Prevalence of MRSA in Emergency and Elective Patients Admitted to a Vascular Surgical Unit: Implications for Antibiotic Prophylaxis

Transcription

1 Eur J Vasc Endovasc Surg 32, 402e407 (2006) doi: /j.ejvs , available online at on Prevalence of MRSA in Emergency and Elective Patients Admitted to a Vascular Surgical Unit: Implications for Antibiotic Prophylaxis B. Muralidhar, 1 * S.M. Anwar, 1 A.I. Handa, 1 T.E.A. Peto 2 and I.C.J.W. Bowler 3 1 Nuffield Department of Surgery, 2 Departments of Infectious Diseases, and 3 Microbiology, John Radcliffe Hospital, Oxford, OX3 9DU, UK Objectives. 1. Audit adequacy of admission screening for MRSA in vascular surgery patients. 2. Establish the prevalence of MRSA carriage at the time of admission in emergency/transfer and elective patients. 3. Establish a threshold prevalence of MRSA that should trigger the use of prophylactic antibiotics active against MRSA. 4. Model some of the costs and efficacy of glycopeptides such as vancomycin, compared to aminoglycosides such as gentamicin, for the prevention of MRSA surgical site infections. Materials and Methods. 200 consecutive emergency/transfer and 150 consecutive elective patients admitted between April 2004 and January 2005, were studied. Data was obtained from departmental Morbidity and Mortality records and the computerised laboratory medicine information system. Results. 261 (75%) of the 350 patients were screened for MRSA on admission (target 100%). The proportions of emergency/transfer and elective patients screened were similar (78% and 72% respectively). The prevalence of MRSA carriage detected by admission screening in emergency/transfer patients 30/153 (20%), was significantly higher (p < ) than in elective patients 2/108 (2%). A simple decision analysis model suggests that gentamicin should be used when the prevalence of MRSA reaches 10% and vancomycin when the prevalence reaches 50%. Conclusions. The high prevalence of MRSA colonisation in emergency/transfer patients has important implications for pre-operative antibiotic prophylaxis. Keywords: MRSA screening; Antibiotic prophylaxis; Vascular surgery. Introduction Surgical site infection is a rare but devastating event for the vascular patient. This may involve the wound or a graft or both. Graft infection is particularly serious occurring at a frequency of 1e6%. 1,2 Treatment usually requires re-operation, prolonged hospital stay and long term antibiotic therapy. Even with such interventions morbidity and mortality remain high. Approximately 30% of patients with aortic graft infection die and 20% lose a limb. 3 Around 17% of patients with femoro-popliteal graft infection die and up to 50% require amputation. 3 *Corresponding author. Balaji Muralidhar, BM, Bch, MA (Oxon), Hutchison/MRC Cancer Cell Unit, Hills Road, Cambridge, CB2 2XZ, UK. addresses: bm239@cam.ac.uk, balim@btinternet.com Methicillin Resistant Staphylococcus aureus (MRSA) has become a major hospital-acquired pathogen in many countries including the UK over the past 15 years. 4,5 In addition, it is now recognised as a community pathogen, 6,7 with MRSA accounting for 24% of community acquired Staphylococcus aureus bacteraemia in the catchment of our hospital. 8 Most surgical site infections after vascular procedures are caused by Staphylococcus aureus, although graft infections may involve Coagulase Negative Staphylococci. 9 Recent audits in the UK show that up to 55% of deep wound and graft infections are caused by Staphylococcus aureus and that 70% of isolates from surgical sites are MRSA. 9,10 Infection with MRSA is associated with a poor outcome in vascular surgery patients, with higher rates of amputation after arterial reconstruction in the leg as compared with methicillin-sensitive Staphylococcus aureus (MSSA) infection. 9 MRSA aortic / $35.00/0 Ó 2006 Elsevier Ltd. All rights reserved.

