Intern Emerg Med DOI 10.1007/s11739-016-1392-z
EM - ORIGINAL
Early lactate clearance for predicting active bleeding in critically ill patients with acute upper gastrointestinal bleeding: a retrospective study Tomoki Wada1 • Akiyoshi Hagiwara2 • Tatsuki Uemura2 Naoki Yahagi1 • Akio Kimura2
•
Received: 21 August 2015 / Accepted: 9 January 2016 Ó SIMI 2016
Abstract Not all patients with upper gastrointestinal bleeding (UGIB) require emergency endoscopy. Lactate clearance has been suggested as a parameter for predicting patient outcomes in various critical care settings. This study investigates whether lactate clearance can predict active bleeding in critically ill patients with UGIB. This single-center, retrospective, observational study included critically ill patients with UGIB who met all of the following criteria: admission to the emergency department (ED) from April 2011 to August 2014; had blood samples for lactate evaluation at least twice during the ED stay; and had emergency endoscopy within 6 h of ED presentation. The main outcome was active bleeding detected with emergency endoscopy. Classification and regression tree (CART) analyses were performed using variables associated with active bleeding to derive a prediction rule for active bleeding in critically ill UGIB patients. A total of 154 patients with UGIB were analyzed, and 31.2 % (48/ 154) had active bleeding. In the univariate analysis, lactate clearance was significantly lower in patients with active bleeding than in those without active bleeding (13 vs. 29 %, P \ 0.001). Using the CART analysis, a prediction rule for active bleeding is derived, and includes three variables: lactate clearance; platelet count; and systolic blood pressure at ED presentation. The rule has 97.9 % & Tomoki Wada [email protected] 1
Department of Emergency and Critical Care Medicine, The University of Tokyo Hospital, 7-3-1 Hongo, Bunkyo-Ku, Tokyo 113-8655, Japan
2
Department of Emergency Medicine and Critical Care, Center Hospital of the National Center for Global Health and Medicine, 1-21-1 Toyama, Shinjuku-Ku, Tokyo 162-8655, Japan
(95 % CI 90.2–99.6 %) sensitivity with 32.1 % (28.6–32.9 %) specificity. Lactate clearance may be associated with active bleeding in critically ill patients with UGIB, and may be clinically useful as a component of a prediction rule for active bleeding. Keywords Lactate clearance Gastrointestinal hemorrhage Prediction rule Emergency endoscopy
Introduction Upper gastrointestinal bleeding (UGIB) is a common emergency condition that ranges in severity from mild symptoms that do not require any interventions, to lifethreatening presentations with hemorrhagic shock that require emergency endoscopy or surgery. The International Consensus Guidelines of 2010 recommend the stratification of patients’ risks using validated prognostic scales [1]. There are several prognostic scales for UGIB. Among these, the Glasgow-Blatchford Score (GBS) is the most reliable prognostic scale for determining outcomes in patients with UGIB (Table 1) [2–4]. The GBS can be calculated in the absence of endoscopic findings, and it predicts the need for interventional hemostasis, which comprises endoscopy and surgery, or transfusion [4]. The GBS can accurately predict which patients are at low risk, and who will not require interventional hemostasis or transfusion [5]. However, the GBS cannot reliably distinguish between patients with active UGIB who require emergency hemostasis and those who do not require hemostasis, but only transfusion [6]. Serum lactate and its serial change (lactate clearance) have been suggested as a possible predictor of patient outcome in several critical care conditions [7]. A recent study
123
Intern Emerg Med Table 1 Glasgow-Blatchford score [4] Risk factor
Score
Blood urea nitrogen (mg/dL) 18.2–22.4
2
22.4–28.0
3
28.0–70.0
4
C70.0
6
Hemoglobin for men (g/dL) 12.0–12.9
1
10.0–11.9
3
\10.0
6
Hemoglobin for women (g/dL) 10.0–11.9 \10.0
1 6
Systolic blood pressure (mmHg) 100–109
1
90–99
2
\90
3
Other markers Heart rate C 100 (bpm)
1
Melena
1
Syncope
2
Hepatic disease
2
Heart failure
2
bpm beats per minute
finds that an initial lactate value at admission may be useful in predicting mortality in patients with UGIB [8]. However, no study has evaluated whether a lactate value or lactate clearance can predict active bleeding in patients with UGIB. Emergency endoscopy should be performed on patients with active UGIB [9–11]. However, emergency endoscopy can represent overtreatment in patients who do not have active bleeding, and can result in unnecessary intervention and cost for these patients [10–12]. Hence, appropriate selection of patients requiring emergency endoscopy is important. The present study determines whether lactate clearance is associated with active UGIB. The study also investigates whether lactate clearance can be clinically useful as a component of a prediction rule for active bleeding in critically ill patients with UGIB.
