Admission blood glucose predicted haemorrhagic shock in multiple trauma patients

Injury, Int. J. Care Injured 46 (2015) 15–20

Contents lists available at ScienceDirect

Injury journal homepage: www.elsevier.com/locate/injury

Admission blood glucose predicted haemorrhagic shock in multiple trauma patients§ Janett Kreutziger a,*, Andreas Rafetseder b, Simon Mathis a, Volker Wenzel a, Rene´ El Attal c, Stefan Schmid d a

Department of Anaesthesia and Critical Care Medicine, Medical University of Innsbruck, Anichstr. 35, 6020 Innsbruck, Austria Internship, General Hospital of Linz, Krankenhausstr. 9, 4020 Linz, Austria c Department of Trauma Surgery, Medical University of Innsbruck, Anichstr. 35, 6020 Innsbruck, Austria d Department of General and Surgical Intensive Care Medicine, Medical University of Innsbruck, Anichstr. 35, 6020 Innsbruck, Austria b

A R T I C L E I N F O

A B S T R A C T

Article history: Accepted 17 September 2014

Introduction: Admission blood glucose is known to be a predictor for outcome in several disease patterns, especially in critically ill trauma patients. The underlying mechanisms for the association of hyperglycaemia and poor outcome are still not proven. It was hypothesised that hyperglycaemia upon hospital admission is associated with haemorrhagic shock and in-hospital mortality. Methods: Data was extracted from an observational trauma database of the level 1 trauma centre at Innsbruck Medical University hospital. Trauma patients (18 years) with multiple injuries and an Injury Severity Score 17 were included and analysed. Results: In total, 279 patients were analysed, of which 42 patients (15.1%) died. With increasing blood glucose upon hospital admission, the rate of patients with haemorrhagic shock rose significantly [from 4.4% (glucose 4.1–5.5 mmol/L) to 87.5% (glucose >15 mmol/L), p < 0.0001]. Mortality was also associated with initial blood glucose [5.50 mmol/L 8.3%; 5.51–7.50 mmol/L 10.9%, 7.51–10 mmol/L 12.4%; 10.01–15 mmol/L 32.0%; 15.01 mmol/L 12.5%, p = 0.008]. Admission blood glucose was a better indicator for haemorrhagic shock (cut-off 9.4 mmol/L, sensitivity 67.1%, specificity 83.9%) than haemoglobin, base excess, bicarbonate, pH, lactate, or vital parameters. Regarding haemorrhagic shock, admission blood glucose is more valuable during initial patient assessment than the second best predictive parameter, which was admission haemoglobin (cut-off value 6.5 mmol/L (10.4 g/dL): sensitivity 61.3%, specificity 83.9%). Conclusions: In multiple trauma, non-diabetic patients, admission blood glucose predicted the incidence of haemorrhagic shock. Admission blood glucose is an inexpensive, rapidly and easily available laboratory value that might help to identify patients at risk for haemorrhagic shock during initial evaluation upon hospital admission. ß 2014 Elsevier Ltd. All rights reserved.

Keywords: Multiple trauma Admission blood glucose Haemorrhagic shock Outcome

Introduction Admission blood glucose is known to be a predictor for outcome in several disease patterns [1–5]. This phenomenon is well documented in trauma patients [6–10], which seem to be more

§ The study was performed at the Medical University of Innsbruck, Dep. of Anaesthesia and Critical Care Medicine. * Corresponding author. Tel.: +43 512 504 80357; fax: +43 512 504 6780357. E-mail addresses: [email protected] (J. Kreutziger), [email protected] (A. Rafetseder), [email protected] (S. Mathis), [email protected] (V. Wenzel), [email protected] (R. El Attal), [email protected] (S. Schmid).

http://dx.doi.org/10.1016/j.injury.2014.09.018 0020–1383/ß 2014 Elsevier Ltd. All rights reserved.

prone to a poor outcome due to hyperglycaemia than other critically ill patients [11]. In a retrospective analysis, it was described that patients who died in haemorrhagic shock had the highest blood glucose levels upon hospital admission [6]. In that case, post-traumatic hyperglycaemia could be a surrogate parameter indicating threatened vital organ blood flow, which could be employed to make emergency medical and hospital personnel aware of life-threatening haemorrhagic shock. Up to now, a possible association of blood glucose and haemorrhagic shock was not analysed in trauma patients. The hypothesis of this study was that admission blood glucose predicts haemorrhagic shock and mortality in multiple trauma patients.