2 Prevalence of MRSA in Emergency and Elective Patients Admitted to a Vascular Surgical Unit 403 graft infection is associated with very high case fatality rates. 11 Because of the serious consequences of MRSA infection in vascular surgery patients, screening for the organism prior to or on admission is recommended. 11,12 Although results of admission screening only become available 3 days or more following admission, they provide useful information to guide empirical therapy should the patient develop a post operative infection, and to reinforce procedures such as isolation or cohort nursing designed to prevent spread of MRSA to other patients. 13 This information can also be used, when aggregated at the population level, to ensure that prophylactic antibiotic regimens are appropriately targeted. 11 The cost effectiveness of this strategy has been demonstrated in both the US and European healthcare systems. 14 In our unit it is policy to screen all vascular patients for MRSA within 48 hours of admission. Antibiotic prophylaxis is recommended for all vascular surgical procedures. 15 In the UK, b-lactams such as a first or second generation cephalosporin or coamoxiclav are often used, with aminoglycosides like gentamicin employed as an alternative for patients allergic to b-lactams. 9,15 Prophylaxis for the prevention of surgical site infections caused by MRSA, which is inherently resistant to all b-lactams, is not universal, though many specialists advise the use of glycopeptides (vancomycin or teicoplanin) in patients at high risk of MRSA associated complications. 9,11,15e17 High risk situations include surgery in patients known to have MRSA; re-operation for graft infection; reexploration of sites adjacent to prosthetic grafts and units with high baseline rates of MRSA. 11,18 The frequency of MRSA carriage in patients admitted for vascular procedures is not well documented, and is likely to change over time and place. 13 Recently a prevalence of >10% MRSA carriage in the population receiving prophylaxis has been suggested as a trigger for the use of glycopeptides. 19 However the widespread use of glycopeptides has several disadvantages. Excessive use is linked to the emergence of vancomycin-resistant Enterococci and Staphylococcus species for which there are few therapeutic options. 19,20 Vancomycin needs to be infused over an hour and hence must be started on the ward prior to theatre. If rapidly infused it may cause hypotension and histamine release which may have adverse effects during anaesthesia. 21,22 Teicoplanin is an alternative but is very costly (British National Formulary (BNF) cost for 400 mg compared to for vancomycin 1 g). These issues have led to calls for a reexamination of the role of aminoglycosides such as gentamicin for MRSA prophylaxis. 19 Gentamicin is usually used for its potent activity against Gram negative organisms such as E. coli and Pseudomonas aeruginosa, with typical minimum inhibitory concentrations (MICs) of 1e4 mg/l and 1e8 mg/l respectively. However it is also highly active against most strains of Staphylococcus aureus including common UK strains of MRSA (MIC 0.12e1 mg/l). 23 A bolus of 1.5 mg/kg at induction gives a peak serum concentration of 5e10 mg/l, exceeding the MIC for MRSA and MSSA by 5e10 fold. 23 Gentamicin resistance in MRSA is a well described but minor problem and currently 91% of MRSA strains in the UK are susceptible. 24 Gentamicin can be given by bolus at induction, intra operative top ups are not required because of its longer half-life compared to cefuroxime, 25 and it is cheap (BNF cost 4.20 for 120 mg). We therefore decided: firstly to audit the adequacy of admission screening for MRSA. Secondly to establish the prevalence of MRSA carriage at the time of admission in elective and emergency/transfer patients. Thirdly to establish a threshold prevalence of MRSA which should trigger the use of prophylactic antibiotics active against MRSA, by modelling the expected burden of MRSA surgical site infections when using b-lactam antibiotics such as cefuroxime, which are not active against MRSA, for prophylaxis. Lastly, to model the efficacy of glycopeptides such as vancomycin, compared to aminoglycosides such as gentamicin, for the prevention of MRSA surgical site infections. Materials and Methods Records of all vascular patients were obtained retrospectively from departmental Morbidity and Mortality meeting data sheets, which log all vascular patients admitted to the John Radcliffe Hospital. Between April 2004 and January 2005, 150 consecutive elective admissions (admitted from the community for a planned surgical intervention), and 200 emergency/transfer admissions (patients admitted via the Emergency Department or transferred from neighbouring regional hospitals) were included in our study. Differences in prevalence of MRSA carriage between the two groups of patients were compared using the chi squared test (STATA 8.2 software). Our departmental policy was to screen all vascular patients admitted to our unit by culturing nose, throat, axillae, perineum and wound swabs for MRSA. Culture included enrichment as recommended by the English Health Protection Agency. 26 If the patient was catheterised a urine sample was also

