Intensive Versus IntermediateGlucose Control in SurgicalIntensive Care Unit PatientsDiabetes Care 2014;37:1516–1524 | DOI: 10.2337/dc13-1771

OBJECTIVE

The optimal perioperative blood glucose range to improve surgical site infection(SSI) in surgical intensive care unit (ICU) patients remains unclear. We sought todetermine whether the incidence of SSI is reduced by perioperative intensiveinsulin therapy (IT).

RESEARCH DESIGN AND METHODS

Patients were randomly assigned to receive perioperative intensive IT, with atarget blood glucose range of 4.4–6.1 mmol/L, or intermediate IT, with a targetblood glucose range of 7.7–10.0 mmol/L in the surgical ICU. We defined theprimary end point as the incidence of SSI.

RESULTS

Study participants were randomly assigned to glucose control with one of twotarget ranges: for 225 patients in the intermediate IT group or for 222 patients inthe intensive IT group, respectively. No patients in either group became hypogly-cemic (<4.4 mmol/L) during their stay in the surgical ICU. In our series, the rate ofSSI after hepato-biliary-pancreatic surgery was 6.7%. Patients in the intensive ITgroup, compared with the intermediate IT group, had fewer postoperative SSIs(9.8% vs. 4.1%, P = 0.028) and a lower incidence of postoperative pancreatic fistulaafter pancreatic resection (P = 0.040). The length of hospitalization required forpatients in the intensive IT group was significantly shorter than that in the in-termediate IT group (P = 0.017).

CONCLUSIONS

We found that intensive IT decreased the incidence of SSI among patients whounderwent hepato-biliary-pancreatic surgery: a blood glucose target of 4.4 to6.1 mmol/L resulted in lower rate of SSI than did a target of 7.7–10.0 mmol/L.

Hyperglycemia is common in acutely ill patients, including those treated in intensivecare units (ICUs) (1). Until 2001, neglecting hyperglycemia was standard ICU carebecause a very impressive large randomized trial involving patients admitted to asurgical ICU showed that intensive insulin therapy (IT), targeting a blood glucose con-centration of 4.4–6.1 mmol/L, significantly reduced in-hospital mortality (2). How-ever, trials examining the effects of tight glycemic control (TGC) have had conflictingresults (1,3–6). Systematic reviews and meta-analyses have also led to differingconclusions (7,8). The main reason these clinical trials and meta-analyses had neg-ative results for TGC was the high incidence of hypoglycemia (10–17%) induced byintensive IT (7,8).

1Center for Innovative and Translational Medi-cine, Regenerative Medicine Group, Departmentof Surgery, Kochi Medical School, Kochi Univer-sity, Kochi, Japan2Center for Innovative and Translational Medi-cine, Regenerative Medicine Group, Departmentof Surgery, Kochi Health Sciences Center, KochiMedical School, Kochi University, Kochi, Japan3Center for Innovative and Translational Medi-cine, Regenerative Medicine Group, Departmentof Biostatistics, Kochi Medical School, Kochi Uni-versity, Kochi, Japan4Center for Innovative and Translational Medicine,Regenerative Medicine Group, Department of An-esthesiology and Intensive Care Medicine, KochiMedical School, Kochi University, Kochi, Japan

Corresponding author: Takehiro Okabayashi,tokabaya@kochi-u.ac.jp.

Received 28 July 2013 and accepted 28 January2014.

Clinical trial reg. no. NCT00735228, clinicaltrials.gov.

A slide set summarizing this article is availableonline.

© 2014 by the American Diabetes Association.See http://creativecommons.org/licenses/by-nc-nd/3.0/ for details.

Takehiro Okabayashi,1,2 Yasuo Shima,2

Tatsuaki Sumiyoshi,2 Akihito Kozuki,2

Teppei Tokumaru,2 Tasuo Iiyama,3

Takeki Sugimoto,1 Michiya Kobayashi,1

Masataka Yokoyama,4 and

Kazuhiro Hanazaki1

1516 Diabetes Care Volume 37, June 2014

CLINCARE/ED

UCATION/N

UTR

ITION/PSY

CHOSO

CIAL

In recent years, it was reported thatthe results of the updated meta-analysisincluding NICE-SUGAR study data alsodo not support widespread adoption ofintensive IT in critically ill patients (9). Inthis meta-analysis, however, intensiveIT may be beneficial to patients admit-ted to a surgical ICU, although the char-acteristics of such patients remain to beclearly defined, as do the effect of dif-ferent blood glucose algorithms, themethod of measuring blood glucose,and the influence of nutritional strate-gies (9). Furthermore, this meta-analysisincluding NICE-SUGAR study data de-scribed that intensive IT significantlyincreased the risk of hypoglycemia. Ef-fective control of surgical site infection(SSI) can reduce the length of hospital-ization and can result in enhancedrecovery after surgery. Perioperativehyperglycemia in critically ill surgery pa-tients increases the risk of SSI, which is acommon surgical complication. In vari-ous surgical settings, it is important toensure optimal blood glucose levels inorder to reduce SSIs. Unfortunately,however, the optimal perioperativeblood glucose range to improve surgicaloutcomes including SSI remains unclear.In the past 10 years, solving the problemof methodology for intensive IT, espe-cially the incidence of hypoglycemiaduring intensive IT, did not progress atall. The true effect of intensive IT in clin-ically ill patients cannot be evaluatedwithout solving the occurrence of hypo-glycemia during intensive IT. The devel-opment of accurate, continuous bloodglucose–monitoring devices and closed-loop systems for computer-assisted bloodglucose control in the ICU will proba-bly help avoid hypoglycemia in thesesituations (10).Based on these findings, we con-