J. Kreutziger et al. / Injury, Int. J. Care Injured 46 (2015) 15–20

16

1241 patients

Admission to resuscitation area No trauma Monotrauma <18 years of age ISS<17 Trauma to admission > 24 h Burns/scalds Surgery in referring hospital

183 patients 179 patients 105 patients 352 patients 13 patients 2 patients 2 patients

Σ 836 patients 405 patients

Multiple injury patients, ISS≥17 Exclusion due to missing data 122 patients Diabetes mellitus 4 patients

Σ 126 patients 279 patients

Study population

Fig. 1. Flowchart of patient population according to inclusion and exclusion criteria.

Patients and methods The study was conducted at the level 1 trauma centre of Innsbruck Medical University hospital in Austria. Data was collected from a prospectively conducted observational hospital database, which was retrospectively analysed. Due to the observational design of the study, consenting was waived by the Institutional Review Board. All patients being admitted to the trauma bay of the emergency room were screened for inclusion during the study phase from 1 December 2008 to 11 November 2010. Inclusion criteria were: adult post-traumatic patients (18 years) with multiple injuries and a calculated Injury Severity Score (ISS [12] 17 according to the Abbreviated Injury Scale (AIS), up-date 2005 [13]. Exclusion criteria were known pre-injury diagnosis of diabetes mellitus or immunologically compromising diseases or therapies, or incomplete data records. The following parameters were recorded: admission blood glucose (in mmol/L, laboratory or blood gas testing within 20 min after admission, normal range 4.11–5.50 mmol/L), age, gender, trauma mechanism, type and site of injury, injury severity, past medical history, rescue and transport times, pre- and postadmission fluid resuscitation and drug administration, various initial laboratory values, vital parameters pre- and in-hospital, mortality until end of hospitalization, causes of death and survival times. The trauma scores ISS, Revised Trauma Score (RTS [14]), and A Severity Characterization of Trauma (ASCOT [15]), and Trauma and Injury Severity Score (TRISS [16]) were calculated for each patient. Haemorrhagic shock was defined by at least two of the following criteria within the first 12 h following admission: systolic blood pressure <90 mmHg, use of catecholamines to increase perfusion pressure, blood loss >20% of estimated body blood volume, requirement of a mass transfusion, or lactate acidosis (2 mmol/L). Mass transfusion was defined as substitution of at least the amount of the patient’s estimated blood volume with blood products; mortality was hospital-mortality. For statistical analysis, IBM SPSS Statistics (Release 20.0, 2011) was used. Testing for normal distribution was done by Shapiro Wilk test. Mann–Whitney U test or Kruskal–Wallis test were used to compare group differences of linear data. Chi-square and

Fisher’s exact test were performed to detect frequency differences. The association of admission blood glucose or laboratory values with haemorrhagic shock was assessed using univariate and multiple logistic regression analysis (forward and backward models, enter method), controlling for age, gender, injury severity measured by the above trauma scores, fluid resuscitation, rescue times, and laboratory values (in case of blood glucose only). These results were compared by receiver-operating-characteristic (ROC) curve analysis. The Youden Index [17] was calculated for investigation of the threshold of admission blood glucose regarding haemorrhagic shock. Odds ratios, sensitivity and specificity from contingency tables were also calculated. Confidence intervals (CI) in this study were 95% CI. A p-value of 0.01 was deemed to be statistically significant. Results During the study phase, 1241 patients were admitted to trauma bay of the emergency room of the level 1 trauma centre at Innsbruck Medical University hospital in Austria, of which 405 were identified as multiple trauma patients with an ISS 17. Due to missing data (mainly pre-hospital data and data from referral hospitals) 122 patients had to be excluded, thus 279 patients were analysed (Fig. 1). Forty two patients (15.1%) died during hospitalisation (Table 1). All patients suffered from blunt trauma;

Table 1 Characteristics of the study population. N = 279

Median/ number

Quartiles/ percentage

Age

47

31–62

Gender female/male Deaths ISS

57/222 42 29

20.4%/79.6% 15.05 22–38

RTS ASCOT TRISS

7.55 6.8 89.8

5.97–7.84 2.0–35.5 63.3–97.0

Intensive care unit length of stay (days) Hospital length of stay (days)