3 404 B. Muralidhar et al. examined. We analysed screening swabs, which were taken within 48 hours of admission. Patients whose admission screen was negative and where MRSA was cultured from any specimen taken to investigate clinical infection after 48 hours of hospital stay were classed as acquisitions (i.e. having acquired MRSA in hospital). This 48-hour cut off is widely used by epidemiologists to define hospital-acquired infection. 27 The MRSA admission screening status and the results of all subsequent microbiology tests performed was obtained from the computerised laboratory medicine information system. A simple decision analysis was used to calculate the number of MRSA wound infections expected at a given prevalence of MRSA in the two groups undergoing surgery 28 : Assumptions underlying the decision analysis The proportion of vascular wound infections caused by Staphylococcus aureus is 75%. 1,9 The wound infection rate when no effective antibiotic prophylaxis is given is 15%. 16,29 The relative risk reduction in rate of post operative wound infection with effective antibiotic prophylaxis is 66% (15% / 5%) 16,29e33 Vancomycin is 100% effective against MRSA in vitro. 19 Gentamicin is at least 80% effective against MRSA strains in the UK in vitro. 24 is ineffective against MRSA in vitro. 19 Model of efficacy of antibiotic prophylaxis in prevention of wound infection in vascular surgery Let the proportion of wound infections caused by Staphylococcus aureus ¼ W Let risk of wound infection with inappropriate or no antibiotic prophylaxis ¼ I Let the relative reduction in risk of wound infection with appropriate prophylaxis ¼ R Let the prevalence of MRSA in the population receiving prophylaxis ¼ P Let the proportion of MRSA strains susceptible to the prophylaxis ¼ S Therefore, the reduction in the risk of wound infection caused by MRSA for any antibiotic prophylactic regimen ¼ WIRPS Therefore, the number needed to treat (NNT) ¼ 1/WIRPS See Appendix for worked example. The model was used to examine some of the costs and effectiveness of three different prophylactic antibiotic regimens (cefuroxime þ vancomycin, cefuroxime þ gentamicin, and vancomycin þ gentamicin) for the prevention of post-operative wound infection caused by MRSA. Cost of antibiotics was calculated using the British National Formulary (BNF). Results Elective admissions Of the 150 patients (M:F 2.8:1, median age 67, range 54e83) in this group, 108 (72%) had evidence of appropriate MRSA admission screening. Only 2 (2%) patients of the 108 screened were colonised with MRSA. No elective patients acquired MRSA during their hospital stay. Emergency/transfer admissions Of 150 emergency patients (M:F 2.2:1, median age 64, range 36e91), 119 (79%) were screened on admission, and of these 26 (22%) were colonised with MRSA. Of 50 patients (M:F 2.3:1, median age 63, range 39e76) transferred from other hospitals, 34 (68%) were screened on admission, and of these 4 (12%) were colonised with MRSA. Because the number of transfer patients was small, their results were combined with the emergency patients. Of the 200 emergency/ transfer patients 153 (77%) were screened for MRSA on admission. and 30 (20%) were colonised with MRSA. Seven (6%) of the 123 patients who screened negative for MRSA on admission, acquired the organism during their stay. The prevalence of MRSA assessed on admission in emergency/transfer patients (30/153 20%) was significantly higher than in elective patients (2/108 2%) (Odds Ratio 12.9 CI 3.1e113.4 c p < ). Using the decision analysis model the numberneeded-to-treat (NNT) to prevent a single MRSA surgical site infection for the two groups using a variety of prophylactic antibiotic regimens was calculated (Table 1, Fig. 1). In the pre MRSA era the prevalence of MSSA carriage in the general population was 10e20%. 34 Using cefuroxime (with 100% activity against MSSA) the NNT to prevent a single MSSA wound infection in vascular surgery was therefore in the order of 135 to 67 (Table 1, Fig. 1). This range of NNTs can be regarded as the gold standard for prophylaxis in the pre MRSA era.

