Predictors of Postoperative Infection in Spinal Deformity Surgery

         Despite efforts to improve infection rates, surgical site infection remains a burdensome complication of spinal surgery. Postoperative spinal wound infections lead to increased morbidity and mortality, increased hospital stay or rehospitalization, and additional cost.1 Reported rates of surgical site infection in the literature range from less than 1% to 15%, varying with procedure, patient population, and the use of instrumentation.2-5 Numerous risk factors for surgical site infection have been reported in the literature. These include patient factors such as a diagnosis of neuromuscular disease, over 60 years of age, smoking, diabetes, previous surgical infection, obesity, Predictors of Postoperative Infection in Spinal Deformity Surgery Which Curves Are at Greatest Risk? Kushagra Verma, M.D., Baron Lonner, M.D., Laura Dean, M.D., David Vecchione, M.D., and Kathryn Kean, M.D. Proofs to: lonnerresearch@gmail.com Author: Please confirm academic degrees of all authors. None were on manuscript, “M.D.,” has been assumed Kushagra Verma, M.D., Baron Lonner, M.D., Laura Dean, M.D., David Vecchione, M.D., and Kathryn Kean, M.D., are in the Department of Orthopaedic Surgery, NYU Hospital for Joint Diseases, New York, New York.

         Correspondence: Baron Lonner, M.D., Department of Orthopaedic Surgery, NYU Hospital for Joint Diseases, 301 East 17th Street, New York, New York 10003; lonnerresearch@gmail.com. 258 Bulletin of the Hospital for Joint Diseases 2013;71(4):00-00 alcohol abuse, long-term steroid use, American Society of Anesthesiologists (ASA) risk class of 3 or higher, and previous spinal surgery.2-4,6 Operative risk factors reported include lumbo-sacral operative level, dural tear during the procedure and the use of glue to repair dural tear, postoperative incontinence, posterior approach, procedure for tumor resection, suboptimal timing of prophylactic antibiotic therapy, a preoperative serum glucose level above125mg/ dL or postoperative level above 200mg/dL, 2 or more surgical residents participating in the procedure, extended preoperative hospitalization, excessive blood loss, receiving a blood transfusion, failure to use a postoperative drain, and prolonged duration of procedure.1,5,7,8 Many of the studies examining spinal surgical site infections have, however, been limited by small numbers, confounded by multiple surgeons and hospitals, or were not specific to spinal deformity.

         The purpose of this study was to identify risk factors associated with surgical site infection from a large, prospectively-collected single-surgeon database of spinal deformity patients. Of particular interest, this study aimed to delineate specific spinal deformity parameters and surgical procedures that may be prone to a higher infection risk. Methods Study Design This is a retrospective review of 806 patients (584 female, 222 male) who underwent 941 total surgical procedures, including revisions (915 single-stage procedures, 13 two-stage procedures) for spinal deformity by a single surgeon from 1999 to 2009. Surgical procedures for all spinal deformities were considered, and both pediatric and adult patients were included. Patients were identified from a prospectivelycollected database of the senior investigator’s surgical cases. Data Collection Data was collected using the database, patient charts, and radiographs. Demographic parameters examined were age, gender, height, weight, and body mass index (BMI).

