International Journal of Nursing Studies 46 (2009) 753–758

Contents lists available at ScienceDirect

International Journal of Nursing Studies

journal homepage: www.elsevier.com/ijns

Intensive insulin therapy on infection rate, days in NICU, in-hospitalmortality and neurological outcome in severe traumatic brain injurypatients: A randomized controlled trial

Yang Menga,1, Guo Qingjieb,1,*, Zhang Xiangtonga, Sun Shuganga, Wang Yaohuaa,Zhao Liweia, Hu Enxia, Li Changyua

a Department of Emergency Neurosurgery and Neurological Intensive Care Unit, The First Affiliated Hospital of HARBIN Medical University, Chinab Department of Neurosurgery, The First Affiliated Hospital of HARBIN Medical University, China

A R T I C L E I N F O

Article history:

Received 10 February 2008

Received in revised form 30 December 2008

Accepted 10 January 2009

Keywords:

Intensive insulin therapy

Severe brain trauma

Hyperglycemia

Nursing

A B S T R A C T

Objectives: Evaluate the impact of an intensive insulin therapy and conventional glucose

control protocol during staying in neurological intensive care unit (NICU) on infection rate,

days in NICU, in-hospital mortality and long-term neurological outcome in severe

traumatic brain injury (TBI) patients.

Methods: A total of 240 patients with severe TBI (GCS score 3–8) admitted to NICU

were prospectively enrolled and randomly assigned either to conventional insulin

therapy or to intensive insulin therapy. Patients in intensive glucose control group

(n = 121) received continuous insulin infusion to maintain glucose levels between

4.4 mmol/l (80 mg/dl) and 6.1 mmol/l (110 mg/dl). Patients in the conventional

treatment group (n = 119) were not given insulin unless glucose levels were greater

than 11.1 mmol/l (200 mg/dl). Both groups were treated with insulin infusion to

maintain normoglycemia after leaving NICU. Comparison was made against conven-

tional insulin therapy using a randomized trial design. The primary outcomes is the

mortality rate at 6 months follow-up. The second outcomes including ICU infection

rate, duration of ICU stay, in-hospital mortality rate and neurologic outcome at 6

months follow-up.

Results: There was no significant difference in gender (66% vs. 67% male), age (46 � 11

years vs. 45 � 10 years), APACHE II score (30 vs. 29), TISS-28 score (47 vs. 46), and Glasgow

Coma Score (GCS, 5.3 vs. 5.3) between the two groups. Overall mortality rates at 6 months

follow-up were similar in the 2 groups (61 of 117, 52.1% vs. 62 of 116, 53.4%; P = 0.8). The

infection rate during the study was significantly higher in patients who received

conventional insulin therapy than that in patients who received intensive insulin therapy

(46.2% vs. 31.4%; P < 0.05). The days stay in NICU was shorter in intensive insulin

control group than that in conventional therapy group [4.2 days vs. 5.6 days (medians)

P < 0.05]. The in-hospital mortality during the study was similar in conventional and

intensive therapy groups (34 of 119, 28.6% vs. 35 of 121, 28.9% in the conventional and

intensive insulin therapy groups; P = 0.85). The neurologic outcome according to Glasgow

Outcome Score (GOS) at 6 months (GOS 5 and 4) was better in the intensive insulin

therapy group (34 of 117, 29.1%) than that in the conventional therapy group (26 of 116,

22.4%, P < 0.05).

* Corresponding author.

E-mail address: guoqingjie188@yahoo.com.cn (Q. Guo).1 These authors contributed equally to the work.

0020-7489/$ – see front matter � 2009 Elsevier Ltd. All rights reserved.

doi:10.1016/j.ijnurstu.2009.01.004

Conclusions: Mortality rates at 6 months follow-up are not affected by intensive glucose

control in patients with severe TBI. Intensive insulin therapy decreases infection rate and

days in NICU and improves the neurological outcome at 6 months follow-up, while has no

obvious influence on in-hospital mortality of severe TBI patients.

� 2009 Elsevier Ltd. All rights reserved.

M. Yang et al. / International Journal of Nursing Studies 46 (2009) 753–758754

What is already known about the topic?