4 Prevalence of MRSA in Emergency and Elective Patients Admitted to a Vascular Surgical Unit 405 Table 1. Table showing number of patients needed to treat (NNT) to prevent a single MRSA wound infection for 3 different prophylactic antibiotic regimens, according to the prevalence of MRSA in the target population. See Appendix for worked example % MRSA in population Vancomycin added to Gentamicin added to Vancomycin added to Gentamicin When using cefuroxime alone, the absolute risk of developing an MRSA surgical site infection in an elective patient in our unit, with carriage rates of 2%, is calculated to be ( see Appendix) or 1 infection for every 444 elective patients undergoing surgery. For our emergency/ transfer patients with a carriage rate of 20%, the risk is tenfold higher, or 1 in 44. This corresponds to a potential burden of 6 MRSA surgical site infections in the 265 emergency/transfer patients admitted to our unit each year, at a cost of 12,225 assuming a cost of 2,017 per infection. 35 The NNT for vancomycin prophylaxis to prevent a single MRSA surgical site infection in the elective group is 673, and in the transfer/emergency group 67 (Table 1, Fig. 1). Since the NNT for cefuroxime used in populations carrying MSSA is in the order of 135 to 67, the use of vancomycin would be justified for our emergency/transfer patients but not those undergoing elective surgery. At a prevalence of 20%, 1,166 ( ) spent on vancomycin in emergency/transfer patients will prevent one MRSA surgical site infection costing 2,017. In elective patients, where the MRSA prevalence is only 2%, the cost of vancomycin would be 11, 656 ( ) for each infection prevented. Given the ecological and financial costs, and the logistical problems of glycopeptide prophylaxis we used the model to examine the efficacy of gentamicin in our patients. We assumed gentamicin was active against 80% of MRSA, a conservative estimate given that in our unit over the study period it was active against 94% of MRSA cultured from admission screens. The NNT for gentamicin to prevent a single MRSA surgical site infection in our elective group is 842, and 84 in the transfer/emergency group (see Table 1). Gentamicin is therefore an attractive option for MRSA prophylaxis in emergency/transfer patients, allowing glycopeptides to be targeted to those patients known to be MRSA colonised at the time surgery, where the NNT for vancomycin is 13 (Table 1 MRSA prevalence 100%). In units using gentamicin in this way, the model suggests there is little benefit to using glycopeptide prophylaxis other than for patients known to be colonised with MRSA at surgery. For example, in our emergency/transfer patients, the NNT to prevent an MRSA surgical site infection when adding vancomycin to a prophylactic regimen that includes gentamicin is estimated to be 337, greater than the gold standard NNT of the pre MRSA era of 135 for cefuroxime used against MSSA strains. However when the prevalence of MRSA colonisation reaches 50%, using vancomycin for all patients may be beneficial, as the NNT for its use in addition to gentamicin is 135 (Table 1), equivalent to the gold standard NNT for routine cefuroxime prophylaxis for MSSA. NNT Number Needed to Treat (NNT) to prevent one MRSA wound infection Prevalence of MRSA (%) Vancomycin added to Gentamicin added to Vancomycin added to Gentamicin Fig. 1. Graph showing number of patients needed to treat (NNT) to prevent a single MRSA wound infection for 3 different prophylactic antibiotic regimens, according to the prevalence of MRSA in the target population. Discussion Approximately 3 out of 4 patients admitted to our unit are screened for MRSA in accordance with our unit policy. We found no evidence that compliance with admission screening was influenced by staff perception of risk of MRSA carriage in the two populations studied. The proportion screened in each group was similar: elective 108/150 (72%) and emergency/transfer 153/200 (78%), rates similar to those recently reported in two adult intensive care units in the UK. 36 The overall prevalence of MRSA carriage in patients admitted to our unit and screened was 32/261 (12%), higher than the 4% documented in patients admitted to the Leicester UK vascular unit in