         Comorbidities were assessed and categorized by system as cardiac, pulmonary, psychiatric, renal or genitourinary, gastrointestinal, endocrine, hematologic, rheumatologic or musculoskeletal, medication allergy, neurological, and tobacco use. The total number of comorbidities was also considered. Table 1 Parameters Entered into Pearson’s Correlation Patient Factors Age Height Gender Body Mass Index (BMI) Comorbidities Cardiac Medication Allergy Pulmonary Neurological Psychiatric Smoking or Drinking History Renal or Genitourinary Rheumatologic or Musculoskeletal Gastrointestinal or Endocrine Total Number of Comorbidities Hematological Diagnoses Scoliosis Syndromic Scoliosis Kyphosis Spondylolisthesis Neuromuscular Scoliosis Hemivertebrae Congenital Scoliosis Curve Parameters Lenke 1 or 2 curves Major Curve Magnitude (scoliosis only) Lenke 3 or 4 curves Kyphosis magnitude (T2-12) Lenke 5 or 6 curves Surgical Procedures Spinal Fusion Revision Surgery Removal of Hardware Spinal Decompression Growth Rod Distraction Vertebral Column or Hemivertebrae Resection Osteotomy Total Number of Procedures Surgery Related Anterior Approach Estimated Blood Loss (EBL) Posterior Approach All Hook Instrumentation Combined Approach All Screw Instrumentation Number of levels Fused Hybrid Instrumentation (> 20% hooks) Total Operative Time Length of Hospital Stay Bulletin of the Hospital for Joint Diseases 2013;71(4):00-00 259 Primary diagnoses were categorized as idiopathic scoliosis, kyphosis, neuromuscular scoliosis, congenital scoliosis, syndromic scoliosis, spondylolisthesis, and revision.

         Patients with AIS were further grouped by Lenke curve types 1 through 6. Surgical approach was categorized as anterior, posterior, or combined approach. Surgical procedures performed included fusion, removal of hardware, osteotomy, revision, decompression, vertebral column resection or hemivertebrae excision, and growth rod procedure (placement or distraction). Total number of procedures was also considered. Other operative parameters analyzed included major curve magnitude (for all types of scoliosis), maximum kyphosis (for kyphosis), number of levels fused, operative time, estimated blood loss (EBL), and length of hospital stay. Instrumentation types were categorized as hooks, screws, or hybrid. A screw construct was defined as a construct ≤ 2 hooks, with greater than 80% pedicle screw anchor density. Hook constructs were defined as constructs with all hook anchors. All other combinations of hooks and screws were considered to be hybrid constructs.

         A full list of variables examined is found in Table 1. Identification and Treatment of Infections Charts were reviewed to determine which patients had deep, superficial, possible wound infections, and no infections. Due to the retrospective nature of the data, it was difficult to apply CDC guidelines for superficial and deep wound infection diagnoses in all cases. The definition of a superficial wound infection was drainage with or without erythema associated with persistent fever, abnormal erythrocyte sedimentation rate (ESR), c-reactive protein (CRP), or white blood count (WBC), or positive superficial wound culture. Patients were considered to have a possible wound infection if they were restarted on antibiotics for persistent wound drainage, but did not have fever, positive wound culture, and abnormal serological study patterns (ESR, CRP, Table 2 Patient Demographics Mean Range Age (years) 22.3 ± 15 1.8 – 82.0 Gender 584 F; 222 M BMI 22.2 ± 5.2 12.1 - 42.0 Sample Size (N = 941) Operative Time (min) 256.9 ± 121.1 20-840 779 Primary EBL (cc) 888 ± 984 1-12,000 162 Revision Length of Stay (days) 6 ± 5 0-54 Major Coronal Curve Magnitude (degrees) 54 ± 15 30-132 * Surgical Approach T2-T12 Kyphosis Magnitude (degrees) 66.2 ± 16 31-112 * 194 Anterior Number of Levels Fused 8.6 ± 4.3 1-23 606 Posterior 141 Combined Number of Comorbidities 0.7 ± 1.1 0-6 Number of Procedures Performed 1.5 ± 0.8 0-5 Most Common Diagnosis Idiopathic Scoliosis (606) Revision (160) Kyphosis (58) Congenital Scoliosis (38) Most Common Comorbidity Psychiatric (87) Pulmonary (86) GI/Endo ( 85) Cardio (76) Most Common Procedure Fusion (873) Osteotomy (162) Revision (162) Removal of Hardware (120) Total Infections (N = 30) MRSA (5) MSSA (3) Other/Unidentified (22) Superficial Infections (N = 17) MSSA (3) MRSA (1) Pseudomonas (1) Other/Unidentified (12) Deep Infections (N = 13) MRSA (4) Pseudomonas (1) Other/Unidentified (8) Deep Infection Instrumentation Stainless steel (10) Titanium (1) Unidentified (2) Mean Range Time of Infection Diagnosis (weeks) Total Infections (N = 30) 10.87 1-84 Superficial Infections (N = 17) 3.86 1-4 Deep Infections (N = 13) 19.62 1-84 * Curve magnitudes for idiopathic, adult, neuromuscular, and congenital scoliosis patients 260 Bulletin of the Hospital for Joint Diseases 2013;71(4):00-00 or WBC).

