Document Eq3YYKkq0Zq0ZEa3XXanaxz4V

Corperati HealthPhysics Corporate OccupationalMedicine Corporate Product Responsibility Corporate Toxicology 3M Medical Department 3M Center,220-2E-02 PO Box 33220 St.Paul,MN 55133-3220 651733 1110 An EpidemiologicAnalysisoftheInpatientand OutpatientClaims Event and EpisodeExperienceof3M Decatur Employees,1993-1998 The purposeofthisresearcphrotocolisto examinetheCorporateHealthStrategies Ingenix(CHS) inpatienatnd outpatientclaimsdatabasesfrom January1,1993 through December 31, 1998 ofDecaturchemicaland filmplantemployees.Claims dataarenot availablferom CHS priorto 1993. The observedclaimsevents(individuacllaims)and episodes(multiplreelatedclaimsgrouped accordingtotheepisodetheyform)willbe compared to an expectednumber calculatebdy indirecsttandardizatiomnethods to adjust forage and gender.Correctedfortheirdifferenatge structurewse, willexamine theratio oftheobservedtoexpectedchemicalplantto theobservedto expectedfilmplant.The many limitationosf theanalysiosf claimsdataaredetaileidnthestudyprotocol.This studydesignwas originallpyroposedin 1998 butwas placed "on hold"to allowforthe completionof therandom sample assessmentof serum fluorochemicalmeasurements of Decatur chemicaland filmplantemployeesas wellasthecompletionof theabstraction, datainputand qualityassuranceoftheDecaturwork historyrecordproject.Both of theseprojectswere requiredto be completedpriortotheclaimsanalysisinorderto providethenecessarysupportinginformationregardingemployee serum fluorochemical levelsand work historyexperience.Estimateddateof thefinalreportforthisstudyis September 15,2000. ADDENDUM EPIDEMIOLOGY PROTOCOL Date: February27,2000 Epidemiology,220-3W-05 Medical Department 3M Company St.Paul,MN 55144 3M Company EPI-0004 Page I TitleA:n EpidemiologicAnalysisoftheInpatienatnd OutpatienCtlaims Event and EpisodeExperienceof 3M DecaturEmployees, 1993-1998 Addendum Study StartDate:February27,2000 EstimatedDate of FinalReport:9/15/2000 ProtocolNumber EPI-0004 IRB Approval Exempt Expedited x PrincipaIlnvestigator: Co-investigators: Geary W. Olsen,D.V.M., Ph.D. Jean M. Burris',R.N.,M.P.H MicheleM. Burlew,M.S.. JeffreyH. Mandel, M.D.,M.P.H. Study Director: Sponsor: JeffreyH. Mandel,M.D., M.P.H. 3M MedicalDepartment,220-3W-05, St.Paul,MN 3M Company EPI-0004 Page 2 INTRODUCTION Thisepidemiologystudyprotocol(Epi-0004)was placed"on hold"in1998 while two supplementalresearchactiviticelsarifietdhepotentiadlifferenceisn fluorochemical exposureamong employees attheDecaturchemicaland filmplants.These two research endeavorswere the "FluorochemicalExposure Assessment of DecaturChemical and Film PlantEmployees (EPI-0006)"and theabstractiodna,tainputand qualityassurance oftheDecaturwork historyrecordproject.As a resultof thesetwo researchactivities, thisstudy'smethods have been modifiedand improved.Therefore,thisAddendum supplantstheoriginalstudyprotocol.The purposeofthisrevisedresearchprotocolisto examine theCorporateHealthStrategieIsngenix(CHS) inpatienatnd outpatienctlaims databasesfrom January 1,1993 throughDecember 31, 1998 of Decaturchemicaland filmplantemployees.The observedclaimseventsand episodes(asdescribedbelow)will be compared toan expectednumber calculatebdy indirecsttandardizatiomnethods to adjustforage and gender. Providedbelow isa detailedescriptioonf thestudymethods. METHODS DescriRtionofCHS CHS islocatedinNew Haven, Connecticut.CHS createsand updateshealth claimsand relateddatabasesfor3M