Clinician-led surgical site infection surveillance oforthopaedic procedures: a UK multi-centre pilotstudy

M. Morgana, J. Blackb, F. Bonec, C. Fryd, S. Harrisa, S. Hogge, A. Holmese,S. Hughese, N. Lookerf, G. McIlvennyg, J. Nixonh, J. Nolani, A. Noonej,J. Reillyj, J. Richardsk, E. Smythg, A. Howardl,*

aNPHS Communicable Disease Surveillance Centre, Cardiff, UKbDepartment of Orthopaedic Surgery, West Wales General Hospital, Carmarthen, UKcDepartment of Microbiology, Dumfries and Galloway Royal Infirmary, Dumfries, Scotland, UKdDepartment of Health, Skipton House, London, UKeImperial College London, Charing Cross, London, UKfNPHS Microbiology Rhyl, Ysbyty Glan Clwyd, Rhyl, UKgNorthern Ireland Healthcare Associated Infection Surveillance Centre, The Royal Hospital, Belfast, UKhThe Ulster Independent Clinic, Belfast, UKiDepartment of Orthopaedic Surgery, Norfolk and Norwich University Hospital, Norwich, UKjScottish Surveillance of HAI Programme, Health Protection Scotland, Glasgow, UKkDepartment of Infection, Prevention and Control, Norfolk and Norwich University Hospital, Norwich, UKlInfection and Communicable Disease Service, Temple of Peace and Health, Cathays Park, Cardiff CF103NW, UK

Received 22 July 2004; accepted 18 November 2004Available online 22 March 2005

01do

KEYWORDSHealthcare-associatedinfection surveillance;Surgical site;Orthopaedic surgery;Arthroplasty; Hemi-arthroplasty fixation;Trochanteric fractures

95-6701/$ - see front matter Q 200i:10.1016/j.jhin.2004.11.024

* Corresponding author. Tel.: C44 2E-mail address: tony.howard@nphs

Summary The UK Department of Health established the Healthcare-associated Infection (HAI) Surveillance Steering Group in 2000 to develop astrategy for implementing a national programme for HAI surveillance inNational Health Service trusts. A subgroup of this committee examined thesurveillance of surgical site infections following orthopaedic surgery. Thisgroup oversaw a pilot scheme that was set up in 12 hospitals around the UKto explore the feasibility of implementing a system of surveillance thatengaged clinical staff in its operation, provided a process for continuousdata collection and could be maintained as part of routine hospital operationover time. Aminimum data set was established by the subgroup, and Centersfor Disease Control and Prevention (CDC) definitions of infection were used.By March 2003, the surveillance had been undertaken continuously in 11 sites

Journal of Hospital Infection (2005) 60, 201–212

www.elsevierhealth.com/journals/jhin

5 The Hospital Infection Society. Published by Elsevier Ltd. All rights reserved.

9 2040 2530; fax: C44 29 2040 2529..wales.nhs.uk

M. Morgan et al.202

for one to two years, depending on the date of implementation. Only onehospital had ceased data collection. The information was collected mainlyby clinical staff, with support and co-ordination usually provided byinfection control teams. Data on more than 5400 procedures were availablefor analysis for four core procedures: arthroplasty of the hip and knee; hemi-arthroplasty of the hip; and internal fixation of trochanteric fractures of thefemur. The data set permitted the calculation of risk-adjusted rates,allowing comparisons between hospitals and within a hospital over time. Themethodology enhanced clinical ownership of the surveillance process, re-inforced infection control as the responsibility of all staff, and providedtimely feedback and local data analysis. The use of CDC definitionspermitted international comparisons of the data.Q 2005 The Hospital Infection Society. Published by Elsevier Ltd. All rightsreserved.

Introduction

A pilot surveillance scheme for healthcare-associ-ated infection (HAI) was established in England in1995; the Nosocomial Infection National Surveil-lance Scheme (NINSS). Up to 2002, approximatelytwo-thirds of English trusts had participated in twoNINSS modules—surveillance of hospital-acquiredbacteraemia and surveillance of surgical site infec-tion (SSI).1 The National Audit Office (NAO)reported on HAI2 in February 2000, and rec-ommended that the Department of Health shouldencourage participation in NINSS so that it becamea comprehensive scheme. The Public AccountsCommittee response to this, published in November2000,3 strengthened this advice, suggesting that thescheme should be mandatory for all trusts inEngland. In consequence, the Department of Healthestablished the Healthcare-associated InfectionSurveillance Steering Group (HAISSG), under thechairmanship of a National Health Service (NHS)trust chief executive, to examine the feasibility ofthis proposal and to develop a broad strategy forHAI surveillance, taking into account experiencegained across the UK and elsewhere.