5 406 B. Muralidhar et al. Rates are likely to vary from region to region and with time, since the MRSA epidemic in the UK and Europe is still evolving, 37 so it is important for each unit to collect local data to inform decisions about antibiotic prophylaxis. We found the prevalence of MRSA in emergency/transfer patients (20%) to be significantly higher than elective patients (2%) p < We did not collect risk factor data but our emergency/transfer vascular patients often have critical ischaemia with skin ulceration and a history of prior hospital admission which are risk factors for MRSA colonisation. 11 The majority of our elective patients do not have these risk factors. Typically they include patients with claudication (not rest pain), abdominal aortic aneurysm and carotid stenosis. It is noteworthy that elective patients, who would be available for pre admission screening, have a low MRSA carriage rate. The emergency/transfer patients, who by dint of their condition are not available for pre admission screening, have a high rate. Decision analysis has been used to examine the cost and efficacy of antibiotic prophylaxis in a neurosurgical unit where the prevalence of MRSA in patients transferred into the unit was found to be 15%. 28 A combination of cefuroxime and gentamicin was found to be satisfactory for these patients, with vancomycin reserved for patients known to be colonised or infected with MRSA. 28 Our model is not definitive or comprehensive. The number of variables has been kept small, we did not carry out a sensitivity analysis to explore the relative impact of each variable on the outcome (NNT), and we did not look at all possible costs and benefits. However, in practice such sophistication is not needed since the aim is to guide rather than determine policy. Antibiotic prophylactic regimens in our unit were changed on the basis of the data and model presented here. The 10% threshold prevalence of MRSA for modification of prophylaxis to include gentamicin, suggested by our model, is in accordance with the 10% threshold recently advised by a working part of the British Society of Antimicrobial Chemotherapy. 19 MRSA carriage is uncommon in our elective vascular surgery patients (2%), and the yield of pre admission screening recommended by some 12 would be correspondingly low. Our model shows that b-lactam prophylaxis for this patient group is likely to be safe. MRSA carriage is common (20%) in vascular surgery patients admitted as an emergency or transferred from another hospital to our unit. The MRSA status of these patients cannot reliably be determined before admission, and since the results of admission MRSA screening are not available for 3 or more days, they usually undergo surgery when their MRSA status is unknown. Our model suggests that using b-lactam antibiotics, such as cefuroxime, for prophylaxis in this group would lead to an unacceptable burden of MRSA surgical site infection. Aminoglycosides such as gentamicin are an attractive option for these patients and other populations where the prevalence of MRSA is between 10 and <50%, allowing glycopeptides to be targeted to those known to be colonised with MRSA. This is in keeping with the 10% prevalence threshold recently suggested by a working party of the British Society of Antimicrobial Chemotherapy. 