Clinical Orthopaedics and Related Research®

, Volume 470, Issue 10, pp 2702–2707

Does Dual Antibiotic Prophylaxis Better Prevent Surgical Site Infections in Total Joint Arthroplasty?


  • Amy Sewick
    • Department of Orthopaedic SurgeryHospital of the University of Pennsylvania
  • Amun Makani
    • Department of Orthopaedic SurgeryHospital of the University of Pennsylvania
  • Chia Wu
    • Department of Orthopaedic SurgeryHospital of the University of Pennsylvania
  • Judith O’Donnell
    • Department of Infectious DiseasePresbyterian Medical Center
  • Keith D. Baldwin
    • Department of Orthopaedic SurgeryHospital of the University of Pennsylvania
    • Department of Orthopaedic SurgeryHospital of the University of Pennsylvania
Symposium: 2011 Musculoskeletal Infection Society

DOI: 10.1007/s11999-012-2255-1

Cite this article as:
Sewick, A., Makani, A., Wu, C. et al. Clin Orthop Relat Res (2012) 470: 2702. doi:10.1007/s11999-012-2255-1



It is unclear which antibiotic regimen provides the best prophylaxis against surgical site infection (SSI) in patients undergoing hip and knee surgery.


Therefore, we determined whether dual antibiotic prophylaxis (1) reduced the rate of SSI compared to single antibiotic prophylaxis and (2) altered the microbiology of SSI.


We retrospectively reviewed 1828 primary THAs and TKAs performed between September 1, 2008 and December 31, 2010. We divided patients into two groups: (1) those who received a dual prophylactic antibiotic regimen of cefazolin and vancomycin (unless allergy), or (2) received cefazolin (unless allergy) as the sole prophylactic antibiotic. There were 701 males and 1127 females with an average age of 56 years (range, 15–97 years). We limited followup to 1 year, presuming subsequent infections were not related to the initial surgery.


During this period, there were 22 SSIs (1.2%). The infection rates for dual antibiotic prophylaxis compared to a single antibiotic regimen were 1.1% and 1.4%, respectively. Of 1328 patients treated with dual antibiotic prophylaxis, only one (0.08%) SSI was culture positive for methicillin resistant Staphylococcus aureus (MRSA), while four of 500 patients (0.8%) receiving only cefazolin prophylaxis had culture positive MRSA infection at the time of reoperation.


The addition of vancomycin as a prophylactic antibiotic agent apparently did not reduce the rate of SSI compared to cefazolin alone. Use of vancomycin in addition to cefazolin appeared to reduce the incidence of MRSA infections; however, the number needed to treat to prevent a single MRSA infection was very high.

Level of Evidence

Level III, therapeutic study. See Guidelines for Authors for a complete description of levels of evidence.


Periprosthetic infections remain a devastating complication following total joint arthroplasty that has caused considerable pain and morbidity [5, 12, 15, 25, 28, 32]. Infections have accounted for an estimated 15% of revision THAs [6] and up to 25% of revision TKAs [7] performed annually in the United States. The treatment of infected joint arthroplasties tends to be costly and poses substantial burdens to the healthcare system: A revision THA performed for infection costs on average 2.8 times greater than that performed for aseptic loosening [5, 12, 32]. Furthermore, some authors have reported lower rates of infection control and lower functional scores in patients with infected THA and methicillin resistant organisms [22]. Thus, these numbers have clearly underscored the importance of infection prevention.

Various methods to minimize postoperative infection [11, 15, 38] have included use of antibiotic-impregnated cement [19, 27], laminar flow [2, 23], and minimizing operating room traffic [24]. However, the most effective way to prevent infection has been administering prophylactic antibiotics within 1 hour of surgical incision and continuing its use during the immediate postoperative period [4, 18, 26]. The American Academy of Orthopaedic Surgeons (AAOS) recommended the use of cefazolin or cefuroxime as preferred prophylaxis in TKA and THA, with consideration given to the local incidence of methicillin resistant Staphylococcus aureus (MRSA) [1, 8, 27, 28, 40]. Vancomycin has been recommended when MRSA incidence was greater than 20%, but the use of this antibiotic as a sole prophylactic agent has lead to incomplete skin flora coverage, systemic reactions and toxicity, and antibiotic resistance [27, 40]. Recently, there has been increased attention towards administration of these two prophylactic antibiotics as a strategy to minimize surgical site infections (SSI) [17, 28].

