Journal of Critical Care 30 (2015) 437.e1–437.e6

Contents lists available at ScienceDirect

Journal of Critical Care

j ourna l homepage: www. jcc journa l .o rg

Insulin infusion therapy in critical care patients: Regular insulin vs short-

acting insulin. A prospective, crossover, randomized, multicenterblind study☆,☆☆

Federico Bilotta, MD, PhD a,⁎, Rafael Badenes, MD b, Simona Lolli, MD a, Francisco Javier Belda, MD b,Sharon Einav, MD c, Giovanni Rosa, MD a

a Department of Anesthesiology, Critical Care and Pain Medicine, Neuroanesthesia and Neurocritical Care, “Sapienza” University of Rome, Rome, Italyb Department of Anesthesiology and Surgical Intensive Care, Hospital Clinic Universitari Valencia, Valencia, Spainc Department of Anesthesiology and General Intensive Care Unit, Shaare Zedek Medical Centre, Jerusalem, Israel

a b s t r a c ta r t i c l e i n f o

☆ Clinical trial registered with: www.ClinicalTrials.gov☆☆ Disclosure: The authors have no conflicts of interestused to support this study.⁎ Corresponding author at: Department of Anesthe

Medicine, “Sapienza” University Rome, Italy. Tel.: +39 33E-mail address: bilotta@tiscali.it (F. Bilotta).

http://dx.doi.org/10.1016/j.jcrc.2014.10.0190883-9441/© 2014 Elsevier Inc. All rights reserved.

Keywords:

Insulin infusion therapyHyperglycemiaHypoglycemiaRegular insulinShort acting insulinCritical care

Introduction: The aim of this multicenter, prospective, randomized, crossover trial is to compare, in critical carepatients receiving insulin infusion therapy (IIT), the pharmacodynamic of Humulin insulin (Hlin), currentlyused as “standard of care,” and Humalog insulin (Hlog), a shorter acting insulin formulation. This was measuredas extent and duration of the carryover effect of insulin treatment, with the latter calculated as ratio betweenblood glucose concentration (BGC) reduction during and after IIT.Materials and methods: Twenty-eight patients treated in an intensive care unit and receiving full nutritional

support were randomly assigned to Hlin or Hlog as first treatment. Insulin was infused at a constant rate in pa-tients presenting with BGC greater than or equal to 180mg/dL (0.04 U/kg per hour) andwas discontinuedwhenBGCwas less than or equal to 140mg/dL (therapeutic BGC drop). Further reductions in BGC after discontinuationof insulin infusionwere recorded (postinfusional BGCdrop). During the study period, whole blood BGCwasmea-sured every 30 minutes. A minimal 6-hour washout interval was maintained between treatments with the 2types of insulin. The primary end point was the extent (calculated as ratio between the therapeutic BGC dropand the postinfusional BGC drop) and duration of the carryover effect.Results: TreatmentwithHlog, as comparedwith Hlin, was associatedwith a less profound carryover effect aswellas a briefer duration of carryover (median, 0.40 vs 0.62; P b .001; median, 1 vs 1.5 hours; P b .001).Conclusions: The use of constant Hlog infusion for IIT, when compared with Hlin at the same dose, is associatedwith a less profound carryover effect on BGC after discontinuation of IIT, a briefer duration of carryover, a fasterBGC drop during infusion, and a quicker BGC rise after discontinuation. These characteristics suggest that Hlog IITmay be preferable for use in critically ill patients.

© 2014 Elsevier Inc. All rights reserved.

1. Introduction

Insulin infusion therapy (IIT) is widely used as “standard of care” totreat hyperglycemia in critical care patients [1,2]. This therapeuticapproach leads to reduced morbidity and to a higher survival rate insome subgroups of critical care patients but is also associated with asubstantial risk of inducing hypoglycemia [3-5]. Various strategieshave been used to minimize the risk of iatrogenic-induced hypoglyce-mia during IIT including wider glycemia target ranges, sliding scale in-sulin titration, increased frequency of blood glucose concentration

(NCT 02165566).. No corporate funds have been

siology, Critical Care and Pain9 33 708 22.