        Patients with a possible or definitive superficial wound infection were treated with oral antibiotics for 1 to 2 weeks. Superficial infections typically did not require an incision and drainage to eradicate the infection. Cultures were not routinely obtained prior to antibiotic administration or prior to surgery for deep infections. All patients with signs of infection including purulent drainage, erythema, swelling with or without laboratory values suggestive of infection, were taken to the operating room. The fascial layer was opened and the deep space explored, and all patients were considered to have an infection involving both the deep and superficial spaces whether or not cultures were positive. Wound cultures were obtained, debridement of necrotic tissue was performed, implants were routinely removed, and pulsatile irrigation with 9 liters of saline solution containing bacitracin was performed. The deep fascia was then closed over suction drains. If the surgeon deemed it necessary, a patient with a deep wound infection may have underwent multiple operative incision and drainage procedures, but that was not routine.

         This method for management of deep wound infections is similar to that described previously.3 Blood cultures were often performed. For cases of deep wound infections, the implant material, organism, and treatment were examined for this study. Organisms were categorized as methicillin-sensitive Staphylococcus aureus (MSSA), methicillin-resistant Staphylococcus aureus (MRSA), Pseudomonas, or other. Blood cultures and operative wound cultures were obtained by an operative incision and drainage and a minimum of 6 weeks of intravenous antibiotics (continuing after hospital discharge) was administered. Statistics Statistical analysis included Pearson’s test to determine significant correlations and multivariate logistic regression to determine independent predictors. Variables entered into the Pearson’s correlation are listed in Table 1. These variables were correlated to three possible outcomes: any infection, superficial wound infection, and deep wound infection.

         Significantly correlated variables for each outcome were then used in a logistic regression to identify statistically significant independent predictors. Results There were 13 deep wound infections (1.4%) and 17 superficial wound infections (1.8%), for an overall infection rate of 3.2%. There were 32 possible infections (3.4%) in which antibiotics were empirically administered. The average age was 22.3 years (range 1.8 to 82). Diagnoses included idiopathic scoliosis (N = 606), neuromuscular scoliosis (N = 20), congenital scoliosis (N = 38), syndromic scoliosis (N = 14) kyphosis (N = 58), spondylolisthesis (N = 22), hemivertebrae (N = 4), and revision (N = 160). There were 194 anterior, 606 posterior, and 141 combined anterior-posterior cases. Procedures performed included spinal fusion (N = 873), removal of hardware (N = 120), growth rod procedure (N = 23), osteotomy (N = 162), decompression (N = 3), and vertebral column resection/hemivertebra excision (N = 32). There were 162 operations that included revision procedures.

         Any given patient in the study was not limited to a single diagnosis, comorbidity, or procedure performed. Demographic parameters are found in Table 2. Any Infection Pearson’s test found the following significant correlations for any infection: increasing age, increasing BMI, increasing total number of levels fused, Lenke 3 and Lenke 4 curve types, osteotomy procedure, hematological comorbidity, rheumatologic or musculoskeletal comorbidity, increasing total number of comorbidities, increasing EBL, and an all screw construct (Table 3). Of these, logistic regression showed the following to be statistically significant predictors of any infection: age, BMI, total number of comorbidities, osteotomy procedure, number of levels fused, and Lenke 3/4 curve types (Table 4). However, Age, BMI, and increasing total number of levels fused had small odds radios, 0.99, 1.11, and 1.13 respectively, that would not intuitively raise or lower the infection risk. Superficial Infection For superficial infection, the following were found to be significantly correlated: increasing age, increasing BMI, diagnosis of kyphosis, Lenke 3/4 curve types, Lenke 5/6 curve types, osteotomy procedure, anterior approach, combined approach, increasing total number of levels fused, and increasing EBL (Table 3).