One of the priority areas for the HAISSG toconsider was surveillance of SSI; a substantial causeof morbidity and mortality among hospitalizedpatients.4 An ability to quantify and monitor theoccurrence of these adverse events is essential toreach objective decisions regarding the scale of theproblem in any given institution, the need fordetailed investigation, the requirement for actionto be taken, and the impact of any intervention thatmight be introduced. This has been shown tocontribute to the achievement of significantreductions in SSI rates.5

Of particular interest to the HAISSG was

surveillance of SSI in relation to orthopaedicsurgery, where the clinical and financial costs ofinfection are high. Orthopaedic SSI can prolongtotal hospital stay by a median of two weeks/patient, can almost double rehospitalization rates,and can increase healthcare costs by more than300%. In addition, patients with orthopaedic SSIhave substantially greater physical limitations andsignificant reductions in their health-related qualityof life.6 The high impact of infections in this areawas noted by the NAO in their report ‘Hip replace-ments: getting it right first time’7 published in April2000, which recommended that all NHS trustsshould monitor infection rates accurately and takecost-effective action to reduce them. A multi-disciplinary subgroup of the HAISSG was thereforeestablished to examine the potential for imple-menting a UK-wide programme that built on theexperience gained by NINSS and other schemes inoperation in the UK.8 The objective was tointroduce an efficient surveillance methodologythat could be undertaken as part of the routineoperation of any UK hospital, utilizing interdisci-plinary co-operation involving nursing, surgical andcommunity staff and infection control teams, thatwould have the following characteristics:

enable all hospitals to collect a minimum amountof data in relation to SSI, that would: (1) besufficient as a measure of surgical performancein this area; (2) indicate where action wasrequired to reduce rates; and (3) monitorprogress following infection controlinterventions;

become the responsibility of clinical workers; utilize a methodology that would allow continu-ous data collection and be incorporated intoroutine hospital activities;

Clinician-led surgical site infection surveillance of orthopaedic procedures 203

provide timely data to users that could besubjected to local data analysis;

establish regional, national and internationalcomparators; and contribute to an increased understanding of therisk factors for SSI.

To inform the subgroup on the range of issuesrelating to the practicality of introducing a schemethat could meet these objectives, a pilot surveil-lance programme was introduced in 12 UK NHShospitals. The methodology is described below.This was developed taking account of the need for asmall core minimum data set that was consistentwith international comparators, with the primaryinvolvement of clinical staff in data collection, thatwas an efficient means of recording and collatingdata, and provided a simple electronic method fordata analysis and feedback.

Methods

The HAISSG SSI surveillance subgroup was respon-sible for deciding which orthopaedic proceduresshould be surveyed, the minimum data set forsurveillance and the definitions to be used.

Procedures

The following operative procedures were the corecomponents of the orthopaedic surveillance pro-gramme: arthroplasty of the hip; arthroplasty ofthe knee; hemi-arthroplasty for fracture of theneck of femur; and fractures of the trochantericregion of the femur that require internal fixationand do not include placement of a joint prosthesis.

Minimum data set and definitions

The core minimum data set used for the pilot studywas based on the surgical patient surveillancecomponent of the Centers for Disease Control andPrevention (CDC) National Nosocomial InfectionSurveillance (NNIS) system,9 with supplementaryquestions specific to orthopaedic surgery decidedby the orthopaedic surgeons involved in thesubgroup in consultation with colleagues. The CDCNNIS surgical site infection definitions10 were usedin the surveillance. The data set included questionson the criteria for diagnosis of infection, whichwere designed to allow sites that had previouslyused NINSS definitions to compare their historicaldata directly with new data obtained using the NNISdefinitions. The primary outcome measure was the

presence or absence of infection at discharge fromhospital. Where prosthetic implants were inserted,the CDC criteria required a record of infectionsoccurring in relation to the operative site for up toone year post surgery.

Surveillance protocol

The subgroup contained representatives from Eng-land, Wales, Scotland and Northern Ireland, andpilot sites were identified in all four countries: six inEngland; two in Northern Ireland; one in Scotlandand three in Wales.

