Risk factors for surgical site infection following orthopaedic spinal operations

Transcription

1 Washington University School of Meicine Digital ICTS Faculty Publications Institute of Clinical an Translational Sciences 2008 Risk factors for surgical site infection following orthopaeic spinal operations Margaret A. Olsen Washington University School of Meicine Jeffrey J. Nepple Washington University School of Meicine in St. Louis K. Daniel Riew Washington University School of Meicine in St. Louis Lawrence G. Lenke Washington University School of Meicine in St. Louis Keith H. Briwell Washington University School of Meicine in St. Louis See next page for aitional authors Follow this an aitional works at: Part of the Meicine an Health Sciences Commons Recommene Citation Olsen, Margaret A.; Nepple, Jeffrey J.; Riew, K. Daniel; Lenke, Lawrence G.; Briwell, Keith H.; Mayfiel, Jennie; an Fraser, Victoria J., "Risk factors for surgical site infection following orthopaeic spinal operations". The Journal of Bone an Joint Surgery, 90, 1, Paper 6. This Article is brought to you for free an open access by the Institute of Clinical an Translational Sciences at Digital It has been accepte for inclusion in ICTS Faculty Publications by an authorize aministrator of Digital For more information, please contact

2 Authors Margaret A. Olsen, Jeffrey J. Nepple, K. Daniel Riew, Lawrence G. Lenke, Keith H. Briwell, Jennie Mayfiel, an Victoria J. Fraser This article is available at Digital

3 This is an enhance PDF from The Journal of Bone an Joint Surgery The PDF of the article you requeste follows this cover page. Risk Factors for Surgical Site Infection Following Orthopaeic Spinal Operations Margaret A. Olsen, Jeffrey J. Nepple, K. Daniel Riew, Lawrence G. Lenke, Keith H. Briwell, Jennie Mayfiel an Victoria J. Fraser J Bone Joint Surg Am. 2008;90: oi: /jbjs.f This information is current as of August 3, 2010 Reprints an Permissions Publisher Information Click here to orer reprints or request permission to use material from this article, or locate the article citation on jbjs.org an click on the [Reprints an Permissions] link. The Journal of Bone an Joint Surgery 20 Pickering Street, Neeham, MA

4 62 COPYRIGHT Ó 2008 BY THE JOURNAL OF BONE AND JOINT SURGERY, INCORPORATED Risk Factors for Surgical Site Infection Following Orthopaeic Spinal Operations By Margaret A. Olsen, PhD, MPH, Jeffrey J. Nepple, MD, K. Daniel Riew, MD, Lawrence G. Lenke, MD, Keith H. Briwell, MD, Jennie Mayfiel, BSN, MPH, CIC, an Victoria J. Fraser, MD Investigation performe at Washington University School of Meicine, St. Louis, Missouri Backgroun: Surgical site infections are not uncommon following spinal operations, an they can be associate with serious morbiity, mortality, an increase resource utilization. The accurate ientification of risk factors is essential to evelop strategies to prevent these potentially evastating infections. We conucte a case-control stuy to etermine inepenent risk factors for surgical site infection following orthopaeic spinal operations. Methos: We performe a retrospective case-control stuy of patients who ha ha an orthopaeic spinal operation performe at a university-affiliate tertiary-care hospital from 1998 to Forty-six patients with a superficial, eep, or organ-space surgical site infection were ientifie an compare with 227 uninfecte control patients. Risk factors for surgical site infection were etermine with univariate analyses an multivariate logistic regression. Results: The overall rate of spinal surgical site infection uring the five years of the stuy was 2.0% (forty-six of 2316). Univariate analyses showe serum glucose levels, preoperatively an within five ays after the operation, to be significantly higher in patients in whom surgical site infection evelope than in uninfecte control patients. Inepenent risk factors for surgical site infection that were ientifie by multivariate analysis were iabetes (os ratio = 3.5, 95% confience interval = 1.2, 10.0), suboptimal timing of prophylactic antibiotic therapy (os ratio = 3.4, 95% confience interval = 1.5, 7.9), a preoperative serum glucose level of >125 mg/l (>6.9 mmol/l) or a postoperative serum glucose level of >200 mg/l (>11.1 mmol/l) (os ratio = 3.3, 95% confience interval = 1.4, 7.5), obesity (os ratio = 2.2, 95% confience interval = 1.1, 4.7), an two or more surgical resients participating in the operative proceure (os ratio = 2.2, 95% confience interval = 1.0, 4.7). A ecrease risk of surgical site infection was associate with operations involving the cervical spine (os ratio = 0.3, 95% confience interval = 0.1, 0.6). Conclusions: Diabetes was associate with the highest inepenent risk of spinal surgical site infection, an an elevate preoperative or postoperative serum glucose level was also inepenently associate with an increase risk of surgical site infection. The role of hyperglycemia as a risk factor for surgical site infection in patients not previously iagnose with iabetes shoul be investigate further. Aministration of prophylactic antibiotics within one hour before the operation an increasing the antibiotic osage to ajust for obesity are also important strategies to ecrease the risk of surgical site infection after spinal operations. Level of Evience: Prognostic Level III. See Instructions to