Tobacco Documents Online

p87/10-044 The New England Journal of Medicine -Nonsmokers - - -smokers a ~ O a I ; f 1 o i 0 4.000I / %• '~ ~ E . . - soo [f n •~ . ~ o . V 3.000 : ~• .. e 4 0 a ; w 5 a 2.0001 ,1- / . 1 2 h°30 00 ,,0 °K°° I°:`° fis?`' Age Group (years) Figure 1. Estimated Annual per Capita Health Gare Costs for Dutch Men in 1988 and for the Male Population in a Life Table, According to Age and Smoking Status. Per capita health care costs for women in the same aga groups era very similar to those for men. . sun•hal is 50 percent and 21 percent, respectively (among women, 67 percent and 43 percent). These differences in the numbers of elderly people have a profound effect on the health care costs for the population, as Figure 1 shows. In the younger age groups, in which mortality even among smokers is quite low, a population of smokers has higher health care costs than a population of nonsmokers, but in the groups of men 70 to 74 and over (and those of women 75 to 79 and over), the lower per capita cost of the nonsmokers is outweighed by the 11 19 percent of totat costs among men and 12 percent the mixed, the smoking, and the nonsmoking pop- ulations are presented according to disease category. All the smoking-related diseases (with the nota- ble exception of stroke among men) arc associated ttith higher costs in a population of smokers and Iower costs in a population of nonsmokers. This rc- lation is particularly strong for the diseases with the highest excess risk: lung cancer and COPD. Howev- er, in the mixed population of smokers and non- smokers, smoking-related diseases account for only greater number of people remaining alive. I of total costs among t%nmen, and the costs of all•the As Figure 1 shotts, the nonsmoking population as a whole is more expensive than the smoking popu- lation. The area bentren the curves in which the smokers have higher health care costs than the non- smokers is smaller than the area between the curves in which the nonsmokers have higher health care other diseases have precisely the opposite relation. In a population of smokers, the costs associated with all the other diseases are less than those in the mixed population: 14 percent less for men and 18 percent less for womcn. Among nonsmokers, the costs of all the other diseases are I S percent higher for men and costs than the smokers. This is shown in greater dc- I 7 percent higher for women. I The risk of the diseases not related to snwking is tail in Table 3, where the total hcalth care costs for I 1054 • Octobcr 9, 1997

P97 /10.044 The New England Journal of Medicine THE HEALTH CARE COSTS OF SMOKING JAN J. BARENDREGT, M•A., t•uc BONNEUX, M.D., AND PAUI J. VAN DER MMS, PH.D. ABSTRACT Bachqround Although smoking cessation is de- ' sirable from a public health perspective, its conse- _ quences with respect to health care costs are still • debated. Smokers have more disease than nonsmok- ; ers, but nonsmokers live longer and can incur more health costs at advanced ages. We analyzed health i care costs for smokers and nonsmokers and esti- ; mated the economic consequences of smoking ces- : sation. Metbodr We used three life tables to.examine the effect of smoking on health care costs - one for a i mixed population of smokers and nonsmokers, one i for a population of smokers, and one for a popula- ; tion of nonsmokers. We also used a dynamic meth- ; od to estimate the effects of smoking cessation on ; health care costs over time. Rcrulrr Health care costs for smokers at a given age are as much as 40 percent higher than those for nonsmokers, but in a population in which no one ! smoked the costs would be 7 percent higher among ' men and 4 percent higher among women than the ' costs in the current mixed population of smokers ; and nonsmokers. If all smokers quit, health care j costs would be lower at first, but after 15 years they would become higher than at present. In the long ! term, complete smoking cessation would produce a I net increase in health care costs, but it could still be ! seen