The majority of sites used optical mark readerquestionnaires for data entry in conjunction with abespoke database for data analysis. Direct dataentry using the patient information systemwas usedby one hospital. However, the majority of sites usedthe following method. Questionnaires, encompass-ing the core data set and any other questions ofinterest to the pilot sites, were designed andprinted using optical mark reader software. Theoptical mark reader was either based at the pilothospital itself or at a separate site serving severalpilot hospitals. Each questionnaire contained threesections: (1) patient identifiers; (2) patient andprocedure details, completed in the operatingtheatre; and (3) infection and discharge details,completed on discharge. The questionnaires weresent to designated surveillance co-ordinators ateach pilot site, who distributed them to theappropriate operating theatres. Training in theuse of the questionnaires and the definitions wasprovided to all staff involved in their completion.The questionnaires were completed by members ofthe surgical teams in the operating theatre for allpatients undergoing the surveillance procedures.Information regarding infections and dischargeinformation was collected by ward staff. A principleof the data collection was that it should be seenprimarily as the responsibility of the clinical teams.Additionally, administrative systems were requiredthat allowed continuous data collection. Hospitalinfection control staff provided varying degrees ofco-ordination and specialist support. After com-pletion on discharge, the questionnaires werereturned to the surveillance co-ordinator, whoremoved and filed the patient identifier sectionbefore sending the rest of the questionnaire forelectronic scanning, either within the hospital or toanother site depending on the optical mark readerlocation. The role of the surveillance co-ordinatorwas discussed by Hogg et al.11 The professionalbackground of the surveillance co-ordinators varied

M. Morgan et al.204

in participating hospitals, but mainly comprisednursing or clerical staff.

A bespoke database, written by the WelshHealthcare-associated Infection Programme, wasused by all the pilot sites in England and Wales forstorage and analysis of the scanned SSI data. Thedatabase consisted of a Microsoft Access Version4.0 (Redmond, USA) Jet Engine, interfaced with aMicrosoft Visual Basic Professional Version 6.0(Redmond, USA) suite of user screens. The databaseincluded predefined reports, written using SeagateCrystal Reports Version 8.0 (Scotts Valley, USA) andan export facility to allow ad-hoc analyses inalternative software packages. The databaseautomatically assigned a risk index for eachpatient record, based on the procedure type,wound class, American Society of Anesthesiolo-gists’ (ASA) classification and procedure durationprovided, using the NNIS risk index classifi-cation.12 Two versions of the database werecreated—a regional version, designed to storeand analyse the results from surveillance atmore than one hospital, and a local version forthe results of SSI surveillance at a singlehospital. In Wales, the files were exportedfrom the optical mark reader into the regionalversion of the SSI database at the NationalPublic Health Service Communicable DiseaseSurveillance Centre for all-Wales analysis. Theidentities of hospitals and individual surgeonsremained confidential within the system throughthe use of codes. The data files were also sentto the surveillance co-ordinators at the hospi-tals, who imported them into their local versionof the database. This was then used by thesurveillance co-ordinator to feed the results ofthe surveillance back to the surgical and infec-tion control teams.

Quality assurance

Questionnaire records were imported into thedatabase, regardless of the number of data fieldscompleted. The level of questionnaire completionthat was acceptable was determined by individualhospitals. In some hospitals, surveillance co-ordi-nators followed-up all missing data items, someonly followed-up data items required for calculat-ing the risk index and the SSI questions, and somedid no follow-up at all. Surveillance co-ordinatorswere encouraged to feed the level of questionnairecompletion back to the surgical teams by means ofone of the predefined reports. They were alsoencouraged to audit and feedback a breakdown ofthe number of questionnaires received compared

with the number of orthopaedic procedures per-formed according to theatre records.

Statistical analysis

The data exported from all four countries werestored and analysed in SPSS (Chicago, USA) andMicrosoft Excel (Redmond, USA). Ninety-five per-cent Poisson confidence intervals were calculatedaround hospital SSI rates using Stata Version 7(College Station, USA). Odds ratios (ORs) and Chi-squared tests for comparison of rates by cliniciantype were calculated in Epi Info Version 6 (Atlanta,USA).

Results

The clinician-led surveillance scheme for SSIsfollowing procedures performed by orthopaedicsurgeons was established successfully in 12 pilothospitals around the UK. A total of 8364 proceduresperformed by orthopaedic surgeons had beensurveyed by the pilot hospitals and aggregated inthe UK database at the time of writing. Somehospitals found it convenient to survey all ortho-paedic procedures, rather than restrict surveillanceto the four core procedures. The extra proceduresincluded prostheses other than hip or knee andother musculoskeletal procedures. There were 2959procedures excluded from the analysis for thefollowing reasons: non-core procedures (NZ2813);patients with a length of stay !24 h (NZ76), i.e.day cases; or no indication given of the presence orabsence of SSI (NZ70). A total of 5405 proceduresremained for analysis after exclusions were made.

Aggregate results

Tables I and II show the core minimum data setquestions with the number and percentage ofmissing values for independent questions (e.g. sexof patient) and dependent questions (e.g. type ofSSI), respectively. The percentages given sub-sequently throughout this paper are the validpercent, i.e. the valid percent is computed withmissing values excluded from the percentage base.