Authors for a complete escription of levels of evience. Surgical site infection is the most common hospitalacquire infection that occurs in the early postoperative perio in surgical patients 1. Unite States hospitals participating in the National Nosocomial Infections Surveillance (NNIS) System, a voluntary performance-measurement system operate by the Centers for Disease Control an Prevention (CDC), monitor rates of surgical site infection following laminectomy an spinal arthroesis. The most recent NNIS Disclosure: In support of their research for or preparation of this work, one or more of the authors receive, in any one year, outsie funing or grants in excess of $10,000 from the Centers for Disease Control an Prevention (UR8/CCU715087), the National Institutes of Health/National Institute of Allergy an Infectious Diseases (Career Development Awars [K01AI an K24AI067794]), an the Washington University School of Meicine Summer Research Program National Institutes of Health training grant (5T35HL007815). Neither they nor a member of their immeiate families receive payments or other benefits or a commitment or agreement to provie such benefits from a commercial entity. No commercial entity pai or irecte, or agree to pay or irect, any benefits to any research fun, founation, ivision, center, clinical practice, or other charitable or nonprofit organization with which the authors, or a member of their immeiate families, are affiliate or associate. J Bone Joint Surg Am. 2008;90:62-9 oi: /jbjs.f.01515

5 63 summary by the CDC reporte a 1.25% rate of surgical site infection after laminectomy an a 2.1% rate following spinal arthroesis 2. Rates of surgical site infection reporte from iniviual institutions have range from 0% to 15%, epening on the reason for the operation, the site, the approach, an the use of instrumentation 3-8. A wie variety of risk factors for surgical site infection after spinal operations have been reporte in the literature. However, many of the stuies were limite by their relatively small sample size, which restricts the ability to perform multivariate analyses to ientify inepenent risk factors for infection 9. Another potential problem with the currently available literature is the use of nonstanar efinitions an variable time-frames for surveillance of surgical site infection, which makes comparison of results between stuies ifficult. A thir problem is that many stuies inclue only a small fraction of all potential risk factors for surgical site infection in their analyses. In orer to accurately ientify inepenent risk factors for surgical site infection, stuies with relatively large numbers of patients with surgical site infection nee to be performe, with the investigators incluing a wie variety of potential risk factors, using stanar accepte efinitions of surgical site infection, an controlling for the occurrence of multiple risk factors within iniviual patients by performing multivariate statistical analyses. We recently escribe inepenent risk factors for surgical site infection following spinal operations performe by neurosurgeons 10. In that stuy, we foun postoperative incontinence, a posterior surgical approach, an operation for tumor resection, an morbi obesity to be associate with an increase risk of surgical site infection in a multivariate analysis of a population of patients treate with spinal surgery in which the overall rate of surgical site infection was 2.76%. We unertook a subsequent retrospective case-control stuy to etermine if we coul ientify unique risk factors for surgical site infection in patients unergoing orthopaeic spinal surgery. We suspecte that the risk factors in our orthopaeic patient population might be ifferent from those in the neurosurgical spine population. This report escribes risk factors for surgical site infection following spinal operations performe by orthopaeic surgeons over a five-year time perio at a tertiary-care university-affiliate hospital. Materials an Methos Stuy Design an Inclusion an Exclusion Criteria We performe a retrospective neste case-control stuy at a tertiary-care university-affiliate hospital after obtaining approval from our institutional review boar. Patients who ha unergone a spinal operation were ientifie by querying the hospital Meical Informatics atabase for amissions coe with International Classification of Diseases, Ninth Revision, Clinical Moification (ICD-9-CM) proceure coes for laminectomy (03.02 an 03.09), iscectomy (80.50 an 80.51), an/or spinal arthroesis (81.00 to 81.09) from January 1998 through December Eligible operations were restricte to those performe by an orthopaeic surgeon TABLE I Demographics an Surgical Characteristics of the 2316 Patients Treate with Orthopaeic Spinal Surgery from 1998 to 2002 Characteristic Mean age (range) (yr) 52.4 ( ) Female gener (no.) 1213 (52.4%) Mean boy mass inex (range) (kg/m 2 ) 28.4 ( ) Type of surgery (no.) Laminectomy only (incluing laminoplasty) 326 (14.1%) Discectomy ± laminectomy 309 (13.3%) Arthroesis ± instrumentation 1657 (71.5%) Instrumentation only 24 (1.0%) Amissions with >1 spinal op. (no.) 87 (3.8%) at our institution in patients fifteen years of age or oler (n = 2316). Other spinal proceures an operations performe by neurosurgeons were exclue. In aition, operations performe in patients with an