and many othercompanies. CHS usesParallaxe softwareforitsanalyticailnterfaceC.HS alsoprovidesanalyticonsultingservicestoits member companies which willbe usedinthisstudy.CHS willserveas theonlyclaims dataresourceto3M forthisresearchstudy. I 3M Company EPI-0004 Page 3 DescriRtionof CHS inratienatnd outratiendtatabases There areseveralCHS databases.These include:1)themonthlycostand utilizatidoantabase;2)thefocusedMSR database(allowsdetailedinvestigatiofntrends found inmonthly costand utilizatio3n);theeligibilidtaytabase;4) theprescriptiodnrug database;5) theinpatiendtatabase;and 6) theoutpatiendtatabase.CHS can onlyprovide claimsdataforDecaturemployees who have chosen the3M Medical Planor BlueCross/BlueShieldof Alabama. Beginningin 1996,3M Decaturemployees had the optionofchoosingan HMO; tothebestofour knowledge,few (<50) have done so. It appearsthatequalpercentagesof chemicaland filmplantemployees choose HMO coverage althoughwe suspecta greaterpercentageof youngeremployees chooseHMO coverage thanolderemployees. The formationof a completeinpatientrecordincludesthehospitarlecordand the physicianrecord.There ar-e27 Major DiagnosisCategories(MDCS) [CHS, 1993]. There areapproximately500 DRGs which aresubcategorieosf the27 MDCS. DRGs were developedatYale Universitaynd group similarpatientsD.iagnosiscodes(ICD-9) can also be obtainedfrom thehospitalrecord.The MDCs thatareofparticulairnterest inthisstudyarethefollowing: MDC I MDC 4 MDC 5 MDC 6 I%MC 7 MDC 10 I%MC II MDC 12 MDC 13 Nervous System RespiratorySystem CirculatorSyystem DigestiveSystem HepatobiliarySystem and Pancreas Endocrine,Nutritionaalnd Metabolic Kidney & UrinaryTract Male Reproductive Female Reproductive NMC 14 MDC 15 MDC 16 MDC 17 3M Pregnancy,Childbirt&h Puerperium Newborns and Other Neonates Blood and Immunological Myeloproliferativaend otherneoplasms Company EPI-0004 Page 4 From theseMDCs more specificDRGs and even more specifiIcCD-9 codes can be selectedforcomparativeanalysiswiththenormativedatabases.The inpatienctlaims databasecan be analyzedforallclaimsfrom January1,1993 (CHS's contractwith3M began atthistime)throughDecember 31, 1998.A singlehospitalizatiroensultsinonly one DRG and MDC code despitethepossibilithyerecouldbe one or more diagnoses made duringhospitalizatioInt.isalsoimportanttorecognizethatmultipleinpatient claimsof a singleDRG may representone employee or more thanone employee withthe same DRG. Thus,each inpatienctlaimcannotnecessarilbye consideredan incident(i.e., new) or prevalencediagnosisduringthe,studytimeperiod.Rather,thisanalysishouldbe viewed as an evaluationof theexistenceof inpatienctlaim eventsfrom January 1993 throughDecember 31 1998. Itwillbe possiblew,hen necessaryt,odeterminehow many individualcsontributetdothenumber ofclaimeventsfora l@MC ora DRG. Thiswillbe helpfulin datainterpretatiwohnen an excessof observedclaimsisreported Most physician-diagnoseidllnessedso notrequirehospitalizatioTnh.erefore,we willalsoexamine,forthesame timeperiod(1/1/93-12/31/98)t,heCHS outpatient claimsdatabase.As withtheinpatienrtecords,an employee may have one or more outpatienrtecordsof thesame ICD-9 code. Furthermore,a claimmay notrepresenta physiciandiagnosisas outpatienItCD-9 codes may be used as a suspecteddiagnosiscode intheorderingof medicaldiagnostipcroceduresforwhich a claimisfiledT.hus,each outpatienctlaim,likeinpatienctlaimscannotnecessarilbye consideredan incident(i.e., 3M Company EPI-0004 Page 5 