Rates of SSIThe crude SSI rate for the four core procedures was2.4% (130/5405). Of these, 77.3% were categorisedas superficial. The number of procedures andpercentage SSI by procedure type are shown inFigure 1.

Table I Observed responses and missing values (%)for independent core data set questions

Field Observations Missing (%)

Sex 5381 24 (0.4)Date of admission 5405 0 (0.0)Date of procedure 5404 1 (!0.1)Operation type 5369 36 (0.7)Anaesthesia 4920 485 (8.8)Wound class 5328 77 (1.4)ASA class 5283 122 (2.3)Procedure type 5405 0 (0.0)Surgeon grade 5129 276 (5.1)Locuma 3329 2076 (38.4)Systemic antibioticprophylaxisa

3351 2054 (38.0)

Thromboprophylaxisa 2638 2767 (51.2)SSI 5405 0 (0.0)Date of discharge/death

5405 0 (0.0)

Age 5397 8 (0.1)

ASA, American Society of Anesthesiologists; SSI, surgical siteinfection.a One hospital did not collect this information.

Figure 1 The number of procedures and the percentageof surgical site infections (SSIs) by procedure type inhospitals participating in the Department of Healthorthopaedic SSI pilot surveillance scheme. Bars, numberof procedures; ——, percentage of SSIs.

Clinician-led surgical site infection surveillance of orthopaedic procedures 205

Patient demographicsMales accounted for 39% (2097/5381) of the patientpopulation. The crude SSI rate in male patients was2.6% (54/2097), compared with 2.3% (75/3284) infemale patients. The number of patients andpercentage SSI by patient age group for the coreprocedures are shown in Figure 2. Over 90% ofprocedures were carried out on patients aged over55 years. Generally, the rate of SSI increased withincreasing age.

Patient risk indexThe number of procedures and percentage SSI forpatients with different risk indices are provided inFigure 3. There were only seven patients recorded

Table II Observed responses and missing values (%)for dependent core data set questions

Field Observations Missing (%)

Type of infection 97 33 (25.4)Date of infection 98 32 (24.6)Criteria used foridentifying SSI

98 32 (24.6)

Duration of procedure 5193 212 (3.9)Patient risk index 5029 376 (7.0)Antibiotic-loadedcementa

2529 2649 (53.2)

SSI, surgical site infection.a One hospital did not collect this information.

with a risk index of 3 in the data set and no SSIs wererecorded in this category.

SSI typesThe majority of SSIs were superficial incisional foreach of the procedures. The proportion of each SSItype for each procedure type is shown in Figure 4.

Clinician typesThe proportions of procedures carried out bydifferent clinician types (see Appendix for defi-nitions) are given in Figure 5. Most (70.9%) ofprocedures were performed by consultants. Thenumber of procedures and percentage SSI byclinician type are shown in Figure 6. The highest

Figure 2 The number of hip and knee prostheses,fixation of fractures and hemi-arthroplasties, and per-centage of surgical site infections (SSIs) by age group inhospitals participating in the Department of Healthorthopaedic SSI pilot surveillance scheme. Bars, numberof procedures; ——, percentage of SSIs.

Figure 3 The number of hip and knee prostheses,fixation of fractures and hemi-arthroplasties, and thepercentage of surgical site infections (SSIs) by patient riskindex in hospitals participating in the Department ofHealth orthopaedic SSI pilot surveillance scheme. Bars,number of procedures; ——, percentage of SSIs.

Figure 5 The proportion of total hip and kneeprostheses, fixation of fractures and hemi-arthroplastyprocedures carried out by different clinician types inhospitals participating in the Department of Healthorthopaedic surgical site infection pilot surveillancescheme.

M. Morgan et al.206

rate of SSI occurred in procedures carried out byspecialist registrars. This was the case for eachprocedure type, as illustrated in Table III (data onlyshown for consultants and specialist registrars). TheSSI rate was lower for consultants (2%) than non-consultants (2.8%), but this was not significant [74/3635 vs. 42/1494; ORZ0.72 (0.48–1.07), PZ0.09].

Approximately 8% (262/3329) of procedureswere carried out by locum clinicians. There was asignificantly increased risk of an SSI in procedurescarried out by the latter group [18/262 vs. 44/3067;ORZ5.07 (2.71–9.11), P!0.001]. A comparison ofthe SSI rates for locums and non-locums for

Figure 4 The proportion of superficial, deep and organ/shospitals participating in the Department of Health orthopaebars, superficial SSIs; open bars, deep SSIs; shaded bars, orga

consultants and specialist registrars is given inTable IV. There was a significantly increased riskof an SSI in procedures carried out by locumconsultants compared with non-locum consultantsand for locum specialist registrars compared withnon-locum specialist registrars. A comparison of theSSI rates by type of procedures carried out bylocums and non-locums is given in Table V. Therewas a significantly increased risk of SSI following hipand knee prostheses carried out by locums.

