Development and Validation of a Preoperative Surgical Site Infection Risk Score for Primary or Revision Knee and Hip Arthroplasty

Everhart, Joshua S. MD, MPH; Andridge, Rebecca R. PhD; Scharschmidt, Thomas J. MD; Mayerson, Joel L. MD; Glassman, Andrew H. MD, MS; Lemeshow, Stanley PhD

Journal of Bone & Joint Surgery - American Volume:
doi: 10.2106/JBJS.15.00988
Scientific Articles

Background: Surgical site infection (SSI) is a major complication following total joint arthroplasty. Host susceptibility to infection has emerged as an important predictor of SSI. The purpose of this study was to develop and validate a preoperative SSI risk-assessment tool for primary or revision knee and hip arthroplasty.

Methods: Data for 6,789 patients who underwent total joint arthroplasty (from the years 2000 to 2011) were obtained from a single hospital system. SSI was defined as a superficial infection within 30 days or deep infection within 1 year. Logistic regression modeling was utilized to create a risk scoring system for a derivation sample (n = 5,789; 199 SSIs), with validation performed on a hold-out sample (a subset of observations chosen randomly from the initial sample to form a testing set; n = 1,000; 41 SSIs).

Results: On the basis of logistic regression modeling, we created a scoring system to assess SSI risk (range, 0 to 35 points) that is the point sum of the following: primary hip arthroplasty (0 points); primary knee (1); revision hip (3); revision knee (3); non-insulin-dependent diabetes (1); insulin-dependent diabetes (1.5); chronic obstructive pulmonary disease (COPD) (1); inflammatory arthropathy (1.5); tobacco use (1.5); lower-extremity osteomyelitis or pyogenic arthritis (2); pelvis, thigh, or leg traumatic fracture (2); lower-extremity pathologic fracture (2.5); morbid obesity (2.5); primary bone cancer (4); reaction to prosthesis in the last 3 years (4); and history of staphylococcal septicemia (4.5). The risk score had good discriminatory capability (area under the ROC [receiver operating characteristic] curve = 0.77) and calibration (Hosmer-Lemeshow chi-square test, p = 0.34) and was validated using the independent sample (area under the ROC curve = 0.72). A small subset of patients (5.9%) had a >10% estimated infection risk.

Conclusions: The patient comorbidities composing the risk score heavily influenced SSI risk for primary or revision knee and hip arthroplasty. We believe that infection risk can be objectively determined in a preoperative setting with the proposed SSI risk score.

Author Information

1Department of Orthopaedics, Wexner Medical Center, The Ohio State University, Columbus, Ohio

2Division of Biostatistics, College of Public Health, The Ohio State University, Columbus, Ohio

E-mail address for A.H. Glassman:

Article Outline

Total joint arthroplasty is a common procedure, with >1 million total hip and knee replacements performed annually in the United States1. Surgical site infection (SSI), occurring in 0.5% to 3% of patients2,3, is a major complication, often requiring prosthesis removal, prolonged antimicrobial therapy, and a several-month delay before reimplantation4. Accurate infection-risk stratification in the setting of modern total joint arthroplasty requires a paradigm shift away from the use of surgical factors to the use of patient factors predictive of host susceptibility. It also requires research dedicated to joint-replacement surgeries alone, as it has become increasingly apparent that individual patient factors that contribute to infection risk vary by surgical procedure2,5-8. The importance of host susceptibility to infection was likely first addressed in orthopaedics by the Cierny-Mader classification system for osteomyelitis, which considers both the severity of the infection and host immune status to select surgical treatment strategies9.

A major barrier to incorporating SSI risk assessment into clinical practice for joint-replacement surgeons is the lack of a method of determining infection risk preoperatively. Specifically, although multivariate regression methods have been used to study independent SSI risk factors after total joint arthroplasty10-18, we are not aware of a current prediction model that can estimate SSI risk after total joint arthroplasty on the basis of preoperative data alone. Current SSI risk assessments rely on factors that are not appropriate or include variables that cannot be measured preoperatively; for example, 2 of 3 variables from the National Nosocomial Infections Surveillance System (NNIS) risk index are the performance of the operation in a contaminated field and an extended operative time compared with similar surgeries17. This is despite multiple examples in the literature of preoperative risk factors, including increased body mass index (BMI)2,19,20, diabetes2,18 tobacco abuse14, malnutrition21, inflammatory arthropathy15, malignancy13, previous SSI12,16, and coagulopathy13. The reporting of such risk factors suggests that the construction of a preoperative SSI prediction model for joint arthroplasty is feasible.