as economically favorable under reasonable assumptions of discount rate and evaluation period. ; Conclusions If people stopped smoking, there : would be a savings in health care costs, but only in ; the short term. Eventually, smoking cessation would lead to increased health care costs. IN Engl J Med I 1997;337:1052-7.) I I S MOKING is a major health hazard, and since nonsmokers are healthier than smokers, it seems only natural that not smoking would sare monn• spent on health care. Yet in eco- nomic studies of health care it has been difficult to determine ufio uses more dollars - smokers, who tend to suffer more from a large variety of diseases, or nonsmokers, who can accumulate more health care costs because they live longer. The Surgeon General reported in 1992 that "the estimated aver- age lifetime medical costs for a smoker exceed those for a nonsmoker by more than S6,000." On the other hand, Lippiatt estimated that a 1 percent dc- cline in cigarette sales increast:s costs for medical care by S405 million among persons 25 to 79 years old.= Manning et al. argued that although smokers incur higher medical costs, these are balanced by to- bacco taxes and by smokers' shorter life spans (and hence their lower use of pensions and nursing homes).+ Leu and Schaub shonved that even when only health care expenditures arc considered, the longer life expectancy of nonsmokers more than offi sets their lower annual expenditures.c We have analyzed comprehensively the health care costs of smoking. in doing so %ve have distinguished between the assessment ofdifPerences between smok- ers and nonsmokers and the assessment of what would happen after intervenaons that changed smok- ing bchaioc Would a nonsmoking population have lower health care costs than one in which some peo- ple smoke? Are antismoking interventions economi- cally attrac[ive? We sought to answer these questions and to determine the consequences for health policy. METHODS Analysis of Smokers end Nonsmokers We examined the effect of smoking in the general population (a mixture of smokers and nonsmokers). We studied the insi- dence, prevalence, and monality associated Nith fnC major cace• gories of disease - heart disease, stroke, lung eancer, a heteroge- ncous group of other cancen, and chronic obswaisc pulmonary disease (COPD). We used data on these diseases, in addition to mortality from all othEr causet, in an extension of the standard life table, the multistate life table, that includes multiple health states, such as "alh+e, heolth•r" and "alne, with hean disease."" Differences in the frcquency of the smoking-related.diseases between smokers and twnsmoken arc commonly expressed as rate ratios. l•ning these rate ratios, the precalence of smoking in the pofwlation, and the age- and sex•spt.ific incidence of the smoking-rclated diseases in the mixed population of smokers and nonsmokers, at can estimate the incidence of the diseases sepa• rately among smokers and nonsmokers.• - Assuming that the «lathe sunfial of persom with these diseas- cs is the same among both makers ard nommokers, nso addi• tional life tables can be calculated -one for smokers and one for nonsmokers. The three life tables differ with regard to the inci- dcnce of the smoking•rclated diseases and therefore in their asso- ciated prcraknce, disease-specific nwrtalin•, and ocer9l mortality. Because of the difference In mortality, more people remain alive in the life table for nonsmokers than in the table for smokers, par• ticulary in the older age groups, and there are corresponding dif• fercnces in life expectancies. In constructing the life tables, e-c used epidemiologie data on the incidence and pm'iknce of the diseases; -t' data on mortalin• from Statistics \ethedanJs,u data on smoking (Table 1).12 and rate ratios from an rnxniexof the litenturc!s tVe tested the sen• From the tkpartment uf htNic Fkalth. Erasmus t'nkersitc, CQ Box 17dA, 1000 DR R,xterdam, the Nnherbnds, wfierc reprint rcquests xh,mtJ Iv aJJrc+seJ w alr aucnJrcgt. 1052 • Octobcr9, 1997