         Diagnosis of kyphosis, combined approach, total number of levels fused, and Lenke 3/4 curve types were statistically significant predictors (Table 4). However, combined approach and levels fused had odds ratios that were relatively close to “one” (0.81 and 1.16, respectively) and did not strongly raise or lower the SSI risk. Of the 17 patients with a superficial infection, the pathogens identified included MSSA (N = 3), MRSA (N = 1), Pseudomonas (N = 1), other or unidentified (N = 12) (Table 2). Time to diagnosis of infection averaged 3.86 weeks (range 1 to 4) in this group (Table 2). Deep Infection The following variables were found to be significantly correlated with deep infection: increasing age, increasing BMI, psychiatric comorbidity, gastrointestinal or endocrine comorbidity, neurological comorbidity, rheumatologic or musculoskeletal comorbidity, total number of comorbidities, neuromuscular scoliosis, Lenke 3/4 curve types, a removal of hardware procedure, increasing EBL, and an all screw construct (Table 3). Of these, only neuromuscular scoliosis and Lenke 3/4 curve types were found to be strong independent predictors of deep infection (odds ratios of 9.2 and 7.4, respectively, Table 4).

         Of the 13 patients with a deep wound infection, the pathogens identified were MRSA (N Bulletin of the Hospital for Joint Diseases 2013;71(4):00-00 261 = 4), Pseudomonas (N = 1), other or unidentified (N = 12) (Table 2). Time to diagnosis of infection in this group averaged 19.62 weeks (range 1 to 84) (Table 2). Instrumentation in most cases was stainless steel (N = 10), followed by titanium (N = 1), and other or unidentified (N = 2) (Table 1). Five patients required removal of all hardware in addition to an operative incision and drainage. One patient required two consecutive operative incision and drainage procedures (MRSA, stainless steel), while another patient required only a partial removal of hardware in addition to an incision and drainage. In all 13 cases, clinical success was achieved in eradicating the infection. Given the relatively few numbers of deep wound infections and incomplete data in this sample, statistical comparisons could not be made based on pathogen or instrumentation type. Discussion Surgical site infection (SSI) is a troubling complication of spinal surgery.

         Policy makers have placed great emphasis on the mitigation of this problem through standardization of hospital protocol and recommendations for prevention. Surgical infection rates have also been used to rate hospitals and by insurance companies to deny reimbursements related to hospital acquired infections.2 In one study, wound infection following spinal surgery was found to increase health care costs by 4-fold.6,9,10 In an effort to manage patient morbidity and increasing health care costs, increased attention has been placed on the causative factors associated with SSI. Research in this arena may help to identify patients at greatest risk, improve patient morbidity, and mitigate healthcare costs and liability associated with this complication. The rates of surgical site infection reported in the literature have ranged from less than 1% to 15%, varying with patient population, diagnosis, and procedure.2,4,5,11 In these studies and others, certain patients have been found to be at increased risk of developing a surgical site infection. Blam and coworkers found a 9.4% SSI rate in patients undergoing spinal trauma surgery compared to 3.7% in patients undergoing elective spine surgery.12

