Document ybzGXzkOYXxQZYqVE8vedNmd

Corporate Health Physics Corporate Occupational Medicine Corporate Product Responsibility Corporate Toxicology 3M Medical Department 3M Center, 220-2E-02 PO Box 33220 St. Paul, MN 55133-3220 651 733 1110 /IR236- WPFoOrkSin(pgpMb)eLmeovrealsndinumCuornreDnat taanQd uHailsittyorAicsasleHssummenatn(Poof pthuelaMtioenasn)and Range of This report documents Battelle's evaluation of the statistical quality and defensibility of 12 data sources of serum perfluorooctanesulfonate (PFOS) levels found in nonoccupational human populations. These data were summarized in the 3M white paper, "Perfluorooctane Sulfonate: Current Summary of Human Sera, Health and Toxicology Data" provided to the EPA in January, 1999. The purpose of this initial collection of human non-occupational serum samples by the 3M Medical Department was to determine the presence or absence of PFOS in human blood from current and historical samples. 3M requested a review of these data by Battelle to determine what additional inferences could be obtained by analyses of these data. Battelle statisticians felt that only the pooled serum from the 18 U S. blood banks was judged defensible for a statistical analysis. Ofthese data, the geometric mean PFOS blood level was 28.2 ppb (95% confidence interval 23.9-33.3). Regional differences could not be explained without additional data and analyses. 001722 WORKING MEMORANDUM on DATA QUALITY ASSESSMENT by Bruce Buxton, Warren Strauss, Owen Chang BATTELLE September 22,1998 cpdsdi(pbaTqfe1n1ltocoeseolu2auhf)otrfltppeetoiiloeedoilsmsueurrscinsdatecsolltrstisataanitateocafictetttcbpbirhneiihadmoseoololesleeobesmlnnpmrpypulUts?icriooonrin.idaoaconcrosfoSnutagediceeAratcaullseodrentdlculebnmordpeytcnmtpsbevorzcotaecoSreeheeehnraqetrstlednahldausastuehitr,tetpefinesnaoeenear(ontc3g-si2Bctfepbtva?rtMt)lpdetoauvaueyojaertdsladbru,itnrMaeisiu3eyfdBnldtfdtllcMaiigregulao3ctbo(e3deoit3Msn,plmo'tMirees)teclnwophalaaIdetslooenoheshcletvacsotamhiefDPtastewclfasaPleFhpofe(,eudnaFdbOlpwfwasadeodOuat3btsShayaaihtarulMoStotsteeokaamdlnwk-erfo;n'sbeeesoseedtpuotteadonnhreagowt1sesacstipwdgep2a:sttlpcdmaeueaemhdrorts(bdleeoc(eua1atpllgPpelthpttes)loehaeidoFeertoceiafisrmIOshnattnrtbstehotafhiSlaahedkraPucveeond.)utnarFetmyadcsmltooiOshbniterwnfaeahesoSiduntsnnsenuioiarsoesftldttrgebonpftnhdeifdesalo-celteotariaoueatrthnerdothccolteansstddaoctrottqtarmtsauowuthegbotsqpchsiaaaeluuhhusaeeslrvoptriitesekicctietgiepmscyoehnhertwoatnrciasipalioerwcraplnotnmleinyalahtdrods.lehkcdiiitdsugrcvaeudei)eIilasht?iaddmenfiodclt,sefoufhitaelt(bnanhoiv2atnnanceshiclev)si3sadaeburiehIlMbytrlsls(ugieyebi3eomlePped)inntoFeasyetnrOnortoaSwlf e Ofoullrofwinindginpgosifnrtosm: this initial assessment of statistical data quality can be summarized in the Wssouonfueefrissccuoeigeusngrpcteresfe,otstre1hn2seutceb1rd9site9beq8yruipa3eoMnwothlsseittcdaahtfisfsed,treinucfomainlseoafrnusotaramcltyeisUswitsi.c.Saas.ljbluyldodgoeedfdebnsasunifbkflsiec(iddeaantttaaossnoouuarlrlccee1s2A. c3Or)i,tfewtrhiaaes, 1jau2ndddgaoetdnaly dTleihvffeeelr1oe9nf9c28e8ps.2oaorpelpeabdpswpeairtruhemnatb9bl5uo%tocdacnboannnfoikdtedbnaectaeexiinpndltaeiicrnvaeatedldowafi2tnh3ao.t9uiottnoaadl3d3gi.te3ioopmnpaeblt.rdiacStamigaennaifdnicaPanFnaOtlyrSesigbsi.loonoadl