Inadequate Blood Glucose Control Is Associated With In-Hospital Mortality and Morbidity in Diabetic...

ISSN: 1524-4539 Copyright © 2008 American Heart Association. All rights reserved. Print ISSN: 0009-7322. Online

72514Circulation is published by the American Heart Association. 7272 Greenville Avenue, Dallas, TX

DOI: 10.1161/CIRCULATIONAHA.107.706416 published online Jun 30, 2008; Circulation

R. Ascione, C. A. Rogers, C. Rajakaruna and G. D. Angelini Morbidity in Diabetic and Nondiabetic Patients Undergoing Cardiac Surgery

Inadequate Blood Glucose Control Is Associated With In-Hospital Mortality and

http://circ.ahajournals.orglocated on the World Wide Web at:

The online version of this article, along with updated information and services, is

http://www.lww.com/reprintsReprints: Information about reprints can be found online at

[email protected]. E-mail:

Fax:Kluwer Health, 351 West Camden Street, Baltimore, MD 21202-2436. Phone: 410-528-4050. Permissions: Permissions & Rights Desk, Lippincott Williams & Wilkins, a division of Wolters

http://circ.ahajournals.org/subscriptions/Subscriptions: Information about subscribing to Circulation is online at

by on May 18, 2011 circ.ahajournals.orgDownloaded from

http://circ.ahajournals.org

http://circ.ahajournals.org/subscriptions/

mailto:[email protected]

http://www.lww.com/reprints


Inadequate Blood Glucose Control Is Associated WithIn-Hospital Mortality and Morbidity in Diabetic and

Nondiabetic Patients Undergoing Cardiac SurgeryR. Ascione, FRCS; C.A. Rogers, PhD; C. Rajakaruna, MRCS; G.D. Angelini, FRCS

Background—Derangement of glucose metabolism after surgery is not specific to patients with diabetes mellitus. Weinvestigated the effect of different degrees of blood glucose control (BGC) on clinical outcomes after cardiac surgery.

Methods and Results—We analyzed 8727 adults operated on between April 1996 and March 2004. The highest bloodglucose level recorded over the first 60 hours postoperatively was used to classify patients as having good (�200mg/dL), moderate (200 to 250 mg/dL), or poor (�250 mg/dL) BGC; 7547 patients (85%) had good, 905 (10%) hadmoderate, and 365 (4%) had poor BGC. Patients with inadequate BGC were more likely to present with advanced NewYork Heart Association class, congestive heart failure, hypertension, renal dysfunction, and ejection fraction �50%(P�0.001). We found that 52% of patients with poor, 31% with moderate, and 8% with good BGC had diabetesmellitus. Inadequate BGC, but not diabetes mellitus (P�0.79), was associated with in-hospital mortality (good, 1.8%;moderate, 4.2%; poor, 9.6%; adjusted odds ratio: poor versus good BGC, 3.90 [95% confidence interval, 2.47 to 6.15];moderate versus good BGC, 1.68 [95% confidence interval, 1.25 to 2.25]). Inadequate BGC also was associated withpostoperative myocardial infarction (eg, odds ratio, poor versus good BGC: 2.73 [95% confidence interval, 1.74 to4.26]) and with pulmonary and renal complications in patients without known diabetes mellitus (eg, odds ratio, poorversus good BGC: 2.27 [95% confidence interval, 1.65 to 3.12] and 2.82 [95% confidence interval, 1.54 to 5.14]respectively).

Conclusions—More than 50% of patients with moderate to poor BGC after cardiac surgery were not previously identifiedas diabetic. Inadequate postoperative BGC is a predictor of in-hospital mortality and morbidity. (Circulation. 2008;118:113-123.)

Key Words: cardiopulmonary bypass � diabetes mellitus � glucose � metabolism � risk factors

Historically, diabetes mellitus (DM) has been associatedwith a poor clinical outcome after cardiac surgery,

including a higher incidence of wound infections, ischemicevents, neurological and renal complications, and mortali-ty.1–5 Over the last decade, the incidence of DM has increasedmarkedly in developed countries. Knowledge of the patient’sdiabetic status preoperatively has led to advances in periop-erative clinical management, including active and continuousblood glucose control (BGC) with improved clinical out-come.6 Nevertheless, derangement of glucose metabolismafter surgery is not specific to patients with DM.7,8 It has beenreported that up to 90% of those without DM had problemswith their blood glucose homeostasis as a result of varioussurgical stresses.7,8 In such patients, the disturbances in bloodglucose homeostasis have been attributed to insulin resistanceand/or a failure of pancreatic �-cell function caused by thesystemic inflammatory response syndrome after cardiopul-monary bypass (CPB) and its effects on systemictemperature.9–11

Clinical Perspective p 123

Over the last decade, a large body of evidence hashighlighted advances in intraoperative and intensive caretechniques for DM patients undergoing cardiac surgery withimproved in-hospital outcome.1,2 More recently, investigatorshave been focusing on undiagnosed DM and non-DM pa-tients and their likelihood of suffering postoperative derange-ment of glucose metabolism leading to postoperativecomplications.

The aim of this study was to investigate the effect ofdifferent degrees of inadequate BGC on clinical outcomes ina large consecutive series of patients undergoing cardiacsurgery.

MethodsPatient SelectionProspectively collected data were extracted from our hospital data-base (Patient Analysis and Tracking Systems, Dentrite Clinical

Received March 29, 2007; accepted April 29, 2008.From Bristol Heart Institute, University of Bristol, Bristol, United Kingdom.Correspondence to Dr Chris A. Rogers, Clinical Trials and Evaluation Unit, Bristol Heart Institute, Level 7, Bristol Royal Infirmary, Bristol, BS2 8HW,

United Kingdom. E-mail [email protected]© 2008 American Heart Association, Inc.

Circulation is available at http://circ.ahajournals.org DOI: 10.1161/CIRCULATIONAHA.107.706416

113

Cardiovascular Surgery



Systems, London, UK) on consecutive adult cardiac surgical patientsoperated on between April 1996 and March 2004. The data collec-tion form comprises 5 sections completed consecutively by anesthe-tist, surgeon, intensive care unit, high-dependency unit, and wardnurses. The cohort was matched to the department of biochemistrydatabase through the use of patient identifiers and operation date toobtain blood glucose results immediately and 12, 48, and 60 hoursafter surgery. All personal identifiers were then removed to make thedata set anonymous. Patients with no postoperative glucose mea-surements were excluded. Study patients were then classified ashaving good, moderate, or poor glucose control if the highestrecorded blood glucose level was �200 mg/dL (11.1 mmol/L),between 200 and 250 mg/dL (11.1 and 13.9 mmol/L), or �250

mg/dL (13.9 mmol/L), respectively. The cutoff points were based onreports in the literature12–14 and were defined prospectively. Thecomplete patient selection and classification process is shown in Figure1. Initially, data were collected on the basis of presumed consent, butsince 2002, when a program of annual follow-up for all survivingpatients was established, consent for the use of data has been sought.