(BGC) measurements, and a higher caloric intake [6-8]. Whether thesetherapeuticmeasures do indeed provide greater safetymargins remainscontroversial as does their impact on clinical benefit of IIT [9,10].

Clinical application of IIT is usually accomplishedwith regular humaninsulin (Hlin) (HumulinR; Eli Lilly, Indianapolis, IN) continuous infusion[11,12]. Shorter acting insulin formulations, such as lispro insulin (Hlog)(Humalog; Eli Lilly) have faster onset and offset kinetics than Hlin andmay thus be more suitable for IIT in critical care patients [13-15]. Themolecular structure of Hlog is characterized by a change in the aminoacid sequence of the insulin B chain—with proline in position 28 andlysine in position 29 inverted Lys(B28),Pro(B29). This pharmacokineticprofile, which resembles that of endogenous insulin, leads to a fasterrise in plasma concentration, a higher peak concentration, and a shorterduration of action than Hlin [12,13,16,17]. The use of Hlog in patientsreceiving chronic insulin therapy is associated with a faster BGC reduc-tion when infusion is started and a reduced “residual effect”when infu-sion is stopped [18]. In the critical care setting, a similar effect may be

437.e2 F. Bilotta et al. / Journal of Critical Care 30 (2015) 437.e1–437.e6

potentially beneficial due to a decreased risk of hypoglycemia in the car-ryover phase, after insulin infusion is discontinued. However, controlledclinical data on the use of shorter acting insulin formulations in criticalcare patients are lacking.

We describe the effect of short-acting insulin on blood glucose dur-ing continuous infusion and after discontinuation of infusion in criticallyill patients, compared with that of regular insulin (the current standardof care).

2. Materials and methods

This prospective, randomized, crossover, multicenter, clinical trialreceived institutional review board approval for human research fromthe University of Rome “La Sapienza,” Italy (approval March 28, 2013,protocol no. 390/13, Chairman: Prof. Aldo Isidori) and from the ValenciaUniversity Hospital Institutional Review Board (President Dr AntonioPelaez) and was registered at the clinicaltrials.gov (NCT 02165566).Written informed consent was obtained from the patients or theirnext of kin (when the patient was sedated or unable to sign by them-selves or when someone from the family was nominated as legal guard-ian). The study took place in 2 academic centers, the neurosurgicalpostoperative intensive care unit (ICU) at the “La Sapienza” Universityof Rome, Italy, and the Surgical ICU at the Hospital Clinic Universitariof Valencia, Spain.

2.1. Study population

All patients who were older than 18 years and receiving full nutri-tional support who presented with a BGC greater than or equal to 180mg/dL were included. Moribund patients and patients enrolled inother studies were excluded as were patients with type 1 diabetes, pa-tients with insulin-dependent diabetes, and patients with glycated he-moglobin greater than 6.5% because of the potential of underlyinginsulin resistance. Simplified Acute Physiology Score II was recorded atICU admission. In both centers, BGC was measured using a point-of-care blood gas analyzer: in Rome, GEM Premier 4000 InstrumentalLaboratories, Barcelona, Spain; in Valencia, Blood Gas Analyzer 825FLEX, Radiometer, Denmark.

Fig. 1. Theprimary outcomemeasurewas the extent of carryover effect (segment2/segment1). Sduration of the carryover effect (segment 3), and the rate of BGC increase from lowest BGC valu

2.2. Study end points

The primary outcome measure was the extent of Hlog and Hlin“carryover effect,” expressed as the ratio between BGC reduction duringinsulin infusion (therapeutic BGCdrop) and BGC reduction after infusiondiscontinuation (postinfusional BGC drop). Secondary outcome mea-sures were the rate of BGC reduction during insulin infusion (milligramsper deciliter per hour), the duration of the carryover effect (ie, the timeelapsing between IIT discontinuation and the lowest BGC value), andthe rate of BGC increase after IIT discontinuation—from the lowest BGCvalue to the first BGC value greater than or equal to 140 mg/dL (Fig. 1,segments 1-3, slope m1, m2).

