Tobacco Documents Online

i AET, EIIDLY' F NEW YORK CITY: SMOlCING-ATTRIBUTABLE MORTALITY, MORBIDITY AND ECONOMIC COSTS Prepared' by: James M. Shul tz „ M. S. Mi'nnesots Department of Health Minneapol is,, Minnesota May, 1986

1.40 INTRODUCTION 1.1!0' SMOK'.I'NGrA'TTRIBUTiABL.E MORTALITY . Csya~;at.`_2 smckiny' is the chief, sinqle pre•.entable 'cause of death in the l'Jni tzd States. A seri es of Surgzon GeneraL '_. Repcrts on smoking and health has implicated cigarette' smo4::ing as a principal cause of death of 'S0.0c.i0Americ_n- each year. This fiy,ure rep:re=_ents the sum of e=_timate= c:f smok:irng7athributab:le mortality from heart disease--1,0,000 deaths (USDHHS, 1980, cancErs of murl ti p1'e sites--1'0, 0-~i deaths (USDHHS, 19SC), and chromic obstructive lung dis_ase--50,,ri00 decths !'tJSDHHG„ 1984') . Recent e:stimates have pL acad' •annaial nati onal deaths f rom smok: i ng at 314,000 (Dffi'.c.e of Technology Assessment, 1985a) and 270.000 (F:ice. Hodqson„ Sinsheimer„ and Ecrowner, 1985). respectively. 1.20 HEALTH ECONOMICS AND EP'IDEMI!OLOI6Y' The ca'Tculation of "costs of iilness" has been reFined a'nd' aystemat:ced during the last two decadles. This endeavor. drterm'ininq'the costs of all diseases in the aggregate. dzterminlnq, the-costs of major categories of diseases (canccrs, h,Part disease), andl determi'ning costs of d'isease_ by primary cost sources (hospitals'~ phy_ician fees, nur_in,- home co4tsJ has app.ropriateTy residedl a~ithin the purviie"-~f heel th ecQnomi cs'. F'r-imaril y wi'thi'n the last decade, several attempts ha•.e bwen made to translate the health toll from cigarette =_mckinq :nt_o economic cost estimates. Calcul ation of the cost:s o:f _m.o!•.ing requires brid'qi'ng across the disciplines or health economics and epidemiolog;.°. This merger of ev:pertise is necessary to define causal 1inb.ages brtwc:en =_moi!ing and multiple diseases, and-t` provide quantitative estimctG= of theproportions of disease cases attributable to smohi;ng.. 1.30 COMPUTER' SOFTWA'RE FOR MORTALITY, MORBIDITY, ANIDi ECONOMIC COST CALCULAT IDNS 5af:ni.ny the dliaease impact of smoking for the group under study can be f'aci 1 i tated b•: appl yi ng state-of-thc--art p.aN,aqyd for ease of use byy heal th prorassionais:. Computer spreadsheets used for New Citv 3.'c'.t l±t. l•1'n' s hd.y been dE:`+'t`31 opFed at the ML (?1'nE?si7t,a De,''Jor t,iT.-r" t ,,,4 ' I-;eal th Qhu1. tc , 198341 whTcfi compute sToPi ny-spec i f. c 85648422