19 When the prevalence of MRSA reaches 50%, vancomycin is likely to be a safer option (NNT 135). However our model is a simple one and needs to be validated by formal cost benefit analysis. The prevalence of MRSA colonisation in patients admitted to hospitals in the UK is increasing and probably varies geographically. 37 Vascular teams should therefore conduct their own local studies of MRSA prevalence to ensure their prophylactic antibiotic regimens are targeted appropriately. We have shown how decision analysis can be used to achieve this targeting once the prevalence data is known. There is a pressing need to validate such models with randomised controlled trials comparing aminoglycosides and glycopeptides for prophylaxis in vascular patients with high rates of MRSA carriage. Appendix. Decision analysis used to determine risk of wound infection and NNTs with different prophylactic antibiotic regimens. Worked Example NNT calculation for efficacy of Gentamicin prophylaxis to prevent a single MRSA wound infection in a vascular surgery patient population where the MRSA carriage rate is 20%: Proportion of wound infections caused by Staphylococcus aureus ¼ W [ 0.75 Risk of wound infection with inappropriate or no antibiotic prophylaxis ¼ I [ 0.15 Relative reduction in risk of wound infection with appropriate prophylaxis ¼ R [ 0.66 Prevalence of MRSA in the population receiving prophylaxis ¼ P [ 0.2 Proportion of MRSA strains susceptible to the prophylaxis ¼ S [ 0.8 NNT [ 1/WIRPS [ 84

6 Prevalence of MRSA in Emergency and Elective Patients Admitted to a Vascular Surgical Unit 407 Acknowledgements I thank Y.R.K.A. Baki and A.D.C. Pilavaki s for their help and advice with statistical techniques. References 1 BUNT TJ. Synthetic vascular graft infections. I. Graft infections. Surgery 1983;93(6):733e HICKS RC, GREENHALGH RM. The pathogenesis of vascular graft infection. Eur J Vasc Endovasc Surg 1997;14(Suppl. A):5e9. 3 ZDANOWSKI Z, DANIELSSON G, JONUNG T et al. Intraoperative contamination of synthetic vascular grafts. Effect of glove change before graft implantation. A prospective randomised study. Eur J Vasc Endovasc Surg 2000;19(3):283e287. 4VOSS A, MILATOVIC D, WALLRAUCH-SCHWARZ C, ROSDAHL VT, BRAVENY I. Methicillin-resistant Staphylococcus aureus in Europe. Eur J Clin Microbiol Infect Dis 1994;13(1):50e55. 5FOLORUNSO ABA-DF, LYON DJ, CHENG AF. Living with methicillinresistant Staphylococcus aureus: a 7-years experience with MRSA in a university hospital. Infect Control Hosp Epidemiol 2000;18: 765e767. 6ADCOCK PM, PASTOR P, MEDLEY F, PATTERSON JE, MURPHY TV. Methicillin-resistant Staphylococcus aureus in two child care centres. J Infect Dis 1998;178(2):577e580. 7MAGUIRE GP, ARTHUR AD, BOUSTEAD PJ, DWYER B, CURRIE BJ. Clinical experience and outcomes of community-acquired and nosocomial methicillin-resistant Staphylococcus aureus in a northern Australian hospital. J Hosp Infect 1998;38(4):273e281. 8WYLLIE DH, PETO TE, CROOK D. MRSA bacteraemia in patients on arrival in hospital: a cohort study in Oxfordshire 1997e2003. BMJ 2005;331(7523):992. 9NAYLOR AR, HAYES PD, DARKE S. A prospective audit of complex wound and graft infections in Great Britain and Ireland: the emergence of MRSA. Eur J Vasc Endovasc Surg 2001;21(4):289e Surveillance of Surgical Site Infection in English Hospitals. Nosocomial Infection National Surveillance Service (NINSS), 1997e EARNSHAW JJ. Methicillin-resistant Staphylococcus aureus: vascular surgeons should fight back. Eur J Vasc Endovasc Surg 2002;24(4): 283e UK MRSA Guidelines Combined Working Party of the British Society for Antimicrobial Chemotherapy, the Hospital Infection Society and the Infection Control Nurses. J Hosp Infect 1998;39: 253e SCRIVEN JM, SILVA P, SWANN RA et al. The acquisition of methicillinresistant Staphylococcus aureus (MRSA) in vascular patients. Eur J Vasc Endovasc Surg 2003;25(2):147e RUBINOVITCH B, PITTET