El Kersh2015
El Kersh2015
El Kersh2015
1–8, 2015
Copyright Ó 2015 Elsevier Inc.
Printed in the USA. All rights reserved
0736-4679/$ - see front matter
http://dx.doi.org/10.1016/j.jemermed.2015.04.008
Clinical Laboratory in
Emergency Medicine
Karim El-Kersh, MD,* Udit Chaddha, MD,† Rahul Siddhartha Sinha, MD,† Mohamed Saad, MD,*
Juan Guardiola, MD,* and Rodrigo Cavallazzi, MD*
*Department of Pulmonary, Critical Care and Sleep Disorders Medicine, University of Louisville, Louisville, Kentucky and
†Department of Internal Medicine, University of Louisville, Louisville, Kentucky
Reprint Address: Karim El-Kersh, MD, Department of Pulmonary, Critical Care and Sleep Disorders Medicine, Ambulatory Care Building,
550 S. Jackson Street, Louisville, KY 40202
, Abstract—Background: The predictive role of lactate in and nonsurvivors was 2.0 (interquartile range [IQR]
critically ill patients with acute upper gastrointestinal 1.2–4.2 mmol/L) and 8.8 (IQR 3.4–13.3 mmol/L; p < 0.01),
bleeding (UGIB) remains to be elucidated. Objective: The respectively. The receiver operating characteristic (ROC)
primary objective of this study was to assess the value of area to predict in-hospital death for clinical Rockall score
lactate level on admission to predict in-hospital death in pa- and lactate level (0.82) was significantly higher than the
tients with UGIB admitted to the intensive care unit (ICU). ROC area for the clinical Rockall score alone (0.69)
The secondary objective was to assess whether lactate level (p < 0.01). Conclusions: In patients admitted to the ICU
adds predictive value to the clinical Rockall score in these with acute UGIB, lactate level on admission has a high sensi-
patients. Methods: This was a retrospective cohort study tivity but low specificity for predicting in-hospital death.
that included 133 patients with acute UGIB admitted to Lactate level adds to the predictive value of the clinical
the ICU. Inclusion criteria were age > 18 years and presence Rockall score. Given its high sensitivity, lactate level can
of UGIB on admission to the ICU. Results: Mean age was be used in addition to other prediction tools to predict
55.4 years old and 64.7% were male. The most common outcomes in patients with UGIB. Ó 2015 Elsevier Inc.
cause of gastrointestinal bleeding was peptic ulcer disease,
followed by erosive esophagitis/gastritis. The in-hospital , Keywords—lactate; gastrointestinal bleeding; Rockall
mortality was 22.6%. Median lactate level in survivors score; mortality in ICU
The results of this study were partly presented in the Amer- INTRODUCTION
ican Thoracic Society 2014 international conference held in
San Diego, CA on May 21, 2014 (Chaddha US, Sinha RS, Acute upper gastrointestinal bleeding (UGIB) accounts
El-Kersh K, Woodford M, Cavallazzi R. Lactate level in criti- for >400,000 hospitalizations per year in the United
cally ill patients with acute gastrointestinal bleeding. Am J States, with an estimated mortality rate ranging between
Respir Crit Care Med 2014;189:A5492).
6% and 10% (1–4). In the intensive care unit (ICU), the
This study has been approved by the appropriate ethics com-
mittee and has therefore been performed in accordance with the
mortality rate can be even higher (5). Risk stratification
ethical standards laid down in the 1964 Declaration of Helsinki in UGIB is important in order to select low-risk patients
and its later amendments. This study was approved by the Uni- for early discharge and high-risk patients for ICU man-
versity of Louisville Institutional Review Board (protocol agement. Several scoring systems have been developed
#13.0231). Informed consent was waived. to predict outcomes after UGIB, but none of these scoring
1
2 K. El-Kersh et al.
systems used serum lactate level as a predictor of out- study were admitted to the same ICU, which had a single
comes (6–8). Although it is well known that in severe team managing all of the patients with UGIB included in
sepsis a high serum lactate level is associated with the study with the same standard protocols for volume
mortality independent of organ dysfunction and shock, resuscitation, blood transfusions, pressor usage, and
the knowledge about the role of serum lactate level interventional procedures, among others.