2.3. Study protocol

All insulin infusions were prepared by diluting 50 U of Hlin orHlog in 500 mL of saline and infused through a volumetric pump.Patients who fulfilled inclusion criteria were prospectively enrolledand randomly assigned to start treatment with either Hlog or Hlinat a dose of 0.04 U/kg per hour [19]. Bolus injection of insulin wasnot allowed.

After full nutritional support was established, continuous IIT wasinitiated provided the BGC was greater than or equal to 180 mg/dL(upper BGC threshold). Insulin infusion therapy was kept constantuntil the BGC was less than or equal to 140 mg/dL (lower BGCthreshold). Insulin infusion was discontinued once the BGC reachedless than or equal to 140 mg/dL. Crossover involved treatment ofthe same patient with both types of insulin (Hlog and Hlin) with awashout interval of at least 6 hours between the 2 treatmentswhen BGC values were greater than 180 mg/dL. Throughout the du-ration of IIT and after insulin infusion was discontinued, the BGCwas measured once every 30 minutes in whole blood until BGCvalues returned to target values (140-180 mg/dL). Two operatorswere used to achieve blinding: The operator in charge of BGC mea-surements was not aware of the type of insulin being used. The op-erator responsible for insulin administration was unaware of thevalues being measured. We defined severe hypoglycemia as BGCless than 90 mg/dL.

econdary endpointmeasureswere rate of BGC reduction during insulin infusion (slopem1),e to the first BGC value greater than or equal to 140 mg/dL (slope m2).

437.e3F. Bilotta et al. / Journal of Critical Care 30 (2015) 437.e1–437.e6

2.4. Nutrition protocol

Parenteral nutrition was initiated upon patient arrival at the ICU.Nutritional support was provided according to a standardizedschedule: 20 to 30 kcal/kg of body weight per 24 hours of nonpro-tein nutrition and a balanced composition including 0.13 to 0.26 gof nitrogen per kilogram per 24 hours and 20% to 40% of nonproteincalories in the form of lipids [20,21]. The treating dietician wasblinded to treatment assignment as were the physician/nurse incharge of BGC measurements.

2.5. Statistical analysis

Data were analyzed according to intention to treat. Demographicand clinical variables are reported, when normally distributed, asmean± SD (continuous variables) or frequency (qualitative variables).The results of continuous outcome variables are given as median, whennot normally distributed. Because of the relatively small sample size andthe nonnormal distribution of some of the outcome variables, the non-parametric Wilcoxon t test was used to verify whether the medianvalues of the 2 treatment groups differed. We planned a study of a con-tinuous response variable from matched pairs of study subjects in2 medical centers. There have been no previous matched studies com-paring 2 types of insulin, so for our sample size calculation, we usedthe difference in response observed in the NICE-SUGAR study, whichalso sought a difference in the rate of hypoglycemia. In the said study,the mean time-weighted blood in 1 group was 115 ± 18 and, in theother, was 144 ± 23 mg [6]. We thus assumed the true difference inthe mean response of matched pairs would be approximately 30 mgwith an SD of 25 mg. Assuming a normal distribution, we would haveneeded to study only 9 pairs of subjects to be able to reject the nullhypothesis that this response difference is zero with a 90%. However,

Fig. 2. Consolidated Standards of Reporting Trials–style flow

the type I error probability associated with this test of this null hypoth-esis is 0.05 assuming a normal distribution. We thus increased thesample size to ensure that the desired type I error be achieved even ifthe central limit theorem were not met and took into consideration a20% likelihood of study noncompletion. Statistical analysis was per-formed using the SPSS Inc., Chicago, Illinois (17th version).