Our objective was to create an SSI prediction model for primary or revision knee and hip arthroplasty by utilizing readily available preoperative variables, including demographic information and medical history. We hypothesized that a model created from these data will have some capability of predicting SSI risk. Such a prediction model could be utilized to identify patients at high risk for SSI and allow the surgeon to provide appropriate counseling for risk modification or nonoperative treatment.

Back to Top | Article Outline

Materials and Methods

Data Sources

Institutional review board (IRB) approval was obtained for this study. All patients included in the study (n = 6,789; 240 with SSI and 6,549 controls) underwent primary total hip (n = 2,920), primary total knee (n = 2,719), revision total hip (n = 652), or revision total knee arthroplasty (n = 498) at a large academic medical center in the Midwestern United States from January 2000 to October 2011. In the case of multiple hip or knee arthroplasties in the same patient, only the initial surgery was included in the data set to ensure that all observations were independent. Therefore, from an initial total of 9,218 surgeries, we excluded 2,429 primary arthroplasties for a different joint or revision arthroplasties in patients already included; the revisions that were included in the data set were revisions of primary arthroplasties that had been performed at our institution before 2000 and revisions of primary arthroplasties performed at other institutions at any date up to October 2011. Medical record data were obtained from a series of queries to the Information Warehouse (a department in our institution chartered to provide patient data sets to researchers, limited to a given IRB-approved study) and included all total joint arthroplasty surgeries, hospitalizations, and International Classification of Diseases, Ninth Revision (ICD-9) diagnosis codes from 1982 to 2012. Also obtained were data pertaining to patient age, height, weight, and race; and basic laboratory values, including albumin and prealbumin, complete blood-cell count (CBC), Staphylococcus screening results, erythrocyte sedimentation rate (ESR), C-reactive protein (CRP) level, and serologic tests for hepatitis B and C, HIV (human immunodeficiency virus), and Clostridium difficile. No patients were recalled for this study; all data were obtained from medical records.

Back to Top | Article Outline

All patients were identified who had an ICD-9 diagnosis code assigned within 365 days after surgery that may be associated with SSI, including codes for osteomyelitis (730.00 to 730.99), septic arthritis (711.0), SSI (998.3 and 998.5), cellulitis (682), and infection or inflammatory reaction resulting from the joint implant or other hardware (996.66 or 996.67). Although multiple criteria have been proposed for diagnosing SSI22-24, an ICD-9 code-based query can have substantially better sensitivity (total hip arthroplasty, 89%; total knee arthroplasty, 81%) compared with that of routine surveillance by hospital epidemiologists (total hip arthroplasty, 56%; total knee arthroplasty, 39%) in detecting SSIs confirmed by record review25. Potential SSI was defined as readmission within 30 days after total joint arthroplasty with a superficial infection-related diagnosis (such as ICD-9 code 998.51: infected postoperative seroma) or within 365 days with a deep or organ-space infection-related diagnosis (such as ICD-9 code 966.66: infection or inflammatory reaction due to internal joint prosthesis) in combination with an elevated inflammatory marker (CRP value of >10.0 mg/L or ESR of >40 mm/hr). The medical records of the 405 patients identified by this method were then individually reviewed for confirmation of superficial infection within 30 days after surgery or deep infection within 365 days after surgery. This included cases of a return to surgery for irrigation, polyethylene exchange, or explantation for presumed infection, regardless of intraoperative culture results. The resulting total number of cases confirmed by individual record review was 240, resulting in a positive predictive value (PPV) of 59.3% for the ICD-9 and inflammatory marker-based screening method.

Medical diagnoses were defined by the presence of a related ICD-9 diagnosis code and by the first date in which the code appeared in a patient’s medical record, resulting in a binary variable (for example, diabetic or nondiabetic) and the duration of time between the diagnosis and the date of surgery (for example, 1,047 days). Weight and BMI were defined by measurements obtained on admission for total joint arthroplasty; in the event of missing data, the most recent weight and height were used. In the case of repeat laboratory data (for example, multiple CBCs over several years), the most recent laboratory value prior to surgery was utilized.