Patients and methods Study design and setting This study was a single-center, retrospective, cohort study held in Center Hospital of the National Center for Global
123
Health and Medicine, Tokyo. The hospital is a tertiary, general hospital with 800 beds, and more than 10,000 patients are brought to the emergency department (ED) by ambulance annually. An emergency physician is in charge of the initial assessment and resuscitation for patients with UGIB who are brought to the hospital by ambulance. The treatment policy for these patients in the ED of the hospital is as follows. First, an emergency physician determines whether a patient is hemodynamically unstable and whether this is possibly related to UGIB at ED presentation. For a patient who is hemodynamically unstable, the emergency physician collects blood samples and starts administering 20–40 mL/kg of Ringer’s acetate solution (fluid challenge). Hemodynamic instability is implicitly determined based on the patient’s vital signs and other physical findings such as peripheral coldness, a prolonged capillary refill time, and altered mental status. The possibility of UGIB is evaluated on the basis of the presence of melena or hematemesis. After a fluid challenge for the patient who is hemodynamically unstable with UGIB at the ED presentation, the emergency physician collects blood samples again to evaluate changes in laboratory parameters, including serum lactate. We used GEM Premier 4000 (Instrumentation Laboratory, Bedford, MA, USA) to measure the serum lactate concentration. The emergency physician then consults with a gastroenterologist regarding critically ill patients with UGIB. The gastroenterologist then decides whether a patient requires an emergency endoscopy or not. With a few exceptions, gastroenterologists perform emergency endoscopy for all UGIB patients who are admitted through the ED. If patients admitted to our hospital in the middle of the night, are hemodynamically stable, and are without hematemesis, they might undergo endoscopy during the closest office hours. This treatment policy is the routine of care for UGIB at our ED prior to planning the current study. The Ethics Committee of the hospital approved this retrospective cohort study. The requirement for written informed consent from patients was waived because the study design formed part of the current standard of care in our ED, and patient data were anonymous. Patient enrollment Medical records of patients who were brought by ambulance with a presenting complaint of melena or hematemesis and admitted to the hospital from April 2011 to August 2014 were reviewed. Of these, we selected patients who were admitted to the hospital with a primary diagnosis of UGIB. Inclusion criteria were as follows: systolic blood pressure (SBP) B 109 mmHg or heart rate (HR) C 100 beats per minute (bpm) at ED presentation;
Intern Emerg Med
lactate evaluations performed at least twice during the ED stay; and the receipt of emergency endoscopy within 6 h after the ED presentation. The cutoffs of SBP and HR were determined with reference to the related components of the GBS (Table 1). In calculating a patient’s GBS, one point was added when the SBP was below 110 mmHg or the HR was 100 bpm or higher. Patients were excluded when they met, at least, one of the following criteria: referred from another hospital; initial endoscopy performed more than 6 h after the ED presentation; ‘‘do not attempt resuscitation’’ order was present; or data were missing. Variables In accordance with the GBS, we determined whether patients had hepatic diseases or cardiac failure as comorbidities based on patients’ histories, physical findings, and laboratory test results [4]. The following laboratory parameters were determined at ED presentation: complete blood count, international normalized ratio, total bilirubin, blood urea nitrogen, and serum lactate (initial lactate). Lactate was also evaluated after a fluid challenge (second lactate). Lactate clearance was calculated as follows: [(initial lactate - second lactate)/initial lactate] 9 100 %. The amounts of Ringer’s acetate solution and red cell concentrate used before the second lactate evaluation were identified individually for each patient. Study outcome The primary outcome was active bleeding as detected via endoscopy. Active bleeding was defined as one of the following: peptic ulcers of Forrest classification IA or IB [13], hemorrhagic varices, or other active bleeding for which a gastroenterologist performed a hemostatic procedure. Patients with active bleeding did not include those who received intervention only for lesions with evidence of recent bleeding. The possible etiology of UGIB in each patient was based on an endoscopy report and a discharge abstract written by the gastroenterologist in charge. Statistical analysis In univariate analyses, the Mann–Whitney U test was used to compare continuous variables, while Fisher’s exact test or the Chi-square test was used to compare categorical variables between patients with and without active bleeding. A multivariate logistic regression analysis was then performed to determine factors associated with active bleeding, with adjustment for hepatic diseases, the presence of hypotension (SBP \ 110 mmHg) at ED presentation, and lactate clearance. This was done because hepatic diseases may affect lactate metabolism [14], and may also
affect outcomes owing to the strong relationship between hepatic diseases and the presence of esophagogastric varices. A classification and regression tree (CART) analysis with standard Gini splitting rule was performed to determine a combination of predictive variables, which distinguished between patients with high and low risk of active bleeding [15]. To perform the CART analysis, we used SPSS Decision Trees version 20.0 (IBM Inc., Armonk, NY, USA). In this analysis, only variables with P \ 0.10 in the univariate analyses were included. The process of the CART analysis was as follows. Initially, the most powerful variable to separate patients into high- and low-risk groups was chosen. If the variable was numerical, an optimal cutoff point was identified. Then, patients were separated into two groups in accordance with the variable. In each group, patients could be further divided into two subgroups with a variable chosen in the same way as that in the first separation. The CART analysis repeated the process, and eventually generated a decision tree model. Tenfold cross validation was used to develop a stable decision tree model. Because problems of overfitting occur if the number of outcome events is too small per predictive variables in a model, we pruned the tree until the number of variables in the model fell below one-tenth of the outcome events (i.e., the number of patients with active bleeding in the present study) [16]. The predictive performance of the model for active bleeding was evaluated by calculating sensitivity, specificity, positive predictive value, and negative predictive value. All statistical tests were two tailed, and factors were considered significant for P \ 0.05. All statistical calculations were performed using SPSS Statistics version 20.0 (IBM Inc., Armonk, NY, USA).
Results The patient selection process is illustrated in Fig. 1, 154 patients were included in the analysis. Active UGIB is found in 31.2 % (48/154) patients (Table 2). Systolic blood pressure at ED presentation tends to be lower (99 vs. 105 mmHg, P = 0.05), and platelet counts are also significantly lower (16.3 vs. 19.0 9 104/mm3, P = 0.04) in patients with active bleeding than in those without active bleeding. Second lactate values are significantly higher (3.0 vs. 2.0 mmol/L, P = 0.03), and lactate clearance is significantly lower (13 vs. 29 %, P \ 0.001), in patients with active bleeding than in those without active bleeding (Table 2). There are no significant differences between patients with and without active bleeding in the time interval between the patients arriving at the ED and second
123
Intern Emerg Med Fig. 1 Patient selection process. Asterisk Critically ill patients were defined as those with systolic blood pressure B 109 mmHg or heart rate C 100 beats per minute at ED presentation. ED emergency department, UGIB upper gastrointestinal bleeding
measurements of the serum lactate level (60 vs. 55 min, P = 0.62). This is also true for the time interval between patient arrival at the ED and emergency endoscopy (146 vs. 143 min, P = 0.74) (Table 2). The multivariate logistic regression analysis reveals that higher lactate clearance is associated with a reduction in the risk of active bleeding in patients with UGIB (adjusted odds ratio: 0.15; 95 % confidence interval [CI] 0.04–0.51) (Table 3). Possible etiologies of UGIB in the patients are described in Table 4. There are no significant differences in the distribution of etiologies between the active and non-active bleeding groups (P = 0.38). The CART analysis for active bleeding was performed with the following variables: SBP at ED presentation, platelet count, second lactate values, and lactate clearance. Consequently, a prediction rule was developed and comprised lactate clearance, platelet count, and SBP at ED presentation (Fig. 2). The rule has 97.9 % sensitivity with 32.1 % specificity for active bleeding (Table 5).
Discussion The present study suggests that high lactate clearance is associated with a reduced risk of active bleeding in critically ill patients with UGIB. Additionally, as a component of a prediction rule for patients with UGIB, lactate clearance may be clinically useful to identify patients without active bleeding.