amission ICD-9-CM iagnosis coe for intraspinal abscess (324.1), osteomyelitis (730.08, , an ), or surgical site infection (998.5, , an ) were exclue. The basic emographic an surgical characteristics of the cohort are shown in Table I. Ientification of Surgical Site Infection Patients likely to have a surgical site infection were ientifie with use of a combination of ICD-9-CM iagnosis coes suggestive of infection, a reamission iagnosis of infection, an/or positive microbiological cultures of specimens from the woun. The ICD-9-CM coes use as inicators of possible surgical site infection inclue coes for surgical site infection (998.5, , an ), cellulitis (682.1, 682.2, an 682.6), osteomyelitis ( an ), ehiscence (998.3 an ), or intraspinal abscess (324.1). The meical recors of patients with inicators of potential surgical site infection uring the hospitalization for the initial surgery or at the time of reamission to the hospital within one year after the operation were reviewe for recore signs an symptoms of surgical site infection. In aition, all microbiology, raiology, pathology, an operative reports were reviewe to etermine if the CDC/NNIS efinitions of surgical site infection were met 11. Inclue in the CDC/NNIS efinition is any physician iagnosis of surgical site infection; therefore, if the spine surgeon or consulting infectious isease physician note the presence of infection in the meical recor, that was consiere proof of surgical site infection. The CDC/NNIS efinitions inclue eep surgical site an organ-space infections if they ha an onset within thirty ays after the operation (or within one year if the operation inclue placement of an implant) an superficial surgical site infection with an onset within thirty ays after the operation 11. Deep surgical site infection involve eep soft tissues (fascia an muscle), whereas organ-space infections inclue osteomyelitis, meningitis, an empyema (following

6 64 anterior thoracic proceures). All surgical site infections were confirme by the nurse infection-control practitioner responsible for spine operations (J.M.). Forty-six patients ha a confirme surgical site infection involving the spinal incision. In aition, four patients with an ICD-9-CM coe inicating surgical site infection were exclue because the infection involve only the bone-graft onor site (the hip), an three were exclue because the infection was at a istant site (unrelate to the spine operation). In five patients with an ICD-9-CM coe inicating surgical site infection at the time of reamission, the infection i not meet the CDC/NNIS efinition of surgical site infection; those patients were exclue as well. The meical recors of 227 control patients without a surgical site infection, selecte with use of a ranom-number generator from the cohort of patients treate with an orthopaeic spinal operation at our institution from 1998 to 2002, were reviewe. This resulte in approximately five uninfecte control patients for each patient with a surgical site infection. Fifteen control patients without a surgical site infection ha ha a two-stage spinal operation performe on separate ays uring the surgical amission. In orer to perform subsequent analyses that inclue surgical variables at the patient level, one of the two stage operations was ranomly selecte for all subsequent analyses. One patient with a surgical site infection ha been scheule to unergo two stage operations, but an infection was iagnose uring the secon operation. Since the surgical site infection was attribute to the first operation, that operation was use for the analysis. Data Collection All ata, incluing the type, approach, an level of the operative proceure; potential risk factors for surgical site infection; an signs an symptoms of surgical site infection were collecte from the meical recors by two investigators (J.J.N. an M.A.O.), using a stanarize ata collection form. In aition, the orthopaeic spine surgeons atabase was use to verify the type of operative proceure, approach, level, source of bone graft, an use of instrumentation. Potential risk factors for surgical site infection inclue a wie variety of emographic, comorbi, operative, an postoperative variables, erive primarily from our previous stuy of surgical site infection after spinal operations 10 an a thorough review of the literature. Data from the first fifty meical recors an a ranom subset of the remaining recors were collecte in uplicate, to ensure comparability of collecte ata between the two investigators. Extensive logic checks were performe to ientify illogical or impossible ata, with resolution of results by repeat review of the meical an computer recors. Data Analysis Associations between surgical site infection an potential risk factors were analyze with use of the chi-square test an the calculation of an os ratio an 95% confience intervals. Significant ifferences between continuous variables were etermine with the t test or the Mann-Whitney U test. A p value of <0.05 was consiere significant. Multivariate logistic regression analysis was use to ientify inepenent risk factors for surgical site infection. Variables eligible for inclusion in the multivariate moels inclue those reporte to be associate with an increase risk of surgical site infection in the literature, those with clinical an/or biologic plausibility, an