Prophylactic antibiotics in elective hip and knee arthroplasty

Current pathogens in hip and knee arthroplasty infections in England

We analysed 189 858 elective primary hip and knee arthroplasty procedures and 1116 inpatient or re-admission SSIs submitted by 184 NHS hospitals (representing 142 NHS Trusts and ten independent NHS treatment centres) to the Public Health England (PHE) national SSI database between April 2010 and March 2013. As the survey on surgical antibiotic prophylaxis was carried out in 2013, the PHE organism data available at the time were for April 2012 to April 2013. The dataset was therefore expanded to include data from the previous two years in order to increase the sample size. Although mandatory orthopaedic data were collected from April 2004, the inclusion of historical data that predated various national policies on healthcare-associated infections would have introduced bias and over-estimation of the burden of S. aureus. Participating hospitals follow a standard protocol of internationally-recognised case definitions for superficial, deep, and organ-space SSIs. Hospitals also undertake systematic prospective follow-up for the capture of cases.^20-22 The standard follow-up period is 30 days for superficial SSIs and up to one year (365 days) for deep and/or organ-space SSIs.

Since April 2004, all NHS Trusts in England have been required to undertake mandatory surveillance in orthopaedic surgery. The four orthopaedic categories are hip arthroplasty, knee arthroplasty, repair of the neck of the femur, and reduction of long-bone fracture. PHE manages the SSI surveillance programme and publishes the rates of SSI by an NHS Trust on an annual basis for the orthopaedic modules. Since public reporting of orthopaedic SSI is at Trust level, the minimum requirement is participation by one hospital site for at least one surveillance quarter in one of the four mandatory orthopaedic categories.

Data are submitted via the PHE secure web-based portal. All data are checked for errors using an inbuilt automated validation system. For example inconsistencies in date values are identified and flagged to the user. SSIs with insufficient SSI criteria entered or superficial SSIs detected beyond 30 days are disallowed to avoid over-reporting of SSIs that do not meet the standard case definitions. Reporting on causative micro-organisms is optional, but must be based on clinically-significant isolates.

The SSIs included in this analysis were those detected during the inpatient stay, or on re-admission following initial hospital discharge, as these methods of detection are a requirement for all participating hospitals. Other forms of post-discharge surveillance (patient wound healing questionnaires or follow-up through review or outpatient clinics) are optional and used inconsistently, and thus were excluded from this study.

We analysed superficial, deep, and organ-space SSI isolates together then conducted a separate subgroup analysis restricted to deep and organ-space SSI isolates. All English NHS Trusts participated in this mandatory orthopaedic surveillance in 2010/2011, three failed to do so in 2011/2012, and two in 2012/2013.²²

The proportion of participating hospitals undertaking continuous surveillance (all four quarters) increased year on year, from 51% in 2010/2011 to 56% in 2012/2013 for hip prosthesis and from 52% to 55% for knee prosthesis over the same time period.²²

The four PHE ‘super regions’ used were London, Midlands and the East of England, the North of England, and the South of England.

Data on primary indications involving trauma/fracture or revision surgery (including revisions for aseptic loosening) were excluded from this analysis. Other indications for surgery were included (avascular necrosis, inflammatory joint disease, osteoarthritis, revision, and other).

Current regimens for prophylaxis in England

As a separate initiative from PHE’s routine surveillance activities, all 144 acute hospital Trusts performing primary hip and knee arthroplasty in England were contacted. Responses were received from 100% of these Trusts. Information governance leads at each Trust were emailed a standard questionnaire in October 2013 and those that did not respond within 30 days were contacted by telephone. Telephone calls were to the duty microbiologist, antibiotic pharmacist, medicines information line, or orthopaedic junior doctor on call. In all cases, sources were asked for information from their hospital’s antibiotic policy. Where the individual contacted was not aware of the local policy, another individual from the list of suitable contacts was contacted.

Contacts were asked ‘Which antibiotics are given prophylactically to patients undergoing elective primary THA or TKA at induction and/or post-operatively? What modifications are made for patients with a serious allergy to penicillin or a history of MRSA infection/colonisation? What doses are administered? What are the dosing intervals? Are there any special circumstances specified in their guidelines, e.g. repeated doses for prolonged surgery or excessive loss of blood?’.