3. Results

3.1. Patients

A total of 36 patients were randomized to 1 of the 2 treatments. Fivepatients assigned as first treatment to Hlin and 3 patients assigned toHlog had no subsequent severe hyperglycemia (BGC N180 mg/dL) andwere, therefore, not assigned to the other treatment (see Fig. 2 forCONSORT diagram). Twenty-eight patients were treated with bothHlog and Hlin. Demographic and clinical data are summarized inTable 1. During the study period, all patients received only parenteralnutritionwith amean calorie load of 26± 1.8 kcal/kg. Mean BGC valuesat the beginning of insulin infusion and at insulin infusion discontinua-tion were similar for Hlog and Hlin treatment: 197.6 ± 19.7 vs 191.9 ±12.8mg/dL, P=.20; 136.5±4.0 vs 134.7±5.6mg/dL, P=.17 (Tables 2and 3). The lowest mean BGC value, after insulin infusion wasdiscontinued, was significantly lower with Hlin than with Hlog treat-ment: 108.7 ± 10.9 vs 117.2 ± 9.3 mg/dL; P = .003 (Tables 2 and 3).Two patients, when assigned to Hlin, developed severe hypoglycemia(BGC b90 mg/dL) 2 hours after IIT discontinuation, showing in bothcases a BGC of 88mg/dL. Severe hypoglycemiawas correctedwith injec-tion of 20 mL of 33% glucose solution, and BGC was controlled shortlyafter to confirm BGC normalization. Repeat BGC measurement wasperformed within 10 minutes shortly after glucose administration toconfirm BGC normalization. The mean duration of Hlog infusion was

diagram of screening, recruitment, and randomization.

Table 3Regular human insulin infusion characteristics and effects

No. ofpatient

Extent ofcarryovereffect

Timeinfusion(h)

Start BGCvalue(mg/dL)

InsulindiscontinuationBGC value(mg/dL)

LowestBGC value(mg/dL)

1 0.82 10.50 189 138 1002 0.00 5.00 189 140 1403 0.71 4.00 191 137 1044 0.67 4.50 197 137 1025 0.24 11.00 202 140 1256 0.44 6.50 197 137 1157 0.62 6.00 182 140 1148 0.41 6.00 189 139 1209 0.57 4.50 191 139 11110 0.40 5.50 203 140 11511 0.32 8.00 247 139 10612 0.47 4.00 189 140 11713 0.51 1.50 210 137 10414 1.00 2.50 182 128 9815 0.64 2.00 190 124 10816 0.57 3.00 182 128 11617 0.52 1.50 182 132 11818 0.83 2.00 188 138 10019 0.62 1.50 187 139 11120 1.18 3.00 184 134 8821 0.63 6.00 191 137 10822 0.60 4.00 187 139 11223 1.00 2.50 182 128 9824 0.64 2.00 190 124 10825 0.57 3.00 182 128 11626 1.18 3.00 184 134 8827 0.73 2.50 191 134 10328 0.73 2.00 196 122 99

0.62(median)

4.20 ± 2.5(mean ± SD)

192 ± 12.9(mean ± SD)

135 ± 5.6(mean ± SD)

109 ± 10.9(mean ± SD)

Table 1Demographic and clinical data

Variables Patients treated with Humalog/Humulin, n = 28 patients

Age groups (y)18-30 131-50 151-70 1771-90 9

Sex (male/female) 10/18SAPS-II 18-56BMI 26.56 ± 3.55Minimum 16.65Maximum 32.69

Reason of admissionSubarachnoid hemorrhage 6Traumatic brain injury 3Traumatic spinal cord injury 1Stroke (brain hematoma) 3Neurosurgery (brain tumors, meningioma/

supratentorial lesion)5

Cardiac surgery (aortic valve substitution) 6General surgery (esophagectomy) 1General surgery (colonic surgery) 2Septic shock 1

The data are expressed as mean ± SD or number of patients; SAPS-II indicates SimplifiedAcute Physiology Score II; BMI, body mass index.