Back to Top | Article Outline
Statistical Analysis

A detailed description of the statistical methods used for model creation and validation can be found in the Appendix. A hold-out sample of 1,000 random observations (41 SSIs) was excluded from the following regression models for later use for validation of our final prediction model. Simple logistic regression models were created for all potential predictors to determine the unadjusted association with SSI. All medical diagnoses were noted on or before the day of admission for total joint arthroplasty. A multivariate logistic regression model was then created (model 1) to create the risk-assessment score (see Appendix). A point scoring system was then created on the basis of the final β coefficients for model 1; points were assigned by dividing the respective coefficients for each predictor by the smallest coefficient in model 1 (diabetes, β = 0.183) and rounding to the nearest half-point. The resulting risk score was determined as the sum of the individual point values, for a possible score ranging from 0 to 35. A model with the risk score as the single predictor was then created (model 2). The discriminatory capability of model 1 and model 2 was assessed by receiver operating characteristic (ROC) curve analysis; goodness of fit was assessed by the Hosmer-Lemeshow (H-L) chi-square test. Finally, the risk-assessment score was validated using the hold-out sample of 1,000 observations.

Back to Top | Article Outline


Descriptive Statistics

The sample had good representation of SSIs (n = 240; 3.5% of the total patient cohort) and uninfected controls (n = 6,549; 96.5%) (Table I). The rate of SSIs was 2.6% for primary total hip arthroplasty, 2.8% for primary total knee arthroplasty, 6.2% for revision total hip arthroplasty, and 8.4% for revision total hip arthroplasty. A total of 113 (47.1%) of the infections occurred within 30 days and 127 (52.9%) occurred from 31 days to 1 year following surgery. The average age at the time of the index procedure (and standard deviation) was 63.1 ± 14.2 years; 43.0% of the patients were male, and 83.4% of the patients were Caucasian. Primary joint replacements were performed for 83.1% of the patients, with revision surgeries performed for the remaining 16.9%. A notable proportion of the patients (12.7%) were morbidly obese (defined as a BMI of ≥40 kg/m2) or diabetic (23.1%). Inflammatory arthropathy (rheumatoid arthritis, psoriatic arthritis, or seronegative inflammatory arthropathy) was present in 6.1% of the patients. A total of 1.9% of the patients had a history of primary bone cancer (most commonly, osteosarcoma or chondrosarcoma), and 4.0% had a pathologic fracture (due to metastatic breast, lung, prostate, or renal cell carcinoma, in most cases).

Back to Top | Article Outline
Univariate Analysis

Many crude associations (unadjusted for confounders) between SSI risk and demographics, surgical procedure, or medical comorbidities were identified (Table II). Age and sex had no association with SSI risk (p = 0.98 and 0.33, respectively). Compared with primary hip arthroplasty, revision hip (odds ratio [OR], 2.84; 95% confidence interval [CI], 1.93 to 4.12) and revision knee arthroplasty (OR, 3.44; 95% CI, 2.32 to 5.06) were at significantly higher risk of SSI (p < 0.001). The highest unadjusted odds of SSI were among patients with a history of staphylococcal septicemia (OR, 18.78; 95% CI, 7.61 to 46.39); this occurred from a variety of potential sources, including infection of the knee or hip, bone or soft-tissue infection of an extremity other than the knee or hip, nonmusculoskeletal infection, and other causes. Other factors with high unadjusted odds of infection included prior reaction or inflammation due to an internal prosthesis or implant (OR, 10.64; 95% CI, 7.51 to 15.09) or a history of lower-extremity osteomyelitis or pyogenic arthritis (OR, 7.04; 95% CI, 4.18 to 11.82) (p < 0.001, all factors). Conversely, several common diagnoses were identified with a smaller but less variable estimated increased odds of SSI, including malnutrition (as defined by ICD-9 code) (OR, 2.96; 95% CI, 1.97 to 4.43), COPD (chronic obstructive pulmonary disease) (OR, 2.37; 95% CI, 1.76 to 3.19), morbid obesity (OR, 3.27; 95% CI, 2.46 to 4.34), depression or bipolar disorder (OR, 1.53; 95% CI, 1.13 to 2.10), and chronic renal failure (OR, 1.86; 95% CI, 1.31 to 2.63).