Non-normally distributed continuous data were described using medians with interquartile ranges. Fisher’s exact test was used to compare differences in categorical outcomes between groups as there were small numbers in some cells. Statistical analyses were performed using Stata 13.0 (StataCorp, College Station, Texas) and p < 0.05 (two-tailed) was adopted as the threshold for significance.

Patient characteristics of primary elective hip or knee procedures

The distribution of the patients’ age, gender and ASA score distribution were broadly similar across the two modules. Patients aged 65 to 74 years accounted for the largest proportion of procedures. Within this group, there were more patients receiving a knee prosthesis, compared with those receiving a hip prosthesis (38.1% and 34.4%, respectively). Patients aged < 45 years accounted for the smallest proportion across both populations, however, it was slightly higher in the THA group (4% and 1%, respectively). Female patients accounted for a slightly higher proportion in the THA than the TKA group (60.2% and 58.0%, respectively). Patients with ASA score of 3 or more were similar across the hip and knee modules (19.4% and 20.3%, respectively).

The median time from procedure date to onset of SSI was based on monomicrobial SSIs. Overall the median time to onset of SSI was 18 days (interquartile range (IQR) 11 to 29, minimum and maximum 1 to 363). The median time to onset of S. aureus SSIs was 19 days (IQR 12 to 29, 1 to 362); 16 days for CoNS SSIs (IQR 11 to 27, 1 to 345) and 17 days for Enterobacteriaceae SSIs (IQR 11 to 25, 2 to 267).

Pathogens reported to cause hip and knee arthroplasty infections in England

There were 1116 inpatient/re-admission SSIs, of which 73.3% (n = 818) included data on causative micro-organisms (Table I). 73.8% (n = 604) of these SSIs had a monomicrobial aetiology (n = 604) and 26.2% (n = 214) were polymicrobial. SSIs with organism data yielded a total of 1083 isolates and, of these isolates, 69.1% (n = 748) related to deep and/or organ-space SSIs.

MSSA was the predominant pathogen across England, accounting for 27.0% of isolates (n = 291) followed by coagulase-negative staphylococci (CoNS) at 25.5% (n = 276). MRSA accounted for 4.2% (n = 45) of total isolates. The seven most common causative organisms accounted for 89% of all SSI isolates following THAs and TKAs across the four PHE super regions.

At regional level, staphylococci (MSSA, MRSA, and CoNS) accounted for 57% of isolates with a similar distribution across the PHE super regions. The burden of MRSA was, however, significantly higher in the Midlands and East of England compared with the other three regions (7.4% vs 2.8%; Fisher’s exact test; p = 0.001).

The burden of Pseudomonas spp. was significantly higher in London compared with the three regions combined (8.9% vs 3.2%; Fisher’s exact test: p = 0.005).

Sub-group analysis limited to deep and organ space SSIs (n = 748) found that CoNS were the predominant pathogens and accounted for 27% of isolates (n = 209) followed by MSSA at 25% (n = 184). MRSA accounted for 3.3% of these isolates. Overall, staphylococci accounted for 56% of isolates in this analysis. Stratified analyses by PHE super region showed that the burden of MRSA was also significantly higher in the Midlands and East of England region compared with the other three regions combined (6.3% vs 2.2%; Fisher’s exact test: p = 0.010). The burden of Pseudomonas spp. was also significantly higher in London compared with other three regions combined (8.2% vs 2.8%; Fisher’s exact test: p = 0.028).

The reasons for the regional differences in MRSA and Pseudomonas spp. are not entirely clear. However, laboratory data reported to PHE’s voluntary surveillance system (LabBase2) from 2010 to 2013 shows that the rate of infections in the bloodstream due to Pseudomonas spp. was consistently higher in London compared with the other three regions over his period even with the declining trend.²³ The corresponding analysis for MRSA was not available although a separate report showing trends by smaller geographical units called Area Teams (ATs) showed that ATs within the Midlands and East England region did not show the highest rates of bloodstream infections due to MRSA than other ATs.²⁴ The MRSA result from the SSI programme is perplexing and needs further study.

Routine prophylaxis

The three most common antibiotics or antibiotic combinations made up 126/146 (87%) of observed practice. Flucloxacillin in combination with gentamicin was the most common regimen, with 57/146 (39%) of Trusts using it as their preferred regimen. Cefuroxime was used as the preferred regimen by 44/146 (30%), with teicoplanin plus gentamicin being the third most popular 25/146 (17%). There were ten further preferred regimens used by the remaining 20 Trusts.