disease i,:Tp;_tLt, incRsl_Fe.' reca11i;1 u'Fijat,ad b`.~` Rice and colleaaucs CF:xce„ Hc:;dy :on', Sir7sheimer, rind Bro:,,rer, ±JST,. 2'. 00-.NEW YORK CITY: SMOKING-ATTRIBUTABLE MORTALITY CiT`ulation of smo4:inq-attributable mortality requires nP definiti!on cf the set of diaqnoses causally linledto .ci.garettz smoking,,, i_) an estimate of the relati .e r:>!-: ronr siTloke.rs coillpared to nonsmokers for e~c;h d'idglnosis, ard ' !7; inf ormwa't ion on smick ; nq' prevaT emce rates. Lorrent smokers are at increased risk for muLtiple dices=_=, primari ly from three dliaynostic cateqories: neoplasms„ respiratory d.iseases, and', cardiovascuolar di=_ea=_es. Fcrmer s„:okars are al so at el evated ri sk compared to never smoE er-, for a period of time po=_t-cessation. That time period varies by di:agma5is and by level of lifetime tob'acco smoE>:e. enwosure. Data from several 1arge prospective epidemiologic studias whi'ch e;ramirrined' the impact of smoking on multipl'ee d'iseasz outcomes permit estimation of relative ri=_l.s for both current and former smokers for approa: i matel y Cfi d'iagno=_es. c^.1CQ THE DIAGNOSIS SET The drzA3nasrs set u'sedl f'or New York Ci ty caLcul ations 1 s identicc,l to that used by Rice, Hodgson, Sinsheimer, cnd' Browner ('1955) .- Earl i er stud i es have i ncl ud'ed addii t i on all smoE:inng-r'elated'~ diaqncases (Rice and Hodqson, 19S7; MD'H. T%4) . The diagnosi=_ set used. by F;i'ce and colleaqucs r_s generally distinquished by robust relative risk estimate-, typ.ically derived from"inulti'ple prospective stad'ies. The folTowinq, diagnoses are inc1!udzdl in"the campuntations: ICD-9 C d no Di is o e 1*1l ;i-<~1C ag s F'espiratory tuber'c'uI'osi s. 1 4 Ci -14!9 Cancers of the L i p, oral cavi ty, phar,;n~: 1 !~%'? Cancer of' the esnphayus 151 Cancer of the stomach 157 C.nnc'er of the pancreas m' 161 Cancer of the 1ar vn:: tJT Cancer s of the-1 unq, trache a, bronc_nus ~, 1 J'( / Cancer of' the uterine cervl:: ~ ~9N9, Cancer of the kidney W' Ah .' ~'. Qa~ .:

C Hypertc+is ion _ tsche:n'ic heart disea5e C'ardiac arrest Carabro.rescular d'ise~~se AtfieroslIerosi s Aortic aneurysm 4?~u-4W 7' F'neumoni a; i nf Z uen: a 4?1-4!aC Chronic bronchitis, emph>/se,ra 4!95 Chronic ai'rways ob=tructi'on~ 5 71 1-C'4 Ul cers The e::'clusion of' liver cirrhosis (appearing in e:~rlier7 iss r'e-asonabl!a because of' the strong, covar'iation ,f and aicoha1l c(Jr'i5UiTlptYon. Other dliagniJses, T1asV iTe'rlt, inclusioni in f'utur'e calculations. Peripheral vascular disease (ICD-9 Code: 44.C) appears to have a stronq 1l1nG: to smoking (USDHHS, 190-') but reZati~xe r'is4:: estimates are generally: 1'ack:ingi. Attributabl'e risks have rerFntly been estimated for per'i'natal and early childhood' cnnditions l i nE: ed to maternal smoG~i ng (Mc Iintosh, 198'4 ). Di aynosEs to consider for inclusion are: short gestation andilow birth weight (ICD~9 C,3(Je: 7-,:r), respir'story di's'tresss s•;zndrcme ('I1CD-9 Code: 76?) , re=_p i ratory cond'i t'i ons of the ne~jt,orn Code: 7'70), and suddP-n infant death svndrome ~'IC'C-'= Code: 792.!:}) . Cor7,.ersel;:, hypertension iis i~ncPuded by Rice end coPlea,;-;es . bPcause of modest