p97/10-044 The New England Journal of Medicine the health care costs associated with "all other dis- ease" at the 65-to-69-year-old level for people over the age of 65. The costs for the mixed population and for the nonsmoking population became virtttal- ly the same, and those for the smoking population ivere still the smallest, albeit by a small margin. Figure 2 shows what the economic consequences would be if all smokers stopped smoking. After this abrupt change, the total health care costs for men (the "no discounting" cun•e) would initially be low- er than they would have been (by up to 2.5 percent), because the incidence of smoking-related diseases among the former smokers would decline to the Iev- el among nonsmokers. Prevalence rates start to de- cline, costs decline, and the intervention shows a benefit. With time, however, the benefit reverses it- self to become a cost. The reason is that along with incidence and prevalence, smoking-related mortality declines and the population starts to age. Growing numbers of people in the older age groups mean higher costs fbr health care. By year 5, the benefit de- rived from the presence of the new nonsmokers starts to shrink, and by year 15 these former smokers are producing excess costs. Eventually a new steady state is reached in which costs are about 7 percent higher - the difference between the mixed and the .nonsmoking populations. Figure 2 shows the consequences of discounting the projected costs and bencfits by various percent- ages. It is apparent that discounting, even at a rate as !mt• as 3 percent, has a huge impact, and this impact becomes greater as the costs become more distant in time. Having all smokers quit becomes economically at- tractive when the future benefits arc larger than the future costs or, in terms of Figure 2, when the area below the x axis is bigger than the area above it. From the figure it is clear that this depends heavily • on the duration of follow-up considered and on the discount rate. With a shorter evaluation period and higher discount rates, stopping smoking looks eco- nomically more attractit•e.1V;th a longer evaluation period and lower discount rates, quitting smoking loses its economic advantages. The break-et•en year, when the initial benefit is exactly balanced by the eventual cost, occurs after 26 years of follow-up when there is no discounting, after 31 years with 3 percent discounting, and after 37 years with 5 percent dis- counting. At 10 percent discounting, the break-even year occurs after more than 50 years and may not oc- cur at all. DISCUSSION This study shows that although per capita health care costs for smokers are higher than those of non- smokcrs, a nonsmoking population would have high- er health care costs than the current mixed popula- tion of smokers and nonsmokers. Yet given a short enough period of follow-up and a high enough dis- count rate, it would be economically attractive to eliminate smoking. Some earlier studies have had differing results, partly because many have focused on costs attribut- able to smoking. From rate ratios and the prevalence ~ of smoking in a population, the proportion of the total number of cases of a disease that can be attrib- uted uted to smoking - the population attributable risk f - can be calculated19 Given the costs according to I disease, one can calculate the' costs attributable to I smoking.r" For instance, in the life-table population ~ of mixed smokers and nonsmokers about 8 percent I of total health eaoe costs among men and almost 3 percent of total costs among wromen can be attrib- uted uted to smoking. Attributable costs, hoast(er, can be interpreted as potential savings only when the dis- eases do not affect mortality. In the case of most smoking-related diseases, reductions in smoking re- duce mortality, creating new possibilities for mor- bidity from other diseases in the years of life gained. Other studies of this subject estimate lifetime I health care costs, taking the differences in life expect- ancy ancy into account, and find