As part of the minimum data set, a question wasincluded on whether a consultant was present in thetheatre suite if a non-consultant was performing

pace surgical site infections (SSIs) by procedure type indic surgical site infection pilot surveillance scheme. Solidn/space SSIs.

Figure 6 The number of hip and knee prostheses,fixation of fractures and hemi-arthroplasties, and thepercentage of surgical site infections (SSIs) by cliniciantype in hospitals participating in the Department ofHealth orthopaedic SSI pilot surveillance scheme. Bars,number of procedures; ——, percentage of SSIs.

Table III Percentage surgical site infection (SSI) followiconsultant clinicians by the type of procedure, and the odfollowing surgery by a specialist registrar compared with a

Procedure type

Specialist registrars

Hip prosthesis 2.4% (10/421)Knee prosthesis 3.7% (12/322)Fixation of fracture 1.9% (6/318)Hemi-arthroplasty 5.6% (13/233)

Table IV Percentage surgical site infection (SSI) followingand internal fixation of trochanteric fractures carried out by95% confidence interval) of an infection following surgery b

Clinician type

Locums

Consultant 6.7% (11/163)Specialist registrar 6.7% (3/45)

Table V Percentage surgical site infection (SSI) following pby the type of procedure, and the odds ratio (with 95% conlocum compared with a non-locum

Procedure type

Locums

Hip prosthesis 7.0% (6/86)Knee prosthesis 7.2% (8/111)Fixation of fracture 4.9% (2/41)Hemi-arthroplasty 8.3% (2/24)

Clinician-led surgical site infection surveillance of orthopaedic procedures 207

the procedure. This question was poorly com-pleted, but the data show very little difference inthe rates by the presence of a consultant. The rateof SSI in non-consultants was 1.4% (3/208) whenthere was a consultant present in the theatre suite,and 1.1% (1/93) when there was not.

Antibiotic useSystemic antibiotic prophylaxis was given to 4.6%(153/3351) of patients undergoing the four pro-cedures. The percentage SSI was 1.9% (61/3198) forthose who received systemic antibiotic prophylaxisand 1.3% (2/153) for those who did not. Noinformation was available centrally on the type ofantibiotics used nor on the length of prophylaxis. Inpatients who underwent procedures using cementand where the field was completed, the majority(87.5%) received antibiotic-loaded bone cement. Forthose who received antibiotic-loaded bone cement,the percentage SSIwas 2.0% (28/1399); for thosewho

ng procedures carried out by specialist registrars andds ratio (with 95% confidence interval) of an infectionconsultant

% SSI

Consultants OR (95% CI)

1.9% (37/1950) 1.61 (0.74–3.40)2.2% (29/1359) 1.78 (0.85–3.67)1.4% (3/212) 1.34 (0.28–8.37)3.5% (4/114) 1.63 (0.49–6.99)

insertion of hip and knee prostheses, hemi-arthroplastylocum and non-locum clinicians, and the odds ratio (withy a locum compared with a non-locum

% SSI

Non-locums OR (95% CI)

1.4% (33/2385) 5.16 (2.30–10.70)0.9% (3/336) 7.93 (1.02–60.52)

rocedures carried out by locum and non-locum cliniciansfidence interval) of an infection following surgery by a

% SSI

Non-locums OR (95% CI)

1.4% (23/1648) 5.30 (1.71–13.86)1.6% (15/932) 4.75 (1.70–12.25)1.5% (6/396) 3.33 (0.32–19.39)0.0% (0/91) –

M. Morgan et al.208

did not, the rate was 2.0% (4/199). However, thesedata should be interpretedwith caution as antibiotic-loaded cement was only recorded for 46.8% ofprocedures.

ThromboprophylaxisThromboprophylaxis was recorded as given to 72.3%of patients (1907/2638). There was no significantdifference in the rate of SSI between patientsreceiving thromboprophylaxis [1.8% (34/1907)] andthose who did not [1.4% (10/731)].

Underlying orthopaedic pathologyThe diagnosis fields were completed on only 12%(682/5178) of procedures. Of these, 61.2% wererecorded as having osteoarthritis, 28.9% hip frac-ture, 2.2% inflammatory arthritis and 7.7% otherdiagnosis. The paucity of data in this categoryprecludes meaningful analysis of the underlyingpathology as a risk factor for infection.