relative risks in several major stuid:es. Hbwever, the 1acE:: of consi'Stant e-vidence for a cau=_ai relationship between smok:i'na and hypertension was noted in the 19S'_• Surgeon Gereral "s Report (USDHHS, 19s') . That report did concede a'possiblv causal relationship between s,nol,: ing arrdi ear ly stroke (pri or to ape, -55) . The R;'i d'a, et al'. , methodbloy,y includes all ages in calcuIlaticns o+ sm'oking-attri!butabl'e' czrebrova=_cuTar diseases. Finallyr„ chronic airways obstruction (ICD-9' Code: 496 + :•,a-- incliuded' i'n addition to emphysema and' chroni_c bronchi:ti=. In Minnesata„ and presumably elsewhere, when a patient presents with chronic obstruct.ive Zung disease (CDLZ% symptoms, the physician may attempt' a d'iff'erential diagnosis to classiify the patient as havingchronlic bronchlitils or emph•fsema.. Very of ten, the case i s not easi l'y cl assi f' i ed into one of these categories and' is classified instead a=_ r_hron tc airways obstruction. W 2.20 ' RELATT VE R I SI` EST IMA'TES C11 . . ~

!r ~`r e'ach' dl 7g r, 0-s 1' =_ 1'n th P_ =a1cu, la tl~7ns't rZla '-_ 1•ia' irl SI estimates are a,Iailablr fr-om major lenigitUdinal epldem.e-' 1ogilc StUdl'es and' these estimates are th'cr-7l.(yhl : in a series of four Surgeon General's Reports !UEDHE'uJ, 13'',; USDHHS, 199C': USrJHHS, 199''; USDI-IHS, 19°4) . 1'ni previ'cus wor:. iF:it_e and Hcdgson, 198__;, MDH, 1984) relative ris4:s u(sea were d1:3',,:,G or T1ed1.3n vdil.leS from' si_lmmar/ tables app',.?_arI!'iq ' ..:. report?s cf ttie Surg,eoni General. In the Upolat.ed wC;i-:: •_ F'ice. Hodyi>>n,i Sinsheimer, and Browrner t t Mr; , rr: l4+ ,.';~ risks for each diagnosis are .actlny: 'i Accardingto the meth:~t ~ ;,;_,_ . L car, • ( epu.,r t , ~,~. L.~_1'ijl L: '~. ~1a7~_. 1..~~~'~~}' «ai~l lIlc?. :. i..~, -. ~ _.~. ~~ fi _. 1~-... -..`~.rc-' -.' . _,. ~ Jht la'iJ~ ......, .~ + ~ .~~.1_.. T:i:: ,'Lri:a 'L :3e1-;rl-'r --l:e?c ~ ". _ ~ - . . . .' ,.,': P ., ._., •"..' , _ _~: . ^.I ....j• C:r~ ! ~.1 ~-a!"t. F-'h.`, :Sli~=1..a~1"i'- =f~.~~a.t_~~'i., 1:-Pj "t_-Llt::, 1jr; t.Cd' ..i : r,rm r:S4Cs f,.;r- fT JL a"j-:I1,J5i:S we;r~?' f,7r c_,-.eIlh. ,:al ~ _- ~ EW `i~t7F.'F' C 1 T'' . 1 `~94 ": EW4 tb -..-3':~~.i .,h ~ r 2:, JC?:At'h,s yimony nan s ni_74 er-s J.i..'7 t, t.,. _aIm-yaars of observation in stuld~r i person-years for smokers in StUid•.. i person-years -for nonsmo4::ers in studyi 2.30 SMOKING PR'EVALENCE' RATES. Smoking, preva'lence rates usedl in the spreadsheet calculations are 1984 New York: Cit, , rates based an projection of' large sample data to the entire popu:ation r=t New York City. These numeric and' percentage zstimat=s or cc:crrent, former, and ne•:•er 3mo4:ers were calculated b•,• ,:ip:d'emiol'ogists at the New York City Diepartment of Hralth. ' - ' T,b.a~ _.'H. SMQF ING F EE .'ALENCE F:MTES "1A LF_8: a''-_4 2 5 -_4' 7 5 -4 4 45 -6 4 ~~+ ~',

