|Year : 2021 | Volume
| Issue : 2 | Page : 55-61
Comparison of complete rockall score, Glasgow–Blatchford score, and AIMS 65 score for predicting in-hospital mortality in patients presenting with upper gastrointestinal hemorrhage at a Tertiary Care Hospital in Southern India
Anand Totagi, Usha Srinivas, Piramanayagam Paramasivan, Sankaranarayanan Krishnan, Kallipatti Ramasamy Palaniswamy, AT Mohan, Sarojini Ashok Parameswaran, Ubal Dhus, Muthuswamy Hariharan, MS Revathy, Natarajan Murugan, Karunakaran Premkumar, Seshadri Venkatesh, Preethi Mahalingam, Santhosh Kumar
Department of Medical Gastroenterology, Apollo Hospitals, Chennai, Tamil Nadu, India
|Date of Submission||03-Sep-2020|
|Date of Decision||04-Oct-2020|
|Date of Acceptance||05-Oct-2020|
|Date of Web Publication||23-Mar-2021|
Department of Medical Gastroenterology, Room 106, Apollo Hospitals, Greams Lane (Off: Greams Road), Chennai - 600 006, Tamil Nadu
Source of Support: None, Conflict of Interest: None
Background: Multiple risk assessment scores are available to triage and stratify patients presenting with upper gastrointestinal (GI) hemorrhage. This study was conducted to compare the accuracy of complete Rockall score (CRS), Glasgow–Blatchford score (GBS), and AIMS65 score (AIMS65-albumin, INR >1.5, impaired mental status, systolic pressure <90 mm Hg, and age >65) in predicting in-hospital mortality in patients presenting with upper GI hemorrhage to the gastroenterology department at a tertiary care hospital in Southern India. The secondary objectives were to compare these three scores in predicting the need for blood transfusion, any intervention (endoscopic or radiological or surgical intervention), or rebleeding in patients presenting with upper GI hemorrhage. Materials and Methods: This was a retrospective analysis of prospectively recorded data which included 207 patients with acute upper GI hemorrhage admitted at a tertiary care hospital at Chennai over 2 years. Demographic, clinical, laboratory, and endoscopic parameters were recorded. CRS, GBS, and AIMS65 scores were calculated. Data regarding in-hospital mortality, need for blood transfusion, endoscopic intervention, radiological intervention, surgical intervention, and rebleeding were collected. Area under receiver operating characteristic curve (AUROC) was compared between the three scores in predicting in-hospital mortality, need for blood transfusion, intervention, and rebleeding. Results: AIMS 65 score >3 (AUROC 0.92) was a better predictor of in-hospital mortality than GBS (AUROC 0.77) and CRS (AUROC 0.69). AIMS65 was a better predictor of rebleeding (AUROC 0.804) than GBS (AUROC 0.676) or CRS (AUROC 0.623). GBS was a better predictor for need of blood transfusion (AUROC 0.785) than AIMS65 (AUROC 0.691) or CRS (AUROC 0.629). Conclusion: AIMS 65 score (>3) was a better predictor of in-hospital mortality than GBS or CRS in patients presenting with acute upper GI hemorrhage due to either variceal and nonvariceal etiology. AIMS 65 was also a better predictor of risk of rebleeding. GBS was a better predictor of need for blood transfusion and need for intervention.