new) or prevalencediagnosisduringthestudytime period.Rather,thisanalysishould be viewed as an evaluationof theexistenceof outpatienctlaimeventsfrom January 1993 throughDecember 31 1998. Itwillbe possiblei,fnecessaryt,oidentifythenumber of employees pereach ICD-9 code.Finallyi,tshouldbe notedthatthesame personcould have severalclaimsforthesame diagnosisin boththeinpatienatnd outpatiendtatabases. The thr-edeigitICD-9 codes thatwillbe examined foroutpatienctlaimsarethe following: ICD-9 Code 140-239 240-279 280-289 320-389 390-459 460-519 520-579 580-629 630-676 710-739 740-759 760-779 780-799 DescriRtion Neoplasms Endocrine,NutritionaMle,tabolic,Immunity Blood and Blood-FormingOrgans Nervous System and Sense Organs CirculatorSyystem Respir-atorSyystem DigestiveSystem GenitourinarySystem Pregnancy,Childbirthand thePuerperium MusculoskeletalSystem and ConnectiveTissue CongenitalAnomalies ConditionsinthePerinataPleriod Symptoms, Signs,and III-DefineCdonditions Should itbe necessarya,n analysisof thefourthdigitoftheICD-9 code isfeasible. Inpatienctlaimeventsregardingnewborns areassignedto theemployee. Neonates (lessthan30 days of age)generallyb,utnot always,have theirinpatienatnd outpatienctlaimsassigneddirectltyo theemployee'srecord.By 30 days ofage,most infantshave theirown separaterecordsand arethen definedas an employee dependent. Employee dependentrecordswillnot be examined. DescfiRtioofCnlinicCaalreGrOURS'(rCmCG)Methodology 3M Company EPI-0004 Page 6 Claims event databy itselfa,s describedabove,arenot an idealrepresentatioonf clinicaelventswhen directluysed. To betterunderstandthetreatmentof diseases, individuaslerviceswith theirespectivdeiagnosiscodes shouldbe linkedtogether accordingto logicthatcombines clinicaelxpertisaend detailekdnowledge of administrativdeata. When claimsrecordsarelinked,theycan form diseaseepisodes, which representalltheservicesprovidedforthediaposis,treatmenta,nd management of a diseaseor condition.Thus,where individuacllaimsdatarepresentan "event",claims datathatisgrouped by diagnosis,treatmentand management of a diseasecan represent an "episode." Because of CHS's recentmerge with Ingenix,Inc.,CHS now has accesstoCCG software.This softwaregroupsallvisitsp,roceduresa,ncillarsyervicesa,nd prescription drugsused inthediagnosis,tr-eatmenatn,d management of more than400 diseasesor conditions.An episodecan be considereda constellatioonf one or more claimsdata recordsrepresentinagn occurrenceof a diseaseor conditionfora particulairndividual. An episodeisassignedtoa CCG Class.A CCG Classhave severalimportantqualitieIst. is:1)clinicalhloymogenous; 2)exhaustive3;)belongtoa diseasehierarchya;nd 4)is classifieads a chronicor acutecondition.SeveralCCG ClassesconstitutaeCCG Category.SeveralCCG CategoriesconstitutaeSpecialtyCategory.Altogethert,hereare 442 CCG Classesthatfallinto103 CCG Categoriesand 20 SpecialtyCategories.The CCG softwareprovidesa summary outputfilewithone recordper episode.The methodology on how theCCG softwareworks isdescribedelsewhere[Valdiviaand Van I Vorst,20001. For thepurposeof thisstudy,theCCG 3M Company EPI-0004 Page 7 softwarewillbe utilizetdo describe theepisodeexperienceof thestudycohorts(describedbelow). DecaturStudy Pot)ulation The initiasltudypopulationconsistosf allfull-timaend inactivemployees atthe Decaturplantas identifieidn the3M epidemiology'Dsecaturwork historydatabaseasof January1,1993. Employees hired,transferretdoa non-Decatursiteor