The recently published WHO and CDC guidelines included some overlapping recommendations for glycaemic control, maintenance of normal body temperature (normothermia) intraoperatively and intraoperative oxygenation targets.

With respect to glycaemic control, the WHO guidelines noted from a review of 15 randomised controlled trials that intensive perioperative glycaemic control was associated with a 57% reduction in the odds of surgical site infections (OR 0.43; 95% CI 0.29–0.64) in patients with and without diabetes mellitus; however, the optimal threshold glucose concentrations were not established [6]. While cautioning about the potential adverse impact of hypoglycaemia, the authors made a conditional recommendation for intensive perioperative glycaemic control supported by low quality of evidence [6]. In contrast, the CDC guidelines recommended perioperative glycaemic control with targets of less than 200 mg/dL for both diabetic and nondiabetic patients [5]. This was graded as a strong recommendation based on high- to moderate-quality evidence derived from two randomised trials in cardiac surgery [5]. While both guidelines emphasised the importance of glycaemic control for both diabetic and nondiabetic patients, the optimal threshold for glucose concentrations appears to be an unresolved issue with divergent literature [5, 6].

With respect to oxygenation, both guidelines included strong recommendation based on moderate-quality evidence about the administration of increased fraction of inspired oxygen (FiO₂) [5, 6]. The administration of high FiO₂ (80%) was associated with a 27% reduction in the odds of surgical site infections compared to standard FiO₂ (OR 0.72; 95% CI 0.55–0.94). Similarly, there was agreement with respect to maintenance of normal body temperature (normothermia) considered a strong recommendation based on analysis of two randomised controlled trials [5, 6].

Surgical hand antisepsis to remove transient flora and reduce resident skin flora is considered standard practice in preparation for surgery. The two recommended methods are surgical hand scrubbing with antimicrobial soap and water and surgical handrubbing with a waterless alcohol-based handrub [36]. There is limited evidence regarding which is the optimal agent for surgical hand preparation.

In a meta-analysis by Tanner et al., two aspects of surgical hand preparation were examined: firstly, the impact of surgical hand preparation on surgical site infections and the impact of different agents on the number of bacteria present on the surgeon’s hands following cleansing [36]. Four randomised trials examining the impact of surgical hand antisepsis on surgical site infections were identified and included in the meta-analysis. The included studies were noted be of low or moderate quality and included a variety of active components and approaches, limiting comparison. Overall, no trial demonstrated superiority of one agent or approach for the prevention of surgical site infections [36].

Ten studies examine the impact of surgical hand antisepsis on the number of bacteria present on the skin, measured as the number of colony-forming units (CFUs). Again the authors noted low to moderate quality of evidence with noted methodological issues identified [36]. Chlorhexidine gluconate was associated with a reduced number of CFUs compared with povidone-iodine, and the effect was sustained out to 2 h following scrubbing (Fig. 5.6). However, these studies did not examine the clinical impact of this observation, in particular whether this observed reduction in hand contamination was associated with fewer surgical site infections [36]. In addition, among the four studies comparing surgical handrubbing to surgical handwashing, there was a suggestion that alcohol-based handrubs were more effective at reducing CFUs. But due to study quality, the authors were reticent to draw definitive conclusions [36].

In the meta-analysis performed in the WHO guidelines, similar conclusions were drawn: namely, while surgical hand preparation was an accepted aspect of overall patient care, there was no evidence to support one method or agent over a comparator [7].

In addition to the potential contamination of the surgical incision with bacteria from the hands of the healthcare worker, the patient’s own skin flora is thought to be a major source of bacterial contamination of the surgical site [37, 38]. This bacterial contamination may lead to surgical site infection, and procedures in which prosthetic material are implanted, such as joint replacement surgery, are particularly prone to infection. In a rabbit model study by Southwood et al., the infectious dose of S. aureus required to establish an infection in 50% (ID₅₀) of animals was 200-fold higher in the absence of prosthetic material. Indeed contamination of the prosthesis with only 50 S. aureus bacteria at the time of implantation induced an infection in 50% of the animals tested [39].

