Journal of Infection (2008) 56, 163e170

www.elsevierhealth.com/journals/jinf

Is MRSA admission bacteraemia community-acquired? A case control study*

Ruth Miller a, Hanif Esmail c, Tim Peto a, Sarah Walker c,Derrick Crook b, David Wyllie b,*

a Nuffield Department of Medicine, University of Oxford, John Radcliffe Hospital, Oxford, UKb Nuffield Department of Clinical Laboratory Sciences, University of Oxford, John Radcliffe Hospital,Oxford OX3 9DU, UKc Department of Microbiology, University of Oxford, John Radcliffe Hospital, Oxford, UK

Accepted 10 December 2007Available online 1 February 2008

KEYWORDSHealthcare associatedinfection;Bacteraemia;MSSA;MRSA

* An abridged version of this paper h* Corresponding author. Tel.: þ44 18

E-mail address: david.wyllie@ndcl

0163-4453/$30 ª 2007 The British Infedoi:10.1016/j.jinf.2007.12.004

Summary Objectives: To compare characteristics of methicillin resistant Staphylococcusaureus (MRSA) and methicillin susceptible S. aureus (MSSA) bacteraemia detected on admissionto a UK hospital and to determine whether these organisms are community-acquired.Methods: Consecutive cases of MRSA bacteraemia admitted to general medicine between2003 and 2006 were identified and compared to MSSA age-matched and unmatched controls(35, 35 and 34 patients, respectively). Demographics, MRSA risk factors, previous health-carecontact and clinical presentation were compared using patient notes. Multi-locus sequencetyping was performed.Results: 34/35 strains of admission MRSA bacteraemia were the health-care associatedSequence Types (ST)-22 (77%) or ST-36 (21%), whereas 20 different MSSA strains were identi-fied. No MRSA cases fitted the CDC definition of community-acquired MRSA. Compatible withhealth-care associated acquisition, after matching for age MRSA cases had significantly higherlevels of previous hospital exposure than MSSA controls, and more co-morbidities. Notably, 63%of MRSA cases were admitted from their own home, as opposed to secondary care facilities.Clinical presentation of MRSA and MSSA bacteraemias was similar.Conclusions: MRSA strains associated with health-care were responsible for almost all casesof MRSA bacteraemia on admission to hospital during the period studied. Despite this themajority of cases with MRSA admission bacteraemia were admitted from their own homes.Further research is needed into the determinants of MRSA bacteraemia among patients out-side hospital.ª 2007 The British Infection Society. Published by Elsevier Ltd. All rights reserved.

as been submitted as an abstract for a poster to FIS 2007.65 221226; fax: þ44 1865 220524.

s.ox.ac.uk (D. Wyllie).

ction Society. Published by Elsevier Ltd. All rights reserved.

164 R. Miller et al.

Introduction differentiating MRSA and MSSA infection among those pa-

Staphylococcus aureus remains a key human pathogen,1

with diverse clones existing among both methicillin sensi-tive (MSSA) and methicillin resistant (MRSA) organisms.2

A recent change with major clinical implications is the ap-pearance of MRSA clones able to spread in the communitycausing infection among previous healthy individuals with-out hospital contact.3 These community-acquired (CA-MRSA) strains are prevalent in the USA where they presentlargely as skin and soft tissue infections in discrete popula-tions.3,4 Small clonal community-based outbreaks have alsorecently been reported in the UK.5 If circulating in the com-munity, these strains might be most easily identifiablewithin hospital isolates from infections presenting shortlyafter admission. Such admission infections accounted for27% of the total MRSA bacteraemias reported in Englandand Wales in 2006/7,6 a similar proportion to two Oxford-shire hospitals from 1997 to 2003.7

The UK epidemic of health-care associated Multi LocusSequence Type (MLST) Sequence Type (ST)-22 and ST-36MRSA has been ongoing since 1990.8 Whilst we previouslynoted hospital exposure to be common in individuals admit-ted with S. aureus bacteraemia,9 given the increasing prev-alence of CA-MRSA globally it is unclear whether this remainsthe case. Additionally, despite the important contribution ofadmission S. aureus bacteraemia to total S. aureus bacter-aemia, no detailed study has described the demographics,characteristics, clinical presentation and strain diversityamong these MRSA cases in the UK, and differences betweenMRSA and MSSA admission bacteraemia are unclear.

