Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

New national strategies for hospital infection control : a critical evaluation Birnbaum, David Wayne 1992

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1992_spring_birnbaum_david_wayne.pdf [ 3.29MB ]
JSON: 831-1.0076835.json
JSON-LD: 831-1.0076835-ld.json
RDF/XML (Pretty): 831-1.0076835-rdf.xml
RDF/JSON: 831-1.0076835-rdf.json
Turtle: 831-1.0076835-turtle.txt
N-Triples: 831-1.0076835-rdf-ntriples.txt
Original Record: 831-1.0076835-source.json
Full Text

Full Text

New National Strategies for Hospital Infection Control: A Critical EvaluationbyDavid Wayne BirnbaumB.A., The University of California at Berkeley, 1966M.P.H., The University of Minnesota, 1974A THESIS SUBMITTED IN PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIESINTERDISCIPLINARY STUDIES PROGRAMMEEPIDEMIOLOGY, STATISTICS, MICROBIOLOGY and INFECTIOUS DISEASES]We accept this thesis as conformingto the required standardTHE UNIVERSITY OF BRITISH COLUMBIAFebruary 1992Q David Wayne Birnbaum, 1992Signature(s) removed to protect privacyIn presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood. that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature).t/ c,F’4I’ •‘ 1hr,5;S‘5’9 5rj /*V’eDepartment of pti#v1’ C’ 44’vf/1 7#2(r-The University of British ColumbiaVancouver, CanadaDate “9 49—DE-6 (2/88)Signature(s) removed to protect privacyPERMISSION TO USE COPYRIGHTED MATERIALPermission is hereby granted to (Name of author of thesis) Dav:Ld. Wayne Birnbaum(Title of thesis) New National StrategiLes for Hospital Infection Control:A Critical EvaluationUniversity of British Columbia (Degree)and to the National Library of Canada to reproduce(Figure/page numbers) text, & figures, pages 465—472in (Title of article/book) Adoption of Guidelines for UniversalPrecau;tions and Body Substance Isolation in CanadianAeute—Care Hospitals(Name, issue number, and year of journod) INFECT CONTROL HO:SPEPIDE1VITOL, 11(9), 1990(Place, publisher and year of book)Return this form to: Library— Special Collections— Thesis Supervisor1956 Main MallUniversity of British ColumbiaVancouver, B.C.CanadaV6T 1Z1The National Library will lend or sell copies of the microfilm. All other publication rights arereserved.Ph.D. (Year) 1992which will appear in this thesis.(Signatures)AddressSignature(s) removed to protect privacyPERIs’IISSION TO USE COPYRIGHTED MATERIALPermission is hereby granted to (Name of author of thesis) David Wayne B irnbauin(Title of thesis) Nw NatiniStrategies for Hospital Infection Control:A Critical EvaluationUniversity of British Columbia (Degree) Fit. D. (Year) 1992and to the National Library of Canada to reproduce(Figure/page numbers) text & figures, pages 319—326in (Title of article/boolç) Epidemiologic Typing Systems forCoagulase—Negative Staphylococci(Name, issue number, and year of puma!) INFECT CONTROL 110SFEPIDEMIOL, 12(5), 1991(Place, publisher and year of book)Library— Special Collections— Thesis Supervisor1956 Main MallUniversity of British ColumbiaVancouver, B.C.CanadaV6T 1Z1The National Library will lend or sell copies of the microfilm. All other publication rights arereserved.which will appear in this thesis.(Signatures)AddressReturn this form to:Signature(s) removed to protect privacyIn presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(SigiI4-1DepartmenU of 4I4( #Ap1g.The University of British ColumbiaVancouver, CanadaDate 72DE-6 (2/88)Signature(s) removed to protect privacyEffectiveness of Infection Control StrategiesAbstract:Isolation of those ill with contagious disease has been a fundamentalinfection control concept for hundreds of years. However, recent studies suggestthat fewer than 50% of health—care workers comply with their hospital&isolation precaution policies and that efficacy of some of those policies isquestionable. In response, two new systems, based upon fundamentally differentgoals, were promoted. The Centers for Disease Control, prompted by health—careworker& concerns about occupational risk of human immunodeficiency virus (HIV)from a growing number of patients with acquired immunodeficiency diseasesyndrome (AIDS), issued formal guidelines in 1987. This formed the basis forUniversal Precautions (UP), a unifying strategy for precautions with all patientsregardless of diagnosis intended to reduce risk to hospital staff members. Alsoin 1987, one hospital issued guidelines for Body Substance Isolation (BSI),hygienic precautions to be used with all patients based on recognition thatcolonized body substances are important reservoirs for cross—infection to bothpatients and staff members. These new strategies have been promoted widely,but there have been no formal assessments to reconcile controversies theyraised nor to confirm their effectiveness. Further, necessary assessment toolshave not been validated.This thesis provides new tools and new information to address three vitalquestions: Have hospitals adopted Universal Precautions or Body SubstanceIsolation? Do their staff members use the new system of precautions in dailypractice? Has reliable use of a new system led to decreased risk of infection?A confidential mailed survey of all acute—care Canadian hospitals wasconducted to measure rates of guideline receipt and adoption. It also obtainedinformation on motivations for and perceived effectiveness of strategies adopted.Page iiEffectiveness of Infection Control StrategiesA self—selected group of responding hospitals subsequently participated instandardized covert observation of their nurses infection control practices, thenhad the observed nurses complete a test examining their knowledge and beliefs.Employee health records were also examined to determine whether needlestickinjury rates had changed since adoption of a new infection control strategy.Most Canadian hospitals adopted and modified new strategies based uponreasonable but unproven extensions of logic to protect health—care workers fromHIV. 74% claimed UP (65%) or BSI (9%) but only 5% of 359 claiming UP and 0of 50 claiming BSI adopted all policies expected. Many hospitals had notreceived key guideline publications. Guideline source, hospital size, and othervariables were significantly associated with receipt. Nurses in 35 hospitalswere observed to wear gloves during only z60% of procedures in which glovingwas expected; rates varied widely among hospitals. Direct examination of sharpsdisposal containers confirmed compliance with a policy to not recap used needles(taken as recapping rate of 25%) in only 47% of 32 hospitals. Paired analysisof needlestick injury rates in 11 hospitals during comparable 90—day periodsbefore versus after implementing UP/BSI showed no significant difference. 489nurses completing a written test achieved their highest scores and leastdiscordance among questions regarding procedural issues established long beforeUP/BSI, and lower scores or greater discordance on UP/BSJ concepts ofphilosophy, risk recognition and newer procedures. Positive correlation betweenknowledge and practice was not evident. UP and BSI now mean different thingsin different hospitals and have not been effective in harmonizing health—careworkers’ infection control practices. Carefully standardized assessment methodsare needed to guide their evolution to cost—effectiveness.Page iiiEffectiveness of Infection Control StrategiesTable of Contents:AbstractList of Tables ViList of Figures viiiAcknowledgements ixPreface XI. Introduction: Strategies for Hospital Infection ControlEvolution of New Strategies for Infection ControlUniversal Precautions and Body Substance IsolationCurrent Knowledge of Hospital Infection Control PracticesScope and Limitations of Thesis ResearchII. Have Hospitals Accepted New Infection Control Strategies? 8Receipt and Adoption of Guidelines by Canadian HospitalsMaterials and MethodsResultsResponse Rate in the Survey of Canadian HospitalsReceipt of Published Guidelines by Canadian HospitalsAdoption of Guidelines for UP and BSIRationale for Adopting a New StrategyOther Perceptions— Disposal of SharpsOther Perceptions— Knowledge of Costs and BenefitsSummary of Major FindingsIII. Have Hospital Staff Members Accepted the New Strategies? 27Infection Control Practices of Critical Care NursesMaterials and MethodsResultsResponse RateUse of GlovesHandling of Used Disposable NeedlesSummary of Major FindingsKnowledge and Beliefs of Critical Care NursesMaterials and MethodsResultsResponse RatesNurses Knowledge and Belief ScoresRelationship Between Knowledge and PracticeSummary of Major FindingsPage ivEffectiveness of Infection Control StrategiesTable of Contents (cont.):IV. Have New Infection Control Strategies Reduced Infection Risks? 49Epidemiologic Assessment of Risk to Hospital Staff— Needlestick InjuriesMaterials and MethodsResultsSummary of Major FindingsEpidemiologic Assessment of Risk to Hospital PatientsTracing the Sources of Infection by Microbiologic MethodsV. Discussion 55Validity, Reliability, Power and Generalization of This ResearchHave Universal Precautions or Body Substance Isolation BeenEffective as Harmonizing Strategies for Hospital Infection Control?Recommendations for Improving the Research ToolsVI. Conclusions and Recommendations 80Recommendations for Improving New Infection Control StrategiesVII. Bibliography 85VIII. AppendicesAppendix 1: Conceptual Models 93Comparison of Isolation StrategiesInfection Control Goals, Objectives and AssumptionsAppendix 2: Hospital Survey Questionnaire Forms 96Appendix 3: Nurse& Knowledge and Belief Test Forms 103Appendix 4: Efficacy of MIDI for Epidemiologic Typingof Staphylococcus epidermidis 113Page vEffectiveness of Infection Control StrategiesList of Tables:Table 1: Evaluation Criteria 4Table 2: Policy Recommendation versusInfection Control Strategy 10Table 3: Survey Response Frequency and Rateby Bed Size Group and Province forNon—Specialty Acute—Care Canadian Hospitals 16Table 4: Receipt of Guidelines by Source and Hospital Size 17Table 5: Hospital Size vs ICP Staffing (in FTE) 19Table 6: Percent of Hospitals Claiming Adoptionof New Strategies 19Table 7: Adoption of Guidelines by Hospital Size 21Table 8: Percent (Number) of Hospitals Adopting PoliciesExpected under UP, BSI and Traditional Strategies 22Table 9: Infection Control Strategy Claimed versus Practiced 22Table 10: Predominant Method for Transporting Used Needlesfrom the Point of Use to the Point of Disposal,by Hospital Size According to Number of Beds 25Table 11: Perceived Effectiveness of Strategies To AvoidRecapping of Used Needles 25Table 12: Nursing Care Procedures Observed 30Table 13: Frequency of Recapped Needles in Disposal Containers 33Table 14: Characteristics of Participating Hospitals 37Table 15: Relative Ranking of Knowledge Test Section Scores 38Table 16: Spearman Correlation Coefficients BetweenTest and Practice Scores 47Table 17: Canonical Variate Coefficients and Correlations 48Table 18: Typing Methods for Coagulase—Negative Staphylococci 54Table 19: Magnitude of a— and 13—error for Hypothesis #3 59Table 20: Status of UP/BSI in Canadian Acute—CareHospitals, 1990 76Page viEffectiveness of Infection Control StrategiesList of Tables: (cont.)Table 21: Comparison of Isolation Strategies 94Table 22: Infection Control Program Objectives and Assumptions 95Table 23: Characterization of Epidemiologically—RelatedIsolates by Lab of Origin 125Table 24: Summary of MIDIs Typing Performance 126Page viiEffectiveness of Infection Control StrategiesList of Figures:Figure 1: Hospitals Motivations for Adopting UP or BSI 23Figure 2: Frequency of Observed Glove Use DuringTypes of Nursing Care 32Figure 3: Effectiveness of Hospitals’ In—Service Program 39Figure 4: Tukey Sum—Difference Graph of CorrectResponse Frequencies 40Figure 5: Nurses’ Overall Test Score, by Site(Ordered by Region and Size) 42Figure 6: Nurses’ Overall Test Score, byPersonal History Claims 45Figure 7: Spearmans Correlation Coefficient (of Knowledgeand Practice Scores) vs Minumum Number of PracticeObservations Required for Inclusion in Analysis 48Figure 8: Needlestick Injury Rates in 11 HospitalsProviding Paired Data 52Figure 9: Magnitude of Confounding in Nurses’ Test Scores 61Figure 10: Normal Probability Plot— Distribution Rejectedby the Kolmogorov—Smirnov Test 62Figure 11: Distribution of Phi Coefficients for 11 PairedResponses, by Site, in 22 Sites Returning 10 Tests 64Figure 12: Section of MIDI Dendrogram ContainingGroup 5 Isolates 127Figure 13: Joining Distance of Repeated Assays 128Page viiiEffectiveness of infection Control StrategiesAcknowledgement:Tasks of this magnitude cannot be completed without considerable encouragementand support. First and foremost, my family has been a partner in the endeavor:I would not have started nor lasted without you Connie, Dad, Jeremy and Jenny.Nor would I have reached this far without being stimulated and challenged byan exceptional group of teachers and colleagues. Dr. Chow, Dr. Kelly, Dr. Noble,Dr. Mathias, Dr. Schechter and Dr. Schulzer, I am in your debt. This work couldnot have been completed without the diligent support of infection controlpractitioners in participating hospitals throughout Canada, and was greatlyfacilitated by endorsement from the Canadian Hospital Association as well as theassistance of friends and colleagues in preparing and translating forms. Lastbut not least, I thank the Government of Canada for funding. This researchwas supported in part by the National Health Research and Development Programthrough a National Health Fellowship.Page ixEffectiveness of Infection Control StrstegiesPreface:Parts of this work have been published as:Birnbaum DW, Schuizer M, Mathias RG, Kelly M, Chow AW. Receipt andAdoption of Universal Precautions and Body Substance Isolation Guidelinesin Canadian Hospitals. Abstract #L22. 90th Annual Meeting, AmericanSociety for Microbiology, May 1990.Blrnbaum DW, Schuizer M, Mathias RG, Kelly M, Chow AW. Receipt andAdoption of Universal Precautions and Body Substance Isolation Guidelinesin Canadian Hospitals. Abstract #C6, 3’rd Decennial InternationalConference on Nosocomial Infections, Atlanta, Georgia, July 1990Birnbaum D, Schulzer M, Mathias RG, Kelly M, Chow AW. Adoption ofGuidelines for Universal Precautions and Body Substance Isolation inCanadian Acute—Care Hospitals. INFECTION CONTROL AND HOSPITALEPIDEMIOLOGY 1990;11(9):465—472Birnbaum D, Schuizer M, Mathias RG, Kelly M, Chow AW. New infectioncontrol strategies in acute care. DIMENSIONS IN HEALTH SERVICE1 990;May:26—29Birnbaum D, Schuizer M, Mathias RG, Kelly M, Chow AW. Needlestick injury:do preventive measures work? DIMENSIONS IN HEALTH SERVICE1990;Nov.:29—32Birnbaum D, Kelly M, Chow AW. Epidemiologic Typing Systems for Coagulase—Negative Staphylococci. INFECTION CONTROL AND HOSPITAL EPIDEMIOLOGY1991; 12(5):3 19—326The research concepts, designs and execution were the original work of DavidBirnbaum; co—authors provided valuable constructive criticism, supervision ofdirected readings and logistical support.Page xEffectiveness of Infection Control StrategiesI. Introduction:Evolution of New Strategies for Infection ControlUniversal Precautions and Body Substance IsolationSegregation and isolation of those ill with contagious diseases has been afundamental part of infection control for hundreds of years.1 Over the lastthirty years. a number of monographs and books from the American Public HealthAssociation, American Hospital Association, the Centers for Disease Control (CDC)in Atlanta, and the Laboratory Centre for Disease Control (LCDC) in Ottawacodified specific isolation techniques for use in hospitals based upon thisheritage and various extensions of logic. However, relatively few of today’shospital—acquired (nosocomial) infections are contagious or even communicable;most, in fact, are latrogenic or endogenous (originating from treatments or apatients own flora). Nevertheless, isolation of patients diagnosed orpresumed as infectious rapidly became a mainstay of hospital infection controlefforts in spite of an absence of proof that isolation per se reduces the risk ofnosocomial infection.2However, in light of persistent absence of proven effectiveness for variousinfection control measures, some experts began to question their continued use.3In response, the CDC initiated a project to review and revise infection controlguidelines on the basis of expert consensus on interpretation of clinical orresearch studies. Notably, the revised CDC Guideline for Isolation Precautionsin Hospitals “did not rank the recommendations by the degree to which theyhave been substantiated by scientific data or the strength of the workinggroup’s opinion on their effectiveness or practical value.”4 There were simplytoo few studies testing the efficacy or effectiveness of recommended measures.These revised guidelines offered hospitals a choice between the traditionalPage 1Effectiveness of Infection Control Strategiescategory—specific system and a potentially more user—flexible disease—specificsystem, and eliminated the traditional Protective Isolation category based uponproof that this time—honored isolation category was ineffective.8In addition to an absence of studies demonstrating efficacy, a growingnumber of anecdotal reports and formal studies confirmed that compliance byhealth—care workers with their hospital’s policies for isolation precautionsfrequently was poor.6 Studies spanning the last ten years all suggest thatfewer than 50% of staff complied with their hospitals’ isolation precautionpolicies.6 Over the last five years, two new systems were proposed based uponfundamentally different goals.In 1987. CDC consolidated its 1982—1986 recommendations in response tohealth—care workers’ concerns about acquiring human immunodeficiency virus(WV) infection from patients with acquired immunodeficiency disease syndrome(AIDS).7 This formed the basis of Universal Precautions (UP) for blood andbody—fluids, a unifying strategy for precautions with all patients, regardless ofdiagnosis, intended to reduce risk to hospital staff members. It deals withblood—borne pathogens only, and is intended as a supplement to the CDCcategory— or disease—specific isolation system. Also in 1987, one hospitalpublished its own recommendations for hygienic precautions to be used with allpatients based on recognition that colonized body substances were importantreservoirs for cross—infection to both other patients and staff members.8 BodySubstance Isolation (BSO, developed In response to bacterial cross—infectionproblems in critical care units, is intended to reduce cross—infection risk withall pathogens, not just blood—borne agents, for all patients and health—careworkers. BSI replaces the CDC category— or disease—specific systems.UP has been promoted widely by the CDC, LCDC, the Association of StatePage 2Effectiveness of Infection Control 5trate7iesand Territorial Public Health Laboratory Directors and others. BSI was promotedin a videotape distributed to hospital administrators by the American HospitalAssociation. Endorsements by such influential agencies have not consideredthese new systems to be equivalent.9 10 Subsequent UP guideline revisions’1have neither reduced this controversy nor received universal acceptance.’2Current Knowledge of Hospital Infection Control PracticesAlthough numerous guidelines have been published, only one descriptivestudy reports the extent of their receipt and adoption among hospitals in theUnited States.’3 Patterns of receipt among Canadian hospitals are unknown.Even though at least 80% of all US hospitals had received and adopted CDCguidelines, compliance rates by their staff members with these measures, asnoted above, tended to be under 50%. Increasing concerns about the safety oftreating AIDS patients in hospital, and increasing doubts about effectiveness ofcurrent isolation precautions have brought North American hospitals to a newcrossroad.’4UP and BSI present new options and new controversies. The extent towhich guidelines for these new systems have been reviewed by individual NorthAmerican hospitals is unknown. Many infection control practitioners (ICP)believe that widespread variations In definitions and practices exist betweenhospitals, but the uniformity or quality of UP or BSI practice is alsoundocumented. Anecdotal reports suggest that UP and BSI have not succeededas unifying strategies to improve compliance, but there have been no formal,multicenter studies. Standardized tools to survey these issues have not beenvalidated. Therefore, expensive new strategies have been promoted widely fornearly five years in the absence of adequate tools or studies to document theirimplied effectiveness. Anecdotal reports, editorials and letters to editors revealPege 3Effectiveness of Infection Control Strateg2escontroversy, confusion and need for formal evaluation. Further, unprovenguidelines may be mandated in accreditation requirements and legislativeregulations.’5 This presents an urgent need for applied research in hospitalepidemiology to guide administrative and infection control decisions.Scope and Limitations of Thesis ResearchThis thesis investigates the effectiveness of UP and BSI as currentlyadopted in acute—care Canadian hospitals. It validates new survey tools andapplies survey sampling and multivariate statistical methods in multi—centerresearch. The approach involves examination of three basic criteria inoperational aspects of structure, process, output and outcome (Table 1).Table 1: EVALUATION CRITERIACRITERION STRUCTURE PROCESS OUTPUT OUTCOME1. Hospitals have All have All have All know UP costs andreceived adequate received reviewed how UP benefits areinformation to CDC/LCDC CDC/LCDC and BSI identified inmake informed guideline guideline differ each hospitaldecisions; for UP for UPHospitals adopted Coherentappropriate policies policy—setconsistent with the in placesystem adopted.2. Hospital staff Hospitals Education High staff Staff acceptunderstand and mandate provided scores on new systemaccept UP or BSI education to staff a test3. Staff adhere to Hospitals Hospitals Gloves wornrecommended UP or provide monitor for riskBSI guidelines in gloves, compliance procedurestheir daily work gowns, eye and providepractices; protection performance Used needlesand devices feed—back not recappedUse of UP or BSI to discouragepractices leads to needle recapping Fewerreduced infection needlestickrisk, injuriesafter UP orBSI adoptedPage 4Effectiveness of Infection Control StrategiesA number of important difficulties and limitations in this work must benoted. First, as implied earlier, UP and BSI are specific systems according totheir respective published documentation, but are likely to mean different thingsas implemented in different hospitals. This research studies the effectivenessof UP and BSI as unifying strategies among all acute—care Canadian hospitals; itdoes not address efficacy nor superiority in reducing infection rates by somehospitals as opposed to others. Effectiveness is defined in terms of compliance,not reduction in nosocomial infection rates, for several reasons:a) Given logistic restraints and the number of confoundingvariables associated with nosocomial infection risk, it was notfeasible to study before—and—after rate differences. Further,it would not have been possible to randomize assignment ofinfection control systems to perform a true experiment.b) Adequate microbiologic methods for tracking certain nosocomialpathogens are not available to support such Part of this thesis research investigates thepotential of a novel microbiologic method that may provide suchsupport in the future.c) As will be discussed later, the driving force motivating adoptionof new systems has been a perceived occupational risk of HIVinfection, not a desire to reduce risk of nosocomial infection ingeneral.d) Finally, since it is not yet documented whether UP or BST havebeen adopted effectively, it is premature to focus evaluation on along—term outcome (infection risk reduction) as opposed to short—term measures (improved compliance with reasonable policies).Page 5Effectiveness of Infection Control StrategiesSecond, the representative nature of hospitals included in the study must beviewed with caution. UP or BSI were promoted widely before this researchbegan. Individual hospitalst implementation costs and constraints, perceivedneeds and potential benefits are likely to vary widely as a function of hospitalsize and location. Infection control systems cannot be assigned in a randomizedor blinded manner, and hospitals cannot be forced to participate. Likelihood,direction and extent of self—selection bias must be considered carefully. Thestudy design must also achieve a balance between sample sizes large enough togenerate convincing statistical power yet small enough to be practical.Third, target events must be selected with regard to reliability of detectionand clear importance in testing relevant hypotheses. Since validated surveytools pertinent to this research were not available, an important part of thework involved development and validation of standardized questionnaires andsampling plans. Further, since there is no criterion ‘goldt’ standard againstwhich to validate these tools, methods to achieve face, consensual and contentvalidity are emphasized.17 Since opportunities to obtain repeated measurementsin participating hospitals were very limited, other approaches to ensuringreliability are used.18 19Finally, this work cannot provide a definitive answer as to whether UP orBSI can succeed. Ideally, infection control guidelines evolve as a function ofexperience and evaluation. The nature of patients, pathogens and procedures inhospitals is dynamic; precautions that have been in force may fall from grace asnew information and situations add to our knowledge. Inadequacies oftraditional isolation systems became well—recognized. This led to a naturalexperiment: New isolation systems were promulgated. My work providesnew tools and new information to indicate what the new systems have achieved.Page 6Effectiveness of Infection Control StrategiesThe underlying null hypothesis being tested is that UP and BSI have notachieved a uniform standard of effective infection control practice among acute—care hospitals in order to protect health—care workers or their patients. This isexamined in several ways through survey of administrative practices, covertobservation of workers practices, formal testing of workers’ knowledge andbeliefs, self—reported disease—exposure histories and review of employee healthincident reports. These data are used to test a number of subordinatehypotheses relating to the elements listed in Table 1. Appendix 1 providesconceptual models for hospital infection control strategies. The followingchapters discuss hypotheses, methods and results as well as provide descriptiveinformation about participating hospitals, their policies and their staff.Page 7Effectiveness of Infection Control StrategiesII. Have Hospitals Accepted New Infection Control Strategies?Receipt and Adoption of Guidelines by Canadian HospitalsUP and BSI involve radical departure from long—standing reliance uponisolation precautions, and they imply potentially large cost increases, so it isimportant that hospital administrators understand their ramifications. UP andBSI both stipulate that all patients are to be treated alike, thus eliminating aneed for traditional “biohazard’ warning labels to prompt special isolationprecautions, but they differ from each other in several important aspects. UPfocuses on minimizing risk from blood—borne infections, primarily hepatitis B andliv, while BSI focuses on minimizing cross—infection risk with all pathogens forboth patients and staff. UP applies only to those body fluids associated withhepatitis B or HIV transmission while retaining traditional isolation categories toprotect against transmission of other diagnosed or suspected infections. It is,in fact, universal application of the traditional CDC category “Blood/Body FluidPrecautions’. BSI applies