Preoperative skin antisepsis is performed in the operating room and aims to reduce the microbial load on the patient’s skin prior to the surgical incision [7]. The three agents commonly used for skin antisepsis are chlorhexidine gluconate , iodophors or alcohol [40]. Alcohol acts rapidly to denature the cell wall protein of a range of microorganisms, but it has no residual activity [41, 42]. Chlorhexidine gluconate is bactericidal and acts through disruption of the outer cell membranes and cytoplasmic membranes of microorganisms [42]. It has a slower onset of action, but it has prolonged residual activity [8, 43]. Iodine is frequently formulated in a polyvinylpyrrolidone-iodine (povidone-iodine) complex which improves the stability of iodine and allows for slower release of free iodine [8, 41, 42]. Iodine disrupts intracellular proteins in microorganisms [42]. Iodophors are inactivated by organic material, including blood and serum which may impact their efficacy [8, 40]. They have minimal residual activity, but they exhibit persistence of bacteriostatic activity when on the skin [8]. Alcohol is frequently combined with chlorhexidine gluconate or iodophors to augment the activity of these agents [40]. Guidelines frequently recommend against the sequential application of iodophors after chlorhexidine as it is theorised that iodophors may inactivate chlorhexidine [38, 44]. However, this has not been supported by in vitro studies: indeed, in a study by Anderson et al., the activity of chlorhexidine gluconate and povidone-iodine was augmented when used in combination [43]. There are limited clinical studies examining the combination of chlorhexidine and povidone-iodine; two cohort studies in patients undergoing neurosurgical procedures have examined these combinations, but there are no randomised controlled trials to date [45, 46].

The optimal agent for skin antisepsis has been the focus of significant research interest. Despite this, a meta-analysis by Dumville et al. in 2015 was unable to provide definitive evidence of superiority of one agent, owing in part to the paucity of high-quality, adequately powered trials [41]. In the meta-analysis performed in the WHO guidelines, alcohol-based antiseptic combinations were compared with aqueous-based preparations. Analysis of twelve randomised controlled trials demonstrated the use of alcohol-based preparations was associated with a 40% reduction in the risk of surgical site infections (OR 0.60; 95% CI 0.45–0.78) [7]. There was low- to moderate-quality evidence suggesting chlorhexidine-alcohol combinations were superior to iodophor-alcohol-based combinations (OR 0.58; 95% CI 0.42–0.80) [7]. Based on these analyses, the WHO guidelines recommended the use of alcohol-based chlorhexidine gluconate antiseptic solutions for skin preparation [7].

In contrast, the CDC guidelines did not demonstrate benefit of chlorhexidine-alcohol combinations over iodophor-alcohol combinations based on the analysis of six randomised controlled trials with high grade of evidence base (OR 0.64; 95% CI 0.24–1.71) [5]. Therefore, while concurring with the need for an alcohol-based antiseptic, the CDC guidelines did not include a specific recommendation for use of chlorhexidine or iodophor [5].

Overall there is an agreement that alcohol-based preparations are optimal, but there remains conflicting data and recommendations about whether chlorhexidine gluconate or an iodophor is the superior agent to use in combination with alcohol.

Given the known association between shoulder surgery and Cutibacterium acnes (formerly known as Propionibacterium acnes ) surgical site infections, a number of research groups have examined the efficacy of surgical antiseptic preparations against this organism. In a study by Saltzman et al., 150 patients were randomised to surgical antiseptic preparation with either chlorhexidine gluconate and alcohol, iodophor and alcohol or iodine scrub and paint. The surgical incision site was swabbed following skin preparation, and the rate of positive cultures for different microorganisms was compared. Coagulase-negative Staphylococcus species and C. acnes were the most common isolates [47]. When comparing the alcohol-based solutions, an organism was isolated in 19% of cases when iodophor-alcohol were used compared with 7% of cases with chlorhexidine-alcohol (P = 0.01): there was no difference observed with respect to the isolation of C. acnes; however, the number of isolates was small (n = 28) [47]. There were no surgical site infections observed in this study; therefore, the clinical impact of the observed difference could not be assessed [47]. A similar study by Savage et al. in lumbar surgery did not find significant difference in rates of positive cultures after skin preparation or wound closure with alcohol-based chlorhexidine preparations compared with alcohol-based iodine preparations [48]. C. acnes has a predilection for deeper dermal structures such as sebaceous glands; therefore, Lee et al. postulated that the failure of skin preparation to eliminate C. acnes in the dermis may act as the potential source of this bacteria in shoulder joint surgical site infections [49]. Dermal punch biopsies were obtained in ten healthy male volunteers following skin antisepsis with an alcohol-based chlorhexidine solution and cultured for C. acnes. C. acnes was isolated in 70% of dermal specimens, despite skin antisepsis [49]. In a follow-up from these observations, Sabetta et al. examined whether the preoperative application of benzoyl peroxide, a commonly used topical preparation for acne treatment, would lead to a reduction in culture positivity for C. acnes [50]. Using the untreated (with benzoyl peroxide), nonsurgical arm as control, Sabetta et al. observed a significant reduction in the number of positive cultures isolating C. acnes [50]. Once again, the clinical implications of these observational studies have not been established and further investigation is required.