Anesthetic and Surgical Technique andPostoperative ManagementAnesthetic and surgical techniques were as reported previous-ly.15–17 Briefly, heparin was given to maintain the activatedclotting time at �480 seconds throughout CPB. A standard CPBcircuit was used with nonpulsatile flow at a rate throughout

Figure 2. Protocol for sliding-scale insulin infusion.

337* patients without glucose values were

excluded from the analysis

8727 patients had at least one glucose value available up to 60 hours after the operation

Linked to database containing glucose values recorded immediately post surgery, 12 hours (next

day), 48 hours, and 60 hours after surgery

Blood glucose control classified according to the highest value recorded:Good: 4 – 11 mmol/lModerate: 11.1 –13.9 mmol/lPoor: >13.9 mmol/l

Good 7457 (85.4%)

Moderate905 (10.4%)

Poor 365 (4.2%)

Blood glucose control

9064 patients underwent cardiac surgery from April 1996 – March 2004

* 50 (14.8%) of the 337 excluded patients died following surgery: 30 (60%) on the day of surgery, 9 (18%) on day 1, 2 (4%) on day 2 and 2 (4%) on day 3. The remaining 7 deaths occurred more than 3 days after the operation.

Figure 1. Patient selection and classificationprocess.

114 Circulation July 8, 2008



bypass of 2.4 L · min�1 · m�2 and a mean arterial pressure of 50to 60 mm Hg.

For conventional coronary artery bypass grafting (CABG) opera-tions, systemic temperature was maintained between 34°C and 36°C.Myocardial protection was achieved with intermittent antegradewarm-blood cardioplegia. The technique used for off-pump CABGprocedures has previously been reported.18 For all other surgeries,myocardial protection was predominantly by a combination ofintermittent antegrade and retrograde cold-blood cardioplegia withmoderate systemic hypothermia (28°C to 32°C).

At the end of surgery, patients were transferred to the intensivecare unit and ventilated with 60% oxygen using volume-controlledventilation. The decision to extubate a patient was at the discretion ofthe consultant anesthetist, who followed a predefined protocol basedon the presence of systemic temp of 36°C, a systolic blood pressure�100 mm Hg, pulse rate �100 bpm, and a blood loss �100 mL/hand decreasing from the chest drains.

Glucose Management ProtocolAll diabetic patients were started on a sliding-scale insulin infusionsoon after surgery to maintain blood glucose levels between 5 and

8 mmol/L according to a standard protocol (Figure 2). This infusionwas continued for the first 24 hours; then, patients were switched totheir baseline medications, and blood sugar levels were monitoredevery 6 hours. For nondiabetic patients, the same sliding-scaleinsulin infusion was begun postoperatively if a single blood glucosemeasurement �8 mmol/L or 2 consecutive blood glucose measure-ments �7 mmol/L were observed.

Management of MedicationHypoglycemic medications, statins, �-blockers, diuretics, antihyper-tensives, angiotensin-converting enzyme inhibitors, and calciumchannel blockers were routinely omitted on the day of surgery.Aspirin was omitted 3 days before surgery. On the first postoperativeday, all patients were started on enoxaparin (20 mg), which wasstopped at discharge, and aspirin (300 mg), which was reduced to 75mg after 3 months. On day 1, statins, �-blockers, diuretics, antihy-pertensives, and angiotensin-converting enzyme inhibitors wererestarted on the basis of accurate monitoring of blood pressure andrenal function. After discharge, these drugs were prescribed andmonitored by the family physician and through routine visits to theCardiology Department outpatient clinic.

Table 1. Baseline Characteristics

BGC, n (%)

Variable Good Moderate Poor P

n 7457 (85.4) 905 (10.4) 365 (4.2)

Age, y* 65.5 (57.6–71.7) 66.3 (59.6–71.6) 66.1 (59.4–70.9) 0.057

Body mass index, kg/m2* 26.7 (24.2–29.5) 27.1 (24.4–30.4) 27.0 (24.3–29.9) 0.013

Female gender 1833 (24.6) 247 (27.3) 103 (28.2) 0.072

Parsonnet score* 7 (3–13) 9 (5–15) 9 (6–16) �0.001

Dyspnea class (NYHA class)

1, 2 4589 (61.7) 471 (52.1) 180 (49.6) �0.001

3, 4 2853 (38.3) 433 (47.9) 183 (50.4)

CCS score

0–2 3884 (52.2) 438 (48.4) 164 (45.0) 0.004

3, 4 3559 (47.8) 467 (51.6) 200 (55.0)

Congestive heart failure 1299 (17.5) 221 (24.4) 101 (27.8) �0.001

Previous myocardial infarction 2627 (35.3) 334 (37.0) 148 (40.6) 0.090

History of diabetes 595 (8.0) 285 (31.5) 189 (51.8) �0.001

Hypertension 3970 (53.4) 542 (59.9) 210 (57.7) 0.001

Hypercholesterolemia† 3956 (61.8) 472 (55.9) 201 (58.8) 0.003

Smoking history

Never smoked 2404 (32.3) 275 (30.5) 126 (34.5) 0.011

Past smoker 4115 (55.3) 546 (60.4) 197 (54.0)

Current smoker 923 (12.4) 82 (9.1) 42 (11.5)

Renal failure 144 (1.9) 37 (4.1) 12 (3.3) �0.001

Arrhythmia 677 (9.1) 108 (11.9) 49 (13.5) 0.001

Ejection fraction, %

�50 5315 (71.3) 556 (61.4) 216 (59.2) �0.001

�50 2065 (27.7) 337 (37.2) 142 (38.9)

Not measured 77 (1.0) 12 (1.3) 7 (1.9)

Operative priority

Elective 3478 (46.7) 427 (47.2) 170 (46.6) �0.001

Urgent 3754 (50.4) 427 (47.2) 169 (46.6)

Emergency 220 (2.9) 50 (5.5) 26 (7.1)

CCS indicates Canadian Cardiovascular Society.*Median and interquartile range.†Not recorded since April 2003.