         Nutritional depletion may also lead to an increased risk for postoperative complications including surgical wound infections. In the most recent review, 2 out of 4 studies found malnutrition to significantly increase SSI risk.2 In one study by Klein and colleagues malnourished patients having a diminished total lymphocyte count (TLC) and serum albumin level preoperatively showed a 15-fold increase in wound infection risk compared to a nutritionally replete group.13 In patients with scoliosis secondary to cerebral palsy and myelomeningocele, Kretzler and associates reported an SSI rate of 11%, as compared with a rate of 1% for idiopathic scoliosis.9,10,14 Similarly, Sponseller and coworkers showed a 10% infection rate in cerebral palsy patients.15,16 Regression analysis identified Table 3 Significant Correlations (Pearson’s test) Any Infection Deep Superficial p-value p-value p-value Patient Factors Age < 0.005 < 0.005 < 0.05 BMI < 0.0001 < 0.005 < 0.05 Comorbidity Psychiatric < 0.01 Hematological < 0.001 Gastrointestinal or Endocrine < 0.01 Neurological < 0.005 Rheumatologic or Musculoskeletal < 0.05 < 0.05 Total Number of Comorbidities < 0.001 < 0.001 Diagnosis Kyphosis < 0.05 Neuromuscular Scoliosis < 0.001 Curve Parameters Lenke 3 or 4 curves < 0.05 < 0.05 < 0.0001 Lenke 5 or 6 curves < 0.05 Procedure Removal of Hardware < 0.05 Osteotomy < 0.001 < 0.01 Total Number of Procedures < 0.05 Surgery Related Anterior Approach < 0.05 Combined Approach < 0.005 Number of levels fused < 0.01 < 0.01 EBL < 0.001 < 0.05 < 0.05 All Screw Construct <.005 < 0.05 262 Bulletin of the Hospital for Joint Diseases 2013;71(4):00-00 the degree of cognitive impairment, elevated preoperative platelet and white blood cell counts, and the presence of metallic implants to be associated with an increased infection risk.

         Of note, nutritional status preoperatively was not predictive of SSI in these cerebral palsy patients. Fang and colleagues identified 61 cases of SSI among 1629 surgeries (3.7%) and found age greater than 60 years, smoking, neuromuscular scoliosis, alcohol abuse, diabetes, elevated BMI, and previous SSI to all increase SSI risk.3 Olsen and coworkers reported a 2.0% SSI rate from another casecontrolled study and found obesity, diabetes, and elevated serum glucose to correlated with SSI.4 Pull ter Gunne and associates also reported obesity and diabetes to be significant risk factors for superficial wound infection.17,18 In a recent review of the infection literature in adults, over 60 years of age, presence of diabetes, malnutrition, obesity, ASA score greater than 2, and elevated blood glucose were associated with SSI. However, gender, steroid therapy, smoking history, and revision surgery were not consistently reported risk factors for SSI, although they have been implicated by some investigators.2 Our study is in agreement with previous reports noting an increased infection rate in neuromuscular scoliosis patients3,15,16 We report a 9-fold increase in deep wound infection risk in these patients.

         The risk of surgical site infection depends on intraoperative factors. Simple discectomy confers the lowest risk, followed by decompression without fusion, and then fusion. Weinstein and coworkers reported the infection rate to be 0.86%, 1.5%, and 2.5%, respectively for these procedures.18 The rate of infection for cervical procedures was also low at 1.2%.8 Pull ter Gunne and associates reported a large increase in deep wound infection rate with vertebral column resection (VCR) (11.1%) versus other types of osteotomies (2.5%). The investigators noted that the SSI risk was also related to increased operative time, blood loss, and transfusion requirements for the VCR patients.17 Our study reports a 3-fold increased risk of SSI with any osteotomy procedure versus no osteotomy, which was not reported by the aforementioned study. Of note, neither VCR nor hemivertebrae excision was an independent predictor of SSI in our study. This may be related to the relatively small numbers of patients undergoing this procedure (N = 32). In an earlier study by the same investigator, deep wound infection risk increased with operative time (odds ratio 2.4 to 2.8).18 Fang and coworkers also noted staged procedures and increased operative times to be significant risk factors for SSI.3