Keywords: Acute upper gastrointestinal hemorrhage, AIMS65, Glasgow–Blatchford score, in-hospital mortality, Rockall score
|How to cite this article:|
Totagi A, Srinivas U, Paramasivan P, Krishnan S, Palaniswamy KR, Mohan A T, Parameswaran SA, Dhus U, Hariharan M, Revathy M S, Murugan N, Premkumar K, Venkatesh S, Mahalingam P, Kumar S. Comparison of complete rockall score, Glasgow–Blatchford score, and AIMS 65 score for predicting in-hospital mortality in patients presenting with upper gastrointestinal hemorrhage at a Tertiary Care Hospital in Southern India. Gastroenterol Hepatol Endosc Pract 2021;1:55-61
|How to cite this URL:|
Totagi A, Srinivas U, Paramasivan P, Krishnan S, Palaniswamy KR, Mohan A T, Parameswaran SA, Dhus U, Hariharan M, Revathy M S, Murugan N, Premkumar K, Venkatesh S, Mahalingam P, Kumar S. Comparison of complete rockall score, Glasgow–Blatchford score, and AIMS 65 score for predicting in-hospital mortality in patients presenting with upper gastrointestinal hemorrhage at a Tertiary Care Hospital in Southern India. Gastroenterol Hepatol Endosc Pract [serial online] 2021 [cited 2021 Apr 22];1:55-61. Available from: http://www.ghepjournal.com/text.asp?2021/1/2/55/311734
| Introduction|| |
Upper gastrointestinal (GI) hemorrhage is a major cause of hospital admission and mortality throughout the world. Optimal treatment of these patients is expensive due to the need for admission, hospital resources for intensive care unit monitoring, blood transfusion, endoscopy, and radiological or surgical intervention. This has necessitated the development of risk assessment scores which stratify patients based on clinical outcomes.
In 1996, complete Rockall score (CRS) was developed and validated to predict risk of mortality. There is a preendoscopy (clinical) and postendoscopy (complete) Rockall score. CRS has shown to be predictor of in-hospital mortality. In 2000, Glasgow–Blatchford score (GBS) was developed to identify patients who will require any intervention (blood transfusions or endoscopic or surgical intervention apart from rebleeding and mortality. Subsequently, in 2011, AIMS65 score was developed with an intent to predict in-hospital mortality. This score included serum albumin <3 g/dl, INR >1.5 impaired mental status, systolic blood pressure <90 mm Hg, and age >65 years.
These scoring systems have been used to identify low risk patients who can be managed on outpatient basis (e, g GBS <1 have very low risk of mortality and need for endoscopic intervention). However, this is likely to be of limited clinical utility in tertiary care centers in India, as <1% of patients have GBS scores <1 and qualify for the management on outpatient basis. However, there is a definite need for identifying sicker cohort of patients at admission, based on risk of mortality, which will help triage them to intensive care and interventions. Multiple studies comparing the utility of these scoring systems in predicting mortality and clinical outcomes have been published.,,, These studies have come to varying conclusions regarding the optimum risk-stratifying score. These differences may be related to the proportion of variceal or nonvariceal cause for hemorrhage, comorbidities and interventions, and the hospital setting (community vs. tertiary care hospital).
Rockall score, GBS, and AIMS 65 have been developed in a study cohort which is different from the cohort of patients presenting at tertiary care center in India. Rockall score was derived from a study cohort, in which majority of patients were >60 years, only 4% were due to variceal hemorrhage, and had ~14% mortality. AIMS 65 score was developed from a study cohort, in which the mean age was 75 years, only 1.7% had variceal hemorrhage, and had ~3% mortality. Recent studies regarding acute upper GI hemorrhage from tertiary care centers in India show an younger population (mean age ~45 years), a higher prevalence of variceal hemorrhage (>40%), and higher mortality (10%–15%).,,, Thus, utility of these scores in triaging patients presenting to tertiary care centers in India needs to be evaluated.
This study was conducted with an objective to assess the utility of complete Rockall score (CRS), Glasgow–Blatchford score, and AIMS65 score in risk-stratifying patients presenting to a tertiary care center in Southern India.
| Materials and Methods|| |
Setting and participants
This study was a retrospective analysis of prospectively recorded data of all patients who presented with acute upper gastrointestinal hemorrhage (UGIH) to a tertiary care center at Chennai from May 2012 to March 2014. Case record of consecutive patients was obtained and data were retrieved for analysis.
Acute UGIH was defined as patients who presented with hematemesis, “coffee ground” emesis, the return of red blood via a nasogastric tube, and/or melena with or without hemodynamic compromise. In-patients who developed upper GI bleed during admission were excluded.