� H

yperglycemia and insulin resistance are common insevere traumatic brain injury patients. � N ormalization of blood glucose with intensive insulin

therapy can decrease mortality and morbidity incritically ill patients.

What this paper adds

� T

his is the original study attempt to elucidate the effectsof intensive insulin therapy during stay in NICU on theinfection rate, days in NICU, infection rate, neurologicaloutcome and mortality in severe traumatic brain injurypatients. � N ormalization of blood glucose should take into con-

sideration as early as possible in severe traumatic braininjury patients based on this research.

1. Introduction

Hyperglycemia at the time of brain injury such asischemic stroke, cerebral hemorrhage, or cerebral traumais long known to be associated with increased morbidityand mortality (Capes et al., 2001; Walia and Sutcliffe, 2002;Bhalla et al., 2003). It is associated with a worse neurologicoutcome after acute severe brain trauma and is con-tributed to brain tissue acidosis in patients with majorhead injury (Zygun et al., 2004). In critically ill patients,hyperglycemia and insulin resistance are common, even ifthey have not previously had diabetes. Normalization ofblood glucose levels with intensive insulin therapyimproves the prognosis for such patients has beendocumented in some researches (Van den Berghe et al.,2001, 2006). It was found that intensive insulin therapyafter surgery reduced morbidity and death in critically illpatients, including patients with acute brain injury in theseresearches. One study of patients under postoperativecardiac intensive care demonstrated lower morbidity(including infection rate) and mortality in patientsreceiving intensive insulin therapy than in patientsreceiving conventional insulin therapy (Gandhi et al.,2007). However, currently no prospective trials conductedin patients with severe brain trauma (Glasgow Coma Scale3–8) admitted to neurosurgical intensive care unit haveevaluated the influence of the 2 insulin regimens. Todetermine the applicability of intensive insulin therapy insevere brain trauma patients in the neurological intensivecare unit (NICU) setting, we designed this prospectiverandomized trial. Herein, we compare conventional insulintherapy and intensive insulin therapy in patients withsevere brain trauma admitted to a neurosurgical ICU. The

primary outcomes is the mortality rate at 6 months follow-up. The second outcomes included ICU infection rate,duration of ICU stay, in-hospital mortality rate andneurologic outcome at 6 months follow-up.

2. Methods

2.1. Study population

Adults with severe brain trauma (Glasgow Coma Scale3–8) who were admitted to an neurological intensive careunit between January 10, 2003, and January 18, 2007, wereeligible for enrollment in the study after written informedconsent had been obtained from the closest familymember. Patients who were participating in other trials,moribund or whom there were do-not-resuscitate orderswere excluded (Bilotta et al., 2008; Van den Berghe et al.,2001). The protocol was approved by the institutionalreview board. The numbers of patients in this study areshown in Fig. 1.

2.2. Study design and data collection

At the time of admission to the neurosurgical intensivecare unit, patients were randomly assigned to receiveeither intensive or conventional insulin therapy. Assign-ments to the treatment groups were made with the use ofsealed envelopes. We stratified GCS scores of severelybrain-injured patients into two subgroups: those with GCSscores of 3–5 and 6–8, respectively (Table 1), and werebalanced with the use of permuted blocks of 10. In theconventional-treatment group, a continuous infusion ofinsulin (50 IU of Actrapid HM [Novo Nordisk, Copenhagen,Denmark] in 50 ml of 0.9 percent sodium chloride), withthe use of a pump (Perfusor-FM, B. Braun, Melsungen,Germany), was started only if the blood glucose levelexceeded 215 mg per deciliter (Shangraw et al., 1989;Mizock, 1995), and the infusion was adjusted to maintainthe level at a value between 180 and 200 mg per deciliter(10.0 and 11.1 mmol per liter). In the intensive-treatmentgroup, an insulin infusion was started if the blood glucoselevel exceeded 110 mg per deciliter, and the infusion wasadjusted to maintain normoglycemia (80–110 mg perdeciliter [4.4–6.1 mmol per liter]). The maximal dose ofinsulin was arbitrarily set at 50 IU per hour. When thepatient was discharged from the neurological intensivecare unit, a conventional approach was adopted (main-tenance of blood glucose at a level between 180 and200 mg per deciliter). Adjustments of the insulin dose werebased on measurements of whole-blood glucose inundiluted arterial blood, performed at one to four-hourintervals with the use of a glucose analyzer. More frequentblood glucose measurements were performed whenever

Fig. 1. Patients in the study. DNR denotes do not resuscitate.