to all body substances and eliminates all othertraditional isolation categories except that for infections spread by an airborneroute. It promotes use of protective measures against risks identified on aprocedure—specific rather than diagnosis—specific basis. BSI provides a detailedguide to patient placement based on evaluation of hygiene, cooperation, anddisease susceptibility of patients and staff. BSI also originally emphasizedgloving as an alternative rather than a supplement to handwashing, a viewsubsequently challenged20 21Rates of receipt and adoption for infection control guidelines by Canadianhospitals had not been reported previously. Therefore, a first step in examiningthe effectiveness of UP and BSI as unifying strategies was to determine whetherCanadian hospitals received and reviewed appropriate publications that definePage 8Effectiveness of Infection Control Strategiesthese new strategies. Next, beyond simply estimating the proportion of hospitalsclaiming adoption of UP or BSI. the specific published guideline statementsadopted by’ each hospital were compared with their claimed adoption of UP or BSIto provide a measure of how well these new unifying strategies have harmonizedhospital infection control policies. Specific objectives of this administrativesurvey were to determine, by size groups, hospitals’ infection control informationresources, current extent of adoption and evaluation of UP or BSI, motivation(s)for adopting a new strategy and knowledge of its cost implications.Materials and Methods:Sampling Frame and QuestionnaireMailing addresses for all Canadian hospitals and size distribution for non—teaching non—specialty hospitals were obtained from the 1988 Canadian HospitalAssociation (CHA) Directory. Size distributions for teaching and for pediatrichospitals were obtained from the Association of Canadian Teaching Hospitals andthe Canadian Association of Paediatric Hospitals membership lists respectively.No data were available regarding the size distribution of 153 specialty hospitals.In March 1989, advance notice letters explaining the study’s purpose andassurance of confidentiality were mailed to the administrator of each acute—carehospital listed in the 1988 CHA membership directory. Questionnaires withcover letter and prepaid return envelope were sent two weeks later.Instructions requested that administrators forward the questionnaire to theirinfection control program director. A prompt for reply was published in CanadaDiseases Weekly Report (CDWR) (April 15, 1989). A second prompt, usingidentical forms, was mailed to non—responders in July, 1989.The questionnaire, a 3—page self—report form, examines consideration andadoption of specific, major, referenced infection control guidelines andPage 9Effectiveness of Infection Control Strategiesrecommendations. It had been pretested and modified for clarity before use, andwas provided in English and French versions (see Appendix 2). A combinationof open—text short responses and closed—response check—off’ boxes was used.Specific terms (i.e.: Universal Precautions, Body Substance Isolation) werepurposely left undefined to test respondents’ knowledge. Questions used todistinguish between UP, BSI and Traditional strategy related to types of bodyfluids perceived as potentially hazardous; to discontinuation of special“isolation’ warning labels for soiled linen, specimens and waste; and to routineglove use and handwashing. The strategy claimed in open—text responses wascompared with objective assignment to the UP, BSI or Traditional strategy, basedupon the specific combinations of fundamental policy adoptions recorded inresponse to closed—response questions (Table 2).TABLE 2: Policy Recommendation Versus Infection Control StrategyINFECTION CONTROL STRATEGY:POLICY RECOMMENDATION: UP BSI TRADITIONAL1. No special warning labels on laboratoryspecimens, waste or soiled—linen bags + + —2. Gloves as alternative to handwashing — + —3. Private Room for staphylococcal pneumonia ± — ±4. Precautions limited to specific body fluids + — —The effect of not adopting a policy to train all staff was also examined.Responding and nonresponding hospitals were compared on the basis of size,location (province, setting), and accreditation status to assess the impact ofself—selection bias.Coding of ResponsesCompleted questionnaires were marked with unique sequential numbers uponreceipt, and these data were stored on computer. Databases were maintainedPage 10Effectiveness of Infection Control Strategieswith an interactive program written in FoxBASE+ (version 2.00, Fox Software,Perrysburg, Ohio); FoxDOC (Fox Software, Perrysburg, Ohio) was used indocumenting and debugging the code. Replies with any one of the three pagesleft blank were not used. Missing responses were coded differently fromnegative responses to questions concerning guideline receipt and review; policyadoption was coded only if an explicit “yes” was indicated (“no’ or non—responseto the single items, or any indication that adoption was planned for a futuredate rather than actually achieved by the survey date were coded as not havingadopted a guideline).Grouping by number of beds was assigned using the same ranges as CHAand Statistics Canada: 1—24, 25—49, 50—99, 100—199, 200—299, 300+; and thelatter further divided into 300—499 vs 500+ to facilitate comparison with asurvey of US hospitals by Celentano and colleagues.13 Total bed number mayinclude long—term as well as acute beds; to be included in the sample, aninstitution must provide acute care service(s). Coding of claimed infectioncontrol strategy (UP, BSI, “Traditional”) and rationale for that choice was basedupon responses to open—text questions. If no indication was given of adoptingone of the newer strategies, the response was coded as “Traditional”. Anindication of formal or informal use of UP or BSI was coded as adoption;indication of consideration rather than use per se was coded as “Traditional”.Several Ontario hospitals indicated using “Body Substance Precautions” ratherthan “Body Substance Isolation”; if their response mentioned retaining category—specific isolation, then they were coded as UP, otherwise as BSI. Some hospitalsindicated using both UP and BSI, which is self—contradictory; they appear to beusing UP without accepting limitations upon which body fluids are considered“infectious” and were therefore coded as UP. Reasons given by each site forPage 11Effectiveness of Infection Control Strategiesadopting UP or BSI were coded under one or more of the following motives:1) Staff Protection, 2) Patient Protection , 3) Compliance with a standard ofpractice expressed in expert guidelines, and 4) Miscellaneous.Two survey questions were asked in relation to discouraging a contra—indicated practice of recapping of used needles. One asked whether therespondent felt that at least half of ward staffs still recap used needlesregardless of policy. A second asked what type of disposal containers wereprovided and how needles were transported from the point of use. The extentto which Canadian hospitals have adopted practices and protective equipmentthat may be effective in reducing needlestick injury risk had not been reportedpreviously. Responses to the second question were coded as either:a) Bedside containers — Point of Use” disposal, in container either mountedin each patient room or small special units hand—carried to each patient;b) Carry unsheathed — used needles to be carried unsheathed by hand or onstandard trays (e.g.: K—basin) to disposal containers located in utilityareas and/or on medication carts;c) Recap and carry — used needles recapped, emphasizing no particulartechnique, and carried by hand or tray to central disposal containers;d) Safe Recapping — “Thimble Technique” or other one—handed method orprovision of special recapping devices for recapping prior to transport;e) Foam Stabbing Block — Used needle stabbed into small containers of rigidfoam for transport (unsheathed) to central disposal containers;f) Special Trays — Specially designed tray to support used needles safelyduring transport to central disposal containers;g) Not Specified — Predominant system not evident from description provided.If two systems were described in a given hospital, the system most immediate toPage 12Effectiveness of Infection Control Strategiesthe point of use was coded (e.g.; safe recapping and carry unsheathed in K—basins would be coded as safe recapping; foam block and bedside (wall—mounted)containers coded as foam block; etc.).Statistical AnalysisPoint estimates and 95% confidence intervals for guideline receipt rates inthe sample received were calculated using poststratification and normalapproximation to the binomial distribution. Poststratification is a surveysampling technique that adjusts for the effect of applying stratification in theanalysis rather than in the sampling step.22 Stratification refers to divisioninto subdomains (strata), in this case into hospital size groups by number ofbeds. The finite population correction factor was applied to variancecalculations. Conservative correction for non—response was also applied toconfidence intervals by assuming that all nonresponders had receivedguidelines.22 This represents an extreme case correction. Approximatedconfidence intervals were also confirmed by exact probabilities computed with P—EXACT software (Kern International) using the hypergeometric distribution withboth the classic and Miettinen’s mid—p algorithm.23 Unlike the classicalgorithm, the mid—p algorithm adds only half of the probability associated withan observed distribution to the probabilities associated with all more extremedistributions.Data extraction and simple descriptive statistics were executed with R&RRelational Report Writer (version 3)(Concentric Data Systems Inc., WestboroughMassachusetts). Cross—tabulations of responses were analyzed by log—linearanalysis, a multivariate technique that models the expected cell values in amultidimensional contingency table from information regarding a number ofcategorical variables and their interactions. Log—linear models were fitted andPage 13Effectiveness of Infection Control Strategiesevaluated by chi—square tests using SAS (version 6.03)(SAS Institute, Cary,North Carolina). SYSTAT (version 4.1) (Systat. Inc., Evanston, Illinois) was alsoused for table analysis with chi—square goodness of fit tests and for Spearman’scorrelation coefficients between guideline receipt or adoption rates and size.BMDP (BMDP Statistical Software, Los Angeles, California) was used for step—wiselogistic regression in order to examine size as an interval as well as acategorical variable and to support inclusion of other variables as predictors ofguideline receipt.Hypotheses TestedNull hypothesis #1: Canadian hospitals show no significant differences inrates for receipt of published guidelines defining UP and BSI in association withhospital size or location. The alternative hypotheses is that receipt rates differby hospital size and/or location. A related objective was to measure Canadianhospitals’ receipt rates for recent CDC or LCDC guidelines in comparison toAmerican hospitals’ receipt rates for previous CDC guidelines in order todetermine whether all hospitals receive this fundamentally important information.Null hypothesis #2: Canadian hospitals show no preference among infectioncontrol strategies (Traditional, UP or BSI). The alternative hypothesis is that amajority have adopted either a Traditional system or one of the new systems. Arelated objective was to compare each hospital’s claim of system used againstspecific policies they adopted in order to determine whether hospitalsimplemented their infection control strategy coherently and completely.ResultsResponse Rate in the Survey of Canadian HospitalsA set of mailing labels for all Canadian hospitals was obtained from CHA.The population sampled consisted of 943 acute—care hospitals after 61 labelsPage 14Effectiveness of Infection Control Strategieswere discarded for specialty hospitals listed as providing long—term care only orfor duplicate addresses (i.e.: parent corporation as well as sites operated by thecorporation). 454 responses were received after the first mailing (48% responserate), and another 125 from a second mailing for a total of 579 (61%). Amongthe 579, 10 indicated provision of long—term care only and these were excluded.Of the remaining 569 responses from acute—care hospitals, one site was not yetin operation and another 12 were excluded as insufficient completion of theforms. These exclusions occurred in the three least bed—size groups.Table 3 summarizes the hospital sampling frame and distribution ofresponses by hospital size and province. The size distribution of respondersapproximates that of all Canadian hospitals. The composition expected versusreceived for the 1—49, 50—99, 100—199, 200—299, and 300 bed groups were 43%vs. 33%, 18% vs. 19%, 14% vs. 16%, 8% vs. 9% and 16% vs. 23% respectively.Mean response rate for urban centers was z80%, with 94% of the largesthospitals responding and lower response rates from smaller and rural hospitals.Receipt of Published Guidelines by Canadian HospitalsCorrelations between rate of receipt for published guidelines and both thehospital size and type of publication are evident in Table 4. Receipt ratesincreased as a function of hospital size, but different publications’ receipt rateswere not equal. Federal publications were received more commonly than topicalreview or medical journals.Two federal publications, CDWR and Morbidity and Mortality Weekly Report(MMWR), are key sources of information that provide primary documentation forUP protocols and timely updates of CDC or LCDC recommendations. Their receiptwas dependent on hospital size (p<0.001, chi—square test). Log—linear analysiswas used to examine the relation between receipt of CDWR or MMWR and hospitalPage 15Effectiveness of Infection Control StrategiesTable 3:Survey Response Frequency and Rate by Bed Size Group and Provincefor Non—Specialty Acute—Care Canadian Bospitals*BED BRITISB BEN NEVPOUNDLAND/ NOVA NNT/ PRINCE EDNARD ALLSIZE ALBERTA COLUMBIA MANITOBA BRUNSN1CK LABRADOR SCOTIA YUKON ONTARIO ISLAND QUEBEC SASKATCEENAN PROVINCESGROUP (118) (93) (77) (32) (35) (45) (7) (189) (7) (121) (132) (856) +1— 49 27 22 23 4 5 5 3 25 1 4 37 16842% 60% 40% 33% 25% 31% 50% 69% 33% 29% 35% 46%50—9921 9 5 7 4 7 2 28 0 3 8 10163% 56% 71% 78% 80% 64% 100% 65% 0% 18% 79% 65%100—199 12 13 5 2 2 13 0 23 1 10 6 88100% 100% 81% 50% 56% 100% + 67% 100% 33% 100% 75%200—299 2 10 3 4 1 2 0 17 0 9 1 4983% 93% 100% 100% 56% 56% + 71% 43% 28% 69%300+ 10 14 5 4 4 1 0 47 1 32 7 129100% 85% 91% 100% 100% 40% + 91% 100% 84% 100% 94%ALL 72 68 41 21 16 28 5 140 3 58 59 535GROUPS 61% 73% 53% 66% 46% 62% 71% 74% 43% 48% 45% 63%NOTE:* Table excludes 10 responses fran chronic care sites, 1 from a site underconstruction, 12 incomplete responses fran acute—care sites, and 21 responses fromspecialty hospitals.+ Total number of non—specialty hospitals listed for each province is shoin inparentheses. Row totals include 24 responses fran unknown provincial location.+ Indicates no hospitals in category.Page 16Effectiveness of infection Control StrategiesTABLE 4:Receipt of Guidelines by Source and Hospital SizeBEDSIZE GROUP:POLICY AREA AND SOURCE <25 25—49 50—99 100—199 200—299 300—499 500+Number of facilities: 80 103 105 89 50 79 50ISOLATION ROOM REQUIREMENTS 25 40 50 61 36 62 45CDC ISOLATION GUIDELINE 31% 39% 48% 69% 72% 78% 90%ISOLATION ROOM REQUIREMENTS 23 46 65 58 36 53 47LCDC ISOLATION GUIDELINE 29% 45% 62% 65% 72% 67% 94%HAZARDOUS BODY FLUIDS LISTED 9 44 49 64 39 75 48HNWR 1988;37:377 11% 43% 47% 72% 78% 95% 96%WARNING LABELS - LABORATORY 25 48 69 69 43 73 48CDWR 1987;13S3 31% 47% 66% 78% 86% 92% 96%WARNING LABELS- HOUSEKEEPING 14 23 41 47 33 51 37ASEPSIS 1986;8:2 18% 22% 39% 53% 66% 65% 74%GLOVING VERSUS HANDWASHING 7 17 14 26 17 43 37ANN INTERN NED 1987;107:243 9% 17% 13% 29% 34% 54% 74%STAFF EDUCATION 17 36 35 41 28 51 42N ENGL J NED 1986;315:1562 21% 35% 33% 46% 56% 65% 84%CDC or LCDC GUIDELINE Rec’d 43% 56% 75% 82% 90% 87% 98%Reviewed if Received 76% 84% 86% 92% 89% 96% 96%CDWR or HNWR Received 34% 62% 76% 85% 90% 97% 98%Reviewed if Received 89% 89% 88% 96% 96% 96% 96%Page 17Effectiveness of Infection Control Strategiescharacteristics in addition to number of beds. Regional location and ruralversus urban setting were not significant factors (p>O.50), but presence of aninfection control practitioner (ICP) was associated with guideline receipt(p<0.000l). The vast majority of Canadian ICPs receiving a guideline reportedreviewing it.Backward step—wise multiple logistic regression was used to examinehospital size as an interval variable and to include more variables than couldbe accommodated in a log—linear model. Default probability limits to remove orenter terms were used (0.15 and 0.10 respectively). Regression models startedwith receipt of CDWR or MMWR as the dependent variable and size, presence ofan ICP, presence of intern or resident teaching programs, provision of riskservices (e.g.: dialysis, sexually—transmitted disease clinic, etc.) and interactionsas independent variables. In decreasing order of significance, terms retained inall final models were size, presence of an ICP and presence of medical teachingprograms. Provision of risk services was dropped as not significant in allmodels, and interaction terms dropped in all but two (ICPsize was retainedwhen size was expressed as the logarithm of the group median; ICPsize andteachingsize were retained when size was expressed as an arbitrary 7—levelcategorical variable in an asymptotic but not in maximum likelihood regression).One—third of hospitals under 200 beds had received neither MMWR nor CDWR(mean receipt rate 63%, 95% confidence interval with finite population correction61% to 65%, and with very conservative correction for nonresponse 61% to 81%).An exact upper limit was within 1 percentage point of the approximation. Inaddition to this low receipt rate, these smaller hospitals were also the leastlikely to have their own full—time ICPs (Table 5).Page 18Effectiveness of Infection Control StrategiesTable 5:Hospital Size vs ICP Staffing (as Full Time Equivalents)Number of Infection Control Practitioners:Numberof Beds<2525— 4950— 99100—199200—299300—499500+All SizesAdoption of Guidelines for UP or BSIThe proportion of hospitals claiming adoption of UP or BSI roseprogressively by size group (Table 6, p<0.001, chi—square test). Overall, 3590.0 <1.0 1.0 1.5 2.0 2.5 3.0 3.5 4.065.0% 31.3% 3.8%32.0 58.3 9.728.6 62.9 8.613.5 64.0 21.4 1.12.0 50.0 44.0 4.0 — — — —1.3 26.6 58.2 6.3 6.3 — — 1.3 —2.0 6.0 36.0 10.0 26.0 2.0 14.0 2.0 2.023.4 46.2 22.8 2.3 3.2 0.2 1.3 0.4 0.2Hospitals Claiming Adoption of New StrategiesSTRATEGY CLAIMED:UP BSI TraditionalTABLE 6:Percent OfNUMBEROF BEDS<2525— 4950— 99100—199200—299300—499500+TOTAL:43.6% 1.3% 55.1%62.1 6.8 31.176.9 3.9 18.369.7 11.2 19.169.4 10.2 20.464.1 15.4 20.561.2 22.5 16.364.4 9.1 26.4Page 19Effectiveness of Infection Control Strategies(64.4%) claimed UP and 50 (9.1%) BSI, ranging from 44.9% claiming UP or BSI inthe <25—bed group to 83.7% in the 500—bed group.Unlike receipt of guideline publications, a positive, correlation betweenhospital size and affirmative response was not evident for adoption of individualpolicy guidelines (Table 7). Fundamental policy differences between UP, BSI andTraditional strategies, together with the percentage of hospitals giving anexpected response for their claimed strategy, are shown in Table 8. Only 20 of359 hospitals (5.6%) claiming UP had adopted the minimum set of policiesexpected under this strategy, and 0 of 50 claiming BSI had adopted all expectedpolicies (Table 9). If a policy to educate staff was also required, then only 16(4.5%) claiming UP adopted expected policies. If BSI was defined without thepolicy to relax handwashing when gloves were worn, then 1 1 of 50 (22%) metthe requirements if an education policy was excluded but only 7 (14%) ifeducation was required.The majority of Canadian hospitals had not adopted guidelines suggestinggloving as a substitute for handwashing, nor limitation of Universal Precautionsto visibly—bloody body fluids (Table 7). A minority of hospitals claiming UP orBSI had eliminated special warning labels for specimens (UP 29%, BSI 36%) andfor trash and linen (UP 32%, BSI 52%); correlation between adoption of thesetwo policies was weak (r=0.10, p>0.50). The proportion of hospitals claimingadoption of UP and claiming a policy to train all health—care workers to use UPwith all patients ranged from just over 50% in the smallest to 90% in thelargest facilities.Page 20Effectiveness of Infection Control StrategiesTable 7:Adoption of Guidelines by Hospital SizeBEDSIZE GROUP:Policy Guideline <25 25—49 50—99 100—199 200—299 300—499 500+UP ends need for warning 13 20 28 26 6 26 15labels on specimens. 16% 19% 27% 29% 12% 33% 30%Special labels & double bags 17 30 25 26 18 29 29not required for waste or 21% 29% 24% 29% 36% 37% 58%soiled linen.Gloves replace handwashing 3 8 9 8 1 6 2unless hands visibly soiled. 4% 8% 9% 9% 2% 8% 4%Private room required for 25 33 37 40 23 45 26Staph pneumonia. 31% 32% 35% 45% 46% 57% 52%+Private room not required 14 33 40 40 20 36 25for Staph pneumonia. 18% 32% 38% 45% 40% 46%UP does not apply to manybody substances unless they 8 23 33 32 23 37 24contain visible blood. 10% 22% 31% 36% 46% 47% 48%Train all health—care workers 29 59 59 52 31 53 42to use UP with all patients. 36% 57% 56% 58% 62% 67% 84%+ Note: Total may exceed 100% as some hospitals claimed adoption of both.Page 21Effectiveness of Infection Control StrategiesTable 8:Percent (Number) of Hospitals Adopting Policies Expected under UP,BSI and Traditional StrategiesEXPECTED RESPONSE UNDER:POLICY: UP BSI TRAD.No Special Warning Labels: ADOPT ADOPT REJECT—for Specimens 29% (107) 36% (18) 94% (138)—for Waste & Soiled Linen 32% (114) 56% (28) 78% (115)Gloves as Alternative to REJECT ADOPT REJECTHandwashing 93% (333) 10% ( 5) 96% (144)Private Room for Staph REJECTPneumonia (n/a) 44% (22) (n/a)Precautions Don’t Apply to ADOPT REJECT REJECTMany Non—Bloody Body Fluids 43% (154) 90% (45) 86% (126)Note: CDC and LCDC guidelines are contradictory regarding Adoption or Rejectionof requiring a private roan for Staph pneumonia under UP or Traditional strategies.As such, this policy is not applicable to assignment of UP or TRAD based uponpolicies adopted.Table 9:Infection Control Strategy Claimed versus PracticedSTRATEGY STRATEGY PRACTICED:CLAIHED: UP BSI TRAD. OTHER TOTALUP 20 1 119 219 359(5.6%)BSI 0 0 14 36 50(0.0%)TRADITIONAL 1 0 96 50 147(65%)Total: 21 1 229 305 556Page 22Effectiveness of Infection Control StrategiesRationale for Adopting a New StrategyHospitals explained their reason(s) in open text for adopting UP or BSI. andall reasons given were tabulated. These are summarized in Figure 1 as fourcategories. In all size groups, protection of staff was consistently themotivation most commonly claimed. Miscellaneous considerations (eg ‘common—sense approach’, “simpler change to introduce”, ‘acceptance by staff”, ‘want tokeep isolation categories”, “eliminate prejudice against labelled patients’, etc.)and compliance with expert guidelines followed, with patient protectionconsistently the least common motivation.Figure 1:Hospitals’ Motivations for Adopting UP or BSIPage 23Effectiveness of Infection Control StrategiesOther Perceptions — Disposal of SharpsWith regard to sharps disposal, only 13 of 556 hospitals (2.3%) describedthe use of cardboard disposal containers. All other descriptions referred todurable plastic (or, rarely, metal or glass) containers which were either obtainedcommercially or recycled from discarded solution containers. However, ICP’s inover half of the facilities claiming adoption of UP or BSI still felt that at least50% of their staff members continued to recap used needles. While manyreported advocating “safe” (one—handed or thimble) techniques when recappingcannot be avoided, several questioned whether these techniques were used inpractice and none claimed any measurement of staff compliance. Table 10summarizes how needles were transported from their point of use to disposalcontainers. In 46% of the hospitals, no special provisions were identified forneedle recapping and transportation to disposal containers on medication cartsor in utility rooms: nurses carried presumably unsheathed needles on anyavailable tray or by hand (36.3%), or they recapped needles at the bedside(9.9%). 40% of the hospitals did advocate safety procedures or devices: 17.8%provided bedside containers to discourage recapping and 21.7% used specialdevices to improve the safety of recapping and/or transportation. Table 11summarizes respondents’ perception of recapping practices in hospitals promotingstrategies to discourage recapping: The alternative most commonly used wasperceived to be the least effective.Overall, 318 respondents (57%) felt that at least half of their hospital’sstaff recap used needles. Of these, 242 specified a strategy other than bedsidecontainers to discourage unsafe recapping en route to disposal containers. 50(21%) promoted recapping safety devices or one—handed technique, and, as notedin Table 11, 10 (4%) provided foam stabbing blocks and 9 (4%) provided specialPage 24Effectiveness of Infection Control StrategiesTable 10:Predominant Method for Transporting Used Needles from the Point of Useto the Point of Disposal, by Hospital Size According to Number of BedsNUMBER OF BEDS:PREDOMINANT 25— 50— 100— 200— 300—METHOD <25 49 99 199 299 499 500 TOTALBedside Container 13.8* 13.6 12.4 15.7 26.0 25.3 28.0 17.8%Carry Unsheathed 36.3 50.5 40.0 41.6 32.0 13.9 30.0 36.3%Recap & Carry 23.8 9.7 10.5 9.0 2.0 6.3 2.0 9.9%Safe Recapping 5.0 4.9 11.4 18.0 18.0 15.2 14.0 11.7%Foam Stabbing Block 2.5 3.9 4.8 3.4 4.0 12.7 4.0 5.0%Special Trays 0.0 5.8 2.9 5.6 10.0 6.3 8.0 5.0%Not Specified 18.8 11.7 18.1 6.7 8.0 20.3 14.0 14.2%* Percent of size group using method indicated.Table 11:Perceived Effectiveness of Strategies To Avoid Recapping of Used NeedlesNUMBER (%) OF HOSPITALSCLAIMING 50% OF STAFF:PREDOMINANT DO NOT DOSTRATEGY: RECAP RECAPBedside Containers 61 (61.6%) 38 (38.4%)Carry unsheathed 83 (41.1) 119 (58.9)Foam Stabbing Block 18 (64.3) 10 (35.7)Special Trays 19 (67.9) 9 (32.1)PEARSON CHI-SQUARE = 17.550, 3 d.f., p < 0.001Page 25Effectiveness of Infection Control Strategiestrays. Thus, only 29% of hospitals where the majority were perceived to recaphad promoted point—of—use safety devices and procedures to reduce risk ofneedlestick injury.Other Perceptions— Knowledge of Costs and BenefitsWhile this survey did not ask for detailed economic evaluations, only 19%of hospitals claiming UP or BSI indicated knowledge of cost implications in theirinstitution. These were primarily gross estimates rather than sophisticatedexamination of marginal costs. Less than half a dozen sites indicated that theywere planning or conducting comprehensive reviews. Measurement of compliancewith infection control policies by health—care workers, and of benefits achievedunder UP or BSI as practiced in Canadian hospitals, largely unknown at thistime, were identified as priorities for further research. 