123
Lactate clearance is lower in patients with active bleeding than in those without active bleeding. Serum lactate is an indicator of the hemodynamic status. Global tissue hypoperfusion and hypoxia induce lactate production [17]. One possible reason for the significant relationship between lactate clearance and active bleeding is that oxygen delivery to the tissues may recover after a fluid challenge more rapidly in patients without active bleeding than in those with active bleeding. Consequently, serum lactate levels may decline more rapidly in patients without active bleeding. One of the notable findings in the present study is that 1-h lactate clearance may be clinically useful at least as a component of a prediction rule for active bleeding in critically ill patients with UGIB. A previous study demonstrates that 2-h lactate clearance is useful as a marker of resuscitation for patients with severe sepsis [18]. Another finding is that lactate clearance may aid physicians in deciding whether or not emergency endoscopy is required in critically ill patients with UGIB. There is a possibility that the prediction rule may guide the decision whether to perform emergency endoscopy, at least for critically ill patients with UGIB. Among critically ill patients with UGIB, emergency endoscopy can be withheld for those who meet none of the following criteria: low lactate clearance, low platelet, and low systolic blood pressure at presentation. Two previously reported studies sought to identify patients with active UGIB [19, 20]. Both studies derive criteria requiring nasogastric tube aspirates to evaluate the
Intern Emerg Med Table 2 Comparisons of the baseline characteristics and lactate values between patients with and without active bleeding
Active bleeding (n = 48)
No active bleeding (n = 106)
P value
Age (years)
60 (56–63)
62 (60–66)
0.13
Men
38 (79.2)
77 (72.6)
0.43
Glasgow-Blatchford Score
11 (9–13)
12 (11–13)
0.77
Comorbidities Hepatic disease
20 (41.7)
35 (33.0)
0.36
Cardiac failure
4 (8.3)
14 (13.2)
0.59
Hematemesis
37 (77.1)
86 (81.1)
0.67
Melena
35 (72.9)
81 (76.4)
0.69
Syncope
4 (8.3)
11 (10.4)
0.78
Presenting symptoms
Vital signs at ED presentation Systolic blood pressure (mmHg) Heart rate (beats per minute)
99 (94–104)
105 (100–110)
0.05
108 (101–115)
110 (105–112)
0.54
Laboratory features at ED presentation Hemoglobin (g/L)
9.5 (8.7–10.3)
8.8 (7.8–9.8)
0.34
16.3 (11.4–21.4)
19.0 (17.3–21.8)
0.04
International normalized ratio
1.2 (1.1–1.4)
1.1 (1.1–1.2)
0.13
Total bilirubin (mg/dL)
0.8 (0.5–1.3)
0.6 (0.5–0.8)
0.32
28.8 (24.7–35.5)
38.2 (32.9–41.0)
0.11
Initial lactate (mmol/L)
3.1 (2.7–4.9)
3.1 (2.5–3.8)
0.56
Second lactate (mmol/L)
3.0 (2.2–4.1)
2.0 (1.7–2.4)
0.03
Lactate clearance (%)
13 (6–22)
29 (23–34)
\0.001
60 (51–73)
55 (51–62)
0.62
146 (133–179)
143 (129–165)
0.74
1500 (1140–1810)
1500 (1180–1550)
0.89
Platelet count (104/mm3)
Blood urea nitrogen (mg/dL) Lactate
Time to second lactate evaluation (min) Time to endoscopy (min) Amount of Ringer’s acetate solution administered before the second lactate evaluation (mL)
Number of patients who used red cell concentrate before the second lactate evaluation 280 mL
1 (2.1)
7 (6.7)
560 mL
3 (6.3)
4 (3.8)
0.41
Categorical variables were described as frequency (%). Continuous variables were described as median (95 % confidence interval)
Table 3 Adjusted odds ratios of lactate clearance, low blood pressure on admission, and comorbidity of hepatic diseases for active upper gastrointestinal bleeding Adjusted OR
(95 % CI)
Lactate clearance
0.15
(0.04–0.51)
Systolic blood pressure \ 110 mmHg
3.26
(1.35–7.89)
Comorbidity of hepatic diseases
1.67
(0.83–3.41)
OR odds ratio, CI confidence interval
risk of active bleeding. However, nasogastric tube aspiration is uncomfortable for patients. Our decision rule requires two blood withdrawals, which may be more tolerable for patients than nasogastric tube insertion.
The present study does not evaluate whether our rule can predict a need for transfusion, the risk of re-bleeding, or mortality. Thus, this rule may not be used to identify those patients with UGIB who can be safely discharged from the ED. A large-scale study finds that the GBS can accurately identify patients who can be safely discharged from the ED without interventional hemostasis or transfusions [5]. However, patients who require transfusions do not necessarily require interventional hemostasis and vice versa. Indeed, the present study shows that the GBS is not significantly associated with active bleeding in our cohort. Validated criteria for the discrimination of patients with and without active bleeding are required to reduce the numbers of unnecessary emergency endoscopies in patients
123
Intern Emerg Med Table 4 Etiologies of acute upper gastrointestinal bleeding
Active bleeding (n = 48)
No active bleeding (n = 106)
P value
Peptic ulcer
25 (52.1)
66 (62.3)
0.38
Esophagogastric varix
16 (33.3)
22 (20.8)
Mallory–Weiss syndrome
4 (8.3)
9 (8.5)
Neoplasm
2 (4.2)
2 (1.9)
Esophagitis
0 (0.0)
5 (4.7)
Angiodysplasia
1 (2.1)
1 (0.9)
Unspecified
0 (0.0)
1 (0.9)
Variables were described as frequency (%)
Fig. 2 Decision tree for predicting active bleeding in critically ill patients with acute UGIB
with UGIB. The present study suggests that lactate clearance is a promising parameter for such criteria. This study has several limitations. First, this is a singlecenter, retrospective, observational study. The generalizability of our findings is limited. Second, the number of patients analyzed is relatively small, which may cause a potential bias. Third, there is a risk of selection bias in the present study because patients were excluded from the main analyses if they did not receive serial lactate evaluation or emergency endoscopy. Additionally, patients were not included in the study if they were not admitted to the
hospital but were discharged from the ED. However, we believe that hemodynamically unstable patients with UGIB were more likely to receive serial lactate evaluation as well as emergency endoscopy and to be admitted to the hospital than hemodynamically stable patients in the study hospital. We assume that the patients included in the present study who were suitable for analysis are representative of critically ill patients with UGIB. Nevertheless, since there is a risk of selection bias, the prediction rule is less reliable when adapted for patients with UGIB as a whole. Fourth, we could not identify whether blood samples used for analyses were from an artery or vein. A previous study shows that arterial lactate values are slightly lower than venous lactate values [21]. However, we believe that blood sampling sites did not differ between patients with and without active bleeding. The present study shows that there are no significant differences in the time interval from ED presentation to the second lactate evaluation or to emergency endoscopy between patients with and without active bleeding. This would support the assumption that clinical practice is similar for all patients. Finally, the suggested prediction rule has not been prospectively validated. Therefore, due to the risks of selection bias and measurement bias, it is currently inappropriate to use the rule for the management of UGIB patients in a clinical setting. The rule should be validated in a future prospective study to confirm its ability to predict active bleeding. In conclusion, this study indicates that lactate clearance may be associated with the presence or absence of active bleeding in critically ill patients with UGIB. Lactate clearance may be useful as a component of a rule for
Table 5 Predictive performance of the new rule for active bleeding in critically ill patients with UGIB Active bleeding
Sensitivity (%) (95 % CI)
Specificity (%) (95 % CI)
PPV (%) (95 % CI)
NPV (%) (95 % CI)
Yes n = 48
No n = 106
High
47
72
97.9
32.1
39.5
97.1
Low
1
34
(90.2–99.6)
(28.6–32.9)
(36.4–40.2)
(86.6–99.5)
CI confidence interval, NPV negative predictive value, PPV positive predictive value
123
Intern Emerg Med
predicting active bleeding. The derived prediction rule using lactate clearance, platelet count, and SBP at ED presentation shows a high sensitivity in predicting active bleeding in these patients. Prospective studies are required to validate the performance of the rule. Acknowledgments The authors wish to thank all the Emergency Department residents and attending staff at Center Hospital of the National Center for Global Health and Medicine for their kind help in this study. The authors acknowledge the contribution of the medical staff in the Department of Gastroenterology at Center Hospital of the National Center for Global Health and Medicine in the implementation of emergency endoscopy for each patient in the present study. Compliance with ethical standards Conflicts of interest The authors have no conflicts of interest to declare. Statement of human and animal rights The present study was conducted in accordance with the amended Declaration of Helsinki. The Ethics Committee of Center Hospital of the National Center for Global Health and Medicine approved this study. Informed consent The requirement for written informed consent from patients was waived because the study design formed part of the current standard of care in the ED, and patient data were anonymous.