D. Screening for methicillin-resistant Staphylococcus aureus in the endemic hospital: what have we learned? J Hosp Infect 2001;47(1):9e Antibacterial prophylaxis in surgery: 3 e Arterial surgery in the abdomen, pelvis and lower limbs. Drug Ther Bull 2004;42(6):43e JENSEN LJ, AAGAARD MT, SCHIFTER S. Prophylactic vancomycin versus placebo in arterial prosthetic reconstructions. Thorac Cardiovasc Surg 1985;33(5):300e FRAISE AP. Guidelines for the control of methicillin-resistant Staphylococcus aureus. J Antimicrob Chemother 1998;42(3):287e Recommendations of the Hospital Infection Control Practices Advisory Committee(HICPAC). Recommendations for preventing the spread of vancomycin resistance. MMWR Recomm Rep 1995;44(Suppl. RR-12):1e GEMMELL CG, EDWARDS DI, FRAISE AP, GOULD FK, RIDGEWAY GL, WARREN RE. Guidelines for the prophylaxis and treatment of methicillin-resistant Staphylococcus aureus (MRSA) infections in the UK. J Anntimicrob Chemother 2006 [Advance Access published February 27]. 20 FRAISE AP. The treatment and control of vancomycin resistant enterococci. J Antimicrob Chemother 1996;38(5):753e LEVY JH, KETTLEKAMP N, GOERTZ P, HERMENS J, HIRSHMAN CA. Histamine release by vancomycin: a mechanism for hypotension in man. Anesthesiology 1987;67(1):122e DAJEE H, LAKS H, MILLER J, OREN R. Profound hypotension from rapid vancomycin administration during cardiac operation. J Thorac Cardiovasc Surg 1984;87(1):145e KUCERS A, CROWE SM, GRAYSON ML, HOY JF. The use of antibiotics. 5th ed. Oxford: Butterworth Heinemann, 1997:450e HENWOOD CJ, LIVERMORE DM, JOHNSON AP, JAMES D, WARNER M, GARDINER A. Susceptibility of gram-positive cocci from 25 UK hospitals to antimicrobial agents including linezolid. The Linezolid Study Group. J Antimicrob Chemother 2000;46(6): 931e WENZEL RP. Preoperative antibiotic prophylaxis. N Engl J Med 1992;326(5):337e Investigation of specimens for screening for MRSA. B.SOP 29, 27 GARNER JS, JARVIS JR, EMORI TG, HORAN TC, HUGHES JM. CDC definitions of nosocomial infections. In: OLMSTED RN, ed. AIPC infection control and applied epidemiology: principles and practice. London: Mosby, 1996:A1eA HAMMOND CJ, GILL J, PETO TE, CADOUX-HUDSON TA, BOWLER IC. Investigation of prevalence of MRSA in referrals to neurosurgery: implications for antibiotic prophylaxis. Br J Neurosurg 2002;16(6):550e SALZMAN G. Perioperative infection prophylaxis in vascular surgery e a randomised prospective study. Thorac Cardiovasc Surg 1983;31(4):239e KAISER AB, CLAYSON KR, MULHERIN Jr JL, ROACH AC, ALLEN TR, EDWARDS WH et al. Antibiotic prophylaxis in vascular surgery. Ann Surg 1978;188:283e PITT HA, POSTIER RG, MACGOWAN AW, FRANK LW, SURMAK AJ, SITZMAN JV et al. Prophlactic antibiotics in vascular surgery. Topical, systemic, or both? Ann Surg 1980;192:356e WORNING AM, FRIMODT-MOLLER N, OSTRI P, NILSSON T, HOJHOLDT K, FRIMODT-MOLLER C. Antibiotic prophylaxis in vascular reconstructive surgery: a double-blind placebo-controlled study. J Antimicrob Chemother 1986;17:105e DOEBBELING BN. The epidemiology of methicillin-resistant Staphylococcus aureus colonisation and infection. J Chemother 1995;7(Suppl. 3):99e WENZEL RP, PERL TM. The significance of nasal carriage of Staphylococcus aureus and the incidence of postoperative wound infection. J Hosp Infect 1995;31(1):13e PLOWMAN R, GRAVES N, GRIFFIN M, ROBERTS JA, SWAN AV, COOKSON B et al. The socio e economic burden of hospital acquired infection. Public Health Lab Serv CEPEDA JA, WHITEHOUSE T, COOPER B et al. Isolation of patients in single rooms or cohorts to reduce spread of MRSA in intensivecare units: prospective two-centre study. Lancet 2005;365(9456): 295e JOHNSON AP, PEARSON A, DUCKWORTH G. Surveillance and epidemiology of MRSA bacteraemia in the UK. J Antimicrob Chemother 2005;56(3):455e462. Accepted 26 March 2006 Available online 23 May 2006