57% of respondentsindicated a desire to participate in the next part of this research: multicenterevaluation of compliance and effectiveness by standardized covert observationsand questionnaires from health—care workers themselves.Summary of Major Findings1. Three—quarters of Canadian acute—care hospitals reported use of UP or BSI.2. Guidelines defining UP or BSI were not received by all of these hospitals:one—third of facilities under 200 beds (those most likely to need externalhelp) had not received them.3. While 65% of hospitals claiming a Traditional isolation program adopted allpolicies expected, <6% claiming UP or BSI adopted all expected policies.Inconsistency in the application of policy to different departments wasnoted, and numerous names other than UP or BSI were substituted.4. Distinctions between UP and BSI, and the costs or benefits of either, werenot well—known.Page 26Effectiveness of fl2fection Control StrategiesIII. Have Hospital Staff Members Accepted the New Strategies?Infection Control Practices of Critical Care Nurses.The major emphasis in guidelines for UP or BSI focuses on appropriate useof gloves, handwashing and handling of sharps. Handwashing is a necessary24but perhaps insufficient measure to protect patients and staff from nosocomialinfection; UP and BSI guidelines therefore propose gloving as an adjunct.Frequent failure to wash hands, and the over—estimation of ones ownhandwashing compliance, remain common findings throughout studies spanningthe last fifteen years.25 - 30 Failure to wear gloves for direct contact withblood or other body substances during nursing care is associated withtransmission of herpes simplex3’ and hepatitis A32 viruses as well as a widevariety of bacteria,33 - and, rarely, hepatitis B39 and humanimmunodeficiency virus.40 41 Gloving, however, is not a panacea. Glovescontaminated after42 or even before use43 may spread contamination andinfection.44 Some guidelines permit reuse of gloves between patients aftereffective washing, but washed gloves soon become sticky45 and even new glovesdo not provide an absolute barrier to virus penetration.46 Bacterialcontamination of hands has been demonstrated after dressing wounds even whengloves were worn.47 This could result from nasal carriage with subsequenttransfer to hands, or from direct contamination of hands upon glove removal.48Latex gloves may cause allergic reactions,49 and detrimental environmentalbyproducts are also generated in glove disposal.5° Current guidelines allowsome individual discretion in choosing between no—touch technique and glovingwith several of the patient—care practices selected for observation in thisstudy. However, it is reasonable to conclude from current knowledge thatuniversal use of gloves during all of the procedures selected might help toPage 27Effectiveness of Infection Control Strategiesprevent nosocomial infection of patients or staff. It is also reasonable toconclude from recent studies that glove use in many hospitals is far fromuniversal.Needlestick injury is the foremost occupational exposure leading tohepatitis B infection among health—care workers. In the United States, it isestimated that 500 to 600 health—care workers will be hospitalized and over 200will die from associated fulminant hepatitis, cirrhosis or liver cancer everyyear.15 A very small but real occupational risk of becoming infected by HIValso exists, and 70—90% of occupational exposures to this fatal infection haveconsistently involved needlestick injuries.41’ 51 - 61 Although it may benecessary to redesign invasive equipment in order to reduce this risk, UP andBSI infection control guidelines focus on admonishing health—care workers tohandle needles carefully and especially to not recap them. This, however, isnot new advice and there is little evidence to suggest that such warnings aboutsafe handling of sharps have been effective.62The few publications reporting understanding, compliance and/oreffectiveness of UP or BSI in individual hospitals present contradictory findings.Programs may fail to achieve their objectives if they are based upon faultytheoretical models or if workers fail to comply with program policies. Asdescribed earlier, the majority of Canadian hospitals introduced myriadmodifications in adopting UP or BSI, to a point where these names no longerhave any specific meaning in practice. Further, the sharps disposal system mostcommonly provided to Canadian health—care workers was found to be the leasteffective in discouraging their recapping of used needles. It is not clearwhether new infection control strategies have induced consistent use of glovesas expected, or whether safer sharps—handling practices have resulted.Page 28Effectiveness of Infection Control StrategiesTherefore, the next step in examining the effectiveness of UP and BSI asunifying strategies was to determine the extent to which health—care workersactually adhere to these precautions in daily practice.Materials and Methods:Recruitment and Participation RequirementsAll acute—care Canadian hospitals had received questionnaires during theadministrative survey described in the previous chapter. In January and March1990, letters inviting participation in this study were sent to administrators ofhospitals that indicated in their response to the administrative survey a desireto take part in subsequent research. Participation requirements were explainedas; (1) submission of hospital policies regarding gloving, sharps disposal,hepatitis B immunization and in—service education; (2) agreement to conduct 60covert observations of nursing care in their critical care unit; (3) inspection of10 filled sharps—disposal containers to estimate the extent of recapping;(4) review of staff needlestick incident reports; and (5) distribution of testquestionnaires (described in the next chapter) to critical care nurses.ICPs in each participating site were asked to conduct 60 covertobservations of specific nursing procedures in their intensive care unit (ICU).The quota of 60 observations requested from each site reflects a balancebetween a number low enough to be feasible yet high enough to provideprecision and power. Sixty observations of nursing care would provide 99%power to detect 50% improvement over random (50%) compliance with approvedprocedures.63 Specific procedures monitored include intravascular therapy,wound care, oral care, and perineal care (Table 12). ICPs were instructed torecord specific observations on a check—list every week over a period of severalmonths, and each recorded observation was dated. ICPs also examined thePage 29Effectiveness of Infection Control Strategiescontents of 10 filled sharps disposal containers from the critical care unit,estimating the proportion of recapped needles as 0. 1—10%, 11—25%, 26—50%, or>50%. 10% and 25% were taken as two limits for tolerable policy compliance.Statistical Analysis and Hypothesis TestingFour observation scores were calculated for each hospital as the simpleproportion of glove use in each of the four categories of procedures observed.These scores relate to hospitals, not individual nurses, as the sampling unit.Notch plots were used to examine the distribution of hospitals scores.64Normality of score distributions was confirmed by the Kolmogorov—Smirnovtest.65 SYSTAT (ver. 4.1) was used to perform these analyses.A null hypotheses (#3) that ICU nurses do not recap used needles wastested at two levels of compliance. Five or more containers with more than 10%of the needles recapped formed one critical region for rejecting the nullhypothesis (a=0.0015, 130.0001 against an alternative hypothesis of 90%recapping). Six or more containers with more than 25% recapped formed thecritical region for a second test (a0.0197, 13=0.0781 against an alternative of75% recapping).Table 12:Nursing Care Procedures ObservedINTRAVASCULAR ORALStart new IV site Mouth careChange IV line Suction airwayUse IV stopcock Retape endotracheal tubeWOUND CARE PERINEALDress dry wound Perineal careDress draining wound Remove bedpanEmpty hemovac Empty foley bagNote: Form supplied to ICPs had 5 lines for recording observations of “Startnew IV site”, 5 for “Change IV line”, etc., so that the 60 total observationswould be composed of equal numbers from each of the four categories and thatnone of the procedures would be observed more than 5 times.Page 30Effectiveness of Infection Control StrategiesResultsResponse Rate57% of over 500 hospitals participating in the administrative surveyexpressed interest in participating in this second phase, but only 72 actuallyenrolled. Half withdrew before completing covert observations and nurses’ tests,for the same stated reason that so few enrolled: workload conflicts of their ICP.The 35 hospitals completing this study were primarily large community—hospitalsfrom urban centers across Canada. Approximately 3% of western, mid—westernand central but 10% of eastern hospitals participated. Smaller facilities tendedto be excluded by the requirement to make observations in a critical care unit,and very few university—affiliated hospitals elected to participate.Use of GlovesThere was considerable variation from hospital to hospital in the frequencyof glove use observed during various types of nursing care procedures: Figure 2shows ranges of 0—100% in all of the categories observed. Distributions ofoverall hospital scores for glove use during IV and Perineal care procedureswere approximately normal (p=0.679 and 0.272, respectively, Kolmogorov—Smirnovtest). Oral and Wound care scores were skewed slightly more by relatively highnumbers of hospitals observing frequent use, but were still reasonably normal intheir distribution (p0.095 and 0.073 respectively). There were large positivecorrelations between several of these four categories of practice.Since the composition of observations from sites providing fewer than 60observations varied, simple overall means would be confounded by the mix ofprocedures observed and therefore not be suitable for direct comparison. Directstandardization,66 applying the four category—specific rates to a “standardpopulation” composed of 15 observations from each of the four categories, yieldsPage 31Effectiveness of Infection Control Strategiesuniformly—biased overall scores suitable for comparison (Figure 2).Figure 2:Frequency of Observed Glove Use During Types of Nursing Care0 25% 50% 75% 100%Iv+WOUNDORALPERINEAL25% 50% 75% 10 %Note: Scale represents the percentage of observations in which gloves wereworn. -. indicates the median among all 35 hospitals, () indicates a95% confidence interval for the median, boxed areas extend from the25th to the 75th percentile, * indicates an extreme value. “DS”is a uniformly—biased directly—standardized overall rate (see text).Handling of Used Disposable NeedlesAs predicted in the administrative survey, nurses frequently did not complywith national guidelines for safe needle disposal. Table 13 summarizes thefrequency with which sharps disposal containers held various proportions ofrecapped needles. Recommended disposal practices were followed 90% of thetime in 9 of 32 hospitals (28%) and 75% of the time in 15 (47%).Page 32Effectiveness of Infection Control StrategiesTable 13:Frequency of Recapped Needles in Disposal ContainersPERCENTAGE OF RECAPPED NEEDLES IN EACH CONTAINER0% 1—10% 11—25% 26—50% 51—100% TOTALCONTAINERS: 32 65 56 63 104 320% OF TOTAL: 10.0 20.3 17.5 19.7 32.5 100%Summary of Major Findings1. All surveyed hospitals had policies in place mandating use of protectiveapparel, but few were prepared to monitor policy compliance and providefeed—back on performance to their staffs.2. On the average, gloves were worn about 60% of the time expected, with widevariation ranging from 0—100% among the hospitals. Fewer than half ofthe hospitals achieved minimally—acceptable compliance (75% usage).3. Fewer than half of the hospitals achieved minimally—acceptable compliance(25% recapping) with policies to not recap used disposable needles.Page 33Effectiveness of Infection Control StrategiesKnowledge and Beliefs of Critical Care NursesPrograms may fail to achieve their objectives if they are based upon faultytheoretical models or if workers fail to comply with program policies. Directobservation of critical care nurses at work found frequent noncompliance withinfection control policies. This could result from failure to know, to understandor to accept hospital policies. Unless pressures of the workplace interfere,strong correlation between knowledge and practice might be expected. However,industrial safety experience often shows discrepancies between knowledge andpractice. In order to test the knowledge and beliefs of nurses whose workpractices had been observed in the 35 hospitals described above, questionnaireswere distributed to those nurses after completion of the covert observations.Materials and Methods:Test FormsTest statements covered knowledge and belief concerning risk recognition,risk control and infection control policy with respect to four knowledge areas:(1) hepatitis B and AIDS in the workplace, (2) safe handling of sharps, (3) useof gloves under UP or BSI, and (4) general aspects of infection control.Concepts and answers were taken from published infection control guidelines.Wording was modified after review of draft statements and their intendedpurpose by nurses experienced in infection control and patient care but notemployed in the hospitals where the test would be applied. The test consistedof 32 statements, and responses were recorded on a 5—point scale of degree of(dis)agreement with each statement (a Likert scale). Some of the statements askabout personal history (e.g.: if the subject has suffered a needlestick injury inthe preceding 30 days, if they had difficulty understanding the questionnaire orusing an optional computer program provided). Each statement, an expectedPage 34Effectiveness of Infection Control Strategiesanswer consistent with corresponding referenced publication(s) and the pertinentreference(s) are listed in Appendix 3.Participating hospitals were offered printed forms or a computer program toconduct the tests. Statements were presented with identical wording on boththe forms and computer screen. All critical care nurses in participatinghospitals were invited by their own ICP to complete a standardized testanonymously at their own convenience; forms were collected a few days aftermass distribution to minimize opportunities for collaboration. The interactivecomputer program, written in F0xBASE+ (version 2.00), does not ask for identityof individual nurses, but does automatically identify the hospital, type ofcomputer, and test completion time. The encrypted and compiled programconfirms integrity of stored coded responses before each subsequent use, anddoes not permit decoding by individual sites. Printed questionnaires, alsoanonymous, were returned in sealed envelopes. Upon receipt, they were markedwith unique sequential numbers and transcribed using duplicate data entry viaF0xBASE+ programs for automated error—checking and scoring.Statistical AnalysisFour section scores (one for each of the knowledge areas tested) and anoverall score were computed for each nurse as the percentage of statementsanswered as expected. Hospital scores were derived as the means of theirnurses’ scores. The distribution of these scores was examined in several ways.Normality of nurses’ and hospitals’ overall and section score distributions wereconfirmed by the Kolmogorov—Smirnov test.65 Notch plots64 and ANOVA ofnurses’ overall test scores (using site and claimed difficulty as independentvariables) were used to determine whether nurses’ overall scores differedsignificantly between hospitals or in association with other variables. A barPage 35Effectiveness of Infection Control Strategiesgraph of the four section scores, grouped by site, was used to look for anynation—wide trends in relative performance among the different knowledge areas.Each concept on the questionnaire was tested by two statements whichappear at variable distances apart from each other. Appendix 3 lists thepairings to which each statement was assigned (0 indicates unscored items).The influence of the order in which statements were presented was not studied;however, potential for learning bias as well as further examination of relativeperformance were explored by comparing scores on the first versus secondstatement in each pair using Tukey’s sum—difference graph.67Correlation coefficients were also computed. A null hypothesis (#4) thatthere is not a strong positive correlation (r0.5) between infection controlknowledge and daily practice scores was tested in three ways: correlationcoefficients between individual knowledge and practice component section scores,between unweighted overall test and observed—practice summary scores, and alsowith weightings derived from canonical correlation using the four knowledgesections of the test arid four care categories of the observations. Canonicalcorrelation is a multivariate technique to produce linear combinations attainingthe maximum correlation possible.68 SYSTAT (ver. 4.1) and SAS (ver. 6.03, PROCGLM) were used for these analyses.ResultsResponse RatesNone of the sites used the interactive software. Seventy—two of 489questionnaires received from 35 sites had one or more questions leftunanswered; eight of these (from eight different sites) left an entire page blankand were excluded as incomplete. Table 14 summarizes characteristics ofparticipating hospitals completing this part of the study.Page 36Effectiveness of Infection Control StrategiesTable 14:Characteristics of Participating HospitalsApprox.Site Number Region Questionnaireof Beds of Canada Response Fraction61 200 West 7/10 (70%)125 200 West 15/16 (94%)123 300 West 5/16 (31%)9 350 West 7/17 (41%)417 450 West 9/14 (64%)257 >500 West 8/27 (30%)258 >500 West 30/67 (45%)19 250 Mid—West 6/30 (20%)103 400 Mid—West 11/29 (38%)225 400 Mid—West 11/50 (22%)420 400 Mid—West 13/22 (59%)527 >500 Mid—West 47/65 (72%)131 50 Central 3/ 6 (50%)78 100 Central 7/10 (70%)192 100 Central 6/12 (50%)95 235 Central 10/15 (67%)165 250 Central 13/20 (65%)408 400 Central 20/30 (67%)481 400 Central 23/50 (46%)541 400 Central 21/61 (34%)81 >500 Central 22/59 (37%)104 >500 Central 29/60 (48%)179 >500 Central 19/40 (48%)67 100 East 5/ 6 (83%)378 100 East 10/25 (40%)296 100 East 6/14 (43%)6 200 East 13/24 (54%)59 200 East 9/18 (50%)189 200 East 14/16 (88%)363 200 East 15/15 (100%)418 250 East 13/25 (52%)285 300 East 15/17 (88%)284 400 East 10/15 (67%)558 450 East 15/40 (38%)452 >500 East 19/29 (66%)Note: Response fractions are the number of ICU nurses returning a completedquestionnaire divided by the number of ICU nurses in each hospital.Page 37Effectiveness of Infection Control StrategiesNurses’ Knowledge and Belief ScoresTest scores by section tended to range from highest to lowest for sharpshandling, gloving, AIDS or hepatitis, and general aspects of infection control,respectively (Table 15). The proportion of nurses in each hospital agreeing thattheir hospital’s infection control measures were practical, effective and well—documented is shown in Figure 3, as well as the proportion claiming receipt ofinstruction in these policies. There was a statistically significant associationbetween nurses agreeing with their hospitals policies and claiming receipt ofinstruction (p<O.OOl, chi—square test; Pearson correlation coefficient O.2) butno such association between the proportion of nurses in agreement with policiesor receipt of instruction and higher overall hospital test or observed—practicescores. Inservice education provided to these nurses appeared to stress sharpshandling more frequently than gloving or general aspects of UP/BSI.Table 15:Relative Ranking of Knowledge Test Section ScoresNumber of Hospitals in Which SectionScore Was Ranked As:Test Section Highest 2nd 3rd LowestSHARPS 24 9 2 0GLOVES 11 24 0 0AIDS/HEPATITIS 0 2 29 4GENERAL 0 0 4 31Nurses’ scores did not improve as a function of exposure to test questions.Figure 4 depicts the distribution of scores for question—pairs horizontally, andfor discordance within pairs vertically. The frequency of correct answers to asecond member in each pair of questions exceeded that for the first member inonly three of eleven pairs. 42—52% of answer—pairs were discordant (onecorrect, the other incorrect) for all pairs except those regarding philosophicPage 38Effectiveness of Infection Control StrategiesFigure 3:Effectiveness of Hospitals’ In—Service Program0.00 0.50 1.00I(+Proportion of nurses agreeing that their hospital’spolicies are practical, effective and well—documented.FProportion of nurses claiming receipt of in—serviceinstruction within past year re: Universal Precautions+ 1Proportion of nurses claiming receipt of in—serviceinstruction within past year re: sharps handlingProportion of nurses claiming receipt of in—serviceinstruction within past year re: proper use of gloves0.00 0.50 1.00Note: Scale represents the proportion of nurses in each hospitalresponded affirmatively. + indicates the median among all35 hospitals, () indicates a 95% confidence interval for themedian, the boxed area extends from the 25th to the 75thpercentile, * indicates an extreme value.Page 39Effectiveness of Infection Control StrategiesFigure 4:Tukey Sum—Difference Graph of Correct Response Frequencies300Difference F IBetween 200Correct 3AnswerFrequency 100 Bfor K APaired C EQuestions 0D G—100H—2000 200 400 600 800 1000Sum of Correct—Answer Frequency forPaired QuestionsLEGEND: Net discordance in paired—item scores is indicated by distance above orbelow the dashed horizontal line; pairs with higher scores appear tothe right and with lower scores to the left on this graph. Questionswere paired on the basis of testing related concepts:A = infectivity of HIV and hepatitis B virus (questions 1 & 5)B = need for screening and warning labels (questions 2 & 26)C = purpose and philosophic differences of UP/BSI (questions 4 & 30)D = immunity to infectious disease (questions 6 & 24)E = operational definitions with UP and BSI (questions 8 & 18)F = frequency and severity of needlestick injury (questions 9 & 12)G = recapping and need].estick risk (questions 10 & 13)H = protection afforded by gloves (questions 15 & 19)I = frequency of handwashing (questions 16 & 23)3 = proper use of gloves (questions 20 & 22)K = definition and disposal of infectious waste (questions 27 & 28)Page 40Effectiveness of Infection Control Strategiesdifferences of UP and BSI (C, 29%), recapping and needlestick risk (G, 23%) andinfectious waste (K, 22%). Scores were lowest on questions regarding infectiouswaste, those pertaining to quantification of risk (F), and purpose or philosophyof specific control measures. Highest scores were achieved on informationrecognized well before the advent of new infection control strategies (D,G,I,J).There were statistically significant differences between scores achieved bynurses in individual hospitals as well as by those claiming difficulty versus nodifficulty understanding the test (2—way ANOVA):DEGREES OF TYPE III SUM ME?N FSOURCE FREEDC1 OF SJ?iRES SQUARE RATIO pHospital 34 12763.08693 375.3849 2.85 0.0001Difficulty 1 688.93145 688.93145 5.22 0.0228Neither region nor hospital size were significant in explaining this variationabout a total mean score of 54%. and very few sites (e.g.: site #6) had adistinctly different median score (Figure 5). The 141 nurses (29%) claimingdifficulty with the test itself achieved a slightly lower median score than thosenot claiming difficulty (Figure 6). Similarly, those claiming receipt of in—serviceeducation or agreeing that their hospital’s infection control measures werepractical, effective and well—documented achieved only slightly higher scoresthan those not. Overall scores achieved by nurses approximated a normaldistribution (p0.05, Kolmogorov—Smirnov test) in 30 of the 35 hospitals.The hospitals’ overall knowledge test scores were normally distributed(p0.3l6, Kolmogorov—Smirnov test), but their directly standardized overallpractice scores less so (p=0.046). Weighted arcsine transformation of thesescores [fniarcsineVpi)] adjusted the raw scores (p1) for differing numbers oftests or observations submitted by each hospital (fli) as well as modifying thedistribution shape of overall test and practice scores (p0.1l4 and 0.185respectively, Kolmogorov—Smirnov test). The Pearson correlation coefficientPage 41Effectiveness of Infection Control StrategiesFigure 5:Nurses’ Overall Test Score, by Site (Ordered by Region and Size)MINIMUM SCORE MAXIMUM SCORE23.0% 50.0% 100.0%Site #619.__________.____I1J—) 417______257( ) [ 258N19D-f 103S 225* 420( .f) 527Page 42Effectiveness of Infection Control StrategiesFigure 5 (emit.):MINIMUM SCORE MAXIMUM SCORE23.0% 50.0% 100.0%Site #FF 192-(- 95165(-f ) 408(..F ) 481* 541_81104-f ) 179Page 43Effectiveness of Infection Control StrategiesFigure 5 (cont.):MINIMUM SCORE MAXIMUM SCORE23.0% 500% 100.0%Site #*____‘13::296C (__6A IH + F189363418285284-F 1 558________* 452Legend: j- indicates medians, () indicates 95% confidence intervals, and *indicates extreme values. Median scores differ significantly iftheir confidence intervals do not overlap.Page 44Effectiveness of Infection Control StrategiesFigure 6:Nurses* Overall Test Score, by Personal Bistory ClaimsMINIMUM SCORE MAXIMUM SCORE Claim23.0% 50.0% 100.0%_____No DifficultyDifficulty-1-) * UnderstandingQuestionnaireIn—Service*_________________(_________________* * * Not ReceivedIn—Service* * I c-I-> * ReceivedNot Satisfied* I (..f )__ _* with PoliciesT (..[ ) I Satisfied withI I PoliciesLegend: -- indicates median score for all nurses regardless of hospital,0 indicates 95% confidence intervals, and * indicates extremevalues. Median scores differ significantly if their confidenceintervals do not overlap. Box width is proportional to thesquare root of the group size.Page 45Effectiveness of Infection Control Strategiesbetween these summary scores was only moderately strong (zO.35) and justachieved statistical significance (p0.04). Spearman’s correlation coefficientsamong test section and care category scores were not statistically significant(Table 16). Table 16A treats all hospitals as equals; 16B (the transformed data)gives greater weight to those with larger nursing staffs (as implied by largernumbers of tests returned). The number of multiple comparisons possible inTable 16 inflates the probability that a spurious association may be found asstatistically significant; Bonferroni adjustment reduces the chances of this.Relationship Between Knowledge and PracticeCorrelation between directly—standardized overall practice and knowledgescores was not statistically significant (Table 16), nor were any of the largercorrelations achievable through reweighting by coefficients derived fromcanonical correlation analysis (Table 17). Further, if the analysis is restrictedto those sites coming closest to the number of observations requested, apersistent negative trend in the correlation coefficient is evident (Figure 7). Atthe extreme of this sensitivity analysis, data from the four sites providing all60 observations requested yield a correlation coefficient of —0.8. This effect isalso evident in the difference between test—practice correlation coefficients onTables 16A and 16B: weighted—arcsine transformed data provides a correction forthis confounding and yields a smaller correlation coefficient.Summary of Major Findings1. In half of the hospitals, no more than 50% of participating nurses agreedthat their policies were practical, effective and well—documented.2. Many nurses held divergent opinions, did not understand or did not agreewith tenets fundamental to UP and BSI. Test scores averaged 54% correct.3. A positive correlation between knowledge and practice was not found.Page 46Effectiveness of Infection Control StrategiesTable 16:Spearman Correlation Coefficients Between Test and Practice ScoresTable 16A: Correlation Coefficients, Raw DataOVERAlLTEST PRACTICE AIDS/IP SHARPS GLOVES OTHER IV WOUND ORAL PERINEALOverallTEST 1.000PRACTICE 0.310 1.000Test SectionAIDS/HEP 0.783 0.275 1.000SHARPS 0.438 —0.061 0.231 1.000GLOVES 0.709 0.246 0.401 0.142 1.000C7l’HFR 0.586 0.363 0.277 0.002 0.332 1.000Care CategoryIV 0.432 0.688 0.298 0.008 0.341 0.503 1.000WOUND 0.267 0.757 0.191 0.089 0.075 0.370 0.501 1.000ORAL 0.145 0.665 0.184 —0.263 0.308 0.087 0.245 0.315 1.000PERINEAL 0.192 0.800 0.201 0.077 0.072 0.172 0.420 0.492 0.500 1.000Table 1GB: Correlation Coefficients, Weighted Arcsirie Transfornd DataOVERALLTEST PRACTICE AIDS/HEP SHARPS GLOVES C1I’HER IV WOUND ORAL PEPJ]EALTEST 1.000PRACTICE 0.287 1.000AIDS/HEP 0.932 0.236 1.000SHARPS 0.908 0.250 0.811 1.000GLOVES 0.980 0.259 0.901 0.872 1.000OTHER 0.967 0.309 0.879 0.860 0.938 1.000IV 0.259 0.756 0.245 0.187 0.237 0.310 1.000WOUND 0.207 0.913 0.154 0.206 0.168 0.254 0.682 1.000ORAL 0.280 0.674 0.256 0.147 0.301 0.272 0.317 0.575 1.000PERINEAL 0.166 0.857 0.168 0.157 0.131 0.180 0.557 0.731 0.622 1.000Note: Values 0. 