437.e4 F. Bilotta et al. / Journal of Critical Care 30 (2015) 437.e1–437.e6

2.9 ± 1.8 hours per patient, and during this period, a total of 5.8 ±3.5 BGC measurements took place; The mean duration of Hlin infusionwas 4.2 ± 2.5 hours per patient, and during this period, a total of8.4 ± 5.1 BGC measurements took place (cumulative infusion time,81 vs 117.50 hours; P = .030; cumulative number of BGC measure-ments, 162 vs 235; P= .030). Insulin infusion durationwas significantlyshorterwhenHlogwas used as comparedwithHlin: 2.9±1.8 hours per

Table 2Lispro insulin infusion characteristics and effects

No. ofpatient

Extent ofcarryovereffect

Timeinfusion(h)

Start BGCvalue(mg/dL)

InsulindiscontinuationBGC value(mg/dL)

Lowest BGCvalue(mg/dL)

1 0.38 4.50 198 139 1182 0.14 4.50 228 139 1283 0.63 3.00 181 139 1144 0.49 2.00 201 140 1105 0.17 9.00 210 139 1286 0.17 4.50 200 140 1307 0.43 5.00 198 140 1158 0.56 3.00 188 140 1139 0.37 3.00 192 140 12110 0.18 5.00 250 138 12011 0.18 5.50 266 138 11712 0.44 3.00 192 139 11713 0.42 1.00 200 137 11514 0.39 2.00 196 136 11815 0.55 1.50 184 132 11616 0.30 2.50 186 138 12617 0.40 1.50 185 138 12218 0.84 2.00 190 139 9819 0.38 1.00 188 134 12220 1.02 2.00 182 127 9721 0.59 3.00 189 135 11122 0.12 1.50 192 140 13423 0.39 2.00 196 136 11824 0.55 1.50 184 132 11625 0.30 2.50 186 138 12626 1.02 2.00 182 127 9727 0.24 1.50 198 128 12628 0.57 1.50 191 134 111

0.40(median)

2.89 ± 1.7(mean ± SD)

198 ± 19.7(mean ± SD)

137 ± 4.0(mean ± SD)

117 ± 9.3(mean ± SD)

patient vs 4.2 ± 2.5 hours per patient; P b .05. The mean time intervalbetween the 2 treatments was 10.29 hours (11.02 hours after Hlin infu-sion and 9.55 hours after Hlog infusion). In none of our patients, BGC in-creased above 180 within the 6 hours of washout.

3.2. Primary outcome

3.2.1. Extent of carryover effectLispro insulin infusion was associated with a significantly less pro-

found carryover effect after discontinuation of IIT (median, 0.40 vs0.62 as ratio between therapeutic and carryover effect; Z = 4.08; P b

.001) (Fig. 3) than Hlin infusion.

Fig. 3. Extent of the carryover effect in lispro insulin and rapid human insulin treated pa-tients after insulin infusion discontinuation. A difference of 20% between the 2 formula-tions has been demonstrated.

Fig. 4. The graph shows the decrease rate of BGC during insulin infusion (therapeuticreduction).

Fig. 6. The graph shows the increase of BGC from the lowest glucose value to the lower gly-cemic threshold of the therapeutic range (140 mg/dL).

437.e5F. Bilotta et al. / Journal of Critical Care 30 (2015) 437.e1–437.e6

3.3. Secondary outcomes

3.3.1. Rate of BGC reductionThe rate of BGC reduction during Hlog infusion (the therapeutic BGC

drop) was more rapid than during Hlin infusion (median, 21.5 vs 13.7mg/dL per hour; Z=4.62; P b .001). Thus, theuse of Hlogwas associatedwith a shorter time span between initiation of insulin infusion andattainment of BGC within the target range (Fig. 4).

3.3.2. Duration of carryoverIn patients assigned to Hlog, when compared with patients who

received Hlin, the duration of carryover was more limited when com-pared with the duration of Hlog carryover (median, 1 vs 1.5 hours;Z = 4.28; P b .001) (Fig. 5).

3.3.3. Rate of BGC increaseAfter the lowest BGC value was registered, the rate of BGC increase

after discontinuation of insulin infusion was more rapid after Hlog infu-sion comparedwithHlin infusion (median, 20.8 vs 13.9mg/dL per hour;Z = 4.01; P b .001) (Fig. 6).