M. Yang et al. / International Journal of Nursing Studies 46 (2009) 753–758 755

the attending nurse considered them necessary andwhenever there had been a steep rise or fall in the bloodglucose level on the previous reading. Blood glucose levelsthat were measured on admission and daily in the morningduring the study, and hypoglycemic events (defined asblood glucose levels of �40 mg per deciliter [2.2 mmol perliter]) were analyzed. The dose was adjusted according to astrict algorithm by a team of intensive care nurses, assistedby a study physician who was not involved in the clinicalcare of the patients. On admission, all patients were fedcontinuously with intravenous glucose (200–300 g per24 h). Thereafter parenteral and/or enteral nutrition wasinitiated as early as possible according to a standardizedschedule (Souba, 1997).

At baseline, data on demographic and clinical char-acteristics of the patients were obtained, including

information necessary to determine the severity of illnessand the use of intensive care resources (Table 1). Thesedata were scored according to the Acute Physiology andChronic Health Evaluation [APACHE II, calculated onInternet. (http://www.icumedicus.com/icu_scores/apa-che.php)] system (Knaus et al., 1985), and simplifiedTherapeutic Intervention Scoring System-28 (TISS-28,Keene and Cullen, 1983; Miranda et al., 1996) with highervalues indicating more severe illness and more therapeuticinterventions, respectively.

2.3. Outcome measures

The primary outcomes is the mortality rate at 6 monthsfollow-up. The second outcomes including ICU infection rate(pneumonia, sepsis, and urinary and wound infections),

Table 1

Base-line character of the two study groups.

Variable Intensive insulin

therapy, N = 121

Conventional insulin

therapy, N = 119

Age, years 46 � 11 45 � 10

Male (%) 66 67

Medical history

Diabetes, n(%) 12(10) 12(10)

Hypertension, n(%) 34(28) 36(30)

Smoking, n(%) 41(34) 46(39)

Diabetes treatment, n(%) 12(10) 12(10)

Insulin only 8(6.6) 6(5.0)

Oral diabetic

medications or both

4(3.4) 6(5.0)

Mean body mass

index (kg/m2)

28 � 5 27 � 5

APACHE II score(median)

First 24 h 30(28–61) 29(26–59)

Second 24 h 26(24–57) 27(24–64)

TISS-28 score (median)

First 24 h 47 46

Second 24 h 43 43

Blood glucose on admission

>110 mg/dl (n, %) 112(92.6) 113(95.0)

>200 mg/dl (n, %) 101(83.5) 97(81.5)

Glasgow Coma Scale, n(%)

3–5 47(38.8) 48(40.3)

6–8 74(61.2) 71(60.0)

M. Yang et al. / International Journal of Nursing Studies 46 (2009) 753–758756

duration of ICU stay, in-hospital mortality rate andneurologic outcome at 6 months follow-up. Infection wasdefined according to the National Nosocomial InfectionSurveillance System (Horan and Gaynes, 2004). The date ofonset and the type of infection was determined by 2consultants for infective disease who were blinded totreatment allocation (Hu EX and Wang YH). To minimize thepossibility of bias in assessing the NICU stay caused bydelays in the transfer of patients to a regular ward because ofthe unavailability of beds, patients were considered to beready for discharge when they no longer needed vital-organsupport and were receiving at least two thirds of theircaloric intake by the normal enteral route or when theywere sent to a ward (Van den Berghe et al., 2006). Physicianson the general wards to which patients were transferredfrom intensive care had no access to the results of bloodglucose testing and were unaware of the study treatmentassignment.

Secondary end point was neurologic outcome andmortality at 6 months follow-up. Neurologic outcome wasmeasured according to the Glasgow Outcome Scale (GOSscale). For the purpose of this analysis, the patients wereseparated into two groups on the basis of their GOS score.The favorable category included patients who made a goodrecovery or who were moderately disabled (GOS = 5 or 4,respectively). The unfavorable outcome category includedthose who were severely disabled, in a persistentvegetative state, or dead (GOS = 3, 2, and 1, respectively).To evaluate the influence of anesthesia and surgical stress,both the admission and the 24-h-postoperative glucosevalues were considered in all analyses.