Clinical history, comorbidities, vitals, systemic examination findings, routine blood tests (complete hemogram, liver function tests, renal function tests, serum electrolytes, and coagulation profile), and ultrasound abdomen findings of all patients admitted with acute upper GI hemorrhage were noted. Patients were resuscitated with crystalloids in the initial period followed by packed cell transfusion if blood hemoglobin was <8 gm% or packed cell volume was <24%. Patients suspected of having nonvariceal UGIH received pantoprazole (80 mg intravenous [IV] bolus followed by 8 mg/h infusion). Patients suspected of having variceal UGIH received antibiotics (IV ceftriaxone 2 gram once daily) and pharmacological treatment (terlipressin 2 mg iv followed by 2 mg every 4 hourly or somatostatin 250 μg IV bolus followed by infusion of 250 μg/hour or octreotide 50 μg IV bolus followed by infusion of 50 μg/hour). The choice of pharmacological treatment was left to the discretion of treating physician.
Esophagogastroduodenoscopy was done within 12 h of admission (Olympus GIF-Q150/GIF-160/GIF-H180/GIF-HQ190, Olympus Medical systems, Tokyo, Japan, was used). If varices were present, they were graded (<5 mm small, >5 mm large) and red color signs were noted. Endoscopic variceal ligation was done for varices; sclerotherapy was used if ligation was technically difficult. In case of gastric or duodenal ulcer, dual therapy consisting of a mechanical method (hemoclips) along with adrenaline (1; 10000) injection was done. Argon plasma coagulation was used for gastric antral vascular ectasia. In patients presenting with melena, if esophagogastroduodenoscopy was normal, colonoscopy was done. CRS, GBS, and AIMS65 scores were calculated based on the demographic data, clinical data, laboratory parameters, and endoscopic findings.
Primary outcome measure was in-hospital mortality. Secondary outcome measures were rate of rebleeding, need for any intervention (endoscopic, radiological, or surgical), and need for blood transfusion.
Proportions were compared using Pearson's Chi-square test and Fisher's exact test. Medians were compared using the Mann–Whitney U-test and the Kruskal–Wallis equality of populations rank test. To assess the predictive ability of the three scores, receiver operating characteristic (ROC) curves were plotted for the primary and secondary outcome measures. Area under ROC curves (AUROC) was compared between the three scores using Delong method. The score thresholds of each of these three scores with best Youden index in predicting primary and secondary outcomes measures were considered optimal. Statistical software SPSS (Version 14.0, SPSS Inc, Chicago, IL, USA) was used for statistical analysis.
Informed consent was obtained from all patients. The study protocol was approved by the Institutional Ethics Committee of Apollo Hospitals, Chennai.
| Results|| |
Two hundred and seven consecutive patients admitted with upper GI hemorrhage were included in the study [Figure 1]. Of 207 patients included in the study, 109 had nonvariceal UGIH and 98 variceal UGIH. Baseline characteristics of the study population are described in [Table 1].
Primary and secondary clinical outcomes are described in [Table 2].
Twenty-five patients (12.1%) died in hospital. Mortality was significantly higher for variceal UGIH (22.4%) as compared to nonvariceal UGIH (2.8%).
Need for blood transfusion
Ninety-five patients (46%) required blood transfusion. Blood transfusion requirement was comparable between variceal and nonvariceal group (52% vs. 40%).
Eighty-four patients (40.5%) required endoscopic intervention. Patients with variceal hemorrhage required endoscopic intervention more often than nonvariceal hemorrhage group (68% vs. 15%). One patient underwent radiological intervention (coiling of gastroduodenal artery pseudoaneurysm). Seven patients (6 in the nonvariceal group and 1 patient in the variceal group) underwent surgical intervention.
Overall, 32 (15.4%) patients had rebleeding. Rebleeding rate was higher in the variceal group as compared to the nonvariceal group (25% vs. 6.4%).
Comparison of the scores in predicting in-hospital mortality
Mortality occurred if CRS ≥2, GBS >12, and AIMS 65 >2. The mean CRS scores among non-survivors and survivors were 4.96 ± 1.89 and 3.65 ± 1.82, respectively (P < 0.05). The mean GBS scores among non-survivors and survivors were 16 ± 3.5 and 11.69 ± 4.53, respectively (P < 0.001). The mean AIMS65 scores among non-survivors and survivors were 3.92 ± 0.81 and 1.52 ± 1.29, respectively (P < 0.001). As the score increased, the risk of mortality increased.