341 are statistically significant (p0 .05, 2—sided) without Bonferroniadjustment; values 0.573 achieve statistical significance with Bonferroni adjustment.69The latter are shown boldface.Page 47Effectiveness of Infection Control StrategiesTable 17:Canonical Variate Coefficients and CorrelationsKNOWLEDGE TEST SECTION SCORES: OBSERVED PRACTICE SCORES:AIDS/HEP SHARPS GLOVES OTHER rl* IV WOUND ORAL PERINEAL0.124 —0.577 0.535 0.519 0.593 —0.703 0.147 —0.804 0.3980.122 0.597 —0.453 0.738 0.409 0.254 0.804 —0.664 0.245*Note: None of the canonical correlations are statistically significant (p>0.3)Figure 7:Spearman’s Correlation Coefficient (of Knowledge and Practice Scores) vs.l4inumum Number of Practice Observations Required for Inclusion in AnalysisI I I0.5CORRELATIONCOEFFICIENT. II I.0.0U—0.5I—1.0I I I I I0 10 20 30 40 50 60[34) [33] [30] [26] [22) [14] [4]MINIMUM NUMBER OF OBSERVATIONS REQUIREDNote: Numbers in brackets indicate number of sites included.Page 48Effectiveness of Infection Control StrategiesIV. Have New Infection Control Strategies Reduced Infection Risks?Although reduced infection risk for patients will undoubtedly provide themajor economic benefit from an effective infection control strategy, reduced riskfor health—care workers was the major focus of this research for two reasons.First, Canadian hospitals indicated that protection of staff, specifically fromHIV. was their primary motivation for adopting new strategies. This can beevaluated indirectly by measuring needlestick injury rates, since the majority ofoccupational HIV and hepatitis B infections involve needlesticks. Second, therewere too many interacting variables associated with most types of nosocomialinfections suffered by patients, too many endogenous and exogenous sources ofpathogens and too many inadequacies in current microbiologic typing methods tomake a study of patients’ risk practicable. The impact of nurses’ glovingpractices on intravascular—associated nosocomial infections was considered forstudy, since this problem involves relatively few confounding variables.However, an adequate typing method was not available for the nosocomialpathogens most commonly associated with intravascular cannulae, coagulase—negative staphylococci.16Epidemiologic Assessment of Risk to Hospital Staff— Needlestick InjuriesNeedlestick injury is a well—recognized occupational hazard for health—careworkers. Published guidelines discourage recapping of used needles as an unsafeand unnecessary practice. However, the strategy most commonly employed byCanadian hospitals appeared to be the least effective in discouraging recappingof used needles (Tables 10 and 11). Subsequent observation of nursing practiceconfirmed the opinion that many nurses still recap used needles in spite ofpolicies discouraging this habit. A test of knowledge among these same nursesconfirmed that most agree that used needles should not be recapped. However,Page 49Effectiveness of Infection Control Strategiesthese were cross—sectional surveys, not before—and—after studies. Therefore,the next step in this research was refinement to see if adoption of UP or BSIresulted in decreased needle recapping or injury rates. Needlestick injuryincidence rates were estimated from data before and after adoption of newinfection control strategies, and incidence density rates were calculated(essentially, number of injuries divided by number of hours worked) on pairedbefore—and—after data. Incidence density is more meaningful than other attackrates which are often published (e.g.: injuries per number of employees, pernumber of full—time equivalent employees, or per number of hospital beds)because it takes the duration of risk—exposure into account.7°Materials and Methods:The nursing test questionnaire described above asked whether respondentshad suffered one or more needlestick injuries in the preceding thirty—day period.ICPs reviewed the hospital’s employee health records after nurses completed thequestionnaires, looking for needlestick injury reports from critical care nursesduring three consecutive preceding thirty—day periods and during correspondingthirty—day periods prior to adoption of their new infection control strategy.In order to control for the confounding influence of differing durations ofexposure to a risk of needlestick injury, incidence density was calculated bydividing the number of needlestick injuries reported among critical care nursesby the number of hours those nurses worked in each thirty—day period. Pairedanalysis of before versus after UP/BSI rates was used to control for differingbaseline rates among the hospitals. The Sign Test for persistence of a trend ofany magnitude, and Wilcoxon’s Signed Ranks Test for a trend of significantmagnitude were used.71 SYSTAT (ver. 4.1) was used for these tests. Thefrequency of reported injuries was compared to the number of questionnaire—Page 50Effectiveness of Infection Control Strategiesrespondents claiming needlestick injury in order to assess the extent of injuryunder—reporting.ResultsEleven hospitals provided complete data (number of needlesticks reportedand hours worked among critical care nurses for three 30—day periods after andthree corresponding periods before adoption of UP or BSI). Thirty—threehospitals provided partial data. Eleven of 312 staff (3.5%) who returnedquestionnaires in hospitals providing complete data claimed one or moreneedlestick injuries during a preceding thirty day period; only four needlestickinjuries were documented in incident reports. This rate is higher than the 2.3%(21 injuries among 929 staff in a 30—day period) documented in employee healthrecords of all responding hospitals.Monthly incidence density rates in the eleven hospitals providing completedata ranged from a minimum of 0 per 1000 hours worked to maximums of 3.94before and 0.41 after UP/BSI adoption. Although the latter numbers suggestlower risk after each hospital adopted a new infection control strategy, paireddata did not show a statistically significant trend based upon persistence ofdifferences of any size (p=0.332, Sign Test). If the magnitude as well asdirection of before— versus after—UP/BSI differences were considered, borderlinestatistical significance was attained (p=0.076, Wilcoxon Signed Ranks Test).Much of’ the latter effect was attributable to one hospital with a before—UP/BSIincidence rate at least one order of magnitude higher than any of the otherhospitals (Figure 8).The proportion of discarded needles that were recapped had also beenassessed in ten of these eleven hospitals. Only four showed evidence of 25%recapping frequency, and decreasing needlestick incidence rates were found inPage 51Effectiveness of Infection Control Strategiesonly one of these. Conversely, four of six hospitals where recapping was stillwidespread reported decreased post—UP/BSI incidence rates.Figure 8:Needlestick Injury Rates in 11 Hospitals Providing Paired DataINJURY RATE4Needlesticksper 1,000hours worked3210Note: Hospital “A” had pre—UP/BSI rates one order of magnitudehigher than the other ten hospitals. Distinct data points arelabelled by hospital (A,B,C,K); * indicates overlapping datapoints from two or more hospitals in this time—series analysis.Summary of Major Findings1. Employee health records documented fewer nurses suffering needlestick injuryduring a 30—day period (2.3% of 929 nurses in 33 hospitals) than wasfound by surveying nurses directly (3.5% of 312 nurses in 11 hospitals).2. Only 1 of 11 hospitals indicated appreciable risk reduction after UP started.3. An association between reduced needle—recapping and reduced needlestickinjury risk was not evident.—90 —60 —30 0 30 60 90Pre—UP/BSI Days Post—UP/BSI DaysERAPage 52Effectiveness of Infection Control StrategiesEpidemiologic Assessment of Risk to Hospital Patients—Tracing the Sources of Infection by Microbiologic MethodsThe marginal cost of UP in excess of existing infection control costs isappreciable,72 and its cost per case of occupational HIV seroconversionprevented has been estimated at over $8—million.73 While patients and staffmembers have a right to a safe hospital environment, and hospitals, in turn,have both legal and ethical obligations to provide this, it is unlikely thateconomic benefits from UP will exceed its costs. Accurate cost—effectivenesscomparison of UP, BSI and other strategies will undoubtedly require assessmentof their impact on the costs associated with patients’ nosocomial infections. Iftreatment costs can be averted by preventive measures, this would motivateimplementation and support of those measures. However, in order to classifypatients into case or referent study groups for such epidemiologic and economicresearch, better microbiologic typing methods are needed.Microbiologic methods to determine whether isolates are epidemiologicallyrelated exploit the diversity within taxonomic relationships. If one or moredifferences can be found, this implies that isolates may be unrelated in spite oftheir common genus—species name. Typing to a subspecies level supportsepidemiologic investigation by determining whether microbes isolated from twosources may have an epidemiologic relationship (e.g.: from different patients,suggesting cross—infection; from patients and an environmental reservoir,suggesting common—source transmission; from different sites of a patient, suchas blood and an intravascular cannula, suggesting routes of transmission, etc.).However, the problem of proving epidemiologic relatedness by microbiologictyping is often analogous to hitting a moving target with tools of undefinedprecision, accuracy and reliability.16Page 53Effectiveness of Infection Control StrategiesThe extent of diversity within a species must be documented and sufficientto generate enough typing categories for adequate discriminatory power. Thetyping method itself must be sufficiently precise, accurate, reliable and practicalto generate useful results. Methods that have been or could be applied totyping coagulase—negative staphylococci were reviewed16 (Table 18). The MIDIsystem for semi—automated fatty acid analysis (Microbial ID, Inc. NewarkDelaware) was felt to show promise, and an evaluation project was initiated.That research is still in progress (see Appendix 4).Table 18:ypiug Methods for Coagulase-Negative StaphylococciKEY: + = Good± = Problematic, but possible—= InadequateElect. = ElectrophoresisREA = Restriction Endonuclease AnalysisRFLP = Restriction Fragment-Length PolymorphismFatty Acids = Fatty Acid ProfileCRITERIONBiotype and/orntibiogramPhageCONVENTIONMJ METHODS NOLECULAR METHODS NOVEL METHODSSerology Plasmid Profile Chromosomal Multilocus FattyElect. REA RFIAP REA RFLP Enzyme Elect. Acids1. CIiN IT DO THE JOB?Discriminatory ± —— ± + + + + ?PowerSpecies Diversity + — — — — — — ——Documented2. DOES IT DO THE JOB?Accuracy andReliability ± ± - ÷ + + + + ?Availability + —— ± — — — —Page 54Effectiveness of Infection Control StrategiesV. Discussion:Validity, Reliability, Power and Generalization of’ This ResearchThis research produced a considerable amount of data to address threebasic questions: Have Canadian acute—care hospitals adopted UniversalPrecautions or Body Substance Isolation; do their staff members use the newsystem of precautions in daily practice; and has reliable use of a new systemled to decreased risk of infection? In order to answer these questions andunderstand the limits to interpretation in those answers, a basic procedure forappraisal of data is important. Abramson suggests that this involves threesteps.74 First, what are the facts, how were they obtained and what is thelikelihood that bias has distorted the true facts? Second, what are the possibleexplanations (i.e.: a fluke of random chance, confounding or a true relationship)?Third, what additional information is required to confirm the possiblerelationship(s)? This chapter reviews the methods and data presented earlier inorder to appraise the information derived from them.Administrative SurveyIn the administrative—level survey, most Canadian acute—care hospitalsreported adopting Universal Precautions or Body Substance isolation. As in allevaluative research, validity and reliability are critical issues. Face andcontent validity of survey questionnaires were high in that questions concerningdiscrete, objective key variables related to stated survey objectives. Consensualvalidity was established by pretesting survey forms at ten geographicallydiverse pilot sites and discussing these preliminary results with an expertresearch supervisory committee. Reliability was supported by using an objectivemailed survey (allowing respondents to check institutional records beforeanswering, and reducing a risk of interviewer bias); including redundantPage 55Effectiveness of Infection Control Strategiesquestions to ensure consistency of response; automating error—checking on dataentry; and random checks of questionnaire data against corresponding computer—stored transcripts. Other than redundant questions, there was no way toconfirm responses in this phase of the research. However, given the anonymousnature of the survey, there was no obvious motive for respondents tomisrepresent their hospital’s position and the nature of responses does notsuggest that unduly favorable pictures were provided. Further, copies ofpolicies received from hospitals participating in the second phase of theresearch confirmed their administrative survey claims. The nature of responsesreceived both in the pretest and main survey were clear and did not suggestmisunderstanding of questions themselves. The questionnaire was created inEnglish, then translated into French by one native—speaker, checked by asecond, and my translation of completed French forms was double—checked by athird person whose native language was French. Neither misrepresentations normisunderstandings in completing or interpreting the questionnaires were likely tohave distorted the conclusions drawn.Non—response and self—selection bias weaken the survey’s external validity;however, use of a sampling frame and separate analysis of responders to asecond mailing indicated that non—responders were primarily small, ruralhospitals with guideline receipt rates lower than responders. Highlyconservative correction for nonresponse was therefore unwarranted. Given thedirection of bias, survey results were unlikely to have under—estimatedguideline receipt or adoption rates.The first two null hypotheses tested were rejected: statistically significantassociations were found between hospital size, preferred infection control systemand receipt of pertinent guidelines. Rejection of a null hypothesis carries thePage 56Effectiveness of Infection Control Strategiesrisk of a—error, a chance that aberrant findings resulted from a quirk of randomchance. However, corroborating evidence reduces this possibility here. Beyondthe level of statistical significance achieved, parallel trends were observed, forexample, in adoption rates for both UP and BSI as opposed to a Traditionalsystem, and in receipt rates for a variety of different publications. Thestratified and multivariate analyses performed explored any influence of themost likely confounding variables which might have distorted true relationships.Other variables examined were ICP—to—bed ratio, provision of risk services (e.g.:dialysis, drug abuse or sexually transmitted disease clinic), hospital location,and size as an interval variable using number of beds, group median or log ofgroup median. Hospital size remained the most significant determinant ofinformation receipt in models that considered all of these variablessimultaneously.Objective comparison of the specific policies adopted versus the systemclaimed provided important additional information about the infection controlsystems of these hospitals. Increasing hospital size and presence of ICPs ormedical teaching programs may indicate a more cosmopolitan outlook, and thusgreater emphasis on staying informed. It is unlikely that smaller hospitalssimply felt that UP or BSI and the underlying problem of HIV were not relevantto them: nearly half of the smallest hospitals claimed to have adopted UP,stated motivations for adoption were the same in hospitals from all size ranges,provision of risk—services was not found to be a significant variable, and trendsrelated to size were evident throughout the entire range of hospital size groups.Some hospitals undoubtedly accepted the UP or BSI system but rejected certainguideline recommendations selectively after careful review, but this does notexplain all of the deviation from expected policies found. The proportion ofPage 57Effectiveness of Infection Control Strategieshospitals that had not received guidelines defining the system they claimed, thelow correlation between use of “biohazard’ labels for housekeeping versuslaboratory materials, the few sites commenting upon a major discrepancy (atypographical error) between two guidelines, and the comments received suggestother explanations for this deviation.Covert Observation of Nursing PracticeIn the study of nurses’ daily practices, covert observation found significantvariation in staff members’ adherence to their hospital’s system of precautions.Validation of observation accuracy and inter—observer agreement for differenthospitals’ ICPs could not be evaluated. These are weaknesses in the internalvalidity of this research. ICPs, however, routinely observe and evaluate asepticpractices as one of their own job responsibilities. Observations required in thisstudy were objective: gloves were worn or not, and needles were capped or not.The open layout of most ICU’s promotes relatively unrestricted view of anynursing procedures being observed. These factors hopefully reduced observerbiases. ICPs were recruited for the study because they are trained observers ofaseptic practice who regularly visit all nursing units, and therefore may notraise nurses’ suspicion or prompt major changes in work practices simply bytheir presence. ICPs were instructed as to what kinds of practices to observe,and to spread their observations of glove use over several months in order toobtain results representative of nursing unit practices. Extended duration ofstudies may introduce problems (e.g.: drop—outs, consistency of performance,temporal effects), but in this case an observation period of several months wasnecessary. When ICPs evaluated the frequency of needle—recapping, no effortwas made to distinguish between needles used solely to draw up medicationsversus those exposed to patients’ blood.Page 58Effectiveness of Infection Control StrategiesIn addition to self—evident descriptive data on glove use, a third nullhypothesis regarding the unsafe practice of needle recapping was rejected in 53%of these highly—motivated hospitals. The probability that significantly worsecompliance with needle disposal policies really existed but was not found (13—error) was low, zO.08 in each site at a sample size of 10 units (Table 19).Table 19:Magnitude of a— and a—error for Hypothesis #3Number of Containers Probability if TrueWith >25% of Needles Recapping Rate Is Nagnitude ofAction Recapped 25% 75% a 13Accept 0 0.0563 0.0000 0.0000Null 1 0.1877 0.0000 0.0000Hypothesis 2 0.2816 0.0004 0.00043 0.2503 0.0031 0.00354 0.1460 0.0162 0.01975 0.0584 0.0584 0.0781Reject 6 0.0162 0.1460 0.0197Null 7 0.0031 0.2503 0.0035Hypothesis 8 0.0004 0.2816 0.00049 0.0000 0.1877 0.000010 0.0000 0.0563 0.0000Staff members’ use of prescribed infection control precautions in daily practicewas far from universal, and this trend was evident in many hospitals with bothgloving for various types of care as well as needle disposal practices. This isconsistent with reports in the recent literature and the historic reports ofnoncompliance under traditional programs which prompted development of newinfection control strategies in the first place.Test of Nurses’ Infection Control Knowledge and BeliefsA test of knowledge and belief found that many nurses in all parts of thecountry held views divergent from published information. Efforts to safeguardinternal validity in the knowledge and belief test phase of this research werePage 59Effectiveness of Infection Control Strategiessimilar to efforts in the administrative survey. Face and content validity ofquestionnaires were high in that questions related to stated objectives and werederived from key publications. Consensual validity was established bypretesting question statements and discussing their intent with an independentpanel of experts in infection control and nursing practice as well as with anexpert research supervisory committee. Two questions were selected for eachconcept tested. Again, the questionnaire was created in English, thentranslated into French by one native—speaker and checked independently by asecond. Finally, the 35 ICPs acting as liaison to their hospital were invited tocomment on any concerns about ambiguity or validity. Automated error—checkingon data entry, automated scoring and random comparisons of questionnaire datawith corresponding computer—stored transcripts again were used to guard againstintroducing transcription errors in data handling.Few sites produced significantly different mean test scores (Figure 5).Confounding of mean scores by attitude toward policies, receipt of instruction ordifficulty with the test (Figure 6) was not significant (Figure 9). Scoredistributions tended to approximate a normal distribution. The Kolmogorov—Smirnov test used is relatively powerful for detecting non—normality in smallsamples.75 The few deviations that it detected were not severe (Figure 10).Two—way ANOVA indicated site and claimed difficulty as determinants of testscore; significance of differences found was confirmed by nonparametric methods.There are two aspects to reliability in psychometric testing: internalconsistency and stability. Internal consistency, that is, cohesiveness among testitems themselves, is best measured by Kuder—Richardson’s formula KR2O orKR2176 which are special cases of Cronbachs alpha coefficient. Stabilityprovides an indication of the degree to which achieved scores on a test measurePage 60Effectiveness of Infection Control StrategiesFigure 9:Magnitude of Confounding in Nurses’ Test Scores—30 —15 0 15 30*_____*Difficulty Understanding TestHAgreement With Hospital Policies* *———L:if] * *Receipt of Inservice Education—30 —15 0 15 30Note: Scale represents the difference (in percentage points)between mean scores for nurses in each hospital who claimedor did not claim each factor. + represents median differencesamong the 35 hospitals. () represents 95% confidence intervalsfor medians. * represents extreme values. These values arecorrected for differences in individual hospital’s scores.individuals’ “true” scores. It may be assessed by giving the same test to eachsubject twice (test—retest method) or using another, equivalent second test(parallel forms), or by correlating scores achieved on each of several equivalentparts of a single test given once to each subject (split—half method). Internalconsistency among the diverse items in a nurses’ knowledge and beliefs testdeveloped for this research was expected to be low, and therefore was of littleinterest. Stability of the new test, as an indication that similar scores wouldbe achieved if nurses were tested repeatedly, was difficult to measure but ofgreater interest.Page 61Effectiveness of Infection Control StrategiesFigure 10:Normal Probability Plot— Distribution Rejected by the Kolmogorov—Smirnov TestExpected 3 IIValueB2(Standard 2Deviations • 2fromthe 1 3 2Mean) 3 24 20 4 355—1 32-2—3I I I20 30 40 50 60 70 80Overall Test Score (% Correct)I I0.4 .• BDeviation B Bfrom . •Expected 0.2 . • •Value B • • • .