Data Characteristics for Statistical Defensibility wemdcdnmoInaoeeufoiucmcnmgotrhsahslhrotdediedwtdgrbweseiwereseicriart,daahermpt(leneodh2tq.etmh)imttuctehToosriartrpedtaerkechageebemoertalnliesnderasdeedsitlwnsfia1itdtftntyhirsesiengepsiurrccntee:PmdiahdnscFenm(staaosO1lbrtiialf)ygenSeyrepntiuPeasdzebsmotsFeseliesomsffOpdooeoto1canfShdt2ritoseeaiabilsostblmestlentitovlcadumeaoehptdkilwdcaarssyeootntiiliabntgsaeochmtlncevanaelsldeiotmumpilrinisssrensmtpgioqifooPlcbopuvniarlsFrliasteletraOvrememifinadnmroStsroefuaieeqmnbrosbnreutlueoaogstaasnsvutaonicsebrrofdteeed.oipntgtsliroaNen.iootdpovgifofIuavnenttdPtlelsheaasaFoeaattdioiOuacsnodevs,rrSnfatouesaehstsmerbasnisvitsslsnphcoeetitthoebsroatiaelshoibdpllmleneilUatelrcsUseyee.tttnvS,uqih.ostSenua.a,nelhnd.its,vraedeeetenworhrfmreadlrfehay,oselis(irfwncn3oti.hrt)gecssitoEtahfntaieeecvshe i 001723 f// Table 1. Statistical Requirements for Defensible Quantitative Conclusions Statistical Requirements 1. Detailed documentation of the study objectives and approach, including sampling design {number and types of samples!, and sampling and chemical analysis protocols 2. Specification of a well defined target population about which inference is to be made te.g., the entire U.S. population, or some specified subpopulatloni 3. Unbiased data collection protocols which ensure that all samples are demographicaily representative of the target population; generally this implies random sampling from the target population 4. Along with PFOS blood levels, accurate collection of important predictor or explanatory Information, such as demographics like age, gender, race, etc. 5. Compositing of samples, when done, is performed only for samples within the same subpopulation so that Important potential differences among subpopulations are preserved in the resulting data 6. Sampling is performed at several (i.e., at least three! points in time which are spaced at regular time intervals te.g., every five years) 7. At each point In time, sampling Is performed cross-sectionally from the entire target population or subpopulations 8 . Sampling is performed to provide uniform coverage of the entire spatial domain of interest (e.g., the entire U.S.) 9. Data is collected both for multiple locations (e.g., regions or cities) across the spatial domain and for multiple individuals within each location 10. All data are collected at approximately the same point in time to eliminate the potential for confounding due to temporal trends 11. Specification of target subpopulations which are well defined in terms of important demographic parameters (e.g., age, gender, race, etc.) and from which sampling is randomly performed 12. Either fix sampling at one time and spatial location {e.g., city) or ensure that sampling for all demographic subpopulations Is performed in a balanced design across the same times and locations (i.e., to eliminate potential confounding due to time or location) Types of Estimation Problems temporal Spatial Demographic Changes Trends Trends OOh* 2 rpeesrqotiubmilreaemtmiosen,nwptsrhoiablreleetmmhe.olraesgt esenvereanlrienqnuairtuemreeanntsdaappppllyymeqouraellsypetociafilclatlhlryeteotyapdeisffoefreensttitmypaetioonf Assessment of 12 Data Sources