Therefore, the purpose of this study was to determine whether (1) dual antibiotic prophylaxis reduced the rate of SSI compared to single antibiotic prophylaxis, and (2) dual antibiotic prophylaxis altered the microbiology of SSI.

Patients and Methods

We retrospectively reviewed 2215 patients who underwent primary THA or TKA between September 1, 2008 and December 31, 2010. We excluded 325 patients who had drug allergies precluding the use of routine antibiotic prophylaxis (ie, cefazolin or vancomycin), and another 62 patients who received a preoperative antibiotic other than cefazolin or cefazolin plus vancomycin, without a documented allergy. This left 1828 patients for analysis. Five hundred patients received cefazolin alone while 1328 patients received both cefazolin and vancomycin at least within 1 hour prior to incision. There were 1174 primary TKAs and 654 primary THAs. All surgeries were performed by one of four fellowship trained arthroplasty surgeons (GCL, JPG, CLN, CLI). Surgical site infections, defined by the Centers for Disease Control and Prevention (CDC) as an infection occurring at the site of surgery within 30 days from the operative date or up to 1 year if an implant was inserted and the infection appears related to the surgery [18], were identified through the use of the official institutional Clinical Effectiveness and Quality Improvement (CEQI) database [16] for infection control, as well as by review of the medical records for each patient who returned to the operating room for infection within this time period. This data was reported to the state as part of the pay-for-quality incentive program. A total of 22 of 1828 (0.01%) patients developed SSI and returned to the operating room with infections related to their index arthroplasty within 1 year. There were 14 hips and eight knees. The average followup was 18 months (range, 12–24 months). The minimum followup was limited to 12 months, presuming subsequent infections were not related to the initial surgery. No patients were lost to followup. We abstracted patient variables that could impact infection, including age, sex, general health (American Society of Anesthesiologist [ASA] classification), and type of procedure (hip or knee), from each medical record. No patients were recalled specifically for this study; all data was obtained from medical records.

We compared the rate of SSI between the two groups using a Pearson’s Chi-square test. Additionally, we examined hips and knees separately by similar methods to determine if the rate of infection was different between the two groups. When there was less than five patients in any cell, we used the Fisher’s exact test. Three of the four surgeons (CLI, JPG, CLN) employed dual antibiotic prophylaxis (n = 1328) compared to one surgeon (GCL) who chose cefazolin as the sole prophylactic agent (n = 500). There were no differences between the two groups regarding age, ASA classification, time of cefazolin infusion, and number of hip and knee procedures (Table 1). There were a greater percentage of females in the dual antibiotic group (p = 0.001).
Table 1

Group demographics


Cefazolin (n = 500)

Cefazolin and vancomycin (n = 1328)

p value

Average age ± SD (years)

59.6 ± 12.0

60.6 ± 12.5










ASA class

 1 or 2




 3 or 4



Surgical site








ASA = American Society of Anesthesiologist.

We used Fisher’s exact test to determine if the rate of MRSA was different between groups, and conducted a univariate analysis to determine the effect of potential confounders on the outcome of variable (infection). These variables were tested with Chi-square tests when dichotomous or categorical (ASA class and sex) and a Mann-Whitney U test if they were continuous variables (age). We used multivariate logistic regression to adjust the risk of infection by age, sex, general health (ASA class), type of surgery (hip or knee), and type of antibiotics. A criterion of 0.10 or less on univariate analysis was used as criteria for entry into the model at each step, and a backwards likelihood ratio method was used to automatically remove variables at each step until all variables were significant. Finally, we performed a post-hoc power analysis to detect subtle differences between the two groups. The statistical analysis was performed using SPSS® software version 16.0 (SPSS®, Chicago, IL, USA).


Infection occurred with a similar incidence (p = 0.636) in patients who received only cefazolin and patients who received both cefazolin and vancomycin: seven of 500 (1.4%) versus 15 of 1328 (1.1%), respectively. This similar incidence held when looking at the rates of SSI in the THA (p = 0.570) and TKA (p = 0.999) subgroups. While in univariate analysis age and the joint operated on predicted SSI (Table 2), the multivariate binary logistic regression showed that only the joint operated on predicted SSI, with hips getting SSI more frequently (p = 0.01) than knees.
Table 2

Demographics of infection groups


Non-SSI (n = 1806)

SSI (n = 22)

p value

Average age ± SD (years)

60.4 ± 12.4

55.6 ± 13.6











 1 or 2




 3 or 4



Surgical site













 Cefazolin and vancomycin



SSI = surgical site infections; ASA = American Society of Anesthesiologist.