• B I B• . I I0.0. I B B B I IB B B B I B BB B• B B B--0.2• BBB—0.4I 1 INote: Normality of the distribution of these 47 nurses’ test scores fromsite #527 was rejected by the Kolmogorov—Smirnov test (p=0.008).Of fifty—five distributions tested, this was one of the more extremep—values among nine rejected. There were more scores near the middleof the distribution than expected, but the extent of this deviationwas not severe.Page 62Effectiveness of Infection Control StrategiesEach ICU nurse could not be tested on two occasions for a test—retest orparallel—forms assessment, a simple split—half single—test method would requiretoo many randomizations to provide statistically valid results,76 and a standardmethod to assess stability would be difficult to establish for tests consisting ofnonequivalent items. Therefore, reliability of the nurses’ knowledge test wasassessed in the following manner. Phi coefficients77 (a correlation coefficientfor two dichotomous variables) were computed for each of the eleven pairs ofquestions for each site returning at least ten completed tests (Figure 11).Questions had been paired on the basis of similar concept. Appendix 3 lists thequestions and their pairings; 0 indicates unscored items. The median of eachsite’s eleven phi values then was taken as an index for that site. Median phivalues were corrected for test length with the Spearman—Brown formula77(“stepped—up reliability”). These stepped—up medians then were used in meta—analysis78 to determine whether a statistically—significant positive correlationwas characteristic of all of the sites. That consistency would suggest a degree ofreliability, indicating that the new test may provide a reproducible measurement.Meta—analysis of phi coefficients reexpressed by Fisher’s z—transformationfailed to find significant differences among the sites (p>0.99, chi—square test forheterogeneity). This suggested that all sites shared a common correlation value,estimated as r0.l4, 95% confidence interval O.03—O.25. This weak butstatistically significant positive correlation between paired—responses providesweak evidence of test reliability. That conclusion was supported by findingcorrelations among the four knowledge section scores all to be positive (Table16). The frequency of discordance found in responses to question pairs suggeststhat many of the paired questions were not of equivalent difficulty. This limitsthe ability to explore test reliability from these data. Paired questions appearPage 63Effectiveness of Infection Control StrategiesFigure 11:Distribution of Phi Coefficients for 11 Paired Responses, by Site,in 22 Sites Returning 1O Tests—1.0 —0.5 0.0 0.5 1.0SITE_________F>6+ 81-( 95—f1—-— — 103— 104125+ 165179189(i]— 2250_____ 258284* 285-1.0 -0.5 0.0 0.5363Page 64Effectiveness of Infection Control StrategiesFigure 11 (contj:—1.0 —0.5 0.0 0.5 1.0SITE— 408418420(.j.f )___ 452- 481( 527_______.j3EJE) 541__* 558—1.0 —0.5 0.0 0.5 1.0at variable distances apart from each other on the test and the influence of theorder in which questions were presented was not studied experimentally.However, gross evidence of learning bias was not evident in a comparison ofdifferences in scores on a first versus second question of each pair (Figure 4).Relationship Between Infection Control Knowledge and PracticeIn examining the relationship between knowledge and practice, this workfailed to find a strong positive correlation. Since null hypothesis #4 was notrejected, one must consider the possibilities of insufficient statistical power todetect a true relationship, and of distortion by other variables (confounding).Page 65Effectiveness of Infection Control StrategiesSixty observations provide z99% power to detect change in a hospital fromrandom (50%) to minimally acceptable (at least 75%) compliance with glovingpolicy; a sample of 35 hospitals provides z93% power to detect strong positivecorrelations (r0.5). Thus, sample sizes appear to have been sufficient.However, knowledge and practice scores of individual nurses could not be linkedbecause of logistical constraints and requirements of anonymity. Comparison ofmean knowledge and practice scores for each hospital was a feasible but lessrefined approach. Pearson’s correlation coefficient provides greater statisticalpower to find statistically significant linear associations between normally—distributed variables than the Spearman coefficient; however, the latter is morerobust to non—normal distribution and more powerful at finding nonlinearassociations. The former was used on data that were normally distributed (ortransformed into a normal distribution), and the latter when normality could notbe assured. Since both coefficients were of similar magnitude and thus led tothe same conclusions with these data, these conclusions are more likely toreflect true facts than to be artifacts of a particular statistical model.Univariate correlations between knowledge and practice scores approached amoderately strong positive level (Table 16), but this appears to have resultedfrom confounding (Figure 7). When analyses corrected for differing numbers andcompositions of observations from the various hospitals, progressive improvementin quality of the data was associated with a negative trend in the correlation.Correlations found between knowledge and practice were not significant wheneach component of knowledge and practice scores was given equal weight (Table16). Canonical correlation analysis suggested that stronger correlations mightbe achievable by contrasting various aspects of knowledge and practice ratherthan giving each aspect equal weight (Table 17). Failure to achieve statisticalPage 66Effectiveness of Infection Control Strategiessignificance in these multivariate models may be due to a small sample size (35hospitals relative to 8 variables), since an accepted rule of thumb formultivariate analyses calls for at least ten observations per variable.79However, more importantly, the models suggested are not readily interpretable byany logic concerning what is known about the importance of these componentaspects of knowledge and practice.Bias in Knowledge, Practice and Injury MeasurementsThe 35 hospitals providing covert observations, knowledge tests andneedlestick rate data comprised a highly—motivated, non—random, self—selectedsample. They were primarily large community—hospitals from urban centers allacross Canada, but are not representative of all hospitals in size or type.Approximately 3% of western, mid—western and central but 10% of easternhospitals participated. Smaller facilities tended to be excluded by therequirement to make observations in a critical care unit, and very fewuniversity—affiliated hospitals elected to participate. Further, nurses testedwithin these hospitals were also volunteers. As a consequence of volunteerbias,8° these results might therefore be considered to represent maximal ratherthan mean values if extrapolated to all hospitals. Given the anonymous natureof the covert observations and nurses’ tests, it was not possible to make anyobjective assessment as to the direction or degree of bias. Self—selection biasmust therefore be considered an important limit to external validity. Selectivereporting bias (e.g.: not reporting injuries from “clean” needles) and recall biasmust also be kept in mind with regard to the needlestick injury data.8°Have Universal Precautions or Body Substance Isolation Been Effective asHarmonizing Strategies for Hospital Infection Control?Historically, numerous anecdotal reports and studies show poor compliancePage 67Effectiveness of Infection Control Strategiesas a common problem under traditional isolation approaches. Indeed, UP/BSIcould be considered behavior modification strategies to achieve bettercompliance. However, most of the few studies documenting performance underthese new strategies find poor compliance with infection control precautionsunder UP. Clements et al., in a covert “before and after” study, found higherrates of colonization and lower compliance with appropriate glove or gown useafter implementing UP.8’ Conversely, Klein et al., in a randomized prospectivestudy, found fewer infections and longer infection—free durations amongpediatric ICU patients under a protocol similar to BSI as compared to standardcare.82 Baroff et al.’s covert observation of a small number of emergencydepartment procedures and subsequent staff interviews report results consistentwith less than 50% compliance, poor exposure—risk knowledge, disagreement withgloving as providing protection during phlebotomy, and complaint that requiredprotective eyewear was not readily available.83 This is consistent with Kelenet al.’s findings.84 Hammond et al. similarly found poor compliance duringtrauma care even if patients were already identified or suspected of HIV.85Kaczmarek et al. found average glove—use rates of 92% (drawing arterial bloodgas) to 71% (phlebotomy) with significant differences between states with highversus low AIDS incidence rates. They conclude that “glove utilization byhealth—care workers during procedures in which they may contact patient bodyfluids is substantial, but not universal.”06 Lynch et al. described improvementin knowledge and compliance on some but not all wards after an intense andcomprehensive inservice effort.87 Wong et al. reported increased use of barriergarments (from 54% to 73%) associated with a decrease in exposure incidentsfollowing adoption of UP.88 However, unlike this thesis research which foundcomposition of observations to be an important confounding variable, thesePage 68Effectiveness of Infection Control Strategiesstudies relied upon convenience samples rather than standardized covertobservations. Further, completion of daily self—report forms in studies, such asthat of Wong et al., may itself prompt behaviour changes artificially. Otherself—report surveys that rely upon memory of past events are subject to well—recognized biases. Single—center studies might be considered as hypothesis—generating whereas multicenter research with standardized covert observationsprovides stronger evidence for or against universal trends.Hospitals have been motivated to adopt either UP or BSI to quell stafffears of occupationally—acquired HIV infection. Protection of staff against HIVhas been the prime motivation for change in Canadian hospitals. This wasevident in the administrative survey from reasons stated for adopting UP or BSI.It was also reflected in the finding that receipt of a staff education guidelinepublished in HIV—related literature was more common in every size group thanreceipt of the BSI handwashing guideline that would not be found in HIV—relatedliterature searches.This research confirms that, in practice, UP and BSI now mean differentthings in different hospitals. At the extremes, one site claiming to haveimplemented UP had eliminated specimen warning labels and all isolationcategories other than ‘Respiratory”: this is essentially BSI. Two others claimedadoption of BSI, one specifically recommending against UP, yet retained all ofthe original isolation categories: this is essentially UP. In many hospitalsclaiming UP, glove use policy was more in line with BSI. A recommendation tolimit UP primarily to visibly—bloody fluids was rejected by approximately half ofthe hospitals in each size group claiming both adoption of UP and receipt ofthis guideline. Further, additional terms were introduced by some respondents(“Routine Infection Control Precautions”, “Hygienic Precautions”, “GeneralPage 69- Effectiveness of Infection Control StrategiesPrecautions’, “Basic Precautions”, “Routine Precautions”, “Body SubstancePrecautions”, “Universal Barrier Precautions”, “Infection Control Precautions”).Some were synonyms for UP with precautions for all rather than just bloodybody fluids; others were synonyms for BSI with more emphasis on handwashing.Several survey responses suggested confusion between the recently—proposed BSI protocol and CDC’s existing “Blood/Body Fluids Precautions”isolation category. This may be due to inadequate receipt of information: tworeports by BSIs originators have not been received by a high proportion ofCanadian ICP’s. Among respondents in the larger size groups claiming adoptionof BSI, 36—60% and 18—40% reported not receiving the ANNALS OF INTERNALMEDICINE and ASEPSIS issues cited. This confusion was also evident amongnurses’ test responses: uniformly low scores were achieved on paired questionsregarding fundamental differences between UP and BSI, and a number of nursesadded written comments about being unfamiliar with the term BSI.A main finding of the administrative survey is that number of beds wasthe most significant determinant of guideline receipt. Fully one—third ofhospitals under 200 beds had not received either CDWR or MMWR guidelines.This is in contrast to the reported American rates of 80% receiving and 50%reviewing CDC Isolation guidelines.’3 All size groups received federal guidelinesmore commonly than those published in topical review or medical journals.Although CDWR and the LCDC Guidelines series were distributed without chargeto Canadian hospitals, neither was sent unless specifically requested. Theoriginal CDC Guidelines series issued in 1981 was mailed, unsolicited, to all U.S.acute—care hospitals. Presence of an ICP and of medical teaching programs werefound to be additional significant determinants for receipt independent ofbedsize, but interpretation of interaction terms (JCPsize and teachingsize) asPage 70Effectiveness of Infection Control Strategiessignificant determinants of guideline receipt will require further study.Conflicting guideline statements regarding staphylococcal pneumoniaprovided an opportunity to consider the depth of review applied by hospitals. ACDC Guideline4 specifies a private room (as well as staff use of masks andgowns) for staphylococcal pneumonia or draining lung abscess. This is intendedto control the extent of environmental contamination by virulent bacteria knownfor their ability to survive drying. An LCDC Guideline,89 derived from the CDCdocument, recommends neither private room nor masks, and specifies gowns forpneumonia but not lung abscess. Only two of the responding sites specificallycommented on this discrepancy between CDC and LCDC documents, one statingthat they had not previously noted it.The majority of Canadian hospitals adopting UP, BSI, or “Body SubstancePrecautions” have not eliminated special warning labels for recognized“infectious” cases, especially with regard to laboratory specimens. This mayinvite a double—standard. It is difficult to find evidence supporting a beliefthat such warning labels are effective prompts to take due caution.90 Over halfof ward staff in one confidential survey admitted to routinely using inadequateinfection control precautions in the care of patients known or suspected to haveHIV infection.91 Limited accuracy of such labeling was documented byHandsfield et al., who found that only 67% of HIV and 28% of HB5Ag seropositivespecimens were labelled.92 However, respondents commented that eliminatingspecimen warning labels is a controversial proposal. Similarly, Miller and Farrhave reported that while 86% of hospital epidemologists agreed with UP and 69%of 121 US hospitals established UP as policy, 74% retained special signs to markrooms of patients with HIV or hepatitis B.93Only two collaborative efforts between neighboring hospitals were evidentPage 71Effectiveness of Infection Control Strategiesin the administrative survey. Development of common guidelines and educationalpackages by an Ontario city’s Task Force for Implementation of Body SubstancePrecautions (BSP) (essentially BSI with greater emphasis on handwashing) wasthe most extensive. One Alberta hospital had initiated comprehensive studiesand acted as a regional resource; while adopting “Universal Precautions”, theirprogram retains only Respiratory Isolation, eliminates specimen warning labels,and rejects the limitation of UP to bloody fluids (thus, comparable to Ontario’sBSP).This lack of consistency between hospitals was reflected in observed staffpractices. Failure to wear gloves at appropriate times has been associated withtransmission of viral hepatitis and HIV.32 39 40 41 Addition of single— ordouble—gloving usually, but not invariably,93 has been associated withresolution in anecdotal hepatitis outbreak reports. While routine use of glovesas a universal precaution is a reasonable extension of logic, it remains to bedemonstrated that the strategy is effective when scaled up. Whether suchdemonstration would improve staff compliance also remains to be shown,especially since a strong correlation between knowledge and practice was notfound in this and another96 study.The optimal type and number of gloves to wear have also been points ofcontroversy. Case reports and, more significantly, outbreak investigations usingcase—control and molecular—typing methods have implicated failure to wear anygloves during direct contact with blood or other body substances in thetransmission of various nosocomial pathogens. While the potential severity ofsuch infections is self—evident from these reports, the exact magnitude of thisinfection risk and of its reduction through routine use of gloves are not known.Handwashing, “no—touch technique” and selective use of gloves were consideredPage 72Effectiveness of Infection Control Strategiesadequate in guidelines prior to the AIDS epidemic. However, in responding tofears triggered by AIDS, UP changed the focus of infection control fromprotecting patients to protecting healthcare workers. An historic emphasis onhandwashing and no—touch technique changed to an emphasis on routine use ofgloves. Blood contact with skin can be prevented by gloving.97 Double—glovinghas been shown to reduce the extent of blood exposure during surgery;98single— or double—gloving may be equally effective in reducing the volume ofblood transferred from a needlestick by 5O%.BSI also stressed routine use of gloves, but for a different reason:handwashing and no—touch technique alone were not proving effective inresolving bacterial cross—infection problems. These new mandates to expand theuse of gloves were based solely on extensions of logic. Logically, the next stepwould be to confirm assumptions of effectiveness. Unfortunately, attentionquickly turned from epidemiologic study of glove effectiveness to laboratorystudy of glove function. “Leakage rates were reported in several papers duringSession 55 (Protecting Health Care Personnel) at the October 1988 InterscienceConference on Antimicrobial Agents and Chemotherapy. While discussion stressedthe epidemiologic fact that no excess risk has been demonstrable given aseemingly adequate though less—than—perfect barrier, Dr. Frank Young,Commissioner of Food & Drugs, reported to the Third National Forum on AIDSand Hepatitis B last November [19881 that leakage levels are ‘alarmingly high’and that FDA plans to use ‘a more stringent and sensitive test for gloves thanthat used for screening condoms.”10° It remains to be demonstrated whetherroutine use of single— or double—gloves can reduce infection risk and whetherany such effect is jeopardized by “alarmingly high” leakage rates.Further evidence against the effectiveness of new infection controlPage 73Effectiveness of Infection Control Strategiesstrategies adopted by Canadian acute—care hospitals was found in needlestickinjury rates. In order to compare results of studies with varying durations ofexposure time, annual risk of one or more needlestick injuries can be estimatedas (1_(1_r)tj where ‘r’ equals a measured risk during shorter time intervals and“t” equals the number of such intervals in one year. The monthly risk of 2.3—3.5% found in this research equates to an annual risk of z245—350 injuries per1000 critical care nurses. The former figure is similar to an annual 273.4injuries per 1000 registered nurses reported recently by McCormick et al.’°1 andhigher than the 153—194 per 1000 nurses in a five—year study by Linnemannet al.102 The latter figure is similar to a recent survey of midwives in which24% claimed one or more needlestick injuries in a preceding six—month period(422 per 1000 nurses annually).’03 Hamory, correcting for 40% underreporting, found an annual incidence rate of 611 per 1000 nurses and cautionedagainst reliance upon employee health records to assess changes in injuryrates.’°4 Correction for under—reporting in the 2.3% health record based ratefound in this research suggests a figure similar to that reported by Hamory.The level of risk found among Canadian critical care nurses is commensuratewith previous reports and remains cause for concern. Beyond a potential fortransmission of hepatitis B and HIV, the direct costs of sharps injurypostexposure investigation and treatment are appreciable.’° 106The perception that needle—handling practices have not improved, theconfirmation that discarded needles are still recapped frequently, a finding ofinjury rates commensurate with rates reported prior to adoption of UP and BSIguidelines, demonstration of only borderline statistical significance in ratereduction trends and no evidence of a relationship between rates of needlerecapping and needlestick injury in this multicenter hospital study reinforce aPage 74Effectiveness of Infection Control Strategiessuggestion that UP and BSI have missed the point on needlesticks.62Needlestick injury is the predominant occupational exposure to HIV, andpreventing occupational HIV exposure was expressed as Canadian hospitals’primary motivation for adopting new infection control strategies.In conclusion, returning to Table l’s list of assessment criteria, theseresults provide little reason to believe that infection control practices havebecome uniform between or even within facilities under UP or BSI (Table 20).Guidelines defining these strategies were not received by all hospitals,especially those smaller facilities most in need of expert guidance, and thedegree to which those receiving guidelines reviewed their content may not beadequate. There were inconsistencies in policies between nearby hospitals aswell as within hospitals (e.g.: the low correlation between warning label use forhousekeeping versus the laboratory). Appreciable proportions of nurses had notreceived inservice education and held divergent views. Appreciable proportionsof nurses were observed to disregard their hospital’s policies on glove use andneedle recapping in spite of expressed agreement with those policies. Thisresearch also failed to produce convincing evidence linking reduced needlerecapping with reduced needlestick injury rates. The low level of effectivenessdiscovered for strategies which may cost as much as $10—million per case of HIVseroconversion prevented is disappointing.Recommendations for Improving the Research ToolsSurvey forms and questionnaires were designed, pretested and validatedbecause established tools were not available. These new tools may be of valuein future research studies and hospital quality assurance projects.The forms developed for the administrative survey could be useful forsimilar cross—sectional surveys in other countries, or for repeated surveys toPage 75Effectiveness of Infection Control StrategiesTable 20:Status of UP/BSI in Canadian Acute—Care Hospitals, 1990EVALUATION FINDINGS WITH REGARD TO:CRITERION STRUCTURE PROCESS OUTPUT OUTCOME1. Hospitals have z37% of <1% of all Few knew 19% couldreceived adequate smaller hospitals how UP identify theirinformation to hospitals noted major and BSI UP/BSI—relatedmake informed didn’t conflict in differ costsdecisions; receive CDC vs LCDCguidelines guidelinesHospitals adopted z75% claimappropriate policies UP or BSI;consistent with the 5% adoptedsystem selected. all policiesexpected2. Hospital staff Most sites Many nurses Mean staff Many nursesunderstand and mandated did not score z54% held divergentaccept UP or BSI education receive on test of opinions andfor all education knowledge disagreed withstaff re: UP/BSI new policies3. Staff adhere to All sites Hospitals Glove—userecommended UP or providing rarely compliance z6O%BSI guidelines in policies monitored but ranged widelytheir daily work mandated compliance among hospitalspractices; protective or providedapparel. feed—back z50% of hospitalsto staff reported 25%needle recappingUse of UP or BSIpractices leads to Needlestickreduced infection injury ratesrisk, comparablebefore andafter UP/BSIPage 76Effectiveness of Infection Control Strategiesstudy changes over time. This 1989 survey provided a picture of hospitaloperation at one point in time. It would be interesting to learn, for example, ifguideline receipt patterns change or if more becomes known about costs andbenefits of evolving new infection control strategies in the future.Two changes might make the questionnaires even more useful. First, sincethe types of “isolation’ used (e.g.: “Strict”, “Respiratory’, etc.) are intrinsic todefining which strategy or system was adopted by each hospital, this areashould be made more explicit. Sufficient information was volunteered in open—text replies, but more explicit interrogation on this point would be prudent.Second, since the marginal costs of UP and BSI are poorly detailed and theirbenefits are even more obscure, a series of questions dealing with economicaspects should be added. A careful cost—effectiveness study of alternativeinfection control strategies would provide important information.The forms used to record covert observations of daily practice could beuseful in hospitals’ on—going quality assurance audits. Substantial agreementamong different ICPs’ independent observations of nursing practices was notconfirmed in this research, but has been reported in a small study by others.107Observer accuracy and agreement, essential to validity and reliability ofinfection control research studies, deserve further study. The composition ofobservations sampled was not evaluated as a potential confounding variable inprevious studies, but was found as an important source of distortion in thisresearch. Future studies should establish observation quotas and schedules, aswas done in this research, rather than rely on convenience samples. It was notfeasible to evaluate observers’ performance, for example in watching a videotapeof specific nursing practices, but this could be considered in future studies.The knowledge and beliefs test validated in this study should be useful forPage 77Effectiveness of Infection Control Strategieshospitals’ educational needs assessments and evaluation of in—service educationprograms. With minor modification, it would provide a practical, valid andsufficiently reliable means to measure knowledge of infection risks and controlmeasures in nursing practice under UP/BSI. Such assessment is important indeveloping and promoting effective infection control programs. Use of this testin additional hospital studies is encouraged.This study used photocopied, double—sided forms with sequentially—numbered questions. Individual—history questions #7 and #32 (unscored items)and, collectively, questions #9—21 (the entire second page) were leftblank withdisproportionate frequency. While only zl.6% of returned tests were spoiled bymissing answers, this might be improved by better typeset quality or use ofsingle—sided forms.Interactive response to a computerized “interview”, unlike use of forms,would allow direct control of quality (no missing answers, out—of—rangeresponses, transcription errors, etc.) and encryption of data (supportingconfidentiality for participating staff). The software provided required neithercomputing nor typing expertise to install or operate. However, results are inagreement with unpublished data suggesting that fewer than one—third of NorthAmerican ICPs have computer support and nurses’ access to other computers islimited (market survey, Applied Epidemiology, Sidney, British Columbia).Computer—based survey of hospitals, whether by modem or mailed disks, is notyet a viable option for hospital epidemiology research.No serious ambiguity or validity concerns were expressed by ICPs aft)erreceiving tabulated summaries of test results. Inclusion of test statementsbetter answered by “yes” or “no” rather than on a Likert scale was identified asa source of difficulty in specific comments written on tests by critical carePage 78Effectiveness of Infection Control Strategiesnurses. Answers to such questions were penned in the margin rather thanindicated on the Likert scale of several tests. Several test items regardinglocal policy were phrased as requiring or prohibiting certain practices; inretrospect, it would be preferable to substitute ‘encouraging’ or “discouraging”since alternatives were allowed in some policies (e.g.: discouraging needlerecapping but offering a preferred method if recapping were deemed necessary).Potentially ambiguous test items specific to knowledge of local policy weretherefore not scored. There is no “gold standard’ against which to establishcriterion validity for this test. Satisfactory review by 35 ICPs is thereforeimportant confirmation of test validity achieved through appropriate test contentand expert consensus, but the test could be improved by rewording its localpolicy questions and substituting “yes/no” boxes for personal history responses.Further refinement and assessment of test reliability would also be desirable.The database management software written for this research is as importantas the forms themselves. Without it, the task of accurately extracting datafrom the forms would be insurmountable. Designing, programming, debugging anddocumenting the code required a formidable number of hours. The existingdatabase management system could be refined by adding user—level help ordocumentation; additional index files and options to search for hospitals byname, region, size, etc., rather than solely by site number; and by creating morestandard output report formats now that the methods of analysis are defined.Speed of operation and size of files could also be improved by upgrading to aproduct release more recent than the version of F0xBASE used.Page 79Effectiveness of Infection Control StrategiesVI. Conclusions and RecommendationsHealth policy is influenced by political, cultural, economic and technologicalfactors as well as by epidemiologic information. The latter is not alwayscompelling and sometimes frankly ignored. Some hospital epidemiologists andinfection control practitioners also have lacked sufficient status, authority,autonomy or support to shape hospital infection control policies efficiently.108Hospitals adopt procedures believed effective to prevent infection andperiodically modify their strategies when epidemiologic evaluations fail todocument effectiveness. Universal Precautions and Body Substance Isolation arethe most recent modifications, and these new national strategies were evaluatedin this multi—center research.Table 20 indicates failures at all levels of structure, process, output andshort—term outcome. The long—term outcome of reduced risk to hospital patientsand health—care workers remains a noble goal. In order to achieve this,improvements are needed in the forces acting upon hospital practice (structure),the measures taken within hospitals themselves (process) and measurement ofwhat actually is done as daily practice (output). A conceptual model of goals,objectives and related assumptions fundamental to all hospital infection controlprograms is shown in Appendix 1. Within this context, the following conclusionsand recommendations are made. Beyond these, the relatively small number ofhospitals completing all phases of this research underscores workload conflictsas a limiting factor. ICP staffing, supervision, computer support and prioritiesare related issues that cannot be addressed directly here. Alternatives to atraditional infection control program structure should