Follow-up neurologic outcome data were recorded afterinterviewing the patients and their relatives, attendingphysician, or rehabilitation physician (by telephone whennecessary). These interviews were conducted in theoutpatient department by the physician responsible for6 months follow-up neurologic outcome evaluation (ZhaoLW and Li CY) who were blinded to the patient’s previoustreatment assignment. We considered patients to be lost tofollow-up if we could not contact them within 10 days ofthe initial attempt. If patients returned to our institutionfor care, we reviewed their medical records and confirmedtheir complications.

2.4. Statistical analysis

From January 2000 to December 2002, 214 patientswith severe brain trauma were treated in our department.The overall mortality rate is about 53.5% at 6 monthsfollow-up. Assuming mortality rate is 54%, we calculatedthat a sample size of at least 115 patients for each groupwould be needed to detect a difference of 30% between the2 treatment groups (with [alpha] = 0.05 and [beta] = 0.2).Data are presented as means � SD or as medians withinterquartile ranges, unless otherwise indicated. The effect ofintensive insulin therapy on the days of NICU was assessed byKaplan–Meier analysis (Mantel–Cox log-rank test). Otherdifferences between study groups were compared with theuse of Student’s t-test, the Chi-square test, and the Mann–Whitney U test, when appropriate. All analyses wereperformed on an intention-to-treat basis. A P value < 0.05indicated statistical significance.

3. Results

3.1. Study patients and treatment

Of the 296 patients who were assessed for eligibility,240 were randomly assigned to intensive treatment(n = 121) or conventional treatment (n = 119, Table 1).We excluded 56 randomly assigned patients (30 in theintensive treatment group and 26 in the conventionaltreatment group) from the final intention-to-treat analysesbecause of the reasons mentioned in the ‘‘study popula-tion’’. Among the patients excluded, 32 were moribund (ofthem 26 died within 24 h) and 24 had the do-not-resuscitate disorders. Among the patients who receivedstudy interventions, 4 of 121 patients in the intensivetreatment group and 3 of 119 patients in the conventionaltreatment group were lost to follow-up after beingdischarged from the hospital. Baseline characteristics didnot differ significantly between the 7 patients who werelost to follow-up and the remaining study patients. Table 1provides clinical and demographic characteristics of studypatients who were included in subsequent analyses (117patients in the intensive treatment group and 116 patientsin the conventional treatment group). Baseline character-istics did not differ statistically significantly between the 2study groups. On admission, about 10 percent of thepatients had a history of diabetes, and about 6 percentwere receiving treatment with insulin (Table 1). There are11 episodes of hypoglycemia during this study. 6 (in 4

M. Yang et al. / International Journal of Nursing Studies 46 (2009) 753–758 757

patients) in the intensive control group and 5 (in 3patients) in the conventional group. All the episodes ofhypoglycemia occurred during the insulin injection. Noobvious difference of base-line character of these patientswas found between the two groups. 3 of 4 in intensivecontrol group and 2 of 3 in conventional group died duringhospital stay. No obvious impact on overall outcome wascontributed to the episodes of hypoglycemia.

3.2. Primary outcome

Overall mortality at 6 months follow-up was similar inthe 2 treatment groups [62/116 (53.4%) in the conventionalinsulin therapy group vs. 61/117 (52.1%) in the intensiveinsulin therapy group, P = 0.8].

3.3. Secondary outcomes

The total infection rate during the study was signifi-cantly higher in the conventional insulin therapy groupthan that in the intensive insulin therapy group [55/119(46.2%) vs. 38/121 (31.4%); P < 0.05]. The most frequenttype of infection in both groups was pneumonia (14.9% inthe conventional insulin therapy group and 23.5% in theintensive insulin therapy group) followed by urinaryinfections (9.1% and 11.8%), wound infections (3.3% and5.0%), and sepsis (4.1% and 5.9%).

Mean days stay in NICU in the conventional insulintherapy and intensive insulin control group were 5.6 and4.2 days (median) respectively (P < 0.05).