terminatedduring theintervaolftimefrom I/l/93through12/31/98willhave theirclaimeventanalyses limitedtotheirtimeofemployment. Full-timeactiveemployeeswho retireodrwent on long-termdisabili(tLyTD) withinthesixyearintervaolftime(1993- 1998)willbe followedthrough12/31/98.However, itshouldbe recognizedthatemployeesplacedon LTD areeligiblfeorMedicare after18 months;thereafte3rM becomes thesecondary providerE.mployees who retirperiorto age 65 may have 3M as theprimaryinsurance provider.Medicare becomes theemployee'sprimaryproviderupon reachingthe employee's65h birthday. Cohort Size Based on areviewof therecentlycomputerizedepidemiologyDecaturwork history database,thenumber of employees eligiblfeorthisstudyarethefollowing(percentsin parentheses): 3M Company EPI-0004 Page 8 Category Chemical Film All Employees 652 659 Gender Male Female 530(81) 122(19) 558(85) 101(15) Age <30 30-39 40-49 50-59 >=60 59(9) 124(19) 238(37) 182(28) 49(7) 36(6) 82(12) 194(29) 267(41) 80(12) Average age 45 years 49 years "Egh" exposurejob* 498(76) 490 (77)* Only worked inchemical(film)plant* 388(60) 424(64) Worked onlyinchemical(orfilm) p* lantand.atleastf,rom 1983-1998 176(27) 269(41) refersto comparablenon-exposedjobsinfilm(i.e.o,peratorsm,aintenanceand supervisorss)imilarto thosejobsinchemicalwhich were considered"high"exposure (seetextfordefinitioonf"high"). *refertsochemicalemployeeswithno work experienceinfilmafter1/l/83and film employees withno work experienceinchemicalafterI/l/83. Data Analysis Claims Events For each yearfrom 1993-1998,thechemicaland filmplantcohortswillbe identifieidnordertocalculatteheobservedand expectedinpatient(NOC, DRG) and outpatient(ICD-9)claimsexperience.The overallclaimeventsexperiencef,oreach MDC, DRG and ICD-9 code,isthesum of theindividuaslixyearsof observedand expectedexperience.Employees who chose an HMO 3M Company EPI-0004 Page 9 in 1996,1997 and 1998 willbe excludedfrom theanalysesfortheyear(s)theyhave HMO coverage. Analysisof theclaimeventsexperiencewillbe examined by whetherthe employee isconsidereda chemicalplantor filmplantemployee. Each employee'swork historyrecordhas been examined by theinvestigator(sJMB, GWO, NUVM) todetermine whethertheemployee: 1)everworked inchemical,filmorboth;2)worked inchemical, filmor bothduringtheclaimsanalysistime periodof January1,1993 throughDecember 31, 1998;and 3)worked continuouslyinchemicalorfilmfortheentire10 yearspriorto thebeginningof theclaimsanalysistime period.For employees who have worked in both plants,theirrecordswere reviewedto identifwyhich planteach employee primarily worked atinhis/hercareeratDecatur.Both chemicaland filmplantemployees willbe subdividedintheanalyses(seebelow)as towhethertheyhave,orhave not,worked in theotherplant.The majorityoffilmplantemployeeswithpriorwork experienceinthe chemicalplantworked atthechemicalplantfora shortperiodoftime(Ito3 months) and thisusuallyoccurredduringthefirsytearoftheiremployment. Film plantworkers were employed,on average,25 years.Like filmplantemployees,many chemicalplant employees have worked atthefilmplantatthebeginningof theircareer.In addition, employees who have site-wideresponsibiliti(etshosewho may work in bothchemical and filmsuch as environmentalhealthand safetyspecialistwse)re assignedtothe chemicalplant. Each employee was assignedajob titlwehich describedtheperson'susualjob activitwyhilea Decaturemployee. Thesejob titlewsere:boileroperatore,nvironmental healthand