be considered so thatapplied research may be supported more extensively in hospital epidemiology.109Page 80Effectiveness of Infection Control StrategiesRecommendations for Improving New Infection Control Strategies1. UP and BSI cannot be cost—effective solely in terms of protecting healthcare workers from blood—borne infection, the motivation for their adoptionclaimed most frequently in this research. UP and BSI are costly, their focus ison gloves and gowns whereas 70—90% of health care workers’ hepatitis B andHIV nosocomial infections derive from sharps injuries and the incidenceof suchinfections is already low. We need to refocus infection control on protectingpatients as well as staff and visitors. BSI may be cost—effective in thisbroader context, but research to explore this must receive greater priority.2. Sharps injuries pose a much greater risk of blood—borne infection forhealth care workers than does skin contact. However, this research found arelatively low proportion of hospitals using promising new protective devices incomparison to the larger proportion accepting increased costs for gloves andgowns. As noted by McCormick et al.,10 there should be commensurate spendingto purchase and evaluate novel devices which could reduce sharps injuries.3. UP and BSI are influential new initiatives, but have lost any specificmeaning in practice. Three—quarters of Canadian hospitals adopted the names,but not all of the policies recommended in published guidelines for thesestrategies. Confusion over the meaning of UP and BSI was evident in surveyresponses. Fewer than half of the nurses in half of the hospitals surveyed feltthat their hospital’s policies were practical, effective and well—documented.ICPs, hospital epidemiologists and other staff members in their hospitals shoulddiscuss the implications of these different strategies at the working level. UPand BSI need to be redefined from the perspective of health care workers, notconsultants who are distant from provision of care itself.4. Expert interpretation of research studies and consensus on recommendedPage 81Effectiveness of Infection Control Strategiespractices is potentially beneficial to all hospitals and health—care agencies.Those with their own resources can use these guidelines as starting points.Those without in—house or networked expertise could accept such guidelines asa basis for responsible policies, but were found to be the least likely to havereceived published guidelines. This presents two problems: How many differentexpert working groups need to produce guidelines independently; how can currentguidelines best be provided to hospitals? Better coordination should be apriority to reduce duplication of effort as well as opportunities to introducecontradictions and typographical errors which may mislead unwary end users. Aplethora of guidelines and revisions for UP has been issued over the past fewyears from federal, state, provincial, regulatory and professional associationswith but limited opportunity for widespread review of preliminary drafts byinterested parties. Since these guideline updates and revisions are published atirregular intervals, are advertised in publications to which only some hospitalssubscribe, and are not received unless specifically requested, a means forplacing standing orders or otherwise improving distribution would help to assurethat appropriate parties always have current versions of recommended practices.5. Administrators need to ensure that a comprehensive, practical systememerges which is consistent with a guiding philosophy and capable of achievingclearly stated, realistic goals. Further, administrators need to ensure thatpolicies in all departments throughout their hospital become consistent with thesystem selected as quickly as possible. Piecemeal implementation, poorknowledge of costs and inconsistency in policies were evident in this research.6. The prospect of regulatory or accreditation agencies stifling appliedresearch by mandating one set of unproven guidelines as the only acceptablePage 82Effectiveness of Infection Control Strategiesapproach is cause for concern. The prospect of guidelines being acceptedwithout careful review by qualified individuals in health—care institutions isequally distressing. This research failed to produce convincing evidence linkingUP, BSI and avoidance of needle recapping with widespread improvement in ratesof needlestick injury, the only quantifiable risk for occupational HIVtransmission. As Gerberding recommends, “Until additional research clarifies thevalue of current infection control policies, standards should maximizeinstitutional autonomy to develop rational strategies consistent with localpractice and perceived needs.”9° Hospitals need tools now, not rules.7. Higher priorities should be placed on having ICPs and hospitalepidemiologists carefully assess implications of changes proposed in newguidelines, discuss new recommendations with other health—care workers so as toprovide liaison with those who must actually live with proposed measures, andassist in establishing institutional process review audit mechanisms. Fewhospitals were prepared to undertake such audits. This research documentedwidespread deficiencies in knowledge and policy compliance as well as lack ofcorrelation between knowledge and practice. Validated audit tools are needed.8. Noncompliance again was found to be a primary impediment to theeffectiveness of infection control strategies. Innovative approaches to motivatebehavior change must be developed and evaluated. Further research shouldseek unique features that distinguish between hospitals exhibiting relatively lowversus high levels of staff compliance with policies.9. In future studies, compliance should be measured by standardized covertobservations. The number and composition of such observations was found to bean important confounding variable in this research.BSI was formulated through a fresh look at unresolved cross—infectionPage 83Effectiveness of Infection Control Strategiesproblems. UP replied to an emotional response triggered by the fatal nature ofAIDS, not by quantified occupational—risk assessment of blood—borne diseases(hepatitis B, with its quantitatively higher risk of morbidity and mortality forhealth care workers, should have motivated the development of such precautionsmuch earlier). This dichotomy, together with a paucity of hospital—initiatedapplied research found in the biomedical literature, prompts one finalrecommendation. Hospitals have traditionally focused on problem identificationas their basis for quality assurance. Some hospitals are now expressing interestin industrial methods of Total Quality Management (TQM) and Continuous QualityImprovement (CQI).11° Infection control, like quality assurance, should beviewed as a journey of evolution, not simply enforced compliance with one set ofrules. Administrators are long overdue in expecting qualified infection controlor hospital epidemiology staff to provide important program planning andevaluation information.11’112 In fact, TQM and CQI bear many similarities toWilliamson’s method for finding “achievable benefit not achieved” opportunitiesto improve quality of service.”3 Hospital epidemiology and infection controlprograms should be expected and supported to provide the tools and mechanismsto take fresh looks at all aspects of hospital service. Their heritage of beingreactive to adverse events will not continue to suffice in a future that requiresa proactive focus on quality improvement.Page 84E’ffectiveness of Infection Control StrategiesBibliography1. LaForce FM. The Control of Infections in Hospitals: 1750 to 1950.In: Wenzel RP, ed. PREVENTION AND CONTROL OF NOSOCOMIAL INFECTIONS.Baltimore. MD: Williams and Wilkins 1987:1—12.2. Bullock WE. Control of Nosocomial Infections Through Patient Isolation.In: PROCEEDINGS OF THE NATIONAL CONFERENCE ON INSTITUTIONALLYACQUIRED INFECTIONS. Washington DC: U.S. Government Printing Office1964:153—1613. Eickhoff TC. International Symposium on Control of Nosocomial Infections:Synthesis and Summary. In: Sacks T, McGowan JE Jr, eds. InternationalSymposium on Control of Nosocomial Infection. REV INFECT DIS1981;3(4): 798—803.4. Garner JS, Simmons BP. Centers for Disease Control Guideline for IsolationPrecautions in Hospitals. INFECT CONTROL. 1983;4(4):245—325.5. Nauseef WM, Maki DG. A Study of the Value of Simple Protective Isolation inPatients with Granulocytopenia. N ENGL J MED 1981;304(8):448—453.6. Pettinger A, Nettleman MD. Epidemiology of Isolation Precautions. INFECTCONTROL HOSP EPIDEMIOL 1991; 12(5):303—307.7. Centers for Disease Control. Recommendations for prevention of HIVtransmission in health—care settings. MMWR 1987;36(suppl 2):3S—18S.8. Jackson MM, Lynch P. Cummings MJ, Stamm WE. Rethinking the Role ofIsolation Practices in the Prevention of Nosocomial Infections. ANN INTERNMED 1987;107:243—246.9. Garner JS, Hughes JM. Options for Isolation Precautions. ANN INTERN MED1987; 107:248—250.10. Campbell B. An Open Letter to Infection Control Personnel. DIMENSIONS INHEALTH SERVICE 1988;Feb.:40.11. CDC. Update: Universal Precautions for Prevention of Transmission of HumanImmunodeficiency Virus, Hepatitis B Virus and Other Bloodborne Pathogensin Healthcare Settings. MMWR 1988;37(24):377—88.12. Manian F. Universal Precautions TClarified”? INFECT CONTROL HOSPEPIDEMIOL l988;9(8):343—4.13. Celentano DD, Morlock LL. Malitz FE. Diffusion and Adoption of CDCGuidelines for the Prevention and Control of Nosocomial Infections in USHospitals. INFECT CONTROL 1987;8(lO):415—423.14. Gerberding JL, Henderson DK. Design of Rational Infection Control Policiesfor Human Immunodeficiency Virus Infection. J INFECT DIS 1987;156(6):861—864.Page 85Effectiveness of Infection Control Strategies15. Department of Labor, Department of Health and Human Services. JointAdvisory Notice: Protection Against Occupational Exposure to Hepatitis BVirus (HBV) and Human Immunodeficiency Virus (HIV). FEDERAL REGISTERl987;52(210)41818—41823.16. Birnbaum D, Kelly M, Chow AW. Epidemiologic Typing Systems for Coagulase—Negative Staphylococci. INFECT CONTROL HOSP EPIDEMIOL 1991;12(5):319—26.17. Suchman EA. EVALUATIVE RESEARCH. New York, NY: Russel Sage Foundation;1967:115.18. Woodward CA, Chambers LW, Smith KD. GUIDE TO IMPROVED DATACOLLECTION IN HEALTH AND HEALTH-CARE SURVEYS. Canadian PublicHealth Association: 1982.19. Knapp TR. Validity, Reliability, and Neither. NURSING RESEARCH1985;34(3): 189—1 92.20. Weinstein RA, Kabins SA. Isolation Practices in Hospitals. ANN INTERN MED1 987;5(107):781 —782.21. Doebbeling BN, Pfaller MA, Houston AK, et al. Removal of NosocomialPathogens from the Contaminated Glove, Implications for Glove Reuse andHandwashing. ANN INTERN MED 1988; 1 09:394—398.22. Cochran WG. SAMPLING TECHNIQUES, 3rd edition. New York NY: John Wiley &Sons 1977:Equations #5A.42, 13.2 and 13.3.23. Miettinen OS. THEORETICAL EPIDEMIOLOGY, PRINCIPLES OF OCCURRENCERESEARCH IN MEDICINE. New York, NY: John Wiley & Sons 1985:pp.136—7.24. Larson E. A Causal Link between Handwashing & Risk of Infection?Examination of the Evidence. INFECT CONTROL HOSP EPIDEMIOL1988;9(1):29—36.25. Fox MK, Langner SB, Wells RW. How Good are Handwashing Practices?. AM JNURSING 1974;74(9):1676—8.26. Albert RK, Condie F. Handwashing Patterns in Medical Intensive—Care Units.N ENGL J MED 1981;304(24):1465—6.27. Mayer JA, Dubbert PM, Miller M, et al. Increasing Handwashing in anIntensive Care Unit. INFECTION CONTROL 1986;7(5):259—62.28. Kaplan LM, McGuckin M. Increasing Handwashing Compliance with MoreAccessible Sinks. INFECTION CONTROL 1 986;7(8):408— 10.29. De Carvalho M, Lopes 3M, Pellitteri M. Frequency & Duration of Handwashingin a Neonatal Intensive Care Unit. PEDIATR INFECT DIS 3 l989;8(3):179—80.30. Simmons B, Bryant 3, Neiman K, et al. The Role of Handwashing inPrevention of Endemic Intensive Care Unit Infections. INFECT CONTROLHOSP EPIDEMIOL 1900;l1(l1):589—94.Page 86Effectiveness of Infection Control Strategies31. Adams G, Stover BH, Keenlyside RA, Hooton TM, Buchman TG, et al.Nosocomial Herpetic Infections in a pediatric intensive care unit. AM JEPIDEMIOL 1981;113(2):126—32.32. Baptiste R, Koziol D, Henderson DK. Nosocomial Transmission of Hepatitis Ain an Adult Population. INFECTION CONTROL 1987;8(9):364—70.33. Casewell M, Phillips I. Hands as Route of Transmission for Kiebsiella species.BMJ 1977;l9Nov:1315—7.34. CDC. Nosocomial Acquisition of Aeromonas hydrophila. National NosocomialInfections Study Report, Annual Summary 1977, Issued November 1979; pp23—5.35. CDC. Acinetobacter calcoaceticus: Common Nosocomial Organism, UncommonPathogen. National Nosocomial Infections Study Report, Annual Summary1976, Issued 1978, ppll—13.36. Church DL, Bryant HE. Investigation of Streptococcus viridansPseudobacteremia Epidemic at a University Teaching Hospital. INFECTCONTROL HOSP EPIDEMIOL 1989;10(9):416—21.37. McFarland LV, Mulligan ME, Kwok RYY, et al. Nosocomial Acquisition ofClostridium diffIcile Infection. N ENGL J MED 1989;320:204—210.38. Johnson S, Gerding DN, Olson MM, et al. Prospective Controlled Study ofVinyl Glove Use to Interrupt Clostridium difficile Nosocomial Transmission.AM J MED 1990;88:l37—40.39. Snydman DR, Hindman SH, Winelanci MD, et al. Nosocomial Viral Hepatitis B:A Cluster Among Staff with Subsequent Transmission to Patients. ANNINTERN MED 1976;85:573—7.40. CDC. Apparent Transmission of Human T—Lymphotropic Virus Type 111/Lymphadenopathy—Associated Virus from a Child to a Mother Providing HealthCare. MMWR 1986;35(5):76—9.41. CDC. Update: human immunodeficiency virus infection in health—careworkers exposed to blood of infected patients. MMWR 1987;36(19):285—9.42. Beaumont LR. Detection of Blood on Nonporous Environmental Surfaces: AnApproach for Assessing Factors Contributing to the Risk of OccupationalExposure to Blood in the Autopsy Suite. INFECTION CONTROL1987 ;8( 10) : 424—9.43. York MK. Bacillus Species Pseudobacteremia Traced to Contaminated GlovesUsed in Collection of Blood from Patients with Acquired ImmunodeficiencySyndrome. J CLIN MICROBIOL 1990;28(9):2114—16.44. Patterson JE, Vecchio J, Pantelick EL, et al. Association of ContaminatedGloves with Transmission of Acinetobacter calcoaceticus var, anitratus inan Intensive Care Unit. AM J MED 1991;91:479—83.Page 87Effectiveness of Infection Control Strategies45. Bagg J, Jenkins S, Barker GR. Laboratory Assessment of the AntimicrobialEffectiveness of Glove Washing and Re—Use in Dental Practice. J HOSP INFECT1990;15:73—82.46. Korniewicz DM, Laughon BE, Cyr WH, et al. Leakage of Virus through UsedVinyl and Latex Examination Gloves. J CLIN MICROBIOL 1990;28(4):787—8.47. Cookson B, Peters B, Webster M, et al. Staff Carriage of EpidemicMethicillin—Resistant Staphylococcus aureus. J CLIN MICROBIOL1989;27(7):1471—6.48. Doebbeling BN, Pfaller MA, Houston AK, Wenzel RP. Removal of NosocomialPathogens from the Contaminated Glove — Implications for Reuse. ANN INTERNMED 1988;109(5):394—8.49. Larson E. Reported Cases of Allergic Reactions to latex Gloves on the Rise.INFECT CONTROL 110SF EPIDEMIOL 1991; 1 2(8):504—5.50. Daschner FD, Habel H. HIV Prophylaxis with Punctured Gloves (letter).INFECT CONTROL HOSP EPIDEMIOL 1 988;9(5): 184—6.51. Thomas GA, Talbot G, Jahre J, et al. Prospective Evaluation of Health—CareWorkers Exposed via Parenteral or Mucous—Membrane Routes to Blood andBody Fluids of Patients with Acquired Immunodeficiency Syndrome. MMWR1 984;33(13): 181—182.52. CDC. Update: Prospective Evaluation of Health—Care Workers Exposed via theParenteral or Mucous—Membrane Route to Blood or Body Fluids from Patientswith Acquired Immuno—Deficiency Syndrome— United States. MMWR1985;34(7): 101 —3.53. Nadler J, Landesman 5, Rechtman D, et al. Update: Evaluation of HTLV—III/LAV Infection in Health-Care Personnel— United States. MMWR1 985;34(38):575—8.54. Stricof RL, Mores DL. HTLV—III/LAV Seroconversion Following A DeepIntramuscular Needlestick Injury. N ENGL J MED 1986;314(17):1115.55. McCray E. Occupational Risk of the Acquired Immunodeficiency SyndromeAmong Health—care Workers. N ENGL J MED 1986;314(7):1127—32.56. Oksenhendler E, Harzic M, LeRoux JM. I-fly Infection with Seroconversion aftera Superficial Needlestick Injury to the Finger. N ENGL J MED1 986;315(9):582.57. Neisson—Vernant C, Arfi 5, Methez D, et al. Needlestick HIV Seroconversionin a Nurse. LANCET 1986;ii;814.58. CDC. Apparent Transmission of Human T—Lymphotropic Virus TypeJIT/Lymphadenopathy—Associated Virus from a Child to a Mother ProvidingHealth—care. MMWR 1986;35(5):76—79.Page 88Effectiveness of Infection Control Strategies59. Elmslie K. OShaughnessy MV. National Surveillance Program on OccupationalExposure to HIV Among Health—Care Workers in Canada. CDWR1987; 13(37): 163—6.60. Marcus R, CDC Cooperative Needlestick Surveillance Group. Surveillance ofHealth—care Workers Exposed to Blood from Patients with the HumanImmunodeficiency Virus. N ENGL J MED 1988;319:1118—23.61. Mangione CM, Gerberding JL, Cummings SR. Occupational Exposure to HIV:Frequency and Rates of Underreporting of Percutaneous and MucocutaneousExposures by Medical Housestaff. AM J MED 1991;90:85—90.62. Jagger J. Pearson RD. Universal Precautions: Still Missing the Point onNeedlesticks. INFECT CONTROL HOSP EPIDEMIOL 1991; 12(4):2 11—13.63. Machin D, Campbell MJ. STATISTICAL TABLES FOR THE DESIGN OF CLINICALTRIALS. Blackwell Scientific Publications, Oxford. 1987. Table 12.1.64. McGill R, Tukey JW, Larsen WA. Variations of Box Plots. THE AMERICANSTATISTICIAN 1978;32:12—16.65. Dallal GE, Wilkinson L. An Analytic Approximation to the Distribution ofLililefors’ Test Statistic for Normality. AMERICAN STATISTICIAN1986;40(4):294—6.66. Rothman KJ. MODERN EPIDEMIOLOGY. Boston, Mass.: Little, Brown & Company;1986:41 —49.67. Cleveland WS. THE ELEMENTS OF GRAPHING DATA. Monterey, CA: Wadsworth,Inc.; 1985:118—23.68. Johnson RA, Wichern DW. APPLIED MULTIVARIATE STATISTICAL ANALYSIS.2nd ed. Englewood Cliffs, New Jersey: Prentice Hall mc; 1988:438—69.69. Geigy Scientific Tables. 7th ed. Switzerland: CIBA—GEIGY Ltd; 1970:66.70. Fishman M. Cathers AP, Stamp D. Needle Punctures — Documentation andIncidence Rate Calculation. INFECTION CONTROL 1 985;6(1 ):35—36.71. Remington RD, Schork MA. STATISTICS WITH APPLICATIONS TO THEBIOLOGICAL AND HEALTH SCIENCES. Englewood Cliffs, New Jersey. PrenticeHall Inc. 1985:316—319.72. Doebbeling BN, Wenzel RP. The Direct Costs of Universal Precautions in aTeaching Hospital. JAMA 1990;264(16):2083—87.73. Stock SR, Gafni A, Bloch RE’. Universal precautions to prevent HIVtransmission to health care workers: an economic analysis. CAN MED ASSOCJ 1990;142(9):937—46.74. Abramson JH. MAKING SENSE OF DATA. New York, NY: Oxford UniversityPress; 1988.Page 89Effectiveness of Infection Control Strategies75. Lilliefors HW. On the Kolmogorov—Smirnov Test for Normality with Mean andVariance Unknown. J AMERICAN STATISTICAL ASSOC 1967 ;62 :399—402.76. Kuder GF, Richardson MW. The Theory of the Estimation of Test Reliability.PSYCHOMETRIKA 1937;2(3):151—60.77. Lord FM, Novick MR. STATISTICAL THEORIES OF MENTAL TEST SCORES.Reading, Mass.: Addison—Wesley 1968:84, 336 (equations 4.2.10 and 15.5.9)78. Hedges LV. STATISTICAL METHODOLOGY IN META-ANALYSIS. ERIC/TM Report83, University of Chicago, 1982. pp.55—61.79. Flury B, Riedwyl H. MULTIVARIATE STATISTICS, A PRACTICAL APPROACH. NewYork, NY: Chapman and Hall; 1988:9.80. Sackett DL. Bias in Analytic Research. J CHRON DIS 1979;32:51—63.81. Clements JS, Hampton KD, Wasilauskas BL, et al. Impact of UniversalPrecautions on the Acquisition of Gentamicin—Resistant Gram NegativeBacilli in a University Intensive Care Unit. abstract #787, 1989Interscience Conference on Antimicrobial Agents and Chemotherapy.82. Klein BS, Perloff WH, Maki DG. Reduction of Nosocomial Infection DuringPediatric Intensive Care by Protective Isolation. N ENGL J MED1989;320(26):1714—21.83. Baraff U, Talan DA. Compliance with Universal Precautions in a UniversityHospital Emergency Department ANN EMERG MED 1989;18:654—7.84. Kelen G, DiGiovanna T, Kalainov D, et al. Adherence to Universal Precautionsby HCWs in an Inner—City Emergency Department with High HIV-1Prevalence among Patients. abstract W.A.O.4, 5th International Conferenceon AIDS, June 1989, Montreal.85. Hammond JS, Eckes JM, Gomez GA, Cunningham DN. HIV, Trauma and InfectionControl: Universal Precautions Are Universally Ignored. J TRAUMA1 990;30(5):555—61.86. Kaczmarek RG, Moore RM Jr, McCrohan J, Lowe JT. Multi—state Investigationof Glove Utilization by Health—care Providers. Abstract Cli, 3’rdInternational Conference on Nosocomial Infections, Atlanta GA, 1990.87. Lynch P, Cummings MJ, Roberts PL, et al. Implementing and evaluating asystem of generic infection precautions: body substance isolation. AM JINFECT CONTROL 1990;18:1—12.88. Wong ES, Stotka JL, Chinchilli VM, et al. Are Universal Precautions Effectivein Reducing the Number of Occupational Exposures Among Health—careWorkers? JAMA 1991;265(9):1123—28.89. Bureau of Infection Control, Health & Welfare Canada. Infection ControlGuidelines for Isolation and Precaution Techniques. 1985.Page 90Effectiveness of Infection Control Strategies90. Gerberding JL. Does Knowledge of Human Immunodeficiency Virus InfectionDecrease the Frequency of Occupational Exposure to Blood? AM J MED1991;91(suppl 3B):308S—311S.91. Gerberding JL, Bryant—LeBlanc CE, Nelson K, et al. Risk of Transmitting theHuman Immunodeficiency Virus, Cytomegalovirus & Hepatitis B Virus toHealth Care Workers Exposed to Patients with AIDS & AIDS—RelatedConditions. J INFECT DIS 1987;156(l):l—8.92. Handsfield HH, Cummings MJ, Swenson PD. Prevalence of Antibody to HumanImmunodeficiency Virus and Hepatitis B Surface Antigen in Blood SamplesSubmitted to a Hospital Laboratory— Implications for Handling Specimens.JAMA 1987;258(23):3395—7.93. Miller P, Farr B. A survey of SHEA Members on Universal Precautions andHIV Screening. INFECT CONTROL HOSP EPIDEMIOL 1988;9(4):163—5.94. Hamm RH, Peare RB, Painter WL, et al. Hepatitis B among Dental Patients—Indiana. MMWR 1985;34(5):73—5.95. Lettau LA, Smith D, Williams D, et al. Transmission of Hepatitis B withresultant Restriction of Surgical Practice. JAMA 1986;255:934—7.96. Gruber M, Beavers FE, Johnson B, et al. The Relationship between Knowledgeabout Acquired Immunodeficiency Syndrome and the Implementation ofUniversal Precautions by Registered Nurses. CLINICAL NURSE SPECIALIST1 989;3(4): 182—5.97. Bell DM. Human Immunodeficiency Virus Transmission in Health Care Settings:Risk and Risk Reduction. AM J MED 1991;91(suppl 3B):294S—300S.98. Gerberding JL, Littell C, Tarkington A, et al. Risk of Exposure of SurgicalPersonnel during Surgery at San Francisco General Hospital. N ENGL J MED1 990;322(25): 1788—93.99. Mast ST, Gerberding JL. Factors Predicting Infectivity Following NeedlestickExposure to HIV in an in vitro Model. CLIN RES 1991;39(l):58A.100. Birnbaum D. The informed use of technology. JOURNAL OF HEALTHCAREMATERIEL MANAGEMENT 1 989;7(4):86.101. McCormick RD, Meisch M, Ircinik F, Maki DG. Epidemiology of Hospital SharpsInjuries: A 14—Year Perspective in the Pre—AIDS and AIDS Eras. Abstract#93, 3’rd Decennial International Conference on Nosocomial Infections.Centers for Disease Control, Atlanta, Georgia, 1990.102. Linnemann Jr. CC, Cannon C, DeRonde M, Lamphear B. Effect of EducationalPrograms, Rigid Sharps Containers, and Universal Precautions on ReportedNeedlestick Injuries in Healthcare Workers. INFECT CONTROL HOSPEPIDEMIOL 1991;12(4):214—219.Page 91Effectiveness of Infection Control Strategies103. Loewen NL, Dhillon GL, Willy ME, Wesley RA, Henderson DK. Use ofPrecautions by Nurse—Midwives to Prevent Occupational Infections withHIV and Other Blood—Borne Diseases. J NURSE MIDWIFERY 1989;34(6):309—317.104. Hamory BH. Underreporting of needlestick injuries in a university hospital.AM J INFECT CONTROL 1983;11(5):174—177.105. McCormick RD, Meisch MG, Ircink FG, Maki DG. Epidemiology of HospitalSharps Injuries: A 14—Year Prospective Study in the Pre—AIDS and AIDSEras. AM J MED 1991;91(suppl 3B):301S—307S.106. Jagger J, Hunt EH, Pearson RD. Estimated Cost of Needlestick Injuries forSix Major Needled Devices. INFECT CONTROL HOSP EPIDEMIOL1990; 11(1 1):584—88.107. Gauthier DK, Turner JG, Langley LG, et al. Monitoring Universal Precautions:A New Assessment Tool. INFECT CONTROL HOSP EPIDEMIOL199 1;12(10):597—601.108. Soule BM. The Evolution of our Profession: Lessons from Darwin. AM JINFECT CONTROL 1991;19(1):45—59.109. Birnbaum D. Infection control consortia. J HEALTHCARE MATERIELMANAGEMENT 1991;9(9):76—77.110. Murdock M. Continuous improvement in health—care: An overview. JHEALTHCARE MATERIEL MANAGEMENT 1991 ;9(1):72-3.111. Massanari RM. Risk Management: An Epidemiological Approach (Editorial).INFECTION CONTROL 1987;8(1):3—6.112. Wenzel RP, Pfaller MA. Infection Control: The Premier Quality AssessmentProgram in United States Hospitals. AM J MED 1991;91(suppl 3B):27S—31S.113. Williamson JW. Formulating Priorities for Quality Assurance Activity. JAMA1978;239(7):631—37.Page 92Effectiveness of Infection Control StrategiesAppendix 1: Conceptual Models:Comparison of Isolation StrategiesInfection Control Goals, Objectives and AssumptionsPage 93Effectiveness of Infection Control StrategiesTable 21:Comparison of Isolation StrategiesSTRATEGY: PURPOSE: PROMPTED BY: APPLIES ¶I): DESCRIPTION:TRADITIC*ThL STRATEGIFS:Category—Specific Prevent the Patient’s Certain body SelectiveIsolation spread of Diagnosis fluids and application ofCDC, 1983 microorganisms substances supplementaryamong patients, (disease- precautionspersonnel and specific) grouped intovisitors, categoriesDisease—Specific Prevent the Patient’s Certain body SelectiveIsolation spread of Diagnosis fluids and application ofCDC, 1983 microorganisms substances supplementaryamong patients, (disease— precautionspersonnel and specific) selected item—visitors. by—itemNEW STRATEGIES:Body Substance Reduce cross- Type of All body ReplacesIsolation infection risk Patient fluids and Category— orLynch, Jackson to patients, Contact substances Disease—Specificet al. 1987 and protect staff System with onefrc*n microorganisms set of hygienicharbored by patients measures for allUniversal Precautions Reduce risk to Type of AU body A basic levelCDC, 1987 staff from Patient fluids and of hygienicbloodhorne Contact substances measures used inpathogens conjunction withtraditional systemUniversal Precautions Reduce risk to Type of Body fluids A basic levelCDC, 1988 Revision staff from Patient associated of hygienicbloodborne Contact with measures used inpathogens and Body hepatitis B conjunction withFluid transmission traditional systemPage 94Effectiveness of Infection Control StrategiesTable 22:Infection Control Goals, Objectives and AssumptionsLong—term Goals:1. Minimized occupational risk of nosocomial infection for staff2. Minimized iatrogenic risk of nosocomial infection for patients3. Minimized risk of widespread antimicrobial resistance for communityAssumptions:1. Staff and patients can be infected through exposure incidents, but theseincidents are preventable through “isolation” and/or other precautionprotocols. UP assumes that isolation of identified cases is necessaryand effective; BSI assumes that all cases cannot be identified reliablyand that special isolation” protocols to prevent direct or indirectcontact transmission are inappropriate.2. Hospital—acquired infections are sufficiently serious to warrantpreventive action.3. Transmission of exogenous infection to patients can be prevented.4. Following expert guidelines will, in fact, reduce infection risk.5. Prevention is more cost—effective than treatment alone.6. Cross—infection contributes more to the prevalence of drug resistancethan de novo development of resistant organisms.7. Hospitals are major sources of novel drug resistance in the community.Intermediate—term Objectives1. Institutions will develop and update infection control policiesand procedures from pertinent published guidelines.2. Staff will follow infection control policies and proceduresin providing patient care.3. Hospitals will monitor infection rates and staff performance.Assumptions:1. Institutional programs review appropriate sources of information.2. Policies developed from guidelines and refined through in—housemonitoring are perceived as practical and beneficial by ward staff.3. UP assumes that “isolation” is an effective behaviour—prompt toimprove handwashing and gloving; BSI assumes that a multitudeof isolation types is unnecessarily confusing.Short—term Objectives1. Pertinent expert guidelines will be available to hospitals.Assumptions:1. Guidelines will be effective in shaping institutional policy.2. Expert interpretation is required.Page 95Effectiveness of Infection Control StrategiesAppendix 2: Hospital Survey Questionnaire FormsEnglish VersionFrench VersionPage 96Effectiveness of Infection Control StrategiesTO: INFECTION CONTROL PRACTITIONERFROM: David Birnbaum, MPHRE: Evaluation of Universal Precautions/Body Substance IsolationINTRODUCTION:I am a Ph.D. student registered in an interdisciplinary program at University ofBritish Columbia, supported by a Fellowship from the National Health Researchand Development Program of Health and Welfare Canada. My thesis studies useand effectiveness of Universal Precautions and Body Substance Isolation. Thisstudy is endorsed by the Canadian Hospital Association. Your assistance inconsenting to participate by completing this confidential questionnaire will begreatly appreciated. It should take no more than 15 minutes, and you are underno obligation to participate in subsequent phases.This first phase investigates current Canadian hospital practices; a second phasewill involve on—site study of implementation in selected hospitals; the finalphase will compare effectiveness in reducing infection exposure risks to patientsor staff. This questionnaire provides the means to indicate if you would like toreceive preliminary results or details of subsequent phases. Published resultswill not identify individual facilities nor infection control practitioners by nameor location. Please return your completed questionnaire in the mailing envelopeprovided or by FAX (604—875—4013). Please include your mailing address if youwish to receive final results of this survey or wish to participate further.FACILITY DESCRIPTION:Please fill in the blank or circle the appropriate number(s) to describe your facility:1. MAJOR FUNCTION: (1) Acute—Care Primary Community Hospital(2) Acute—Care Regional Referral Hospital(3) University Teaching Hospital(4) Residential Care Facility(5) Other (specify)___________________________2. NUMBER OF BEDS:______3. NUMBER OF INPATIENTS/YEAR:_____4. NUMBER OF OUTPATIENTS/YEAR:5. NUMBER OF INFECTION CONTROL PRACTITIONERS: FULL-TIME— PART-TIME —6. LOCATION: (1) City Centre (3) Town/Village(2) City Suburb (4) Unincorporated Area7. DO YOU PROVIDE: (1) Hemodialysis (4) Internships(2) Drug Abuse Clinics (5) Residency Training(3) STD/AIDS Clinics8. WHEN WERE YOU LAST ACCREDITED? Date_ _ _AgencyStatusPage 97Effectiveness of Infection Control StrategiesUNIVERSAL PRECAUTIONS/BODY SUBSTANCE ISOLATIONHas your facility made a formal decision to adopt or reject UNIVERSALPRECAUTIONS or BODY SUBSTANCE ISOLATION? Which one? What factorswere influential in the decision?What advantages and problems do you foresee in implementing thesestrategies?Do you have any data regarding implementation or operating costs as comparedto category— or disease—specific isolation programs?Regardless of policy, do you feel that at least half of ward staffs stillrecap used needles?What type of disposal containers are in use, and how are needlestransported from the point of use to the point of disposal (ie: recapped,placed on tray, stabbed into foam block, hand—carried, etc.)