Within the Emergency Department (ED), rapid identification and stabilization of critically ill patients with upper gastrointestinal bleeding (UGIB) is essential to patient survival. While there are some clinical prediction scores, ED physicians most often depend on clinical gestalt when risk stratifying and managing these patients. The most commonly used scores are the Glasgow-Blatchford score, AIMS65 and the Rockall score. The Rockall score, developed in 1996 to predict mortality, is a well-known inpatient predictor score that has been validated to risk stratify patients with UGIB. Given this score requires endoscopic results, the Clinical Rockall score has been developed for ED patients. However this score has come under scrutiny for its low specificity (Johnston 2015)and inability to identify patients requiring endoscopic intervention (Masoaka 2007, Pang 2010).

Recently, the discussion has turned to using lactate for risk stratification. The association between lactate levels and mortality in sepsis is well known and has been extensively analyzed in past literature. The use of lactate in UGIB however, has not yet been clearly elucidated. This study seeks to assess the utility of elevated lactate (>2.1 mmol/L) in predicting mortality. Furthermore, it seeks to determine if the addition of lactate criterion adds to the clinical prediction of the clinical Rockall score.

Clinical Question

Can the serum lactate level on admission predict in-hospital death in patients with UGIB admitted to the ICU? Can lactate level add to the predictive value of the clinical Rockall score?


Patients 18 yrs and older admitted to the study site ICU from the ED with primary diagnosis of acute UGIB


In hospital mortality


Single center, retrospective cohort study


Lower GI Bleeding and those who did not have a lactate level on admission to the ICU

Primary Results

Primary Results

  • 133 patients included
  • Median lactate level in survivors and non-survivors was 2.0 mmol/L and 8.8 mmol/L, respectively

Critical Findings

  • Predicting in hospital mortality:
    • Elevated lactate (>2.1 mmol/L) had a sensitivity of 87% and specificity of 55%
    • Rockall score >1 had a sensitivity of 93% and specificity of 28%
  • ROC areas to predict in-hospital death:
    • Clinical Rockall score = 0.69
    • Lactate Level = 0.80
    • Clinical Rockall + Lactate level = 0.82
  • Odds Ratio for in-hospital death:
    • Elevated Lactate = 8.1 (p <0.01)
    • Rockall score >1 = 5.5 (p = 0.026)



  • Well defined, patient centered clinical end-point
  • Objective outcome measure
  • Measurements adjusted for abnormal distribution to avoid the influence of outliers


  • Retrospective design
  • Small sample size
  • Exclusion of pts without lactate drawn may lead to selection bias
  • Single institution

Author's Conclusions

“Lactate level on admission has a high sensitivity but low specificity for predicting in-hospital death. Lactate adds to the predictive value of the clinical Rockall score. Given its high sensitivity, lactate level can be used in addition to other prediction tools to predict outcomes in patients with UGIB.”

Our Conclusions

In this retrospective cohort study of acute UGIB patients sent to the ICU, median lactate levels were noted to be sensitive, but not specific, for in-hospital mortality. This study succeeded in identifying a way to improve UGIB evaluation and management through enhancement of a scoring system. The improvement of the clinical Rockall predictive score with the addition of lactate is notable, and warrants further assessment with a large, multicenter study. This study also showed a direct correlation between increased lactate and mortality, a result that is consistent with prior studies.
Some aspects of this study do warrant clarification, such as the decision to make the Rockall cut-off score greater than 1. The literature, including the paper cited for this decision (Rockall 1996), typically use a Rockall score of 2 or more as the cut-off for low risk individuals, stating a score of 2 or less has “negligible risk of death” with a mortality of 0.3%. Further studies should be conducted to determine whether alteration of this score to >2 improves the specificity, while maintaining sensitivity. The lactate level cut-off also needs further justification. This cut-off was based on a study using lactate to risk-stratify sepsis patients (Miklsen 2009). It remains to be seen if these same cut-offs can be applied to patients with UGIB. This remains questionable, given prior gastrointestinal literature cites a lactate level of >4 as the crucial cut-off point (Shah 2014). Further studies can also evaluate the utility of measuring lactate clearance with these patients as well using lactate in patients with varying degrees of liver dysfunction.

Potential Impact To Current Practice

The Rockall score in combination with lactate level in patients with UGIB may assist in risk stratifying patients in the ED. However, clinical gestalt must be implemented, as these two measurements require additional modifications and validation before becoming standard of practice.

Bottom Line

The Rockall score and lactate levels may be used to assist, but not determine level of care for patients with UGIB in the ED. Further work should be done to develop a better scoring system.

Read More

EM Crit: Episode 5 Upper GI Bleeding guidelines


Blatchford, O., Murray, W.R., Blatchford, M. A risk score to predict need for treatment for upper gastrointestinal haemorrhage. Lancet. 2000;356:1318–1321. PMID: 11073021

Masaoka T, Suzuki H 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 Hepatol. 2007;22(9):1404-1408 PMID: 17716345

Meltzer, AC., Bernett, S., et al, Pre-endoscopic Rockall and Blatchford scores to identify which emergency department pateints with suspected gastrointestinal bleed do not need endoscopic hemostasis. J Emerg Med. 2013 Jun; 44(6):1083-7. PMID: 23360648

Miklsen, M.E., Miltiades, A.N., Gaieski, D.F. et al, Serum lactate is associated with mortality in severe sepsis independent of organ failure and shock. Crit Care Med. 2009;37:1670–1677. PMID: 19325467

Johnston M.R., Murray I.A, et al. Does Preendscopy Rockall Score Safely Identify Low Risk patients following Upper Gastrointestinal Haemorrhage Gastroenterol Res Pract 2015: 2015. Epub 2015 May 5. PMID: 26089867

Pang, S., Ching J. et al, Comparing the Blatchford and pre-endoscopic Rockall score in predicting the need for endoscopic therapy in patients with upper GI hemorrhage. Gastrointest Endosc. 2010 Jun;71(7):1134-40. PMID: 20598244

Rockall, T.A., Logan, R.F.A., Devlin, H.B., Northfield, T.C. Risk assessment after acute upper gastrointestinal haemorrhage. Gut. 1996;38:316–321. PMID: 8675081

Shah, A., Chisolm-Straker, M., Alexander, A., Rattu, M., Dikdan, S., Manini, A.F. Prognostic use of lactate to predict inpatient mortality in acute gastrointestinal hemorrhage. Am J Emerg Med. 2014;32:752–755. PMID: 24813902