We have therefore conducted a clinical and molecularstudy of S. aureus admission bacteraemias, particularlyconsidering whether the organism was likely to be commu-nity acquired, rather than associated with hospital or otherhealth-care contact.10 In order to exclude any potentialhealth-care associated routes of S. aureus transmission,a sensitive definition of health-care exposure is required.Diverse definitions have been used previously, which com-plicate comparisons;10 here we used the widely acceptedand stringent US Centres for Disease Control (CDC) defini-tion (see Box 1).11 We investigated characteristics

Box 1. CDC definition of CommunityAcquired MRSA

� MRSA diagnosed in outpatient or within 48 h afteradmission to hospital� No medical history of MRSA infection or colonisation� No history in the past year of

e Hospitalisatione Admission to a nursing home, skilled nursing

facility, or hospicee Dialysise Surgery

� No permanent indwelling catheters or medicaldevices that pass through the skin into the body1

tients admitted with S. aureus bacteraemia.

Method

Case and control selection

Individuals were included if they had MRSA (case) or MSSA(control) bacteraemia on admission to the general medicalservice of the Oxford Radcliffe Hospital Trust, Oxfordbetween April 2003 and September 2006. This group waschosen because they are responsible for most admissionMRSA bacteraemias in our centre.7 Admission bacteraemiawas defined as S. aureus isolated from blood cultures takenwithin 48 h of arrival; consistent with the CDC definition ofCA-MRSA11 and UK mandatory surveillance.6 Cases com-prised all consecutive MRSA isolates meeting inclusioncriteria, selected from electronic records. Two groups ofMSSA admission bacteraemia controls were identified. Onewas unmatched and consisted of an equal number of MSSAadmission bacteraemias selected randomly from the sametime period. The second group was age-matched with agewithin 6 years of each MRSA case. Age-matched controlswere re-sampled until every MRSA case had an age-matchedcontrol.

S. aureus characteristics

A structured questionnaire identifying demographics, riskfactors for S. aureus infection,10e15 previous health-carecontact, clinical presentation and medication was com-pleted from case notes by two independent reviewers,resolving disagreements by consensus. Previous hospitaladmissions to the Oxford Radcliffe Hospital Trust (includingthe Horton General Hospital, Banbury) and MRSA isolationwere determined from electronic records from 1997 on-wards. If previous hospital admissions were not recordedelectronically and were also not recorded in patient notes(where admission to other hospitals were also identified),patients were assumed not to have been admitted to anyhospital in the last 3 years, and were censored at thistime point. Four separate measures of health-care contactwere considered; (i) previous hospital exposure, includingtime since most recently in hospital and length of timespent in hospital in the last year. (ii) health care portalsof entry, any of: invasive surgery or intervention in thelast year, concurrent vascular access, current use of a uri-nary catheter, any previous renal dialysis or any previousadmission into an intensive care unit; (iii) co-morbidityscore from 0 to 6, with one point for each of the followingconcurrent conditions: diabetes, heart failure, oral ste-roids, vascular disease, chronic obstructive pulmonary dis-ease (COPD) and chronic renal failure; (iv) the measuresof health-care contact listed in the CDC CA-MRSA definition(Box 1).11

Microbiology

S. aureus isolates were recovered from glycerol stocks.Identity and sensitivity testing was performed using

MRSA admission bacteraemia 165

colonial morphology, tube coagulase, DNAse testing, MRSA-Select� chromogenic agar (Bio-Rad, Limerick, Ireland), andgrowth on oxacillin plates (E and O Laboratories, Bonny-bridge, UK) as described.16 DNA was extracted from singlecolonies and grown overnight in a 5% salt broth (NaCl)(E and O Laboratories, Bonnybridge, UK) at 37 �C using aDNEasy tissue kit (Qiagen, Crawley, UK) as recommendedby the manufacturer.

PCR

All isolates were tested for 16S rRNA, lukS/F-PV and mecAusing a multiplex PCR,17 modified as described (Table 1).A clinical MRSA isolate from our hospital, isolate 212 (16SrRNAþ, lukS/F�PV�, mecAþ) and V7 (16S rRNAþ, lukS/F�PVþ, mecA�) (a kind gift of Dr A. Kearns, HPA) wereused as positive controls for PCR.

