Statistical Quality Control Methods in Infection Control and Hospital Epidemiology, Part I: Introduction and Basic Theory Author(S): James C
Total Page:16
File Type:pdf, Size:1020Kb
Statistical Quality Control Methods in Infection Control and Hospital Epidemiology, Part I: Introduction and Basic Theory Author(s): James C. Benneyan Source: Infection Control and Hospital Epidemiology, Vol. 19, No. 3 (Mar., 1998), pp. 194-214 Published by: The University of Chicago Press Stable URL: http://www.jstor.org/stable/30143442 Accessed: 25/06/2010 18:26 Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=ucpress. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to Infection Control and Hospital Epidemiology. http://www.jstor.org 194 INFECTONCONROL AD0EIDEM00LGYMarch1998 Statistics for Hospital Epidemiology EDITEDBY DAVID BIRNBAUM, PHD, MPH Statistical Quality Control Methods in Infection Control and Hospital Epidemiology, Part I: Introduction and Basic Theory James C. Benneyan,PhD ABSTRACT This articleis the firstin a two-partseries discussing ences for furtherinformation or statisticalformulae. Part II and illustratingthe applicationof statisticalprocess control discusses statisticalproperties of control charts, issues of (SPC)to processes often examinedby hospitalepidemiolo- chart design and optimalcontrol limit widths, alternate pos- gists. The basic philosophicaland theoreticalfoundations of sible SPC approachesto infectioncontrol, some common statisticalquality control and their relationto epidemiology misunderstandings,and more advancedissues. The focus of are emphasizedin order to expand mutualunderstanding both articles is mostly non-mathematical,emphasizing and cross-fertilizationbetween these two disciplines.Part I importantconcepts and practical examples rather than acad- providesan overviewof qualityengineering and SPC,illus- emic theoryand exhaustive calculations (Infect Control Hosp trates commontypes of controlcharts, and providesrefer- Epidemiol1998;19:194-214). At its most basic level, statisticalquality control is controlcharts to such epidemiologicalconcerns as surgical- rooted in the graphicaland statisticalanalysis of process site infections,bacteremia, Clostridium difficile toxin-positive data for the purposes of understanding,monitoring, and stool assays, medicalintensive-care-unit (ICU) nosocomial improvingprocess performance-general objectives that infections, and needlestick injuries. Additionally,several in essence are quite similar to those of epidemiology. other authors17-26have discussed surveillanceand related Particularadvantages of qualitycontrol charts over other epidemiologytopics that in essence are quite similarto the analysis methods are that they offer a simple graphical philosophyand methods of SPC. manner by which to display process behavior and out- For example,Birnbaum18 recently stated that a "sta- comes, they examine these data chronologicallyas a time tisticallyvalid, systems approachto surveillanceanalysis is series, and, althoughbased in valid statisticaltheory, they the key to determiningwhether occurrenceof unexpected are easy to constructand use. Moreover,once constructed, events is generic ... or an exception"and discussed some even the more complexmethods discussed in partII of this approaches-dependentupon whetherthe underlyingrate series' remainrelatively easy to interpret. is constant,very rare,and so on-for determiningwarning Severalprevious articles that have appearedin this2-5 and thresholdlimits for sentinel events that would signal- and otherjournals' 16discuss healthcare applications of sta- such exceptions.More recently,in a reviewof surveillance tistical quality control charts and other tools generally methodsfor detectingdisease clusters,Jacquez et a124sug- associatedwith statisticalprocess control(SPC), total qual- gested examiningthe time between successive infectious ity management(TQM), and continuousquality improve- diseases with respect to an appropriatenull reference dis- ment (CQI).(Many of these applicationsare describedfur- tribution(assumed by the aboveauthors to be exponential) ther in the "SuggestedReferences" section later in this for non-randombehavior. That is, a deviationfrom the theo- article.)As one recent example,Sellick2 and a subsequent retical exponentialmodel-and in particularan excess of letter by Lee3discussed the basic applicationof statistical several consecutiveshort waitingtimes-would be a low- From NortheasternUniversity, Boston, Massachusetts. Address reprint requests to James C. Benneyan, PhD, Mechanical, Industrial, and ManufacturingDepartment, 334 Snell EngineeringCenter, Northeastern University, Boston, MA 02115. 96-SX-198.Benneyan JC. Statistical qualitycontrol methods in infectioncontrol and hospital epidemiology,part I: introduction and basic theory.Infect Control Hosp Epidemiol 1998;19:194-214. Vol. 19 No. 3 STATISTICSFOR HOSPITAL EPIDEMIOLOGY 195 probabilityevent if no clusterswere occurringand infec- neering andthe closely relatedfield of operationsresearch tiousdiseases were assumed to occurotherwise over time spanmany industries and utilize a host of methods,increas- accordingto a Poissonprocess. Such deviant observations ing their widespreadgeneral value but makingexplaining thuswould be takenas signalsof the presenceof one or their specificssomewhat of a difficulttask. Nonetheless,to moreinfectious disease clusters. establish some context for later discussion,the remainder As the aboveand other authors indicate, these con- of this section provides a basic overview of some typical cernsare quitesimilar to thosein SPCand therefore also skills and methods used by industrialengineers and com- couldbe handledwith "industrial" statistical quality control mon applicationsof these tools in health care, manufactur- charts.The intent of the current series, therefore, is to relate ing, and other settings. the use of controlcharts to thesegeneral types of epidemio- Most generally,industrial engineers and operations logicalissues and to expandon muchof what has been writ- researcherscan be describedas being concernedwith the ten aboutSPC in the healthcareliterature to dateby clini- scientificstudy, improvement, and optimizationof processes cians, consultants,and other healthcarepractitioners. It and process outcomes of any type and in any industry. shouldbe emphasized,however, that, much like epidemiolo- Clearly,this quite broad definitionmight describe many gy,statistical quality control is a broadfield, and not all tech- types of individualsin a varietyof industries,perhaps includ- nicalissues can possiblybe coveredin satisfactorydepth ing epidemiologistsand health care. Of interestto the pre- here.The currentseries instead aims to providea broad sent article,TQM and SPC fit within the abovedefinition and overviewof subjectsat the foundationof SPC,so thatepi- are an integralpart of most industrialengineers' training. In demiologistsand other clinical researchers, based on their additionto these areas, industrialengineers engage in sev- healthcareexpertise, can consider if andhow to bestincor- eral other activitiesthat can range considerablyfrom basic porateSPC into their other methodologies. Particular atten- managementsupport services through advanced mathemat- tion,for example,is givento providinga broaderunder- icaland optimization techniques. For example, a "traditional" standingof industrialand quality engineering, the general industrialengineer might use tools such as process flow theoreticalbasis of SPC, various approaches to applyingSPC analyses,time and motionstudies, hypothesis tests, design to infectioncontrol data, and the role of controlcharts in of experiments,reliability methods, networktheory, com- establishingand improving consistent processes. For addi- putersimulation, queueing analysis, Markov chain analysis, tionalbackground information, Benneyan6 recently provided game theory, decision analysis,economic analysis,linear a thoroughgeneral introduction tothe use and interpretation and nonlinearprogramming, mathematical modeling and of statisticalprocess control charts in a wide varietyof optimization,and regressionand other statisticalanalyses to healthcareapplications, and several references are provided studyand improveissues relatedto inventorymanagement, herefor readers wishing to pursueany of the discussedtop- productionscheduling, facility and location, scheduling, and ics in greaterdepth. capacityand throughputanalysis."7 Note that many of the A moregeneral objective is to stimulateincreased dia- same mathematicalmodeling and optimization methods also logue,collaboration, and cross-fertilization between