ducted this prospective study to investi-gate which perioperative glycemiccontrol range was better for reducingSSIdbetween intensive IT, with targetblood glucose range of 4.4–6.1 mmol/L,and intermediate IT,with targetblood glu-cose range of 7.7–10.0 mmol/Ldusing aclosed-loop glycemic control system.

RESEARCH DESIGN AND METHODS

Study DesignWe conducted a parallel-group, ran-domized, controlled trial involving adultsurgical patients admitted to the ICU ofour hospital. Patients were randomized

at the time of the starting operation.Also, blood glucose control of recruitedpatients was begun from the start ofsurgery. All eligible patients were thosesurgically treated for hepato-biliary-pancreatic diseases and admitted tothe surgical ICU overnight until post-operative day 1 after hepatic and/orpancreatic resection to control postop-erative blood glucose concentrationlevels by using closed-loop glycemiccontrol system. Patient exclusion crite-ria included a body weight loss .10%during the 6 months prior to surgery,the presence of distant metastases, orseriously impaired function of vital or-gans due to respiratory, renal, or heartdisease. Patients were informed of thepurpose and details of the study, andwritten consent was obtained fromthem prior to enrolment. The studywas approved by the local ethics com-mittee at the Kochi Medical School andcarried out in accordance with the Dec-laration of Helsinki. All studies were per-formed at the Kochi Medical Schoolbetween August 2008 and August 2012.

A complete physical examination wasundertaken and clinical histories wereobtained from all patients. Laboratorytests included the measurement of se-rum levels of albumin, total bilirubin,total cholesterol, and alanine and as-partate aminotransferases, as well asperipheral blood cell counts and pro-thrombin time. The patients’ clinicalhistory included details of glucose me-tabolism, including medication for thetreatment of diabetes, fasting blood glu-cose level, and the monitoring of hemo-globin A1c (HbA1c) levels in plasma. Ofall patients included in this study, dia-betes mellitus (DM) status was checkedby diabetologist, and DM were alltype 2. Operation-related parametersof liver and pancreatic surgery werealso evaluated.

Postoperative Nutrition in SurgicalICU and Glycemic ControlIn our study, all patients were given par-enteral nutrition (PN) postsurgery justafter bring admitted to the surgical ICU.The total caloric requirement was calcu-lated according to the Harris-Benedictequation (11). All of the patients whowere assigned to this study received in-travenous PN and 20% intravenous fatemulsions and 25% glucose solution,which included an adequate amount of

protein, glucose, vitamins, minerals, andtrace element such as zinc, iron, copper,manganese, and iodine (rate of glucoseadministration ,5 mg/kg/min). Enteralfeeding was attempted as soon as possi-ble when the patients were hemody-namically stable. The amount of PN wascalculated daily as the difference be-tween the total energy intake that waseffectively delivered by enteral nutri-tion and the calculated caloric goalwhen postoperative oral intake wouldbe restarted.

The Nikkiso Company (Tokyo, Japan)developed the closed-loop glycemiccontrol system in 1984. The closed-loop glycemic control system is a reli-able and accurate device that measuresblood glucose concentration comparedwith the ABL 800FLEXmachine (Radiom-eter Medical ApS, Brønshøj, Denmark)recommended by the National Commit-tee for Clinical Laboratory (12,13). Theclosed-loop glycemic control system iscomposed of a glucose sensor, which per-forms glucose detection/monitoring,and pumps for infusing the appropriateamount of insulin or glucose (14). Theinsulin and glucose pumps are computerregulated based on a target blood glu-cose value predefined prior to the sys-tem initiation. Peripheral blood forglucose monitoring was sampled contin-uously at 2 mL/h during at least the firstpostoperative 18 h in the surgical ICU.Furthermore, the closed-loop artificialpancreas system was used to evaluatethe patient’s insulin requirements.

Study participants were randomly as-signed to glucose control with one oftwo target ranges: the intensive (i.e.,tight) control target of 4.4–6.1 mmol/L(intensive IT group), based on that usedin previous studies (2,15), or an interme-diate control target of 7.7–10.0 mmol/L(intermediate IT group), based on thatused in a previous study (16). Blood glu-cose concentration (intermediate vs. in-tensive) was controlled by a closed-loopglycemic control system from the startof surgery, and either intermediate orintensive IT was performed during theoperation. After patients were removedfrom the closed-loop glycemic controlsystem at postoperative day 1, bloodglucose levels were routinely checkedby nursing staff and were controlled bythe subcutaneous injection of regularhuman insulin every 2 h: the dose wasdetermined by the commonly used

care.diabetesjournals.org Okabayashi and Associates 1517

sliding scale, and the target blood glu-cose level to avoid hypoglycemia was8.3–11.1 mmol/L (17).