No difference was found in mortality rate duringhospitalization in the conventional and intensive insulintherapy groups [34/119 (28.6%) vs. 35/121 (28.9%)];P = 0.85).

The neurologic status at 6 months follow-up measuredwith Glasgow Outcome Scale (Glasgow Outcome Scale 5and 4) were better in intensive insulin control group (34 of117, 29.1%) than that in conventional therapy group (26 of116, 22.4%, P < 0.05) (Table 2).

Table 2

Outcome measure in the two groups.

Variable Intensive insulin

therapy, N = 121

Conventional

insulin therapy,

N = 119

P value

Infections, n(%) 38(31.4) 55(46.2) <0.05

Pneumonia 18(14.9) 28(23.5)

Urinary infections 11(9.1) 14(11.8)

Wound infections 4(3.3) 6(5.0)

Sepsis 5(4.1) 7(5.9)

Days in NICU

(median)

4.2(2.5–9.8) 5.6(3.7–12.5) <0.05

Mortality, n(%) 61(52.1)* 62(53.4)* >0.05

In-hospital 35(28.9) 34(28.6) >0.05

Follow-up at

6 months

26(22.2)* 28(24.1)* >0.05

GOS scale at 6 months, n(%)

5 and 4 34(29.1)* 26(22.4)* <0.05

3, 2 and 1 83(70.9)* 90(77.6)* <0.05

Note: Results of ‘‘*’’ was calculated at the bases of 6 months follow-up

patients number (n = 117 in intensive insulin control group and n = 116 in

conventional treatment group).

4. Discussion

Our main finding in this prospective randomized trialis that severe traumatic brain injury patients whoreceived intensive insulin therapy had reduced infectionrates, days in NICU and improved neurological outcomeat 6 months follow-up in comparison with similarpatients who received conventional insulin therapy. Butthere is no significant difference in mortality duringhospitalization and 6 months follow-up between thetwo groups. This negative finding may be due to thestudy population, which are severe traumatic braininjury patients in this study (GCS score 3–8). Theprimary insults of brain is the most important factorin determining the mortality and insulin control was noteffective enough to reduce the mortality rate. Only 11hypoglycaemia episodes were found in 7 patients (2.9%)of the two groups. This demonstrated the safety for theintensive insulin control therapy in severe brain trau-matic patients. In the research conducted by Bilotta et al.(2008), rate of hypoglycaemia was higher than that inthe present study. This may be accounted by thedifferent standard of hypoglycemic episodes. Hypogly-caemia was defined as blood glucose < 80 mg/dl(4.44 mmol/l) by Bilotta et al. (2008), while it wasdefined as blood glucose �40 mg/ml (2.2 mmol/l) byothers (Van den Berghe et al., 2006) and our study.Otherwise, frequent measurement of blood glucose(from 1 to 4 h interval) ensured the discovery ofhypoglycaemia episodes. The results of this studycomplement and extend of those previous publications(Van den Berghe et al., 2006; Gunjan et al., 2007; Sperryet al., 2007; Scalea et al., 2007).

Glucose is the main energy source for the brain.Consideration of brain energy supply after severe braintrauma, glucose was injected with the mixture of insulin.To minimized the up-regulation effects of blood glucoselevel by intravenous injection of glucose, insulin wasadded at a ratio of 4:1 (glucose (g):insulin (IU)).

Traumatic brain injury is associated with a stressresponse that includes hyperglycemia, which has beenshown to worsen neurological outcome during cerebralischemia and hypoxia (Lam et al., 1991; Zygun et al.,2004). Early hyperglycemia is a frequent component ofthe stress response to head injury, a significant indicatorof its severity, and a reliable predictor of outcome(Young et al., 1989; Lam et al., 1991; Aristedis andSerafim, 2000). Patients with severe head injury hadsignificantly higher serum glucose levels than did thosewith moderate injury. Patients who subsequently had anunfavorable outcome had significantly higher glucoselevels than did those with a better prognosis. Among thepatients with more severe head injury, a glucose levelgreater than 200 mg/dl was associated with a worseoutcome. In the study conducted by Aristedis andSerafim (2000), glucose level was more than 200 mg/dlin 160 of 184 patients with severe head injury (86.96%).60 of the 160 patients had a poor prognosis (37.5%)(Aristedis and Serafim, 2000). In our study population,glucose level is greater than 200 mg/dl in 198 of 240patients (82.5%) on admission. A poor prognosis is 70.9%