those with p values of <0.20 in the univariate analyses. After ientification of the main effects in the logistic regression moels, all clinically meaningful two-way interaction factors were teste in the moels. The final moel(s) were checke for gooness of fit with the Hosmer an Lemeshow test an by colinearity an resiuals iagnostics, to ensure they were well specifie an fit the ata 12. Results The incience of surgical site infection following orthopaeic spinal operations performe from 1998 to 2002 was 2.0% (forty-six of 2316). Twenty (43%) of the forty-six infections were classifie as eep incisional (involving fascia an/or muscle); eight (17%), as organ space (involving an anatomic space opene uring the surgery other than the incision, an incluing osteomyelitis, empyema, an meningitis); an eighteen (39%), as superficial incisional (involving only skin or subcutaneous tissues). The meian time from the operation to the iagnosis of the infection was eleven ays, with a minimum of two ays an a maximum of 236 ays for a patient with osteomyelitis. All surgical site infections were treate with intravenous antibiotics in the hospital, an thirtysix (78%) of the forty-six patients ha a repeat operation to treat the infection. Seven of the ten patients who i not have a repeat operation were iagnose with a superficial surgical site infection an respone to intravenous antibiotic therapy. The patient-level factors that were foun to be associate with a significantly increase risk of surgical site infection in the univariate analysis are shown in Table II. They inclue iabetes, an elevate serum glucose level, a perioperative transfusion, postoperative incontinence (bowel or blaer, or both), an any incontinence (preoperative or postoperative). There was no ifference in the risk of surgical site infection between patients with insulin-epenent iabetes mellitus an iabetic patients treate with oral therapy (a surgical site infection evelope in two of four insulinepenent iabetic patients compare with twelve of twentytwo iabetic patients treate with oral therapy only). Only three iabetic patients were manage solely with iet, an a surgical site infection i not evelop in any of them. Obesity, efine as a boy mass inex of 30 to 35 kg/m 2, was associate with an increase risk of surgical site infection, although morbi obesity (a boy mass inex of >35 kg/m 2 ) ha only a marginal association with surgical site infection (p = 0.075). Diagnoses of herniate isc an nerve root compression were associate with a significantly lower risk of surgical site infection, whereas a iagnosis of vertebral fracture was associate with a higher risk of surgical site infection. More severe illness, as inicate by an American Society of Anesthesiologists score of 3 or 4, was associate with an increase risk of surgical site infection. Malnutrition, efine as a serum albumin level of

7 65 TABLE II Univariate Comparisons of Iniviual Risk Factors in Patients with an without Surgical Site Infection Following Orthopaeic Spinal Operations Patient-Level Characteristics No. (%) of Patients with Surgical Site Infection (N = 46) No. (%) of Uninfecte Patients (N = 227) Os Ratio (95% Confience Interval) P Value Boy mass inex Normal ( 25 kg/m 2 ) 9 (20) 84 (37) 1.0 Overweight (>25 an <30 kg/m 2 ) 11 (24) 77 (34) 1.3 (0.5, 3.4) Obese ( 30 an 35 kg/m 2 ) 18 (39) 37 (16) 4.5 (1.9, 11.0) Morbily obese (>35 kg/m 2 ) 8 (17) 29 (13) 2.6 (0.9, 7.3) Diabetes an hyperglycemia Neither (no iabetes or hyperglycemia) 20 (43) 179 (79) 1.0 Serum glucose >75th percentile 12 (26) 33 (15) 3.3 (1.5, 7.3) Diabetes 14 (30) 15 (7) 8.4 (3.5, 19.8) <0.001 Diagnoses Herniate isc 6 (13) 63 (28) 0.4 (0.2, 1.0) Fracture 6 (13) 9 (4) 3.6 (1.2, 10.8) American Society of Anesthesiologists class 3 or 4 22 (48) 59 (26) 2.6 (1.4, 5.0) Nerve root compression 22 (48) 153 (67) 0.4 (0.2, 0.8) Transfusion Cell Saver or autologous bloo 10 (22) 35 (15) 2.2 (0.9, 5.0) Packe re bloo cells or platelets 15 (33) 33 (14.5) 3.4 (1.6, 7.4) Postoperative incontinence 8 (17) 14 (6) 3.2 (1.3, 8.2) Preoperative or postoperative incontinence 13 (28) 27 (12) 2.9 (1.4, 6.2) <2.5 g/l (<25 g/l) in bloo collecte uring the most recent clinic visit within thirty ays before the operation or the surgical amission, was not associate with surgical site infection (p = 1.000). The univariate associations of selecte surgical-level factors an the risk of spinal surgical site infection are shown in Table III. Operations on the cervical spine, intravenous use of sterois intraoperatively, an use of cefazolin alone for infection prophylaxis were all associate with a significantly lower risk of surgical site infection. Performance of the operation through a posterior approach was associate with a significantly increase risk of surgical site infection. There was no association between surgical site infection an the use of bone graft (p = 0.479), the use of instrumentation (p = 0.901), or a previous operation at the same site (p = 0.775). Suboptimal timing of prophylactic antibiotics therapy, efine as the aministration of cefazolin more than sixty minutes before the incision or any antibiotic(s) first given after the incision, was associate with an increase risk of surgical site infection, as was a suboptimal ose of prophylactic cefazolin in obese patients (1 g of cefazolin in persons