Ascione et al Poor Blood Glucose Control After Cardiac Surgery 115



DefinitionsDiabetic patients were defined as those admitted to the hospital witha comorbid diagnosis of DM not controlled by diet alone. No patientwas classified as having DM on the basis of glucose testingimmediately before surgery. The highest postoperative glucose levelwas the single highest value from all the available postoperativerecordings for each patient. Perioperative death was defined as anydeath occurring in hospital. In-hospital complications were recordedand defined as previously reported.15,17

Statistical AnalysisBaseline and operative characteristics and postoperative outcomeswere compared by use of a �2 or Fisher’s exact test (categoricalvariables) or the Kruskal-Wallis test (continuous variables). Out-comes for the BGC subgroups also were compared by use of multiplelogistic regression (STATA version 9.2). A backward stepwiseprocedure, with significance levels of 0.05 and 0.10 for the additionand exclusion of variables, respectively, was used to determinewhich prognostic variables were retained in the regression models.BGC and DM were included in all regression models regardless ofthe statistical significance. Prognostic variables considered werechosen because they showed imbalance between the groups and/orthey were thought to affect outcome. Missing prognostic data wereimputed with the median for continuous measures and the mode forcategory measures. With 1 exception (hypercholesterolemia), datawere missing for �1% of cases; for hypercholesterolemia (13%missing), a separate missing data category was created. Outcomedata were available for �99% of patients. The effect of BGC onoutcome for DM and non-DM patients was examined by addinginteraction terms to the regression models. This was done both aspart of the stepwise procedure and subsequent to the stepwisemodeling with consistent results. The significance of interactionterms also was assessed with bootstrapping, again with consistentresults. All regression analyses used robust SEs clustered by thesurgeon to allow for any nonindependence between patients operatedon by the same surgeon. Results are presented as adjusted odds ratios(ORs) with 95% confidence intervals derived from 200 bootstrapsamples. When a significant interaction was found (P�0.05), theeffects are presented separately for the DM/non-DM and BGC

subgroups. For BGC, the P values reported are for the overall effectof BGC on the outcome. To ensure that the adjusted analyses hadsufficient power to identify important predictors of the outcome,only those postoperative outcomes experienced by at least 150patients were considered. The number of outcomes of interestresulted in a large number of statistical comparisons. No correctionwas made for multiple comparisons, but confidence intervals andexact P values are presented throughout, and the Bonferroni-corrected probability value for overall 5% statistical significance isincluded as a footnote. The interpretation of the findings takes intoaccount the consistency of the findings and their magnitude, as wellas their statistical significance.

The authors had full access to and take full responsibility for theintegrity of the data. All authors have read and agree to themanuscript as written.

ResultsA total of 9064 patients underwent cardiac surgery betweenApril 1996 and March 2004 at our institution. Postoperativeblood glucose measurement(s) were available for 8727 ofthese patients; 7457 (85.4%) had good, 905 (10.4%) hadmoderate, and 365 (4.2%) had poor BGC in the first 60 hoursafter surgery. The distribution of baseline characteristics ofthe 3 groups is shown in Table 1. Patients with moderate orpoor BGC were more likely to have an advanced New YorkHeart Association class and Canadian Cardiovascular Societyclass, a history of congestive heart failure, hypertension,arrhythmia, renal failure, and an ejection fraction �50%(P�0.004). We found that 48.2% of patients in the poor BGCgroup and 68.5% in the moderate group were nondiabetic,although moderate BGC and poor BGC were more prevalentamong DM patients (P�0.001; Figure 3a). Overall, thenumber of patients with inadequate BGC declined over time(P�0.001; Figure 3b).

Operation details and postoperative outcomes are shown inTable 2. The majority of patients underwent CABG (�70%

89.6

8.12.3

55.7

26.717.7

020

4060

8010

0

Pe

rcen

tage

of p

atie

nts

Non-DM DM

(a) by diabetic status .

73.6

18.38.1

86.8

9.73.4

91.8

5.8 2.4

020

4060

8010

0

Pe

rcen

tage

of p

atie

nts

4002-10021002-99919991-6991

(b) by year of surgery .

Good BGC Moderate BGC Poor BGC

Figure 3. BGC (a) in patients with and without DM and (b) by year of surgery.




in each group). Overall, 35.5% of CABG patients wereoperated off pump. In total, 3962 patients (45.5%) requiredinotropic support after surgery: 3263 (43.8%), 482 (53.3%),and 217 (59.5%) in the good, moderate, and poor BGCgroups, respectively (P�0.001). In-hospital mortality was2.3% overall; 131 (1.8%), 38 (4.2%), and 35 (9.6%) deathsoccurred in the good, moderate, and poor BGC groups,respectively (P�0.001). All postoperative complications var-ied significantly across the BGC groups (P�0.006). Exceptfor transient ischemic attacks and gastrointestinal complica-tions, all complications occurred most often in the poor BGCgroup. The intensive care unit and total postoperative stayalso were significantly longer in the poor BGC group(P�0.001). Table 3 shows the independent predictors ofin-hospital mortality. Two models were fitted: 1 with DMfitted as a single group and 1 with the DM group subdividedinto non–insulin-dependent DM and insulin-dependent DM.DM per se was not identified as an independent predictor forin-hospital mortality (P�0.79), and the risk was similar forthe non–insulin-dependent and insulin-dependent DM sub-groups (P�0.47; data not shown). After controlling forconfounding factors associated with in-hospital death and

diabetic status, inadequate BGC was found to be an indepen-dent predictor of in-hospital death (P�0.001). The mortalityrisk associated with poor BGC was greater than with moder-ate control (OR, 2.32; 95% CI, 1.28 to 4.20; P�0.005) andwas greater than the difference between moderate and goodBGC control (OR, 1.68; 95% CI, 1.25 to 2.25; P�0.001).Other independent predictors of in-hospital death identifiedwere age �65 years, female gender, advanced New YorkHeart Association class, renal failure, arrhythmias, ejectionfraction �50%, presence of left main stem disease, aorticprocedures, CPB time �90 minutes, operative priority, andthe need for inotropes (P�0.022 for all).

The effects of DM and inadequate BGC on mortality andmorbidity after multivariate adjustment controlling for con-founding factors associated with outcome are shown inTables 4 and 5, respectively. In non-DM patients, inadequateBGC was found to be an independent predictor of pulmonary,renal, and gastrointestinal complications (overall effect ofBGC, P�0.001). In this patient group, both moderate andpoor BGC carried a higher risk of complications comparedwith good BGC, and the risk associated with moderate BGCand poor BGC were similar (P�0.19). In contrast, for

Table 2. Type of Operation and Clinical Outcomes

BGC, n (%)

Variable Good Moderate Poor P *

n 7457 (85.4) 905 (10.6) 365 (4.2)

Operation type

CABG 5367 (72.0) 635 (70.2) 258 (70.7) �0.001

Valve surgery 1167 (15.7) 137 (15.1) 42 (11.5)

CABG and valve 671 (9.0) 111 (12.3) 57 (15.6)

Congenital and other 252 (3.4) 22 (2.4) 8 (2.2)