PrsssOueoutsFuntcueyOdcdrMpiceSeetessoadssb.yflewauoHs1sloht,o2isdimtwhc,o1dhoaeo9awtvtebt9heantPe8ra,tyiF,ihn3niOehnMFytaShmidghrs.boauatlnaudrBoefygoaafhd1csvf.eaaallidxsietSleeevaosudrebnabnllttsseootueh,irqBnaiaesudsanseditwoentnieantfetilorl-ollycorlenhm,caaae3csalsuMttawaimpsobarssnolitoett,iratcoeBelfininfaesaatstxtdeliettndoehneglcuntwluemst1omria2viattBeentdteneaadpnmtttaootadetpcpilsoutouleoecmnlduaue,ecrtmlicnoaeoeetuncnasndltstdrtti.oawhobtnenTiaeiyoschonfbipnofseoiaultltnrenhiatesdeibtavhs.wlleeeeliyhmtd1ha2aisttgahbheet ebne jhtigrlmtisGluenneihiooistmdvviideaelwttveeiegetiirpeecalantdetyowspnalhtuirdtaweeso.netarrathanslehtgutTIceiaeidcnctsthavhthhiyueshbneaaoeaaloewnlfatnruelfowrrcatghdrtrhth2rshcremihageiu.,ceestteebhTaa,tdIwe1tdsmsnaasicn2apeootboomaasfaudndluleojetlolseorpinsmedocafc2rlesceeleiuubeinttnssbrtemyetsruccsidweetlaimeoeaoiebntspnetfmdhdmiyeBrptlcoesseaisraacat,,etobyssrrittnsoopitieleozetorornlnsreeeelmtio,dorlqeascwedinetedtuatoemoeebemisaaudrtyafwwoerseromtetmge3her`oraVrtipMmerseatabetltsndpaheeeee?atsdashnetss"sb,nsrisb,ocdceellbdmyierdcsnoustisoioertttctataieeeotncutntnhrodndutemgiooedaodmoi3brjtesnndufaMtd)edn.aoh,sTneotsit.caoanfoTtacuaetnow,ubuAhitnmdnoerlmiiselcowseniessiteenbohona1tew,nyi,.setmllodtalpasriiaItrentesaeetinfyfsyinoarfeososoconToconmhnfmhrtuamsaw.ieBentbiovsebe3nadIlaetfrenisMtictiltmabotbih2aatetnenhysyve,aoldee,letaareBcni1sswisotrl2eooahwaifnhttobecdbwetieputaellarlaeejreslinituoilateaidnaedecdbnsfesera,"idlgaoaodssenXbiaertcumcubofdnro""rarriauco?ttobet(sheeth"timiess.entaedihrgts.te,hae pmwraoosrbefleodmuentsadiwlienoduthlndeontweosrtiatbtbeeonaupdtpoelcaiuccamhboelenf.ttahTteiohdneastteoacswaosauerrsrcaeansrtefcodoremnthmooteesndetc.bryitaeria wihneTreabsuleff2ic.ieTnatbilnefo3rmlisattsion 3 001725 S4/ Summary of Mean and Range of PFOS (ppb)FLigeuvreels1in Current and Historical Human Populations. All Data Analyzed In February - April 1998 Populations 1A. Current Populations (blood collected in 1998) . Non-occupatlonally exposed 3M employees 2. Commercial pooled serum samples Intergsn Laboratory Sigman Laboratory 3. Pooled serum from 21 separate U.S. blood banks (sea Figure 1 for more detail) Description of Sample 31 individuals 3 samples each with 1 0 0 donors 3 samples each with 1 0 0 donors 3 to 6 pooled samples per location with 5 to 10 donors per sample M ean* (ppb) 47 44 33 29 Range* (ppb) 28-96 43-44 26-45 9-56 B. Historical Samples (chronological order) 1. Korean War era U.S. military recruits, 1948 to 1951 2. Swedish samples, 1957 3. Michigan Breast Cancer Study, 1969-1971 4. Swedish samples, 1971 5. MRFIT pooled calibration samples, 1976 6. MRFIT pooled calibration samples, 1980 7. China samples (Linxian, rural province), 1984 8. MRFIT individual samples, 1985 9. China samples (Shandong, rural province), 1994 * Rounded to nearest ppb * * Not detected lO pooled samples with 10 donors per sample 10 individual samples 5 individual samples 10 individual samples 6 pooled samples with unknown number of donors per sample 3 pooled samples with unknown number of donors per sample 6 individual samples 3 individual samples 6 individual samples N.D.* * 2 33 1 31 23 N.D. 31 N.O. N.D. N.O. - 2 N.D. - 59 N.D. - 1 14-56 14-41 N.D. N.D. - 44 N.D. Note: This figure was retyped from a fax sent to Battede on 5/12/98 without changing possible mistakes. 