M. Yang et al. / International Journal of Nursing Studies 46 (2009) 753–758758

(83 of 117 patients) in intensive insulin control groupaccording to GOS score (GOS 1–3), and this rate is 77.6%(90 of 116 patients) in conventional insulin therapygroup at 6 months follow-up. The reduction of poorprognosis between the two groups has a significantdifference (P < 0.05).

Early hyperglycemia is a marker of severe physiologicinsult after injury, and that strict glycemic control mayreduce or prevent the infectious complications pre-viously shown to be associated with hyperglycemiaearly after injury (Sperry et al., 2007). It has beenreported that infection rates were decreased withintensive insulin control during cardiac surgery andSAH patients (Gunjan et al., 2007; Bilotta et al., 2007).These results are consistent with our study. The detailedmechanisms which contributed to the effects of inten-sive insulin control on infection need to be studied in thefuture.

Another finding of this study is patients receivedintensive insulin control had an improved neurologicaloutcome at 6 months follow-up according to GOS scale.This effect on neurological outcome can be explained bythe CNS protection of intensive insulin control. Intensiveinsulin therapy protected the CNS, as it reduced mean andmaximal intracranial pressure in patients with isolatedbrain injury (Van den Berghe et al., 2005). The mostcommon complication of severe traumatic brain injurypatients is increased intracranial pressure. The beneficialeffect of intensive insulin control on intracranial pressureoccurred in the presence of similar cerebral perfusionpressures that were achieved with significantly lessnorepinephrine as a vasopressor (Van den Berghe et al.,2005). This is the first randomized controlled studyproviding evidence for an effective metabolic measure toprevent secondary insults after severe traumatic braininjury. There are several potential mechanisms involved,including prevention of glucose toxicity as well as directeffects of insulin independent of glycemic control (Zygunet al., 2004). There also appeared to be a long-term benefitof intensive insulin therapy during intensive care as alarger fraction of survivors after isolated brain injuryrehabilitated to a level of independent living after 12months (Van den Berghe et al., 2005). Although the samplesize of our study was relatively small, the clinical relevanceof this observation, if confirmed in a larger study, isenormous.

Conflict of interest: There was no conflict interest among

authors of this manuscript.

Funding: This research was supported partly by the Nature

Science Foundation of Heilongjiang Province (D200847).

Ethical approval: This research was approved by the Medical

Ethical Committee of the First Affiliated Hospital of Harbin

Medical University.

References

Aristedis, R., Serafim, K., 2000. The influence of hyperglycemia on neu-rological outcome in patients with severe head injury. Neurosurgery46, 335.

Bhalla, A., Tilling, K., Kolominsky-Rabas, P., Heuschmann, P., Megherbi, S.E.,Czlonkowska, A., Kobayashi,A., Mendel,T., Giroud,M.,Rudd, A.,Wolfe,C.,2003. Variation in the management of acute physiological parametersafter ischaemic stroke: a European perspective. Eur. J. Neurol. 10, 25–33.

Bilotta, F., Spinelli, A., Giovannini, F., Doronzio, A., Delfini, R., Rosa, G.,2007. The effect of intensive insulin therapy on infection rate, vasos-pasm, neurologic outcome, and mortality in neurointensive care unitafter intracranial aneurysm clipping in patients with acute subar-achnoid hemorrhage: a randomized prospective pilot trial. J. Neuro-surg. Anesthesiol. 19, 156–160.

Bilotta, F., Caramia, R., Cernak, I., Paoloni, F.P., Doronzio, A., Cuzzone, V.,Santoro, A., Rosa, G., 2008. Intensive insulin therapy after severetraumatic brain injury: a randomized clinical trial. Neurocrit. Care9, 159–166.

Capes, S.E., Hunt, D., Malmberg, K., Pathak, P., Gerstein, H.C., 2001. Stresshyperglycemia and prognosis of stroke in nondiabetic and diabeticpatients: a systematic overview. Stroke 32, 2426–2432.

