Preoperative Glucose Levels Predict 1-Year Mortality After Surgery

Jim Kling

October 27, 2010

October 27, 2010 (San Diego, California) — At 1 year after noncardiac surgery, high preoperative glucose levels are associated with higher mortality rates. A related study found that diabetics with normal glucose levels are at higher risk for mortality. The study was conducted in noncardiac surgeries. The results were presented here at the American Society of Anesthesiologists 2010 Annual Meeting.

Preoperative hyperglycemia has been linked to higher rates of pulmonary embolism in major joint surgery and to mortality in noncardiac surgery. "We wanted to confirm a relationship between glucose levels before surgery and poor outcomes in a large number of patients," Basem Abdelmalak, MD, staff anesthesiologist and director of anesthesia for bronchoscopic surgery at the Cleveland Clinic, in Ohio, told Medscape Medical News. Dr. Abdelmalak presented 1 study and was involved in the other.

The studies focused on the same group of patients, who were ASA grade I to IV and underwent elective noncardiac surgery between January 2005 and November 2009. Primary outcomes were 1-year mortality and a composite of in-hospital mortality and cardiovascular, neurologic, pulmonary, urological, and infectious complications.

The researchers used additive logistic regression models (ALRMs). They initially used an ALRM without covariates to estimate the unadjusted incidence of each outcome as a function of preoperative glucose. Covariable-adjusted incidence was then estimated using a multivariable ALRM. The researchers adjusted for type of surgery and baseline comorbidities univariably significant with preoperative glucose at the P < .05 level.

The analysis included 61,536 patients. The researchers found a relationship between composite in-house outcomes and preoperative glucose. Euglycemic patients had outcome rates of 8% to 11%, whereas hyperglycemic patients had rates of 12% to 16% (P < .001). After adjustment for covariables, incidence was found not to be related to preoperative glucose levels (P = .37).

In all, 44,461 patients completed 1-year follow-up, and 1-year mortality was estimated to be 5.35% (Bonferroni-adjusted 95% confidence interval, 5.11% - 5.59%). Among patients with preoperative glucose levels in the euglycemic range (at 85 mg/dL), the incidence was 3.5%. It was greater than 9% for patients with glucose values higher than 160 mg/dL (P < .001 after covariable adjustment).

The absence of a relationship between preoperative glucose levels and in-hospital outcomes was surprising and contradicted previous studies. "Further studies are needed to confirm these findings and identify a plausible explanation," the researchers write in the abstract.

A second study, conducted by members of the same group, looked at hyperglycemia in nondiabetic and diabetic patients undergoing noncardiac surgery, and its relationship to morbidity and mortality. They used the same patient population. For each outcome, they estimated the unadjusted and covariable-adjusted incidence as a function of preoperative blood glucose (BG) for diabetics and nondiabetics. They adjusted for type of surgery and baseline comorbidities and estimated odds ratio functions comparing diabetics with nondiabetics.

Overall, 15.8% of patients were diabetic, and they had a crude incidence of composite in-hospital outcome of about 15%, regardless of baseline blood glucose. Among nondiabetic patients, the incidence was 7% to 9% for euglycemic patients and 13% to 15% for patients whose blood glucose levels were higher than 150 mg/dL. However, when adjusted for covariables, there was no significant difference between diabetics and nondiabetics in the relationship between preoperative BG and postoperative complications (P = .048; Bonferroni-adjusted significance criterion of 0.025).

In diabetics, 1-year mortality was 8% to 11% across blood glucose concentrations, with the exception of those near the lower euglycemic range, whose rate was 10% to 18%. In nondiabetics, incidence was strongly related to baseline BG: 3% to 5% for euglycemic patients and more than 11% for patients with BG levels of more than 200 mg/dL. The difference between diabetics and nondiabetics remained after adjustment for covariables (P < .001).

The finding that diabetics in the euglycemic range had higher mortality was surprising. "We think perhaps diabetics who have lived with high glucose levels for quite some time have become accustomed to it. Perhaps their bodies can't tolerate lower glucose levels," said Dr. Abdelmalak.

Physicians routinely advise diabetics on how to adjust medications before surgery to achieve target glucose levels. "If these findings get reproduced and this is not the best glucose level for them, we need to revisit these recommendations," said Dr. Abdelmalak.

He hopes that the findings will stimulate further research to determine optimal presurgical management for diabetics and nondiabetics. "There may be different targets for different patients, and we need to study the implication of the strategies on improving patient outcomes and safety."

The study points to the power of anesthesiologists to influence surgical outcomes, according to John F. Dombrowski, MD, president of the Washington Pain Center in Washington, DC, who attended the presentation. "You don't think of someone's blood sugar being related to what happens a year after surgery. It shows you the power of anesthesia in preoperative care. If we can [better understand] glucose, we can truly make a difference in survival a year later," Dr. Dombrowski told Medscape Medical News.

The finding that nondiabetics with high glucose levels had worse survival should also spur action among anesthesiologists, according to Arthur Boudreaux, MD, clinical professor of anesthesiology at the University of Alabama at Birmingham. "It might mean we should further investigate that patient before we proceed with whatever elective procedure we're planning," Dr. Boudreaux said.

Dr. Abdelmalak, Dr. Dombrowski, and Dr. Boudreaux have disclosed no relevant financial relationships.

American Society of Anesthesiologists 2010 Annual Meeting: Abstract 720. Presented October 17, 2010. Abstract 794. Presented October 19, 2010.