Table 15

Impact of Perioperative Blood Glucose Concentration on Outcomes

StudyStudy TypeNo. of PtsStudy DesignMean Glucose, mg/dL% of Pts With DiabetesMajor Findings
van Den Berghe et al., 2001 (507)RCT1548 Surgical ICU ptsIntensive intravenous insulin vs. conventional treatmentIntensive 103 ± 19 vs. conventional 153 ± 3313 vs. 13Intensive insulin compared with conventional treatment decreased mortality (8.0% to 4.6%) and major morbidity.
Lazar et al., 2004 (508)RCT141 On-pump CABG ptsGIK vs. SQ insulinGIK 134 ± 4 vs. SQ insulin 267 ± 6100GIK improved 5-year survival and decreased major morbidity.
Lell et al., 2002 (509)RCT46 Off-pump CABG ptsGIK vs. salineGIK 386 ± 152 vs. saline 211 ± 7528 vs. 45No differences in TnI or CK-MB between groups; study terminated owing to concerns of persistent hyperglycemia in the GIK group.
Finney et al., 2003 (510)Prospective observational531 ICU ptsIntravenous insulinN/A16Increased administration of insulin was an independent predictor of ICU mortality; regression models demonstrated a mortality benefit if blood glucose was maintained less than 144 to 200 mg/dL.
Ouattara et al., 2005 (520)Prospective observational200 On-pump CABG ptsInsulin by standardized protocolTightly controlled 147 ± 42 vs. poorly controlled 208 ± 54100Poor intraoperative control of blood glucose was an independent predictor of severe morbidity; mortality rate was increased in pts with poorly controlled glucose (11.4%) vs. those with tightly controlled glucose (2.4%).
McGirt et al., 2006 (511)Retrospective1201 Pts undergoing CEAPostoperative insulin use was nonstandardizedN/A27Multivariate analysis demonstrated that preoperative glucose greater than 200 mg/dL was an independent predictor of 2.8-, 4.3-, and 3.3-fold increases in risk of stroke/TIA, MI, or death.
Gandhi et al., 2005 (512)Retrospective409 Cardiac surgery ptsNonstandardized intraoperative use of insulin in 6%Any adverse event 141 ± 37 vs. no events 127 ± 2528.6 vs. 18Multivariate analysis demonstrated that mean and maximal intraoperative glucose predicted increased mortality. A 20-mg/dL increase in mean intraoperative glucose was associated with a 30% increase in adverse events.
Krinsley 2004 (513)Retrospective1600 Med-Surg ICU ptsHistorical control vs. standardized glucose control protocolHistorical 152 ± 93 vs. protocol 131 ± 5516 vs. 18Decreased mortality, renal insufficiency, and ICU length of stay were observed in the standardized insulin protocol compared with the historical group.
Hill et al., 2000 (514)Retrospective2862 CABG ptsNonstandardized glucose management79–65331Univariate analysis showed no association between maximum blood glucose concentration and mortality.
Krinsley 2003 (515)Retrospective1826 Med-Surg ICU ptsNonstandardized glucose managementSurvivors 138 vs. nonsurvivors 17222Progressive increase in in-hospital mortality rate as blood glucose concentration increased, up to 42.5% among patients with mean glucose values in excess of 300 mg/dL.
Furnary et al., 2003 (516)Retrospective3554 CABG ptsSQ insulin vs. continuous intravenous insulinSQ 213 ± 41 vs. intravenous 177 ± 30100Continuous intravenous insulin was an independent predictor of survival.
Estrada et al., 2003 (517)Retrospective1574 CABG ptsNonstandardized glucose managementDiabetes 214 ± 47 vs. no diabetes 157 ± 3735Hyperglycemia did not predict increased mortality but was associated with increased resource utilization.
McAlister et al., 2003 (518)Retrospective1574 CABG pts92% received intravenous insulin by protocol164–209100Hyperglycemia was an independent predictor of adverse outcomes.

CABG indicates coronary artery bypass graft; CEA, carotid endarterectomy; CK-MB, creatine kinase MB fraction; GIK, glucose insulin potassium; ICU, intensive care unit ; Med-Surg, medical-surgical; MI, myocardial infarction; N/A, not available; pts, patients; RCT, randomized, controlled trial ; SQ, subcutaneous; TIA, transient ischemic attack; and TnI, troponin I.