with a boy mass inex of >30 kg/m 2 ). Other operative variables associate with an increase risk of surgical site infection inclue aminoglycosie prophylaxis, irrigation of the surgical woun with an antibiotic solution (cefazolin or bacitracin), use of a rain for three or more ays after the operation, an two or more surgical resients participating in the operation. The meian uration of the operation was significantly longer (181 compare with 150 minutes, p = 0.009) an the meian estimate bloo loss was significantly higher (275 ml compare with 150 ml, p = 0.033) in the patients with a surgical site infection than in the control patients. As shown in Table III, an extensive operation involving seven or more intervertebral levels was associate with a higher risk of surgical site infection than was an operation involving only one intervertebral level. This fining is consistent with the association between the uration of the operation an the risk of surgical site infection since the meian number of intervertebral levels involve in operations with a uration of longer than the 75th percentile was four compare with two levels for operations lasting less than the 75th percentile. Participation in the operation by two or more surgical resients was also associate with a significantly longer uration of the operation an an operation involving a larger number of intervertebral levels (p < for both). Of all of the patient-level an operative characteristics, only a boy mass inex of 30 to 35 kg/m 2, iabetes, an transfusion of packe re bloo cells or platelets met the criterion for significance after correction for multiple testing (a = 0.001). The association between preoperative an postoperative serum glucose levels an surgical site infection was assesse. The results of serum glucose tests were not available for all patients at all time-points, so the number of subjects varie epening on the timing of the glucose testing. Patients with a surgical site infection ha significantly higher serum glucose

8 66 TABLE III Univariate Comparisons of Surgical Risk Factors in Patients with an without Surgical Site Infection Following Orthopaeic Spinal Operations Operative Characteristics No. (%) of Patients with Surgical Site Infection (N = 46) No. (%) of Uninfecte Patients (N = 227) Os Ratio (95% Confience Interval) P Value Cervical level 11 (24) 97 (43) 0.4 (0.2, 0.9) Posterior approach 42 (91) 172 (76) 3.4 (1.2, 9.8) Suboptimal timing of prophylactic antibiotic therapy* 15 (33) 31 (14) 3.1 (1.5, 6.3) Suboptimal osage of prophylactic antibiotic 20 (43.5) 51 (22.5) 2.7 (1.4, 5.1) Only cefazolin use as prophylactic antibiotic 24 (52) 168 (74) 0.4 (0.2, 0.7) Aminoglycosie use as prophylactic antibiotic 17 (37) 41 (18) 2.7 (1.3, 5.3) Intravenous sterois uring operation 8 (17) 85 (37) 0.4 (0.2, 0.8) Woun irrigate with antibiotic-containing solution 40 (87) 162 (71) 2.7 (1.1, 6.6) No. of intervertebral levels 1 10 (22) 83 (37) (46) 94 (41) 1.9 (0.8, 4.2) (15) 30 (13) 1.9 (0.7, 5.5) (17) 20 (9) 3.3 (1.2, 9.5) Duration of operation >75th percentile 18 (39) 49 (22) 2.4 (1.2, 4.6) Hemovac rain place 26 (56.5) 95 (42) 1.8 (1.0, 3.4) Drains in place 3 ays 33 (72) 108 (48) 2.8 (1.4, 5.6) resient surgeons 32 (70) 109 (48) 2.5 (1.3, 4.9) *Cefazolin given more than sixty minutes before the incision or after the incision, or another antibiotic given after the incision. Two hunre an twenty-nine (84%) of the 273 patients receive prophylactic cefazolin, alone (192), in combination with an aminoglycosie (thirty-four), or in combination with another antibiotic (three). One gram of cefazolin use as a prophylactic antibiotic in patients with a boy mass inex of 30.0 kg/m 2. The 75th percentile for fusion was minutes; the 75th percentile for non-fusion operations was 145 minutes. levels at the time of the most recent preoperative clinic testing (within thirty ays before the surgical amission) an significantly higher postoperative serum glucose levels (with use of the highest value within five ays after the operation for the analysis) (Table IV). The bloo collecte for glucose testing at the most recent preoperative clinic visit an the postoperative bloo were obtaine ranomly, since patients ha not been tol to fast. Since few patients (20%) ha serum glucose tests within twenty-four hours before the operation, the results of the ranom preoperative laboratory testing were combine with the results of fasting serum glucose tests performe on the ay before the operation to create a preoperative serum glucose variable (with the most recent result use if both ha been obtaine). The 75th percentile for this combine preoperative serum glucose level was 125 mg/l (6.9 mmol/l), an the 75th TABLE IV Association Between an Elevate Serum Glucose Level an the Risk of Surgical Site Infection Following Orthopaeic Spinal Operations Glucose Level (mg/l*) Patients with Surgical Site Infection Mean ± Stanar Deviation Meian (Range) Uninfecte Patients Mean ± Stanar Deviation Meian (Range) P Value At most recent preoperative visit (n = 189) 131 ± (73-267) 101 ± (56-300) <0.001 Within 24 hr before operation (n = 53) 154 ± ( ) 126 ± (72-311) Within 5 ays after operation (n = 146) 206 ± ( ) 169 ± (99-460) *The conventional unit (mg/l) is converte to the SI unit (mmol/l) by multiplying by Derive with the Mann-Whitney U test.