safetyspecialisetn,gineer,milloperator,maintenance,officeworker,operator, 3M Company EPI-0004 Page 10 qualitycontrolworker,shippingclerkand supervisorB.ased on thefindingsfrom the Decatur exposure assessmentstudy(Olsenet al.,1999),maintenanceworkers,operators (includescellc,hemical,milland waste)and supervisorisnthechemicalplanthad significanthliygherserum fluorochemicalevels[e.g.a,pproximategeometricmean of I to2 ppm ofPFOS (perfluorooctanesulfonaantde/)orPFOA (perfluorooctanoaiccid)] thantheotherjob categorie(sapproximategeometricmean levelsof 0.2to0.5ppm for PFOS and/orPFOA). Thereforet,hesetwo job groupswillbe thehigh and low fluorochemicalexposurecategorieisn thechemicalplantanalyses.Allchemicalplant workers,regardlesosfjob titlew,ere found tohave,on average,significanthliygher serum fluorochemicalevelsthanemployees assignedtothefilmplant.The geometric mean forfilmplantemployees approximated0.1 ppm PFOS and lessthan0.1 ppm for PFOA. Thus,thefilmplantworkersareconsideredtobe theleastexposed and willbe consideredthecomparison population. The observednumber ofclaimsbasedon MDC'S, DRG's and ICD-9 codeswillbe compared toan expectedvalueusingindirecsttandardizatimoenthods [Tsaiand Wen, 1986] toadjustforage (<40,40-49 and >=50) and gender(male,female).Two normativedatabases willbe used tocalculatteheexpectednumber ofclaims:1)theentireU.S.3M population excludingtheCottageGrove (locatiocnode = 064)and Cordova (locatiocnode = 034)sites due to theirfluorochemicaplroductionactivitieasn;d 2) theU.S.3M populationexcluding employeesattheSt.Paul(locatiocnode = 089),Woodbury (locatiocnode = 343),Cottage Grove (locatiocnode = 064)and Cordova (locatiocnode= 034)sitesT.he lattedratabase, hereaftecralledthe3M manufacturingplantnormativedatabase,willbe developedto providea normativedatabasethatwould likelybe more representatiovfe3M manufacturing I 3M Company EPI-0004 Page II plantemployeesas itexcludestheSt.Paul3M Center(corporataend researchemployees).it alsoexcludesthedowntown St.Paulmanufacturingplantasthe3M St.Paul sitesuse the same locatiocnode (089). A thirndormativedatabasew,hich consistosfothercompanies thatbelong totheCHS aggregatedatabase,was not consideredappropriatteouse by CHS staffbecausethedifferenhtealthinsuranceplansthattheseothercompanies usemay greatly influenctehelikelihooodfhospitalizatioTnh.isisespecialltyruewhen studyingsmall populationsuch as thoseidentifieidnthisDecatur study. StandardizeCdlaimsRatios(SCR) forMDCs (inpatienDtR)G,s (inpatienatn)d ICD9 codes (outpatientw)illbe calculatedT.he SCRd@e.,iijstheratiooftheobservedto expectedclaimsexperienceforthechemicalplant.The SCRfilmistheratiooftheobservedto expectedclaimsexperienceforthefilmplant.Ninety-fivpeercentConfidenceIntervals (95% Cl)of theSCRche,@,wand SCRfil.willbe calculateudsingmethods describedby Rothman and Boice [1979].However, theestimatethatismost importantistheratioofthese two indirecsttandardizerdatiowshich we willdefineastheClaimsEvent Ratio: CER = SC&he..i.,S/CRfilm The directcomparisonoftwo indirecsttandardizerdatios(e.g.S,tandardizedMortality Ratios)can be made onlyiftheagestructureasresimilarorage-specifimcortalitryatiosare homogeneous. Because thechemicaland filmplantpopulationhsave somewhat differenatge structureussed forindirecsttandardizati(osneecohortsizeabovewhich shows chemical plantemployeesaresomewhat youngerthanfilmplantemployees),theSCRs forthe chemicaland filmplantpopulationasrenotnecessarildyirectlcyomparable.We willusethe methods by Tsaiand Wen (1986)tocompare whethertheratiooftwo indirecsttandardized ratiosc,orrectedfortheirage structuries,similartoan unadjusterdatioT.he ratiooftwo 3M Company EPI-0004 Page12 indirescttandardirzaetdiousp,oncorrectfioorntheiargestructduirfeferenccaensb,e consideraendunbiasedstimatoofrarisrkati[oTsaaindWen,1986]. CERc.ctw= ..oe SCR,@he.