MLST

MLST was performed essentially as described,18 notingimproved performance using primer pairs we developed(Table 1). DNA amplification was performed in a PTC-200 Peltier Thermal Cycler (MJ Research, Boston, Mass)in a final volume of 25 ml containing 2.5 ml 10� PCRbuffer (Qiagen, Crawley, UK), 0.125 ml Taq DNA polymer-ase (Qiagen), 0.05 ml forward and reverse primer (OperonScientific, Cologne, Germany), 0.0125 ml each dNTP (Invi-trogen, Paisley, UK) and 1 ml DNA. Cycle sequencing wasperformed using amplification primers (Table 1) andsequences read using an ABI 3730xl DNA instrument. ST

Table 1 Sequences of primers used in the PCR

Gene name Primer

16S rRNA Staph765FStaph750R

lukS/F-PV Luk-PV-1LukPV-2

mecA MecA1MecA2

Carbamate kinase (arcC)a arcc Farcc R

Shikimate dehydrogenase (aroE)a aroE FaroE R

Glycerol kinase (glpF)b glpF Fglpf R

Guanylate kinase (gmk)b gmk Fgmk R

Phosphate acetyltransferase (pta)a pta Fpta R

Triosephosphate isomerase (tpi)a tpi Ftpi R

Acetyl coenzyme A acetyltransferase(yqiL)a

yqil Fyqil R

a Samples were denatured for 2 min at 94 �C, followed by 35 cyclesextension at 72 �C for 1 min, with a final extension at 72 �C for 5 min

b Samples were denatured for 2 min at 94 �C, followed by 35 cyclesextension at 72 �C for 1 min, with a final extension at 72 �C for 5 min

was determined by comparing the seven alleles withstrains of known allele number on the MLST database(http://www.mlst.net).

Statistical analysis

Results were analysed using SPSS 14.0 for Windows. Fisher’sexact and ManneWhitney rank sum tests were used tocompare categorical and continuous variables, respec-tively. Time to event outcomes were analysed usingKaplaneMeier plots and log rank tests. P values of <0.05were considered statistically significant. Recognising thatwith 70 cases and controls our study had relatively lowpower to detect small effects, a multivariable model wasconstructed for cases and age-matched controls, usingbackwards elimination (P < 0.05) on all factors significantat P < 0.10 in Table 2 (excluding the CDC definition as noMRSA cases met this criteria, and whether the patient hadbeen in hospital in the last year as this was nearly collinearwith being in hospital in the past 3 years).

Results

Using electronic hospital admission and microbiologyrecords, 57 patients fulfilled inclusion criteria as MRSAadmission bacteraemia cases and 57 unmatched MSSAadmission bacteraemia unmatched controls were randomlyselected from the same time period. On further examina-tion 14 MRSA cases and 11 MSSA unmatched controls did notin fact meet inclusion criteria because they were eithermoved within the hospital (5 MRSA, 3 MSSA) or admitted to

Sequence Source

AACTCTGTTATTAGGGAAGAACA 17CCACCTTCCTCCGGTTTGTCACCATCATTAGGTAAAATGTCTGGACATGATCCA 17GCATCAAGTGTATTGGATAGCAAAAGCGTAGAAATGACTGAACGTCCGATAA 17CCAATTCCACATTGTTTCGGTCTAATTGATTCACCAGCGCGTATTGTC 18AGGTATCTGCTTCAATCAGCGGCAGTTATCGGAAATCCTATTTCAC This studyCTCATTAAAGTATTGGGAGAAAGATGCTTTGGTGGTGGCGTTTGTG This studyCCTAATAAACCACCGGCAATTGGGTTAATCGTTTTATCAGGACCATC This studyGTTCATCAATTTCACGCGCTCGTTAAAATCGTATTACCTGAAGG pta F18

GCTTCTTGAACTTTTGTCACGTCG pta R:This study

TCGAAGATAATGGTGCGTTCACAG This studyACCATGTTCGCTTTCGCGGTTCGCGAGAGTCGTATTAGCAGCAGC This studyGGTTCACCTTTACGTTGAGGAATCG

of denaturation at 94 �C for 30 s, annealing at 50 �C for 30 s and.of denaturation at 94 �C for 30 s, annealing at 55 �C for 30 s and.