Ability of CRS, GBS, and AIMS 65 scores to predict in-hospital mortality was compared using AUROC [Figure 2]. AUROC was highest for AIMS65 0.926 ([95% confidence interval (CI) [0.874–0.978]), followed by GBS 0.777 (95% CI [0.682–0.872]) and CRS 0.690 (95% CI; [0.579–0.800]). As shown in [Table 3], in the overall study population, AIMS65 score was better than GBS and CRS in predicting in-hospital mortality (P = 0.00 and 0.01 respectively). In the variceal hemorrhage subgroup, AIMS65 performed better than GBS and CRS in predicting in-hospital mortality (P = 0.00 and 0.01, respectively). In the non-variceal hemorrhage subgroup, AIMS65 was better than CRS and was equivalent to GBS in predicting in-hospital mortality.
|Figure 2: Comparison of ROC curves of the three scores in predicting in-hospital mortality. AIMS65-Albumin, INR >1.5, Impaired mental status, systolic pressure <90mm Hg, age >65; GBS: Glasgow–Blatchford score, TCRC: Complete Rockall score UGIH: Upper gastrointestinal hemorrhage, ROC: Receiver operating characteristic|
Click here to view
Based on ROC, the optimal cutoff for AIMS 65 score for predicting mortality was 3 with sensitivity of 90% and specificity of 84%. Mortality for patients with AIMS65 score >3 was 60% as compared to 2.3% for patients with AIMS65 score <3. Optimal cutoff for GBS score for predicting mortality was 13 with sensitivity of 82% and specificity of 48%. Mortality for patients with GBS score >13 was 23.3% as compared to 3.3% for patients with GBS score <13. Optimal cutoff for CRS score for predicting mortality was 5 with sensitivity of 57% and specificity of 77%. Mortality for patients with CRS score >5 was 28.5% as compared to 7.8% for patients with CRS score <5.
Comparison of scores for predicting needing for blood transfusion, intervention, and rebleeding
Need for blood transfusion
In the overall study population, need for blood transfusion was better predicted by GBS (AUROC 0.78) as compared to CRS (AUROC 0.63) and AIMS65 (AUROC 0.69) (P = 0.00 and 0.007, respectively). In the non variceal hemorrhage subgroup, GBS scored better than CRS and AIMS65 (P = 0.00 and 0.001, respectively). In variceal hemorrhage subgroup, there was no significant difference between the three scores in predicting need for blood transfusion [Table 4].
|Table 4: Comparison of scores for predicting need for blood transfusion, intervention and rebleeding|
Click here to view
Need for intervention (endoscopic/radiological/surgery)
GBS was better than CRS and AIMS65 in predicting the need for intervention in the overall study population (P = 0.04 and 0.01, respectively). In the non variceal hemorrhage subgroup, GBS was better than AIMS65 (P = 0.01) and was comparable to CRS in predicting need for intervention. In the subgroup analysis of variceal hemorrhage, there was no significant difference between the three scores in predicting need for intervention [Table 4].
AIMS65 was a better predictor of rebleeding than CRS and GBS in the overall study population (P = 0.003 and 0.028, respectively) and in variceal hemorrhage subgroup. There was no significant difference in predicting the rebleeding rates among these scores in the non variceal hemorrhage subgroup [Table 4].