@@:.I..ted/SCRfil.-,r .d EachagecorrectSeCdR isnotmeaningfbuylitselofn,lytheratiCoE,R.,,,Wo,ftwo correctSeCdRS.AlthougThsaiandWen suggestthattheconfidenicnetervoafClER,.,,,,t@d canbecalculautseidntgheprocedurdeesvelopebdyEderearndMante(l1974f)ortheratio oftwoPoissovnariablseusc,hmethodsi,nfactc,annobteused.UnlikCeER whichisa Poissovnariab(lseumObsd.-3 ge .p, /sum Obsfil.-3 ge .p,) Multiplbiyedaconsmn(tsum ExPfilm-3 agegroups/sumExPcbeamg-eg3roups),thecalculatoifoanconfidenicnetervfaolr CERc.r,ei,sn,owtreadialpypareansteachObscon-@.ed-chedObsc,@ed-sru,b,g.roup (n = 3 age groups)isnot a Poisson variableand neitheristhesum (personalcommunication, Dr. Tim Church, biostatisticiUani,versityof Minnesota). Therefore,our analysisstrategywillbe to calculateboth CER and a CERc.,r,,,eIdf.thetwo ratiosarerelativelsyimilarw,hich we suspectthey willbe given thefactthattheirage structureasrenotsubstantialldyifferentt,hen we willonly calculatetheconfidencelimitof CER (SCRchnicdSCRfiln)using themethods proposed by Ederer and Mantel (197). We willcalculateCER and CE&.recledforthose inpatient(MDC and DRG) and outpatien(tICD-9) claim categoriewshere therearetwo or more observed claimsin eitherthechemicalor filmplantpopulations. Because of confidentialictoyncernsregardingemployee medical informationt,he identificationf theinpatientand outpatienctlaimsexperienceisknown onlyby CHS. Therefore,thecohortsmust be establishebdy 3M priorto thedeterminationof theclaims data. Record linkage with theCHS claims data willincludethe employee's name, gender, socialsecuritynumber, employee number and date of birth. 3M Company EPI-0004 Page 13 The observedand expectedinpatienatnd outpatienctlaimsexperience(andtheir SCRS, CERs and CERs,.r,,,ewdi)llbe determinedforseveralcohorts(excludingthose individualwsho have HMO coverage).These willincludethefollowingcomparisons: 1)allchemicalplantworkers(n = 652)compared toallfilmplantworkers(n= 659); 2) allchemicalplantworkerswho neverworked infilm(n= 388)versusallfilm plantworkerswho neverworked inchemical(n= 424); 3)all'high('definedabove)exposurechemicalplantworkerswho havenotworked in filmforatleastsinceJanuary1,1983,ifever(n = 498),compared totheijrob counterpart(sconsidered'leasetxposed')inthefilmplantwho have notworked in chemicalforatleastsinceJanuary1,1983,ifever(n= 490,thiswould include operatorss,upervisorasnd maintenanceworkersinthefilmplant)a;nd 4) alllong-term'highe'xposurechemicalplantworkerswho have notworked infilm sinceatleastJanuary1,1983,ifever(n= 211),compared totheijrob counterpartosf leastexposed'workersinthefilmplantwho have notworked inchemicalsinceat leasJtanuary1,1983,ifever(n= 345); Claims Episodes Comparable indirecsttandardizatitoenchniquesa,s describedabove fortheclaims eventsdata,willbe used toanalyzetheclaimsepisodedataderivedfrom theCCG software. We willutiliztehesame two 3M normativepopulationsforcalculatioonfexpectedvalues. Basicallyi,nsteadof calculatinagn observedtoexpectedratioofclaim'eventsw'e, will calculataen observedtoexpectedratioofepisodesbasedon theCCG softwareI.nother words:CEPR = SCEPRt,@ni.I/SCEpRfiwlh.e)re CEPR = Claims EpisodesRatio;SCEp&h,n-@,w StandardizedEpisode Claims Ratioforchemical;and SCEPRfit=m StandardizedEpisode Claims Ratioforfilm.The CEPR willincludethe CCG Class,CCG 3M Company EPI-0004 Page 14 Categoryand Specialty Categoryepisodes. AdditionalStudyLogistics JenniferBilodeauand Doug BateswillcoordinatetheCHS aspectsofthisproject. Geary Olsen willserveasthe3M principailnvestigatorJ.ean Burris,MicheleBurlew and JeffMandel willserveas3M co-investigatorFso.r each cohortthat3M provides, CHS willdeterminetheobservedand calculatteheexpectednumber of claimsby MDC (inpatientD)R,G (inpatientI)C,D-9 (outpatiencto)desaswellastheCCG ClassesC,CG Categoriesand CCG SpecialtyCategories(asdiscussedabove)pereach age group and gender. Calculationof theSCRch,., SCRffl.and CER (andtheirassociated95% confidenceintervalsw)illbe performedby 3M investigatoarswellasthecalculatioonf theSCRchem-corrected, SCRfit..ctd and CER .. ctw.The titloef therevisedstudyprotocol ischanged (fromtheoriginaplrotocolt)oincludeoutpatienctlaims,claimepisodesand extendthestudytimeperiodone additionaylear(through1998). OualityAssurance This studyprotocol,any addenda totheprotocold,ataanalysesa,nd a copy of the finalreportwillundergo a QualityAssuranceaudit.Permanent recordsof allotherdata generatedduringthecourseof thisstudyaresubjecttoprivacyand confidentiality considerationsA.ll datagatheredor generatedincludingprotocoladdendum and the finalreportwillbe archivedby theMedicalDepartment,3M Company, St.Paul, 3M Company EPI-0004 Page 15 Minnesota.Upon reviewand approvalof thefinalreport,studyresultswillbe communicated to theDecaturemployees. DISCUSSION The purposeof thisstudyistoexamine theinpatienatnd outpatienmtedicalclaim eventexperienceof Decaturchemicaland filmemployees from January 1,1993 through December 31, 1998. Sincethewritingof theoriginasltudyprotocols,oftwarebecame availablteo alsoanalyzeclaimsdataintoepisodes.The overallhypothesisisthatthe inpatienatnd outpatienctlaimseventexperienceas wellas theepisodeexperienceis comparablebetween employees of the3M Decaturchemicaland filmplantswho diffeirn theirserum fluorochemicalmeasurements by,on average,one orderof magnitude,butdo notdifferintheiraccesstothesame medicalcaredeliverysystem (i.e.t,heDecatur medicalpracticeasvailablteothem).With thecomputerizatioonftheDecaturwork historydatabase,we can alsorestricttheclaimsanalysistothosechemicalplant employees who have worked solelyinthechemicalplantforatleast10 yearsprior (startinIg/l/83)totheclaimsanalysitsimeperiodinjobsthatareconsideredtohave the highestpotentiaolf exposure. ...This claimeventanalysiscan providea much broaderperspectiveof possible employee healthoutcomes thana retrospectivceohortmortalitystudy,especialliyn instancess,uchas this,where employees'serum fluorochemicalexposurelevelsb,ased on findingsfrom toxicologystudiesw,ould not suggestan increasedriskformortalityW.e can alsoaddresssome questionsof femalechemicalemployees about thepotentiaelffect, ifany,of placentatlransfeorf theiroccupational-derivfelduorochemicalserum levels. I 3M Company EPI-0004 Page 16 The CHS claims eventdatawillallowforan evaluationof theclaimeventsrelatedto childbirtahnd theneonatalperiod.We arenot certainwhethertheCCG episodesdata willprovidedataregardingtheneonatalperiod. The analysiosf claimsdataisless subjectivtehandirectemployee interviewtshatwould lacktheextremelytime-consuming validatioonf