Length of stayFor each of the core procedures, patients with SSIhad longer median stays from procedure to dis-charge than those who did not (Table VI). Thedifference in length of stay from procedure todischarge was greatest in patients undergoingfixation of fractures.

Results by hospital

The number of procedures and crude percentage SSIby pilot hospital are shown in Figure 7. The averagenumber of questionnaires returned per month bypilot hospital is shown in Figure 8. Hospitals weredivided into three categories: high volume, O130procedures/month (two hospitals); mediumvolume, 30–40 procedures/month (four hospitals);and low volume, !20 procedures/month (six

Table VI Median lengths of stay (in days) fromprocedure to discharge for patients with and withoutsurgical site infection (SSI) by procedure type

Procedure All patients Patientwithout SSI

Patientwith SSI

Hip pros-thesis

7 7 13

Knee pros-thesis

7 7 12

Hemi-arthro-plasty

12 12 18

Fixation offracture

8 8 13

hospitals). The SSI rate for high-, medium- andlow-volume facilities was 2.3%, 2.8% and 2.1%,respectively.

In hospitals with a high volume of procedures,the proportion of deep incisional and organ/spaceSSIs (31%) was greater than the proportion frommedium-volume facilities (10%) and low-volumefacilities (11.1%). The proportion of each SSI typefor hospital by volume of procedure is shown inFigure 9.

Discussion

This paper describes a methodology for SSI surveil-lance that has been tested in 12 UK hospitals. It hasproved capable of continuous operation over time,with some pilot hospitals having run this surveil-lance for over two years. The low infection ratesassociated with this category of surgery emphasizethe need for long-term data collection if significantvariations are to be identified and new interven-tions assessed. Work needs to be undertaken todetermine whether other analytical approaches,such as the use of statistical process control charts,could further assist in the identification of outliersboth between and within hospitals when used inconjunction with continuous data collection.

Further work will also be required to developrobust systems for quality assurance of the datacollected by different units. This will need to coverassessments of the completeness of inclusion ofpatients undergoing the procedures subject tosurveillance, the adequacy of form completion,and the degree to which clinical diagnoses ofinfections conform to the agreed definitions.However, it is reassuring to note that the overallinfection rates observed in this project were in linewith those recorded in other national surveillancesystems.13,14 In addition, the results have high-lighted several issues worthy of more detailedinvestigation; a key objective of surveillance ofthis kind. These include the validity and possiblereasons for the interhospital variation and rates ofinfection associated with particular surgeon cat-egories, such as investigating whether theincreased risk associated with procedures carriedout by locum clinicians was associated with lesserskills, experience and training of the locums oras a result of unfamiliarity with standard operat-ing procedures in different hospitals. The resultshave also highlighted aspects of the impact of SSIson patient outcomes by illustrating the increasedlength of hospital stay associated with SSIs and thelarge proportion of potentially serious deep-seated

Figure 7 The number of hip and knee prostheses, fixation of fractures and hemi-arthroplasties, and the percentage ofsurgical site infections (SSIs) by hospitals participating in the Department of Health orthopaedic SSI pilot surveillancescheme. 95% confidence intervals are shown for SSIs. Shaded bars, total procedures; open bars, total SSIs; &,percentage of SSIs.

Clinician-led surgical site infection surveillance of orthopaedic procedures 209

infections. The impact of the surveillance pro-gramme on clinical outcomes will take some time toassess, but the results to date have emphasized theessential requirement of long-term data collection.

The methodology devised relies on substantialclinical involvement from a wide range of disci-plines, with a support role from specialist infectioncontrol teams. Information on this is publishedelsewhere.11 The provision of a database tohospitals and the regular feedback of data for

Figure 8 The average number of hip and kneeprostheses, fixation of fractures and hemi-arthroplastiesreturned for surveillance per month by hospitals partici-pating in the Department of Health orthopaedic surgicalsite infection pilot surveillance scheme.

local report production are also designed toimprove timeliness and relevance of the data tohospital staff as well as their sense of responsibility.The local surveillance co-ordinators can producereports tailored to the requirements of particularhospitals or units.

The core data set allows the production of risk-adjusted rates, which in time will allow improved

Figure 9 The proportion of superficial, deep andorgan/space surgical site infections (SSIs) by volume ofprocedures performed in hospitals participating in theDepartment of Health orthopaedic surgical site infectionpilot surveillance scheme. Solid bars, superficial SSIs;open bars, deep SSIs; shaded bars, organ/space SSIs.