Data CollectionBaseline demographic and clinical char-acteristics of the patients were wellmatched between the two study groups(Table 1). Local study coordinators atour institution collected the data;source data were verified by study mon-itors from our coordinating center at thedepartment of biostatistics. Daily re-cords were kept regarding all intensivecare treatments and surgical proce-dures, new bacterial or fungal infec-tions, the results of blood and urinechemical analyses and hematologicstudies, and markers of inflammation.Also recorded were the total energy in-take delivered daily by means of enteraland PN, interruptions of delivery ofenteral nutrition, and feeding-relatedcomplications. In addition, wheneverpractically feasible, the functional statusof patients before hospital dischargewas quantified.

Outcome Measures

Safety End Points

Safety end points included vital status(the rates of death in the ICU and thehospital) and especially the rates of hy-poglycemia. Hypoglycemia that wasresistant to parenteral glucose adminis-tration during the intervention windowwas considered to be a serious adverseevent. And a blood glucose level of#40mg/dL was considered a serious adverseevent.

Primary Efficacy End Point

The primary end point of this study wasto determine whether the incidence ofSSI is reduced by perioperative inten-sive IT. The Centers for Disease Controland Prevention’s National NosocomialInfections Surveillance has developedstandardized surveillance criteria for de-fining SSI (18). According to these crite-ria, the infection control team evaluatedwhether postoperative SSI in patientswho underwent liver and pancreatic re-section had occurred.

Secondary Efficacy End Point

The secondary end point was to evalu-ate the duration of hospital stay in eachpatient group. To reduce bias that mightresult from variability in the availabilityof beds on regular wards, we definedthe time to discharge from our instituteas the time by which patients wereready for hospital discharge, accordingto the suggestion by trained expert staffof the Department of Diagnostic Medi-cine at our institute and nutritional sup-port team blindly.

Statistical AnalysisGiven that the primary outcome of thestudy is SSI, we aim to detect a clinicallyrelevant difference in 5% of SSI inci-dence concretely from 10 to 5% ofpatients during hospitalization. This dif-ference is in accordance with previousreports (19,20) and is considered clini-cally significant (21). Consequently, therequired number to detect this differ-ence with 80% is 219 patients per group.Allowing for attrition of 10% of caseswith SSI, we have to recruit 438 patients

and randomly assign 219 patients intothe intensive IT group and 219 patientsinto the intermediate IT group (Fisherexact test, two-sided, intention-to-treatbasis). Because of the probably rate ofdropout that is estimated, ~10% of thepatients initially enrolled into the study,additional patients were randomized,resulting in a total subject number of482 across the two groups. Continuousvariables are presented as the mean 6SD. Dichotomous variables are pre-sented as both number and percentagevalues. P , 0.05 was considered signif-icant. Data were analyzed using theStudent t test (two-tailed), with dichot-omous variables analyzed by the x2 test(two-tailed) or Fisher exact test (two-tailed), as appropriate.

Subgroup analyses on intent to treatfor the primary outcome were based onan unadjusted test of interaction in alogistic model to evaluate which pa-tients would benefit from intensive ITduring the surgical ICU to reduce the in-cidence of postoperative SSI. All analy-ses were performed using SPSS (SPSS,Chicago, IL). The data were analyzed byan independent statistician of theDepartment of Biostatistics (Kochi Uni-versity). The analyses were reviewed bythe independent data and safety-monitoring committee, which waschargedwith recommending that the tri-al be stopped if there was evidencebeyond a reasonable doubt of a dif-ference in the rate of postoperativecomplications, such as hypoglycemia,hospital death, bile leakage from liverstump after hepatic resection, and pan-creatic fistula after pancreatic resection,from any cause between the two treat-ment groups.

RESULTS

Study ParticipantsParticipants were recruited and hadfollow-up during the period fromAugust2008 through August 2012; 502 wererandomly assigned to one of the twotreatment groups: 252 to intermediateglucose control and 250 to intensive glu-cose control (Fig. 1). Study treatmentwas discontinued prematurely in 27 of252 patients (10.7%) in the intermediateIT group and 28 of 250 patients (11.2%)in the intensive IT group. There werefive in-hospital deaths in the intermedi-ate IT group (mortality rate 1%). Fourpatients underwent hepatectomy for

Table 1—Baseline characteristics of the patients

VariablesIntermediate IT

(BS 140–180 mg/dL)Intensive IT

(BS 80–110 mg/dL) P

n 225 222

Age (years) 66.4 6 10.4 66.7 6 10.1 0.781

Male/female (n/n) 151/74 142/80 0.484

BMI (kg/m2) 23.1 6 3.4 23.3 6 3.6 0.723

Blood chemistryAlbumin 3.9 6 0.4 4.0 6 0.4 0.104Hemoglobin 12.5 6 1.6 12.3 6 1.8 0.205Prothrombin time (%) 88.5 6 17.1 87.9 6 14.1 0.743

Creatinine 0.8 6 0.2 0.8 6 0.5 0.235

Blood urea nitrogen 14.1 6 4.5 13.5 6 4.2 0.289

HbA1c 5.6 6 1.1 5.6 6 0.9 0.459

Medication for DM, n (%) 59 (26.2) 54 (24.3) 0.724

Data are means 6 SD unless otherwise indicated. BS, target levels of blood glucoseconcentration.