001726 4 Table 2. Overall Assessment of Defensibility of 12 Data Sources Statistical Requirement . DataiSource A l A2 A3 81 82 83 84 B5 86 87 B8 89 1. Detailed documentation ? 7 7 7 7 7 7 7 ? ? 7 ? 2 . Defined target population / 7 7 7 7 7 7 3. Unbiased sampling protocols 7 ? X 7 7 ? 7X X X7 4. Accurate explanatory information 5. Appropriate compositing 7? 77? 7 ? ? ? ? 7 - ? 7 7- 7- 7 7- - - 6 . Regular sampling in time - 7- 77 7 ? ? 7- ? 7 7. Complete sampling at each time - ?- ? ? 777? - 7? ** 8 . Uniform sampling across space - ? 7 7- ?-- - - 7 9. Multiple samples locally - ? 77- ? - - - - 7 10. Sampling at same timels) - ?7? ?- ?- - - - ? 11. Defined demographic subpopulations ? 7- 12. Sampling at same time(s) and locationls) ?- / = Adequately designed and implemented X = Inadequately designed or implemented ? = Unknown due to insufficient documentation = Not applicable to this data source 77? ? 7? 777777 ? ? 7? ? ? Table 3. Notes Regarding Defensibility of 12 Data Sources Al.: Non-occupatlonaflyexposed 3M employees 1. Detailed documentation - The only documentation is in Geary Olsen's 8/15/98 letter 2 . Defined target population - The target population was 3M employees with minimal occupational exposure 4. Accurate explanatory information ~ Some demographic information was collected and included, such as age and gender 12. Sampling at same time(s) and locations(s) - The data appears to be only from employees at the corporate center A2. Commerclalpooled srum samples 1. Detailed documentation - The only documentation is in Geary Olsen's 8/15/98 letter . A3. Pooled seram from 21 separete U-S. blood banks 1. Detailed documentation - The only documentation is in Geary Olsen's 8/15/98 letter 3. Unbiased sampling protocols - Thera is potential bias as a result of the 5 0 % refusal rate 8. Uniform sampling across space - The samples are geographically dispersed 9. Multiple samples locally--There are multiple regions and multiple individuals within each region B1. Korean War sire U.$: military recruhe.1948 to 1961 1. Detailed documentation - The only documentation is in Geary Olsen's 8/15/98 letter 2 . Defined target population - The target population was U.S. military recruits from the Korean War era : B2. Swedish samples, 1987' 1. Detailed documentation - The only documentation is in Gearv Olsen's 8/15/98 letter B3. Mlchlaah Breast CancerUtudv. 1969>1971 1. Detailed documentation - The only documentation is in Geary Olsen's 8/15/98 letter 84. Swdlsh:samples. 1971 ; ; 1. Detailed documentation - The only documentation is in Geary Olsen's 8/15/98 letter 5 001727 Table 3. Notes Regarding Defensibility of 12 Data Sources (Continued) ______________________ ______ BB. MRFIT pooled calibration sitipIm , 1976 1. Detailed documentation - MRFIT is documented in Geary Olsen's 8/15/98 letter and in multiple journals including JAM A, Vol. 235, No. 6, Feb. 23, 1976, but subsampling for 3M is not documented 2. Defined target population The target population was men aged 35-57 with multiple risks for heart disease 3. Unbiased sampling protocols - The sampling protocol selected certain characteristics, with multiple screenings 4. Accurate explanatory information - Information