Back to Top | Article Outline
Multivariate Model and SSI Risk Score

A 16-item multivariate logistic regression model was successfully created on the basis of 4 binary variables for surgical procedure and 12 binary variables for preoperative medical comorbidities (Table III). No demographic variables or laboratory values were found to significantly contribute to the final multivariate model. The resulting SSI risk score corresponding to the multivariate model was calculated as the point sum for the planned procedure and all present comorbidities, with a possible score ranging from 0 to 35. For each additional point in the SSI risk score, the estimated odds of SSI increased 1.43-fold (95% CI, 1.38 to 1.51-fold) (Table III). For example, a non-insulin-dependent patient with diabetes (1 point) and COPD (1 point) who is to undergo a primary total knee arthroplasty (1 point) would have a total SSI risk score of 3 points (Table III) and an estimated probability of SSI of 2.8% (Table IV).

The multiple logistic regression model (model 1) had good discriminatory capability (area under the ROC curve = 0.7734), as did the SSI risk score based on the multivariate model (model 2) (area under the curve = 0.7699) (Fig. 1). The multiple logistic regression model (model 1) was well-calibrated (meaning that it fit the data well across varying levels of SSI risk) (H-L chi-square, 10.68; degrees of freedom [df] = 9; p = 0.30), as was the SSI risk score (model 2) (H-L chi-square, 10.12; df = 9; p = 0.34). When varying the cutoff used to predict SSI, a moderately elevated score such as 6.5 demonstrated 37.7% sensitivity and 94.1% specificity for SSI, whereas a high-risk score such as 10 demonstrated a high level of specificity for SSI (99.4%) but not sensitivity (14.1%) (Table V). Finally, validation of the SSI risk score using the 1,000-patient random hold-out sample demonstrated that it maintained good discriminatory capability (area under the curve = 0.7215) and calibration (H-L chi-square, 9.31; df = 9; p = 0.41).

On the basis of the risk score, most patients had a relatively low estimated risk of SSI, and a small but notable subset of patients was at markedly elevated risk (Figs. 2-A and 2-B and Table VI). A total of 4.8% of the patients who underwent a primary arthroplasty had a risk score of ≥6 points (infection risk rate of ≥7.8%) and accounted for 24% of all SSIs following a primary arthroplasty procedure. A total of 8.3% of the patients who underwent a revision arthroplasty had a risk score of ≥9 points (infection risk rate of ≥20.1%) and accounted for 42% of all SSIs following a revision procedure. The utilization of a 6-point cutoff for primary arthroplasties and a 9-point cutoff for revisions would have resulted in the denial of 5.4% of the surgeries and a reduction of 30.8% of SSIs overall.

Back to Top | Article Outline


Since the advent of total joint arthroplasty, improvements in surgical technique and the use of such prophylactic measures as perioperative antibiotics have drastically reduced SSI rates. However, with increasing yearly volumes of total joint arthroplasty in the United States, there remains a substantial population of patients who have infection-related complications following these procedures. Accurate preoperative infection-risk stratification is an important step toward further reducing SSI rates. We successfully created and validated a simple preoperative SSI risk scoring system for knee and hip arthroplasty with a large and diverse sample of patients from a single academic medical center. We believe that this represents a major advancement from earlier, qualitative methods of determining host susceptibility to orthopaedic infections, such as the Cierny-Mader classification system9.

We believe that the SSI risk score developed in this study can be utilized to identify high-risk patients and allow the surgeon to provide appropriate counseling for risk modification or nonoperative treatment. In our study, primary arthroplasty patients with a risk score of ≥6 and revision patients with a score of ≥9 represented 5.4% of all procedures but 30.8% of SSIs. There are certainly clinical situations in which high infection risk may be acceptable for total joint arthroplasty, but the use of a cutoff such as 6 points for primary arthroplasty and 9 for revision may be appropriate to cue clinicians to thoroughly discuss with patient the risks of surgery, possible nonsurgical alternatives, and especially, possible means of risk reduction. A logical extension of the current study is to implement this tool in clinical practice to determine its impact on preoperative risk counseling and institutional SSI rates. Additionally, patient comorbidities that are potentially modifiable (such as obesity) represented a substantial portion of our risk-assessment models. Treatment of some SSI risk factors, such as Staphylococcus colonization, are considered a standard of care26, while policies to address other potentially modifiable factors in our model, such as the requirement of smoking cessation, occur inconsistently in current clinical practice. It is unknown whether risk-factor modification would result in an actual reduction in SSI rates; additional research is therefore indicated.

