Glucose Control Lowers the Risk of Wound Infection in ...Document Transcript
A single copy of this article can be downloaded and printed only for the readers personal research and study. Glucose Control Lowers the Risk of Wound Infection in Diabetics After Open Heart Operations Kathryn J. Zerr, MBA, Anthony P. Furnary, MD, Gary L. Grunkemeier, PhD, Stephen Bookin, MD, Vivek Kanhere, MD, and Albert Starr, MD Albert Starr Academic Center for Cardiac Surgery, Providence St. Vincent Hospital & Medical Center, Portland, Oregon Background. Elevated blood glucose levels in the postoperative period are associated with an increased risk of deep wound infection in diabetic individuals undergoing open heart operations at Providence St. Vincent Hospital. Methods. Of 8,910 patients who underwent cardiac operations between 1987 and 1993, 1,585 (18%) were diabetic. The rate of deep sternal wound infections in diabetic patients was 1.7%, versus 0.4% for nondiabetics. Nine hundred ninety patients had their operation before implementation of the protocol and 595 after implementation. Charts of all diabetic patients were reviewed. Mean blood glucose levels were calculated from documented results of ﬁnger-stick glucometer testing. Results. Thirty-three diabetic patients suffered 35 deep wound infections: 27 sternal (1.7%) and eight at the donor site (0.5%). Infected diabetic patients had a higher mean blood glucose level through the first 2 postoperative days than noninfected patients (208 ± 7.1 versus 190 ± 0.8 mg/dL; p < 0.003) and had a greater body mass index (31.5 ± 1.4 versus 28.6 ± 0.1 kg/m2; p < 0.05). Multivariable logistic regression showed that mean blood glucose level for the first 2 days (p = 0.002), obesity (p < 0.002), and use of the internal mammary artery (p < 0.02) were all independent predictors of deep wound infection. Institution of a protocol of postoperative continuous intravenous insulin to maintain blood glucose level less than 200 mg/dL was begun in September 1991. This protocol resulted in a decrease in blood glucose levels for the first 2 postoperative days and a concomitant decrease in the proportion of patients with deep wound infections, from 2.4% (24/990) to 1.5% (9/595) (p < 0.02). Conclusions. The incidence of deep wound infection in diabetic patients was reduced after implementation of a protocol to maintain mean blood glucose level less than 200 mg/dL in the immediate postoperative period. This article was published in: Ann Thorac Surg; 63:356–361,1997, and is posted with permission from “The Society of Thoracic Surgeons.” Annals of Thoracic Surgery Home Page: http://www.sciencedirect.com/science/journal/00034975
2IntroductionIntensive diabetic therapy with improved blood glucose control has been shown to delay the long-term detrimental effects of microvascular complications  and to prevent impairment of the whiteblood cell’s ability to phagocytose and effectively kill bacteria [2, 3]. The importance of defects inleukocyte function in diabetics was addressed in 1982 by Rayﬁeld and associates  at the MountSinai Medical Center, when they showed a significant positive correlation between mean plasmaglucose levels and the frequency of acute infections. The diminution of intracellular bactericidalactivity of leukocytes to both Staphylococcus aureus and Escherichia coli was shown to have a directrelation to glucose control. Studies by Hennessy and colleagues  at the University of TexasMedical School showed the detrimental effects of short-term hyperglycemia on the ability ofimmunoglobulin G to ﬁx complement, one aspect of antimicrobial immune function. Of importance,their studies showed significant increases in glycation after only 16 hours . The same groupdemonstrated that pretreatment of human immunoglobulin with 240 mg of glucose per deciliter ofsaline solution for as short as 8 hours under physiologic conditions stops its ability to extend survivalin an asplenic, infant rat model . Further, extracellular glycosylation of proteins in diabetics hasbeen shown to impair wound healing and is associated with increased collagenase activity anddecreased wound collagen content .Material and MethodsClinical MethodsA diabetic protocol (Appendix 1) was designed and implemented that standardized the titration ofintravenous insulin to maintain the mean blood glucose (BG) level less than 200 mg/dL. Bloodglucose levels were determined by glucometer measurements, and insulin was titrated by critical carenurses. There were no changes in antibiotic protocols during this time, and procedures wereperformed by the same surgical group. Data were collected by retrospective chart review on alldiabetic patients who had cardiac operations between 1987 and 1993 at Providence St. VincentHospital. Table 1 shows baseline data for the patients before and after implementation of theprotocol. Daily mean BG levels were calculated by averaging the levels obtained clinically by ﬁngerstick every 1 to 2 hours and were recorded in the medical record. Levels were compared before andafter implementation of the diabetic patients with deep wound infections were compared with thosewho did not become infected (Fig 2).