upon admission in predicting outcomes in patients with
UGIB is still evolving (5,9–11). We hypothesized that Measurements
the use of serum lactate level upon admission can be a
valuable tool for the prediction of outcomes in patients We abstracted information from the charts of patients us-
with upper gastrointestinal hemorrhage. The primary ing a structured data-collection form. Information
aim of this study was to evaluate the value of the collected included demographics, cause of gastrointes-
admission lactate level to predict in-hospital death in pa- tinal hemorrhage, comorbidities, vital signs, and lactate
tients with UGIB admitted to the ICU. The secondary aim level on admission to the ICU, the clinical Rockall score,
was to evaluate whether lactate level adds predictive and outcomes.
value to the clinical Rockall score in these patients. The medical record review and data abstraction were
performed by two internal medicine residents who had at
METHODS least 1 year of internal medicine training. They underwent
data-collection training that included defining eligibility
This was a retrospective cohort study that included criteria and other variables that were included in the study
consecutive patients with acute UGIB admitted to our via initial supervised data collection of randomly selected
university hospital ICU from 2010 to 2013. We per- charts. After completion of data collection, random chart
formed a secondary analysis of a database created by reviews were performed to ensure data accuracy.
our group of patients with gastrointestinal hemorrhage The clinical Rockall score (before endoscopy) was
admitted to the ICU. The study was approved by the uni- calculated from three clinical variables that included pa-
versity Institutional Review Board (protocol #13.0231). tient’s age (score 0 to 2), presence of shock (systolic
Informed consent was waived. blood pressure and heart rate) (score 0 to 2), and presence
of comorbid conditions (score 0 to 3), with a maximum
Patients additive score of 7 (Table 1).
The venous lactate level was measured by VITROS-
Patients admitted to the ICU from the emergency depart- 5600 analyzer (Ortho Clinical Diagnostics, Rochester,
ment (ED) with a primary diagnosis of acute UGIB were NY). The outcome for this study was in-hospital death.
identified through a computerized search using Interna-
tional Classification of Diseases, Ninth Revision, and Statistical Analysis
Clinical Modification (ICD-9-CM) codes. Medical re-
cords were subsequently reviewed. Inclusion criteria for We present normally distributed continuous variables as
the study were age > 18 years, presence of UGIB (evi- mean and standard deviation (SD). When they are not nor-
denced by hematemesis or an endoscopic evidence of mally distributed, we present them as median and inter-
UGIB in patients presenting with either hematemesis or quartile range (IQR). We employed the Wilcoxon rank-
melena) on admission to the ICU, and a lactate level sum (Mann-Whitney) test to compare continuous vari-
that was obtained on the same day of ICU admission. ables. For categorical variables, we employed Fisher’s
We excluded patients with lower gastrointestinal exact test. We provided the accuracy of both lactate level
bleeding and those who did not have a lactate level and the clinical Rockall score to predict in-hospital death.
upon admission to ICU. All the patients included in the For lactate level, we used a cutoff of >2.1 mmol/L, as
Score
Variable 0 1 2 3
CHF = congestive heart failure; HR = heart rate; IHD = ischemic heart disease; SBP = systolic blood pressure.
Lactate in Acute Upper Gastrointestinal Bleeding 3
RESULTS
SD = standard deviation.
4 K. El-Kersh et al.
Lactate, mmol/L, median (IQR) 2.6 (1.4–6.8) 2 (1.2–4.2) 8.8 (3.4–13.3) <0.01
Clinical Rockall score, mean (SD) 3.0 (1.7) 2.8 (1.7) 3.8 (1.5) <0.01
and the clinical Rockall score + lactate level, respectively remained significantly higher in nonsurvivors compared
(Figure 2). The ROC area for clinical Rockall score + to survivors (9.7; IQR 4.5–15.2 vs. 2.25; IQR 1.55–5,
lactate level was significantly higher than the ROC area respectively; p = 0.003).
for clinical Rockall score (p < 0.01). However, there
was no statistical significance between the ROC areas DISCUSSION
for lactate and clinical Rockall score + lactate (p = 0.68).
Our study found that lactate level of 2.1 mmol/L on
Relationship Between Lactate Level, Clinical Rockall admission has high sensitivity but low specificity for pre-
Score, and In-Hospital Mortality dicting in-hospital death in patients with acute upper
gastrointestinal hemorrhage admitted to the ICU. The
Four patients with a lactate level < 2.1 mmol/L died in the addition of the clinical Rockall score to lactate level did
hospital. All 4 patients had a clinical Rockall score $2. not significantly improve the discriminative property of
Table 4 stratifies the numbers of patients with clinical lactate. However, with the combined use of clinical Rock-
Rockall score > 1 and outcomes according to lactate level all score and lactate level, no case of in-hospital death
quartiles, and Figure 3 shows the lactate level according would have been missed. Such a highly sensitive strategy
to clinical Rockall score in survivors and nonsurvivors. comes at the cost of a very low specificity.