11 21

4–20 10–35

J. Kreutziger et al. / Injury, Int. J. Care Injured 46 (2015) 15–20

trauma causes were mainly car and motorcycle accidents, skiing accidents and falls from large heights. With increasing blood glucose upon hospital admission, the rate of patients presenting with or developing haemorrhagic shock rose significantly (p < 0.0001 for blood glucose 10.01 mmol/L; Fig. 2). In-hospital mortality was also associated with initial blood glucose level (blood glucose groups: 5.50 mmol/L 8.3%; 5.51– 7.50 mmol/L 10.9%; 7.51–10 mmol/L 12.4%; 10.01–15 mmol/L 32.0%; 15.01 mmol/L 12.5%, p = 0.007). Admission blood glucose was a better predictor for existence or development of haemorrhagic shock than admission haemoglobin, standard base excess, pH, lactate, or vital parameters at the accident site or in-hospital (Table 2). Admission blood glucose was a better indicator for the incidence of haemorrhagic shock (calculated cut-off value 9.4 mmol/L: sensitivity 67.1%, specificity 83.9%, mortality with blood glucose <9.4 mmol/L 22/204 patients = 10.8%, mortality with blood glucose >9.4 mmol/L 20/75 patients = 26.7%) than admission haemoglobin (arbitrary cut-off value 5 mmol/L (8 g/dL): sensitivity 26.7%, specificity 98.1%, arbitrary cut-off value 6.2 mmol/L (10 g/dL): sensitivity 53.3%, specificity 88.2%; calculated cut-off value 6.5 mmol/L (10.4 g/dL): sensitivity 61.3%, specificity 83.9%) or lactate (calculated cut-off value 2.4 mmol/L (21.8 mg/dL): sensitivity 77.0%, specificity 68.3% (Tables 2 and 3). Logistic regression analysis showed that blood glucose (p < 0.0001), haemoglobin (p = 0.001) and the lowest systolic

17

Fig. 2. Rate of haemorrhagic shock in association to admission blood glucose, patients were arbitrarily stratified in the depicted groups.

blood pressure in the emergency department (p = 0.008) predicted haemorrhagic shock; shock prediction by admission blood glucose was independent of age, gender, injury severity measured by TRISS and ASCOT, pre-hospital fluid resuscitation, emergency medical service mission times and the laboratory values. A patient with an

Table 2 Results of important laboratory values and vital parameters and their predictive value and sensitivity and specificity at the cut-off value (Youden index) for haemorrhagic shock. N = 279

Median quartiles

R2 (Nagelkerke)

AUC 95% CI

Cut-off value

Sensitivity at cut-off 95% CI

Specificity at cut-off 95% CI

Blood glucose mmol/L

8.1 6.7–9.6

0.333

0.811 0.742–0.867

9.4

67.1 52.4–77.4.1

83.9 78.1–88.4

Haemoglobin mmol/L (g/dL) (mean, range) n = 269

7.4 (11.9) 2.6–10.4 (4.2–16.7)

0.306

0.795 0.732–0.859

6.5 (10.4)

61.3 48.1–73.1

82.9 78.2–88.5

3.3 5.4 to

0.204

0.742 0.6766–0.817

5.4

61.1 48.1–73.1

83.1 78.3–88.7

Base excess n = 269

1.2

Bicarbonate n = 245

22 20–24

0.182

0.719 0.637–0.802

20

48.4 35.7–61.3

80.4 73.5–85.8

Lactate in mmol/L n = 245

2.18 1.5–3.3

0.213

0.787 0.723–0.850

2.40

77.0 65.5–87.9

68.3 61.0–74.9

pH n = 246

7.32 7.27–7.37

0.195

0.707 0.617–0.787

7.249

45.2 32.7–58.2

88.7 83.0–92.7

Systolic blood pressure in mmHg at accident site (mean)

119 100–140

0.056

0.657 0.565–0.749

125

61.3 48.1–73.1

40.8 33.0–49.1

Heart frequency in counts per minute at accident site

87.5 73–100

0.105

0.716 0.627–0.808

98

55.8 43.0–65.4

77.2 70.4–81.9

Lowest systolic blood pressure in mmHg in emergency department

105 84–130

0.222

0.629 0.517–0.784

85

48.4 35.7–61.3

80.8 72.4–87.2

Table 3 Comparison of critical blood glucose and haemoglobin levels upon admission, and the incidence of haemorrhagic shock (n = 62) together with sensitivity and specificity of these values. Haemorrhagic shock

