Best Evidence Topics
  • Send this BET as an Email
  • Make a Comment on this BET

Glasgow-Blatchford risk scoring in upper GI bleed

Three Part Question

In [patients with non-variceal GI bleed] is [Glasgow-Blatchford score better than Rockall score] in identifying [low risk patients]?

Clinical Scenario

A previously health 25 years old man present to the A&E department with a chief complain of vomiting a small amount of fresh blood. You consider doing a risk assessment for outpatient treatment in him, but wonder whether the use of Glasgow-Blatchford risk scoring system is sensitive in recognition the need for admission?

Search Strategy

Medline using the OVID interface 1950 - week 3 November 2009:
[Glasgow-Blatchford$.mp. or Blatchford$.mp.] AND [rockall$.mp.]
EMBASE 1980-week 49 2009:
[glasgow-blatchford$.mp. or Blatchford$.mp.] AND [rockall$.mp.]
Cochrane Library:
Search for the terms Glasgow-Blatchford / Blatchford / Rockall
Google scholar:
Search for Blatchford and Rockall, LIMIT to English language
References search

Search Outcome

7 papers were found on the Medline search, of which 3 were relevant to the three part question.
7 papers were found on the EMBASE search, 3 were relevant (all included in the Medline search).
1 further relevant paper was found from the references of papers used.
Google search revealed: 162 results, 1 abstract was relevant (only abstract was published).
5 papers were therefore available and are summarized in the table below:

Relevant Paper(s)

Author, date and country Patient group Study type (level of evidence) Outcomes Key results Study Weaknesses
Blatchford et. al
Two stage study. Initial stage not relevant to three part question. The second stage 197 consecutive adult patients with UGIB in three hospitals during 3 months, comparing GBS vs. IRS & CRS Prospective Internal validating trial(second stage of the study) Predicting need for treatment (Area under ROC curve)Glasgow Blatchford score=0.92 (95%CI= 0.88-0.95); Initial (clinical or pre endoscopic) Rockall score=0.71 (95%CI= 0.64-0.78); Initial (clinical or pre endoscopic) Rockall score=0.71 (95%CI= 0.64-0.78)Done by the inventor of the GBS Small number of patients Basic demographic information and number patients who had endoscopy not mentioned
Stanley et. al.
Two phase study, (second phase, not relevant to three part question). First phase, 676 consecutive patients attending with UGIB from 4 hospitals. Compared Glasgow Blatchford score vs. Initial (clinical or pre endoscopic) Rockall score vs. Complete (post endoscopic) Rockall scoreMainly prospective, but one hospital (18% patients was retrospectivePredicting need for treatment:Glasgow Blatchford score=0.92 (95%CI= 0.90-0.94); Initial (clinical or pre endoscopic) Rockall score=0.72 (95%CI= 0.68-0.76)Ambiguous design in phase one (retrospective & prospective, and different time frames in different hospitals)
Predicting need for treatment:Glasgow Blatchford score=0.90 (95%CI= 0.88-0.93); Initial (clinical or pre endoscopic) Rockall score=0.81 (95%CI= 0.77-0.84)
Chen et. al.
354 adult patients admitted with non variceal UGIB who had endoscopy during 7 months period in one hospitalRetrospective (data collection by blinded research assistant)Need for intervention (sensitivity)Glasgow Blatchford score=99.6% (95%CI=97.9-99.9); Initial (clinical or pre endoscopic) Rockall score= 90.2% (95%CI= 85.9-93.4); Complete (post endoscopic) Rockall score=91.1% (95%CI= 86.8-94)Retrospective Short period of follow-up
Need for intervention (specificity)Glasgow Blatchford score= 25% (95%CI= 17.8-13.9); Initial (clinical or pre endoscopic) Rockall score= 38% (95%CI= 29.4-47.4); Complete (post endoscopic) Rockall score= 77.8% (95%CI= 69.1-84.6)
Need for intervention (Negative predicting value)Glasgow Blatchford score=96.4% (95%CI= 82.3-99.4); Initial (clinical or pre endoscopic) Rockall score=63.1% (95%CI= 50.9-73.8); Complete (post endoscopic) Rockall score= 79.2% (95%CI= 70.6-85.9
Gralnek Dulai
175 consecutive adult patients admitted with non-variceal upper GI bleed, during 2 yearsHistorical Cohort Retrospective studyYield of identifying low risk casesGlasgow Blatchford score=8%; Initial (clinical or pre endoscopic) Rockall score=12%; Complete (post endoscopic) Rockall score=30% (p<0.0001)Retrospective Short follow-up period Limited statistical results
Recurrent bleeding or death (Negative predictive value)Glasgow Blatchford score=100%; Initial (clinical or pre endoscopic) Rockall score=100%; Complete (post endoscopic) Rockall score=96.2%
Unknown country
111 patients with upper GI bleeding presenting to the emergency department during one year Compared Glasgow Blatchford score vs. Complete (post endoscopic) Rockall score vs. Clinical triage decision (physician’s clinical decision) in predicting need for endoscopic therapy Retrospective studySensitivityGlasgow Blatchford score=100%; Complete (post endoscopic) Rockall score=94%; Clinical triage decision (physician’s clinical decision) =67%Abstract only available Retrospective Variceal bleeding included Used ICU admission as surrogate marker for risk assessment
SpecificityGlasgow Blatchford score=2.6%; Complete (post endoscopic) Rockall score=2%; Clinical triage decision (physician’s clinical decision) =71%
AccuracyGlasgow Blatchford score=32%; Complete (post endoscopic) Rockall score=31%; Clinical triage decision (physician’s clinical decision) =69%