4. Discussion

In this prospective randomized pilot study performed in critical carepatients, the clinical effects of IITwith Hlog or Hlin—during and after in-sulin infusion—were compared.With Hlog, as comparedwith Hlin, dur-ing insulin infusion, the rate of BGC reduction was faster, and afterinsulin discontinuation, the extent and duration of carryover effectwere more limited and the rate of BGC increase was more rapid. Theseresults suggest that shorter acting insulin analogs such as Hlog have a

Fig. 5. The duration of carryover effect (expressed in hours), after drug treatment, resultedmore limited in time with Hlog infusion than Hlin infusion.

safer profile when used as continuous infusion for glycemic control inICU patients.

The use of IIT in ICU patients became “standard of care” after clinicaltrials demonstrated an association between tight glucose control andwith lower mortality and morbidity. However, IIT has also been associ-ated with a significant increase in the rate of hypoglycemic episodes[11,22,23]. Various approaches have been proposed to reduce IIT-related risks, including targeting higher BGC values, more frequentBGC measurement, and increased caloric intake [3,6,24,25]. In this clin-ical trial, we compared the use of insulin with a shorter half-life (Hlog)to standard care and found that this approach may be safer for BGCmanagement in the ICU setting. Our findings suggest that Hlog infusionmay be associated with a possible reduction in the risk of inducingiatrogenic hypoglycemia.

In the ICU setting, where patients' clinical status can change rapidly,the use of short-acting drugs is implemented in the treatment of variousaspects of the acute illness. Examples include the use of short-actingvasodilators and antihypertensive drugs (such as nitroglycerin andesmolol), short-acting vasopressors (eg, noradrenalin and vasopres-sin), and short-acting sedatives and opioids (such as propofol andremifentanyl) [26-28].

The strengths of our study are its randomized crossover design(ensuring control of covariates and naivety to both treatment arms)and in the inclusion of more than 1 ICU. Several certain methodologicalfeatures in of this study deviate from standard clinical practice. Thesedeviations were aimed to minimize errors in insulin dosing and BGCmeasurements. To minimize the risk of involuntary insulin bolus ad-ministration and to ensure correct dosing, we used amore diluted solu-tion of insulin (50 U in 500-mL saline infused through a volumetricpump). This approach differs from the insulin dilution reported insome of previous studies (50 U in 50 mL in saline infused through amicrometric pump). Infusion of a more diluted solution minimizes theamount of insulin injected should bolus injection occur during thechanging of the syringe in the micrometric pump. We used continuousinsulin infusionwith a “fixeddose” protocol instead of titrating the insu-lin dose according to the BGC value with a “sliding scale” approach,which is generally used in this clinical setting [11,23]. We derived thedose of continuous insulin infusion used in this study from a validatedprotocol [19]. To minimize confounding factors, related to the totalamount of insulin infused, the sliding scale used in that protocol wasnot adopted. This protocol allowed us to record comparable results forthe 2 forms of insulin tested, and because of the target range used inthis study (insulin start at BGC≥180 mg/dL and insulin discontinuationat BGC≤140mg/dL), itwas not associatedwith any case of clinically sig-nificant hypoglycemia (the lowest BGC value recordedwas 88mg/dL). Afixed dose protocol for IIT in ICU setting is also associated with lesspronounced BGC variability and has been suggested as “standard ofcare” when intermittent BGC measurements are used [29]. For BGC

437.e6 F. Bilotta et al. / Journal of Critical Care 30 (2015) 437.e1–437.e6

measurements, in both centers, a point-of-care blood gas analyzer wasused. Blood gas analyzers located in the ICU have been reported tohave a greater accuracy than point-of-care BGC measurement instru-ments and provide results more rapidly than a central laboratory[30,31]. Blood gas analyzers also provide data on other relevant biolog-ical variables, in particular, potassium concentration, which can be af-fected by insulin infusion [32].

4.1. Methodological limitations

This study included a limited number of cases; however, the cross-over design of the study lends additional power to our sample size andreduces the influence of confounding covariates because each patientserves as his or her own control. Our study included mostly surgicaland neurosurgical patients. Whether our findings may be extrapolatedto the general intensive care population remains to be seen.