Admission Admission Admission Admission Admission Admission Admission Admission

blood glucose  9.4 mmol/L (n = 203) blood glucose > 9.4 mmol/L (n = 76) haemoglobin >5 mmol/L (8 g/dL) (n = 254) haemoglobin 5 mmol/L (8 g/dL) (n = 20) haemoglobin > 6.2 mmol/L (10 g/dL) (n = 216) haemoglobin  6.2 mmol/L (10 g/dL) (n = 58) haemoglobin > 6.5 mmol/L (10.4 g/dL) (n = 202) haemoglobin  6.5 mmol/L (10.4 g/dL) (n = 72)

Yes

No

21 41 44 16 29 33 24 38

182 35 210 4 187 25 178 34

(10.3%) (53.9%) (17.3%) (80%) (13.7%) (58.2%) (11.9%) (52.8%)

(89.7%) (46.1%) (82.7%) (20%) (86.3%) (41.8%) (88.1%) (47.2%)

Sensitivity 95% CI

Specificity 95% CI

66.1 52.4–77.4.1 26.7% 16.5–39.9 53.3% 40.2–65.8 61.3 48.1–73.1

83.9 78.1–88.4 98.1% 95.0–99.4 88.2 82.9–92.21 83.9 78.2–88.5

18

J. Kreutziger et al. / Injury, Int. J. Care Injured 46 (2015) 15–20

trauma), one due to hypoxic brain oedema following cardiopulmonary resuscitation, and one due to complications following trauma (pulmonary embolism). Table 4 gives an overview of common laboratory values of patients with and without haemorrhagic shock. Discussion

Fig. 3. ROC analysis of the three best predicting laboratory values for haemorrhagic shock.

Fig. 4. Box plot of important laboratory values in patients with and without haemorrhagic shock.

admission blood glucose  9.4 mmol/L had a 10.2 times increased chance (odds ratio, CI 5.4–19.2) of suffering a haemorrhagic shock than patients with an admission blood glucose below that value (Figs. 3 and 4). In addition, of the patients with an admission blood glucose  9.4 mmol/L, only 14.5% (n = 11) had an admission haemoglobin <5.0 mmol/L (8.0 g/dL), but 54.0% (n = 41) of them had or developed a haemorrhagic shock. Of the 35 patients with admission blood glucose  9.4 mmol/L without haemorrhagic shock, 18 (51.4%) had a moderate to severe traumatic brain injury with an AIS  3, six of them deceased due to traumatic brain injury. In addition, off the 35 patients without haemorrhagic shock, two patients deceased due to cardiogenic shock (in addition to the

This observational study demonstrated that admission blood glucose in multiple trauma patients indicated the presence or development of a haemorrhagic shock during initial in-hospital care more precisely than other laboratory parameters such as admission haemoglobin, lactate, standard base excess, pH, bicarbonate or vital parameters. Furthermore, admission blood glucose predicted the presence or development of haemorrhagic shock quite sensitively and specifically. The sensitivity of admission haemoglobin – a common parameter of clinically evaluating haemorrhage – was distinctly lower, especially when usual clinical limits (e.g., 8 or 10 g/dL) instead of the calculated limit of 10.4 g/dL were used. All other common parameters such as blood pressure, heart rate, base excess, lactate, pH or bicarbonate were even less predictive. Taken together, admission blood glucose is an inexpensive, rapid and reliable indicator of haemorrhagic shock; although its specificity is not very high for haemorrhagic shock, in patients without haemorrhagic shock it identified patients at high risk for poor outcome such as patients with moderate to severe traumatic brain injury, or cardiovascular problems in addition to multiple injuries. Is ‘‘the’’ marker for patients at risk suffering multiple injuries now identified? Unfortunately, that is only half of the story, because several limitations have to be considered: such as two exceptions in this model: one patient with devascularizing liver injury (Moore V, avulsion of the liver from the vena cava); his admission blood glucose was 3.83 mmol/L despite severe haemorrhagic shock. Very probably, liver function in regards of delivery of blood glucose after injury was profoundly restricted in this case. A second patient had a blood glucose of 24.4 mmol/L without any sign of haemorrhagic shock and an injury pattern without relevant blood loss. No diabetes mellitus was diagnosed. It is assumed that this patient received an inadvertent glucose infusion during prehospital care. Also, there may be further reasons for exceptions of this model: patients with poorly controlled diabetes mellitus, insulin or anti-diabetic drug overdose, severely limited liver function, e.g., due to alcohol abuse or hepatitis, severe hypothermia or others [18]. Bearing these exceptions in mind, admission blood glucose is, nevertheless, a very helpful, additional parameter during clinical evaluation, inexpensive and simple to measure, and its value as an independent predictive factor for multiple trauma patients being at risk may be underestimated until now. This seems to be an independent indicator of the physiological stress response to trauma, especially due to acute blood loss. This was also proven by logistic regression results, which showed an independence of shock prediction by admission blood glucose