References 1. Barkun AN, Bardou M, Kuipers EJ et al (2010) International consensus recommendations on the management of patients with nonvariceal upper gastrointestinal bleeding. Ann Intern Med 152:101–113. doi:10.7326/0003-4819-152-2-201001190-00009 2. Laursen SB, Hansen JM, Schaffalitzky de Muckadell OB (2012) The Glasgow Blatchford score is the most accurate assessment of patients with upper gastrointestinal hemorrhage. Clin Gastroenterol Hepatol 10(1130–1135):e1. doi:10.1016/j.cgh.2012.06.022 3. Stanley AJ, Dalton HR, Blatchford O et al (2011) Multicentre comparison of the Glasgow Blatchford and Rockall Scores in the prediction of clinical end-points after upper gastrointestinal haemorrhage. Aliment Pharmacol Ther 34:470–475. doi:10.1111/ j.1365-2036.2011.04747.x 4. Blatchford O, Murray WR, Blatchford M (2000) A risk score to predict need for treatment for upper-gastrointestinal haemorrhage. Lancet 356:1318–1321. doi:10.1016/S0140-6736(00)0 2816-6 5. Stanley AJ, Ashley D, Dalton HR et al (2009) Outpatient management of patients with low-risk upper-gastrointestinal haemorrhage: multicentre validation and prospective evaluation. Lancet 373:42–47. doi:10.1016/S0140-6736(08)61769-9 6. Attar A, Sebbagh V, Vicaut E et al (2012) Urgent endoscopy in severe non-variceal upper gastrointestinal hemorrhage: does the
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17. 18.
19.
20.
21.
Glasgow-Blatchford score help endoscopists? Scand J Gastroenterol 47:1086–1093. doi:10.3109/00365521.2012.703237 Zhang Z, Xu X (2014) Lactate clearance is a useful biomarker for the prediction of all-cause mortality in critically Ill patients. Crit Care Med 42:2118–2125. doi:10.1097/CCM.0000000000000405 El-Kersh K, Chaddha U, Sinha RS et al (2015) Predictive role of admission lactate level in critically Ill patients with acute upper gastrointestinal bleeding. J Emerg Med. doi:10.1016/j.jemermed. 2015.04.008 Sacks HS, Chalmers TC, Blum AL et al (1990) Endoscopic hemostasis. An effective therapy for bleeding peptic ulcers. JAMA 264:494–499 Jairath V, Kahan BC, Logan RFA et al (2012) Outcomes following acute nonvariceal upper gastrointestinal bleeding in relation to time to endoscopy: results from a nationwide study. Endoscopy 44:723–730. doi:10.1055/s-0032-1309736 Lin HJ, Wang K, Perng CL et al (1996) Early or delayed endoscopy for patients with peptic ulcer bleeding. A prospective randomized study. J Clin Gastroenterol 22:267–271 Bjorkman DJ, Zaman A, Fennerty MB et al (2004) Urgent vs. elective endoscopy for acute non-variceal upper-GI bleeding: an effectiveness study. Gastrointest Endosc 60:1–8 Forrest J, Finlayson ND, Shearman DJ (1974) Endoscopy in gastrointestinal bleeding. Lancet 304:394–397. doi:10.1016/ S0140-6736(74)91770-X Almenoff PL, Leavy J, Weil MH et al (1989) Prolongation of the half-life of lactate after maximal exercise in patients with hepatic dysfunction. Crit Care Med 17:870–873 Grobman WA, Stamilio DM (2006) Methods of clinical prediction. Am J Obstet Gynecol 194:888–894. doi:10.1016/j.ajog. 2005.09.002 Stiell I, Wells G (1999) Methodologic standards for the development of clinical decision rules in emergency medicine. Ann Emerg Med 33:437–447. doi:10.1016/S0196-0644(99)70309-4 Kreisberg RA (1980) Lactate homeostasis and lactic acidosis. Ann Intern Med 92:227–237 Jones AE, Shapiro NI, Trzeciak S et al (2010) Lactate clearance vs central venous oxygen saturation as goals of early sepsis therapy: a randomized clinical trial. JAMA 303:739–746. doi:10. 1001/jama.2010.158 Adamopoulos AB, Baibas NM, Efstathiou SP et al (2003) Differentiation between patients with acute upper gastrointestinal bleeding who need early urgent upper gastrointestinal endoscopy and those who do not. A prospective study. Eur J Gastroenterol Hepatol 15:381–387. doi:10.1097/01.meg.0000050005.68425.22 Iwasaki H, Shimura T, Yamada T et al (2013) Novel nasogastric tube-related criteria for urgent endoscopy in nonvariceal upper gastrointestinal bleeding. Dig Dis Sci 58:2564–2571. doi:10. 1007/s10620-013-2706-x Bloom BM, Grundlingh J, Bestwick JP, Harris T (2014) The role of venous blood gas in the emergency department: a systematic review and meta-analysis. Eur J Emerg Med 21:81–88. doi:10. 1097/MEJ.0b013e32836437cf
123
EM - ORIGINAL
Early lactate clearance for predicting active bleeding in critically ill patients with acute upper gastrointestinal bleeding: a retrospective study Tomoki Wada1 • Akiyoshi Hagiwara2 • Tatsuki Uemura2 Naoki Yahagi1 • Akio Kimura2
•
Received: 21 August 2015 / Accepted: 9 January 2016 Ó SIMI 2016