3Table 1. Comparison of Patients Before and After Implementation of the Diabetic Protocol Before After Protocol Protocol Variable (n = 990) (n = 595) Not found significant ( p > 0.05) Age 65 65 Sex—male 62% 60% Weight (kg) 83 84 Smoked within 30 d of operation 17% 13% Renal failure 6% 8% Peripheral vascular disease 14% 15% Diabetes type Insulin dependent 35% 38% Oral agent 49% 47% Diet only 11% 10% No treatment 6% 6% Procedure CABG without IMA 32% 33% CABG with IMA 56% 54% Valve only 7% 7% Valve plus CABG 5% 6% Other 1% 1% Redo operation 14% 13% Pump time (min) 90 87 Ventilator > 48 h 8% 11% Inotropic agents > 48 h 14% 11% Mortality 6% 5% Significant (p < 0.05) History of hypertension 54% 64% History of congestive heart failure 28% 23% Renal insufficiency 3% 6% Status at operation Elective 49% 22% Urgent 43% 68% Emergent 6% 7% Salvage 3% 2% Body mass index (kg/cm2) 28.4 29.2CABG = coronary artery bypass grafting; IMA = internal mammary artery.
4Statistical MethodsMultiple variables were considered as possible predictors of deep wound infection. Variables andtheir univariate significance were as follows: body mass index, 0.0015; average BG 48 hours afteroperation, 0.0024; mean BG level on first postoperative day, 0.003; mean BG level on secondpostoperative day, 0.0163; use of internal mammary artery, 0.0183; operation before or after protocolimplemented, 0.077; inotropic agents for more than 48 hours, 0.1243; body surface area, 0.1768;diabetes requiring insulin or oral agents, 0.2256; redo operation, 0.2519; status at time of operation,0.2548; packed red blood cells transfused, 0.2551; steroids, 0.3675; sex, 0.4065; age, 0.4356; andpump time, 0.8173. Univariate and multiple logistic regression analyses were used to develop amodel to predict the risk of deep wound infection for each patient. Analysis was done by paired t test,X2 test, and multivariable logistic regression using SPSS (Chicago, IL) statistical software.Definitions were as follows: Body mass index = weight/height2; body surface area = body weight0.425x height0.725 x 0.007184 (Dubois’ equation); obesity = body mass index greater than 27.5. “Deepwound infection” included mediastinitis, sternal wound infections involving the sternum and deeper,and vein donor-site infections involving Scarpa’s fascia and deeper. “Diabetic patients” includedthose who were insulin dependent and non–insulin-dependent at the time of operation. We excludedpatients who were not diabetic but temporarily required insulin in the postoperative period related tothe administration of total parenteral nutrition or inotropic agents (i.e., epinephrine), and who did notrequire therapy after the discontinuation of these treatments.“Elective status at operation” denoted one that was performed on a patient with cardiac function thathad been stable in the days or weeks before operation. Cases are usually scheduled at least 1 daybefore the surgical procedure. “Urgent status at operation” included any patient who did not have togo to the operating room emergently but required operation on the basis of medical necessity beforedischarge. These patients had unstable symptoms or critical anatomy and often required intravenousintervention, ie, nitroglycerin, heparin, or inotropic support, for stabilization before operation. Thesepatients did not ﬁt into the elective, emergent or salvage categories. “Emergent status at operation”denoted cases that permitted no delay in operative intervention. Patients requiring emergencyoperation had ongoing, refractory, unrelenting cardiac compromise, with or without hemodynamicinstability, and were not responsive to any form of therapy except cardiac operation. “Salvage statusat operation” involved any patient in extremis or in cardiogenic shock going into the operating room.This included patients who came to the operating room with cardiopulmonary resuscitation in
5progress. “Renal insufficiency” was deﬁned as physician-documented renal insufficiency or patientswith an increase in the creatinine level to greater than 2.0 mg/dL.Clinical MaterialIn all, 8,910 patients underwent cardiac operations between 1987 and 1993; 1,585 were diabetic(18%): 35% insulin dependent, 47% taking oral agents, 10% diet controlled, and 6% on no treatmentbefore admission. The mean age was 65 ± 9.7 years; 61% (963/1,585) were male. There were 1,378coronary artery bypass grafting operations (63% using the internal mammary artery), 110 valveoperations, 84 valve/coronary artery bypass grafting, and 12 other operations; 213 (14%) were redo-sternotomies. The overall mortality rate in the diabetic population was 5.7% (90/1,585). Thirty-threepatients suffered 35 deep wound infections (2.1%): sternal in 27 and donor in eight. Of those patientswho died, 6.7% (6/90) had deep wound infection. Of 33 patients who had deep wound infections, 6died (18%). Diabetic patients who were admitted with endocarditis (n = 6) were included, but none ofthem had deep wound infection.Fig 1. Mean blood glucose levels in all diabetic patients before and after implementation of thediabetic protocol.