Table 4 Overview of common laboratory values in patients with and without haemorrhagic shock. N = 279 Median (Quartiles)

Patients without haemorrhagic shock

Patients with haemorrhagic shock

p

Blood glucose mmol/L Haemoglobin mmol/L (g/dL) n = 269 Base excess n = 269 Bicarbonate n = 245 Lactate in mmol/L n = 245 pH n = 246 Systolic blood pressure in mmHg at accident site Heart frequency in counts per minute at accident site Lowest systolic blood pressure in mmHg in emergency department

7.56 (6.35–8.71) 12.9 (10.9–14.3) 2.8 ( 4.5 to 0.9) 22.5 (20.4–24.5) 1.97 (1.32–2.76) 7.33 (7.28–7.38) 120 (100–140) 84 (70–97) 110 (90–135)

10.3 (8.4–12.8) 9.9 (8.0–11.9) 5.9 ( 10.2 to 3.1) 19.4 (17.6–22.5) 3.70 (2.42–4.82) 7.28 (7.17–7.34) 110 (80–121) 100 (90–115) 80 (60–85)

<0.0001 <0.0001 <0.0001 <0.0001 <0.0001 <0.0001 0.001 <0.0001 <0.0001

J. Kreutziger et al. / Injury, Int. J. Care Injured 46 (2015) 15–20

19

Table 5 Comparison of haemorrhagic shock and mortality rate of the historic with the current trauma population.

Rate of haemorrhagic shock Mortality rate Mortality due to haemorrhagic shock

Current trauma population n = 279

Historic trauma population n = 592

p

n = 62; 22.2% n = 42; 15.1% n = 4; 6.5%

n = 99; 16.7% n = 124; 20.9% n = 42; 42.4%

0.063 0.049 <0.0001

from age, gender, injury severity, pre-hospital fluid resuscitation, emergency medical service mission times and other laboratory values. However, patients with vs. without haemorrhagic shock had comparable injury severity (measured by AIS) of the head, face, neck and thorax, but significantly higher AIS values of the abdomen (liver, spleen) or extremities (pelvis, femur) (p < 0.001). As in previous investigations [6,7], admission blood glucose showed only a low correlation, and no co-linearity to other welldocumented factors for outcome and/or injury severity prediction such as lactate, pH, bicarbonate, standard base excess, and haemoglobin (data not shown) [19–22]. These factors are, at least in part, influenced by the type and amount of fluids being administered pre-hospitally. For example, a multiple trauma patient barely receiving fluids outside the hospital may present upon hospital admission with a treacherously by normal haemoglobin level, while hypoperfusion is already occurring. This phenomenon of non-dilution versus high-dilution of haemoglobin levels with aggressive fluid resuscitation may have probably contributed to the lower predictive power of the traditional tool ‘‘haemoglobin level upon hospital admission’’, as the patients received in a pre-hospital mission time (including referrals) of 103  94 min widely varying levels of fluid resuscitation (range 100– 5500 mL). In contrast, admission blood glucose seems to be relatively independent from pre-hospital fluid administration (no correlation or co-linearity to pre-hospital fluid administration) and may be instead influenced by shock development and its treatment. Therefore, these data give rise to demand admission blood glucose measurement in every trauma patient, and the result should have influence on the initial clinical evaluation and decisions. The study population was comparable to recent trauma populations [2,6,7,19–22,10,23]. However, in a historic study, hospital admission blood glucose was highly predictive for inhospital mortality, [6] which is, in part, in contrast to the current data. There were significantly less surviving patients with high blood glucose levels, but especially in very high hyperglycaemia (>15 mmol/L) the mortality rate was lower than expected. There might be two possible explanations: on the one hand, a detailed comparison with historic data showed that the mortality rate due to haemorrhagic shock was much higher in the historic patient population than in the current. (Table 5) The mechanism for this profoundly decreased mortality due to uncontrollable bleeding within several years may be mainly damage control surgery, improved logistics and coagulation management. On the other hand, the lower patients number in this investigation may cause a lack of power (type 2 error) and yielding statistical significance based on a relatively high vs. low mortality rate is easier to achieve. The main difference between both populations were the amount of blood products administered to patients with haemorrhagic shock (mean sum of blood products: historic population 27.6, median 20, (4–150), current population mean 18.4, median 11 (range 0–111), p < 0.0001). Instead, in the current population coagulation factors such as fibrinogen and prothrombin complex were administered. Limitations of this study were the relatively small patient number, and its single centre design. Especially the missing significance of the association of admission blood glucose and mortality as well as the very low mortality rate due to haemorrhagic shock might have been biased by the small patient population. Unfortunately, many patients had to be excluded due