Abstract Not all patients with upper gastrointestinal bleeding (UGIB) require emergency endoscopy. Lactate clearance has been suggested as a parameter for predicting patient outcomes in various critical care settings. This study investigates whether lactate clearance can predict active bleeding in critically ill patients with UGIB. This single-center, retrospective, observational study included critically ill patients with UGIB who met all of the following criteria: admission to the emergency department (ED) from April 2011 to August 2014; had blood samples for lactate evaluation at least twice during the ED stay; and had emergency endoscopy within 6 h of ED presentation. The main outcome was active bleeding detected with emergency endoscopy. Classification and regression tree (CART) analyses were performed using variables associated with active bleeding to derive a prediction rule for active bleeding in critically ill UGIB patients. A total of 154 patients with UGIB were analyzed, and 31.2 % (48/ 154) had active bleeding. In the univariate analysis, lactate clearance was significantly lower in patients with active bleeding than in those without active bleeding (13 vs. 29 %, P \ 0.001). Using the CART analysis, a prediction rule for active bleeding is derived, and includes three variables: lactate clearance; platelet count; and systolic blood pressure at ED presentation. The rule has 97.9 % & Tomoki Wada [email protected] 1
Department of Emergency and Critical Care Medicine, The University of Tokyo Hospital, 7-3-1 Hongo, Bunkyo-Ku, Tokyo 113-8655, Japan
2
Department of Emergency Medicine and Critical Care, Center Hospital of the National Center for Global Health and Medicine, 1-21-1 Toyama, Shinjuku-Ku, Tokyo 162-8655, Japan
(95 % CI 90.2–99.6 %) sensitivity with 32.1 % (28.6–32.9 %) specificity. Lactate clearance may be associated with active bleeding in critically ill patients with UGIB, and may be clinically useful as a component of a prediction rule for active bleeding. Keywords Lactate clearance Gastrointestinal hemorrhage Prediction rule Emergency endoscopy
Introduction Upper gastrointestinal bleeding (UGIB) is a common emergency condition that ranges in severity from mild symptoms that do not require any interventions, to lifethreatening presentations with hemorrhagic shock that require emergency endoscopy or surgery. The International Consensus Guidelines of 2010 recommend the stratification of patients’ risks using validated prognostic scales [1]. There are several prognostic scales for UGIB. Among these, the Glasgow-Blatchford Score (GBS) is the most reliable prognostic scale for determining outcomes in patients with UGIB (Table 1) [2–4]. The GBS can be calculated in the absence of endoscopic findings, and it predicts the need for interventional hemostasis, which comprises endoscopy and surgery, or transfusion [4]. The GBS can accurately predict which patients are at low risk, and who will not require interventional hemostasis or transfusion [5]. However, the GBS cannot reliably distinguish between patients with active UGIB who require emergency hemostasis and those who do not require hemostasis, but only transfusion [6]. Serum lactate and its serial change (lactate clearance) have been suggested as a possible predictor of patient outcome in several critical care conditions [7]. A recent study
123
Intern Emerg Med Table 1 Glasgow-Blatchford score [4] Risk factor
Score
Blood urea nitrogen (mg/dL) 18.2–22.4
2
22.4–28.0
3
28.0–70.0
4
C70.0
6
Hemoglobin for men (g/dL) 12.0–12.9
1
10.0–11.9
3
\10.0
6
Hemoglobin for women (g/dL) 10.0–11.9 \10.0
1 6
Systolic blood pressure (mmHg) 100–109
1
90–99
2
\90
3
Other markers Heart rate C 100 (bpm)
1
Melena
1
Syncope
2
Hepatic disease
2
Heart failure
2
bpm beats per minute
finds that an initial lactate value at admission may be useful in predicting mortality in patients with UGIB [8]. However, no study has evaluated whether a lactate value or lactate clearance can predict active bleeding in patients with UGIB. Emergency endoscopy should be performed on patients with active UGIB [9–11]. However, emergency endoscopy can represent overtreatment in patients who do not have active bleeding, and can result in unnecessary intervention and cost for these patients [10–12]. Hence, appropriate selection of patients requiring emergency endoscopy is important. The present study determines whether lactate clearance is associated with active UGIB. The study also investigates whether lactate clearance can be clinically useful as a component of a prediction rule for active bleeding in critically ill patients with UGIB.