Aortic procedures 247 (3.3) 28 (3.1) 16 (4.4) 0.49

Inotropic support 3263 (43.8) 482 (53.3) 217 (59.5) �0.001

In-hospital death 131 (1.8) 38 (4.2) 35 (9.6) �0.001

MI 133 (1.8) 25 (2.8) 16 (4.4) 0.001

Arrhythmias 1982 (26.8) 256 (28.4) 123 (34.2) 0.006

Pulmonary complication 989 (13.4) 184 (20.5) 84 (23.4) �0.001

Chest infection 468 (6.3) 99 (11.0) 42 (11.7)

Reintubated 201 (2.7) 51 (5.7) 29 (8.1)

Tracheostomy 134 (1.8) 36 (4.0) 24 (6.7)

Neurological complication 277 (3.7) 67 (7.5) 29 (8.1) �0.001

Permanent stroke 46 (0.6) 15 (1.7) 7 (2.0)

Transient ischemic attack 57 (0.8) 14 (1.6) 2 (0.6)

Infective complication 295 (4.0) 55 (6.1) 30 (8.4) �0.001

Renal complication 385 (5.2) 88 (9.8) 54 (15.0) �0.001

Gastrointestinal complication 110 (1.5) 34 (3.8) 12 (3.3) �0.001

Multisystem failure† 74 (1.2) 27 (3.2) 20 (5.9) �0.001

Reoperation for bleeding/tamponade 290 (3.9) 45 (5.0) 33 (9.1) �0.001

ITU stay, d‡ 1 (1–2) 1 (1–2) 1 (1–3) �0.001

Postoperative stay, d‡ 7 (6–9) 8 (6–11) 8 (6–12) �0.001

*Uncorrected P values; variables with a value of P�0.0042 are significant at the 5% level after Bonferronicorrection for multiple comparisons.

†Not recorded since April 2003.‡Median and interquartile range.




patients with diagnosed DM, the risk of pulmonary and renalcomplications was similar across the 3 BGC groups(P�0.07). Complication rates by BGC for non-DM and DMpatients are shown in Figure 4. The risk of these complica-tions in the moderate and poor BGC groups was lower for apatient with DM compared with a non-DM patient; thedifference in risk between non-DM and DM patients wasgreatest among those with poor BGC (Table 4).

No significant interaction between BGC and DM wasfound for postoperative myocardial infarction (MI), arrhyth-mias, neurological or infective complications, or reoperationfor bleeding/tamponade (P�0.05). DM was found to carry anincreased risk for neurological and infective complications(estimated increased risk, 63% [P�0.003] and 34%[P�0.032], respectively) but not for postoperative MI, whichwas less common among DM patients (estimated 33% re-duced risk; P�0.034). In contrast, inadequate BGC wasstrongly associated with postoperative MI; the risk associatedwith poor BGC was greater than with moderate control (68%;P�0.037) and was similar to the difference between moder-ate and good BGC control (62%; P�0.006). Inadequate BGCalso was associated with neurological complications (overalleffect of BGC, P�0.003) but not with infective complica-

tions (P�0.20). Neither DM nor BGC was independentlyassociated with arrhythmias, and DM was not predictive ofthe need for reoperation for bleeding/tamponade. However,inadequate BGC was found to be associated with reoperationfor bleeding/tamponade (overall effect of BGC, P�0.026).Poor BGC carried a higher risk compared with both good andmoderate BGC, but the risk associated with good BGC andwith moderate BGC was similar (P�0.38).

DM, ejection fraction �50%, advanced New York HeartAssociation class, type of operation, use of CPB, operativepriority, and inotropic support were found to be associatedwith poor BGC (Table 6). The effect of DM differed betweenthose patients given and those not given inotropes, as evi-denced by a statistically significant interaction between DMand inotropic support (P�0.001). Poor BGC was mostcommon among DM patients not requiring inotropes (OR,22.7); in DM patients given inotropes, the risk of poor BGCwas reduced by �75%, although it remained significantlyhigher than for a non-DM patient who required inotropes(OR, 5.71). Similarly, in non-DM patients, the risk of poorBGC increased 3-fold when inotropes were given (OR, 3.10)compared with DM patients in whom the risk reduced.

DiscussionMore than 50% of patients with poor and moderate BGCwere not previously identified as diabetic. This is consistentwith the Euro Heart Study of cardiology patients withcoronary artery disease in which 50% of patients not previ-ously identified as diabetic were found to have diabetes orimpaired glucose tolerance19 but is in contrast to the study byLauruschkat and colleagues,20 who suggested that the preva-lence of undiagnosed DM in patients undergoing CABG isonly 5.2%. This discrepancy can be only partly explained byour classifying patients with diet-controlled DM as nondia-betic because patients with diet-controlled DM made up only4.5% of the moderate and 6.6% of the poor BGC groups. Thestress of cardiac surgery might uncover a borderline diabeticstatus causing marked transient or permanent imbalance inbody sugar control and leading to hyperglycemia. Becausehyperglycemia has been associated with poor in-hospital andlong-term outcome, including further progression of nativecoronary artery disease,4 our results raise the question of theneed for a more accurate preoperative diagnosis of silentdiabetic status. Hyperglycemia early after cardiac surgery, inthe absence of excessive glucose infusion, would occur as aresult of either insulin resistance or decreased insulin produc-tion caused by pancreatic �-cell failure.10 Disturbances inpancreatic �-cell secretion in the absence of autoimmunediabetes have been shown to be related to hypothermia duringCPB,21 which recovers fully at the end of operation, withnormal insulin secretion potential in the postoperative peri-od.10 Insulin resistance, however, also is caused by catechol-amines and cortisone secretion (surgical stress),22,23 CPB, theaccompanying systemic inflammatory response syndrome,and the effects of systemic heparinization.9 Some of thefindings of the present study also support this hypothesisbecause the use of CPB was found to be independentlyassociated with poor BGC.

Table 3. Independent Predictors of In-Hospital Mortality

Multivariate*

Variable OR 95% CI‡ P †

Age �65 y 1.54 1.00–2.38 0.050

Female gender 1.64 1.23–2.19 0.001

Diabetes mellitus 0.93 0.53–1.61 0.79

BGC

Good 1.00 �0.001

Moderate 1.68 1.25–2.25

Poor 3.90 2.47–6.15

Dyspnea class, NYHA class 3, 4 1.82 1.37–2.41 �0.001

Renal failure 3.47 1.56–7.70 0.002

Arrhythmia 1.45 1.05–1.99 0.022

Ejection fraction �50% 1.95 1.25–3.03 0.003

Left main stem disease 1.71 1.32–2.21 �0.001

Operation type

CABG 1.00 0.015

Valve surgery 1.39 0.67–2.91

CABG and valve 1.58 1.14–2.17

Congenital and other 2.01 0.91–4.40

Aortic procedure 2.87 1.67–4.95 �0.001

CPB used 0.74 0.53–1.04 0.08

CPB time �90 min 1.44 1.06–1.96 0.020

Operative priority

Elective 1.00 �0.001

Urgent 1.53 1.09–2.15

Emergency 6.64 4.62–9.56

Inotropic support 4.72 2.97–7.49 �0.001

*Hosmer-Lemeshow goodness-of-fit test, P�0.38; c statistic�0.87.†Estimated from 200 bootstrap samples.