about the original sample population was recorded, but no information about the pooled calibration sample is known____________________________ ,________ _________ B6:. MRFIT pooled calibration sample, 1980 / ______ _____ 1. Detailed documentation - MRFIT is documented in Geary Olsen's 8/15/98 letter and in multiple journals including JAM A, Vol. 235, No. 6, Feb. 23, 1976, but subsampling for 3M is not documented 2. Defined target population - The target population was men aged 35*57 with multiple risks for heart disease 3. Unbiased sampling protocols - The sampling protocol selected certain characteristics, with multiple screenings 4. Accurate explanatory information - Information about the original sample population w as recorded, but no information about the pooled calibration sample is known______ ___ ________________________________ _______ B7. Chlna samples (Unxlan. rural provlnce), 1984 ____________ . 1. Detailed documentation - Documented in Geary Olsen's 8/15/98 letter and in Journal of the National Cancer Inst., Vol. 85, No. 18, Sept 15, 1993, but subsampling for 3M is not documented 2. Defined target population - The target population was all the residents of Linxian aged 40*69 without debilitating diseases or prior esophageal or stomach cancer 3. Unbiased sampling protocols - The sample appears to be representative of the target population 4. Accurate explanatory information - Information about the original sample population w as recorded, but no information about the individual sample is known_________________________ ______ B8 . MRFIT individual samples. 1985 __________ 1. Detailed documentation MRFIT is documented in Geary Olsen's 8/15/98 letter and in multiple journals including JAM A, Vol. 235, No. 6 , Feb. 23, 1976, but subsampling for 3M is not documented 2. Defined target population - The target population was men aged 35-57 with multiple risks for heart disease 3. Unbiased sampling protocols - The sampling protocol selected certain characteristics, with multiple screenings 4. Accurate explanatory information - Information about the original sample population was recorded, but no information about the pooled calibration sample is known__________________________________________ B9. China samples (8fjandona. rural province), 1994 : ' 1. Detailed documentation ** The only documentation is in Geary Olsen's 8/15/98 letter __________________ jbwatcsGdTtuohanoiohiiademaldvlceslgroleueeeepyoemlcnmdlzykietw,siaetesndhftidnshhnogiiti,tfesoemrafbc3tneltbh3ouMpairleoMtnctleohrynkc'nnoestltofndihuoddossoesfdtaartaiteudotrtocvmaaeedocnhansyucoapeiuofulmnbmoroalumdottrbbegiaiamnclnejpihleenagn,trcaloewetosdttadsupainiratevonraiibtntteontaaothyistnleavieyian,nl3saesr,ogntMeridthorsadhosise.rlegtltcwrtandaiaeIuanttierfimsailiagmslsessvlteteeiaiaiocttscnsclhiugtspaslatueoarramloldrarbltcpeybipyetloonuheenhdnnutletilbaoeccnyootofinldiaetuffrolreshniisttentgerhhiostnsoidliewewnebdna,(alshs1smietl)ti.i.2hu,ecfsoied.hdPsoaa,nynmasronsgautyonynodapb,tolstehlijssettisnrueeo.anneigsmmgugtcstIinrtiotnrapoucfhemtialredesicisncsointmaeedhadmgittsalesiel.sootwenytaasnhfducmatabaeoabsfctpldososopreilraobnioehsmnssreditmsfg(gcapoesahtsplo.riaiugtmsrsnuom.sott,gdrieetpccMiodtweaaoloceslsRtofaolseoy(stlFslwhessoI.eegawTtons.)),d. pOufrpalolstehseidsatthaesdoautracseestporofv1i9d9e8dpboyo3leMd,sperruombabfrloymthUe -mS.obstlodoedfebnasnibklse(fdoartastsaotiusrticcealAa3n)a.lyAslitshough 6 001728 psrdTis(cneaetooehacfpltnseeteoapsiccrrisgireeelttboneenirlacxdndicezraiieewminilnsnngoa,araginlanitttsnenhatt-ohelrleytyueenphssdnshoedipeebstaoomne.iplacnnefoosusxtotegiemetafdrnbslaettemhipbniacaohitastaalsiinsoetcpsinesneoocesrtnoowriero,nfneieusattstahitsuhlacealladltetclevidaaykndetosiag)itnnsndwwgouftre,rbrhiosoftijimihtedoccciduhifntteocehtndeaehdecesdepcitpeahbalsrsipreayontpmwbiertceehadphuasbelirultecloaleltsharthsirlwbanleagitlgncmheotkdihoPncitedo3FuhoMmfOsbwuacaSbrmoentehranbkprebetl.lrooiseosloeWfsialtndbmymbtelylopttaeaooolyveredemae3shclMbcatahdhaovk.aenbneetyklcHabovsereoaotreoelnwdqinedauebdevalsiwenmtreki,idtth.ed Statistical Analysis of Blood Bank Data rTi1ne7hpcelgrueerdseoeicgnnergtnaa:tptlihyvieccoamlllleyeacdnteiPsdpFeaOnrsdSeda(pnRpaelbyd)zCleedrvoepslos.oHblleoodwobdelvobeoarnd, kassavhmaarpvieleetsybfeorefonamscsoaumnmbipnitndieeodpnestnoadreeesntnitmeMcaeitsnesnaaernsyoa,ttaioannadlly 1) pEoapchulpaotioolnedhasdamanpleeqiusarlecphraensecnetoatfivbeeionfgtrheeprreegseionnteadl pinopthuelastaiomnp(lie.e).., each member of die 2) Ethaecshapmopollee.d sample has a known number of donors, and each donor is represented equally in 3) Tsahmepslaems.pling and analytical methods for obtaining serum PFOS levels is the same across all IAtlissou, nthliekedlaytathhaatvtehethseamfopllloews ionbgtaaidndeidtiboyna3lMlimcoitmatpiolentse:ly satisfy all of the above assumptions. 1) pruTgeenehfonlueipeksbrleeealdlolywoptotdhohoppbauatadrlntoathkintecisaoisptrneaeampitbnerlpeohelsoeaaedscdnhottnobgootetnahcoliheyngaerRtnadhepcodfehsrieooCcmtfrrhobeapsegtsaii.ndrotgonicn.riaepRtpaearndetthsbbeelnlroo,toeotdhdde-ibsneai.n3eskM.a,smc'asiptrilseeeassrmeawpprrheleeer.sreeepFntruthearesttiehvbneeltroamootifdovtrehbeea,oniftki(ss) 2) Iwtaiss usenlkenctoewd.n how many donors are in each pooled sample, nor is it known how each donor 3) msMmaeemottshhet.oooddfss.tahrAeessethaselalmsoapfmlthees.e caTanhmaeleyMsfirisonmwneaRssoeptdaerCsfaormormspsleebdslobpoyrdo3bbMaabn,lktyhsed, osaonnaiotltyihstiaclivakelemslyiegttnhhioofisdcesasnsahtmloyupdlldiifnfbgeerethnet Aeflftehcotsugmhotdheelsewdasatuasmedaytoneosttismataistefyaanllatoiofnthael gaessoummeptrtiiconmsedaentaPiFleOdSableovveel,:the following random ln(PFOSjj) p + B j +Ey, where 7 001729 dslnaif(mPfeFprlOeenSfcjrejo)m(roenpthrteehseiethnntaBstultohraoeldnloaBtgaunrsackla,-lleeox)gpb(terptaw)nisesfeotnhrmethneeadmtiPoeFnaOanlSPgFleeOovmSelelatersvisceolmcaieat attehndePwiFthiOthbSltohleoevdje-tlbh,aBpnojkoisalentdhdethe ntbhaaentikion)n.dailIvtmidiseuaaansl,ssuaamnmdepdleijthPisaFttOhBeSj rfmeosleildaosuwuasrleaemrnreoonrrtmsleaaflntdduinsthetrxeibpmulaetiianonendPwFbiyOthSthmeleemvaenoldzaeetlroi(ttsahcneoddrrvifeafsrepiraoennncdceienobg2eBbtewltoweoeeednn BioodBanks, and that Ejj follows a normal distribution with mean zero and variance a2withinBloodBanks- Tcinoadnbiflviedid4eunpacrleorveingidtieoersnvatghleseofemosrteittmhrieactoemvseeoarbanltsal iwgneheodicmhfreowtrmiecrfemitfteoianungn.dthtios fmaloldoeulttsoidtehethdea9ta5.