that smokers have high- er medical costs.3,2ta: In our study, lifetime costs for smokers can be calculated as $72,700 among men and $94,700 among xomen, and lifetime costs among nonsmokers can be calculated as $83,400 and 5111,000, respectively. This amounts to lifetime costs for nonsmokers that arc higher by 15 percent among men and 18 percent among ammcn. The studies cited above apply discounting to the lifetime cost estimate. Because costs incurred at old- er ages are discounted more, this approach reduces lifetime costs for nonsmokers more than those for. smokers. For example, when one applies discount- ing to our Gfe tables for smokers and nonsmokers, smokers have higher health cane costs when the dis- count rate is at least 4.5 percent in men or at least 5.5 percent in eromen. We disagree with this ap- pnoach, however. Discounting should be used for purposes of evaluation and should'not be applied in a descriptive context, such as the estimation of life- time costs. Our analysis is not very sensitive to substantially different values in the rate ratio. Neither is it very sensitive to the age-related increase in the cost of "all other diseases"; that is, an increase that is less steep in the United States than in the Netherlands will not lead to different conclusions. Including additional smoking-related diseases could change 'the results only if those diseases generate morbidity and costs without raising the excess risk of mortality. There may be some of these eonditions, such as cataracts, but they are unlikely to change outcomc. For exam- ple, in our data all eyre diseases, most of which arc not related to smoking, account for about I percent of total health care costs. 1056 - Uttnhcr 9, 1997

p97/10-044 THE HEALTH CARE COSTS OF SMOKING sitl.9n• of the analysis M• recalculating the life tables with excess risks (the rate utio-1) that werc 50 percent higher and 50 per- cent kmer (Tabk 2). The medical costs we used werc based on a study that allocated the total costs for health carc in the \ethcdands in 1988 (39.8 billion guilders, or $19.9 billion, at the present exchange rate) to categories of age, sex, and diseasc.1• We used the Dutch popula- tion in 1988 and the pm'alcnce rates of the smoking•rclatad d'o- cascs from the life table for mixed smokers and nonsmokers to estimate the costs per case of disease according to age and sex. The remaining costs werc assigned to'per capita costs for all oth- er diseases" (in categories according to age and sexl by dividing the costs by the number of people in the eategory in question. Using the per capita costs for each disease and the `all other d-n- ease" costs, we calculated the health care costs for the populations included in the three life taWes. TaSEE 1. t'ttE\:UE\CE OF SMOxIS4'.' . AaE m) SwoeEa lIUE rxftLLE percant 0-la 0 0 15-19 20 20 20-34 35-19 . 39 42 37 36. -e0-64 39 27 a63 34 13 i Assessment of the Effect of Complete Smoking Cessation ` The estimated health care cost derimed from the life table of ; ( nonsmokers can be seen as an estimate of the cost of health care if no one ever smoked. It does not ptoside an estimate of the ° health care cost if all smokers stopped smoking. In the latter ease, ; the size of the elderly population would initially be the same as ~ in the mixed population of smokers and nonsmokers. For it to be- : come similar in size to the elderly population among nonsmokers, ~ in which more elderly people arc alhe, would take several lYars, ' even if mortality declined rapidly• ' To describe the epidemiologic changes asHt the changes in the : population over time, a dtramic model is needed. For this purpose, . we needed a series of 4nked life tables, one for each point in tiose, with the population at a given age (a) and time (t) depending on : the population at age a-I and time t-I, and on the incidence of ' disease and the associated mortality between r-I and t. We used ', the Prevent Plus computer prognm, witich is designed to evaluate ~ interventions concerning risk faaors dytlamially:us ~ This