Patients and methods Study design and setting This study was a single-center, retrospective, cohort study held in Center Hospital of the National Center for Global
123
Health and Medicine, Tokyo. The hospital is a tertiary, general hospital with 800 beds, and more than 10,000 patients are brought to the emergency department (ED) by ambulance annually. An emergency physician is in charge of the initial assessment and resuscitation for patients with UGIB who are brought to the hospital by ambulance. The treatment policy for these patients in the ED of the hospital is as follows. First, an emergency physician determines whether a patient is hemodynamically unstable and whether this is possibly related to UGIB at ED presentation. For a patient who is hemodynamically unstable, the emergency physician collects blood samples and starts administering 20–40 mL/kg of Ringer’s acetate solution (fluid challenge). Hemodynamic instability is implicitly determined based on the patient’s vital signs and other physical findings such as peripheral coldness, a prolonged capillary refill time, and altered mental status. The possibility of UGIB is evaluated on the basis of the presence of melena or hematemesis. After a fluid challenge for the patient who is hemodynamically unstable with UGIB at the ED presentation, the emergency physician collects blood samples again to evaluate changes in laboratory parameters, including serum lactate. We used GEM Premier 4000 (Instrumentation Laboratory, Bedford, MA, USA) to measure the serum lactate concentration. The emergency physician then consults with a gastroenterologist regarding critically ill patients with UGIB. The gastroenterologist then decides whether a patient requires an emergency endoscopy or not. With a few exceptions, gastroenterologists perform emergency endoscopy for all UGIB patients who are admitted through the ED. If patients admitted to our hospital in the middle of the night, are hemodynamically stable, and are without hematemesis, they might undergo endoscopy during the closest office hours. This treatment policy is the routine of care for UGIB at our ED prior to planning the current study. The Ethics Committee of the hospital approved this retrospective cohort study. The requirement for written informed consent from patients was waived because the study design formed part of the current standard of care in our ED, and patient data were anonymous. Patient enrollment Medical records of patients who were brought by ambulance with a presenting complaint of melena or hematemesis and admitted to the hospital from April 2011 to August 2014 were reviewed. Of these, we selected patients who were admitted to the hospital with a primary diagnosis of UGIB. Inclusion criteria were as follows: systolic blood pressure (SBP) B 109 mmHg or heart rate (HR) C 100 beats per minute (bpm) at ED presentation;
Intern Emerg Med
lactate evaluations performed at least twice during the ED stay; and the receipt of emergency endoscopy within 6 h after the ED presentation. The cutoffs of SBP and HR were determined with reference to the related components of the GBS (Table 1). In calculating a patient’s GBS, one point was added when the SBP was below 110 mmHg or the HR was 100 bpm or higher. Patients were excluded when they met, at least, one of the following criteria: referred from another hospital; initial endoscopy performed more than 6 h after the ED presentation; ‘‘do not attempt resuscitation’’ order was present; or data were missing. Variables In accordance with the GBS, we determined whether patients had hepatic diseases or cardiac failure as comorbidities based on patients’ histories, physical findings, and laboratory test results [4]. The following laboratory parameters were determined at ED presentation: complete blood count, international normalized ratio, total bilirubin, blood urea nitrogen, and serum lactate (initial lactate). Lactate was also evaluated after a fluid challenge (second lactate). Lactate clearance was calculated as follows: [(initial lactate - second lactate)/initial lactate] 9 100 %. The amounts of Ringer’s acetate solution and red cell concentrate used before the second lactate evaluation were identified individually for each patient. Study outcome The primary outcome was active bleeding as detected via endoscopy. Active bleeding was defined as one of the following: peptic ulcers of Forrest classification IA or IB [13], hemorrhagic varices, or other active bleeding for which a gastroenterologist performed a hemostatic procedure. Patients with active bleeding did not include those who received intervention only for lesions with evidence of recent bleeding. The possible etiology of UGIB in each patient was based on an endoscopy report and a discharge abstract written by the gastroenterologist in charge. Statistical analysis In univariate analyses, the Mann–Whitney U test was used to compare continuous variables, while Fisher’s exact test or the Chi-square test was used to compare categorical variables between patients with and without active bleeding. A multivariate logistic regression analysis was then performed to determine factors associated with active bleeding, with adjustment for hepatic diseases, the presence of hypotension (SBP \ 110 mmHg) at ED presentation, and lactate clearance. This was done because hepatic diseases may affect lactate metabolism [14], and may also
affect outcomes owing to the strong relationship between hepatic diseases and the presence of esophagogastric varices. A classification and regression tree (CART) analysis with standard Gini splitting rule was performed to determine a combination of predictive variables, which distinguished between patients with high and low risk of active bleeding [15]. To perform the CART analysis, we used SPSS Decision Trees version 20.0 (IBM Inc., Armonk, NY, USA). In this analysis, only variables with P \ 0.10 in the univariate analyses were included. The process of the CART analysis was as follows. Initially, the most powerful variable to separate patients into high- and low-risk groups was chosen. If the variable was numerical, an optimal cutoff point was identified. Then, patients were separated into two groups in accordance with the variable. In each group, patients could be further divided into two subgroups with a variable chosen in the same way as that in the first separation. The CART analysis repeated the process, and eventually generated a decision tree model. Tenfold cross validation was used to develop a stable decision tree model. Because problems of overfitting occur if the number of outcome events is too small per predictive variables in a model, we pruned the tree until the number of variables in the model fell below one-tenth of the outcome events (i.e., the number of patients with active bleeding in the present study) [16]. The predictive performance of the model for active bleeding was evaluated by calculating sensitivity, specificity, positive predictive value, and negative predictive value. All statistical tests were two tailed, and factors were considered significant for P \ 0.05. All statistical calculations were performed using SPSS Statistics version 20.0 (IBM Inc., Armonk, NY, USA).
Results The patient selection process is illustrated in Fig. 1, 154 patients were included in the analysis. Active UGIB is found in 31.2 % (48/154) patients (Table 2). Systolic blood pressure at ED presentation tends to be lower (99 vs. 105 mmHg, P = 0.05), and platelet counts are also significantly lower (16.3 vs. 19.0 9 104/mm3, P = 0.04) in patients with active bleeding than in those without active bleeding. Second lactate values are significantly higher (3.0 vs. 2.0 mmol/L, P = 0.03), and lactate clearance is significantly lower (13 vs. 29 %, P \ 0.001), in patients with active bleeding than in those without active bleeding (Table 2). There are no significant differences between patients with and without active bleeding in the time interval between the patients arriving at the ED and second
123
Intern Emerg Med Fig. 1 Patient selection process. Asterisk Critically ill patients were defined as those with systolic blood pressure B 109 mmHg or heart rate C 100 beats per minute at ED presentation. ED emergency department, UGIB upper gastrointestinal bleeding
measurements of the serum lactate level (60 vs. 55 min, P = 0.62). This is also true for the time interval between patient arrival at the ED and emergency endoscopy (146 vs. 143 min, P = 0.74) (Table 2). The multivariate logistic regression analysis reveals that higher lactate clearance is associated with a reduction in the risk of active bleeding in patients with UGIB (adjusted odds ratio: 0.15; 95 % confidence interval [CI] 0.04–0.51) (Table 3). Possible etiologies of UGIB in the patients are described in Table 4. There are no significant differences in the distribution of etiologies between the active and non-active bleeding groups (P = 0.38). The CART analysis for active bleeding was performed with the following variables: SBP at ED presentation, platelet count, second lactate values, and lactate clearance. Consequently, a prediction rule was developed and comprised lactate clearance, platelet count, and SBP at ED presentation (Fig. 2). The rule has 97.9 % sensitivity with 32.1 % specificity for active bleeding (Table 5).
Discussion The present study suggests that high lactate clearance is associated with a reduced risk of active bleeding in critically ill patients with UGIB. Additionally, as a component of a prediction rule for patients with UGIB, lactate clearance may be clinically useful to identify patients without active bleeding.