?What additional information would you want, if any, in order toconfidently recommend for or against Universal Precautions?.to recommend for or against Body Substance Isolation?Page 98Effectiveness of Infection Control StrategiesCURRENT POLICIES:FOR ERCR POLICY PLEASR FILL I XITBXR OE YIS’ OR O BOX I 11CR COLUfl YO HiDICITX IF YOU Bill ALRRW:RECEIVED PUBLICATION REVThE) PUBLICATION ADOPTED GIJIDELThTEPOLICY GUIDELINE & SOURCE: YES NO YES NO YES-DATE (nn/yj) NOUniversal Precautions endsneed for warning labels onspecinens. CANADA DISEASESWEEKLY REPORT 1987;1353Handwashing is unnecessaryafter removing gloves wornfor anticipated contact withblood, secretions • mucousmembranes or lesions unlesshands are visibly soiled.ANN I1T1’ERN MED 1987;107:243Private room is recommendedfor Staph. pneumonia. 1983CDC “GUIDELINE FOR ISOLATIONPRECAUTIONS IN HOSPITALS”Private room not reccnanendedfor Staph. pneumonia. 1985HEALTH & WELFARE CANADA“ISOLATION AND PRECAUTIONTECHNIQUES” INFECTIONCONTROL GUIDELINESSecurely bag all trash &linen to prevent leakage;no additional labels, doublebags or special handling areneeded for infectious cases.ASEPSIS 1986;8(4):2Health care workers shouldbe trained in infection—control procedures designedto prevent transmission ofblood-borne pathogens andbe advised to use theseprocedures for ail patients.N FGL NED 1986;315:1562Universal Precautions don’tapply to feces, urine, nasalsecretions, sputum, sweat,tears, voinitus, saliva orbreast milk unless theycontain visible blood.MMWR 1988;37(24):377LI LIo oo oII LILI 0LI LIo o11LILILILILILILILILILILILILILI_____LILILILILILIYES NOLILILILILILILIWould you like to receive preliminary results of this study?Would you like to participate further in this study? YES_ NOPage 99Effectiveness of Infection Control StrategiesAU: Directeur du programme de contrôle des infectionsDE: David Birnbaum, MPHRE: Iva1uation des Precautions élémentaires/Isolation des substances organiquesINTRODUCTION:Je suis un étudiant au doctorat, inscrit a un programme interdisciplinaire aPUniversité de Colombie—britannique. Je suis également boursier du Programmenational de recherche et de développement en matière de sante. Mon sujet dethese est l’utilisation et l’efficacité des Precautions élémentaires et del’Isolation des substances orgarLiques. Cette recherche est approuvée parlAssociation des hôpitaux du Canada. Votre participation sera fort appreciée; ii&agit de completer ce questionnaire confidentiel. Cela ne prendra que 15minutes et vous nétes pas oblige de participer aux etapes subséquentes.Cette premiere étape est une étude des pratiques actuelles dans les höpitauxcanadiens; une seconde etape consistera a limplantation dans des hôpitauxsélectionnés; Ia dernière étape comparera lefficacité a réduire les risquesdinfection pour les patients et le personnel. Ce questionnaire vous permettradindiquer si vous désirez recevoir les résultats préliminaires ou le detail desetapes subséquentes. Les résultats nidentifieront pas les individus et lesinstitutions ayant participé, ni par leur nom, ni par leur emplacement. Veuillezsil—vous—plaIt retourner ce questionnaire rempli dans lenveloppe ci—incluse oupar FAX (604—875—4013). N’oubliez pas dajouter votre adresse Si VOUS désirezrecevoir les résultats finals de cette enquête ou Si vous voulez participer auxetapes subséquentes.DESCRIPTION DE L’ITABLISSEMENT:Prdre do rdpondre dans lespace laissd en blanc on encore dencercier [e ou les nundros dbcrivant votre établissenent:1. PRINCIPALE FONCTION: (1) hãpital communautaire de soins aigus(2) hôpital regional de références(3) hôpital denseignement universitaire(4) établissement résidentiel(5) Autre (spécifiez)__________________________2. NOMBRE DE LITS:_______3. NOMBRE DE PATIENTS PAR ANNE:_____4. NOMBRE DE CONSULTATIONS EN CLINIQUES EXTERNES PAR ANNE:______5. NOMBRE DE PRATICIENS AU CONTROLE DES INFECTIONS:a plein temps: a temps partiel:6. EMPLACEMENT: (1) Centre—yule (3) Ville/village(2) Banlieue (4) Region non—incorporée7. OFFREZ—VOUS: (1) Hémodialyse (4) internat(2) Clinique de desintoxication (5) postes de residence(3) Clinique de MTS et SIDA8. A QUAND REMONTE VOTRE DERNIERE ACCRDITATION? DateAgenceStatutPage 100Effectiveness of Infection Control StrategiesPRCAUTIONS ILIMENTAIRES/ISOLATION DES SUBSTANCES ORGANIQUESEst—ce que votre établissement a formellement décidé dadopter ou derejeter les Precautions élémentaires ou l’Isolation des substancesorganiques? Laquelle des deux strategies a été adoptée ou rejetée? Quelsfacteurs ant influence cette decision?Quels avantages et queues difficultés prévoyez—vous dans limplantationde ces strategies?Avez—vous des informations concernant les coGts dimplantation oud’opération de ces strategies comparativement aux programmes dIsolationspécifique par categories ou maladies?Indépendamment des politiques, croyez—vous qu’au moms 50% du personnelaffecté aux salles remballe les aiguilles utilisées?Quel type de contenant est utilisé pour disposer des aiguilles et commentces aiguilles sont—elles transportees de lendroit dutilisation a lendroitoü on en dispose (remballées, placées sur des plateaux, piquées dans desblocs de caoutchouc mousse, portées dans les mains, etc.)?Queues informations supplémentaires voudriez—vous recevoir, au besoin,pour pouvoir recommander ou déconseiller lusage des Precautionsélémentaires?.pour recommander ou déconseiller lusage de lIsolation des substancesorganiques?Page 101Effectiveness of Infection Control StrategiesPOLITIQUES COURANTES:Pour chaque politique. priere de placer on X dans la case ‘OPI cu ‘ION dans chacune des colannes pour indiquer si vous aye: dëjà:RE1J LA PUBLICATION REVtJ LA PUBLICATION ADOPT LA POLITIQUEPOLITIQUE EW SOURCE: OUI NON OUI NON OUI-DATE (rn/a) NONLes Precautions élémentaires nerequiCrent plus ‘avertissernentsur les étiquettes des échantilions.CANADA DISEASES WEEKLY REPORT1987; 1333Layer les mains n’est pas nécessaireaprés avoir enlevé des gants portéspour empécher le contact avec le sang,les sécrétions, les membranes rnuqueu—ses ou les lesions, a moms que lesmains ne soient visiblernent souillées.ANN INTERN MEl) 1987;107:243Chambre privée reconirnandée pourpneumonic a staph.1983 CDC “GUIDELINE FOR ISOLATIONPRECAUTIONS IN HOSPITALS”Chambre privée n’est pas nécessairepour pneumonic a staph.1985 HEALTH & WElFARE CANADAISOLATION AND PRECAUTIONTECHNIQUES” INFECTIONCONTROL GUIDELINESBien emballer tout déchet ou lingepour prévenir les fuites; les casd’infection ne nécessitent pas d’autreétiquetage, ni sacs doubles, nimaniement special.ASEPSIS 1986;8(4):2Les travailleurs de la sante doiventétre renseignés sur les procedures decontràle des infections afin deprévenir la transmission des agentspathogènes par le sang; on doit aussiles aviser d ‘utiliser ces procedurespour tous les patients.N EIGL J MED 1986;315:1562Les Precautions élémentaires nes’appliquent pas aux excrements,urine, sécrétions nasales, larmes,expectorations, transpiration, salive,ou lait de fenuie a moms que ceux-cine contiennent du sang de faconvisible. NMWR 1988;37(24):377Voulez—vous recevoir les résultats préliminaires de cette étude? OUI NONVoulez—vous participer aux étapes subséquentes de cette étude? OUI NONPage 102Effectiveness of Infection Control StrategiesAppendix 3: Nurses’ Questionnaire FormsEnglish VersionFrench VersionExpected Responses, References and Question PairingsPage 103Effectiveness of Infection Control StrategiesTO: NURSING STAFF MEMBERSFROM: David Birnbaum, MPHRE: INFECTION CONTROL PRACTICES SURVEYINTRODUCTION:My work toward a Ph.D. at University of British Columbia, supported by aFellowship from the National Health Research and Development Program ofHealth and Welfare Canada, involves researching the effectiveness ofseveral different approaches to infection control measures. Your facilityhas elected to participate in this national survey; your help in answeringthis confidential questionnaire will be greatly appreciated.Published results will not identify individual facilities nor staff membersby name. You have been selected at random to receive a questionnaire.Please respond to every statement but do NOT write your name on theform. You are free to decline without repercussions if you do not wish toparticipate; to decline, please simply return the blank form.PLEASE CIRCLE A NUMBER ‘it) INDICATE WHErHER YOU (STRONGLY) AGREE, DON’T KfW, OR (STRONGLY) DISAGREEWITH EACH OF THE FOLLOWING STATEMEITrS:STRONGLY DON’T DIS- STRONGLYSTATE1FI AGREE AGREE KNOW AGREE DISAGREE COMMEIS (OPTIONAL)1. Risk of AIDS and hepatitis B are equal:nst needlestick exposures lead to infectionwith either disease. 1 2 3 4 52. In order to implement necessary “isolation”precautions, patients should be screened forAIDS and hepatitis B upon admission. 1 2 3 4 53. This hospital’s policies require a specialwarning label on laboratory specimens frompatients with AIDS or viral hepatitis. 1 2 3 4 54. The primary purpose of “Body SubstanceIsolation” is to protect healthcare workersfrc& AIDS and hepatitis B in the workplace. 1 2 3 4 55. Hepatitis B virus in blood that has driedon environmental surfaces may still be infectious. 1 2 3 4 56. All healthcare workers who receive threeproperly administered doses of hepatitis Bvaccine will gain life—long immunity. 1 2 3 4 57. I have received a complete series (3 doses)of hepatitis B vaccine. 1 2 3 4 58. Body Substance Isolation is a system forinfection precautions that eliminates the needfor isolation categories (Strict, Enteric, etc.)except for one category for airborneccemunicable infections. 1 2 3 4 5Page 104Effectiveness of Infection Control StrategiesSTRONGLY DON’T DIS- STRONGLYSTATEY1ENT AGREE AGREE KNOW AGREE DISAGREE CONNENTS (OPTIONAL)9. Needlestick injury is the single mostsignificant source of occupational infectionwith hepatitis B or. HP! among healthcare workers. 1 2 3 4 510. Used needles should be recapped innediately afteruse with patients to prevent needlestick injury. 1 2 3 4 511. This hospital ‘s policies prohibit recapping ofused disposable needles. 1 2 3 4 512. The risk of needJ.estick injury in manipulatingthe connections within IV lines (eg: mini-bag,heparin lock use, etc.) is higher than the risk ofinjury in giving injections with hypodermic syringes. 1 2 3 4 513. A significant proportion of needlestickinjuries result from missing or penetrating acap while recapping used syringe needles. 1 2 3 4 514. I have received instruction during the pastyear on this hospital’s policies for safe handlingofsharps. 1 2 3 4 515. When suctioning, a single sterile glove isworn on the douinant hand to protect the patientfrom infection. 1 2 3 4 516. Handwashing is I necessary after removinggloves unless the hands are visibly soiled. 1 2 3 4 517. This hospital’s policies require gloving forany activity likely to cause contact with blood,secretions, excretions, mucous membranes, or openlesions. 1 2 3 4 518. The primary purpose of gloving under“Universal Precautions” is to protect patientsfrom bacterial and viral cross—infections. 1 2 3 4 519. Gloves do NOT provide any real protectionsince they leak and there isn’t evidence showingthat their use prevents transmission of infection. 1 2 3 4 520. Gloves may be worn from patient to patient ifgloved hands are washed between patients, 1 2 3 4 521. I have received instruction during the pastyear on this hospital’s policies for proper useand disposal of gloves. 1 . 2 3 4 5Page 105Effectiveness of Infection Control StrategiesSTRONGLY DON’T DIS- STRONGLYAGREE AGREE KNOW AGREE DISAGREE CONMFI’ITS (OPTIONAL)22. Universal Precautions and Body SubstanceIsolation require that vinyl or latex glovesbe iorn for every patient—care activity. 1 2 3 4 523. Most studies have shown that nurses wash theirhands adequately between patient care activities. 1 2 3 4 524. Staff members who are !I’ imune to cbickenpoxmay safely care for patients with chickenpox orherpeszoster. 1 2 3 4 525. I have received in-service instruction frctnthis hospital explaining the application of“Universal Precautions”. 1 2 3 4 526. Universal Precautions or Body Substance Isolationeliminates a need for “bithazard” warning labelson specimens frctn infected patients. 1 2 3 4 527. Everything discarded in any “isolated”patient’ s roon should be incinerated orsterilized as “infectious waste”. 1 2 3 4 528. Universal Precautions prohibits disposal of anypatient wastes in a x*iiminity’s sanitary landfill. 1 2 3 4 529. This hospital’s infection control measuresare practical, effective, and well-documented. 1 2 3 4 530. The primary purpose of Universal Precautionsis to protect patients and staff fron all types ofnosoconial infections. 1 2 3 4 531. I have had one or ire needlestick injuries atthis hospital during the past 30 days. 1 2 3 4 532. This questionnaire was difficult to understand. 1 2 3 4 5Please return, whether completed or not, in the envelope provided.Page 106Effectiveness of Infection Control StrategiesA: Membres du Personnel/InfirmièresDE: David Birnbaum, MPHSUJET: Enquête sur les Strategies pour Contrôle d’JnfectionINTRODUCTION:Mes travaux envers un Ph.D. a l’université de la Colombie Brittanique,appuyés par un bourse du Programme National de Recherche et deDèveloppement en Matière de Sante, Sante et Bien—être Social Canada,impliquent une recherche sur l’efficacité d’un certain nombre de strategiesdifférents utilisées pour le contrôle d’infection. Votre section de servicea choisi de participer a ce questionnaire national; nous apprécionsbeaucoup votre collaboration en répondant a ce questionnaire confidentiel.La publication des résultats se gardera d’identifier, soi les sections deservice particulières ôu les membres du personnel. Votre nom a étéselectionné a la pige pour recevoir ce questionnaire. Veulllez s.v.p.répondre a chacune des questions, mais prier de ne pas écrire votre nomsur le formulaire. Le choix de répondre a ce questionnaire est facultatifet sans repercussion, quel que soi ce choix. Si vous optez de ne pas yrépondre, veuillez renvoyer le formulaire comme tel.V1JILU2 2F21 LE NLO DThIThN ‘,1YIRE CHODC DE PRICE: (1) BSIAENF D’ACCORD, (2) D’ACCORD,(3) NE SAIS PAS, (4) PAS D’ACCC*D aJ (5) DEFINIvFNP PAS D’ACRD PC*JR CHACUN DL 2’ccs SUIVANTS.NE DFD1VENABSOLUNN SAIS PASD’AccORD PAS D’AcD iNfAIRFS1. Les risques d’ enquérir le CIDA i 1 ‘hépitateB sont équitables: la plupart des ouverturesd’aiguilles sont précurseurs de l’infection desoi l’une ou l’autre des maladies. 1 2 3 4 52. Afin d’implanter les precautions d’isolationnécéssaires, les patients devraient subir unexamen scrutaire pour determiner 1 ‘existencede CIDA et d’hépatite B avant admission. 1 2 3 4 53. Les politiques cle cet hopital exigent uneetiquette d’avertissement identifiant lesspéciiuèns de laboratoire provenant depatients avec le CIDA ou l’hépatite viral. 1 2 3 4 54. Le but primrdial d’ Isolation de SubstanceOrganicjues est de protéger ceux et celles auservice de la santée, de l’infection du CIDA etd’hépatite B dans le milieu du travail. 1 2 3 4 55. Le virus d’hépatite B pout être encoreinfectueux même s’ ii se trouve dans le sangasséché sur une surface environnementale. 1 2 3 4 5Page 107Effectiveness of Infection Control StrategiesNE DFINITIVEMENABSOUJME’1T SAIS PASEK)NCS D’ACCORD PAS D’ACCORD CX)MMENI7JRES6. Toute personne au service de la sante qulrecoit 3 doses de vaccin Hépatite B proprementadministrées est iniiunisée pour la vie. 1 2 3 4 57. J’ai reçu l’iinunization conplète du vaccinHépatite B (3 doses). 1 2 3 4 58. L’Isolation de Substances Organiques est unsystème pour les precautions envers 1 ‘infection,qui élimine la nécéssité de categories d’isolation(eg. STRIcT, ENTERIC, etc.) sauf une catégoried’infection transmissible aéroportée. 1 2 3 4 59. Dans ce métier, le foyer d’infection avecl’hépatite B ou le liv qui est le plus significatifentre ceux ou celles au service de la sante est labléssure d’aiguille. 1 2 3 4 510. Les aiguilles usager doivent étre recapsuléesinédiatement suivant leurs usage sur le patientaf in de prévenir toute blêssure d’aiguille. 1 2 3 4 511. Les politiques de cet hâpital interdisent lerecapsulement des aiguilles—à—jeter usagées. 1 2 3 4 512. Le risque de blessure d’aiguilles, subit durantla manipulation des connections de ligne IVcontrairement au même risque provenant de picpres,est beaucoup plus élevé. 1 2 3 4 513. Une proportion considerable de blessured’aiguilles sont le résultat de la pénétrationde la capsule ou de manquer la capsule avecl’aiguille durant le recapsulage. 1 2 3 4 514. Durant cette année passée j ‘ai reçu del’instruction des politiques de cet hôpitalenvers le maniement des objets a pointe aigüe. 1 2 3 4 515. Pour protéger le patient contre l’infection,ii est nécéssaire de porter qu’ un gant sterile surla main principale durant 1 ‘aspiration. 1 2 3 4 516. Aprés avoir enlevé les gants, ce n’est pasnécéssaire de se layer les mains a nin que lesmains soient visiblement souillées. 1 2 3 4 5Page 108Effectiveness of Infection Control StrategiesNE DFINITIVEMENTABSOLUMENT SAIS PASEN0NS D’ACcORD PAS D’ACCORD CO!4MENTMRES17. Les politiques de cet hâpital reguient que,pour toutes activitées ou le contacte avec le sang,les sécrétioris, les excretions, les membranesmuqueuses, ou les blCssures ouvertes est possible,les gants doivent ètre portés. 1 2 3 4 518. Le but primordial des “Precautions lémentaires”gui éxige qu’on portent les gants, est de protégerles patients contre les infections—croisées viralet bactériens. 1 2 3 4 519. Les gants, puisque us coulent, n’offrent pasde vrai protection et ii n’a pas d’évidence que l’usagedes gants prévient la transmission des infections. 1 2 3 4 520. On peut porter les gants de patient a patientsi on se laves entre chaque patient. 1 2 3 4 521. Durant cette année passée jt ai reçu del’instruction des politiques de cet hópital enversl’usage et la disposition de gants. 1 2 3 4 522. “L’ Isolation de Substances Organiques” et les“Precautions lémentaires” requient que je porte degants de vinyle ou de latex pour chaque activitéeavec mes patients. 1 2 3 4 523. Les infirmières se lavent les mains suffisamentbien entre chaque patient, ce qui est indiqué parla plupart des etudes. 1 2 3 4 524. Tout personnel de l’hâpital gui ne sont pasuTaTunisé contre la varicelle peuvent quand mêmesoigner, sans risque, ceux gui ont la varicelleou les herpes zoster. 1 2 3 4 525. J’ai deja reçu de cet hôpital des instructionsexpliquant 1 ‘application des “Precautionslémentaires”. 1 2 3 4 526. Les “Precautions Elénientaires” ou “l’Isolationdes Substances Organigues” élmniinent le besoind’ etiquette d ‘ avertissement “bio—hazard”, pour lesspécimèns des patients infectés. 1 2 3 4 527. Tout objets abandonnés dans la chambre “isolée”d’un patient, devraient êtres incinérés ou stérilizésccanme “déchéts infectueux”. 1 2 3 4 5Page 109Effectiveness of Infection Control StrategiesNE DFINITIVEM\TTABSOUJMENT SAIS PASEN0NCS D’ACCORD PAS D1ACCORD CONMENTAIRES28. Les “Precautions 1érnentaires’ empêchent quetoutes ordures ou déchéts des patients soient enlevésau terrain dépotoir sanitaire de la coiunauté. 1 2 3 4 529. Les mesures utilisés pour le controled’infections de cet hopital sont éfficace,pratique et bien documentées. 1 2 3 4 530. Le but prordia1 de les “PrecautionsE].émentaires” est de protéger les patients etle personel de ],‘hopital contre tout les typesd’infections nosocomiales. 1 2 3 4 531. Durant les derniers trente jours a cet hôpital,j’ai subis une blessure d’aiguille. 1 2 3 4 532. J’ai eu de la difficultée a comprendre cequestionnaire. 1 2 3 4 5Veuillez renvoyer ce formulaire dans Fenveloppe fourni, rempli ou non.Page 110Effectiveness of Infection Control StrategiesQUESTION: EXPECTED REPLY REFERENCE PAIR1 Risk of AIDS and hepatitis 3 are equal: most needlestick exposures lead to DISAGREEinfection vith either disease.2 In order to implement necessary “isolation” precautions, patients should be DiSAGREE 2 26screened for AIDS and hepatitis S upon admission.3 This hospital’s policies require a special yarning label on laboratory (site specific) —specimens from patients vith AIDS or viral hepatitis.4 The primary purpose of “Body Substance Isolation” is to protect healthcare DISAGREE 3 30yorkers from AIDS and hepatitis B in the vorkpiace.5 Bepatitis B virus in blood that has dried on environmental surfaces may AGREEstill be infectious.6 All healthcare yorkers vho receive 3 properly administered doses of DISAGREE 5,6 24hepatitis B vaccine vill gain life—long immunity.71 have received a complete series (3 dosesi of hepatitis B vaccine. (site specific) — 0S Body Substance Isolation is a system for infection precautions that eliminates AGREE 3 18the need for isolation categories (Strict, Enteric, etc.) except for onecategory for airborne communicable infections.9 Needlestick injury is the single most significant source of occupational AGREE 78 12infection vith hepatitis B or BIY among healthcare yorkers.10 Used needles should be recapped immediately after use vith patients to prevent DISAGREE 7,8,13,16, 13needlestick injury.11 This hospital’s policies prohibit recapping of used disposable needles. (site specific) —12 The risk of needlestick injury in manipulating the connections vithin IV lines AGREE 9 9(eg: mini—bag, beparin lock use, etc.) is higher than the risk of injury ingiving injections vith hypodermic syringes.13 A significant proportion of neediestick injuries result from missing or AGREE 9 10penetrating a cap nile recapping used syringe needles.141 have received instruction during the past year on this hospital’s policies (site specific) — 0for safe handling of sharps.15 ben suctianing, a single sterile glove is yarn on the dominant band to protect DISAGREE 10 19the patient from infection.16 Bandvashing is NOT necessary after removing gloves unless the hands are DISAGREE 11 23visibly soiled.17 This hospital’s policies require gloving for any activity likely to cause (site specific) 0contact vith blood, secretions, excretions mucous membranes, or open lesions.18 The primary purpose of gloving under ‘Universal Precautions” is to protect DISAGREE 12 8patients from bacterial and viral cross—infections.19 Gloves do NOT provide any real protection since they leak and there isn’t DISAGREE 12,13 15evidence shoving that their use prevents transmission of infection.20 Gloves may be vorn from patient to patient if vashed betveen patients. DISAGREE 3,11,12,13 22211 have received instruction during the past year on this hospital’s policies (site specific) — 0far proper use and disposal of gloves.22 Universal Precautions and Body Substance Isolation require that vinyl or latex RISAGREE 12,13 20gloves be yarn for every patient—care activity.23 Most studies have shovn that nurses vash their hands adequately betveen patient DISAGREE 14,15 16care activities.24 Staff members no are NOT immune to chickenpox may safely care for patients DISAGREE 16 6vith chickenpox or herpes zoster.251 have received in—service instruction tram this hospital explaining the (site specific) — 0application of “Universal Precautions”.26 Universal Precautions or Body Substance Isolation eliminates the need for AGREE 12 2“biohazard” yarning labels on specimens from infected patients.27 Everything discarded in any “isolated” patient’s room should be incinerated or DISAGREE 17 20sterilized as “infectious vaste”.28 Universal Precautions prohibits disposal of any patient vastes in a community’s DISAGREE 17 27sanitary landfill.29 This hospital’s infection control measures are practical, effective, and yell— AGREE - 0documented.30 The primary purpose of Universal Precautions is to protect patients and staff DISAGREE 1,13,12 4from all types of nosocomial infections.311 have had one or more needlestick injuries at this hospital during the past 30 days. —— 032 This computer program vas difficult to use. —- 0Page 111Effectiveness of Infection Control StrategiesReferences for Nurses’ Questionnaire1. Hughes JM: Universal Precautions, CDC Perspective. OCCUPATIONAL MEDICINE STATEOF THE ART REVIEWS 1989;4:13-27.2. Gerberding JL, UC San Francisco Task Force. Recommended Infection—ControlPolicies for Patients with Human Immunodeficiency Syndrome, an Update. N ENGL JMED 1986;315(24):1562—43. Lynch P, Jackson MM, Cummings MJ, et al. Rethinking the Role of