;:,JF:FEVT . 6 5 '. 7'1.5~.. 'O W=,. ii.'S7.: 21.15. FORMER . e.67 -O. 08 :. 21. 17 26.71 40.75 NEVER 5;'.68 42.22 34.00~~4'. 1C 78. W _ ' . , ^i ~ '. . T ~ ^ _. .. .... ~ ~. ~~ ~ ~ : ' FEMC,LES _ - ~4 - 4 . .J J~`t '7" ~v; ~ '7..J''t 6W . CUF:F'EhiT 26.62% =3.'u?;'. _. 16;. Em.27% 1 a'.W: FORMER 12494 11,51 M I6.54 1b.?6 22. 4' NEVER 60. 44l 54. 8C' 50. 20 51. -7 6'.7"' . 2. 4'O ATTR I BUTABG.E R ISY< CA4.CULAT IDN The critical cal!cuTation in this spreadlsheet is the smokiny- attributable fr.,cticn of deaths from each smoking-related dia.gnosis. The attributab1'e ri5l: calculation manipulatcs the data elzmenf s di scussedl above; for each di agnosi s in the diag;nosis set, the gender-specific smo~~!;ing-attri5atable risl:: i s calicuTated from currant, former„ and never smcker prevaLence rates and from current and' former smcG er rei''ati fe risl;s according Wthe following formula: CpO +p 1( RR 1) + Q 4RR'_')7- I Httributab1!e F:i sk = ---------------- ----- Cpca' + p1 (F:R1) + pZ'CRR2) 7 where: p!'i: - percent of never smokers in group under stud-i p l: percent of' current smok ers i n group under =_tud'/ p2': '- per-cent of former smokers in group under st+_id, RR2: • weiyhted relative risk for former smokers* . RF'1: weighted relative risk for current .smckerst * rel ative to never smoE! er=_ 2'.60 DA'TA SUMMARY Tab l e r^..6r+: ANNUAL SMOFr.:I'NG-ATTFi I L+<UTF',BLE "1OFi'TAL I,T`+ roEW YORY CITY . m AGE GF:GUF' MALES FEMALES FOTN' SEYEJ ~ ~ 20-a- 7294 1954 112 4 3

ZO:-'44 48` 177 :5' 45-64 ., 897 l.=u5--' .-. • 3.' 1 4~J.•.,7~ 2.50 SUMMAFeY OF METHODS i+rom g:nder-spec't f lc siToki,ngprt-"6val,enice rat,r=5 f(or =orr -'o" and former =_mokers ('for al L ages _0-and over, for a,;e 0- - 6F, and far aqes 6C-and over) and gendcr-spar_i f u<< mor ta1 t`. data for C0' d'iagnoses, es.timates of s!moGrinq"attrtbutabl _ rort a1 i t.y wern computed for mal es, fema'1 ts, and both :,emAr= 7E b: i ndli •, i dua'1 ddagross s, by d i agnost i c cat:egory, and f or the t,_t,aii dii,,,gnov:s sYt. The numbersof deaths are presente.d tn ?y'e b'.racE, ets: : 0-441„ 4'5-54, 6C-74, of these bracr:e'ts, a'ges 2~~:> and over., 75 andi over 3.00: NEWYORkC CI?Y: Sf9OK'IN6-ATTR'IBUTABLE . YEARS OF'POTEN'TIAL LIFE'LOST (YPLL) 3.10 METHOD'OLOCiY' AND SOFTWARE OPERATI'ON'. tssi~_~e's inuolved in the cal'cul ation of _nok:.nq-autributa'b1e mortality are' g,ermane to the .:a'1zur: xtion' of years of potential 1'ife lost. hi._..-i7t .. .Ji.i se~nC.e Wll-1w 1995s. I,1~9~O..cJb~.J'~. Havi^g d!ef'ined the diagnos'is set of diseases cauu_aI1v 1tnl=t to sanoking, and ha4ing inserted the two comp'on?n* .:ar:;tlps of the attributable ris1: formula--smoF:_ing p,~reval:ence, rat__ and gender and d i seaye-spe'c &ic, wce,i ghted'. rel at i4n r i=', est'imates--smoking-attributablr mortality ha'= been cal''culated4 Cigaret te smok i ng i s tCnown to ca!