9 67 TABLE V Risk of Spinal Surgical Site Infection Accoring to Categorization of Serum Glucose Results Glucose Level* (mg/l [mmol/l]) No. of Patients with Surgical Site Infection/ No. Teste (%) No. of Uninfecte Patients/ No. Teste (%) Os Ratio (95% Confience Interval) P Value Preoperative >125 (>6.9) 20/39 (51) 30/182 (16) 5.3 (2.5, 11.2) <0.001 Postoperative >200 (>11.1) 14/35 (40) 21/111 (19) 2.9 (1.2, 6.5) Preoperative >125 or postoperative >200 25/45 (56) 45/215 (21) 4.7 (2.4, 9.3) <0.001 *These values represent the 75th percentiles, which were use as the cutoffs in the analysis. The result of testing twenty-four hours before the operation, if performe, or the result of the most recent preoperative laboratory test. TABLE VI Multivariate Logistic Regression Moel for the Development of Spinal Surgical Site Infection* Risk Factor Ajuste Os Ratios (95% Confience Interval) P Value Diabetes 3.5 (1.2, 10.0) Suboptimal timing of prophylactic antibiotic therapy 3.4 (1.5, 7.9) Elevate serum glucose level (>125 mg/l [>6.9 mmol/l]) 3.3 (1.4, 7.5) preoperatively or >200 mg/l [>11.1 mmol/l]) postoperatively Obesity (boy mass inex >30.0 kg/m 2 ) 2.2 (1.1, 4.7) resient surgeons 2.2 (1.0, 4.7) Operation involving cervical levels 0.3 (0.1, 0.6) *The c-statistic for the moel = The Hosmer an Lemeshow gooness-of-fit chi-square p = (7 egrees of freeom), an the Nagelkerke R 2 = percentile for the ranom postoperative serum glucose level was 200 mg/l (11.1 mmol/l). When the serum glucose levels were categorize accoring to these cutoffs, a preoperative level of >125 mg/l was associate with a 5.3-fol increase risk of surgical site infection, an any postoperative glucose level (within five ays after the operation) of >200 mg/l was associate with a 2.9-fol increase risk of surgical site infection (Table V). Either a preoperative or any postoperative serum glucose level of >75th percentile was associate with a 4.7-fol increase risk of surgical site infection. We also analyze the association of the preoperative glucose level with surgical site infection after taking into account receipt of total parenteral nutrition before the surgery. Twenty-seven patients receive total parenteral nutrition uring their hospital stay, although only six of the twentyseven patients receive total parenteral nutrition before the eligible operation. Using a cutoff of 200 mg/l for the preoperative serum glucose level in patients receiving total parenteral nutrition instea of 125 mg/l ecrease the effect size for the association of the preoperative glucose level an the risk of surgical site infection only slightly (os ratio = 4.7, p < 0.001, compare with os ratio = 5.3; Table V). The results of the multivariate analysis to ientify inepenent risk factors for spinal surgical site infection are shown in Table VI. Diabetes ha the strongest association with surgical site infection, with an ajuste os ratio of 3.5 after we controlle for the other variables in the moel. Other variables that remaine inepenently associate with an increase risk of surgical site infection inclue suboptimal timing of prophylactic antibiotic therapy (os ratio = 3.4), an elevate serum glucose level (a preoperative ranom or fasting serum glucose level of >125 mg/l or a postoperative ranom serum glucose level of >200 mg/l) (os ratio = 3.3), obesity (os ratio = 2.2), an participation in the operation by two or more surgical resients (os ratio = 2.2). An operation involving cervical levels was associate with a significantly lower risk of spinal surgical site infection (os ratio = 0.3). The moel ha goo preictive ability, with a c-statistic of Discussion This stuy extens the work that we i previously to etermine inepenent risk factors for surgical site infection after spinal operations 10. We performe this secon stuy to etermine whether there were unique risk factors in our patients unergoing orthopaeic spinal surgery as compare with patients unergoing spinal neurosurgery base on unerlying ifferences in the patient populations. An aitional rationale for unertaking a secon stuy was to collect more etaile ata regaring some potential risk factors, such as hyperglycemia, rain utilization, an local an systemic steroi use, than ha been collecte in our initial stuy. In the present stuy, iabetes, suboptimal timing of prophylactic antibiotic

10 68 therapy, elevate serum glucose levels, obesity, an participation in the operation by two or more surgical resients were foun to be inepenently associate with infection involving the spinal incision after laminectomy, iscectomy, an/or spinal arthroesis. Surgery at a cervical level was inepenently associate with a significantly ecrease risk of surgical site infection following orthopaeic spinal operations. Contrary to some reports in the literature, we foun no association between arthroesis an an increase risk of surgical site infection. Our hospital is a regional referral center for spine operations, an it is likely that more complex laminectomies an iscectomies are performe at our institution than at community or smaller tertiary-care hospitals. This is consistent with our surgical site infection rates as compare with the CDC/NNIS rates; the surgical site infection rate after laminectomies an iscectomies performe by orthopaeic surgeons at our institution from 1998 to 2002 was 2.3%, higher than the mean CDC/NNIS rate of 1.5% 2. The surgical site infection rate at our institution after orthopaeic spinal arthroeses was 1.9%, which is slightly lower than the mean CDC/NNIS rate of 2.1% 2. Diabetes an Increase Risk of Surgical Site Infection A iagnosis of iabetes was associate with the greatest inepenent risk of spinal surgical site infection, an elevate serum glucose levels remaine