         While operative time has been reported to increase SSI risk in previous reports, it has not been consistently reported in most recent studies2,19,20Schwarzkopf and associates have identified blood product usage to increase SSI risk by 6-fold.21,22 This finding, however, was only reported in only 2 out 7 studies in a recent review article.2 Olsen and colleagues identified suboptimal timing of prophylactic antibiotics, and greater than 2 resident physicians to be a risk factor for SSI.4 The use of allograft may or may not be associated with increased SSI rate and has been inconsistently reported.2,20,23 Recently, Table 4 Independent Predictors (Logistic Regression) Any Infection Deep Superficial Odds ratio P value Odds ratio P value Odds ratio P value BMI 1.11 < 0.005 Total Number of Comorbidities 1.24 < 0.05 Diagnosis Kyphosis 4.38 < 0.005 Diagnosis Neuromuscular Scoliosis 9.2 Osteotomy Procedure 3.14 < 0.001 Combined Approach 0.81 < 0.05 Number of levels fused 1.13 0.002 1.16 < 0.005 Lenke 3 or 4 curves 1.67 < 0.05 7.40 < 0.001 4.54 < 0.0001 Table 5 Infection Rates by Group N Maybe Infection Deep Superficial All Idiopathic Scoliosis 606 19 (3.1%) 6 (0.9%) 8 (1.3%) All Kyphosis 58 3 (5.1%) 0 3 (5.1%) Adult Idiopathic Scoliosis 82 4 (4.8%) 2 (2.4%) 3 (3.6%) Adult Kyphosis 26 2 (7.7%) 0 1 (3.8%) All adults 236 10 (4.2%) 8 (3.4%) 7 (2.9%) All Adolescents 671 22 (3.3%) 0.7 (%) 10 (1.5%) Primary Surgery 781 27 (3.4%) 9 (1.1%) 13 (1.6%) Revision surgery 160 5 (3.1%) 4 (2.5%) 4 (2.5%) Bulletin of the Hospital for Joint Diseases 2013;71(4):00-00 263 however, Borkhuu and associates reported a drastic decrease in deep wound infection rate in cerebral palsy patients by using antibiotic-loaded allograft versus allograft alone (15% to 4%).5

         In this study, the number of levels fused, operative time, estimated blood loss, and surgical approach all were correlated with infection risk but were not strong independent predictors of SSI. The use of instrumentation may increase the infection risk although this has not been uniformly reported.2,19 Weinstein and coworkers found that fusion without instrumentation was associated with a 0.4% rate, while the rate for fusion with instrumentation was 3.2%.7 Similarly, Abbey and associates reported an instrumented spinal fusion infection rate of 3.7%.24 While the increased infection risk has been well documented for instrumented fusions, no prior study has compared “all hook,” versus “hybrid,” versus “all pedicle screw” constructs in terms of SSI risk. Our study did not find any one of these options to influence SSI risk significantly. The use of a closed wound suction drain has not been shown to decrease infection risk in two randomized trials.25,26 Cheng and colleagues reported a 0% infection rate after use of iodine solution for wound irrigation prior to closure; however, details regarding the methodology were unclear.2,27 This technique was not employed in our series.

           Compared with previous studies, this study has many potential advantages: a large single surgeon cohort of pediatric and adult spinal deformity patients, inclusion of numerous surgical and curve parameters, and use of correlation analysis in conjunction with multivariate logistic regression. The overall SSI rate was 3.2% for all patients and procedures, while the deep wound infection rate was 1.4%. These rates may have reportable value for policy makers; however, SSI varies significantly with patient population and surgical procedure. Physicians agree that the risk for SSI cannot be eliminated but rather minimized and expeditiously managed. To our knowledge, no prior study has investigated curve type and kyphosis diagnosis as a risk factor for SSI. We report Lenke 3 or 4 curve types as a risk factor for both superficial (4.5-fold increased risk) and deep wound infection (7.4-fold increased risk). We also find an increase in superficial wound infection with a diagnosis of kyphosis (4.4-fold increased risk). This analysis of a large database of only spinal deformity patients allows for delineation of curve parameters prone to increased infection rates, not reported in previous studies. While BMI, age, and the number of patient comorbidities were statistically significant predictors for SSI, the odds ratios were too small to suggest a clinically significant increase in SSI risk.