_~Ja' nr~m :r _:=e mortali ty i n the sense that nonsmoker=_ are 1 ess L:. Wel , t.= dl e f'rom certal n s+To4: i ng-rCl at'zd diseases and ntilnsilrihet' _. "; .:; do succumb to thes= di ag'noses yeneral l y d'i e at amor-_ e-. .;d r-;nced age than th'ra.~i r coun terp<art s who amoke. Hawe', re1ative ris1:; data' bu a:aa isy only a•.nailable for a.mal 11 na,nber <<f :dia'ynose_. Estima:ti'on of e:;cas=_ smoking-- ,ttl.rt ib~':;t.abte mcrtaliity by age of' death has been perfvr,ra.-' f'c,r r_;rrcer-= c_~f sa:era'1' s'ites (Coll and F'etW. 1TE'Li', for Lsclhow.c heart disasse !^1DH, 1184), and for botlh cacaer _ :' lJ,(',`~ ,1f RtL~O ar,~d~ n~a'.l.l.r.~.6ig'l-Dn^3 1~1?35i, d~:.di, rYG1t, osr3 nCl.o.- :p<-VIc y.c re::at7..C' "l si> data (a?;.Sept fur- t'wio, alE rang•_-

C a.hy.a 1 i ed~d t..i7 i.si=.h~aiG~1, -., h~ _ .. - _. : . _.- _ _:c3~, . 0. - sd annd 6J+I 1 . 4~~1 t..^.=. yasis for the skra..d _nart „+.h„-drL<_qw, the' wcr'k of h-hE=_. ~LinVestity.ytCr'a c_.n i.,•? _. ..:_yd to undc?'"ast1,^=t^a YPLL,. .`.c l iimif:,lng a'ssu!amF't ionl-1!S t;'+at, for each SiT,(DkLny-;^'e,.I at=6 d!lngnos'17, tirlre smWE:ing-,3tF_r.:,b'ut•3bl,e fr'.y,i=tlon r7f ijeatfis.[ s dii=_tri'biited_ in e;:act prapm_rtion to the di stribution cf al i~ d.=_ .,th.s Fr- ..m that cause. Th iazverrtuates a consar : at i._ e,;timnaa' of Y'PLL •based on the' prc_iumption of a di ..prroN'mr}i ]nate amount of earU':' death from"s,nokTngi. rates for r_urr?nt and f,orrner smoE:ar-, and aye and yendo.- - sFeci f l c mortaL i t}; data fcr thE' dl agnes:. s set--chE: -:=r'?a1sheet s[7ft.Wctr-n interlial'1'y' ca1L:l.llatrr_'s siTlokinqattribatable mc,rta;:ity f'or males and females' bl rndi . Z_r_.1. di3".1'UzI','s.. Re5.1.(jc?nt .. l,nthiespr-andshe.e'tn are tC1b.l,es .. ot' ,<ptcted years of' iifa remainin'y for males and f'erri;les b. 5- y••,.r age', •jroUp. F'.r each d2'aLqnosl s, tCr' e'aCh' c-year a'yc groufa, the software multipli'es the number of deaths in thw aNl t b;. the :exprzcted years of I i f'e rem,,i niny for that age Jf d'e.'ath, and mul tipl i es thi s product, in turn, by the smo4: ing- afi.t.ributable risk for the d'iagno=_is. YPLL estimates are aqgregated into four age brracke}':.; 20-44, 45-641„ 6C-74, 70 .A rrd1 o•:.er. These age br<_,chet.s are summed to pro~ide the estimate of totall smoking-attiributable YF'LL for- each Using f'Jew Yurk City data--gender-specific smuki'ng pre•:a di.aynosi's. The spreadsheet provi des gender-speci f i r_ `rF'LL e'sti mate _.~or thQse age brackets by individual diagnosis nndl 5-> diay zstic crateg¢,ry. and' dlspllys total smcF:lng-attribUtable YPLL. Fur: -ha_rmore', these data ?.re al3o pr,-sen'ted tor bcrhisr ez- combinrd.. Additronai't f, all data-is computed' baaedl on twoo e`_s w'f nati'onal ye-zr=.