- Click here for explanation of classification scheme.

Authors hypothesized that a continuous intravenous insulin infusion (CII) in the perioperative period would substantially reduce the incidence of hyperglycemia and therfore significantly lower the rate of deep sternal wound infection (DSWI)in diabetic patients.

All known diabetic patients consecutively admitted to Portland St Vincent Medical Center for open heart surgical procedures between January 1987 and November 1997 were entered into the study.  A diabetic patient was defined as one who experienced chronic glucose intolerance, either insulin dependent or non-insulin dependent, at a time of operation.

Patients who did not carry the preoperative diagnosis of diabetes but who temporarily required insulin in the postoperative period relative to the administration of total parental nutrition or inotropes (eg. epinephrine) and did not have glucose intolerance after the cessation of these modalities were not included.

Recruitment All eligible patients (see inclusion criteria) operated between January 1, 1987 and September 1, 1991 recruited in Control group.  All eligible patients operated on between September 1, 1991 and Nobember 30, 1997 recruited in the Study group.

Design Prospective cohort of diabetic patients

Blinding used (if applicable) none described

Statistical Analysis

Data variables were analyzed using SPSS statistical software.  Mean ± standard error of the mean, were calculated; univariate analysis performed using x² tests for categoric variables, t tests for continuous variables, and the F test to compare variance.  Forward stepwise multivariable logistic regression was used to test the independent association of multiple variables against DSWI.  Multivariable results are given with 95% confidence intervals.

Timing of Measurements mean glucose level on day of operation & POD 1 through 5.  Prior to surgery, weight, height, BMI

Dependent Variables

• Variable 1: Incidence of deep sternal wound infection - defined as chest wound infections involving the sternum or mediastinal tissues, including mediastinitis.
• Variable 2:  Glucose control - blood glucose levels prospectively monitored with Accu-check Easy (Boehringer Mannheim, Indianapolis, IN) every 1 to 2 hours through fingerstick or arterial line drop sample and recorded.  Daily mean blood glucose levels were calculated by averaging all glucose levels obtained clinically during the day of operation and each of the first through fifth POD.

Independent Variables Continuous intravenous insulin infusion (CII).  The insulin drip was titrated on the basis of the most recent fingerstick glucose measurement to maintain blood glucose levels between 150 and 200 mg/dL.  Standardized procedure was developed so that it could be instituted and administered entirely by nursing staff.  The Portland CII protocol consists of starting intravenous insulin infusion dosage, blod glucose testing frequency requirements, insulin infusion titration and cessation orders.

Control Variables Individualized sliding-scale guided subcutaneous insulin injections as method of postoperative glucose regulation.  Treatment was administered every 4 h with goal of keeping blood sugar levels at or below  the "safe" limit of 200 mg/dL.

Initial N: 2,467

Attrition (final N):

Age: 65±10 years with 62% male

Ethnicity: not given

Other relevant demographics: none given

Anthropometrics BMI - Control group 28.4 & Study Group 29.6 with p = 0.0001, weight loss >15% Control Group =8 & Study Group = 11 with p=0.08

Location: Portland St. Vincent Medical Center

1. Deep sternal wound infection occurred in 31 of the 2,467 diabetic patients (1.3% overall).  74% (23 of the 31) of DSWI occurred after initial discharge and thus required a second admission.
2. Mean blood glucose levels on day of operation through POD 3 were significantly lower within the CII (study) group than in the control group (199 ±1.4 versus 241 ±1.9 mg/dL on the day of operation, 176±0.8 versus 206±1.2 mg/dL on POD 1, 181 ±1.2 versus 195±1.3 mg/dL on POL 2 and 179 ±1.5 versus 188±1.4 mg/dL on POD 3, CII group versus control, respectively; p <0.0001 for all comparisons.
3. CII resulted on overall tighter glucose control, with less daily variance, as evidenced by the smaller standard deviation in the CII group (standard deviation, 36 mg/dL for control on POD 1 versus standard deviation, 26 mg/dL for CII on POD 1, p<0.001.
4. Comparison of mean blood glucose levels between patients with DSWI and those without DSWI showed patients without DSWI on POD 1 with a mean blood glucose of 190±1 mg/dL and those with DSWI at 209 mg±7 mg/dL (p=0.02).  On POD 2 patients without DSWI had blood glucose of 188±1 mg/dL and those with DSWI at 206±8 mg/dL (p=0.01).
5. Implementation of the CII protocol resulted in a 2.5-fold decrease in the rate of DSWI compared with that for the control group.  The rate of DSWI dropped from 1.9% (19 of 968) with control to 0.8% (12 of 1,499) with CII (x² test, p-0.011).
6. The final multivariate model run in a nonstepwise fashion revealed that CII (p=0.005; relative risk, 0.34) produced a significant decrease in the risk of DSWI.  Obesity, as measured by BMI, increased the risk of DSWI (p=0.03; relative risk, 1.06), as did the use of an ITA (internal thoracic artery) pedicle (p=0.1; relative risk, 2.0).

The present study demonstrated that (1) hyperglycemia in the immediate postoperative period remains an independent predictor of and may be a causal factor in DSWI in diabetic patients; (2) postoperative hyperglycemia in diabetic patients may be safely and effectively managed through the institution of the Portland CII protocol; and (3) the incidence of DSWI in diabetic patients can be normalized to that of the nondiabetic population through the use of CII protocol that aims to aggressively eliminate postoperative hyperglycemia.

University/Hospital:

Saint Vincent Medical Center