Table 2 Characteristics

Characteristic MRSA (n Z 35),n (%) ormedian(IQR)

MSSAunmatchedcontrols(n Z 34)

P valuevs.MRSAcases

MSSAage-matchedcontrols(n Z 35)

P valuevs.MRSAcases

DemographicsSex (male) 26 (74%) 20 (88%) 0.21 14 (39%) 0.007Age 81 (75; 87) 70 (57; 81) <0.001 80 (70; 84) 0.34

Admitted fromNursing home 5 (14%) 0 (0%) 0.08 2 (6%) 0.70Residential home 4 (11%) 3 (9%) 4 (11%)Community hospital 4 (11%) 2 (6%) 3 (9%)Own home 22 (63%) 28 (82%) 26 (74%)Hostel 0 (0%) 1 (3%) 0 (0%)

Previous hospital exposureAdmitted to hospital in the past year 34 (97%) 16 (47%) <0.001 19 (54%) <0.001Admitted to hospital in the past 3 years 34 (97%) 21 (62%) <0.001 21 (60%) <0.001Days since last in hospitala 51 (10, 118) 591 (14; 1361) <0.001 236 (19; 1361) <0.001No. of in-patient appointments in last year 1 (1;2) 0 (0;1) <0.001 0 (0;1) 0.001No. of out-patient appointments in last year 1 (0;3) 0 (0;1.25) 0.05 0 (0;1) 0.09No. of days in hospital in the past year 17.4 (1.9; 33.4) 0 (0; 2.5) <0.001 0 (0; 2.8) <0.001

Co-morbiditiesHealthcare portal of entry 25 (71%) 12 (35%) 0.004 15 (43%) 0.03

Invasive surgery 12 (34%) 4 (12%) 8 (23%)Invasive intervention 7 (21%) 6 (18%) 4 (11%)Vascular access 0 (0%) 0 (0%) 0 (0%)Urinary catheter 11 (31%) 2 (6%) 4 (11%)Renal dialysis 1 (3%) 0 (0%) 0 (0%)Admission into intensive care unit 5 (14%) 1 (3%) 0 (0%)

Ulcers, eczema or psoriasis 8 (23%) 11 (32%) 0.43 7 (20%) 1.0Intra-venous drug use 0 (0%) 3 (9%) 0.11 0 (0%) 1.0Co-morbidity score 1 (1;2) 0 (0;1) 0.002 0 (0;2) 0.05

Diabetes 6 (17%) 4 (12%) 4 (11%)Heart failure 6 (17%) 2 (6%) 3 (9%)Oral steroids 2 (6%) 2 (6%) 2 (6%)Vascular disease 20 (57%) 9 (26%) 13 (37%)COPD 4 (11%) 2 (6%) 3 (9%)Chronic renal failure 5 (14%) 1 (3%) 4 (11%)

CDC Community AcquiredFit CDC definition of

Community-Acquired0 (0%) 16 (47%) <0.001 13 (37%) <0.001

MRSA previously isolatedPrevious MRSA 9 (26%) 1 (3%) 0.01 2 (6%) 0.05

Clinical presentationPatient judged clinically to have an

infection on the day of presentation31 (89%) 28 (82%) 0.51 29 (83%) 0.73

Temperature, �Cb 37.2 (36.3; 38.5) 37.8 (36.1; 38.5) 0.80 37.4 (36.0; 38.3) 0.62Pulse 88 (71; 105) 109 (88; 122) 0.003 97 (81; 115) 0.09Systolic BP 126 (105;143) 122 (100; 140) 0.70 130 (100; 147) 0.74Diastolic BP 61 (54; 75) 67 (57; 82) 0.12 65 (57; 85) 0.16WCC, 109/L 15.3 (10.9; 17.8) 14.4 (8.8; 20.8) 0.85 16.4 (10.3; 21.4) 0.82CRP 177 (92; 253) 164 (80; 267) 0.97 210 (143; 285) 0.23Neutrophils, 109/L 12.6 (8.7; 16.3) 12.5 (7.1; 18.5) 0.93 13.9 (7.6; 18.9) 0.80Lymphocytes, 109/L 0.9 (0.6; 1.4) 0.8 (0.6; 1.5) 0.92 0.8 (0.6; 1.2) 0.62