significantly associate with surgical site infection after we controlle for iabetes an other variables. To our knowlege, this is the first stuy to emonstrate the inepenent risk of surgical site infection after spine operations associate with hyperglycemia. Numerous authors have foun that the increase risk of eep sternal surgical site infection following cariac operations can be ameliorate in iabetic patients by careful perioperative monitoring an control of serum glucose levels We are aware of no stuies on the effects of such a strategy for patients unergoing spinal surgery, although the univariate association of iabetes with spinal surgical site infection has been reporte in a number of stuies 8, We previously foun an association between perioperative hyperglycemia an surgical site infection following spinal neurosurgery 10. In our previous stuy, this association i not remain significant in the multivariate analysis, possibly because of a more strict efinition of hyperglycemia (a glucose level of >200 mg/l uring the surgical amission) compare with that use in the current stuy. Confirmation of these finings might lea to stuies of intensive perioperative glucose control in patients with iabetes unergoing orthopaeic operations. Aitional stuy is neee to confirm the risk of surgical site infection ue to hyperglycemia in patients not previously iagnose with iabetes. Other Inepenent Preictors of Surgical Site Infection Suboptimal timing of prophylactic antibiotic therapy was associate with a 3.4-fol increase risk of surgical site infection in the multivariate moel, a fining very similar to the results reporte by Classen et al. 20. Because of the relatively small size of this case-control stuy, we inclue receipt of a cephalosporin more than one hour before the incision or any prophylactic antibiotic given after the incision in the suboptimal category. The current recommenation for antibiotic prophylaxis for neurosurgical an orthopaeic proceures is for 1 to 2 g of cefazolin to be given in the hour before the incision in nonallergic patients 21 ; thus, aministration of an antibiotic outsie of this perio woul be consiere suboptimal. The fining of an increase risk of surgical site infection associate with prophylactic antibiotic aministration outsie of the onehour winow before the incision supports the recommenations of the Surgical Care Improvement Project (SCIP) to establish quality-improvement measures to ensure timely aministration of prophylactic antibiotics 22. Obesity was associate with an increase risk of spinal surgical site infection in this stuy, a fining similar to those after other operations 11,23. The SCIP avisory panel recommens a 2-g ose of cefazolin for prophylaxis in patients who weigh 80 kg 24. We coul not accurately etermine the effect of increase cefazolin osage on the risk of surgical site infection in obese persons because of the small number of obese persons who were given a 2-g ose. However, given the minimal sie effects of cefazolin in nonallergic patients, it appears reasonable to give 2 g to all patients weighing 80 kg to ecrease the risk of surgical site infection associate with obesity. Participation in the operative proceure by two or more surgical resients was also associate with an increase risk of surgical site infection. We assume that this variable was a proxy for the uration of the operation an the complexity of the operative proceure. We coul not etermine from the chart review if the resients were present at the same time or move in an out of the operating room. An operation at the cervical level was inepenently associate with a ecrease risk of surgical site infection. Zeiman et al. previously reporte a low rate of surgical site infection following cervical spinal operations 25. In our previous stuy of spinal operations by neurosurgeons, we ientifie the posterior surgical approach as a risk factor for surgical site infection 10. The posterior approach may not have remaine inepenently associate with an increase risk of surgical site infection in the present stuy because of the inclusion of cervical operations in the moel an the relatively small number of anterior operations. Limitations an Strengths of This Stuy The observational nature of this stuy preclue complete analysis of some potentially important risk factors for surgical site infection, such as malnutrition. The serum albumin level was measure at the iscretion of the operating surgeon an was therefore not available for all patients. The analyses of serum glucose levels were also hampere by incomplete testing. It was thus not possible to etermine if the risk of surgical site infection associate with persistently high serum glucose levels was higher than that associate with only a single high value. Sequential serum glucose testing nees to be performe before an after spinal operations in orer to more accurately etermine the association between hyperglycemia an surgical site infection.