IsolationPractices in Prevention of Nosocomial Infections. ANN INTERN MED 1987;107;243—6.4. Bond WW, Favero MS, Petersen NJ, et al. Inactivation of Hepatitis B Virus byIntermediate — to —High Level Disinfectant Chemicals. J CLIN MICROBIOL1983; 18(3) :535—8.5. Hadler SC, Francis DP, Maynard JE, et al. Long—Term Immunogenicity and Efficacyof Hepatitis B Vaccine in Homosexual Men. N ENGL J MED 1986;315(4):209—14,6. Barnas GP, Hanacik U. Hepatitis B Vaccine, Persistence of Antibody FollowingImmunization. INFECT CONTROL HOSP EPIDEMIOL 1 988;9(4): 147—50.7. Elmslie K, Mulligan L, O’Shaughnessy MV. Occupational Exposure to the HumanImmunodeficiency Virus Among Health—Care Workers in Canada. CDWR1 988;14(43): 197—200.8. Marcus R, CDC Cooperative Needlestick Surveillance Group. Surveillance of HealthCare Workers Exposed to Blood from Patients Infected with the HumanImmunodeficiency Virus. N ENGL J MED 1988;319:1118—23.9. Jagger J, Hunt EH, Brand—Elnaggar J, et al. Rates of Needle—Stick Injury Causedby Various Devices in a University Hospital. N ENGL J MED 1989;319(5):284—8.10. INFECTION CONTROL GUIDELINES, PREVENTION OF NOSOCOMIAL PNEUMONIA. HEALTHAND WELFARE CANADA, 1988, p. 41.11. Doebbeling BN, Pfaller MA, Houston AK, et al. Removal of Nosocomial Pathogensfrom the Contaminated Glove, Implications for Glove Reuse and Handwashing. ANNINTERN MED 1988;109(5):394—8.12. LCDC. Universal Precautions, Report of a Consensus Committee Meeting. CDWR1989;15(5):23—28.13. LCDC. Update— Universal Precautions for Prevention of Transmission of HumanImmunodeficiency Virus, Hepatitis B Virus, and Other Bloodborne Pathogens inHealth—Care Settings. CDWR 1988;14(27):1 17—24.14. Fox, MK, Langner SB, Wells RW. How Good Are Handwashing Practices? AM J NURS1974;74(9):1676—8.15. Albert RK, Condie F. Hand—Washing Patterns in Medical Intensive—Care Units.N ENGL J MED 1981;304(24):1465—66.16. INFECTION CONTROL GUIDELINES, OCCUPATIONAL HEALTH IN HEALTH CAREFACILITIES. HEALTH AND WELFARE CANADA, 1986. p.4317. Canadian Standards Association. HANDLING OF WASTE MATERIALS WITHIN HEALTHCARE FACILITIES. CAN/CSA—Z3 17.10—88Page 112Effectiveness of Infection Control StrategiesAppendix 4:Efficacy of MIDI for Epidemiologic Typing of Staphylococcus epidermidisDavid Birnbaum, MPHWith the assistance of:Leona Ayers, MDJohn Boyce, MDLoreen Herwaldt, MDDon Low, MDMichael Noble, MDMichael Pfaller, MDRobert Sherertz, MDAnthony W. Chow, MDPage 113Effectiveness of Infection Control StrategiesABSTRACTLack of an adequate typing system has hampered our understanding of theepidemiology of infections caused by coagulase—negative staphylococci (CNS).CNS have become recognized as important nosocomial pathogens, and theprincipal cause of infections associated with invasive devices. Sensitive,specific and convenient methods are needed to evaluate whether implementationof guidelines for infection control reduce the risk of nosocomial infections fromCNS and other pathogens. Existing typing methods are either not convenient foruse in most hospital laboratories or lack adequate specificity to distinguishbetween strains. The Microbial Identification System (MIDI, Microbial ID Inc.,Newark, Delaware), a semi—automated system for fatty acid methyl ester (FAME)analysis, shows considerable promise for clinical and epidemiologic applications.Its predictive accuracy and reliability were tested using epidemiologically—related and replicated CNS isolates as well as CNS from epidemiologicallyunrelated clinical infections. These isolates were obtained from five establishedhospital culture collections in diverse geographic locations. Two—hundred isolateswere fully characterized in five days by one person using MIDI, and results wereconsistent with those produced by more expensive and time—consumingconventional typing methods. MIDI, an attractive alternative to molecularmicrobiology methods, may offer important advantages to hospital epidemiologistsand infection control programs.Page 114Effectiveness of Infection Control StrategiesINTRODUCTIONUniversal Precautions (UP) is an expensive strategy for preventingoccupational infection of healthcare workers with bloodborne pathogens. Thenation—wide cost of UP per human immunodeficiency virus infection prevented isestimated at between one—million and ten—million dollars.1 2 It has beensuggested that UP or another strategy, Body Substance Isolation (BSI), might alsoreduce patients’ risk of nosocomial infections, thus improving the cost—effectiveness of these strategies. However, the few studies of compliance oreffectiveness under UP or BSI that are available present contradictory findings.Therefore, although UP and BSI advocate gloving for patient—care activities, itsvalue as an infection control measure is not precisely known. Methods to tracethe sources of specific infections are needed to further our understanding.Staphylococcus epidermidis, for example, is clearly the most commonpathogen in intravascular (IV) device—associated sepsis, but the relativeimportance of different routes of infection is uncertain.3 If entry—site skinflora migrating along cannulae cause most IV CNS infections, then infectioncontrol measures must focus on antiseptics, time limits for site use andminimizing trauma of the insertion site. If contamination of IV hubs byexogenous flora is more frequent, then a focus on asepsi, including use ofgloves under either UP or BSI, may reduce the risk of common source or cross—infection. Evidence supporting the former view may be biased by use ofmicrobiologic rather than clinical definition of infection in some studies.Further, it is not difficult to imagine that variation in hospital practices wouldlead to variation in the extent of hub contamination at different times orplaces. A convenient typing system would allow individual hospitals to monitorthe origin of such infections, periodically adjusting the emphasis of their ownPage 115Effectiveness of Infection Control Strategiesinfection control program accordingly.The basis for an epidemiologically useful typing system is to exploitdocumented species diversity, but this is often comparable to hitting a movingtarget with tools of uncertain precision, accuracy or reliability. Existing typingsystems are not adequate for tracing the origin of nosocomial ONS infections.4Fatty acid methyl ester (FAME) analysis as a qualitative pattern—recognitionmethod is well—established for species—level identification (eg: anaerobes).Eerola and Lehtonen recently demonstrated its reliability in distinguishingbetween species of different aerobic genera as well as between strains withinspecies, but noted that different univariate correlation methods used for dataanalysis are not equiva1ent. Subsequent work suggested the future ability todiscriminate between epidemiologically related and unrelated strains.6 MIDI(Microbial ID Inc., Newark, Delaware) automates quantitative analysis of over200 fatty acid metabolic products and comparison with proprietary databaselibraries by multivariate statistical methods. New library entries can be createdby individuals when isolates with an unrecognized pattern are encountered, andsubsequent samples are automatically compared to this updated library.MIDI establishes species—level identifications through principal componentanalysis of ratios of FAME peak areas. This multivariate, quantitative approachpromises superior ability over simple pattern recognition to differentiate betweenorganisms. In addition to providing genus—species identifications, MIDI softwarecan also display the degree of relatedness of isolates on two—dimensionalprincipal component plots and on cluster analysis dendrograms. However, MIDIhas not been applied to epidemiologic typing of CNS. Since it may offeradvantages to both clinical laboratories and hospital epidemiologists, itsdiscriminatory power, predictive accuracy, intra— and inter—laboratory reliabilityPage 116Effectiveness of Infection Control Strategiesfor subspecies typing of CNS need to be documented.MATERIALS & METHODS:DESCRIPTION OF ORGANISMSOne—hundred and eighty—three well—characterized coagulase—negativestaphylococci (eNS) from clinical specimens were obtained from culturecollections in five geographically distinct areas. Their identification andantimicrobial susceptibility profile were determined in the contributinglaboratories using conventional methods approved by the National Committee forClinical Laboratory Standards (NCCLS). Identifications were confirmedsubsequently by one of us (MP) using the scheme of Kloos and Schielfer. Someof these collections have been described elsewhere.7 8Among the 183 are six groups containing epidemiologically related isolates:ten S. haemo]yticus blood culture isolates from neonates in Ontario; three pairsof matching blood and intravascular cannula tip S. epidermidis isolates fromBritish Columbia; five S. epidermidis isolates from an outbreak of surgical woundinfections attributed to a surgeon’s hand flora in Rhode Island and twelveconsecutive S. epidermidis isolates from one immunosupressed patient in Iowa.Their epidemiologic relatedness was assessed on the basis of the followingcriteria, in descending order of importance (see Table 23):1. Epidemiologic evidence of time—space clustering2. Matching biotype/susceptibility profilea) No differences = no evidence that isolates are differentb) 1 difference = isolates might be differentc) 2 differences = isolates are different3. Molecular methods, if done, show no differencesa) Plasmid Profileb) Restriction endonuclease analysis (REA)c) Hybridization (probe or DNA [RFLPI)Other isolates obtained from each collection differ from each other byconventional microbiological typing and/or lack of epidemiologic suspicion ofPage 117Effectiveness of Infection Control Strategiescommon—source or cross—infection relatedness among them.The remaining 155 isolates were judged to be epidemiologically unrelated bymethods that meet or exceed current standards of infection surveillance practice.Forty—eight unrelated blood culture isolates collected over a 6 month period inone Ontario children’s hospital exhibited unique combinations of biotype,susceptibility profile and plasmid profile, and no evidence of time—spaceclustering. Fifty—seven unrelated blood culture isolates from a universityhospital in Iowa were unique in their combined biotype, susceptiblity profile,slime production, synergistic hemolysis, plasmid profile and restriction digestpatterns. Forty—two unrelated blood culture isolates in a British Columbiauniversity hospital were selected from a collection spanning 3 years; they wereconsidered unrelated episodes on the basis of no time—space clustering. Threeisolates from intravascular lines in a North Carolina university hospital werealso judged to be unrelated, in spite of comparable biotype and susceptibiltyprofiles, on the basis of epidemiologic assessment.The majority of these isolates, primarily Staphylococcus epidermidis, meetCDC criteria for clinical significance9. Twenty—six of the 183 were bloodculture contaminants (included to study how closely they would be related toclinically—significant isolates). One isolate was replicated eighteen times (group#7) to test intra—laboratory reliability. The entire set of 200 randomly—numbered clones was analyzed blindly and independently in two facilities withestablished MIDI systems (under supervision of Microbial ID Inc. staff in theirlaboratory, and in a university hospital clinical laboratory) to test inter—laboratory reliability. Media inoculation was done in Class II laminar airflowhoods, and colony morphology was inspected on subcultures (to check forcontaminants) immediately after receipt of isolates from contributing laboratoriesPage 118Effectiveness of Infection Control Strategiesand again just prior to MIDI analyses.MIDI TECHNIQUEAnalysis was conducted in accordance with MIDI instructions.10 Briefly,isolates were quadrant—streaked onto trypticase soy agar (TSA) plates (BBL) andharvested in log—phase growth after 24 (±2) hours’ incubation at 28CC in air.Approximately one 4mm loop—full of bacteria was scraped for each sample. Eachsample was then saponified, methylated, extracted and cleaned before removingits organic layer for injection into MIDI’s gas chromatography system.Calibration standards and a negative control blank were run with each batch,and one positive control specimen (Xanthomonas maltophilia, ATCC 13637) wasrun each day. MIDI’s aerobic bacteria library (Ver. 3.30) was used forinterpretation, Samples with suboptimally low gas chromatography results (totalpeak area <80,000 or <85% of peak area used by the naming algorithm) wererepeated after concentration by evaporation.STATISTICAL ANALYSISMIDI computes an index for each isolate through principal componentanalysis of ratios of cellular fatty acid content. Cluster analysis of theseindices was applied’1 with Euclidian distances calculated by the simple andWard’s’2 linkage methods. Empirically, the replicated (group #7) isolates’joining distances were examined to determine an optimal strain—level joiningdistance. The MIDI system software and SYSTAT (version 4.1, Systat Inc.,Evanston IL) were used for statistical analysis. Discriminatory power wascalculated as Simpson’s Index, the probablity that two isolates selected atrandom would be placed into different typing categories.13 Sensitivity,specificity, positive and negative predictive accuracy were evaluated bycomparing MIDI’s predictions with the known related or unrelated nature ofPage 119Effectiveness of Infection Control Strategiesisolates.RESULTSAll 200 isolates were typable and their analysis was completed in 5 daysby one of us (DB). Nearly two—thirds of an inexperienced operator’s first batchof extracts had low peak areas and required concentration, but subsequentbatches rarely suffered this problem.A critical distance to distinguish between related and unrelated isolateswas established emperically by examining the Euclidian distance at which knownepidemiologically—related isolates joined. Group #7 was taken as the ‘goldstandard” for clonal origin. These replicated isolates all joined at 5.4 units,but exhibited independent clustering at 4.5 units on a dendrogram generated byWard’s method. Thus, any isolates joining at less than 4.5—5.4 units could beconsidered clustered into the same typing category, whereas isolates joining atmore than 5.4 units would be assigned into different categories. This isillustrated in Figure 12.Single and Ward’s linkage methods yielded similar groupings. At least 46typing categories were readily apparent, with a Simpson’s Index of 0.93 (Table24). Approximately half of the categories contained a single isolate.There are seven groups of related isolates in the test set. Group sevenwas created artificially by replicating one isolate eighteen times. MIDI foundthree clusters joined at 0.7—2.1 units, and linkage of these clusters at 4.5—5.4units. The former range is consistent with previous experience for repeatedanalysis of the same isolate (personal communication, Myron Sasser), and thepossibility of a mixed culture was noted.Group one consisted of 10 blood culture isolates from one neonatal nursery,identified by the contributing laboratory as S. haemolyticus with identicalPage 120Effectiveness of Infection Control Strategiessusceptibility profile, restriction endonuclease pattern and hybridization to threedrug—resistance probes. This resistant strain was endemic in several adjacenthospitals, the affected infants had overlapping hospitalization dates during a 5month period, but there was no conclusive evidence to establish or rule outcommon source or cross—infection. MIDI identified two isolates (obtained withinthree days of each other) as related to each other and to a third (joined at 1.9Euclidian distance units) and fourth (at 5.4 units).Group two consisted of matching blood culture and intravascular cannulaisolates from a single patient. MIDI placed these two (and otherepidemiologically—unrelated isolates) in a single category. This patient hadthree CNS isolated from an intravascular line; only one matched the bloodisolate’s susceptibility profile. MIDI correspondingly typed the isolates withmatching profiles as related and the other two as unrelated. The correctlypaired isolates joined at 2.6 Euclidian distance units, and subsequently wereidentified as S. epidermidis whereas the other two cannula isolates wereconfirmed as S. intermedius and S. auricularis. On the basis of susceptibilityprofile differences, the two related isolates could be distinguished from most ofthe epidemiologically—unrelated isolates in this MIDI typing category.Group three similarly consisted of paired blood and cannula isolates fromanother patient. MIDI placed these two and other epidemiologically—unrelatedisolates in a single category. The correctly paired isolates joined at 1.2Euclidian distance units. On the basis of susceptibility profile differences, foursubcategories correctly separated the two related isolates from three mutuallyunrelated isolates.Group four also consists of paired blood and cannula isolates from onepatient. These were correctly matched, again with unrelated isolates as well, atPage 121Effectiveness of Infection Control Strategiesa distance of 3.8 units, and susceptibility profile differences createdsubcategories correctly separating related from unrelated isolates.Group five consisted of S. epidermidis from a surgeon’s hands, twoinfected surgical wounds, and two blood cultures. Temporal clustering, plasmidanalysis, EcoRI restriction endonuclease digests, antibiograms and biotypingdocument this as a common—source outbreak. MIDI typed the four patient—isolates as related (3 joined at 1.4 units and the 4’th at 2.8) but the surgeon—isolate was in a different category (joined at 10.6 units). Susceptibility profiledifferences created subcategories correctly distinguishing the outbreak patientisolates from unrelated isolates. This is illustrated in Figure 12: isolatesjoining to the right of the distance marked as did not cluster into the typingcategory containing the outbreak patient isolates.Group six, a series of twelve S. epidermidis isolates from oneimmunosupressed patient, was identified by MIDI as two clusters and twounrelated isolates. One cluster consisted of two isolates from blood culturesdrawn on the same day (joined at 2.1 distance units). The second containedeight isolates obtained over several months (joined at 4.5 units). Since thesusceptibility profiles differ, at least two distinct clusters are correctlyidentified by MIDI for group six.DISCUSSIONA “gold standard” for comparison was established by available epidemiologicand microbiologic information with clinical isolates. It is reasonable to assumethat such information is valid for ruling out relatedness. However, it cannot beconsidered conclusive proof of relatedness when it failed to find differences:testing with another substrate or enzyme, or a more powerful method mightreveal differences. MIDI’s grouping together of isolates shown by other methodsPage 122Effectiveness of Infection Control Strategiesto be epidemiologically unrelated may be an error resulting from detectingtaxonomically similar (but epidemiologically unrelated) strains. MIDI’s separationof isolates grouped together by other methods may be an error or may indicatethat MIDI is a more powerful typing method. Although MIDI’s predictions closelyparalleled the “gold standard” results, MIDI’s failure to cluster all of Group 5’sisolates into one category is of concern. The surgeon’s hand isolate had beenfound identical to the patient isolates by whole plasmid gels, plasmid REA andpulsed field electrophoresis analysis of chromosomal DNA. These results arepromising, but further work is required to document MIDI’s accuracyMIDI conveniently accomodates large numbers of isolates, can supply atyping result for every isolate, and its material costs for complete identificationare low (about $1.30 per isolate). One—tube sample preparation (saponification[sodium hydroxide in methanoll, methylation [hydrochloric acid in methanol],extraction [hexane in methyl tert—butyl ether], sample cleanup [sodiumhydroxide]) and automated analysis accomodate identification of up to 45samples per technologist—day per machine. One person can process as many as30—50 isolates in about 4 hours, depending upon experience. Performance of gaschromatography requires an additional thirty minutes per sample. MIDI’s wastecan be recycled or biodegraded. Compared to other typing systems, a relativelysmall amount of plastic is discarded. Glass, water and sodium chloride are themajor waste products.Growth of CNS on TSA plates at 24 hours was often relatively light.Better growth might be achieved on blood agar at 37CC, but MIDI’s library forthis condition is not as well established as their aerobic 28CC TSA library.Improved reproducibility with blood agar at 37 has been reported for otheraerobic bacteria.14 This problem was easily overcome by using a heavierPage 123Effectiveness of Infection Control Strategiesinoculum. Streaking two rather than one TSA plate is another possible strategywhen poor growth is anticipated, and offers the additional advantage of allowingan immediate repeat extraction in case a critical sample is ruined by an errorin technique. Concentrating extracts for a second chromatography run, thestandard procedure when MDI’s printout warns of unacceptably low peak areas,was also simple and effective: 84% of flagged” results attained acceptable peakareas after concentration. Difference in joining distance between “flagged’results versus concentrate results (due to new peaks emerging above detectionlimits) changed the typing category assigned in 16% of these samples (Fig. 13).FAME analysis by gas—liquid chromatography may be particularly cost—effective for epidemiological typing of clinical or environmental isolates ascompared to current alternatives. Successful application to the subspecies levelhas been reported with staphylococci.5’15 However, discerning epidemiologicrelationships is more difficult than simply establishing taxonomic relationships.MIDI was used recently in a small study of Pseudomonas cepacia in cysticfibrosis centers; a low but promising discriminatory ability was determined among42 isolates from 5 centers (Simpson’s index, the probability of two randomly—selected isolates being assigned to different typing categories, was O.775).16 Thepresent study of CNS involves a larger number of isolates selected specificallyto minimize the chance that they were related by unrecognized common source orcross—infection. In comparison to molecular typing methods now available tocharacterize isolates, MIDI promises to be a relatively powerful, rapid andinexpensive tool.ACKOWLEDGEMENTSTechnical advice: Drs. Richard Mathias, Michael Kelly, Myron Sasser and MichaelSchulzer.Logistical assistance: Robin Barteluk, Linda Boyken, Christine Corten, RobertaDickenson, Shelley Scriver and Marcie Sponholtz.Page 124Effectiveness of Infection Control StrategiesTable 23:Characterization of Epidemiologically-Related Isolates by Lab of OriginGROUP NUMBERCRITERION 1 2 3 4 5 6Epidemiology ShowsTime—Space Clustering ± + + + + ±Biotype/SusceptibilityProfiles Match + + + + + ±Plasmid Profile + - — — + +REA + - - - + -DNA Probe/RFLP probe - - - RFLIP —Note: + = clearly documented by contributing laboratory,± = incompletely documented,-= not done.Page 125Effectiveness of Infection Control StrategiesTable 24:Summary of MIDI’s Typing PerformanceCLUSTER ANALYSIS HETHOD Ward’s Method Single LinkageTHRESHOLD LIHIT Euclidean Distance 45 54* 4.6Number of Categories Created 66 46 57Number containing only 1 isolate 34 (52%) 21 (46%) 29 (51%)Simpson’s Index zO.96 zO.93 zO.96Sensitivity + + +Specificity + + +Positive Predictive Accuracy + + +Negative Predictive Accuracy + + +Inter-Laboratory Agreement + + +* Group #7’s isolates all joined at 5.4 units, but exhibitedindependent clustering at 4.5 units. These isolates were allsubcultured from isolate #1, which appeared mixed whensubcultured for MIDI analysis. Further work will be requiredto determine whether these eighteen reference isolates werepure or mixed. It is unlikely that the threshold limit couldbe set any lower than 3.8, the joining distance for group #4.+ Qualitative results described in text; quantitative results notyet available.+ Results not yet available.Page 126Effectiveness of Infection Control StrategiesFigure 12:Section of MIDIIsolate GroupID#167 0100 0100 0101 073 091 0109 5108 5107 5129 0169 0110 5120 0159 0166 0128 074 057 048 0111 5Dendrogram Containing Group #5 IsolatesEuclidian Distance0.00 2.35 4.70 7.05 9.40 11.75+-—--+——-—+—---+---.+--1-+----+---—+—-—-+---—+----+---—+0.00 2.35 4.70 7.05 9.40 11.75KEY: indicates 4.5—5.4 unit joining distance limitd = epidemiologically—unrelated isolates5 = epidemiologically—related isolatesPage 127Effectiveness of Infection Control StrategiesFigure 13:Joining Distances of Repeated AssaysSAMPLETYPES C * *10 15 20JOINING DISTANCELegend: C = concentrated vs. unconcentrated sampleN = morphotype “a versus “b” for isolatesexhibiting variants in colony size, textureor colorNote: Two morphotypes of isolate #1 join at distance 9.4Page 128Effectiveness of Infection Control StrategiesReferences for Appendix 4:1. Wenzel RP. Interaction of Man and Microbe, Implications of the AIDSEpidemic for Hospital Epidemiology. AM J INFECT CONTROL.1988; 1 6(5):214—20.2. Stock SR, Gafni A, Bloch RF. Universal Precautions to Prevent HIVTransmission to Health Care Workers; An Economic Analysis. CAN MEDASSOC J 1990;142(9):937—463. Sitges—Serra A, Liñares J; Brun—Buisson C; Collignon P, Munro R.Limitations of Semiquantitative Method for Catheter Culture (letters).J CLIN MICROBIOL 1988;26(5):1074—6.4. Birnbaum D, Kelly M, Chow AW. Epidemiologic Typing Systems forCoagulase—Negative Staphylococci. INFECT CONTROL I-IOSP EPIDEMIOL1991; 12(5):3l 9—26.5. Eerola E, Lehtonen 0. Optimal Data Processing for Automatic BacterialIdentification by Gas—Liquid Chromatography of Cellular Fatty Acids.J CLIN MICROBIOL 1988;26(9):1745—53.6. Kotilainen F, Huovinen F, Eerola E. Application of Gas—LiquidChromatographic Analysis of Cellular Fatty Acids for SpeciesIdentification and Typing of Coagulase—Negative Staphylococci. J CLINMICROBIOL l991;29(2);315—322.7. Sherertz RJ, Raad II, Belani A, Koo L, Rand KH, et al. Three—YearExperience with Sonicated Vascular Cathester Cultures in a ClinicalMicrobiology Laboratory. J CLIN MICROBIOL 1 990;28( 1):76—82.8. Boyce JM, Potter—Bynoe G, Opal S, Dziobek L, Medeiros AA. A Common—Source Outbreak of Staphylococcus epidermidis Infections amongPatients Undergoing Cardiac Surgery. J INFECT DIS 1990;161;493—99.9. Garner JS, Jarvis WR, Emori TG, et al. CDC Definitions for NosocomialInfections, 1988. AM J INFECT CONTROL l988;16(3);l28—40.10. Sasser M. Technical Note #101— Identification of Bacteria by GasChromatography of Cellular Fatty Acids. MIDI 1990.11. Johnson RA, Wichern DW. APPLIED MULTIVARIATE STATISTICALANALYSIS, 2nd Edition. Prentice Hall Inc., New Jersey. 1988. p543—72.12. Ward JH Jr. Hierarchical Grouping to Optimize an Objective Function.AMERICAN STATISTICAL ASSOC J 1 963;58;236—44.13. Hunter PR, Gaston MA. Numerical Index of the Discriminatory Ability ofTyping Systems; An Application of Simpsons Index of Diversity.J CLIN MICROBIOL 1988:26(1 1);2465—66.Page 129Effectiveness of Infection Control Strategies14. Osterhout GJ, Shull VH, Dick JD. Identification of Clinical Isolates of Gram—Negative Nonfermentative Bacteria by an Automated Cellular Fatty AcidIdentification System. J CLIN MICROBIOL 1991 ;29(9): 1822—30.15. OtDonnell AG, Nahaie MR, Goodfellow DE, et al. Numerical Analysisof Fatty Acid Profiles in the Identification of Staphylococci.J GEN MICROBIOL 1985;131:2023—2033.16. Mukwaya GM, Welch DF: Subgrouping of Pseudomonas cepacia byCellular Fatty Acid Composition. J CLIN MICROBIOL 1989;27:2640—46.Page 130


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items