1518 Intensive Versus Intermediate Glucose Control Diabetes Care Volume 37, June 2014

hepatocellular carcinoma. Two of fourhepatectomized patients, whowere suf-fering from severe liver cirrhosis, fellinto hepatic failure owing to limited liverfunction of the remnant liver and diedafter surgery (42 days and 47 days afterliver resection for hepatocellular carci-noma). One patient of hepatectomizedpatients, who had a gastrointestinalbleeding after liver resection, deve-loped sequential hepatic failure anddied 63 days after surgery. Anotherhepatectomized patient, who had an ac-cidentally dissecting aneurysm of aorta,died 2 days after surgery. The remainingpatient, who had a 4-cm-sized adeno-carcinoma located in the head of thepancreas, with insufficiency of pan-creato-enteric anastomosis, was rela-parotomized for diffuse peritonitis27 days after the initial pancreatico-duodenectomy, ran a fatal course dueto subsequent intra-abdominal septiccomplications, and died on the 91st

postoperative day. Reasons for discon-tinuation were withdrawal because offive in-hospital deaths in the intermedi-ate IT group or closed-loop glycemic con-trol system errors in patients whounderwent IT during the surgical ICUfor 22 of the 252 patients (8.7%) as-signed to intermediate IT and 28 of the250 patients (11.2%) assigned to inten-sive IT; thus, study data were availablefor 225 and 222 patients, respectively.Of the 50 patients for whom study datawere unavailable, we could recognize42 cases (8.5%) of an insufficiency forblood sampling from peripheral vesselsand 8 cases (1.6%) of trouble with theglucose sensor of the closed-loop glyce-mic control system itself.

The baseline characteristics of thetreatment groups were similar (Table1). The mean 6 SD age was 66.4 610.4 and 66.7 6 10.1 years in the inter-mediate IT group and the intensive ITgroup, respectively; the percentage of

male patients 67.1% and 64.0%; andthe mean BMI score, 23.1 6 3.4 and23.36 3.6. Also, both preoperative liverand renal function were similar in thetwo groups. Furthermore, the presenceof a previousmedical history for DMwasequally distributed between the twogroups. In the current study, becauseall of the surgeries were scheduled pro-cedures, DM status in patients whowere recruited in our study was preop-erativelywell controlled by the diabetol-ogist; the mean HbA1c levels were 5.661.1% in the intermediate IT group and5.6 6 0.9% in the intensive IT group,respectively.

All patients underwent either hepa-tectomy or pancreatectomy consistingof curative resection of both hepaticand pancreatic tissue for the removalof a tumor. There was no significant pre-disposition to these operative proce-dures between the two groups; liversurgery for 142 cases in intermediate

Figure 1—Assessment, randomization, and follow-up of the study patients. Patient excluded for a number of medical reasons, including a bodyweight loss .10% during the 6 months prior to surgery, the presence of distant metastases, or seriously impaired function of vital organs due torespiratory, renal, or heart disease. Study treatment was discontinued prematurely in 27 of 252 patients in the intermediate IT group and 28 of 250patients in the intensive IT group. Study data were available for 225 in the intermediate IT group and 222 patients in the intensive IT group,respectively. BS, target levels of blood glucose concentration.

care.diabetesjournals.org Okabayashi and Associates 1519

IT group and 148 cases in intensive ITgroup and pancreatic surgery for 83 casesin intermediate IT group and 74 cases inintensive IT group, respectively (Table 2).The operation time and estimated vol-ume of blood loss did not differ significan-tly between the two groups (operationtime 36.2 6 144.2 min in intermediateIT group and 337.26 142.1 min in inten-sive IT group; estimated blood loss vol-ume 768.7 6 764.5 mL in intermediateIT group and 762.1 6 720.0 mL in inten-sive IT group) (Table 2).

Postoperative Blood Glucose LevelsAll patients received PN just after beingadmitted to the surgical ICU. The meandaily amount of calories administeredwas 1,479 6 207 kcal in the intermedi-ate IT group and 1,4656 216 kcal in theintensive IT group, respectively. Enteralfeeding was attempted as soon as pos-sible when the patients were hemody-namically stable. Postoperative bloodglucose levels in the intermediate andintensive IT groups during the first 18 hafter surgery are shown in Fig. 2A. Theaverage blood glucose levels in patientsfrom the intermediate IT group werecontrolled to the target zone (7.7–10.0mmol/L). The percentage in blood glu-cose target at intermediate glucose con-trol by a closed-loop system duringsurgical ICU was 96.8% (Fig. 2A), al-though blood glucose levels graduallyincreased during the first 6 h after thehepatectomy and pancreatectomy and

reached a plateau of ~13.8 mmol/L be-tween 4 and 8 h by the ordinary bloodglucose control methods by using slid-ing scale and then the concentrationsgradually decreased toward 8.3 mmol/Lby 18 h after surgery in our previousstudies (21,22). Also, the average bloodglucose levels in patients from the in-tensive IT group were well controlledto the target zone (4.4–6.1 mmol/L)but did not decrease to hypoglycemiclevels, and the percentage within theblood glucose target for intensive glu-cose control by a closed-loop systemduring surgical ICU was 85.8% (Fig.2A). Patients undergoing intensive glu-cose control received a larger mean in-sulin dose (1016 88 units [range 8–390]in the surgical ICU vs. 77 6 72 units[range 0–355] for the intermediate ITgroup) for 18 h after the hepatic resec-tion (Fig. 2B). This represents a sig-nificant difference in total insulinconsumption between the two groups(P = 0.015).