6ResultsImplementation of the diabetic protocol in September 1991 resulted in a decrease in the mean BGlevels on the first postoperative day (206 versus 172 mg/dL; p < 0.005) and the second postoperativeday (195 versus 176 mg/dL; p < 0.002) (see Fig 1), and over 48 hours (201 versus 174 mg/dL; p <0.003). Glucose levels were measured by glucometer tests, as indicated in the protocol (see Appendix1). The rate of deep sternal wound infection in diabetic patients dropped from 2.8% in 1987 to 0.74%in 1993, the third year after implementation. When measured over a 5-year period beforeimplementation of the diabetic protocol, the rate in diabetic patients was 2.1% versus 0.98% in the 3years after implementation. In nondiabetic patients, the 5-year rate was 0.4%, versus 0.31% for the 3years after implementation. Elevated BG at 48 hours was found to be significantly associated with anincreased risk of deep wound infection (p < 0.002) (see Fig 2; Fig 3). When looked at over time, therate of infection in diabetic patients dropped after implementation of the diabetic protocol (Fig 4).Univariate regression analysis of variables considered as possible predictors of deep wound infectionin diabetic patients revealed the following to be significant at p < internal mammary artery. Variablesin the multivariable 0.05: body mass index, average BG at 48 hours, BG on the predictive modelfound to be significant at p < 0.05 were first and second postoperative days, and use of the body massindex, use of the internal mammary artery, and mean BG levels at 48 hours greater than 200 mg/dL(Table 2). Postoperative BG and the deep wound infection rate showed a significant direct relation (p= 0.002) (Fig 5). This model demonstrates the ability to identify diabetic patients at higher risk ofdeep wound infection after cardiac operations (Fig 6).Fig 2. Comparison of mean blood glucose levels in diabetic patients (Pts) with and without deepwound infection (includes mediastinitis, sternal wound infections involving the sternum and deeper,and vein donor-site infections involving Scarpa’s fascia and deeper).
7Fig 3. Mean blood glucose levels on the first postoperative day (POD) for patients with deep woundinfection and for all patients.Fig 4. Deep sternal wound infection (DSWI) in patients having cardiac operations: all patientsversus diabetic patients versus nondiabetic patients.
8 aTable 2. Variables in the Final Predictive Model Significant at p < 0.05 Regression StandardVariable Coefficient Error SignificanceBody mass index (BMI) 0.0799 0.0268 0.0029Internal mammary artery 0.9752 0.4135 0.0183(IMA)Average blood glucose 0.0159 0.0051 0.0017levels 48 h after operation (BG48)a Probability of deep wound infection = eBX/1 + eBX, where BX = 10.0 + 0.0799 x IMA + 0.0159 xBG48. Model provides an estimate of the risk of deep wound infection for each diabetic patienthaving cardiac operation.Fig 5. Significant direct relation shown between postoperative blood glucose level and deep infectionrates.Fig 6. Reliability of the model to assign appropriate risk is shown by comparing groups according tothe expected risk of infection versus actual infection rates.
9CommentDeep wound infection causes increased mortality, lengthy hospital stays, repeated trips to theoperating room, and greater hospital costs, not to mention the pain and suffering of patients and theirfamilies during prolonged convalescence. It is well documented that an increased body mass index(measure of obesity) [8, 9] increases the risk of infection. Reports of increased risk with use of theinternal mammary artery have differed and occasionally conflict among institutions [10 –12]. Whathas not been shown in clinical practice is the effect of intensive control of BG levels in diabeticpatients in the postoperative period. We hypothesize that elevated BG levels after cardiac operationsin diabetic patients are associated with a higher incidence of infectious complications. The 5-yearrate of deep wound infection in patients having cardiac operations at Providence St. Vincent Hospital& Medical Center was six times higher in diabetics than in nondiabetics. The diabetic protocol wasproven successful in lowering the mean BG levels in the immediate postoperative period (see Fig 1).The decrease in infection rates for patients with average BG levels less than 200 mg/dL on the firstpostoperative day was significant (p = 0.018). The rate of deep sternal wound infection in diabeticshas decreased since implementation of the protocol, whereas the rate in nondiabetic patients remainsstable at less than 1%; this shows that the decrease is likely not due to other factors.This was an observational study using retrospective chart review. We recognize the inherentproblems of lack of recognition of important variations among patient populations and the potentialbias to which this type of investigation may be prone. However, we believe that the significantimpact on mean BG levels with use of the protocol is a determining factor in the dramatic decrease inour infection rates. We are confident that the decreased rates before and after implementation of theprotocol (p = 0.14) will show higher significance as more patients are added to the study (see Fig 3).In conclusion, elevated BG levels immediately after cardiac operations in diabetic patients in ourstudy were associated with a higher incidence of deep wound infection. We suggest that protocols formaintaining BG less than 200 mg/dL in the immediate postoperative period may be a factor inreducing the incidence of deep wound infection in diabetic patients.We thank Kenneth Melvin, MD, Endocrinologist and Chief of Medicine, and Peter Fuchs, MD,Director of Laboratory and Hospital Epidemiologist, for their guidance, advice, and support; andStephanie Zerr for data management support.Address reprint requests to Ms Zerr, Providence St. Vincent Hospital & Medical Center, 9205 SWBarnes Rd, Portland, OR 97225.