166 R. Miller et al.

Table 2 (continued )

Characteristic MRSA (n Z 35),n (%) ormedian(IQR)

MSSAunmatchedcontrols(n Z 34)

P valuevs.MRSAcases

MSSAage-matchedcontrols(n Z 35)

P valuevs.MRSAcases

Platelets, 109/L 252 (176; 336) 207 (88; 274) 0.09 235 (129; 357) 0.53Haemoglobin, g/dl 12.3 (10.5; 13.4) 12.5 (10.2; 13.4) 0.93 12.1 (10.0; 13.0) 0.49

Values are number (%) for categorical/dichotomous variables and median (25th;75th percentile) for continuous variables. Significancetests are 2-sided Exact Tests for categorical variables and ManneWhitney Rank Sum tests for continuous variables. A maximum of 2 casesare missing in each group.

a KaplaneMeier medians and quartiles, and log rank test.b Temperature missing from 2 MRSA and 1 unmatched MSSA.

MRSA admission bacteraemia 167

the Horton General Hospital where isolates were not stored(9 MRSA, 8 MSSA). This left 43 MRSA cases and 46 MSSAunmatched controls, for whom study data were obtainedfor 35 (81%) and 34 (74%) respectively, losses being due tomissing patient notes (4 MRSA, 5 MSSA), lost/unrecoverablemicrobial isolates (2 MRSA, 7 MSSA) and two individualswhose admission could not be found in the notes. A secondgroup of MSSA admission bacteraemia controls age-matchedwithin 6 years were identified for the 35 MRSA cases withcomplete data, re-sampling until each MRSA case had anage-matched control. Twenty of these age-matched con-trols were already part of the unmatched MSSA controlgroup and a further 15 controls were identified from thesame time period.

MRSA cases had a median age of 81 years, compared to70 years for unmatched MSSA controls (P < 0.001) (Table 2).Thus, age is a potential confounder when comparingco-morbidities and characteristics of MRSA and MSSA bac-teraemia. To adjust for this age-matched MSSA controls

33332222211111

Co

un

t

a

MRSA case MSSA age-matched control matched control

Reference lines at 3 years, 1 year, 6 months, 30 days and 1 week

10,000

1,000

100

10

1

0

MSSA un-

Days sin

ce p

atien

t w

as last in

h

osp

ital

Figure 1 Demographic details of cases of admission S. aureus bacpital, including in-patient admissions to Oxford and other hospitalsSix unmatched MSSA controls and 7 age-matched MSSA controls w3 years. (b) Location the patient was admitted from.

were used when comparing demographics and othercharacteristics.

One interesting feature was that individuals with admis-sion MRSA bacteraemia were more likely to be malethan those with admission MSSA bacteraemia (P Z 0.21unmatched, P Z 0.007 age-matched) (Table 2). Even aftermatching for age, patients with MRSA bacteraemia hadsignificantly more health-care associated characteristicsthan those with MSSA (Fig. 1, Table 2). This was notablein numerous measures of health-care exposure includingthat those with MRSA had more co-morbidities (P Z 0.05),more health care portals of entry (P Z 0.03) had been inhospital more recently (P < 0.001) (Fig. 1a) and had beenin hospital for more time in the past year (P < 0.001) thanage-matched controls with MSSA bacteraemia (Table 2).All but one of the MRSA cases had been in hospital in thepast year, compared to only 16 (47%) of the unmatchedand 19 (54%) of the age-matched MSSA controls (Table 2).The individual with MRSA who had not been in hospital in

6420864208642086420

Residential HomeOwn HomeNursing HomeHostel

Community Hospital

bMSSA age-

matched control matched controlMRSA case MSSA un-

teraemia. (a) Number of days since the patient was last in hos-as well as community hospitals, and out-patient appointments.ere without record of previous admission and were plotted at

168 R. Miller et al.

the past year had other health-care associated characteris-tics including admission from a nursing home, a urinarycatheter, and the nosocomial MRSA strain ST-22. Conse-quently no MRSA cases were classed as CA-MRSA by theCDC definition, however, 13/35 (37%) of age-matched and16/34 (47%) unmatched MSSA controls fitted the CDCdefinition of community-acquired infection11 (Box 1, Table 2).MRSA cases were also more likely to have had MRSA isolatedpreviously than MSSA controls (P Z 0.01 unmatched,P Z 0.05 age-matched). Previous MRSA isolation fromMRSA cases was from skin and superficial samples a medianof 185 days before admission (not shown). Finally it is nota-ble that, among the cases of MRSA bacteraemia, over half(63%) were admitted from their own home, as opposed tofrom a nursing or residential home, a community hospitalor a hostel (Fig. 1b, Table 2).