Safety OutcomesThe study groups had similar rates ofdeath in the ICU and the hospital. How-ever, mortality occurred solely in theintermediate IT group, although postop-erative complications occurred in thesepatients after surgery. This includedfour patients in postoperative liverfailure after hepatic surgery and onepatient in postoperative acute respira-tory distress syndrome after pancreatic

surgery. No patients in either group be-came hypoglycemic (,2.2 mmol/L) dur-ing their stay in the surgical ICU orduring the hospitalization.

Primary OutcomeThe short-term outcomes of patientswho underwent hepatic and pancreaticresection are detailed in Table 2. Withrespect to postoperative characteristics,when blood glucose levels were tightlycontrolled using the closed-loop glyce-mic control system, the incidence of SSIin the intensive IT group was signifi-cantly lower than that in the intermedi-ate IT group (P = 0.028). There was noreadmission of patients who had post-operative SSI in the current study.

Secondary OutcomePostoperative morbidities were exam-ined by abdominal ultrasonography orcomputed tomography. There was nosignificant difference between the twogroups in the incidence of bile leakageafter liver resection. Interestingly, theincidence of postoperative pancreaticfistula formation after pancreatic sur-gery in the intensive IT group was sig-nificantly lower than that in theintermediate IT group (P = 0.040). It isof note that patients in the intensive ITgroup required a significantly shorterhospitalization than patients in the in-termediate IT group (P = 0.017). Therewas one operation-related readmissionof patients in the intensive IT group dueto the abdominal abscess formation fordelayed pancreatic fistula in the currentstudy.

Subgroup AnalysisIn postsubgroup intent-to-treat analy-ses, we evaluated association betweenthe incidence of postoperative SSI andperioperative predictable factors. Threeparameters were identified as indepen-dentmarkers for the occurrence of post-operative SSI by the Spearman correlationanalysis: 1) elderly patients (.65 yearsold) (P = 0.039), preoperative predict-able factor; 2) patients with DM (P =0.036), preoperative predictable factor;and 3) the presence of pancreatic fistulaformation after pancreatic resection (P =0.047), postoperative factor after thesurgical management. Considering DMstatus, which was a preoperatively pre-dictable factor, the rate of SSI was lowerin patients without DM (3.7%) than inpatients with DM (19.0%, P , 0.001).

Table 2—Operation-related parameter and postoperative outcomes

VariablesIntermediate IT

(BS 140–180 mg/dL)Intensive IT

(BS 80–110 mg/dL) P

n 225 222

Liver surgery (n = 290) 142 148Hemihepatectomy 20 23Segmentectomy 49 51 0.936Limited resection 73 74

Pancreatic surgery (n = 157) 83 74 0.939Total pancreatectomy 4 6Pancreaticoduodenectomy 47 43Distal pancreatectomy 26 20Minimal invasive resection 6 5

Operation time (min) 362.2 6 144.2 337.2 6 142.1 0.196

Estimated blood loss (mL) 768.7 6 764.5 762.1 6 720.0 0.948

Complications (%)Bile leakage 18/142 (12.7) 16/148 (10.8) 0.622Pancreatic fistula 22/83 (26.5) 9/74 (12.3) 0.040SSI 22/225 (9.8) 9/222 (4.1) 0.028

Hospitalization (days) 23.0 6 16.6 18.2 6 15.8 0.017

Data are n, means 6 SD, or n/n (%).

1520 Intensive Versus Intermediate Glucose Control Diabetes Care Volume 37, June 2014

CONCLUSIONS

We found that there was a significantdifference in the incidence of postoper-ative SSI between intermediate and in-tensive IT group in the surgical ICUamong patients who underwent hepato-biliary-pancreatic surgery. The majorstrength of this study is the restrictionto ICU patients admitted after hepaticand pancreatic surgery. ICU patientscomprise a heterogeneous population,and different effects in subgroups ofICU patients may partly explain the con-flicting findings in previous trials (1,2,15).The NICE-SUGAR trial was criticized for

the overlapping blood glucose levels inthe two treatment arms (1). In our study,because the use of a closed-loop glyce-mic control system was promising formaintaining blood glucose within a rea-sonable narrow range, 55 (11.0%) of 502were excluded after randomization toremove the potential impact of any se-lection bias from dropouts from thestudy, and we recalculated again the ex-act number to obtain the adequateresults by a biostatistician (23). Stratifi-cation by diabetes status is very relevant,as the impact of intensive IT may be dif-ferent in this subgroup. Intensive IT in

the surgical ICU was also associatedwith a lower incidence of postoperativepancreatic fistula and a shorter durationof hospital stay compared with the inter-mediate IT group. The current work ex-amined both the safety of intensive ITusing the closed-loop glycemic controlsystem and postoperative SSI in an ex-clusively surgical population. Also, thisnovel intensive IT using a closed-loopglycemic control system enabled us tomaintain stable glycemic controldnotonly no hypoglycemia but also no hyper-glycemiadwith less variability of bloodglucose concentration. Moreover, our