123
Lactate clearance is lower in patients with active bleeding than in those without active bleeding. Serum lactate is an indicator of the hemodynamic status. Global tissue hypoperfusion and hypoxia induce lactate production [17]. One possible reason for the significant relationship between lactate clearance and active bleeding is that oxygen delivery to the tissues may recover after a fluid challenge more rapidly in patients without active bleeding than in those with active bleeding. Consequently, serum lactate levels may decline more rapidly in patients without active bleeding. One of the notable findings in the present study is that 1-h lactate clearance may be clinically useful at least as a component of a prediction rule for active bleeding in critically ill patients with UGIB. A previous study demonstrates that 2-h lactate clearance is useful as a marker of resuscitation for patients with severe sepsis [18]. Another finding is that lactate clearance may aid physicians in deciding whether or not emergency endoscopy is required in critically ill patients with UGIB. There is a possibility that the prediction rule may guide the decision whether to perform emergency endoscopy, at least for critically ill patients with UGIB. Among critically ill patients with UGIB, emergency endoscopy can be withheld for those who meet none of the following criteria: low lactate clearance, low platelet, and low systolic blood pressure at presentation. Two previously reported studies sought to identify patients with active UGIB [19, 20]. Both studies derive criteria requiring nasogastric tube aspirates to evaluate the
Intern Emerg Med Table 2 Comparisons of the baseline characteristics and lactate values between patients with and without active bleeding
Active bleeding (n = 48)
No active bleeding (n = 106)
P value
Age (years)
60 (56–63)
62 (60–66)
0.13
Men
38 (79.2)
77 (72.6)
0.43
Glasgow-Blatchford Score
11 (9–13)
12 (11–13)
0.77
Comorbidities Hepatic disease
20 (41.7)
35 (33.0)
0.36
Cardiac failure
4 (8.3)
14 (13.2)
0.59
Hematemesis
37 (77.1)
86 (81.1)
0.67
Melena
35 (72.9)
81 (76.4)
0.69
Syncope
4 (8.3)
11 (10.4)
0.78
Presenting symptoms
Vital signs at ED presentation Systolic blood pressure (mmHg) Heart rate (beats per minute)
99 (94–104)
105 (100–110)
0.05
108 (101–115)
110 (105–112)
0.54
Laboratory features at ED presentation Hemoglobin (g/L)
9.5 (8.7–10.3)
8.8 (7.8–9.8)
0.34
16.3 (11.4–21.4)
19.0 (17.3–21.8)
0.04
International normalized ratio
1.2 (1.1–1.4)
1.1 (1.1–1.2)
0.13
Total bilirubin (mg/dL)
0.8 (0.5–1.3)
0.6 (0.5–0.8)
0.32
28.8 (24.7–35.5)
38.2 (32.9–41.0)
0.11
Initial lactate (mmol/L)
3.1 (2.7–4.9)
3.1 (2.5–3.8)
0.56
Second lactate (mmol/L)
3.0 (2.2–4.1)
2.0 (1.7–2.4)
0.03
Lactate clearance (%)
13 (6–22)
29 (23–34)
\0.001
60 (51–73)
55 (51–62)
0.62
146 (133–179)
143 (129–165)
0.74
1500 (1140–1810)
1500 (1180–1550)
0.89
Platelet count (104/mm3)
Blood urea nitrogen (mg/dL) Lactate
Time to second lactate evaluation (min) Time to endoscopy (min) Amount of Ringer’s acetate solution administered before the second lactate evaluation (mL)
Number of patients who used red cell concentrate before the second lactate evaluation 280 mL
1 (2.1)
7 (6.7)
560 mL
3 (6.3)
4 (3.8)
0.41
Categorical variables were described as frequency (%). Continuous variables were described as median (95 % confidence interval)
Table 3 Adjusted odds ratios of lactate clearance, low blood pressure on admission, and comorbidity of hepatic diseases for active upper gastrointestinal bleeding Adjusted OR
(95 % CI)
Lactate clearance
0.15
(0.04–0.51)
Systolic blood pressure \ 110 mmHg
3.26
(1.35–7.89)
Comorbidity of hepatic diseases
1.67
(0.83–3.41)
OR odds ratio, CI confidence interval
risk of active bleeding. However, nasogastric tube aspiration is uncomfortable for patients. Our decision rule requires two blood withdrawals, which may be more tolerable for patients than nasogastric tube insertion.
The present study does not evaluate whether our rule can predict a need for transfusion, the risk of re-bleeding, or mortality. Thus, this rule may not be used to identify those patients with UGIB who can be safely discharged from the ED. A large-scale study finds that the GBS can accurately identify patients who can be safely discharged from the ED without interventional hemostasis or transfusions [5]. However, patients who require transfusions do not necessarily require interventional hemostasis and vice versa. Indeed, the present study shows that the GBS is not significantly associated with active bleeding in our cohort. Validated criteria for the discrimination of patients with and without active bleeding are required to reduce the numbers of unnecessary emergency endoscopies in patients
123
Intern Emerg Med Table 4 Etiologies of acute upper gastrointestinal bleeding
Active bleeding (n = 48)
No active bleeding (n = 106)
P value
Peptic ulcer
25 (52.1)
66 (62.3)
0.38
Esophagogastric varix
16 (33.3)
22 (20.8)
Mallory–Weiss syndrome
4 (8.3)
9 (8.5)
Neoplasm
2 (4.2)
2 (1.9)
Esophagitis
0 (0.0)
5 (4.7)
Angiodysplasia
1 (2.1)
1 (0.9)
Unspecified
0 (0.0)
1 (0.9)
Variables were described as frequency (%)
Fig. 2 Decision tree for predicting active bleeding in critically ill patients with acute UGIB
with UGIB. The present study suggests that lactate clearance is a promising parameter for such criteria. This study has several limitations. First, this is a singlecenter, retrospective, observational study. The generalizability of our findings is limited. Second, the number of patients analyzed is relatively small, which may cause a potential bias. Third, there is a risk of selection bias in the present study because patients were excluded from the main analyses if they did not receive serial lactate evaluation or emergency endoscopy. Additionally, patients were not included in the study if they were not admitted to the
hospital but were discharged from the ED. However, we believe that hemodynamically unstable patients with UGIB were more likely to receive serial lactate evaluation as well as emergency endoscopy and to be admitted to the hospital than hemodynamically stable patients in the study hospital. We assume that the patients included in the present study who were suitable for analysis are representative of critically ill patients with UGIB. Nevertheless, since there is a risk of selection bias, the prediction rule is less reliable when adapted for patients with UGIB as a whole. Fourth, we could not identify whether blood samples used for analyses were from an artery or vein. A previous study shows that arterial lactate values are slightly lower than venous lactate values [21]. However, we believe that blood sampling sites did not differ between patients with and without active bleeding. The present study shows that there are no significant differences in the time interval from ED presentation to the second lactate evaluation or to emergency endoscopy between patients with and without active bleeding. This would support the assumption that clinical practice is similar for all patients. Finally, the suggested prediction rule has not been prospectively validated. Therefore, due to the risks of selection bias and measurement bias, it is currently inappropriate to use the rule for the management of UGIB patients in a clinical setting. The rule should be validated in a future prospective study to confirm its ability to predict active bleeding. In conclusion, this study indicates that lactate clearance may be associated with the presence or absence of active bleeding in critically ill patients with UGIB. Lactate clearance may be useful as a component of a rule for
Table 5 Predictive performance of the new rule for active bleeding in critically ill patients with UGIB Active bleeding
Sensitivity (%) (95 % CI)
Specificity (%) (95 % CI)
PPV (%) (95 % CI)
NPV (%) (95 % CI)
Yes n = 48
No n = 106
High
47
72
97.9
32.1
39.5
97.1
Low
1
34
(90.2–99.6)
(28.6–32.9)
(36.4–40.2)
(86.6–99.5)
CI confidence interval, NPV negative predictive value, PPV positive predictive value
123
Intern Emerg Med
predicting active bleeding. The derived prediction rule using lactate clearance, platelet count, and SBP at ED presentation shows a high sensitivity in predicting active bleeding in these patients. Prospective studies are required to validate the performance of the rule. Acknowledgments The authors wish to thank all the Emergency Department residents and attending staff at Center Hospital of the National Center for Global Health and Medicine for their kind help in this study. The authors acknowledge the contribution of the medical staff in the Department of Gastroenterology at Center Hospital of the National Center for Global Health and Medicine in the implementation of emergency endoscopy for each patient in the present study. Compliance with ethical standards Conflicts of interest The authors have no conflicts of interest to declare. Statement of human and animal rights The present study was conducted in accordance with the amended Declaration of Helsinki. The Ethics Committee of Center Hospital of the National Center for Global Health and Medicine approved this study. Informed consent The requirement for written informed consent from patients was waived because the study design formed part of the current standard of care in the ED, and patient data were anonymous.