to missing data. Mainly, missing vital parameters from the accident site were reasons for exclusion (no TRISS calculation possible). Therefore, a bias due to missing data seems implausible. The limited inclusion criteria (adult multiple injured patients with an ISS 17, no diabetes mellitus) renders generalisation to all trauma patients impossible and further data are needed to implement blood glucose measurements eventually in evaluation algorithms upon hospital admission of all trauma patients. Furthermore, hospital admission blood glucose represents only a very short time window within the course of treatment of multiple trauma patients. Also, the dynamics of shock, the impact of ‘‘second hits’’ such as several surgical procedures or miscellaneous complications were not analysed. In addition, ongoing, untreated hyperglycaemia following multiple injury, the so called ‘‘diabetes of injury’’ [10,24,25], may have more influence on outcome, although all patients with blood glucose levels > 8.3 mmmol/L (150 mg/dL) received insulin in the intensive care unit. In addition, multiple injury patients are a very heterogeneous population in regards to injury patterns, severity, age, past medical history, and other miscellaneous factors. Up to now, it is unknown, how pre-traumatic food and drink intake influences post-traumatic blood glucose levels. It is conceivable that, e.g., long and intensive physical exercise in a cool environment such as ski touring may lead to exhaustion of glycogen reserves and lower blood glucose levels following severe trauma. In contrast, a high pre-trauma glucose intake may lead to the contrary. Both extremes may explain in parts the imprecision of the prediction model. However, it is highly improbable that all patients with haemorrhagic shock had high glucose intakes before trauma. In addition, down regulation of gut circulation during shock with consequent decreased digestion and absorption is a well-known phenomenon that may counteract a high glucose uptake from the gut during shock. Conclusions In multiple trauma, non-diabetic patients, admission blood glucose predicted the incidence of haemorrhagic shock. Admission blood glucose is an inexpensive, rapidly and easily available laboratory value that might help to identify patients at risk for haemorrhagic shock during initial evaluation upon hospital admission. Key messages  Admission blood glucose predicted haemorrhagic shock.  Admission blood glucose identified high risk patients.  Admission blood glucose is superior to any other clinical parameter upon hospital admission in identifying high risk patients, especially in patients with haemorrhagic shock. Authors’ contribution JK was involved in conceiving the study, trial design, statistical analyses, and was also responsible for the paper as a whole. Data collection was done by AR, SM, RA and SS. VW, RA and SS were responsible for quality control. The manuscript was drafted by VW and SS. All authors contributed substantially to manuscript revision.