The main scoring systems that predict outcomes
Regression Analysis after UGIB, such as clinical Rockall score, Glasgow-
Blatchford score, and AIMS65 score use systolic blood
On univariate logistic regression, the odds ratio for in- pressure and heart rate as a part of the scoring system,
hospital death was 8.1 (95% CI 2.6–24.7; p < 0.01) for pa- but none of these scores used serum lactate level to
tients with lactate level > 2.1 mmol/L, and 5.5 (95% CI predict outcomes (6–8).
1.2–24.5; p = 0.026) for patients with clinical Rockall Koch et al. found elevated serum lactate level in
score > 1. On multivariate regression analysis, both patients with acute gastrointestinal hemorrhage on admis-
lactate level and the clinical Rockall score remained inde- sion to the ICU to be superior to some of these standard
pendently associated with in-hospital death (Table 5). prognostic scores in predicting mortality (5). Further-
more, elevated serum lactate level on presentation to
Subgroup Analysis the ED with acute gastrointestinal hemorrhage was found
to be associated with higher in-hospital mortality and it
We assessed the lactate level according to outcomes in the was independently correlated with death (11).
group of patients with liver cirrhosis. Median lactate level It is well known that, in severe sepsis, high serum
lactate is associated with mortality independent of organ
dysfunction and shock (9). In addition, in septic patients,
increased clearance of lactate within the first 6 h has been
shown to be associated with decreased 60-day mortality,
even in the absence of arterial hypotension (13). In
trauma patients, survival rate was 100% for those who
had normalization of their serum lactate level in 24 h (14).
Animal studies showed that the gut is sensitive to
hypoperfusion. During hemorrhage, intestinal oxygen
uptake is compromised early, even without changes in
systemic oxygen consumption, due to reflex mesenteric
vasoconstriction (15). Furthermore, progressive vasocon-
striction of microvasculature can persist after resuscita-
tion and restoration of hemodynamics (16,17).
The concept of tissue hypoperfusion despite restora-
Figure 2. Area under the receiver operating characteristic
(ROC) of predictive models for in-hospital mortality. CRS = tion of hemodynamics was further supported by clinical
Clinical Rockall Score. studies that showed evidence of tissue hypoperfusion,
Lactate in Acute Upper Gastrointestinal Bleeding 5
Table 4. Relationship Between Outcome and Clinical Rockall Score According to Lactate Level Quartiles
evidenced by lactic acidosis and decreased mixed production cannot be measured or predicted based
venous oxygen saturation, despite normalization of vital solely on hemodynamics as well.
parameters, including blood pressure, heart rate, and
urine output (18,19). Also, some patients with acute Implications for Future Research
gastrointestinal hemorrhage have an abnormally ele-
vated lactate level, despite normal hematocrit and heart Our study encourages further investigation of the predic-
rate, which point to the potential use of lactate level to tive role of serum lactate level in the setting of acute
identify occultly ill patients (11). UGIB. Developing scoring systems for an acute UGIB
Besides tissue hypoxia, other mechanisms have been incorporating serum lactate level and other biomarkers
proposed for lactic acidosis in the setting of sepsis can be beneficial for predicting outcomes. The role of
(20,21). These mechanisms include mitochondrial serial measurements of serum lactate levels for predicting
dysfunction and increased aerobic glycolysis via tissue outcomes in acute UGIB and the role of using serum
cytokine-mediated glucose uptake or catecholamine- lactate to help in triage of patients with acute UGIB can
enhanced Na-K skeletal muscle pump activity, which is be investigated in future studies.
a proposed mechanism of lactic acid production in hem- Prospective studies can assess the impact of lactate use in
orrhage as well (22–28). These mechanisms of lactic acid acute UGIB on both resource utilization and patient-
production are not reflected in vital parameters, such as oriented outcomes before it is recommended for routine use.
blood pressure and heart rate. These findings support
the role of serum lactate as a marker of tissue Limitations
hypoperfusion in the setting of normal blood pressure
and heart rate in early stages of hypoperfusion and after Our study has a number of limitations. The retrospective
resuscitation. Other proposed methods of lactate design leads to a higher risk of misclassification bias with
Figure 3. Lactate level according to Clinical Rockall score in survivors and nonsurvivors.