Upper GI bleeding is a common presentation. It would be very useful for the emergency medicine practice to have a simple and accurate way of differentiating the low risk patients who can be safely investigated and treated as outpatients. A Glasgow-Blatchford scoring system is promising as it is based on the clinical and simple laboratory data and does not rely on endoscopic findings, hence can be used easily in the emergency department. As Rockall is the most commonly used scoring system in upper GI bleed, the aim was to compare the Glasgow-Blatchford score with the Rockall score. Although the literature reviewed revealed only few comparative studies between these two scoring systems, the findings were interesting as almost all of the above mentioned studies demonstrated a very high sensitivity and negative predicting values for Glasgow-Blatchford scoring system in detecting high risk patients. Again concluded from the above studies, the drawback of the Glasgow-Blatchford score was its poor specificity and positive predicting factor. Most studies were retrospective, included a small number of patients and lacked the required follow-up for the low risk group. One study was not clear in the design and some were missing sufficient statistical findings. To be able to use the Glasgow-Blatchford scoring system as the standard method in triaging upper GI bleeding, more well designed, multicenter, large scale studies should be performed.

Editor Comment


Clinical Bottom Line

Glasgow-Blatchford scoring is a useful screening tool in identifying the low risk upper GI bleeding patients.

Level of Evidence

Level 3 - Small numbers of small studies or great heterogeneity or very different population.


  1. Blatchford O. Murray WR. Blatchford M. A risk score to predict need for treatment for upper-gastrointestinal haemorrhage. The Lancet 2000;356(9238):1318-21
  2. Stanley A.J. Ashley D. Dalton H.R. Mowat C. Gaya D.R. Thompson E. Warshow U. Groome M. Cahill A. Benson G. Blatchford O. Murray W. Outpatient management of patients with low-risk upper-gastrointestinal haemorrhage: multicentre validation and prospective evaluation The Lancet 2009;373(9657):42-47
  3. Chen IC. Hung MS. Chiu TF. Chen JC. Hsiao CT. Risk scoring systems to predict need for clinical intervention for patients with non-variceal upper gastrointestinal tract bleeding American Journal of Emergency Medicine 2007; 25(7):774-9
  4. Gralnek IM. Dulai GS. Incremental value of upper endoscopy for triage of patients with acute non-variceal upper-GI hemorrhage Gastrointestinal Endoscopy. 2004;60(1):9-14.
  5. Farees T. Farooq, Michael H. Lee, Rahul Dixit, Ananya Das and Richard C. Wong Physician Clinical Decision-Making and Triage Is a More Accurate Predictor of Need for Endoscopic Therapy (ET) Than Clinical Rockall Score (CRS) and Blatchford Score (BS) in Patients with Acute Upper Gastrointestinal Endoscopy 2007;65(5): AB122