The current literature regarding total joint arthroplasty-related infection relies primarily on data from single institutions or general implant registries, often resulting in insufficient data sets for broad data mining and the creation of a robust total joint arthroplasty-specific model of infection prediction. Use of a premature end point for tracking SSI outcomes will result in the misclassification of patients with eventual SSI as healthy controls and will considerably diminish the predictive value of a risk-assessment calculator based on these data. Even high-quality national data sets such as the U.S. National Surgical Quality Improvement Program (NSQIP) data set have limited utility for predicting infection risk in the setting of total joint arthroplasty, as this program only follows patients 1 month after surgery. The SSI risk score developed in the current study was based on comprehensive medical records data for a large sample and the identification of SSIs up to 1 year after surgery, which proved to minimize misclassification, as 52.9% of the SSIs were not confirmed until >30 days after joint replacement surgery.

There is strong evidence in the literature that the relationship between SSI risk and the comorbidities included in our model is biologically plausible. A relationship between SSI and morbid obesity (a BMI of ≥40 kg/m2)2,10,14,19,20,27 or diabetes2,11,12,14,18,27 has consistently been reported; although both conditions are often concurrent, the current study as well as the results of other multivariate analyses demonstrated that both diabetes and morbid obesity independently contribute to SSI risk14,27,28. Tobacco abuse14, inflammatory arthropathy15, primary bone malignancy12,14, and prior lower-extremity fracture29-32 are all previously reported SSI risk factors. A history of total joint arthroplasty-related SSI is a known risk factor for repeat SSI, even after delayed surgery12,16; our results suggest that the risk of repeat SSI remains elevated at least 3 years after a diagnosis of periprosthetic infection or inflammatory response, even for remote-site surgeries (for example, primary left total knee arthroplasty in a patient with a history of right total hip arthroplasty SSI 2 years prior). Although lower-extremity osteomyelitis or septic arthritis may be clinically resolved, the association between these conditions and SSI risk appears to be permanent. Similarly, a history of staphylococcal septicemia, regardless of etiology, had a durable association with SSI in the present study. Additional research is warranted to determine the physiologic basis for the lasting association of prior staphylococcal septicemia, lower-extremity musculoskeletal infection, or inflammatory or infectious complication after total joint arthroplasty with future total joint arthroplasty SSI risk.

One limitation of our SSI risk score is the inability to incorporate several likely risk factors because of limitations in medical coding; for example, no coding distinction is made between a standard and oncologic prosthesis for total joint replacement, although high infection rates have been reported following use of megaprostheses33. The SSI risk score was constructed from data from 1 surgery per patient and may underestimate the infection risk in established patients in a clinical practice who require multiple revisions or who present for subsequent primary arthroplasties in different joints. Institutional SSI rates frequently rely on same-hospital readmission data and often fail to capture patients who present to another hospital for treatment. Although our status as a tertiary referral center allows for a high rate of follow-up for patients with SSI who have undergone total joint arthroplasty within our hospital system, this remains a limitation of any single-institution SSI surveillance system.

In conclusion, we believe that the risk of SSI following primary or revision knee and hip arthroplasty can be reliably estimated in a preoperative setting. A prediction tool for SSI within 1 year following primary or revision knee or hip arthroplasty was created and validated with a large, diverse sample of patients from a single academic medical center. The resulting SSI risk score is a point system based on the planned procedure and the presence or absence of 12 medical comorbidities, with a resulting score of 0 to 35. This preoperative risk score relies on readily obtainable clinical information and addresses an important barrier to reducing SSI rates.

Back to Top | Article Outline

Appendix Cited Here...

Details of the statistical methods used for model creation and validation are available with the online version of this article as a data supplement at

Investigation performed at the Wexner Medical Center, The Ohio State University, Columbus, Ohio

A commentary by James A. Keeney, MD, is linked to the online version of this article at

Disclosure: No external funding was received for this study. On the Disclosure of Potential Conflicts of Interest forms, which are provided with the online version of the article, one or more of the authors checked “yes” to indicate that the author had a relevant financial relationship in the biomedical arena outside the submitted work and “yes” to indicate that the author had a patent and/or copyright, planned, pending, or issued, broadly relevant to this work.