20

J. Kreutziger et al. / Injury, Int. J. Care Injured 46 (2015) 15–20

Conflict of interest statement All authors state that they have no conflict of interest regarding the content discussed in this manuscript. The study was supported by departmental resources only. Acknowledgments The authors thank Dr. Tobias Kastenberger for his help in data editing. References [1] Laird AM, Miller PR, Kilgo PD, Meredith JW, Chang MC. Relationship of early hyperglycemia to mortality in trauma patients. J Trauma 2004;56:1058–62. [2] Yendamuri S, Fulda GJ, Tinkoff GH. Admission hyperglycemia as a prognostic indicator in trauma. J Trauma 2003;55:33–8. [3] Van den Berghe G, Wilmer A, Hermans G, Meersseman W, Wouters PJ, Milants I, et al. Intensive insulin therapy in the medical ICU. N Engl J Med 2006; 354:449–61. [4] Capes SE, Hunt D, Malmberg K, Pathak P, Gerstein HC. Stress hyperglycemia and prognosis of stroke in nondiabetic and diabetic patients: a systematic overview. Stroke 2001;32:2426–32. [5] Capes SE, Hunt D, Malmberg K, Gerstein HC. Stress hyperglycaemia and increased risk of death after myocardial infarction in patients with and without diabetes: a systematic overview. Lancet 2000;355:773–8. [6] Kreutziger J, Wenzel V, Kurz A, Constantinescu MA. Admission blood glucose is an independent predictive factor for hospital mortality in polytraumatised patients. Intensive Care Med 2009;35:1234–9. [7] Kreutziger J, Schlaepfer J, Wenzel V, Constantinescu MA. The role of admission blood glucose in outcome prediction of surviving patients with multiple injuries. J Trauma 2009;67:704–8. [8] Jeremitsky E, Omert LA, Dunham CM, Wilberger J, Rodriguez A. The impact of hyperglycemia on patients with severe brain injury. J Trauma 2005;58:47–50. [9] Kinoshita K, Kraydieh S, Alonso O, Hayashi N, Dietrich WD. Effect of posttraumatic hyperglycemia on contusion volume and neutrophil accumulation after moderate fluidpercussion brain injury in rats. J Neurotrauma 2002;19:681–92. [10] Bochicchio GV, Sung J, Joshi M, Bochicchio K, Johnson SB, Meyer W, et al. Persistent hyperglycemia is predictive of outcome in critically ill trauma patients. J Trauma 2005;58:921–4.

[11] Vogelzang M, Nijboer JM, van der Horst IC, Zijlstra F, ten Duis HJ, Nijsten MW. Hyperglycemia has a stronger relation with outcome in trauma patients than in other critically ill patients. J Trauma 2006;60:873–7. discussion 878–9. [12] Baker SP, O’Neill B, Haddon Jr W, Long WB. The injury severity score: a method for describing patients with multiple injuries and evaluating emergency care. J Trauma 1974;14:187–96. [13] The Abbreviated Injury Scale. Update 2005 association for the advancement of automotive medicine; 1990, http://www.carcrash.org. [14] Champion HR, Sacco WJ, Copes WS, Gann DS, Gennarelli TA, Flanagan ME. A revision of the trauma score. J Trauma 1989;29:623–9. [15] Champion HR, Copes WS, Sacco WJ, Lawnick MM, Bain LW, Gann DS, et al. A new characterization of injury severity. J Trauma 1990;30:539–45. discussion 545–6. [16] Boyd CR, Tolson MA, Copes WS. Evaluating trauma care: the TRISS method. Trauma score and the injury severity score. J Trauma 1987;27:370–8. [17] Youden WJ. Index for rating diagnostic tests. Cancer 1950;3:32–5. [18] Wouters M, Posma RA1, van der Weerd L, van Putten TS, Wendt KW, Nijsten MW. Incidence, causes and consequences of early hypoglycaemia in severe trauma patients. Paris: Poster-abstract at the ESICM; 2013. [19] Maegele M, Lefering R, Yucel N, Tjardes T, Rixen D, Paffrath T, et al. Early coagulopathy in multiple injury: an analysis from the German Trauma Registry on 8724 patients. Injury 2007;38:298–304. [20] FitzSullivan E, Salim A, Demetriades D, Asensio J, Martin MJ. Serum bicarbonate may replace the arterial base deficit in the trauma intensive care unit. Am J Surg 2005;190:941–6. [21] Rixen D, Raum M, Bouillon B, Lefering R, Neugebauer E. Arbeitsgemeinschaft polytrauma of the Deutsche gesellschaft fur unfallchirurgie. Base deficit development and its prognostic significance in posttrauma critical illness: an analysis by the trauma registry of the Deutsche gesellschaft fur unfallchirurgie. Shock 2001;15:83–9. [22] Rixen D, Siegel JH. Metabolic correlates of oxygen debt predict posttrauma early acute respiratory distress syndrome and the related cytokine response. J Trauma 2000;49:392–403. [23] Yucel N, Lefering R, Maegele M, Lehnen DM, Kawel N, Riess P, et al. Trauma associated severe hemorrhage (TASH)-score: probability of mass transfusion as surrogate for life threatening hemorrhage after multiple trauma. J Trauma 2006;60:1228–36. discussion 1236–7. [24] McCowen KC, Malhotra A, Bistrian BR. Stress-induced hyperglycemia. Crit Care Clin 2001;17:107–24. [25] Thorell A, Nygren J, Ljungqvist O. Insulin resistance: a marker of surgical stress. Curr Opin Clin Nutr Metab Care 1999;2:69–78.