The microbiology of SSI in the two groups differed (Table 3). Patients receiving cefazolin alone were more likely to develop MRSA infections compared to patients receiving cefazolin and vancomycin for prophylaxis. We found a higher percentage (p = 0.022) of patients with MRSA positive SSI in patients who received only cefazolin compared to those who received both cefazolin and vancomycin: four of 500 (0.008%) versus one of 500 (0.002%). The number needed to treat with vancomycin additional prophylaxis to prevent one MRSA infection was 138 (CI 101.5, 2828.2). There were no known complications attributed to the use of vancomycin in addition to cefazolin for infection prophylaxis.
Table 3

Bacteria cultured in the 22 SSIsa



Cefazolin and vancomycinb

No infection













 Streptococcus varieties






SSI = surgical site infection; MRSA = methicillin resistant Staphylococcus aureus; MSSA = methicillin-sensitive Staphylococcus aureus; CNS = coagulative negative Staphylococcus

aThere were seven SSI in the cefazolin group. Multiple organisms grew in multiple sites.

bThere were 15 SSI in the cefazolin and vancomycin group. Multiple organisms grew in multiple sites.

cOdds ratio 95% CI, 29.0 (4.0, 593.3); numbers needed to treat 95% CI, 138.0 (101.5, 2828.2).

dOne each of E coli, Klebsiella, B fragilis.

eOne each of Diptheria, E aerogenes, Enterococcus, Pseudomonas, Acinectobacer, Proteus.


Surgical site infections have caused substantial morbidity and patient dissatisfaction following joint arthroplasty [3, 10, 34, 41]. Furthermore, these infections have reportedly lead to deep prosthetic infections, requiring additional surgeries at considerable costs to the healthcare system [31, 36]. While various techniques, such as use of laminar flow, antibiotic cement, and minimizing operating room traffic, have helped minimize infections [2, 11, 15, 19, 23, 24, 27, 38], the single most effective strategy for infection reduction has been timely administration of prophylactic antibiotics during the perioperative period [8, 29, 33, 37]. The AAOS recommended use of either cefazolin or cefuroxime as preferred prophylactic agents in patients undergoing primary THA or TKA, with consideration given to the incidence of MRSA and patient allergies [1, 29, 33, 37]. However, due to changing skin flora and increasing prevalence of MRSA in the community, these cephalosporins may not provide adequate prophylaxis. Therefore, we determined whether dual antibiotic prophylaxis (1) reduced the rate of SSI compared to single antibiotic prophylaxis and (2) altered the microbiology of SSI.

There were several limitations to our study. First, this was a retrospective study with data obtained from medical records and the institutional CEQI data submitted to the state for quality improvement assurance. We did not recall any patients specifically for this study; therefore, it was possible that some patients with SSI were treated at outside hospitals and did not factor into our calculations. Second, most patients receiving cefazolin as the sole prophylactic agent underwent THA and TKA by a single surgeon (GCL), which may have introduced a selection bias. However, because the infection rates varied minimally amongst the four surgeons and the ASA classification of the patients in these two groups were similar, we believed these comparisons to be valid. Third, we did not monitor the effect of increasing vancomycin use on the emergence of resistant organisms. These data were not available to us, but we recognized the importance of these data in order to truly comment on the benefits and/or detriments of single versus dual antibiotic prophylaxis. Fourth, patients in this study were not screened for MRSA or treated for MRSA colonization prior to joint arthroplasty. Patients colonized with MRSA may have been at higher risk of developing postoperative MRSA infection [13, 17, 20]. Also, other factors, in addition to antibiotic choice, could have affected the rate of infection, including host factors (colonization, recent infection) and environmental/social factors (recent hospitalization or nursing home residence). Thus, the pathogenesis of infection was multifactorial and antibiotic choice was one of many factors that could have affected its incidence. Finally, due to the relative paucity of SSI cases, power to detect small changes in infection rate (in this case between 1.4% and 1.1%) required exceptionally large numbers. A post hoc power analysis showed that in order to detect changes this small, assuming a balanced design at 80% power and a 5% Type I error rate, would have required over 8000 patients per group. These numbers, even if they provided a result, would have resulted in a number needed to treat of over 300. This represented a highly clinically inefficient treatment [21]. However, this study had strengths. It was comprised of a large number of patients undergoing primary THA or TKA. We highlighted some of the issues associated with optimizing antibiotic prophylaxis (ie, increasing coverage at the potential expense of increasing resistant organisms). We found that, even for MRSA, treating with vancomycin represented at best an inefficient treatment strategy [21]. A larger, prospective, multicenter study would certainly help answer these questions more definitively.