To make the study more reliable, other sources of glucose/dextrose(medications, IV fluids, diet, etc) were eliminated during the experi-ment. Despite this, in our study, severe hypoglycemia was extremelyrare. The interval between the 2 treatmentsmay have affected the treat-ment impact. However, random assignment to the first treatment arm,the duration between the 2 treatments, and the increase of BGC tovalues greater than 180 mg/dL in between treatments adequatelycover the likelihood of clinically relevant residual insulin effect relatedto the first treatment.

5. Conclusions

In conclusion, in ICU patients, the use of short-acting insulin Hlog forIIT at a fixed dose, when comparedwith Hlin at the same dose, is associ-atedwith a less profound carryover effect on BGC after insulin infusion isdiscontinued. The use of Hlog, when compared with Hlin, is also associ-atedwith a higher rate of BGC reduction during insulin infusion—leadingto faster attainment of the target BGC range—a shorter duration of carry-over and a higher rate of BGC increase after insulin infusion discontinu-ation. The latter characteristic is potentially associated with a quickerreversal of hypoglycemia should IIT induce BGC to drop below thelower threshold. The use of short-acting insulin Hlog, when comparedwith Hlin—the “standard of care” in most ICUs—may be associatedwith a safer pharmacodynamic profile. Future studies should be carriedout to further confirmwhether Hlog is a more appropriate choice for IITin surgical and medical ICU patients.

References

[1] Bilotta F, Giovannini F, Caramia R. Glycemia management in neurocritical carepatients: a review. J Neurosurg Anesthesiol 2009;21(1):2–9.

[2] Klonoff DC. Intensive insulin therapy in critically ill hospitalized patients: making itsafe and effective. J Diabetes Sci Technol 2011;5(3):755–67.

[3] OkawaM, Kunimoto F, Kanamoto M, Narahara H, Hinohara H, TobeM, et al. Effect ofdifferent blood glucose target levels on the incidence of hypoglycemia during insulintherapy in the intensive care unit. J Diabetes 2013;5:51–6.

[4] Heller S. Reducing hypoglycaemia with insulin analogues. Int J Obes (Lond) 2002;26:S31–6.

[5] Jacobi J, Bircher N, Krinsley J, AgusM, Braithwaite SS, Deutschman C, et al. Guidelinesfor the use of an insulin infusion for the management of hyperglycemia in criticallyill patients. Crit Care Med 2012;40(12):3251–76.

[6] NICE-SUGAR Study Investigators, Finfer S, Chittock DR, Su SY, Blair D, Foster D, et al.Intensive versus conventional glucose control in critically ill patients. N Engl J Med2009;360(13):1283–97.

[7] Bilotta F, Rosa G. Glycemia management in critical care patients. World J Diabetes2012;15:130–4.

[8] Shetty S, Inzucchi SE, Goldberg PA, Cooper D, Siegel MD, Honiden S. Adapting to thenew consensus guidelines for managing hyperglycemia during critical illness: theupdated Yale insulin infusion protocol. Endocr Pract 2012;18(3):363–70.

[9] Bilotta F, Rosa G. Glucose management in the neurosurgical patient: are we yet anycloser? Curr Opin Anaesthesiol 2010;23(5):539–43.

[10] Egi M, Finfer S, Bellomo R. Glycemic control in the ICU. Chest 2011;140(1):212–20.[11] Van Den Berghe G, Wouters P, Weekers F, Verwaest C, Bruyninckx F, Schetz M, et al.

Intensive insulin therapy in critically ill patients. N Engl J Med 2001;345:1359–67.[12] Howey DC, Bowsher RR, Brunelle RL, Woodworth JR. [Lys(B28), Pro(B29)]-

human insulin. A rapidly absorbed analogue of human insulin. Diabetes 1994;43:396–402.

[13] Home PD. The pharmacokinetics and pharmacodynamics of rapid-acting insulin an-alogues and their clinical consequences. Diabetes Obes Metab 2012;14:780–8.