Figure 2—Postoperative blood glucose levels and total insulin requirement. A: Postoperative blood glucose levels using the closed-loop glycemiccontrol system in the intermediate IT (solid line) and intensive IT (dashed line) groups. The percentage of subjects within the blood glucose targetrange for intermediate glucose control by a closed-loop system during surgical ICU was 96.8%, and also, the average blood glucose levels in patientsfrom the intensive IT group were well controlled to the target zone (80–110mg/dL) but did not decrease to hypoglycemic levels, and the percentagewithin the blood glucose target range for intensive glucose control by a closed-loop system during surgical ICUwas 85.8%. B: Total insulin usage withthe closed-loop glycemic control system in the intermediate and intensive IT groups. Patients undergoing intensive glucose control received a largermean insulin dose (1016 88 units [range 8–390] in the surgical ICU vs. 776 72 [0–355] with the intermediate IT group) for 18 h after the hepaticresection (P = 0.015).

care.diabetesjournals.org Okabayashi and Associates 1521

findings support the hypothesis by vanden Berghe that accurate continuousblood glucose–monitoring devices andclosed-loop systems for computer-assistedblood glucose control in the ICU willlikely reduce the incidence of hypoglyce-mia in intensive IT (10).Although complication rates in hepato-

biliary-pancreatic surgery are improved,wound infections remain a commoncomplication (24). One of the most im-portant risk factors for developmentof any SSI in patients undergoing gen-eral surgery was DM (25,26). Also, ourstudy demonstrates significant asso-ciations between the development ofSSI and patients who were sufferingfrom DM, and the incidence of SSI waslower in patients without DM than inpatients with DM, whereas, in ourstudy, the incidence of postoperativeSSI was acceptable in both groups,even if perioperative blood glucose lev-els were controlled by either interme-diate IT or intensive IT. Interestingly,

the rate of postoperative SSI in patientswith DM was significantly reduced byintensive IT using a closed-loop glyce-mic control system by subgroup analy-sis (Fig. 3). The current study suggeststhat intensive IT might be beneficial topatients with DM admitted to a surgicalICU after hepato-biliary-pancreatic sur-gery because there were correlative as-sociations between the intensive IT andthe reduction of the incidence of post-operative SSI in patients with DM.

Appropriate nutritional support andglycemic control play important rolesin the promotion of enhanced recoveryafter surgery for surgical patients. Variousconsensus statements and guidelinesfrom international societies recommendthe early and adequate initiation of spe-cialized nutrition therapy for critically illpatients admitted for surgical condi-tions (27,28). Although the formulationand caloric goals of PN solutions havechanged in recent years, debate still ex-ists about the timing of the initiation of

PN when enteral nutrition either is im-possible or does notmeet the nutritionalgoals. Thus, the total caloric energy in-take per day during the perisurgical pe-riod was assumed to be equal betweenthe two groups. However, there is a di-lemma because promotion of nutritionalsupport is likely to result in hyperglyce-mia, and TGC with the open-loop systemalways has the risk of hypoglycemia.Indeed, in previous reports of conven-tional intensive IT without the closed-loop glycemic system, there was risknot only of hypoglycemia associatedwith TGC but also of insufficient nutri-tional support mainly because of fearof hyperglycemia (7,9). Therefore, com-patibility of sufficient nutritional supportand intensive IT seemed to be difficult incritically ill patients. Surprisingly, even insuch patients, this study showed thatglycemic control using a closed-loopglycemic control system enabled us tocomplete intensive IT with a targetblood glucose range of 4.4–6.1 mmol/L,

Figure 3—Forest plots in subgroup analyses. Forest plots show treatment effects on the reduction of the incidence of postoperative SSI in each groupby intent-to-treat analysis. C, treatment responses. PF, pancreatic fistula.

1522 Intensive Versus Intermediate Glucose Control Diabetes Care Volume 37, June 2014

although postoperative nutritional sup-port in every patient undergoing hepaticresection or pancreatic resection wascarried out by calorie intake accordingto the calculations of the Harris-Benedictequation. Namely, if we used the closed-loop glycemic control system, compati-bility of appropriate nutrition supportand TGC including intensive IT was pos-sible simultaneously even in critically illsurgical patients. Virtually all large stud-ies comparing an intensive versus an in-termediate glucose control range haveshown that the incidence of hypoglyce-mia is higher when targeting a lower goalrange of 4.4–6.1 mmol/L. In the past 10years, themain problem discussed aboutmethodology for intensive IT was the in-cidence of hypoglycemia during inten-sive IT; however, that was nonsense. Inour study, because we used the closed-loop glycemic control system in the sur-gical ICU, the true effect of intensive IT inthe clinically ill patients could be evalu-ated without the solution of the occur-rence of hypoglycemia during intensiveIT. Furthermore, the mean amount of in-sulin infusion requirement (range 32–71units/day) was too small to perform theTGC, although the incidence of hypogly-cemia differs between surgical and med-ical populations and patients’ individualinsulin resistance should be evaluated(27). At least 70 units i.v. insulin wasneeded to control the blood glucose lev-els with the postoperative adequate ca-loric requirement that was calculatedaccording to the Harris-Benedict equa-tion, even if the targets of blood glyce-mic concentration were set between 7.7and 10.0 mmol/L (intermediate IT).What is the optimal blood glucose

