A Comparative Analysis of Risk Scoring Systems in Predicting Clinical Outcomes in Upper Gastrointestinal Bleed

Background Upper gastrointestinal bleed (UGIB) is a life-threatening condition that presents as hematemesis (fresh blood), coffee-ground vomiting, or melena. Multiple scoring systems are developed to predict different clinical outcomes, which are important to managing UGIB and are essential to determining low and high-risk patients. The study aimed to compare the sensitivity and specificity of risk scoring systems and their optimum cut-off values in the assessment of UGIB. Methods The prospective cross-sectional study included patients (N = 81) with acute UGIB. Four different proposed scores [Glasgow-Blatchford score (GBS), AIMS65, pre-endoscopic Rockall, and full Rockall scoring system] were used for evaluating patients with UGIB. The optimum cut-off values of these risk scores were used to predict the clinical outcomes. Results The AIMS65 score [Area Under the Receiver Operating Characteristic curve (AUROC): 0.91, cut-off: >1, sensitivity: 100%, specificity: 76.62%] and pre-Rockall were similar (AUROC: 0.91, cut-off: >0, sensitivity: 100%, specificity: 93.51%) at predicting mortality. The GBS (cut-off: >9, AUROC: 0.79, sensitivity: 69.23, specificity: 87.50) and AIMS65 scores (cut-off: >0, AUROC: 0.67, sensitivity: 72.31, specificity: 62.5) were good predictors of need for ICU care. Conclusion GBS was superior in predicting categorization into high risk and low risk, and endoscopic intervention, blood transfusion, and intensive care unit (ICU) care in UGIB patients. Pre-Rockall score and AIMS65 score were similar in predicting the mortality rate in UGIB.


Introduction
Upper gastrointestinal bleed (UGIB), defined as bleeding occurring from the gastrointestinal tract, presents as hematemesis (fresh blood), coffee-ground vomiting, or melena (black stools) [1]. The presentation of patients varies widely with an insignificant bleed or may have significant bleed which may lead to death. The estimated number of UGIB is 48-165 per 100,000 adults/year with a mortality rate of 6-10% overall. However, the number varies in different regions of the world [2,3]. Peptic ulcer disease and gastrointestinal variceal bleeding secondary to portal hypertension are the two leading causes of UGIB [3,4].
Multiple scoring systems are developed to predict different clinical outcomes in UGIB patients. It is increasingly noticed that early identification of high-risk patients is an essential part of management, as it directly recommends suitable patient care, and also the timing of endoscopy. With multiple risk scoring systems being in place for UGIB, there are ones that can be calculated without the endoscopy as a component [5]. An ideal risk score is one that is easy to calculate, one with high sensitivity for determining outcomes, and can be calculated early during a presentation with UGIB. However, the accuracy and generalisability of these scores and the optimum cut-offs to distinguish low-risk from high-risk patients, remain unclear [2,6].
Glasgow Blatchford score (GBS), Rockall score (RS), and the AIMS65 score are the typically used risk scoring systems for UGIB [7,8]. Our study intended to frame optimum threshold values of the scoring systems to assess predetermined composite endpoints, which would help in the quick assessment and risk stratification in patients with UGIB [5]. The study aims to compare the sensitivity and specificity of risk scoring systems and their optimum cut-off values in the assessment of UGIB.

Materials And Methods
The prospective cross-sectional study was conducted in acute UGIB patients attending a tertiary care hospital from October 2018 to July 2020. After Institutional Ethics Committee approval from Ramaiah Medical College (vide letter number EC/PG-27/2018) and written consent, all consecutive patients attending the hospital with acute UGIB (defined by hematemesis, coffee-ground vomiting, or melena, within 7 days of onset) were included [9,10]. Patients who did not undergo upper gastrointestinal endoscopy or attended the hospital after 7 days of onset of UGIB were excluded. Subjects (N = 81) included were calculated based on the reported prevalence of gastrointestinal bleeding of 4.7 [9], and 7% absolute allowable error at 95% confidence level and 80% power.