6 K. El-Kersh et al.
Table 5. Univariate and Multivariate Logistic Regression Results for the Outcome In-Hospital Death
regard to the characteristics of the patients. Because of 3. Lin HJ, Wang K, Perng CL, Lee CH, Lee SD. Heater probe ther-
mocoagulation and multipolar electrocoagulation for arrest of
the reliance on administrative search to identify cases,
peptic ulcer bleeding: a prospective, randomized comparative trial.
it is possible that some patients have been left out. J Clin Gastroenterol 1995;21:99–102.
Another limitation is that a high number of patients 4. Kankaria AG, Fleischer DE. The critical care management of non-
variceal upper gastrointestinal bleeding. Crit Care Clin 1995;11:
were not included because a lactate test was not available 347–68.
on admission. As a result, there is a potential for selection 5. Koch A, Buendgens L, Dückers H, et al. Bleeding origin, patient-
bias, as the sicker patients might be more likely to be related risk factors, and prognostic indicators in patients with acute
gastrointestinal hemorrhages requiring intensive care treatment. A
tested for serum lactate levels; therefore, the study popu- retrospective analysis from 1999 to 2010. Med Klin Intensivmed
lation might be biased toward selection of a population Notfmed 2013;108:214–22.
with a higher risk of death, which can lead to underesti- 6. Saltzman JR, Tabak YP, Hyett BH, Sun X, Travis AC, Johannes RS.
A simple risk score accurately predicts in-hospital mortality, length
mating negative predictive and overestimating positive of stay, and cost in acute upper GI bleeding. Gastrointest Endosc
predictive value, compared with an unselected popula- 2011;74:1215–24.
tion. Although we documented the admitting systolic 7. Blatchford O, Murray WR, Blatchford M. A risk score to predict
need for treatment for upper gastrointestinal haemorrhage. Lancet
blood pressure, heart rate, and presence of comorbid con- 2000;356:1318–21.
ditions that are required to calculate clinical Rockall 8. Rockall TA, Logan RFA, Devlin HB, Northfield TC. Risk assess-
score, we did not document the presence of other acute ment after acute upper gastrointestinal haemorrhage. Gut 1996;
38:316–21.
comorbid conditions that might have an impact on lactate
9. Miklsen ME, Miltiades AN, Gaieski DF, et al. Serum lactate is asso-
level and mortality. Median lactate level was significantly ciated with mortality in severe sepsis independent of organ failure
higher in nonsurvivors compared to survivors in patients and shock. Crit Care Med 2009;37:1670–7.
10. Chaddha US, Sinha RS, El-Kersh K, Woodford M, Cavallazzi R.
with liver cirrhosis, but the severity of underlying liver Lactate level in critically ill patients with acute gastrointestinal
disease could not be assessed. bleeding. Am J Respir Crit Care Med 2014;189:A5492.
11. Shah A, Chisolm-Straker M, Alexander A, Rattu M, Dikdan S,
Manini AF. Prognostic use of lactate to predict inpatient mortality
CONCLUSIONS in acute gastrointestinal hemorrhage. Am J Emerg Med 2014;32:
752–5.
According to our study results, serum lactate test on 12. Rockall T, Logan R, Devlin H, Northfield T. Selection of patients for
early discharge or outpatient care after acute upper gastrointestinal
admission can be a useful adjunct to the often difficult haemorrhage. Lancet 1996;347:1138–40.
task of predicting outcomes in patients with upper gastro- 13. Nguyen HB, Rivers EP, Knoblich BP, et al. Early lactate clearance is
intestinal hemorrhage. Given its high sensitivity, it ap- associated with improved outcome in severe sepsis and septic
shock. Crit Care Med 2004;32:1637–42.
pears that lactate level may be particularly useful as a 14. Abramson D, Scalea TM, Hitchcock R, Trooskin SZ, Henry SM,
screening test. Greenspan J. Lactate clearance and survival following injury.
We found that patients with normal lactate level are J Trauma 1993;35:584–9.
15. Nelson DP, King CE, Dodd SL, Schumacker PT, Cain SM. Systemic
unlikely to suffer in-hospital death. However, because and intestinal limits of O2 extraction in the dog. J Appl Physiol
there are patients who can develop bad outcomes despite 1987;63:387–94.
a normal lactate level, the isolated use of this test for pre- 16. Fruchterman TM, Spain DA, Wilson MA, Harris PD, Garrison RN.