Figure 1 illustrates the range of routine prophylactic antibiotic regimens used throughout England. Two Trusts employed two different regimens in this category of patient based on age. Therefore the denominator is 146.

Fig. 1

Graph showing the prophylactic antibiotic regimens in general use for patients undergoing hip and knee arthroplasty.

Prophylaxis in patients with penicillin allergy

The two most common antibiotic/antibiotic combinations made up 128/145(88%) of observed practice. Teicoplanin in combination with gentamicin was the most common regimen, with 90/145 (62%) of Trusts using it as their preferred regimen. Teicoplanin alone was used as the preferred regimen by 36/145 (26%). There were 12 further preferred regimens used by the remaining 20 Trusts.

Figure 2 illustrates the spread of prophylactic antibiotic regimens employed throughout England for patients who are allergic to penicillin. One Trust employed two different regimens in this category. Therefore the denominator is 145.

Fig. 2

Graph showing prophylactic antibiotic regimens in use for patients who are allergic to penicillin

Prophylaxis in patient with a high risk of developing MRSA SSI infection

The two most common antibiotic/antibiotic combinations made up 123/146 (84%) of observed practice. Teicoplanin, in combination with gentamicin, was the most common, with 87/146 (60%) of Trusts using it as their preferred regimen. Teicoplanin alone was used as the preferred regimen by 34/146 (24%). There were 11 further preferred regimens used by the remaining 20 Trusts.

Figure 3 illustrates the distribution of prophylactic antibiotic regimens employed throughout England for patients at high risk of developing MRSA SSI infection. Two Trusts employed two different regimens in this category, therefore the denominator is 146.

Fig. 3

Graph showing prophylactic antibiotic regimens in use for patients at a high risk of MRSA colonisation.

Current evidence

The evidence for different antibiotic regimens as prophylaxis for wound infections following joint arthroplasty surgery was last reviewed in 2005. This systematic review did not find any statistically significant difference in rates of infection when comparing cephalosporins with teicoplanin, cephalosporins with penicillin derivatives, or first-generation with second-generation cephalosporins. However, this review was based on poor-quality studies with variable follow-up and unsatisfactory definitions of infection.¹⁵

Dose and duration of therapy: cefuroxime

There is strong evidence for the use of 1.5 g cefuroxime at induction, however, two randomised controlled trials examining the effectiveness of post-operative doses of cefuroxime found no statistically significant difference in the prevention of SSIs.^29,30

Dose and duration of therapy: flucloxacillin

Use of a single prophylactic dose of flucloxacillin (1g) is supported by one RCT in which it compared favourably with cefazolin in clean, semi-elective orthopaedic surgery involving the implantation of metal work.³¹

Dose and duration of therapy: gentamicin

There is no evidence for the use of systemic gentamicin as prophylaxis in primary elective THA and TKA surgery.

Dose and duration of therapy: teicoplanin

Four randomised controlled trials provide strong evidence for the use of a single dose of 400 mg of teicoplanin at induction.^32-35 Although there is no evidence to suggest that higher doses or prolonged courses of treatment result in fewer SSIs, studies have shown that this dose may be inadequate for patients weighing over 70 kg.³⁶

Complication profiles: cefuroxime

Although there is strong evidence for an association between cefuroxime and CDI in elderly inpatient populations and trauma patients receiving implantation of metal work, studies have not shown any association in the elective orthopaedic setting.^18,37-39 Despite this, our analysis of PHE data showed that 25.5% of SSI isolates were reportedly due to CoNS and 4.2% to MRSA. Cefuroxime is ineffective against MRSA, and may not be effective against CoNS. Additional arguments against the continued use of cefuroxime include its lack of activity against enterococci and Pseudomonas spp, and the increasing number of infections caused by extended spectrum beta-lactamase-producing organisms.

Complication profiles: flucloxacillin with gentamicin

There has been an increase in the percentage of Trusts using flucloxacillin in combination with gentamicin – from 1.3% in 2005 to 38.4% in 2013.⁴⁰The efficacy of gentamicin depends on local strains and sensitivities, but it is usually active against Enterobacteriaceae, Pseudomonas spp. and MRSA in the United Kingdom, although rates of resistance are increasing.