-o'f-e>:p'ected-life-remaini'ng data MCH.:, 10''34'1 re'siid'ent in the spreadsheet: all races cr_anbi'ned and' whites. 3.20 DATA' SUMMA'RY ''ab~1 e 1~~.CAz ANNUAL SMDKI NG-H~TMI SUTM~LE~. 1"EraRS' QF' F'OTE"1TINL LIFE LOST (`rPLL)' F:ELAT I VE TO EXPECTED YEAF:S OF LIFE FE"'?!5L;;!_: New `!'crG:. Cvty (IifE' e;:p:ectancy for all r,___: AGE C;FOUF f1A LES FEt1ALES ' RQTH '_ E'! E' ; _ 9`* 111_5C b:4o5 ' 1,'=71-" _>,: ~ -4 4 1,_~r_,._ , . 4.a--'a.1 J1P/4 ~. - i 7?~. ~.~. _.'t~J'4 ` CV7 14M.,

4.00 NEWYORK CITh':' SMOKIh1IGr-ATTR'IHU'TAEtLE DIRECT HEALTH CARE COSTS 4.1C1 DIRECT COSTS' :rjl~u}: ing-<ittv i',:;Lttablt-~ direi_t c7sts ira ~r,u]_`._., __ ?,... th,e , ,?:.1.o^'',, d'.i.3g inus .1.° anld t:rC?a ,t'i71en1"_ ~f S7ic.r! ;i y-..:..._ _-_.. ~1~:da~=~l ~llll iT?GLi=a1 condltlloris., r'1,11'e:•_'_ l ~D`:>It._= inedi~.= al c=.'i'e^-per'sonai'; l1e;.~1,tj-i, :` lln .1~_lr~I'ry sUch c17st t=t='nt.e?rz _irs hGSpltct.lTze,tlGn,, SlLtr~c~tL=~i}' L1in'Tc-=i1 care?, nttr'siny' hiome cc7re, home health c-;r_., ser:ic~~ ro+ prlmarv ph, sic.ilans and': jpecial ist:s, servic;~~ of non- p'hysici•an h,,al1th practitioner-s, and medl =atiorr co=_ts., ,?Lrrect costs al'.sn inctt_tde the "St_spport" costs o+ proaram a.d;ni^isrration, resear'ch, public he'aTth acti:rtie~, and maaiir_al f,~tt_i l ities constrt_iction costs, a...riie;- =_''_tdi:es ap!plked the "mortali'ty ,_'cmparlson" tA3f:1hi:db1'oM3'f lL!:Ce dn'tr' jL.t`lwelt.'.er, 1977; F';1't=e and Hr;d4,s''._„-. which as=_umed that the proportion u+ attr iL-1_ttab ae ;ncrt•a'1 it f fur 1' aryz dise;,se grrurp.=_ ,~.- a arc::•;y tor esti.matiny the proportron of smaL. Ln~ ~it;~ rlbi_t} - afl'i r3' t~:~~t.'~,. i i re rCi=trnt.) y ~_Lpt~r ~ded' m~?t!z ;i;ci ~,~'. =r ~'1~i:~; an d' CJ11?,agU.ts s :19SJ)' applLa~ dlrect' me_is1jrc_ :at di~f~?r er~tia'1 medical care t_ttili-ation by cur;^'ant ard ._,n•.~Eers relative to never smokers. ' MeCica1 uti1l~~,ti.cn h, a~nc4. in•a st?:tt!rs was obtairred from speci al anal .se= o+ t1~Lii.~n~.T Heal'th Int'er•~iew Si_tre^a erf data. The met`Icd ent,lla =.tzps, (1) Di fferences be+:wken smol ;Ar=_ and ne•,-er smc~ ar ; 1 n capita rates of' ut'11 ilzation of hospital carF and ph ;s; -, services were e::amined b•.: age and gender cateq,-~r-le,s ,:nd tested f'mr stati=_ti'caI =_ignir'icance. (~? Attributable risk., a me'asure of the ma;:irnal pr,,G~r'.i ~r uf hospiital utllization and physiJciani servrt_es attrlb..t~:a:_ tG S,nol: tnL] waa t_al cu1 ated . The Treva1 ence r,_te usatd w.: =_ `.t,et ccinfi';.nCct proportion of c;urr'ent and former smc~.et"S •.-• jJ_i~?ntl'a,l si71~_4rc~c-~ of Ova)rrst,lifla't,lon). Pela'tl,:?, r:Sl s 0-4tCJ the attr:bU.table rls:E> farmula were srrrO~:er' C=ne_-' =mcE,er ratios f,3r hospital' da•j+= and ph'y_ici,an ;) =rc,E iny-<;ttrihUtable hospital diiz•.= __ acd r'~•~~. UR1_At~-?~i]. +'~-Dr t•ha~* d L S:Ca~S??' rQ ur~l.:-5~ c'F 7- ~C,7~p L 85648429