| Discussion|| |
In this study, AIMS65 was a better predictor of in-hospital mortality compared to CRS and GBS in patients presenting with acute upper GI hemorrhage to a tertiary care center in southern India. This cohort was different from the multicenter studies and original Rockall, GBS, and AIMS65 cohorts in which variceal hemorrhage was the etiology in <5% of patients.,, The study cohort was similar to those reported from tertiary care centers in Northern, Western, and Southern India, in which patients were younger and varices constitute >40% of etiology of upper GI hemorrhage.,,, In this study, in-hospital mortality was noted in 12%, need for blood transfusion in 46%, need for intervention in 40%, and rebleeding in 15%. These outcomes were similar to those reported by Rout et al. and Chandnani et al.,
The primary clinical outcome measure chosen in this study was in-hospital mortality since two of the three scores (CRS and AIMS65) used in this study were developed and validated with aim to predict in-hospital mortality. Four earlier studies from India had evaluated the utility of scoring systems in risk-stratifying patients presenting with upper GI hemorrhage.,,, These studies used different primary outcome measures and hence had come to different conclusions. Of these, three studies by Anchu et al., Sharma et al., and Rout et al. used composite endpoint (of mortality, need for intervention, and rebleed), whereas a study by Chandnani et al. used 30-day mortality as the primary outcome measure. The studies by Anchu et al. and Rout et al., which used composite endpoint of mortality and need for intervention, have concluded that GBS was better at overall risk stratification. However, AIMS65 score was not compared in the study by Anchu et al. The study by Sharma et al. compared the utility of Clinical and CRS in predicting composite end point and concluded CRS performed better. GBS and AIMS65 scores were not compared in this study. In the study by Chandnani et al., in which 30-day mortality was the primary outcome measure, CRS was found to be better than GBS, AIMS65, and Progetto Nazionale Emorragica Digestiva (PNED) score.
AIMS65 was found to be equivalent to CRS and better than GBS in predicting in-hospital mortality in the study by Robertson et al. Hyett et al., and Stanley et al. AIMS65 was noted to be similar to GBS and CRS in predicting in-hospital mortality and was better than GBS in predicting delayed mortality in the study by Martínez-Cara et al.
Earlier studies from India have reported that scoring systems performed better in subgroup of non variceal hemorrhage than variceal hemorrhage. In post hoc subgroup analysis of the present study, AIMS65 was better at predicting in-hospital mortality for both variceal and non-variceal hemorrhage than CRS or GBS. AIMS65 was noted to be better at predicting 6-week mortality than GBS, Child score, or Rockall score and equivalent to MELD-Na score in cirrhotic patients presenting with variceal bleed in the study by Wang et al. Similar to the present study, Alexandrino et al. noted that AIMS65 was a better predictor of in-hospital mortality than other scores in both variceal and non variceal hemorrhage subgroups. In another study by Robertson et al., AIMS65 performed better than liver risk scores (Child score and MELD) in predicting in-hospital mortality.
AIMS65 score >3 was identified as the cutoff with best sensitivity (90%) and specificity (84%) in predicting in-hospital mortality in patients with either variceal or non variceal hemorrhage. Overall, patients with AIMS65 score >3 had 60% mortality as compared to 2.3% for patients with AIMS score <3. Robertson et al. also had concluded that AIMS65 cutoff of 3 best differentiated high-risk groups from low-risk groups in terms of in-hospital mortality (37% vs. 4%). This cutoff needs to be validated in future studies from India.
GBS was a better predictor of need for blood transfusion and intervention than CRS and AIMS65, in the overall study population and in non variceal hemorrhage subgroup. There was no significant differences in predicting the need for transfusion or intervention between the three scoring systems in the variceal hemorrhage subgroup. GBS was uniformly noted to perform better than other scores in predicting need for blood transfusion in most studies reported in the literature.,,,
In this study, AIMS 65 was a better predictor of rebleed in the overall study population and in the variceal hemorrhage subgroup. There was no significant difference among the three scores in predicting rebleed in the non variceal hemorrhage subgroup. In contrast, GBS and CRS have been shown to predict rebleed better than AIMS65 in studies by Hyett et al. and Robertson et al. PNED was noted to be a better predictor of rebleed than AIMS65 or GBS or CRS in the study by Chandnani et al. In a few studies (Rout et al. and Stanley et al.), none of the scores were good at predicting rebleed.,
There are a few limitations of this study. First, this is a retrospective analysis of prospectively collected data from a tertiary care center. This needs to be prospectively studied in cohorts in multicenter studies. Second, the mortality in the patients with non variceal group was 2.75% as compared to 22% in variceal group. The mortality in each of these subgroups is similar to that reported in other studies.,, However, the factors predicting in-hospital mortality might have been skewed due to the relatively higher mortality in variceal group. Third, there was no follow-up of patients after discharge. This limits the comparison of this study with other studies from India which have reported 30-day or 42-day mortality. Fourth, multiple other scores such as PNED have been reported to be useful in risk-stratifying patients with upper GI hemorrhage, which have not been included in this study.