         Smoking or drinking and diabetes were also not predictors of SSI as reported previously.2-4 This finding is consistent with a previous study from our institution.21 This suggests regional variations in patient profiles and a relatively small number of patients having these comorbidities in our patient population (endocrine N = 85, smoke/drink N = 54). However, we believe that these risk factors would be better delineated in a large cohort of only adult patients. Patients at high risk for infection should be appropriately targeted for preventative therapies including nutritional optimization, weight loss, blood sugar control, prophylactic antibiotics, betadine irrigation, and use of antibiotics in allograft. Disclosure Statement None of the authors have a financial or proprietary interest in the subject matter or materials discussed, including, but not limited to, employment, consultancies, stock ownership, honoraria, and paid expert testimony.

Reference

1. Kirkland KB, Briggs JP, Trivette SL, et al. The impact of surgical-site infections in the 1990s: attributable mortality, excess length of hospitalization, and extra costs. Infect Control Hosp Epidemiol. 1999 Nov;20(11):725-30.

2. Schuster JM, Rechtine G, Norvell DC, Dettori JR. The influence of perioperative risk factors and therapeutic interventions on infection rates after spine surgery: a systematic review. Spine (Phila Pa 1976). 2010 Apr 20;35(9 Suppl):S125-37.

3. Fang A Risk factors for infection after spinal surgery. Spine (Phila Pa 1976). 2005 Jun 15;30(12):1460-5.

4. Olsen MA, Nepple JJ, Riew KD, et al. Risk factors for surgical site infection following orthopaedic spinal operations. J Bone Joint Surg Am. 2008 Jan;90(1):62-9.

5. Borkhuu B, Borowski A, Shah SA. Antibiotic-loaded allograft decreases the rate of acute deep wound infection after spinal fusion in cerebral palsy. Spine (Phila Pa 1976). 2008 Oct 1;33(21):2300-4.

6. Capen DA, Calderone RR, Green A. Perioperative risk factors for wound infections after lower back fusions. Orthop Clin North Am. 1996 Jan;27(1):83-6.

7. Weinstein MA, McCabe JP, Cammisa FP Jr. Postoperative spinal wound infection: a review of 2,391 consecutive index procedures. J Spinal Disord. 2000 Oct;13(5):422-6.

8. Wimmer C, Gluch H, Franzreb M, Ogon M.Predisposing factors for infection in spine surgery: a survey of 850 spinal procedures. J Spinal Disord.1998 Apr;11(2):124-8.

9. Calderone RR, Thomas JC Jr, Haye W, Abeles D. Outcome assessment in spinal infections. Orthop Clin North Am. 1996 Jan;27(1):201-5.

10. Calderone RR, Larsen JM. Overview and classification of spinal infections. Orthop Clin North Am. 1996 Jan;27(1):1-8.

11. Hodges SD, Humphreys SC, Eck JC, et al. Low postoperative infection rates with instrumented lumbar fusion. South Med J. 1998 Dec;91(12):1132-6.

12. Blam OG, Vaccaro AR, Vanichkahorn JS, et al. Risk factors for surgical site infection in the patient with spinal injury. Spine (Phila Pa 1976). 2003 Jul 1;28(13):1475-80.

13. Klein JD, Hey LA, Yu CS, et al. Perioperative nutrition and postoperative complications in patients undergoing spinal surgery. Spine (Phila Pa 1976). 1996 Nov 15;21(22):2676-82.

14. Calderone RR, Garland DE, Capen DA, Oster H. Cost of medical care for postoperative spinal infections. Orthop Clin North Am. 1996 Jan;27(1):171-82.

15. Sponseller PD, LaPorte DM, Hungerford MW, et al. Deep wound infections after neuromuscular scoliosis surgery: a multicenter study of risk factors and treatment outcomes.