Data collection
Patient demographic including clinical history recorded. Physical, systemic, and routine blood investigations were performed. Upper gastrointestinal endoscopy was done in all the UGIB patients, and findings were noted. If any growth is noted, then a biopsy was taken and sent for histopathological analysis.

Scores in UGIB
Four different proposed scores (GBS, AIMS65, pre-endoscopic Rockall, and full Rockall scoring system) were used for evaluating UGIB. GBS includes clinical (history, comorbidities, and systolic blood pressure) along with laboratory variables (blood urea, hemoglobin), which have a score ranging from 0-23 [7,11,12]. A detailed description of GBS can be found in Table 1. AIMS65 includes five clinical and laboratory parameters (serum albumin <3g/dL, INR >1.5, altered mental status [Glasgow Coma Scale (GCS) <14/15, stupor, coma] systolic blood pressure <90 mmHg, age >65 years), each corresponding to 1 point [8,11]. Admission Rockall (pre-endoscopic Rockall) scoring system ranging from 0-7 score has three variables i.e., age (<60 years, 60-79 years or ≥80 years), comorbidities (no major comorbidity, any comorbidity except renal failure, liver failure, and/or disseminated malignancy, renal failure, liver failure, and/or disseminated malignancy) and shock (no shock, tachycardia or hypotension) [11,13]. Table 2 depicts, that the full Rockall scoring system has a score ranging from 0-11 and has variables including age, comorbidity, shock, diagnosis, and major stigmata of most recent hemorrhage [11,13].    The optimum cut-off values of these risk scoring systems were studied for the following clinical outcomes: requirement of blood transfusion; endoscopic treatment, interventional radiology, or surgery; in-hospital death or duration of in-hospital stay; and mortality.
Statistical analysis R 4.0.3 (R Foundation for Statistical Computing, Vienna, Austria) was used to analyze the data. QQ plot/Shapiro-Wilk's test was used to check the normality of variables. Continuous variables are presented as mean ± SD form and categorical variables as a frequency table. GBS Score, AIMS65 score, pre-Rockall, and full Rockall scores were further analyzed using the receiver operating characteristic for determining optimal cut-off points to calculate sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). The comparison of >3 groups was done with the Kruskal-Wallies test. p<0.05 represents the statistical significance.

Results
Patients (N=81) with UGIB were studied.  No significant difference was observed in the distribution of AIMS65, GBS, pre-Rockall, and full-Rockall over the duration of hospitalization (p>0.05; Table 4). Even though the GBS score was higher, it did not show any association with the duration of hospital stay. Although patients with a lower score (<8) stayed in the hospital for a longer duration, it is for other reasons such as electrolyte imbalances and their other comorbidities, i.e., chronic kidney disease on maintenance hemodialysis, myocardial infarction, and sepsis.   Table 5). AIMS65 (p = 0.01) and pre-Rockall (p = 0.02) were better at predicting mortality ( Table 5). The GBS score was better at predicting requirement for endoscopic intervention (p = 0.049), requirement for ICU (p < 0.001), and blood transfusion (p = 0.00) when compared with other scoring systems ( Table 5). AIMS65 was also better at predicting the need for ICU (p = 0.03), however, quite low compared to GBS.