dynamic analysis produces a projection of future health ! care costs. To asuss the economic attracthxness of an intencn- j tion that would make smokers quit, these costs arc compared i with those expected when no intervention is made. One ditliculty : in such an evaluation is the fact that most people prefer to rcceixn benefits as soon as possible and to postpone payments. E.roms- 'mists call this phenomenon "time prcfercnce,"'a' and it is taken into account by discounting the future benefits and costs - that . is, those further aw•ay in time arc given lower weights in the os•er- ail evaluation. . . The degree of time preference is expressed in the discount rate. Typical values range from 0 to 10 percrnt, with 0 pcrxnt mean- ing that thero is no discounting and no time prdercnce and 10 • percent mcaning that there is a strong time prckrenee. Since there is no genenlly agrecd-upon discount rate, we used various rates (0, 3, 5 and l0 percent) in cvaluating the intervention. A second difficulty in evaluating future costs and berKlits is de- ~ ' ciding how far into the future the anahsis should go. Therc is no generally agreed-upon duration of follow--up in this type of analysis. For each projection of discounted costs and benefits, we thercfure report the duration of follow-up at which the bcncfits and costs ex- pected in the future exazly balance each other. (the break-nrn ~ year) - the point at which cartying out the intcnention is neither more nor less economically attractive than not doing so. RESULTS Figure I shows the annual per capita health care costs for male smokers and nonsmokers 40 to 89 years old, in 5-year age groups (the costs for women in the same age groups arc very similar). Per capita costs rise sharply Nvith age, increasing almost 10 times from persons 40 to 44 years of age to those 85 to 89 years of age. In each age group, smokers 'Data arc atvragea for 1988-1993 in the 4ther- IandsP Tatxt 2. R+TE Rarios a..D SE.srrnTn RI.GFS ASSOCUTED wiTH Fn'E CATt00Mt5 I OF DISEASE. eattccar R.R R.ne ISrsmsmts• RanuEN Heart . 342-a) Lungeancer 10l3.3-1J.ss Struk . 2 t I S-2S) Other eanasst 211.3-2.31 COPDj 23113-37) 'Rate rn&w'elcr u, the nte uf the disease in ulsuken a.vmparcJ with nansmuken. The bw<r and upper bmnds nf the sensitdny nnge wesc eal- avbsed a 1+0.31RK-11 and I+I.3/RR-1). n:- . sr[RIYelY, wtlQt RK ,ICINKR the nle ratio. tThis categuwy includes ne,plas/m etspt Fw aonu.h. ulkxYaal, lung. bnau. rnxitalr, and skin canccn•and k•nign tunxes, zcorn deM,cr chnmic ubarucais< pulmun.sry• disease. incur higher costs than nonsmokers. 71te difference varies with the age group, but among 65-to-74-ycar- olds the costs for smokers are as much as 40 percent higher among men and as much as 25 percent high- cr among women. Hmt'L%'er, the annual cost per capita ignores the dif- fercnces in longetity between smokers and nonsmok- ers. These difl'ercnces are substantial: for smokers; the life expectancies at birth are 69.7 )rars in men and 75.6 tears in women; for nonsmokers, the life expect- ancies are 77.0 and 81.6 vtars (these life-table esti• mates agree %'rn' w'rll %tith the empirica) 'findings of Doll et al!"). This means that many more nonsmok- ers than smokers live to old age. At age 70, 78 percent of male nonsmokers are still alit'e, as compared with only 57 percent of smokers (among women, the fig- ures are 86 percent and 75 percent); at age 80, men's Volume 337 Number 15 - 1053