11 69 The strengths of this stuy inclue the wie variety of potential risk factors that were analyze an the relatively large number of patients with spinal surgical site infection compare with the numbers in most stuies of this complication. We use multivariate logistic regression analysis to etermine inepenent risk factors for surgical site infection, which is particularly important when etermining the magnitue of risk associate with factors, such as iabetes an obesity, that ten to cluster within iniviual patients 26. We extracte ata for many patient an surgical risk factors, incluing the type an uration of use of surgical rains, irrigant solutions, prophylactic antibiotics, an serum glucose levels, in orer to perform etaile analyses of these variables. To our knowlege, this is the most in-epth an comprehensive analysis of risk factors for surgical site infection following spinal operations that has been publishe. Aitional stuies are warrante to etermine whether careful monitoring an control of serum glucose levels in the perioperative perio are associate with a ecrease risk of infection following spinal operations. While the risk of surgical site infection can never be reuce to zero, establishing quality-improvement programs to monitor an ensure compliance with recommenations regaring antibiotic prophylaxis an maintenance of normoglycemia may prevent a subset of these infections an improve outcomes for patients unergoing spinal operations. n Margaret A. Olsen, PhD, MPH Victoria J. Fraser, MD Division of Infectious Diseases, Campus Box 8051, Washington University School of Meicine, 660 South Eucli Avenue, St. Louis, MO aress for M.A. Olsen: molsen@im.wustl.eu Jeffrey J. Nepple, MD K. Daniel Riew, MD Lawrence G. Lenke, MD Keith H. Briwell, MD Department of Orthopaeic Surgery, Washington University School of Meicine, One Barnes-Jewish Hospital Plaza, Suite West Pavilion, St. Louis, MO Jennie Mayfiel, BSN, MPH, CIC Department of Hospital Epiemiology an Infection Prevention, Barnes- Jewish Hospital, Mailstop , 4905 Forest Park Avenue, St. Louis, MO References 1. Horan TC, Culver DH, Gaynes RP, Jarvis WR, Ewars JR, Rei CR. Nosocomial infections in surgical patients in the Unite States, January 1986-June National Nosocomial Infections Surveillance (NNIS) System. Infect Control Hosp Epiemiol. 1993;14: National Nosocomial Infections Surveillance System. National Nosocomial Infections Surveillance (NNIS) System Report, ata summary from January 1992 through June 2004, issue October Am J Infect Control. 2004;32: Abbey DM, Turner DM, Warson JS, Wirt TC, Scalley RD. Treatment of postoperative woun infections following spinal fusion with instrumentation. J Spinal Disor. 1995;8: Hoges SD, Humphreys SC, Eck JC, Covington LA, Kurzynske NG. Low postoperative infection rates with instrumente lumbar fusion. South Me J. 1998;91: Picaa R, Winter RB, Lonstein JE, Denis F, Pinto MR, Smith MD, Perra JH. Postoperative eep woun infection in aults after posterior lumbosacral spine fusion with instrumentation: incience an management. J Spinal Disor. 2000;13: Tenney JH, Vlahov D, Salcman M, Ducker TB. Wie variation in risk of woun infection following clean neurosurgery. Implications for perioperative antibiotic prophylaxis. J Neurosurg. 1985;62: Weinstein MA, McCabe JP, Cammisa FP Jr. Postoperative spinal woun infection: a review of 2,391 consecutive inex proceures. J Spinal Disor. 2000;13: Wimmer C, Gluch H, Franzreb M, Ogon M. Preisposing factors for infection in spine surgery: a survey of 850 spinal proceures. J Spinal Disor. 1998;11: Peuzzi P, Concato J, Kemper E, Holfor TR, Feinstein AR. A simulation stuy of the number of events per variable in logistic regression analysis. J Clin Epiemiol. 1996;49: Olsen MA, Mayfiel J, Lauryssen C, Polish LB, Jones M, Vest J, Fraser VJ. Risk factors for surgical site infection in spinal surgery. J Neurosurg. 2003;98(2 Suppl): Mangram AJ, Horan TC, Pearson ML, Silver LC, Jarvis WR. Guieline for the prevention of surgical site infection, Hospital Infection Control Practices Avisory Committee. Infect Control Hosp Epiemiol. 1999;20: Hosmer DW, Lemeshow S. Applie logistic regression. 2n e. New York: John Wiley an Sons; Furnary AP, Zerr KJ, Grunkemeier GL, Starr A. Continuous intravenous insulin infusion reuces the incience of eep sternal woun infection in iabetic patients after cariac surgical proceures. Ann Thorac Surg. 1999;67: Carr JM, Sellke FW, Fey M, Doyle MJ, Krempkin JA, e la Torre R, Liicoat JR. Implementing tight glucose control after coronary artery bypass surgery. Ann Thorac Surg. 2005;80: Hruska LA, Smith JM, Heny MP, Fritz VL, McAams S. Continuous insulin infusion reuces infectious complications in iabetics following coronary surgery. J Car Surg. 2005;20: Simpson JM, Silveri CP, Balerston RA, Simeone FA, An HS. The results of operations on the lumbar spine in patients who have iabetes mellitus. J Bone Joint Surg Am. 1993;75: Kuo CH, Wang ST, Yu WK, Chang MC, Liu CL, Chen TH. Postoperative spinal eep woun infection: a six-year review of 3230 selective proceures. J Chin Me Assoc. 2004;67: Fang A, Hu SS, Enres N, Brafor DS. Risk factors for infection after spinal surgery. Spine. 2005;30: Kanafani ZA, Dakouki GK, El-Dbouni O, Bawwab T, Kanj SS. Surgical site infections following spinal surgery at a tertiary care center in Lebanon: incience, microbiology, an risk factors. Scan J Infect Dis. 2006;38: Classen DC, Evans RS, Pestotnik SL, Horn SD, Menlove RL, Burke JP. The timing of prophylactic aministration of antibiotics an the risk of surgical-woun infection. New Engl J Me. 1992;326: Antimicrobial prophylaxis for surgery. Treat Guiel Me Lett. 2006;4: Bratzler DW, Hunt DR. The surgical infection prevention an surgical care improvement projects: national initiatives to improve outcomes for patients having surgery. Clin Infect Dis. 2006;43: Thomas EJ, Golman L, Mangione CM, Marcantonio ER, Cook EF, Luwig L, Sugarbaker D, Poss R, Donalson M, Lee TH. Boy mass inex as a correlate of postoperative complications an resource utilization. Am J Me. 1997;102: Bratzler DW, Houck PM; Surgical Infection Prevention Guieline Writers Workgroup. Antimicrobial prophylaxis for surgery: an avisory statement from the National Surgical Infection Prevention Project. Am J Surg. 2005;189: Zeiman SM, Ducker TB, Raycroft J. Trens an complications in cervical spine surgery: J Spinal Disor. 1997;10: Hennekens CH, Buring JE. Epiemiology in meicine. Mayrent SL, eitor. Boston: Little, Brown; 1987.