Plastic, adhesive, incise drapes are commonly used in surgery based on a theoretical risk of contamination of the wound by skin flora and the thought that the adhesive drapes act as a ‘microbial barrier’ [51]. However, a number of studies have raised concerns as to whether the presence of adhesive drapes may promote bacterial growth and, therefore, conversely increase the risk of surgical site infection [52, 53]. A randomised controlled trial by Falk-Brynhildsen et al. examined bacterial recolonisation of the skin with adhesive drapes in 140 patients undergoing cardiac surgery [52]. Significant differences were observed in the proportion of skin swabs recolonised with C. acnes when adhesive drapes were used compared to no drapes after 120 min (63.1% versus 44.4%, respectively; P = 0.034), and also an increased proportion of subcutaneous tissue specimens isolated coagulase-negative staphylococci at the end of surgery in patients with adhesive drapes (14.7% versus 4.5%, respectively; P = 0.044) [52].

This question was recently addressed in a Cochrane review by Webster et al. [51]. The initial analysis compared the incidence of surgical site infections with and without adhesive drapes in a range of surgical procedures. Five randomised controlled trials with 3082 participants were included, and the analysis demonstrated an increased risk of infection with adhesive drapes (RR 1.23; 95% CI 1.02–1.48) with high-quality evidence [51]. The second analysis compared iodine-impregnated adhesive drapes compared to no adhesive drapes and included two randomised controlled trials with 1133 participants. The pooled analysis did not demonstrate any increased risk of infection with iodine-impregnated drapes (RR 1.03; 95% CI 0.66–1.60) with moderate-quality evidence [51].

In contrast, the meta-analysis conducted by Berríos-Torres et al. in the CDC guidelines did not demonstrate an increased risk of infection with the use of drapes compared with no drapes (n = 1,742, RR 1.05; 95% CI 0.66–1.60) [5]. This meta-analysis included four randomised controlled trials (which were also included in the Cochrane review); however, the meta-analysis did not include the study by Cordtz et al., which was included in the review by Webster et al. [5, 51]. Cordtz et al. conducted a factorial randomised controlled trial in 1340 patients undergoing caesarean section comparing incision drape or no drape with or without skin disinfection with 2.5% iodine in 70% alcohol [54]. The frequency of infections was higher in patients with adhesive drapes (15.0% versus 10.9%; RR 1.37; 95% CI 1.03–1.82) [51, 54]. When examining iodine-impregnated drapes, Berríos-Torres et al. identified the same two randomised controlled trials as Webster et al. with similar results [5, 51]. The CDC recommended that adhesive drapes were not necessary to prevent surgical site infections as a weak recommendation [5]. Webster et al. agreed with the conclusion that there was no evidence to support the use of drapes adding that nonimpregnated may increase the risk of infection [51].

Optimal surgical antimicrobial prophylaxis prescribed requires adoption of seven key principles (Table 5.1). In regard to joint replacement surgery, the specific literature regarding these elements is quite broad in terms of quality. The current literature predominantly references hip and knee joint replacement surgery as opposed to elbow, ankle and shoulder arthroplasty. However, the evidence pertaining to the surgical antimicrobial principles for lower limb joint replacement surgery can be broadly applied to arthroplasty in general.

Joint replacement surgery is classified as a class I/clean procedure by the CDC surgical wound classification system, but given that the procedure involves the implantation of foreign material, surgical antimicrobial prophylaxis is recommended [8, 55].

An early meta-analysis, AlBuhairan et al. pooled the data from seven randomised controlled trials and demonstrated that the use of antimicrobial prophylaxis was associated with an 81% risk of wound infection compared with no antibiotics among patients undergoing arthroplasty (RR 0.19; 95% CI 0.12–0.31) [58]. This included studies in which the prophylaxis was administered in the form of antibiotic-loaded cement with or without intravenous antimicrobials. In addition, the authors conducted multiple pooled analyses and did not identify any significant difference when comparing the following: cephalosporins with teicoplanin (RR 1.22; 95% CI 0.64–2.34), cephalosporins with penicillin derivatives (RR 1.17; 95% CI 0.31–4.41) and second-generation with first-generation cephalosporins (RR 1.08; 95% CI 0.63–1.84) [58]. It is important to note that the studies included in these pooled analyses ranged from 1979 to 1999 [58]. The authors also highlighted a number of methodological issues impacting on the quality and interpretation of the included trials [58].