Gandhi, G.Y., Nuttall, G.A., Abel, M.D., Mullany, C.J., Schaff, H.V., O’Brien,P.C., Johnson, M.G., Williams, A.R., Cutshall, S.M., Mundy, L.M., Rizza,R.A., McMahon, M.M., 2007. Intensive intraoperative insulin therapyversus conventional glucose management during cardiac surgery: arandomized trial. Ann. Intern. Med. 146, 233–243.

Horan, T.C., Gaynes, R.P., 2004. Surveillance of nosocomial infections. In:Mayhall, C.G. (Ed.), Hospital and Infection Control. 3rd ed. LippincottWilliams & Wilkins, Phladelphia, pp. 1659–1702.

Keene, A.R., Cullen, D.J., 1983. Therapeutic Intervention Scoring System:update 1983. Crit. Care Med. 11, 1–3.

Knaus, W.A., Draper, E.A., Wagner, D.P., Zimmerman, J.E., 1985. APACHE II: aseverity of disease classification system. Crit. Care Med. 13, 818–829.

Lam, A.M., Winn, H.R., Cullen, B.F., Sundling, N., 1991. Hyperglycemia andneurological outcome in patients with head injury. J. Neurosurg. 75,545–551.

Miranda, D.R., de Rijk, A., Schaufeli, W., 1996. Simplified TherapeuticIntervention Scoring System: the TISS-28 items—results from a multi-center study. Crit. Care Med. 24, 64–73.

Mizock, B.A., 1995. Alterations in carbohydrate metabolism duringstress:a review of the literature. Am. J. Med. 98, 75–84.

Scalea, T.M., Bochicchio, G.V., Bochicchio, K.M., Johnson, S.B., Joshi, M.,Pyle, A., 2007. Tight glycemic control in critically injured traumapatients. Ann. Surg. 246, 605–610.

Shangraw, R.E., Jahoor, F., Miyoshi, H., Neff, W.A., Stuart, C.A., Herndon,D.N., Wolfe, R.R., 1989. Differentiation between septic and postburninsulin resistance. Metabolism 38, 983–989.

Souba, W.W., 1997. Nutritional support. N. Engl. J. Med. 336, 41–48.Sperry, J.L., Frankel, H.L., Vanek, S.L., Nathens, A.B., Moore, E.E., Maier, R.V.,

Minei, J.P., 2007. Early hyperglycemia predicts multiple organ failureand mortality but not infection. J. Trauma 63, 487–493.

Van den Berghe, G., Wouters, P., Weekers, F., Verwaest, C., Bruyninckx, F.,Schetz, M., Vlasselaers, D., Ferdinande, P., Lauwers, P., Bouillon, R.,2001. Intensive insulin therapy in critically ill patients. N. Engl. J. Med.345, 1359–1367.

Van den Berghe, G., Schoonheydt, K., Becx, P., Bruyninckx, F., Wouters, P.J.,2005. Insulin therapy protects the central and peripheral nervoussystem of intensive care patients. Neurology 64, 1348–1353.

Van den Berghe, G., Wilmer, A., Hermans, G., Meersseman, W., Wouters, P.J.,Milants, I., Van Wijngaerden, E., Bobbaers, H., Bouillon, R., 2006. Inten-sive insulin therapy in the medical ICU. N. Engl. J. Med. 354, 449–461.

Walia, S., Sutcliffe, A.J., 2002. The relationship between blood glucose,mean arterial pressure and outcome after head injury: an observa-tional study. Injury Int. J. Care Injured 33, 339–344.

Young, B., Ott, L., Dempsey, R., Haack, D., Tibbs, P., 1989. Relationshipbetween admission hyperglycemia and neurologic outcome ofseverely brain-injured patients. Ann. Surg. 210, 466–472.

Zygun, D.A., Steiner, L.A., Johnston, A.J., Hutchinson, P.J., Al-Rawi, P.G.,Chatfield, D., Kirkpatrick, P.J., Menon, D.K., Gupta, A.K., 2004. Hyper-glycemia and brain tissue pH after traumatic brain injury. Neurosur-gery 55, 877–881.