Back to Top | Article Outline


1. Fingar KR, Stocks C, Weiss AJ, Steiner CA. Most frequent operating room procedures performed in U.S. hospitals, 2003. Agency for Health Care Policy and Research (U.S.). 2014(12).
2. Malinzak RA, Ritter MA, Berend ME, Meding JB, Olberding EM, Davis KE. Morbidly obese, diabetic, younger, and unilateral joint arthroplasty patients have elevated total joint arthroplasty infection rates. J Arthroplasty. 2009 ;24(6)(Suppl):84–8. Epub 2009 Jul 15.
3. Cui Q, Mihalko WM, Shields JS, Ries M, Saleh KJ. Antibiotic-impregnated cement spacers for the treatment of infection associated with total hip or knee arthroplasty. J Bone Joint Surg Am. 2007 ;89(4):871–82.
4. Ritter MA, Farris A. Outcome of infected total joint replacement. Orthopedics. 2010 ;33(3):149–54. Epub 2010 Mar 10.
5. Richet HM, Chidiac C, Prat A, Pol A, David M, Maccario M, Cormier P, Bernard E, Jarvis WR. Analysis of risk factors for surgical wound infections following vascular surgery. Am J Med. 1991 ;91(3B):170S–2S.
6. Minnema B, Vearncombe M, Augustin A, Gollish J, Simor AE. Risk factors for surgical-site infection following primary total knee arthroplasty. Infect Control Hosp Epidemiol. 2004 ;25(6):477–80.
7. Harrington G, Russo P, Spelman D, Borrell S, Watson K, Barr W, Martin R, Edmonds D, Cocks J, Greenbough J, Lowe J, Randle L, Castell J, Browne E, Bellis K, Aberline M. Surgical-site infection rates and risk factor analysis in coronary artery bypass graft surgery. Infect Control Hosp Epidemiol. 2004 ;25(6):472–6.
8. Lola I, Levidiotou S, Petrou A, Arnaoutoglou H, Apostolakis E, Papadopoulos GS. Are there independent predisposing factors for postoperative infections following open heart surgery? J Cardiothorac Surg. 2011;6:151. Epub 2011 Nov 14.
9. Cierny G 3rd, Mader JT, Penninck JJ. A clinical staging system for adult osteomyelitis. Clin Orthop Relat Res. 2003 ;414:7–24.
10. Pulido L, Ghanem E, Joshi A, Purtill JJ, Parvizi J. Periprosthetic joint infection: the incidence, timing, and predisposing factors. Clin Orthop Relat Res. 2008 ;466(7):1710–5. Epub 2008 Apr 18.
11. Lai K, Bohm ER, Burnell C, Hedden DR. Presence of medical comorbidities in patients with infected primary hip or knee arthroplasties. J Arthroplasty. 2007 ;22(5):651–6.
12. Berbari EF, Hanssen AD, Duffy MC, Steckelberg JM, Ilstrup DM, Harmsen WS, Osmon DR. Risk factors for prosthetic joint infection: case-control study. Clin Infect Dis. 1998 ;27(5):1247–54.
13. Poultsides LA, Ma Y, Della Valle AG, Chiu YL, Sculco TP, Memtsoudis SG. In-hospital surgical site infections after primary hip and knee arthroplasty—incidence and risk factors. J Arthroplasty. 2013 ;28(3):385–9. Epub 2012 Nov 8.
14. Everhart JS, Altneu E, Calhoun JH. Medical comorbidities are independent preoperative risk factors for surgical infection after total joint arthroplasty. Clin Orthop Relat Res. 2013 ;471(10):3112–9.
15. Carroll K, Dowsey M, Choong P, Peel T. Risk factors for superficial wound complications in hip and knee arthroplasty. Clin Microbiol Infect. 2014 ;20(2):130–5. Epub 2013 Apr 10.
16. Bongartz T, Halligan CS, Osmon DR, Reinalda MS, Bamlet WR, Crowson CS, Hanssen AD, Matteson EL. Incidence and risk factors of prosthetic joint infection after total hip or knee replacement in patients with rheumatoid arthritis. Arthritis Rheum. 2008 ;59(12):1713–20.
17. Culver DH, Horan TC, Gaynes RP, Martone WJ, Jarvis WR, Emori TG, Banerjee SN, Edwards JR, Tolson JS, Henderson TS, et al..; National Nosocomial Infections Surveillance System. Surgical wound infection rates by wound class, operative procedure, and patient risk index. Am J Med. 1991 ;91(3B):152S–7S.