References 1. Barkun AN, Bardou M, Kuipers EJ et al (2010) International consensus recommendations on the management of patients with nonvariceal upper gastrointestinal bleeding. Ann Intern Med 152:101–113. doi:10.7326/0003-4819-152-2-201001190-00009 2. Laursen SB, Hansen JM, Schaffalitzky de Muckadell OB (2012) The Glasgow Blatchford score is the most accurate assessment of patients with upper gastrointestinal hemorrhage. Clin Gastroenterol Hepatol 10(1130–1135):e1. doi:10.1016/j.cgh.2012.06.022 3. Stanley AJ, Dalton HR, Blatchford O et al (2011) Multicentre comparison of the Glasgow Blatchford and Rockall Scores in the prediction of clinical end-points after upper gastrointestinal haemorrhage. Aliment Pharmacol Ther 34:470–475. doi:10.1111/ j.1365-2036.2011.04747.x 4. Blatchford O, Murray WR, Blatchford M (2000) A risk score to predict need for treatment for upper-gastrointestinal haemorrhage. Lancet 356:1318–1321. doi:10.1016/S0140-6736(00)0 2816-6 5. Stanley AJ, Ashley D, Dalton HR et al (2009) Outpatient management of patients with low-risk upper-gastrointestinal haemorrhage: multicentre validation and prospective evaluation. Lancet 373:42–47. doi:10.1016/S0140-6736(08)61769-9 6. Attar A, Sebbagh V, Vicaut E et al (2012) Urgent endoscopy in severe non-variceal upper gastrointestinal hemorrhage: does the
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17. 18.
19.
20.
21.
Glasgow-Blatchford score help endoscopists? Scand J Gastroenterol 47:1086–1093. doi:10.3109/00365521.2012.703237 Zhang Z, Xu X (2014) Lactate clearance is a useful biomarker for the prediction of all-cause mortality in critically Ill patients. Crit Care Med 42:2118–2125. doi:10.1097/CCM.0000000000000405 El-Kersh K, Chaddha U, Sinha RS et al (2015) Predictive role of admission lactate level in critically Ill patients with acute upper gastrointestinal bleeding. J Emerg Med. doi:10.1016/j.jemermed. 2015.04.008 Sacks HS, Chalmers TC, Blum AL et al (1990) Endoscopic hemostasis. An effective therapy for bleeding peptic ulcers. JAMA 264:494–499 Jairath V, Kahan BC, Logan RFA et al (2012) Outcomes following acute nonvariceal upper gastrointestinal bleeding in relation to time to endoscopy: results from a nationwide study. Endoscopy 44:723–730. doi:10.1055/s-0032-1309736 Lin HJ, Wang K, Perng CL et al (1996) Early or delayed endoscopy for patients with peptic ulcer bleeding. A prospective randomized study. J Clin Gastroenterol 22:267–271 Bjorkman DJ, Zaman A, Fennerty MB et al (2004) Urgent vs. elective endoscopy for acute non-variceal upper-GI bleeding: an effectiveness study. Gastrointest Endosc 60:1–8 Forrest J, Finlayson ND, Shearman DJ (1974) Endoscopy in gastrointestinal bleeding. Lancet 304:394–397. doi:10.1016/ S0140-6736(74)91770-X Almenoff PL, Leavy J, Weil MH et al (1989) Prolongation of the half-life of lactate after maximal exercise in patients with hepatic dysfunction. Crit Care Med 17:870–873 Grobman WA, Stamilio DM (2006) Methods of clinical prediction. Am J Obstet Gynecol 194:888–894. doi:10.1016/j.ajog. 2005.09.002 Stiell I, Wells G (1999) Methodologic standards for the development of clinical decision rules in emergency medicine. Ann Emerg Med 33:437–447. doi:10.1016/S0196-0644(99)70309-4 Kreisberg RA (1980) Lactate homeostasis and lactic acidosis. Ann Intern Med 92:227–237 Jones AE, Shapiro NI, Trzeciak S et al (2010) Lactate clearance vs central venous oxygen saturation as goals of early sepsis therapy: a randomized clinical trial. JAMA 303:739–746. doi:10. 1001/jama.2010.158 Adamopoulos AB, Baibas NM, Efstathiou SP et al (2003) Differentiation between patients with acute upper gastrointestinal bleeding who need early urgent upper gastrointestinal endoscopy and those who do not. A prospective study. Eur J Gastroenterol Hepatol 15:381–387. doi:10.1097/01.meg.0000050005.68425.22 Iwasaki H, Shimura T, Yamada T et al (2013) Novel nasogastric tube-related criteria for urgent endoscopy in nonvariceal upper gastrointestinal bleeding. Dig Dis Sci 58:2564–2571. doi:10. 1007/s10620-013-2706-x Bloom BM, Grundlingh J, Bestwick JP, Harris T (2014) The role of venous blood gas in the emergency department: a systematic review and meta-analysis. Eur J Emerg Med 21:81–88. doi:10. 1097/MEJ.0b013e32836437cf
123