| Conclusion|| |
An ideal risk-stratifying score for patients presenting with upper GI hemorrhage should be easy to use, applicable at emergency and should be valid irrespective of etiology. AIMS65, a pre-endoscopy score, can be easily calculated at admission with three clinical and two lab parameters. It is useful in predicting in-hospital mortality in both variceal and nonvariceal etiology of upper GI hemorrhage. This study validates the use of AIMS65 as the preferred risk-stratifying score for predicting in-hospital mortality in patients presenting with acute upper GI hemorrhage to tertiary care centers in India.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Rockall TA, Logan RF, Devlin HB, Northfield TC. Risk assessment after acute upper gastrointestinal haemorrhage. Gut 1996;38:316-21.
Blatchford O, Murray WR, Blatchford M. A risk score to predict need for treatment for upper-gastrointestinal haemorrhage. Lancet 2000;356:1318-21.
Saltzman JR, Tabak YP, Hyett BH, Sun X, Travis AC, Johannes RS. 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.
Rout G, Sharma S, Gunjan D, Kedia S, Nayak B, Shalimar. Comparison of various prognostic scores in variceal and non-variceal upper gastrointestinal bleeding: A prospective cohort study. Indian J Gastroenterol 2019;38:158-66.
Chandnani S, Rathi P, Sonthalia N, Udgirkar S, Jain S, Contractor Q. et al
. Comparison of risk scores in upper gastrointestinal bleeding in western India: A prospective analysis. Indian J Gastroenterol 2019;38:117-127.
Anchu AC, Mohsina S, Sureshkumar S, Mahalakshmy T, Kate V. External validation of scoring systems in risk stratification of upper gastrointestinal bleeding. Indian J Gastroenterol 2017;36:105-12.
Stanley AJ, Laine L, Dalton HR, Ngu JH, Schultz M, Abazi R, et al
. Comparison of risk scoring systems for patients presenting with upper gastrointestinal bleeding: International multicentre prospective study. BMJ 2017;356:i6432.
Sharma V, Jeyaraman P, Rana SS, Gupta R, Malhotra S, Bhalla AB. Utility of clinical and complete Rockall score in Indian patients with upper gastrointestinal bleeding. Trop Gastroenterol 2016;37:276-82.
Robertson M, Majumdar A, Boyapati R, Chung W, Worland T, Terbah R, et al
. Risk stratification in acute upper GI bleeding: comparison of the AIMS65 score with the Glasgow-Blatchford and Rockall scoring systems. Gastrointest Endosc 2016;83:1151-60.
Hyett BH, Abougergi MS, Charpentier JP, Kumar NL, Brozovic S, Claggett BL, et al
. The AIMS65 score compared with the Glasgow-Blatchford score in predicting outcomes in upper GI bleeding. Gastrointest Endosc 2013;77:551-7.
Martínez-Cara JG, Jiménez-Rosales R, Úbeda-Muñoz M, de Hierro ML, de Teresa J, Redondo-Cerezo E. Comparison of AIMS65, Glasgow-Blatchford score, and Rockall score in a European series of patients with upper gastrointestinal bleeding: performance when predicting in-hospital and delayed mortality. United European Gastroenterol J 2016;4:371-9.
Wang F, Cui S, Wang F, Li F, Tang F, Zhang X, et al
. Different scoring systems to predict 6-week mortality in cirrhosis patients with acute variceal bleeding: a retrospective analysis of 202 patients. Scand J Gastroenterol 2018;53:885-90.
Alexandrino G, Carvalho R, Reis J. Comparison of the AIMS65 Score with other risk stratification scores in upper variceal and nonvariceal gastrointestinal bleeding. Gut Liver 2018;12:111-3.
Robertson M, Ng J, Abu Shawish W, Swaine A, Skardoon G, Huynh A, et al
. Risk stratification in acute variceal bleeding: Comparison of the AIMS65 score to established upper gastrointestinal bleeding and liver disease severity risk stratification scoring systems in predicting mortality and rebleeding.Dig Endosc 2020;32:761-8. doi: 10.1111/den.13577.
[Figure 1], [Figure 2]
[Table 1], [Table 2], [Table 3], [Table 4]