Selective microvascular endothelial cell dysfunction in the small
diction of outcomes should not be encouraged. Instead, intestine after resuscitated hemorrhagic shock. Shock 1998;10:
lactate level may be a suitable test to be used in addition 417–22.
to other prediction tools to predict outcome in patients 17. Zakaria ER, Spain DA, Harris PD, Garrison RN. Resuscitation reg-
imens for hemorrhagic shock must contain blood. Shock 2002;18:
with acute upper gastrointestinal hemorrhage. 567–73.
18. Scalea TM, Maltz S, Yelon J, Trooskin SZ, Duncan AO, Sclafani SJ.
Resuscitation of multiple trauma and head injury: role of crystalloid
REFERENCES fluids and inotropes. Crit Care Med 1994;22:1610–5.
19. Abou-Khalil B, Scalea TM, Trooskin SZ, Henry SM, Hitchcock R.
1. Gralnek IM, Barkun AM, Bardou M. Current concepts: manage- Hemodynamic response to shock in young trauma patients: need for
ment of acute bleeding from a peptic ulcer. N Engl J Med 2008; invasive monitoring. Crit Care Med 1994;22:633–9.
359:928–37. 20. Levy B, Sadoune LO, Gelot AM, Bollaert PE, Nabet P, Larcan A.
2. Silverstein FE, Gilbert DA, Tedesco FJ, Buenger NK, Persing J. The Evolution of lactate/pyruvate and arterial ketone body ratios in
National ASGE Survey on Upper Gastrointestinal Bleeding II. Clin- the early course of catecholamine-treated septic shock. Crit Care
ical prognostic factors. Gastrointest Endosc 1981;27:80–93. Med 2000;28:114–9.
Lactate in Acute Upper Gastrointestinal Bleeding 7
21. Rivers E, Nguyen B, Havstad S, et al. Early goal-directed therapy in 25. Haji-Michael PG, Ladriere L, Sener A, Vincent JL, Malaisse WJ.
the treatment of severe sepsis and septic shock. N Engl J Med 2001; Leukocyte glycolysis and lactate output in animal sepsis and
345:1368–77. ex vivo human blood. Metabolism 1999;48:779–85.
22. Brealey D, Brand M, Hargreaves I, et al. Association between mito- 26. Levy B, Gibot S, Franck P, Cravoisy A, Bollaert PE. Relation be-
chondrial dysfunction and severity and outcome of septic shock. tween muscle Na+K+ ATPase activity and raised lactate concentra-
Lancet 2002;360:219–23. tions in septic shock: a prospective study. Lancet 2005;365:871–5.
23. Crouser ED, Julian MW, Blaho DV, Pfeiffer DR. Endotoxin- 27. Luchette FA, Robinson BR, Friend LA, McCarter F, Frame SB,
induced mitochondrial damage correlates with impaired respiratory James JH. Adrenergic antagonists reduce lactic acidosis in response
activity. Crit Care Med 2002;30:276–84. to hemorrhagic shock. J Trauma 1999;46:873–80.
24. Taylor DJ, Faragher EB, Evanson JM. Inflammatory cytokines 28. Luchette FA, Friend LA, Brown CC, Upputuri RK, James JH.
stimulate glucose uptake and glycolysis but reduce glucose Increased skeletal muscle Na+, K+–ATPase activity as a cause of
oxidation in human dermal fibroblasts in vitro. Circ Shock increased lactate production after hemorrhagic shock. J Trauma
1992;37:105–10. 1998;44:796–801.
8 K. El-Kersh et al.
ARTICLE SUMMARY
1. Why is this topic important?
This topic is important because predicting outcome in
acute upper gastrointestinal bleeding (UGIB) is essential
for triaging and clinical management. In addition, the pre-
dictive value of combining lactate level to clinical Rockall
score to predict outcomes after UGIB was not evaluated
before.
2. What does this study attempt to show?
This study evaluated the value of lactate level on admis-
sion to predict in-hospital death in patients with UGIB
admitted to the intensive care unit (ICU). Also, it evalu-
ated whether lactate level adds predictive value to the clin-
ical Rockall score in these patients.
3. What are the key findings?
In patients admitted to the ICU with acute UGIB,
lactate level on admission has a high sensitivity but low
specificity for predicting in-hospital death. Lactate level
adds to the predictive value of the clinical Rockall score.
4. How is patient care impacted?
Lactate level might be a suitable test to be used in addi-
tion to other prediction tools to predict prognosis patients
with acute upper gastrointestinal hemorrhage. Developing
scoring systems for an acute UGIB incorporating serum
lactate level and other biomarkers can be beneficial for
predicting outcomes.