10References1. The effect of intensive treatment of diabetes on the development and progression of long-termcomplications in in-sulin-dependent diabetes mellitus. The Diabetes Control and Complications TrialResearch Group. N Engl J Med 1993;329:977– 86.2. Sima AA, O’Neill SJ, Naimark D, Yagihashi S, Klass D. Bacterial phagocytosis andintracellular killing by alveolar macrophages in BB rats. Diabetes 1988;37:544–9.3. Masuda M, Murakami T, Egawa H, Murata K. Decreased ﬂuidity of polymorphonuclearleukocyte membrane in strep-tozocin-induced diabetic rats. Diabetes 1990;39:466–70.4. Rayﬁeld EJ, Ault MJ, Keusch GT, et al. Infection and diabetes: the case for glucose control[Review]. Am J Med 1982;72: 439–50.5. Hennessey PJ, Black CT, Andrassy RJ. Nonenzymatic glycosylation of immunoglobulin Gimpairs complement ﬁxation. JPEN J Parenter Enteral Nutr 1991;15:60– 4.6. Black CT, Hennessey PJ, Andrassy RJ. Short-term hyperglycemia depresses immunitythrough nonenzymatic glycosylation of circulating immunoglobulin. J Trauma 1990;30:830–3.7. Hennessey PJ, Ford EG, Black CT, Andrassy RJ. Wound collagenase activity correlatesdirectly with collagen glycosylation in diabetic rats. J Pediatr Surg 1990;25:75– 8.8. Gadaleta D, Risucci DA, Nelson RL, et al. Effects of morbid obesity and diabetes mellitus onrisk of coronary artery bypass grafting. Am J Cardiol 1992;70:1613– 4.9. He G-W, Ryan WH, Acuff TE, et al. Risk factors for operative mortality and sternal woundinfection in bilateral internal mammary artery grafting. J Thorac Cardiovasc Surg 1991; 102:342–7.10. Grossi EA, Esposito R, Harris LJ, et al. Sternal wound infections and use of internalmammary artery grafts. J Thorac Cardiovasc Surg 1991;102:342–7.11. Singh AK, Feng WC. Sternal wound infection after myocardial revascularization with internalmammary artery. Indian Heart J 1993;45:29–31.12. Lust RM, Sun YS, Chitwood WR Jr. Internal mammary artery use. Sternal revascularizationand experimental infection patterns. Circulation 1991;84(Suppl 3):285-289.
11Appendix 1. Postoperative Insulin Protocol for Diabetic Patients1. Start infusion by pump piggyback to maintenance intravenous line as follows. Test blood glucoseby finger-stick method. Blood glucose level (in mg/dL) determines insulin rate: <150, 0 U/h; 150–200, 1 U/h; 201–250, 2 U/h; >251, 3 U/h.2. Frequency of blood glucose testing: (a) postoperatively every hour until stable (when frequent changes in insulin dosage are no longer necessary), then may test every 2 hours; (b) when weaning vasopressor agents (eg, adrenalin), check every 30 minutes until stable.3. Insulin titration based on blood glucose level (in mg/dL): (a) <75—stop insulin, give 25 mL D50 and recheck blood glucose in 30 minutes; when blood glucose >150 mg/dL, restart with rate 50% of previous rate; (b) 75–100—stop insulin; recheck blood glucose in 30 minutes; when blood glucose >150 mg/dL, restart with rate 50% of previous rate; (c) 101–150—decrease rate by 0.5 U/h or, if 10 mg/dL lower than last test, decrease rate by 50%; (d) 151–200—same rate; (e) 201–250—if lower than last test, same rate; if higher than last test, increase rate by 0.5 U/h; and (f) >250—if lower than last test, same rate; if higher than last test, increase rate by 1 U/h. If blood glucose is >251 mg/dL and has not decreased after three hourly increases in insulin, then double the insulin rate.4. Continue intravenous insulin administration postoperatively until patient is taking a full liquid diet.Consult MD for new orders at that time.5. Diabetic diet starts with any oral intake.