Discussion
Various prospective studies have proved the effectiveness of these risk scoring systems in predicting prolonged hospitalization, the requirement for blood transfusion, endoscopic interventions, and mortality [14,15]. International consensus guidelines also stated that early stratification of low from high-risk patients is essential for the management of UGIB with timely interventions to decrease the morbidity as well as mortality burden [16]. However, the available scores have a few limitations; hence comparison between older scores, their simplified versions, and newer risk scores is necessary to direct evidence-based clinical decisions [16,17]. Hence we compared different risk scoring systems (GBS score, AIMS65 score, and the preendoscopy and full Rockall scores) to predict certain clinical outcomes.
Our study had the majority of the subjects (42%) in the 51-65 years age group with a male to female ratio of 3.05:1. In a study done by Chandnani et al., the mean age of the patients was 43.5 years with male predominance (69%) [18]. Thandassery et al. and Nagaraja et al. also reported mean age of 46.16 years in the patients with male predominance [19,10].
In our study, most patients had liver disease (43.2%) in UGIB and the least common was malignancy (3.7%). Similarly, in Chandnani et al., the study showed liver disease and malignancy in 43.3% and 2.3% of the patients, respectively [19]. The incidence of non-variceal bleed is more frequent, secondary to peptic ulcer disease (41.4%) when compared to variceal bleed (27.9%) [20]. One Chinese study also found that nonvariceal UGIB (84.4%) was more commonly observed in patients than variceal UGIB (15.6%) [21]. While our study had comparable variceal and non-variceal bleed cases.
AIMS65 score was better at determining the hospital stay when compared to other risk scoring systems, however not to a statistically significant degree. In our study, patients stayed around 1-15 days in the hospital. However, underlying disease conditions (dyselectrolytemia, regular hemodialysis in chronic kidney disease (CKD) patients in volume overload, sepsis secondary to pneumonia, and urinary tract infections) were the reasons behind longer hospital stay rather than the UGIB.
We found that both GBS and AIMS65 scores are able to predict ICU requirements in UGIB patients. GBS was the best at predicting the requirement for ICU care (cut-off >9, AUROC 0.793, p<0.001), followed by AIMS65 (cut-off >0, AUROC 0.67, p=0.03). Both scores had a good PPV of 95.7 and 88.7% for GBS and AIMS65, respectively. This helps the clinicians to categorize high from low-risk patients and thus could determine the patients requiring ICU admission. Pines et al in their study also stated both GBS and AIMS65 were able to determine ICU requirements, however, AIMS65 was relatively more accurate compared to GBS [12].
GBS had a higher ability to predict the endoscopic intervention requirement in UGIB patients when compared to other risk scores [7]. Similarly, we also found that GBS was better at predicting the requirement for endoscopic intervention (AUROC 0.618, p 0.06). Chandnani et al. in western India and Tham et al in Glasgow, UK also found that GBS is superior to other risk scores to predict blood transfusion, endoscopic, and radiological or surgical interventions in UGIB [18,13]. A similar study by Martínez-Cara et al. showed that AIMS65 and GBS were quite identical (0.62 vs. 0.62) in predicting endoscopic intervention and GBS alone was better at predicting blood transfusion requirement [22]. In our study, AIMS65 score cut-off value ≥2 (AUROC .553, p = 0.408, statistically not significant) predicts the blood transfusion requirement. Thandassery et al. in who also had a cut-off value of ≥2 better predicted the blood transfusion requirement (AUROC 0.59) [19]. Our study found that GBS is superior to other risk scores at predicting the requirement for blood transfusion with AUROC being 0.713 with a high sensitivity of 97.62% with the cut-off value of >6, which is also statistically significant (p=0.001).
Our study assessed four pre-and post-endoscopy scores that showed the most promising for clinical use. Hence, we can recognize the optimal way to risk assess UGIB early after the presentation, following endoscopic diagnosis, and following treatment. Comparison of scoring systems might provide invaluable information to the clinicians to keenly identify those who are at high risk of endoscopic intervention and to direct such patients immediately to the intervention. The study has a few limitations that need to be regarded. First, this is a single-center study. Hence, a multi-centric design involving more consecutive patients attending to the hospital needs to be included to validate the current results.

Conclusions
GBS was superior in predicting categorization into high risk and low risk, and endoscopic intervention, blood transfusion, and ICU care in UGIB patients. Pre-Rockall score and AIMS65 scores were quite comparable in predicting mortality in UGIB patients. GBS and AIMS65 scores help in predicting the requirement for ICU care; AIMS65 being a simple score will also reduce the cost burden of unnecessary ICU admissions.

Additional Information Disclosures
Human subjects: Consent was obtained or waived by all participants in this study. Ramaiah Medical College Institutional Ethics Committee issued approval EC/PG-27/2018. Animal subjects: All authors have confirmed that this study did not involve animal subjects or tissue. Conflicts of interest: In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work. Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.