Using a multivariate model we found that independentpredictors of MRSA versus MSSA age-matched admissionbacteraemia from Table 2 were male versus female (oddsratio (OR): 3.34, 95% confidence interval (CI): 0.91e12.27),the number of days the patient had been in hospital in thelast year (OR: 1.07 per day, 95% CI: 1.00e1.13), and thenumber of months since the patient was last in hospital(OR: 0.94 per 30 days, 95% CI: 0.89e1.00). When the modelwas fitted to include whether the patient had been in hos-pital in the last year instead of in the last 3 years resultswere very similar. The major difference being that whetherthe patient had been in hospital in the last 3 years was anindependent predictor, rather than duration the patienthad spent in hospital in the last 3 years.

9259246304642561881827845363430252220151210

8521

ST

302826242220181614121086420Count

a b

MSSA bac

MRSA ba

Sequence Type

Ag

e o

f p

atien

t o

n ad

missio

n in

years

Figure 2 Sequence type of admission S. aureus isolates. (a) MLSpatients on admission in years with MRSA ST-22 and ST-36 with line

Despite the excess of co-morbidity in the MRSA group(data not shown), the clinical presentations of MRSA andMSSA bacteraemia were very similar, as judged by objec-tive, prospectively collected measures of illness severity(temperature, pulse, blood pressure, white cell counts andC-reactive protein) (Table 2). All patients had abnormalitiesin one or more measures, and over 80% were judged clini-cally to have an infection on the day of presentation,with no differences between MRSA and MSSA bacteraemia(Table 2). It is therefore unlikely that these blood streamisolations reflect contamination.19

Extensive, clonal spread of successful S. aureus clonescan change the clinical presentation and prevalent strainsof S. aureus disease, as has occurred following communityMRSA spread in the USA (associated with MLST ST-8) andhealthcare associated MRSA in the UK (ST-22, 36).20,21 To in-vestigate which clones of S. aureus were associated withadmission bacteraemia, we performed MLST typing. Wefound that the STs for MRSA were much less varied thanfor MSSA with all but one being ST-22 (n Z 27) or ST-36(n Z 7). One individual with MRSA ST-256 was in theirsixties, admitted from their own home, with last hospitaladmission 6 months previously, no previous MRSA isolation,no portals of entry and two co-morbidities; COPD and heartfailure. In contrast, unmatched MSSA strains had 20 differ-ent STs of much lower frequency, 2 of which were novel tothis study (ST-924 and ST-925) (Fig. 2a). The most frequentMSSA STs were ST-15 (n Z 6) and ST-30 (n Z 6) with theother MSSA STs having frequencies of three or fewer.Additionally all of the MRSA isolates and none of the MSSA

teraemia

cteraemia

ST-22 ST-36

100

95

90

85

80

75

70

65

60

55

50

45

40

Median ST 22

Median ST 36

T STs of MRSA cases and MSSA unmatched controls. (b) Age ofs showing the median (83 and 70 years respectively).

MRSA admission bacteraemia 169

isolates had the MecA gene and no isolates had the PVLgene.

Comparison of the MRSA ST-22 and ST-36 showed thatindividuals with MRSA ST-22 admission bacteraemia weresignificantly older than those with ST-36 with median agesof 83 and 70 years for ST-22 and ST-36 respectively(P Z 0.004) (Fig. 2b).

Discussion

This study has identified a specific population admitted tohospital with MRSA bacteraemia; the patients are pre-dominantly elderly with high levels of previous hospitalexposure, but mostly resident in their own homes prior toadmission. MRSA cases fitting the CDC definition forcommunity acquisition were not identified and in keepingwith this, 34/35 had STs corresponding to UK health-careassociated epidemic MRSA strains.