M. Morgan et al.210

comparisons between surgical units in differenthospitals, and surgeons, and will provide trend datafor a specific hospital, which can be used to monitorthe success of interventions. The use of CDC NNISdefinitions enhances the capacity for comparisonsinternationally, since these definitions are used inmany other countries around the world15–18 andhave been adopted by the European Union Hospitalsin Europe Link for Infection Control throughSurveillance III SSI surveillance programme.

Several problems were encountered whilst intro-ducing the surveillance in individual hospitals. Theloss of key individuals during the initial phasecritically impaired implementation in two hospi-tals, and emphasizes the need for incorporation ofhospital infection surveillance programmes intoroutine hospital activities. The support from con-sultant orthopaedic surgeons was variable and, onoccasions, required tact and persistence to gaintheir full and continued support. It is hoped that thesuccess of the programme to date will encourageincreasing support from this group. There was alsosome difficulty in the organization of form com-pletion at ward level.

For long-term viability and extension of thesurveillance described, such a programmeneeds to be integrated within the clinicalgovernance activities within a trust. This shouldensure staff commitment, the necessary admin-istrative and organizational support, the use ofdata to influence practice and the avoidance ofundue reliance on key individuals. At govern-mental level, decisions have now been taken inrelation to England, Northern Ireland, Scotlandand Wales regarding the introduction of manda-tory SSI surveillance for orthopaedics tostrengthen and support the organizational com-mitment to the programme within the NHS. Thecore dataset agreed by HAISSG and described inthis paper has been adopted for use as thestandard for all UK national orthopaedic SSIprogrammes.

Acknowledgements

The authors acknowledge the support receivedfrom clinical, infection control, epidemiological,IT and administrative colleagues. Encouragementwas also received from the Department ofHealth and the three devolved health adminis-trations. We thank the HAISSG SSI subgroup forthe expertise and time they devoted to thisproject, namely Professor Sean Hughes, Dr TonyHoward, Carole Fry, David Howell, Will Niblett,

Dr Judith Richards, Dr Georgia Duckworth, MsJenny Wilson, Dr Edward Smyth, Mr John Black,Dr Ahilya Noone, Mr John Nolan, Professor JamesNixon, Professor Jimmy Hutchinson, Dr FrankBone, Dr Ian Gould, Dr Nick Looker, Dr AlisonHolmes, Dr Richard Gair and Dr Deidre Lewis.

Appendix: Definitions used fororthopaedic SSI surveillance

Types of anaesthesia

General anaesthesiaAdministration of drugs or gases that enter thegeneral circulation and affect the central nervoussystem to render the patient pain-free, amnesiac,unconscious and often paralysed with relaxedmuscles.

Local anaesthesiaSubcutaneous or deeper infiltration with analgesicsolutions to produce localized areas of loss ofsensation.

Regional anaesthesiaUse of local anaesthetic solution(s) to producecircumscribed areas of loss of sensation includingnerve blocks, spinal, epidural and field blocks.

Types of diagnosis

OsteoarthritisPrimary degenerative disease of the hip or kneejoint or secondary degeneration caused by con-ditions such as trauma.

Inflammatory arthritisDegenerative disease of the hip or knee joint causedby inflammatory processes, such as rheumatoid orpsoriatic arthritis.

Hip fractureFracture of the trochanteric or cervical region ofthe proximal femur.

Other diagnosisDiagnoses that do not fall into the above categories.

Specific site of organ/space infections

Specific sites are assigned to organ/space SSI tofurther identify the location of the infection.

Clinician-led surgical site infection surveillance of orthopaedic procedures 211

Joint or bursa infectionsJoint or bursa infections must meet the followingapplicable criteria:

patient has organisms cultured from joint fluid orsynovial biopsy;

patient has evidence of joint or bursa infectionseen during a surgical operation or histopatholo-gical examination; or patient has at least two of the following signs andsymptoms with no other recognized cause: jointpain, swelling, tenderness, heat, evidence ofeffusion or limitation of motion.

And at least one of the following:

organisms and white blood cells seen on Gramstain of joint fluid;

positive antigen test on blood, urine or joint fluid; cellular profile and chemistries of joint fluidcompatible with infection and not explained byan underlying rheumatic disorder; or radiographical evidence of infection, e.g. abnor-mal findings on x-ray, computerized tomographic(CT), magnetic resonance imaging (MRI) orradiolabelled scans (gallium, technetium, etc.).

Osteomyelitis infectionsOsteomyelitis infections must meet at least one ofthe following criteria:

patient has organisms cultured from bone;

patient has evidence of osteomyelitis on directexamination during a surgical operation orhistopathological examination; or patient has at least two of the following signs andsymptoms with no other recognized cause: fever(O38 8C), localized swelling, tenderness, heat ordrainage at suspected site of bone.