[14] Bilotta F, Guerra C, Badenes R, Lolli S, Rosa G. Short acting insulin analogues in ICUpatients. World J Diabetes 2014;5(3):230–4.

[15] Barnett AH, Owens DR. Insulin analogues. Lancet 1997;349:47–51.[16] Sheldon B, Russell-Jones D, Wright J. Insulin analogues: an example of applied

medical science. Diabetes Obes Metab 2009;11:5–19.[17] Di Marchi RD, Mayer JP, Fan L, Brems DN, Frank BH, Green LK, et al. Synthesis of a

fast-acting insulin based on structural homology with insulin-like growth factor 1.In: Smith JA, Rivier JE, editors. Peptides: chemistry and biology, proceedings of the12th American Peptide Symposium. Leiden, Netherlands: ESCOM; 1992. p. 26–8.

[18] Mudaliar SR, Lindberg FA, Joyce M, Beerdsen P, Strange P, Lin A, et al. Insulin aspart(B28 asp-insulin): a fast-acting analog of human insulin: absorption kineticsand action profile compared with regular human insulin in healthy non diabeticsubjects. Diabetes Care 1999;22:1501–6.

[19] Texas Department of Health State Services. IV insulin infusion protocol for critically-ill adult patients in the ICU setting. http://www.acponline.org/about_acp/chapters/mt/insulin_texas.pdf.

[20] Cahill NE, Dhaliwal R, Day AG, Jiang X, Heyland DK. Nutrition therapy in the criticalcare setting: what is “best achievable” practice? An international multicenter obser-vational study. Crit Care Med 2010;38(2):395–401.

[21] Souba WW. Nutritional support. N Engl J Med 1997;336:41–8.[22] Bilotta F, Rosa G. Optimal glycemic control in neurocritical care patients. Crit Care

2012;16(5):163.[23] Bilotta F, Caramia R, Paoloni FP, Delfini R, Rosa G. Safety and efficacy of intensive in-

sulin therapy in critical neurosurgical patients. Anesthesiology 2009;110:611–9.[24] Lena D, Kalfon P, Preiser JC, Ichai C. Glycemic control in the intensive care unit and

during the postoperative period. Anesthesiology 2011;114:438–44.[25] Derde S, Vanhorebeek I, Ververs EJ, Vanhees I, Darras VM, Van Herck E, et al. Increas-

ing intravenous glucose load in the presence of normoglycemia: effect on outcomeand metabolism in critically ill rabbits. Crit Care Med 2010;38:602–11.

[26] Chitilian HV, Eckenhoff RG, Raines DE. Anesthetic drug development: novel drugsand new approaches. Surg Neurol Int 2013;4:S2–S10.

[27] Vuylsteke A, Vincent JL, De La Garanderie DP, Anderson FA, Emery L, Wyman A, et al.Characteristics, practice patterns, and outcomes in patients with acute hypertension:European registry for Studying the treatment of Acute hyperTension (Euro-STAT).Crit Care 2011;15:R271.

[28] Flaherty JT, Reid PR, KellyDT, TaylorDR,WeisfeldtML, Pitt B. Intravenous nitroglycerinin acute myocardial infarction. Circulation 1975;51:132–9.

[29] Maury E, Vitry P, Galbois A, Ait-Oufella H, Baudel JL, Guidet B, et al. Continuous insu-lin administration via complex central venous catheter infusion tubing is anotherrisk factor for blood glucose imbalance. A retrospective study. Ann Intensive Care2012;2:16.

[30] Pitkin AD, Rice MJ. Challenges to glycemic measurement in the perioperative andcritically ill patient: a review. J Diabetes Sci Technol 2009;6:1270–81.

[31] Corstjens AM, Ligtenberg JJ, van der Horst IC, Spanjersberg R, Lind JS, Tulleken JE,et al. Accuracy and feasibility of point-of-care and continuous blood glucose analysisin critically ill ICU patients. Crit Care 2006;10(5):R135.

[32] Gunst J, Van den Berghe G. Blood glucose control in the intensive care unit: benefitsand risks. Semin Dial 2010;23(2):157–62.