Our findings suggest that inadequate BGC, regardless ofdiabetic status, is an independent predictor of in-hospitalmortality, postoperative MI, and neurological complicationsin patients undergoing cardiac surgery. Our data alsosuggest that in patients without diagnosed DM, inadequateBGC is a predictor of renal, pulmonary, and gastrointesti-nal complications.

Our results can be explained in several ways. Acutehyperglycemia occurring intraoperatively abolishes ischemicpreconditioning24 and amplifies reperfusion injury25 to theheart. In addition, during ischemia, glucose is the preferredsubstrate for the myocardium, but marked insulin resistanceleads to hyperglycemia as a result of impaired cell uptake ofglucose, which in turn leads to increased concentrations offree fatty acids. Fatty acids are detrimental to the ischemic

myocardium because of the increased oxygen consumptionrequired to metabolize the new substrate.26 Hyperglycemiaalso leads to increased free radical release and hence in-creased oxidative stress, causing endothelial dysfunction,which may further affect myocardial ischemia.27–29

Several mechanisms might explain the hyperglycemia-related brain damage suggested by our study. The flowreductions during hyperglycemia could be due to the hyper-osmolality of glucose. However, intraperitoneal glucose in-jected into rats to produce hyperglycemia was associated with24% reduction in regional blood flow, whereas injection ofD-mannitol to produce an equivalent osmolality reducedcerebral blood flow by only 10% compared with controls.30

Hyperglycemia also increases local edema and causesglucose-mediated oxidative stress and inflammation.31

Table 4. Effect of Diabetes Mellitus on Postoperative Outcomes: Adjusted Effect Sizes

Adjusted Effect Size*P, Interaction Between

BGC and DM‡Outcome OR 95% CI† P ‡

In-hospital death 0.93 0.53–1.62 0.79 0.51

MI 0.66 0.45–0.97 0.034 0.12

Arrhythmias 1.06 0.91–1.24 0.43 0.15

Pulmonary complications§ �0.001 �0.001

Good BGC 1.44 1.15–1.80

Moderate BGC 0.63 0.38–1.03

Poor BGC 0.44 0.17–1.12

Chest infection§ �0.001 0.003

Good BGC 1.29 0.93–1.77


Poor BGC 0.46 0.16–1.28

Neurological complications 1.63 1.18–2.24 0.003 0.068

Infective complications 1.34 1.02–1.76 0.032 0.42

Renal complications§ 0.001 0.042

Good BGC 2.20 1.42–3.41


Poor BGC 0.72 0.24–2.15

Gastrointestinal complications§ 0.002 0.002

Good BGC 2.50 1.25–5.00


Poor BGC 0.15 0.04–0.58

Reoperation forbleeding/tamponade

0.86 0.59–1.24 0.43 0.15

*The following variables were considered for inclusion in the adjustment: age �65 years, gender, body mass index(categorized as �25, �25 and �30, �30), NYHA class 3 or 4, Canadian Cardiovascular Sociaty class 3 or 4, previousMI, hypertension, hypercholesterolemia, smoker (grouped as never, past, current), renal failure, arrhythmia, ejectionfraction �50%, triple-vessel coronary disease, left main stem disease, operative priority (grouped routine, urgent,emergency), use of cardiopulmonary bypass, bypass time �90 minutes, operation performed (grouped as CABG,valve, CABG and valve, congenital and other), aortic procedure, and need for inotropic support after surgery; thosevariables significant at P�0.10 were chosen (see Methods).

†Estimated from 200 bootstrap samples.‡Uncorrected probability values estimated from 200 bootstrap samples; variables with a value of P�0.005 are

significant at the 5% level after Bonferroni correction for multiple comparisons.§ORs suggest that DM is “protective” for patients whose blood glucose is poorly controlled (OR decreases with

decreased BGC). This phenomenon is illustrated in Figure 3. In non-DM patients, the complication rate increases withdecreasing BGC; in the DM group, the rate is similar across the 3 BGC groups. The difference in complication ratesbetween DM and non-DM groups is greatest for patients with poor BGC, with the DM group having the lower rate ineach case; hence the OR for DM is �1 and lowest for the poor BGC group.




DM has been shown to be a catalyst for infective compli-cations after cardiac surgery,32–34 and our study supports thisfinding. Although the association between BGC and infectivecomplications was not statistically significant (P�0.20), atrend toward a higher risk of complications with inadequateBGC was observed, which is in keeping with another report.5

Carr et al7 demonstrated that tight glucose control in bothdiabetic and nondiabetic patients leads to a marked reductionin the incidence of mediastinitis. Acute hyperglycemia hasseveral effects on innate immunity: It reduces neutrophil andcomplement activity; it increases proinflammatory cytokinestumor necrosis factor-� and interleukin-6; and by reducingendothelial nitric oxide formation, it decreases microvascularreactivity to dilating agents such as bradykinin and impairscomplement function.35

The pathophysiology of ischemic acute renal failure in-volves a complex interplay between renal hemodynamics,tubular, endothelial cell injury, and inflammatory process.29

Endothelial dysfunction caused by hyperglycemia contributesat the microvascular level, initiating and subsequently extend-ing the tubular injury.36 These detrimental endothelial effects

also might explain the increased incidence of pulmonary andgastrointestinal complications observed in our study. Othershave reported that hyperglycemia may be associated withimpairment of renal blood flow and glomerular filtration ratethrough a tubuloglomerular feedback mechanism.37

Our findings suggest that our insulin infusion protocolwas not effective in maintaining tight blood sugar controlin all patients regardless of their diabetic status. Since thisanalysis, we have extended our insulin infusion protocol to48 hours after surgery to all patients regardless of theirdiabetic status and have adopted a stricter attitude towardinitiating insulin infusion in presence of outliers (ie, patientswith blood glucose level �8 mmol/L) regardless of theirpreoperative diabetic status. Furthermore, we now aim tokeep blood glucose levels between 4.4 and 6.1 mmol/L incritically ill patients in intensive care, as suggested by Vanden Berghe et al.38

The study has limitations. First, it is a retrospectiveanalysis, although the data were collected prospectively.Those responsible for data collection were not blinded to thepatient’s blood glucose levels, so the possibility of observer