%Taanbdle959%lists the Table 4. Parameter Estimates for Blood Bank Data Param eter Estim ate U 9 6 % C o n fldenee In te rval 9 9 * C o n fid e n ce In te rval N a tio n a l G e o m e tric M e a n P F O S L e ve l (a**) (ppb) 2 8 .2 (23.9. 33.3) (2 2 .5 . 35.4) o* B e tw e e n B loo d B an k a [ln (p p b ll2 .0 9 5 - o1 W ith in B lo o d B a n k s Iln (p p b )) .0 5 2 - - Table 5. Regions With Geometric Means Falling Outside Confidence Intervals for Overall Geometric Mean Region , Santa Barbara, CA Omaha, NE Billings, MT Grand Rapids, Ml Cheyenne, W Y East Orange. NJ Newark, DE Anchorage, AK Las Vegas, NV Santa Rosa, CA Corpus Christi, TX Kansas City, MO Davenport, IA Scottsdale, AZ Meridian, M S Minneapolis, MN Lafayette, LA Greenville, SC h Geometric Mean (ppb) :;; Outside 9 5 % Cl Outsld# 9 9 % Cl 3 T5.9 5 16.2 y3 22.4 3 23.1 3 23.3 3 23.5 5 23.9 3 25.3 3 26.6 3 27.3 3 28.9 3 29.8 3 3 3 6 3 3 31.9 38.5 37.5 45.6 45.7 51.5 ------------------- y7 ---------7 ---------- --------- 7 ------------------------------777 --------------------------------------- PtfTheFhlelOeooSavubetcosroaivdnlelecmretenhseteuaraln9tts5iios%dnea.mcroTeonshnufisilsdttrehoanitfgcethehtheninaufttame1crb2vt eatohrlfaoct1oft8nrhesbegtrlrioeuoocinsdtaeslbdiamgfnnoekirafsnitchhsaeafndaetlsgvltiieanmorgimaastteeiigotdrnnincifbaimetcitaoewnnaetnalelyPngaFedbOoioemSvgeeleetarovinmcedlemsbtrteeihlacoantw 8 001730 MtddmfOihnaiimeffcediffntaueoeaonnsrrhbreteesPasranni,eFtpecchrNsOoeevaslsEettSi,hdsicaaa,ncavtdnMontaddahnrpNieecitBxpaie,odipblnlLynoliltaliraan3inailftonMgytadrisyoaabp,epnteopMtarsttpooweaoTurd,ntebluLiaaemsocltAnenyitarossbvyoani(lesnfbosdaduteowthcadGiieihtsstabhoreacvaelasmhaaencrntckbhoeivasalrdbr.oibelplofiloTeelerdiooxt,otymdpbmSi.laCnbaasnvFnanukmneauocstkmfrhaota.yieracegxyIytbxauatepitbmrneioeselftopsrophsuaclroe)enram,,esotrwsetaetshidhtaobhieesouliiorelnbnrenpcltcwosehoSlosaosafo.dstosnuietrtHbblhadtplaeoheBnrbrweoeekaesxexrsearbvipremiraeelnqearriguma,ntyii,aroeCettnaiodoAna.nli,nfgofrul Resampling the 18 Blood Banks mhd3eaMeltaaps,aeustvhkreaeedlrdueiaBnataerdettithefvfleaelreruieatnnhtciteoeiprnrtosaooiopnleiftndy2i0osinanpmPpoFpbflOteohsSrefmbsrtoloaomrtoeisdtbthileceetavwsleaemlmseenewritgithtehoeoifmngrregiansepaiohmmgicurpaamlplinhlagoincctdaahltmeilosoaencx.a1itm8iobunlmso.oPdIFnbOtahSneklscevutoerrlesnt rpspb(dpsfImueeroaofruernos.eomcepsagtdamcivsidu.memogpi,tidlsreonltabsosepeihcuitttterslnrioaeibseinooagnnnlwuentn1yidgoitfme.8iafimtshfytltchephfBhliiioiiobgneoaenccycuyhtlsr,iffateetolttdaeatlyoandhrtcan1pkcmtetebt3t8o.oiaflmsiMnayuprafelwtgsaootsui.tufcoegcholimartandanreWueoctatfoltiricemlposdoorsueasaehrnneatnbtmssiishsnjsmc)ueo,baacpigtrbidmaestiolngntiueeatencfpegomnhtfosogslfbva.ierttleothttmpilsctagrewehrtotifcarhoanraonPtttegohteipmuFmoyrmdaihlatOntedmoimfincvtSrwrsnhatoesaiamoeyob,mnrlastiulw,focaecoholbtobcsdeohrecetunimdleinlhhirptsimiaeeatabtleyieteelvtltspepivoxduios,ooeopstoqnwtreralldoutaatmoveaahalibknna,nnampbanrtnttltinitoiiiahaauhwfkgnwtbsetyeshgreeti,iphltthsdtrichBooiibteooasyseouseawnan.pulrvtmeascttolameCidlrtemi,rpsyl(noilu.sldsaeptiltac)ebnloAlimrhemoyngieclcfvpcisogtototeioitalygennmehfcrmnrstgtaiooarehaetmwnnopnbertvdleaheoiehaglttiwnsiciihrcbetcnhdio,yeneahbgosnsetbltuetcwtmrooiehlaelfodreiolearatdrnyxtrUceeep.dtxw.oeiSpFerice.stothc,rt 001731