In a subsequent meta-analysis in primary hip and knee joint replacement by Voigt et al., the use of systemic intravenous surgical antimicrobial prophylaxis was associated with a 77% reduction in the risk of infection after 6–12 months of follow-up compared to placebo (RR 0.23; 95% CI 0.12–0.43) [56]. Similar findings were demonstrated when assessing infections in the longer term (up to 6.5 years). However, this pooled analysis was for primary hip joint replacements, and all included studies were of moderate grade quality [56].

Vancomycin has been studied as a prophylactic antibiotic in randomised controlled trials in cardiothoracic surgery including a study by Finkelstein et al., in which vancomycin was compared to cefazolin in patients undergoing sternotomy. Overall there was a 73% reduction in methicillin-resistant S. aureus and 82% reduction in Enterococcus. There was, however, a significant increase in methicillin-susceptible S. aureus in patients receiving vancomycin [61]. Data from the Victorian Healthcare Associated Surveillance System, in Australia, by Bull et al. have identified a similar trend [62]. In this large retrospective observational study, including 10,973 hip and 7369 knee joint replacement surgeries, there was a significant increase in methicillin-susceptible S. aureus in patients receiving vancomycin alone as surgical antimicrobial prophylaxis compared with beta-lactam antibiotics [62].

A systematic review and meta-analysis were performed by Saleh and colleagues comparing glycopeptide and beta-lactam surgical antibiotic prophylaxis in cardiovascular and orthopaedic surgery [63]. Overall fourteen randomised controlled trials were included in the meta-analysis; including six studies examining patients undergoing orthopaedic procedures. No included trial examined combination prophylaxis with a beta-lactam plus a glycopeptide antimicrobial. There was no difference in the overall incidence of surgical site infections between glycopeptide and beta-lactam surgical antimicrobial prophylaxis (RR 0.87; 95% CI 0.63–1.18; P = 0.37). There was, however, a reduction in the incidence of methicillin-resistant staphylococcal infections (RR 0.52; 95% CI 0.29–0.93) and enterococcal infections (RR 0.36; 95% CI 0.16–0.80) with glycopeptide surgical antimicrobial prophylaxis [63]. Therefore, current data does not demonstrate an overall efficacy with vancomycin when used alone. Indeed, data suggest that use of vancomycin in isolation may conversely lead to an increase in methicillin-susceptible S. aureus surgical site infections.

Based on these observations, the potential benefit of combination prophylaxis to cover both sensitive and resistant staphylococci is currently being examined. A Cochrane review examined surgical antimicrobial prophylaxis for prevention of methicillin-resistant S. aureus surgical site infections following other surgical procedures. The review included twelve randomised controlled trials, including two studies examining combination surgical antimicrobial prophylaxis. The authors could not draw definitive conclusions about the role of combination surgical antimicrobial prophylaxis, due to the high risk of bias, very low quality of evidence, significant heterogeneity and lack of patient and economic outcome data. The authors concluded that there was a need for well-designed randomised controlled trial in this arena [64].

In the absence of such a trial, there is evidence from observational studies demonstrating a reduction in the incidence of infections after introduction of combination prophylaxis in total joint replacement surgery [65–68]. A retrospective review of 1828 patients by Sewick et al. compared the efficacy of a combination antibiotic prophylaxis regimen (cefazolin and vancomycin) to cefazolin alone [69]. Infection rates were comparable, 1.1% and 1.4%, respectively; methicillin-resistant S. aureus infections were reduced when comparing the combination prophylaxis to the cefazolin regimens, 0.008% and 0.8%, respectively (P = 0.022) [69]. However, the number needed to treat with additional vancomycin prophylaxis to prevent one methicillin-resistant S. aureus infection was high (138, 95% CI: 101.5–2828.2) [69].

Similarly, Liu et al. examined the role of combination prophylaxis with vancomycin and cefazolin in 414 patients undergoing revision hip or knee replacement surgery . The introduction of combination prophylaxis reduced the rate of infection from 7.89% to 3.13% (Fisher’s exact test P = 0.046) [67]. The proportion of methicillin-resistant Staphylococcus decreased from 53% to 29% following introduction [67].

A third group examined the incidence of prosthetic joint infection after the introduction of combination prophylaxis with teicoplanin and cefuroxime in a retrospective cohort of 1896 patients undergoing total hip or knee arthroplasty. Tornero et al. reported a 64% reduction in the overall risk of prosthetic joint infection following the adoption combination prophylaxis (hazard ratio 0.355; 95% CI 0.170–0.740). There was a significant reduction in infections due to Gram-positive bacteria following the introduction of combination prophylaxis (2.9% versus 0.9%; P = 0.002). In particular, there was a reduction in both methicillin-susceptible and methicillin-resistant S. aureus infection in the combination prophylaxis cohort (1.6% to 0%; P < 0.0001) [68].