Table 5. Effect of BGC on Postoperative Outcomes: Adjusted Effect Sizes

Adjusted Effect Size*

Moderate vs GoodBGC

Poor vs GoodBGC

Poor vs ModerateBGC

Outcome OR 95% CI† OR 95% CI† OR 95% CI† P ‡P, Interaction Between

BGC and DM§

In-hospital death 1.68 1.25–2.25 3.90 2.47–6.15 2.33 1.28–4.20 �0.001 0.51

MI 1.62 1.15–2.29 2.73 1.74–4.26 1.68 1.03–2.72 �0.001 0.12

Arrhythmias 0.95 0.82–1.10 1.18 0.89–1.57 1.25 0.98–1.57 0.15 0.15

Pulmonary complications �0.001

Non-DM 1.71 1.28–2.27 2.27 1.65–3.12 1.33 0.86–2.05 �0.001

DM 0.74 0.50–1.10 0.70 0.32–1.51 0.94 0.34–2.59 0.07

Chest infection �0.001

Non-DM 1.90 1.39–2.61 2.18 1.15–4.12 1.14 0.61–2.15 �0.001

DM 0.94 0.57–1.55 0.66 0.31–1.40 0.75 0.24–2.27 0.64

Neurological complications 1.54 1.16–2.05 1.35 0.96–1.90 0.87 0.56–1.35 0.003 0.068

Infective complications 1.19 0.95–1.49 1.42 0.72–2.78 1.19 0.59–2.42 0.20 0.42

Renal complications 0.042

Non-DM 1.70 1.08–2.66 2.82 1.54–5.14 1.65 0.74–3.70 �0.001

DM 0.85 0.49–1.47 0.92 0.40–2.10 1.08 0.42–2.77 0.84

Gastrointestinal complications 0.002

Non-DM 2.25 1.70–2.97 2.70 1.80–4.06 1.20 0.81–1.77 �0.001

DM 0.99 0.47–2.04 0.16 0.04–0.60 0.16 0.06–0.46 0.003

Reoperation for bleeding/ tamponade 1.17 0.82–1.68 2.08 1.20–3.59 1.78 1.10–2.87 0.026 0.15

*The following variables were considered for inclusion in the adjustment: age �65 years, gender, body mass index (categorized as �25, �25 and �30, �30)NYHA class 3 or 4, Canadian Cardiovascular Society class 3 or 4, previous MI, hypertension, hypercholesterolemia, smoker (grouped as never, past, current), renalfailure, arrhythmia, ejection fraction �50%, triple-vessel coronary disease, left main stem disease, operative priority (grouped routine, urgent, emergency), use ofcardiopulmonary bypass, bypass time �90 minutes, operation performed (grouped as CABG, valve, CABG and valve, congenital, and other), aortic procedure, andneed for inotropic support after surgery; those variables significant at P�0.10 were chosen (see Methods).

†Estimated from 200 bootstrap samples.‡Test for an overall effect of blood glucose control on outcome. Uncorrected probability values estimated from 200 bootstrap samples. Variables with a value of

P�0.0042 are significant at the 5% level after Bonferroni correction for multiple comparisons.§Uncorrected probability values estimated from 200 bootstrap samples. Variables with a value of P�0.005 are significant at the 5% level after Bonferroni correction

for multiple comparisons.




bias cannot be discounted. Nonetheless, when the data werecollected, this analysis had not been planned. Second, it couldbe argued that some of our findings are due to the misclas-sification of DM patients. However, classifying patientsaccording to their diabetic history, as provided by the refer-ring cardiologist, reflects clinical practice. Using fastingglucose also has the potential for misclassification. The

glucose level depends on the accuracy of the fasting durationand the choice of cutoff for DM. Furthermore, obtainingfasting glucose results is not feasible for urgent and emer-gency patients (�50% of study cohort). Third, operationscarried out since work on this study was started were notincluded. The risk profile of patients treated and operativeprocedures have not changed since 2004, but protocolchanges made in light of these findings precluded ourextending the study cohort to include more recent operations.Fourth, it is possible that the differences or similaritiesobserved between the groups were a result of unmeasuredconfounders or variables we did not consider for inclusion inthe analysis. However, we considered all variables thatshowed imbalance between the groups and/or were thought toaffect outcome, including the need for inotropic support aftersurgery. However, we were unable to comment on thespecific effect of �-1 agonists, which are known to beassociated with glucose homeostasis, because this informa-tion was not available; only the use of inotropes, not the type(ie, �-1 agonist, selective phosphodiesterase inhibitor, orother type), is recorded in the database. In addition, antidia-betic medications may affect the perioperative period becausethey are continued until the day of surgery, but BGC ismonitored in all DM patients in the intensive care unit, and aninsulin infusion is started when the BGC rises above thethreshold. Although multivariate analysis is a theoreticallysound statistical method of accounting for differences be-tween groups in the absence of random allocation, it isrecognized that no statistical model can fully account for allthe differences in risk profile between the diabetic andnondiabetic patients in our study cohort. Nevertheless, thisstudy was based on a large cohort of patients, and theanalyses were limited to those outcomes that occurred in atleast 150 patients, thereby ensuring the study had sufficientpower to identify factors significantly associated with the

010

2030

40P

erc

enta

ge o

f pa

tient

s


NonDM

DM NonDM

DM NonDM

DM

Pulmonary complications

05

1015

20P

erc

enta

ge o

f pa

tient

s


NonDM

DM NonDM

DM NonDM

DM

Chest infection

05

1015

2025

Pe

rcen

tage

of p

atie

nts


NonDM

DM NonDM

DM NonDM

DM

Renal complications

02

46

Pe

rcen

tage

of p

atie

nts


NonDM

DM NonDM

DM NonDM

DM

Gastrointestinal complications

Note: the scale of the y-axis varies across the four charts

Figure 4. Prevalence of complications byBGC and DM.