One large, good quality study showed that a single dose of gentamicin caused a significant increase in the number of patients suffering from transient acute kidney injury (AKI).¹⁸ Another non-randomised study found an association between combined high-dose flucloxacillin with single-dose gentamicin and renal impairment; including three patients that subsequently required short-term haemodialysis.⁴¹This was supported by a prospective study in 2013, which showed a highly significant increase in patients suffering AKI – from 1.7% with cefuroxime to 9.5% with combined flucloxacillin and gentamicin.⁴¹

Complication profiles: teicoplanin with gentamicin

Use of teicoplanin alone is not associated with significant complications, although it may cause AKI when combined with gentamicin. Advantages of teicoplainin over vancomycin include a reduced risk of nephrotoxicity and a quicker speed of pre-operative intravenous administration, despite increased time for reconstitution. It is administered as a five-minute intravenous bolus, rather than a one-hour infusion. Teicoplanin is highly active against both MRSA and MSSA, although resistance is increasing.²⁸ This regime is also useful in those who are allergic to penicillin.

The limitations of this study include the absence of data on the use of antibiotic impregnated cement, and that microbial profiles provided by super regions may not directly correspond with causative organism distribution at Trust level, due to variation between Trusts in each region. Reporting on micro-organism is optional, and may have been a potential source of bias. However, the microbial aetiology among deep-seated SSIs was similar to the overall analysis, and so the possibility of bias in the type of organism being reported does not seem convincing. In addition, there is no guarantee that the antibiotics administered as prophylaxis always comply with Trust protocols.

The use of an ecological analysis to examine the correlation between interventions and outcomes at population level (prophylaxis choice and organism distribution) does not establish cause and effect. However, it has been useful in defining a problem in public health (variation in practice) for future investigation.

In conclusion, this survey outlines current practice with regard to antibiotic prophylaxis for elective primary TKA and THA in England. It reveals a disparity in the choice of antibiotic(s), duration of therapy and interpretation of the available evidence in the literature without clear justification. A number of regimens currently in use do not appear to take account of the most recent evidence, and could potentially result in avoidable complications and adverse events. The median time to onset of SSI suggests that a prophylaxis regimen is a relevant factor, hence practice changes remain indicated. Extremely late onset of SSI occurs much less frequently and, for these patients, surgical antibiotic prophylaxis practices may be less relevant, although this should not preclude optimisation of peri-operative practices.

Despite the lack of high-level evidence proving one antibiotic superior to others, there have been efforts in North America to establish a consensus. In Canada, a 2009 survey of antibiotic prophylaxis for total joint arthroplasty (TJA) surgery showed that 97.3% of surgeons surveyed routinely administered cefazolin as their first line prophylaxis.⁴² In the United States, the American Academy of Orthopaedic Surgeons recommend that cefazolin or cefuroxime are the preferred intravenous antibiotics to be used as prophylaxis in primary TJA.⁴³ With rates of SSI reported as approximately 1% in the United States and Canada, their results are comparable with the United Kingdom and Europe. It is therefore difficult to understand why such variation exists in the United Kingdom. One possibility is that antibiotic choice is influenced by organisational aversion to certain antibiotics, e.g., following high-profile CDI outbreaks. For example, those that are averse to cephalosporins may be more likely to use the dual combination flucloxacillin and gentamicin. The 2014 English surveillance programme for antimicrobial utilisation and resistance (ESPAUR) report includes a national survey on antimicrobial stewardship in secondary care in 2014. A total of 99 (67.8%) of 146 contacted acute NHS Trusts responded to the survey. Of those, 98% of Trusts reported the use of surgical antibiotic prophylaxis policy. However, just over 80% of Trusts implemented audits of compliance to antibiotic guidelines (dose, route and duration). Although this is high further improvement in the process, monitoring is needed as this activity is key in order to influence changes in practice. The report also found that 24% of survey respondents had a written antimicrobial education and training strategy.²³ These results may, in part, explain the variation in surgical prophylaxis that we observed in our study. Another explanation that could account for some of the variation in regimens is that it could reflect local analyses of infecting organisms. There is, however, no obvious reason why the United Kingdom should not aim to establish a consensus in the same way as has been achieved in North America.

It is arguable that the financial burden, morbidity, and mortality associated with SSIs following THA and TKA are sufficient to support efforts to undertake further research in this field. This research should focus on establishing the safest regimen/dose, by comparing complications such as SSI, AKI, CDI rates, and rates of MRSA infection. We would suggest that the National Joint Registry collect data on antibiotic(s) used including dose, route, duration, and timings. This data could then be combined with that already collected by PHE including rates of SSI, CDI and MRSA.