Clinical assessment and measures of host response toinfection were all compatible with a bacteraemic illness, asopposed to contamination of blood cultures.22,23 Theseresults support previous anonymised analyses in the samepopulation demonstrating that the population at highestrisk of admission MRSA bacteraemia are those with previoushospital contact. Notably, among these patients, MRSA bac-teraemia occurred at rates comparable with Streptococcuspneumoniae, MSSA and Escherichia coli bacteraemia.Such population-based analysis,24 together with nationalreports6 and our current study establish MRSA as a key path-ogen causing bacteraemia in individuals with previous hos-pital exposure.

Limitations of this study arise from its conduct in a singlecentre and its retrospective design. However, the epidemi-ology we observe in this and our previous report9 is similarboth to contemporaneous but more limited national data-sets,6 and to reports from other regions,25 supporting gen-eralisability of our conclusions. Nevertheless, stratifiedsampling from other areas and hospital types, perhaps aspart of the national mandatory surveillance scheme, mightreveal regional variation or additional complexity. Antibi-otic usage before admission may affect the susceptibilityof the S. aureus bacteraemia acquired and would havebeen interesting to investigate. Unfortunately, preliminarystudies showed that in this region, it was difficult to obtaincomplete, prospectively recorded antibiotic histories onthis patient group who typically had had multiple health-care contacts in diverse locations. Finally, it is importantto note that this study deals only with bacteraemic illness;the prevalence of CA-MRSA may be higher than identified inthis study if, as reported,4 CA-MRSA strains most frequentlypresent with skin and soft tissue infection.

S. aureus can have a prolonged carriage state, and MRSAstrains acquired in hospital can persist for many months oreven years in some individuals.10 Since the recently hospi-talised form a substantial part of the population (an esti-mate of >5% was derived from a geographical regionsurrounding Oxford),24 it is important to consider the po-tential for post-discharge sequelae following in-hospitalMRSA acquisition. Bacteraemia is not the most commonform of S. aureus disease,26 and we would speculate thatcases readmitted with bacteraemia represent the tip of

an iceberg of largely unrecognised MRSA disease in thehealth-care exposed community. Estimates of the extentof any such problem are hard to derive due to uncertainlyabout carriage half-life, prevalence, the relationship be-tween disease risk and time since acquisition, and difficul-ties ascribing microbiological aetiology in lung, skin andsoft tissue infections. Adding to the difficulty of estimatingthe extent of MRSA are problems interpreting urinary isola-tion in elderly populations where both catheter use andasymptomatic polymicrobial urinary isolation is common.Nevertheless, studies addressing some of these issues arefeasible; the data presented show they are clearlywarranted.

Acknowledgements

R.M. was funded by NIHR comprehensive biomedicalresearch award to the Oxford Radcliffe Hospital andUniversity of Oxford.

References

1. Grundmann H, Aires-de-Sousa M, Boyce J, Tiemersma E. Emer-gence and resurgence of methicillin-resistant Staphylococcusaureus as a public-health threat. Lancet 2006;368(9538):874e85.

2. Robinson DA, Enright MC. Multilocus sequence typing and theevolution of methicillin-resistant Staphylococcus aureus. ClinMicrobiol Infect 2004;10(2):92e7.

3. Fridkin SK, Hageman JC, Morrison M, Sanza LT, Como-Sabetti K,Jernigan JA, et al. Methicillin-resistant Staphylococcus aureusdisease in three communities. N Engl J Med 2005;352(14):1436e44.

4. Diederen BM, Kluytmans JA. The emergence of infections withcommunity-associated methicillin resistant Staphylococcusaureus. J Infect 2006;52(3):157e68.

5. Holmes A, Ganner M, McGuane S, Pitt TL, Cookson BD,Kearns AM. Staphylococcus aureus isolates carrying Panton-Valentine leucocidin genes in England and Wales: frequency,characterization, and association with clinical disease. J ClinMicrobiol 2005;43(5):2384e90.

6. Health Protection Agency. Commentary on quarterly, six-monthly and annual data for MRSA bacteraemia derived frommandatory surveillance. HPA 2007.

7. Wyllie DH, Crook DW, Peto TE. Mortality after Staphylococcusaureus bacteraemia in two hospitals in Oxfordshire,1997e2003: cohort study. BMJ 2006;333(7562):281.