The use of dual antibiotics (cefazolin and vancomycin) did not decrease the incidence of SSI in patients undergoing primary THA or TKA at our institutions. During the study period, seven patients who received only cefazolin developed a SSI (as defined by the CDC) compared to 15 patients who received both cefazolin and vancomycin (1.4% versus 1.1%). Prior to 2008, patients undergoing THA and TKA at our institution received cefazolin as the sole prophylactic agent, except in cases of drug allergy, in which case vancomycin was usually given. Due to the high incidence of MRSA infections (30%) at our institution, a Committee for Infection Control recommended the addition of vancomycin to the prophylactic regimen. Miller et al. [30] developed a decision analysis model to determine what MRSA prevalence might benefit from the use of vancomycin as antibiotic prophylaxis for cardiothoracic SSIs. Their findings suggested that vancomycin should be considered in populations with MRSA prevalence greater than 3%. Furthermore, Ritter et al. [35] reported on a series of 201 consecutive patients undergoing total joint arthroplasty treated with a single dose of 1 gram of vancomycin and 80 grams of gentamicin. The trough levels of vancomycin up to 24 hours exceeded the minimum inhibitory concentration for all sensitive organisms, and they reported no postoperative infections. Finally, Finkelstein et al. [14] reported similar efficacy between vancomycin and cefazolin in preventing SSIs in 885 patients undergoing cardiac surgery. Consequently, while vancomycin should be considered in settings of high prevalence of MRSA or colonization, the addition of vancomycin to cefazolin in this study did not reduce the incidence of SSIs compared to use of cefazolin alone.

The microbiology of SSIs in patients receiving only cefazolin differed to that of patients treated with a combination of cefazolin and vancomycin. Of the seven SSIs in 500 patients who received only cefazolin, MRSA was isolated in four (0.8%) patients compared to one of 1328 (0.08%) patients who received both cefazolin and vancomycin. The most common infecting organism in patients receiving dual prophylaxis was methicillin-sensitive Staphylococcus aureus (MSSA), followed by MRSA. We found MRSA infection to be more common among patients receiving single prophylaxis. Similar changes in infecting organisms were observed by others [9, 14, 39]. Finkelstein et al. [14] reported in a series of cardiac patients that those receiving cefazolin for prophylaxis were more likely to have infections with beta lactam resistant organisms, and those who received vancomycin were more likely to have methicillin-susceptible Staphylococci. However, while there were no apparent complications of dual antibiotic therapy in this study, the addition of vancomycin to routine prophylactic regimen should be used with caution: the AAOS has recommended against the routine use of vancomycin because it could promote the development of vancomycin resistant enterococcus (VRE) colonization and infections [1]. Furthermore, it recommended that vancomycin be reserved for the treatment of serious infection with ß-lactam-resistant organisms or for treatment of infection in patients with life-threatening allergy to ß-lactam antimicrobials [1]. So, while the addition of vancomycin to the prophylactic antibiotic regimen did appear to decrease the incidence of MRSA infections in our patient population, the number needed to prevent one single MRSA infection was very high (138). As such, this was clinically inefficient, and the risks of increasing bacterial resistance should be carefully weighed against the benefit of fewer MRSA infections.

In conclusion, dual antibiotic prophylaxis (cefazolin and vancomycin) did not appear to affect the rate of SSI in patients undergoing primary THA and TKA compared to patients receiving single agent prophylaxis (cefazolin only). The addition of vancomycin did alter the profile of the infecting organism, as patients receiving only cefazolin were more likely to get MRSA infections postoperatively compared to those receiving dual prophylaxis. These findings were consistent with those reported in the cardiac literature [14, 39]. However, because the risk of increased bacterial resistance is almost certain and the number needed to treat is so high, we concluded the use of dual antibiotic prophylaxis is not sufficiently effective to reduce infection rates for routine use. Our data support the AAOS position of applying vancomycin prophylaxis only in cases of known MRSA carrier status. Further studies to include preoperative screening and decolonization programs should be performed to ascertain the best antibiotic prophylaxis for patients undergoing elective primary THA and TKA.


The authors thank Jonathan P Garino MD, Charles L Nelson MD, and Craig L Israelite MD for contributing patients for this study.

Copyright information

© The Association of Bone and Joint Surgeons® 2012