18. Iorio R, Williams KM, Marcantonio AJ, Specht LM, Tilzey JF, Healy WL. Diabetes mellitus, hemoglobin A1C, and the incidence of total joint arthroplasty infection. J Arthroplasty. 2012 ;27(5):726–9.e1. Epub 2011 Nov 4.
19. Dowsey MM, Choong PF. Obesity is a major risk factor for prosthetic infection after primary hip arthroplasty. Clin Orthop Relat Res. 2008 ;466(1):153–8. Epub 2008 Jan 3.
20. Namba RS, Paxton L, Fithian DC, Stone ML. Obesity and perioperative morbidity in total hip and total knee arthroplasty patients. J Arthroplasty. 2005 ;20(7)(Suppl 3):46–50.
21. Greene KA, Wilde AH, Stulberg BN. Preoperative nutritional status of total joint patients. Relationship to postoperative wound complications. J Arthroplasty. 1991 ;6(4):321–5.
22. Horan TC, Gaynes RP, Martone WJ, Jarvis WR, Emori TG. CDC definitions of nosocomial surgical site infections, 1992: a modification of CDC definitions of surgical wound infections. Infect Control Hosp Epidemiol. 1992 ;13(10):606–8.
23. Mangram AJ, Horan TC, Pearson ML, Silver LC, Jarvis WR; Centers for Disease Control and Prevention (CDC) Hospital Infection Control Practices Advisory Committee. Guideline for prevention of surgical site infection, 1999. Am J Infect Control. 1999 ;27(2):97–132, quiz :133-4; discussion 96.
24. Parvizi J, Zmistowski B, Berbari EF, Bauer TW, Springer BD, Della Valle CJ, Garvin KL, Mont MA, Wongworawat MD, Zalavras CG. New definition for periprosthetic joint infection: from the Workgroup of the Musculoskeletal Infection Society. Clin Orthop Relat Res. 2011 ;469(11):2992–4.
25. Bolon MK, Hooper D, Stevenson KB, Greenbaum M, Olsen MA, Herwaldt L, Noskin GA, Fraser VJ, Climo M, Khan Y, Vostok J, Yokoe DS; Centers for Disease Control and Prevention Epicenters Program. Improved surveillance for surgical site infections after orthopedic implantation procedures: extending applications for automated data. Clin Infect Dis. 2009 ;48(9):1223–9.
26. Parvizi J, Matar WY, Saleh KJ, Schmalzried TP, Mihalko WM. Decolonization of drug-resistant organisms before total joint arthroplasty. Instr Course Lect. 2010;59:131–7.
27. Jämsen E, Nevalainen P, Kalliovalkama J, Moilanen T. Preoperative hyperglycemia predicts infected total knee replacement. Eur J Intern Med. 2010 ;21(3):196–201. Epub 2010 Mar 15.
28. Dowsey MM, Choong PF. Obese diabetic patients are at substantial risk for deep infection after primary TKA. Clin Orthop Relat Res. 2009 ;467(6):1577–81. Epub 2008 Oct 8.
29. Massin P, Bonnin M, Paratte S, Vargas R, Piriou P, Deschamps G; French Hip Knee Society (SFHG). Total knee replacement in post-traumatic arthritic knees with limitation of flexion. Orthop Traumatol Surg Res. 2011 ;97(1):28–33. Epub 2010 Dec 16.
30. Ranawat A, Zelken J, Helfet D, Buly R. Total hip arthroplasty for posttraumatic arthritis after acetabular fracture. J Arthroplasty. 2009 ;24(5):759–67. Epub 2008 Jun 4.
31. Sermon A, Broos P, Vanderschot P. Total hip replacement for acetabular fractures. Results in 121 patients operated between 1983 and 2003. Injury. 2008 ;39(8):914–21. Epub 2008 Jul 2.
32. Weiss NG, Parvizi J, Hanssen AD, Trousdale RT, Lewallen DG. Total knee arthroplasty in post-traumatic arthrosis of the knee. J Arthroplasty. 2003 ;18(3)(Suppl 1):23–6.
33. Mavrogenis AF, Pala E, Angelini A, Calabro T, Promagnoli C, Romantini M, Drago G, Ruggieri P. Infected prostheses after lower-extremity bone tumor resection: clinical outcomes of 100 patients. Surg Infect (Larchmt). 2015 ;16(3):267–75. Epub 2015 Mar 26.
Copyright 2016 by The Journal of Bone and Joint Surgery, Incorporated