Table 6. Independent Predictors of Poor BGC

Multivariate*

Variable OR 95% CI† P †

Diabetes mellitus‡

Without inotropic support 22.7 14.0–36.8 �0.001

With inotropic support 5.71 4.11–7.93 �0.001

Inotropic support‡

Non-DM patient 3.10 2.15–4.46 �0.001

DM patient 0.78 0.56–1.08 0.13

Dyspnea class, NYHA class 3, 4 1.16 0.98–1.37 0.077

Ejection fraction �50% 1.22 0.95–1.57 0.11

Operation type

CABG 1.00 0.001

Valve surgery 0.85 0.61–1.17

CABG and valve 1.48 1.08–2.01

Congenital and other 0.76 0.41–1.39

CPB used 2.49 1.58–3.94 �0.001

Operative priority

Elective 1.00 0.037

Urgent 0.78 0.57–1.06

Emergency 1.90 1.06–3.41

*Hosmer-Lemeshow goodness-of-fit test, P�0.76; c statistic�0.81.†Estimated from 200 bootstrap samples.‡Test for interaction between diabetes and inotropic support, P�0.001.




outcome. A fifth limitation is the classification of patientsaccording to their postoperative BGC. We defined the clas-sification in advance of any analysis for the purpose of thisstudy because no recognized guidelines are available on thistopic. The 200-mg/dL (11.1-mmol/L) threshold for moderateBGC was selected because it is the cutoff point used todiagnose DM from random glucose levels,12 whereas the250-mg/dL (13.9-mmol/L) threshold for poor BGC was basedon previous reports indicating that patients with glucosevalues exceeding this level were at higher risk of postopera-tive complications, particularly deep wound infections.13,14 Itwould have been useful to classify patients by their hemo-globin A1c results along with the BGC data, but unfortunately,this information was not available for our study cohort. Wealso decided before any analysis was undertaken to excludepatients with diet-controlled diabetes (2.5% of the studycohort) from the diabetic group and include them with thepatients without known diabetes. We acknowledge that this isa conservative strategy that may have diluted any effect ofdiabetes on outcome. Nevertheless, reclassifying these pa-tients as diabetic in a posthoc analysis did not alter our overallconclusions; the estimated ORs followed a pattern consistentwith our original analysis, and no new statistically significanteffects were found. Finally, although we have identifiedsignificant associations between inadequate BGC and out-come, we cannot infer that poor BGC necessarily causes thepoor outcome; increased blood sugar levels can result frompostoperative complications in addition to contributing totheir occurrence.

ConclusionsOur study shows that �50% of patients developing moderateto poor BGC after cardiac surgery were not previouslyidentified as diabetic. Moderate to poor BGC is an indepen-dent predictor of in-hospital mortality and is strongly associ-ated with morbidity in patients not known to be diabetic. Ofthe baseline variables, use of CPB, type of surgery, advancedNew York Heart Association class, and poor left ventricularejection fraction were identified as predictors of poor BGCand might be used preoperatively to risk stratify patients andto optimize their clinical management.

Sources of FindingWe thank the British Heart Foundation and the Garfield WestonTrust for their financial support.

DisclosuresNone.

References1. Furnary AP, Gao G, Grunkemeier GL, Wu Y, Zerr KJ, Bookin SO, Floten

HS, Starr A. Continuous insulin infusion reduces mortality in patientswith diabetes undergoing coronary artery bypass grafting. J Thorac Car-diovasc Surg. 2003;125:1007–1021.

2. Furnary AP, Zerr KJ, Grunkemeier GL, Starr A. Continuous intravenousinsulin infusion reduces the incidence of deep sternal wound infection indiabetic patients after cardiac surgical procedures. Ann Thorac Surg.1999;67:352–360.

3. Lazar HL, Chipkin SR, Fitzgerald CA, Bao Y, Cabral H, Apstein CS.Tight glycemic control in diabetic coronary artery bypass graft patientsimproves perioperative outcomes and decreases recurrent ischemicevents. Circulation. 2004;109:1497–1502.

4. Ouattara A, Lecomte P, Le Manach Y, Landi M, Jacqueminet S, PlatonovI, Bonnet N, Riou B, Coriat P. Poor intraoperative blood glucose controlis associated with a worsened hospital outcome after cardiac surgery indiabetic patients. Anesthesiology. 2005;103:687–694.

5. Zerr KJ, Furnary AP, Grunkemeier GL, Bookin S, Kanhere V, Starr A.Glucose control lowers the risk of wound infection in diabetics after openheart operations. Ann Thorac Surg. 1997;63:356–361.

6. Flaherty JD, Davidson CJ. Diabetes and coronary revascularization.JAMA. 2005;293:1501–1508.

7. Carr JM, Sellke FW, Fey M, Doyle MJ, Krempin JA, de la Torre R,Liddicoat JR. Implementing tight glucose control after coronary arterybypass surgery. Ann Thorac Surg. 2005;80:902–909.

8. Smith CE, Styn NR, Kalhan S, Pinchak AC, Gill IS, Kramer RP, SidhuT. Intraoperative glucose control in diabetic and nondiabetic patientsduring cardiac surgery. J Cardiothorac Vasc Anesth. 2005;19:201–208.

9. Anderson RE, Brismar K, Barr G, Ivert T. Effects of cardiopulmonarybypass on glucose homeostasis after coronary artery bypass surgery. EurJ Cardiothorac Surg. 2005;28:425–430.

10. Doenst T, Wijeysundera D, Karkouti K, Zechner C, Maganti M, Rao V,Borger MA. Hyperglycemia during cardiopulmonary bypass is an inde-pendent risk factor for mortality in patients undergoing cardiac surgery.J Thorac Cardiovasc Surg. 2005;130:1144.

11. Lehot JJ, Piriz H, Villard J, Cohen R, Guidollet J. Glucose homeostasis:comparison between hypothermic and normothermic cardiopulmonarybypass. Chest. 1992;102:106–111.

12. Peters AL, Davidson MB, Schriger DL, Hasselblad V. A clinicalapproach for the diagnosis of diabetes mellitus: an analysis using glyco-sylated hemoglobin levels: Meta-analysis Research Group on theDiagnosis of Diabetes Using Glycated Hemoglobin Levels. JAMA. 1996;276:1246–1252.

13. Fish LH, Weaver TW, Moore AL, Steel LG. Value of postoperative bloodglucose in predicting complications and length of stay after coronaryartery bypass grafting. Am J Cardiol. 2003;92:74–76.

14. Golden SH, Peart-Vigilance C, Kao WH, Brancati FL. Perioperativeglycemic control and the risk of infectious complications in a cohort ofadults with diabetes. Diabetes Care. 1999;22:1408–1414.

15. Ascione R, Caputo M, Calori G, Lloyd CT, Underwood MJ, AngeliniGD. Predictors of atrial fibrillation after conventional and beating heartcoronary surgery: a prospective, randomized study. Circulation. 2000;102:1530–1535.

16. Lim KH, Caputo M, Ascione R, Wild J, West R, Angelini GD, Bryan AJ.Prospective randomized comparison of CarboMedics and St Jude Medicalbileaflet mechanical heart valve prostheses: an interim report. J ThoracCardiovasc Surg. 2002;123:21–32.

17. Narayan P, Caputo M, Rogers CA, Alwair H, Mahesh B, Angelini GD,Bryan AJ. Early and mid-term outcomes of surgery of the ascendingaorta/arch: is there a relationship with caseload? Eur J CardiothoracSurg. 2004;25:676–682.

