma scale score of 14 or a designation of disorientation, lethargy, stupor, or coma by a physician. 13 This judgment has to be made in the acute setting, often on severely ill patients in shock. This seems feasible, yet the value and ease of use of this marker needs prospective confirmation. Furthermore, although the authors point out that their low mortality is in line with findings of several other recent studies, other data such as the recent U.K. audit reported a considerably higher mortality of 10%, together with an 11% instead of 1.7% incidence of variceal bleeding. 15 The AIMS65 system has to be tested under such circumstances and preferentially compared with the Blatchford and full Rockall systems, in particular. Its ease of use and predictive capacity may help clinicians to follow consensus recommendations to use risk assessment scores for targeted management. A recent, large, U.K. survey including 6750 patients with upper GI bleeding showed that this is, at present, not routinely done everywhere. 15 In conclusion, Saltzman et al 13 in this issue of the Journal present a carefully derived and validated new prediction rule for mortality of patients with upper GI bleeding. Their system is of major interest, given its simplicity and its capacity as a pre-endoscopy predictor of mortality. As such, the system asks to be widely applied in other populations and tested against older systems. Confirmation of its performance may help clinicians to implement the routine use of risk assessment for clinical management of upper GI bleeding. DISCLOSURE The author disclosed no financial relationships relevant to this publication. Ernst J. Kuipers, MD Departments of Gastroenterology & Hepatology and Internal Medicine Erasmus MC University Medical Center Rotterdam, The Netherlands Abbreviation: INR, international normalized ratio. REFERENCES 1. Barkun AN, Bardou M, Kuipers EJ, et al. International consensus recommendations on the management of patients with nonvariceal upper gastrointestinal bleeding. Ann Intern Med 2010;152:101-13. 2. Sung JJ, Barkun A, Kuipers EJ, et al. Intravenous esomeprazole for prevention of recurrent peptic ulcer bleeding: a randomized trial. Ann Intern Med 2009;150:455-64. 3. Rockall TA, Logan RF, Devlin HB, et al. Risk assessment after acute upper gastrointestinal haemorrhage. Gut 1996;38:316-21. 4. Stanley AJ, Ashley D, Dalton HR, et al. Outpatient management of patients with low-risk upper-gastrointestinal haemorrhage: multicentre validation and prospective evaluation. Lancet 2009;373:42-7. 5. Romagnuolo J, Barkun AN, Enns R, et al. Simple clinical predictors may obviate urgent endoscopy in selected patients with non-variceal upper gastrointestinal tract bleeding. Arch Intern Med 2007;167:265-70. 6. Pang HS, Ching JYL, Lau JYW, et al. Comparing the Blatchford and preendoscopic Rockall score in predicting the need for endoscopic therapy in patients with upper GI hemorrhage. Gastrointest Endosc 2010;71: 1134-40. 7. Chen IC, Hung MS, Chiu TF, et al. Risk scoring systems to predict need for clinical intervention for patients with nonvariceal upper gastrointestinal tract bleeding. Am J Emerg Med 2007;25:774-9. 8. Masaoka T, Suzuki H, Hori S, et al. Blatchford scoring system is a useful scoring system for detecting patients with upper gastrointestinal bleeding who do not need endoscopic intervention. J Gastroenterol 2007;22:1404-7. 9. Gralnek IM, Dulai GS. Incremental value of upper endoscopy for triage of patients with acute non-variceal upper GI hemorrhage. Gastrointest Endosc 2004;60:9-14. 10. Saeed ZA, Winchester CB, Michaletz PA, et al. A scoring system to predict rebleeding after endoscopic therapy of non-variceal upper gastrointestinal hemorrhage. Am J Gastroenterol 1993;88:1842-9. 11. Hay JA, Lyubashevsky E, Elashoff J, et al. Upper gastrointestinal hemorrhage clinical guideline: determining the optimal hospital length of stay. Am J Med 1996;100:313-22. 1228 GASTROINTESTINAL ENDOSCOPY Volume 74, No. 6 : 2011 www.giejournal.org
Kuipers Editorial 12. Camellini L, Merighi A, Pagnini C, et al. Comparison of three different risk scoring systems in non-variceal upper gastrointestinal bleeding. Dig Liver Dis 2004;36:271-7. 13. Saltzman JR, Tabak YP, Hyett BH, et al. A simple risk score accurately predicts in-hospital mortality, length of stay and cost in acute upper GI bleeding. Gastrointest Endosc 2011;74: 1215-24. 14. Shingina A, Barkun AN, Razzaghi A, et al. Systematic review: the presenting international normalised ratio (INR) as a predictor of outcome in patients with upper nonvariceal gastrointestinal bleeding. Aliment Pharmacol Ther 2011;33:1010-8. 15. Hearnshaw SA, Logan RF, Lowe D, et al. Use of endoscopy for management of acute upper gastrointestinal bleeding in the UK: results of a nationwide audit. Gut 2010;59:1022-9. GIE on Facebook GIE now has a Facebook page. Fans will receive news, updates, and links to author interviews, podcasts, articles, and tables of contents. Search on Facebook for GIE: Gastrointestinal Endoscopy and become a fan. www.giejournal.org Volume 74, No. 6 : 2011 GASTROINTESTINAL ENDOSCOPY 1229