range to improve morbidity and mortal-ity in surgical patients? Nobody knowsthat a glycemic target range of 7.7–10.0mmol/L (intermediate IT) for a patientin an ICU could be recommended, al-though neglecting hyperglycemia wasstandard ICU care (29). In our study, itremained unclear that the advantage ofintensive IT for patients who underwenthepato-biliary-pancreatic surgery wasan effect in intraoperative IT, postoper-ative IT, or both. Also, it has not beenclarified how long should postoperativeblood glucose levels were controlled ad-equately. In the current study, TGC wasperformed for 18 h in patients with liverresection at the surgical ICU, and excel-lent glucose control was successfully

observed without hypoglycemia byusing a closed-loop glycemic control sys-tem. A corollary issue that has arisenfrom this study is why an abbreviatedperiod of glycemic control had an ef-fect on infection. Note the increase ingrowth of bacteria, including Escherichiacoli, Streptococcus, Proteus, Staphylo-coccus, and Pseudomonas, displayedfor between only 2 h and 18 h at growthcurves of aerobic and facultative organ-isms (30). We believe that the durationof hospitalization after both liver andpancreatic surgery was reduced in pa-tients who received perioperative TGCwith intensive IT and that this could beassociated with the reduction in postop-erative complications due to SSI.

Our results support the conclusionsfrom previous observational studiesthat intensive IT improves the outcomefor critically ill patients, especially in sur-gical ICU patients (2,9). Because all ofthe scheduled patients were recruitedin the current study, the mortality ratewas very low and acceptable comparedwith previous large studies (25). Ofcourse, limitations of our trial includethe use of a subjective criterion thesame as previous reviews and meta-analysis describing bias, chance, andatypical clinical practices (7–9). Al-though our trial was not blinded, whichcould lead to bias, postoperative calorieadministration and surgical manage-ments were matched in both groups bythe same staff at our single center. Wedid not collect specific data to addresspotential biologic mechanisms of the tri-al interventions, and on the basis of theresults in the predefined pairs of sub-groups, we believed that intensive TGCmay benefit some surgical patients. Ourresults suggest that the prospectivelarge randomized control trials shouldbe reprogrammed to evaluate the trueeffect of TGC for patients in both thesurgical and medical ICU for the conclu-sion of the argument over the incidenceof hypoglycemia during intensive IT andpropose that the closed-loop glycemiccontrol system is an easy and effectiveway to control the blood glucose levelsand perform insulin treatment in theICU without occurrence of hypoglyce-mia during intensive IT.

In conclusion, with use of a closed-loop glycemic control system, perioper-ative intensive ITwas significantly betterthan intermediate IT to reduce SSI

and morbidity, resulting in shorter post-operative hospitalization. An importantquestion that has arisen from this studyis how to choose candidates for inten-sive IT using a closed-loop system. Fur-ther studies will be required to addressthis issue.

Funding. This work was supported by the KochiOrganization for Medical Reformation and Re-newal grants.Duality of Interest. No potential conflicts ofinterest relevant to this article were reported.Author Contributions. T.O. wrote the manu-script and researched data. Y.S. reviewed andedited the manuscript. T.Sum., A.K., T.T., and T.I.researched data. T.Sug., M.K., and M.Y. contrib-uted to discussion and reviewed and edited themanuscript. K.H. contributed to discussion, re-viewed and edited the manuscript, researcheddata, and contributed to discussion. T.O. is theguarantor of this work and, as such, had fullaccess to all the data in the study and takesresponsibility for the integrity of the data andthe accuracy of the data analysis.

References1. Finfer S, Chittock DR, Su SY, et al.; NICE-SUGAR Study Investigators. Intensive versusconventional glucose control in critically ill pa-tients. N Engl J Med 2009;360:1283–12972. van den Berghe G, Wouters P, Weekers F,et al. Intensive insulin therapy in critically illpatients. N Engl J Med 2001;345:1359–13673. Patel A, MacMahon S, Chalmers J, et al.; AD-VANCE Collaborative Group. Intensive bloodglucose control and vascular outcomes in pa-tients with type 2 diabetes. N Engl J Med2008;358:2560–25724. Duckworth W, Abraira C, Moritz T, et al.;VADT Investigators. Glucose control and vascu-lar complications in veterans with type 2 diabe-tes. N Engl J Med 2009;360:129–1395. Brunkhorst FM, Engel C, Bloos F, et al.; Ger-man Competence Network Sepsis (SepNet). In-tensive insulin therapy and pentastarchresuscitation in severe sepsis. N Engl J Med2008;358:125–1396. Preiser JC, Devos P, Ruiz-Santana S, et al.A prospective randomised multi-centre con-trolled trial on tight glucose control by intensiveinsulin therapy in adult intensive care units: theGlucontrol study. Intensive Care Med 2009;35:1738–17487. Wiener RS, Wiener DC, Larson RJ. Benefitsand risks of tight glucose control in critically illadults: a meta-analysis. JAMA 2008;300:933–9448. Langley J, Adams G. Insulin-based regimensdecrease mortality rates in critically ill patients:a systematic review. Diabetes Metab Res Rev2007;23:184–1929. Griesdale DE, de Souza RJ, van DamRM, et al.Intensive insulin therapy and mortality amongcritically ill patients: a meta-analysis includingNICE-SUGAR study data. CMAJ 2009;180:821–82710. van den Berghe G. Insulin therapy in theintensive care unit should be targeted to