And at least one of the following:

organisms cultured from blood;

positive blood antigen test on blood (e.g. H.influenzae, S. pneumoniae); or radiographical evidence of infection, e.g. abnor-mal findings on x ray, CT, MRI or radiolabelledscans (gallium, technetium, etc.).

BothInfections must meet the criteria for both joint orbursa infections and osteomyelitis infections above.

Infection date

The date when the first clinical evidence of the SSI

appeared or the date the specimen used to make orconfirm the diagnosis was collected, whichevercomes first.

Clinician types

ConsultantA hospital doctor who has completed higherspecialist training in their chosen field and holds apermanent consultant position at a UK hospital.

Specialist registrarA hospital doctor undertaking higher specialisttraining of 4–6 years, depending on the specialty.

Senior house officerA hospital doctor undertaking basic specialisttraining of 2–3 years, depending on the specialty.

Non-consultant career gradeA hospital doctor with a permanent contract, butwho does not hold a consultant post. The doctormay or may not have completed their higherspecialist training. This encompasses staff grades,associate specialists and trust grades.

References

1. Wilson JA, Ward VP, Coello R, Charlett A, Pearson A. A userevaluation of the Nosocomial Infection National SurveillanceSystem: surgical site infection module. J Hosp Infect 2002;52:114—121.

2. National Audit Office. The management and control ofhospital acquired infection in acute NHS trusts in England.Report by the Controller and Auditor General. London: TheStationery Office; 2000.

3. The management and control of hospital acquired infectionin acute NHS trusts in England. Public Accounts CommitteeReport. London: House of Commons; 2000.

4. Plowman R, Craves N, Griffin M, et al. The socioeconomicburden of hospital acquired infection. London: Public HealthLaboratory Service; 2000.

5. Schneeberger PM, Smits MHW, Zick REF, Wille JC. Surveil-lance as a starting point to reduce surgical-site infectionrates in elective orthopaedic surgery. J Hosp Infect 2002;51:179—184.

6. Whitehouse JD, Friedman D, Kirkland KB, Richardson WJ,Sexton DJ. The impact of surgical site infections followingorthopaedic surgery at a community hospital and a universityhospital: adverse quality of life, excess length of stay andextra cost. Infect Control Hosp Epidemiol 2002;23:183—188.

7. National Audit Office.Hip replacements: getting it right firsttime. Report by the Controller and Auditor General. London:The Stationery Office; 2000.

8. Smyth ETM, McIlvenny G, Barr JG, Dickson LM, Thompson IT.Automated entry of hospital infection surveillance data.Infect Control Hosp Epidemiol 1997;18:486—491.

9. National Nosocomial Infections Surveillance (NNIS) manual.

M. Morgan et al.212

Atlanta: U.S. Department of Health and Human Services,Public Health Service, Centers for Disease Control andPrevention; 1999.

10. Horan TC, Gaynes RP, Martone WJ, et al. CDC definitions ofnosocomial surgical site infections, 1992: a modification ofCDC definitions of surgical wound infections. Infect ControlHosp Epidemiol 1992;13:606—608.

11. Hogg S, Baird N, Richards J, et al. Developing surgical siteinfection surveillance within clinical governance. ClinGovern Int J 2005;10:24—36.

12. Haley RW, Culver DH, Morgan WM, et al. Identifying patientsat high risk of surgical wound infection: a simple multi-variate index of patient susceptibility and wound contami-nation. Am J Epidemiol 1985;121:206—215.

13. National Nosocomial Infections Surveillance System report,data January 1992 to June 2003. Am J Infect Control2003;31:481—498.

14. Nosocomial Infections National Surveillance Service. Sur-veillance of surgical site infection in English hospitals 1997–2001. London: Public Health Laboratory Service.

15. Ronveaux O, Mertens R, Dupont Y. Surgical wound infectionsurveillance: results from the Belgian hospital network. ActaChir Belg 1996;96:3—10.

16. McLaws ML, Taylor PC. The hospital infection standardisedsurveillance (HISS) programme: analysis of a two year pilot.J Hosp Infect 2003;53:259—267.

17. Sodano L, Sefafini C, Petrosillo N, et al. Active surveillanceof hospital infection in a cardiac surgery unit: the experienceof San Camillo-Forlanini Hospital, Rome. Bollettino Epide-miologico Nazionale, Notiairio ISS 2001;14:n3.

18. Abramczyk M, Carvalho W, Carvalho E. Nosocomial infectionin a pediatric intensive care unit in a developing country.Braz J Infect Dis 2003;7:375—380.