8. Johnson AP, Pearson A, Duckworth G. Surveillance andepidemiology of MRSA bacteraemia in the UK. J AntimicrobChemother 2005;56(3):455e62.

9. Wyllie DH, Peto TE, Crook D. MRSA bacteraemia in patients onarrival in hospital: a cohort study in Oxfordshire 1997e2003.BMJ 2005;331(7523):992.

10. Salgado CD, Farr BM, Calfee DP. Community-acquired methicil-lin-resistant Staphylococcus aureus: a meta-analysis of preva-lence and risk factors. Clin Infect Dis 2003;36(2):131e9.

11. Naimi TS, LeDell KH, Como-Sabetti K, Borchardt SM, Boxrud DJ,Etienne J, et al. Comparison of community- and health care-associated methicillin-resistant Staphylococcus aureus infec-tion. JAMA 2003;290(22):2976e84.

12. Tacconelli E, Venkataraman L, De Girolami PC, EM DA.Methicillin-resistant Staphylococcus aureus bacteraemiadiagnosed at hospital admission: distinguishing between

170 R. Miller et al.

community-acquired versus healthcare-associated strains.J Antimicrob Chemother 2004;53(3):474e9.

13. Lodise Jr TP, McKinnon PS, Rybak M. Prediction model to iden-tify patients with Staphylococcus aureus bacteremia at risk formethicillin resistance. Infect Control Hosp Epidemiol 2003;24(9):655e61.

14. Pan ES, Diep BA, Charlebois ED, Auerswald C, Carleton HA,Sensabaugh GF, et al. Population dynamics of nasal strains ofmethicillin-resistant Staphylococcus aureusdand their rela-tion to community-associated disease activity. J Infect Dis2005;192(5):811e8.

15. O’Sullivan NP, Keane CT. Risk factors for colonization withmethicillin-resistant Staphylococcus aureus among nursinghome residents. J Hosp Infect 2000;45(3):206e10.

16. Health Protection Agency. National standard methods. HPA2006.

17. McClure JA, Conly JM, Lau V, Elsayed S, Louie T, Hutchins W,et al. Novel multiplex PCR assay for detection of the staphylo-coccal virulence marker Panton-Valentine leukocidin genesand simultaneous discrimination of methicillin-susceptiblefrom -resistant staphylococci. J Clin Microbiol 2006;44(3):1141e4.

18. Enright MC, Day NP, Davies CE, Peacock SJ, Spratt BG. Multilo-cus sequence typing for characterization of methicillin-resis-tant and methicillin-susceptible clones of Staphylococcusaureus. J Clin Microbiol 2000;38(3):1008e15.

19. Neish B, Gillespie T, Liddell KG. Improving surveillance of MRSAbacteraemia: Scottish data prompt query of significance ofMRSA isolated from blood in acute admissions. BMJ 2005;331(7525):1143.

20. Cox RA, Bowie PE. Methicillin-resistant Staphylococcus aureuscolonization in nursing home residents: a prevalence study inNorthamptonshire. J Hosp Infect 1999;43(2):115e22.

21. Crum NF, Lee RU, Thornton SA, Stine OC, Wallace MR,Barrozo C, et al. Fifteen-year study of the changing epidemiol-ogy of methicillin-resistant Staphylococcus aureus. Am J Med2006;119(11):943e51.

22. Wyllie DH, Bowler IC, Peto TE. Bacteraemia prediction inemergency medical admissions: role of C reactive protein.J Clin Pathol 2005;58(4):352e6.

23. Wyllie DH, Bowler IC, Peto TE. Relation between lymphopeniaand bacteraemia in UK adults with medical emergencies.J Clin Pathol 2004;57(9):950e5.

24. Wyllie DH, Walker AS, Peto TE, Crook DW. Hospital exposure ina UK population, and its association with bacteraemia. J HospInfect 2007;67(4):301e7.

25. Karas JA, Enoch DA, Emery MM. Community-onset healthcare-associated MRSA bacteraemia in a district general hospital.J Hosp Infect 2006;62(4):480e6.

26. Drews TD, Temte JL, Fox BC. Community-associated methicil-lin-resistant Staphylococcus aureus: review of an emergingpublic health concern. WMJ 2006;105(1):52e7.