care.diabetesjournals.org Okabayashi and Associates 1523

maintain blood glucose between 4.4 mmol/land 6.1 mmol/l. Diabetologia 2008;51:911–91511. Harris JA, Benedict FG. A biometric study ofhuman basal metabolism. Proc Natl Acad SciU S A 1918;4:370–37312. Yamashita K, Okabayashi T, Yokoyama T,et al. The accuracy of a continuous blood glu-cose monitor during surgery. Anesth Analg2008;106:160–16313. Yamashita K, Okabayashi T, Yokoyama T,et al. Accuracy and reliability of continuousblood glucose monitor in post-surgical patients.Acta Anaesthesiol Scand 2009;53:66–7114. Hanazaki K, Nose Y, Brunicardi FC. Artificialendocrine pancreas. J Am Coll Surg 2001;193:310–32215. Van den Berghe G, Wilmer A, Hermans G,et al. Intensive insulin therapy in the medicalICU. N Engl J Med 2006;354:449–46116. Van den Berghe G, Schetz M, Vlasselaers D,et al. Clinical review: Intensive insulin therapy incritically ill patients: NICE-SUGAR or Leuvenblood glucose target? J Clin Endocrinol Metab2009;94:3163–317017. Alfonso A, Koops MK, Mong DP, VigerskyRA. Glycemic control with regular versus lisproinsulin sliding scales in hospitalized Type 2 dia-betics. J Diabetes Complications 2006;20:153–15718. Horan TC, Gaynes RP, Martone WJ, JarvisWR, Emori TG. CDC definitions of nosocomial

surgical site infections, 1992: a modification ofCDC definitions of surgical wound infections.Infect Control Hosp Epidemiol 1992;13:606–60819. Furnary AP, Zerr KJ, Grunkemeier GL, Starr A.Continuous intravenous insulin infusion reducesthe incidence of deep sternal wound infection indiabetic patients after cardiac surgical proce-dures. Ann Thorac Surg 1999;67:352–360; dis-cussion 360–36220. Greif R, Akça O, Horn EP, Kurz A, Sessler DI;Outcomes Research Group. Supplemental peri-operative oxygen to reduce the incidence ofsurgical-wound infection. N Engl J Med 2000;342:161–16721. Okabayashi T, Nishimori I, Maeda H,Yamashita K, Yatabe T, Hanazaki K. Effect of in-tensive insulin therapy using a closed-loop gly-cemic control system in hepatic resectionpatients: a prospective randomized clinical trial.Diabetes Care 2009;32:1425–142722. Okabayashi T, Nishimori I, Yamashita K,et al. Continuous postoperative blood glucosemonitoring and control by artificial pancreas inpatients having pancreatic resection: a prospec-tive randomized clinical trial. Arch Surg 2009;144:933–93723. Bell ML, Kenward MG, Fairclough DL,Horton NJ. Differential dropout and bias in rand-omised controlled trials: when it matters andwhen it may not. BMJ 2013;346:e8668

24. BoltzMM,Hollenbeak CS, Julian KG, Ortenzi G,Dillon PW. Hospital costs associated with surgi-cal site infections in general and vascular surgerypatients. Surgery 2011;150:934–94225. de Blacam C, Ogunleye AA, Momoh AO,et al. High body mass index and smoking predictmorbidity in breast cancer surgery: a multivari-ate analysis of 26,988 patients from the nationalsurgical quality improvement program data-base. Ann Surg 2012;255:551–55526. Wukich DK, McMillen RL, Lowery NJ,Frykberg RG. Surgical site infections after footand ankle surgery: a comparison of patientswith and without diabetes. Diabetes Care2011;34:2211–221327. American Society for Parenteral and EnteralNutrition (A.S.P.E.N.) Board of Directors. Clinicalguidelines for the use of parenteral and enteralnutrition in adult and pediatric patients, 2009.JPEN J Parenter Enteral Nutr 2009;33:255–25928. Singer P, Berger MM, Van den Berghe G,et al. ESPEN guidelines on parenteral nutrition:intensive care. Clin Nutr 2009;28:387–40029. Kavanagh BP, McCowen KC. Clinical prac-tice. Glycemic control in the ICU. N Engl J Med2010;363:2540–254630. Bowman RL, Blume P, Vurek GG. Capillary-tube scanner for mechanized microbiology.A photoelectric scanner measures growth inagar-filled capillaries and gives a new approachto microbiology. Science 1967;158:78–83

1524 Intensive Versus Intermediate Glucose Control Diabetes Care Volume 37, June 2014