.CcU'1at.=rrV ~.~ in~v 1,~:.~ p att.ribt:rta'ble sr.•I- :~<a: tstfs f,zr tha condi t l o rs ~.b`r aye ; g vrr.cG?r- -.± (4? (tii3es, 17-64 ~e'ar=- '•UFre' =oiT14TY-?s ~cdtr't'ta s:lng 1e Crc f:- r s; =ttnc ; w;t'hi a.ail:;ble t--, ealth rr.ar~ _ast dat; vt l.3kJie ' fL,r z gr?,s 6-4 r and under at3,=c_' OL- a ~Q (c E-tima+lit:~n rf s;noG:i'n~-attriU+_ttabl'e dir-pct herlth •-ara crs`= wr.•s acc-lt-lpltshed' using the following cor,.~ent-Lon: -'p:t '.<Kl L',t} L e--r1,i r, i_frS:l ny hojne,, and other Fri-, teSls 1 arl-3 L =er•~ice cc_it_ u4rre estimated Usi'lllig the attributable' rls. L.~r It:,_;pital ph:;sician ser.icas an'd' drt_ty _osts w_-r : Asti.,natr-d tht_' attribL.itablre risl:: for physician .1s.ti-i. Thi = procr.li,rw repre-ents an irnportant upqrade o-f =,,noG in-3- at'tri~utat,l'e LQ'I method~.5'1oy,1, the shi.' ft from trcrtal ityr to inurbid'i t,t rnea>.ures ~f ;d''irot_t ;.osts'. Hhwe,rer, use of t:torbid'it•/ comparisoni5 is not aComplete solutlon. Exce=_s l;~rb:c~lty in :--_in;:;G ers is not as well, doCLlmentedb'y age, gender, arrdl dl aqncai s code as e ccess mortal i ty. Data on use' 134"::t_Ltt7=1t1'~pnf mr?d1 =-31 ser_..iliCes 1s partl'CLIl':c;r1'V sP:4?1-ch'J. ruurthermori:?, thP logic of cost estimation a=_=_umes that, for a,;i ,e;-i degr~2e af i l ln?_•s, smol!:er , and nons;-~ :.it ct7iTlp?ra't:e ra'tes. Thl's ass!!lTiptLC.- 1=, m . + vU;-h.,+=^'t ed'' ',.- . ~loyt., artd Sch,W-?'i yzt•~r'' 3 ( 1'3Q.) stLtd , u+ 51C:F:ltiy, =^_'n.rci,- -t li=t C5 and Utlllzat'lon' pat'iFr'1E _tr,11 F:uis:- r-F'erm'ar~ntheallth p1<,n picpcllatiun. ~;r• . ra ~-,ad, --r-- e-rs; r.:f ~To-= hospital d~; = 1, ?r:/~J person-.•ear u =,-iri'..iyT :; t(:. n7 n•=in0 V ?r=- ',1r- =s f,h'a 'n ?,'•.pect'ed:l and' ;o4.1'tac;tl,-?nt -_t:.ilL-ation' drd n';-,t difiar between current and neoer S . Sm-;F.er swere .ess l il: el to see~ preventl.,e' ;-ai error in the appl'ica'ticin of prevalence-bas _c-, 1-1., I ,rc-:t`,'cds to lifestyl!e beha~iors such as cigarztte :ric:ude i'1l imprecisiLn in a'ttribi-~'table r1sE! e=_tim~te=. t'~`,`' i~ncompl'ete data on dlfferential mcrbidity for smoVers and n,onsmokers, (3) inc~mnl~_te d'atc on p~;tt.ern= ot util'iaation by smotring statL.ts, (4!" the gross agyr'egatlc~n n- econQlmilc dc3ta NJhich pcj,71 s ci'..,:Sts f:-3r l:ar.~ra i_at?y:or'ies _• d'i sease rather than f or i rrd i' .1 d~taI d'l seaJe=_ . (1S ~ erri-or : n the under'1ying tot.aL L0[1 astimates, and' (,3) error lnt, U; es-trapol'a'tilny from national data: to the yroup undler- St.i-ol :;. (z 4.20 METHODOLOGY AND SOFTWARE OPER'AT I'ONI N' ~. dlt . vuPP: _ed for' r9ewi _,_ -,. ~ L'-....11 .3~L'.:- Jr-i ~ ~i~T •]L "r...='L..*_ Fn.R~~A.y r~n, cc;/"~-- L:7~B t--:1 ~r~ L'1"1~-2 ~ t_~r`2 ~-. :...=r-. r ~ .;C