18. Watters MP, Ascione R, Ryder IG, Ciulli F, Pitsis AA, Angelini GD.Haemodynamic changes during beating heart coronary surgery with the“Bristol Technique.” Eur J Cardiothorac Surg. 2001;19:34–40.

19. Bartnik M, Ryden L, Ferrari R, Malmberg K, Pyorala K, Simoons M,Standl E, Soler-Soler J, Ohrvik J, for the Euro Heart Survey Investigators.The prevalence of abnormal glucose regulation in patients with coronaryartery disease across Europe: the Euro Heart Survey on diabetes and theheart. Eur Heart J. 2004:25;1880–1890.

20. Lauruschkat AH, Arnrich B, Albert AA, Walter JA, Amann B, RosendahlUP, Alexander T, Ennker J. Prevalence and risks of undiagnosed diabetesmellitus in patients undergoing coronary artery bypass grafting.Circulation. 2005;112:2397–2402.

21. Mills NL, Beaudet RL, Isom OW, Spencer FC. Hyperglycemia duringcardiopulmonary bypass. Ann Surg. 1973;177:203–205.

22. Svensson S, Svedjeholm R, Ekroth R, Milocco I, Nilsson F, Sabel KG,William-Olsson G. Trauma metabolism and the heart: uptake of sub-strates and effects of insulin early after cardiac operations. J ThoracCardiovasc Surg. 1990;99:1063–1073.

23. Hjemdahl P. Stress and the metabolic syndrome: an interesting butenigmatic association. Circulation. 2002;106:2634–2636.

24. Kersten JR, Toller WG, Gross ER, Pagel PS, Warltier DC. Diabetesabolishes ischemic preconditioning: role of glucose, insulin, and osmo-lality. Am J Physiol Heart Circ Physiol. 2000;278:H1218–H1224.

25. Verma S, Maitland A, Weisel RD, Li SH, Fedak PW, Pomroy NC, MickleDA, Li RK, Ko L. Rao V. Hyperglycemia exaggerates ischemia-reperfusion-induced cardiomyocyte injury: reversal with endothelinantagonism. J Thorac Cardiovasc Surg. 2002;123:1120–1124.




26. Liu Q, Docherty JC, Rendell JC, Clanachan AS, Lopaschuk GD. Highlevels of fatty acids delay the recovery of intracellular pH and cardiacefficiency in post-ischemic hearts by inhibiting glucose oxidation. J AmColl Cardiol. 2002;39:718–725.

27. Koltai MZ, Hadhazy P, Posa I, Kocsis E, Winkler G, Rosen P, Pogatsa G.Characteristics of coronary endothelial dysfunction in experimentaldiabetes. Cardiovasc Res. 1997;34:157–163.

28. Gross ER, LaDisa JF Jr, Weihrauch D, Olson LE, Kress TT, HettrickDA, Pagel PS, Waritier DC, Kersten JR. Reactive oxygen speciesmodulate coronary wall shear stress and endothelial function duringhyperglycemia. Am J Physiol Heart Circ Physiol. 2003;284:H1552–H1559.

29. Sack MN, Yellon DM. Insulin therapy as an adjunct to reperfusion afteracute coronary ischemia: a proposed direct myocardial cell survival effectindependent of metabolic modulation. J Am Coll Cardiol. 2003;41:1404–1407.

30. Duckrow RB, Beard DC, Brennan RW. Regional cerebral blood flowdecreases during hyperglycemia. Ann Neurol. 1985;17:267–272.

31. Garg R, Chaudhuri A, Munschauer F, Dandona P. Hyperglycemia,insulin, and acute ischemic stroke: a mechanistic justification for a trial ofinsulin infusion therapy. Stroke. 2006;37:267–273.

32. Brandt M, Harder K, Walluscheck KP, Fraund S, Boning A, Cremer J.Coronary artery bypass surgery in diabetic patients. J Card Surg. 2004;19:36–40.

33. Carson JL, Scholz PM, Chen AY, Peterson ED, Gold J, Schneider SH.Diabetes mellitus increases short-term mortality and morbidity in patientsundergoing coronary artery bypass graft surgery. J Am Coll Cardiol.2002;40:418–423.

34. Szabo Z, Hakanson E, Svedjeholm R. Early postoperative outcome andmedium-term survival in 540 diabetic and 2239 nondiabetic patientsundergoing coronary artery bypass grafting. Ann Thorac Surg. 2002;74:712–719.

35. Turina M, Fry DE, Polk HC Jr. Acute hyperglycemia and the innateimmune system: clinical, cellular, and molecular aspects. Crit Care Med.2005;33:1624–1633.

36. Molitoris BA, Sutton TA. Endothelial injury and dysfunction: role in theextension phase of acute renal failure. Kidney Int. 2004;66:496–499.

37. Woods LL, Mizelle HL, Hall JE. Control of renal hemodynamics inhyperglycemia: possible role of tubuloglomerular feedback.Am J Physiol. 1987;252:65–73.

38. van den Berghe G, Wouters P, Weekers F, Verwaest C, Bruyninckx F, SchetzM, Vlasselaers D, Ferdinande P, Lauwers P, Bouillon R. Intensive insulintherapy in the critically ill patients. N Engl J Med. 2001;345:1359–1367.

CLINICAL PERSPECTIVEOur study demonstrates that inadequate blood glucose control (BGC) after cardiac surgery is not specific to patients withdiabetes mellitus (DM). Inadequate BGC, regardless of DM status, was independently associated with in-hospital mortalityand morbidity. Our findings have epidemiological, clinical, academic, and financial implications. We suggest that DMpatients represent only a fraction of those suffering derangement of glucose metabolism after surgery. The projected futurenumber of adults with DM is an underestimate of the number likely to be affected by deranged glucose metabolism andits related complications. Inadequate BGC after surgery seems to represent a separate clinical entity that is explained onlypartially by undiagnosed and diet-controlled diabetes. Our data suggest that strict protocols to maintain BGC should beused for all patients. However, the efficacy of these protocols and the pathophysiologic mechanisms of this condition needfurther research. In addition, further research and guidelines as to how best to manage these patients are needed. Currently,important clinical decisions such as choice of screening test, strategy for maintaining adequate BGC, and the ideal targetlevel of BGC are often left to the individual clinician. This has resulted in inconsistencies in the definition of undiagnosedDM, stress hyperglycemia, and inadequate BGC; marked variation in estimates of prevalence; and significant variation intreatment, the impact of which remains uncertain. Our findings also may apply to patients admitted for major noncardiacsurgery. The impact on life expectancy and on hospital resources is potentially enormous.




Inadequate Blood Glucose Control Is Associated With In-Hospital Mortality and Morbidity in Diabetic...

Documents

Transcript of Inadequate Blood Glucose Control Is Associated With In-Hospital Mortality and Morbidity in Diabetic...