self-reporthsrougha medicalrecordreview.The studydesignalso maintainstheconfidentialiotfyemployees'medicalclaimsdata. There areseverallimitationtshatmust be understood.Firstand foremost,an analysiosfcwms datashouldnotbe considereda studyofrisk(i.e.t,heprobabilitoyf diseaseinan 'exposed'group compared to theprobabilitoyf diseaseinan 'unexposed' group). Claims data shouldnotbe consideredincidenceorprevalencedata.Rather,the studyshouldbe viewed as an analysiosf thecomparison ofmedicalclaim eventsbetween two populations(chemicaland film).A claimdoes notmean thatitrepresentasn accuratediagnostidcescriptioonfthereasonforthehospitalizatioorntheoutpatienvtisit. As previouslydiscussed,ICD-9 codes foroutpatienctlaimsmay be relatedtothereason why a procedurewas done,notbecausethediagnosiswas made astheresultof the procedure.The claimsepisodesdataalsodoes notprovideinformationregarding incidentrisk.Itmay, however, providea reasonableprevalencedescriptiodnuringthe sixyeartime period.The employees'claimeventand episodeexperiencewillonly be examined fora recent6 yeartimeperiodata chemicalplantthatbegan fluorochemical productionin theearly1960's.Itisnotpossibletoconducta comprehensivelongitudinal analysisof claimeventsand episodesdata. The studyisalsolimitedwithinthe6 year time frame as we can examine onlyfortheexistenceof claimsdataforindividualwsho remain full-timaectiveemployees or areinactiveemployees between 1/1/93and 3M Company EPI-0004 Page17 12/31/9T8h.eclaimeventandepisodeexperienocfeemployeewsho subsequentqluyit, terminatedor transferredduringthesixyearsarenotconsidered.Employees who are placedon LTD willlikelyonlyhave claimeventsreportedby CHS forthefirst18 months as Medicare would thentakeoveras theprimaryprovider.Likewiseclaimevent and episodeexperienceof thoseindividualwsho retiraend reachtheage of 65 willlikely not be observedas theseindividualasreprimarilycoveredby Medicare. Part-timeand summer employees as wellascontractorse'mployees arenotincludedin theanalysis. Anotherlimitatioinsthesmallsizeofthestudypopulation(totaolf652 chemicaland 659 filmplantemployees).The person-yearwsillbe furthereducedby theexclusionof thoseemployees who have chosenHMO coverage. Despitetheabove shortcomingst,hisclaim eventand episodeanalysiscan provide3M Decaturemployees a betterperspectiveas towhether thehealthexperience ofchemicalplantemployees,asmeasured by theexistenceof CHS claimsinformationi,s significantdliyfferentthantheclaimsexperienceoffilmplantemployees. Additional investigatiocnasn be designedshouldtherebe significantmloyre observedthanexpected claimseventsand/orepisodesforchemicalthanfilmplantemployees forspecifiMcDC, DRG, ICD-9 and CCG codes. REFERENCES 3M Company EPI-0004 Page 18 CHS (1993).Parallaxanalystinterfac1e997 codes catalog.New Haven:Corporate HealthStrategiesI,nc. EdererF,MantelN (1974).Confidencelimitson theratioftwo PoissonvariablesA.m J Epidemiol 100:165-167. Olsen GW, Logan PW, Simpson CA, Hansen KJ, BurrisJM, Burlew NM, Schumpert JC, Mandel JH (1999).FluorochemicalexposureassessmentofDecaturchemicaland film plantemployees. St.Paul:3M Company. Rothman KJ,BoiceJD (1979).EpidemiologicAnalysiswitha Programmable CalculatorU..S.DBEW NIH,PublicatioNno. 79-1649. TsaiSP, Wen CP (1986).A reviewofmethodologicalissuesofthestandardized mortalitryatio(SNM) inoccupationaclohortstudies.IntJ Epidemiol15:8-21. ValdiviaT, Van VorstK (2000).ClinicaClare GroupsTm Methodology. IngenixInc., San Francisco.