The reduction in the incidence of surgical site infections must be balanced against the potential unintended consequences of combination surgical antimicrobial prophylaxis. These concerns include serious adverse outcomes such as acute kidney injury, reported by Courtney et al. following the introduction of combination surgical antimicrobial prophylaxis (13% versus 8% for cefazolin prophylaxis; P = 0.002) [70]. Another major concern with vancomycin prophylaxis is the emergence of vancomycin-resistant Enterococcus (VRE) and other antibiotic-resistant microorganisms [71]. Exposure to vancomycin has been linked with colonisation with organisms such as VRE [72]. Literature regarding subsequent colonisation of patients with VRE after exposure to vancomycin surgical antimicrobial prophylaxis is conflicting; in a study by Merrer et al., use of vancomycin surgical antimicrobial prophylaxis was not associated with subsequent VRE colonisation [73]. In contrast, Kachroo et al. demonstrated a 4% incidence of VRE colonisation after vancomycin surgical antimicrobial prophylaxis, but the authors did not provide data on the prevalence of VRE within their centre; therefore, establishing a causal link is challenging [71]. Thus, further research is required to assess the risks and benefits for the prevention of methicillin-resistant staphylococcal infections .

Another area of current debate is the role of extended prophylaxis coverage for Gram-negative bacilli. In a before and after study, the incidence of surgical site infection following the addition of weight-based dosing of gentamicin to cefazolin in patients undergoing hip arthroplasty was examined. The study group includes 4122 hip arthroplasties receiving cefazolin and 1267 hip arthroplasties receiving cefazolin plus gentamicin. In parallel, the researchers also include 4695 knee arthroplasties receiving cefazolin throughout both time periods as a quasi-‘control’ group, presumably as an attempt to account for secular trends. With the inclusion of gentamicin prophylaxis, the incidence of surgical site infections decreased from 1.19% to 0.56% (Fisher’s exact test P = 0.05). Of note, the rate of infections due to Gram-negative bacilli decreased from 0.32% to 0% (P = 0.048) [74]. There was no increase in toxicities associated with aminoglycoside therapy [75].

Finally, there is ongoing concern about the association between cephalosporin use, including as surgical antimicrobial prophylaxis, and Clostridium difficile infections [76]. Earlier research of 108 participants receiving a single dose of cephalosporin as prophylaxis observed a high proportion of patients with C. difficile detected on subsequent faecal samples [77]. Following a single dose of cefazolin, 14.3% of patients had toxin-positive C. difficile detected; of note, however, no patient had symptomatic disease [77].

In response to concerns about increasing rates of C. difficile infections, the antibiotic regimen at many Scottish hospitals was altered as part of a broader policy to limit cephalosporin use [78]. Bell and colleagues examined the effect on preoperative acute kidney injury following the change in surgical antimicrobial prophylaxis protocols from cefuroxime to flucloxacillin and gentamicin for total joint replacement surgery. In this large cohort study (n = 12,482) applying time-series analysis methodology, the alteration of the surgical antimicrobial prophylaxis regimen to flucloxacillin and gentamicin was associated with 94% increase (95% CI 93.8–94.3%) in the incidence of acute kidney injury in patients undergoing orthopaedic procedures (excluding patients that had surgery for fractured neck of femur, in whom co-amoxiclav was given as prophylaxis) [78]. This increased incidence of acute kidney injury was also noted to have a higher mortality rate within the first year of surgery (20.8% vs 8.2%) [78]. The authors also noted C. difficile rates fell in all patients including those undergoing repairs for a fracture neck of femur (who received co-amoxiclav); therefore, the authors suggested that factors other than the prophylaxis might be driving the development of C. difficile infections [78]. Consequently, such findings led to a change back to the original prescribing policies [78].

Overall, the current literature including general expert consensus and guidelines still recommends a first- or second-generation cephalosporin such as cefazolin and cefuroxime , respectively, to cover the most common pathogens [55]. Current recommendations regarding vancomycin therapy suggest its use should be reserved for patients with documented immediate hypersensitivity to beta-lactams and those with known methicillin-resistant S. aureus colonisation [55].

One group has investigated intraosseous administration of vancomycin in a small (n = 30), poor-quality randomised controlled trial that was not powered to assess clinical outcomes [79]. There are limited data to support alternative routes of administration, and intravenous administration of surgical antimicrobial prophylaxis remains the standard method of administration [55].