THE HEALTH CARE COSTS OF SMOKINO . - considered equal for smokers and nonsmokers, but ~ the nonsmoking population lives longer and there- fore incurs more costs due to those diseases, partic- ularly in old age, when the costs arc highest. On bal- ance, the total costs for male and female nonsmokers are 7 percent and 4 percent higher, respectively, than for a mixed population, whereas for smokers the to- tal costs are 7 percent and 11 percent lower. Table 3 also shows that changing the assumptions about the excess risk associated with smoking-relat- ed diseases by as much as 50 percent in either direc- tion does not change the conclusion, except in the case of stroke. The age-related increase in incidence is steepest for stroke, and there is also an age-related increase for stroke in the cost per case; therefore the health care costs associated with stroke are the most sensitive to changes in Hfe expectancy. Because of the costs of other diseases, the popu- lation of nonsmokers has higher health earc costs, partly because these costs increase with age. -Ib test the sensitivity of the analysis to this age-related itt- creasc, we recalculated the three life tables, keeping - No discounting -3%digeountrgte ' --SR ditequnt nta --- t0%diuount rate Break-even year with 5% direounlir Years since Smoking Cessation Figure 2. Pertrent Changes in Total Health Care t:osta for the Male Population after Smoking Cessation, as Determined In a Dynamic Analysit, According to the Number o( Years since Cessation, with No Discounting and with Three Discount Rates. The labels show the 'break-iwn' years, when the coet and benefit of the Inter»ntion balance each other. Shorter follow- up times make gmokhg cessation anractive economically, and , longer follow-up makes @ OnaitraetWe. With 10 percent dis- counting, the break-even year is later than 50 years. TA&E 3. HEUTH CARE COSTS FOR THE THREE POPtMT1wS Sn'DIED wnH L1rE TAeLES, AccoRDwc To Srx A.\D DIfEA(E CATECORt, NTIH THE RATIOS OF THE C.06[s FOR SMOKEas AND NOSSR1oIiFS TO TNOSE FOR THE MIxED POFUUTIO\ COYLV%*iNG SOTH. Soc wo DrufE CanoGal^ lxEn POIwAIqu fMU1ERS uwAOC[as SMOImu: Mm6 PptMtqN Norsr6nq: Mpro F6naA1nM . millione of g . eoft ratio (aeroitivky range)t Men Heart . $26 676 371 129U.18-L3s1 0.71(0.IM.0.61) Sttokc 416 390 428 0.94(0.91-0.97) 1.03(1.06-0.98) Lungcarrca 114 211 33 t.BS(1.76-1.87) 0.29(0.44-0.22) OtAernncea COPD - 226 165 254 275 203 ' 23 1.12(1.05-1.18) 1.67(1.72-Ib3) 0.90(0.97-0.83). 0.14 f0.23-0.111 Ag othrr . Total 6,360 7,806 5,463 . 7,270 7.284 . 8,342 0.86 (0.89-0.84) 0.93 (0.95-0.921 l.l S 1 I.09-1.181 1.07 (1 A4-1.09) Women Heart 386 538 330 1.39 (1.24-1.48) 0.g610.93-0.79) Svak 510 571 502 1.12(1.04-L19) 0.9811.01-0.961 Lungeancer 23 70 9 3.10(2A1-3.32) 0.39(0.55-0.31) Oncwrcancers 297 387 264 1.30(1.16-1.42) 0.g9(09a-0.84) COPD 102 254 20 3.48(2.45-2.451 0.20(0.31-0.15) All ocher 9353 . 7,653 10.013 0.12 (0.87-0.79) 1.07 (1.04-1 A9) Total 10.676 9,473 11,138 0.89 (0.92-0.87( . 1.0i U.03-1.05) •COPD denou. chronic obsmictive puhtwnary dwcuc. . . . . . tThc mmiwuy range was nkvlarcd aith the bwer and u(per toundi of the rarc ntiaa in Table 2. A bwer nu ratio ndum the dilkrence bnween mwken and nomnwken In the inciderla, prer aknce, and morWiry Gan mwking•rd.ted disease. Therefare, nonmoken t,.en k.vreaee uf mmk- ing.mlarcd diuau (kading to kr.er uvinge) but rimuh+ncauM, gdn ka in Ore especww~, (kading to lorer added can Oom'odier' dheaer). For most unokinpreb(ed diseases an6 `ahK' disan, lower rate nriot make the diHcrence in cosn una0er. . Volume 337 Number 15 - 7055 Break-even year whh 3% diacountir ..J-----r-----. .. ..: ...... .i...... L .. 15 20 25 30 35 40 iS 50 Braak-ewn year , with no diaaounli

i THE HEALTH CARE COSTS OF SMOKING This study relied on rate ratios from epidemiologic studies to express the differences between smokers and nonsmokers. To the extent that the rate ratios do not describe these differences sufficiently, the results will be affected. For example, the much lower cost for lung cancer among female smokers than among male smokers (Table 3) is hard to explain physiolog- ically. But as long as the smokers have higher rates of lung cancer than the nonsmokers, such shortcom- ings of the data will not affect the overall conclu- siorts. The results of this study illustrate the ambiguities in any economic method of evaluation. Even a well- designed study of this type is marred by inevitable arbitrariness concerning what costs to include, which discount rate to apply, and what duration of follow- up to use. There are differences of opinion - on the discounting of Gfetime costs, for example, and the evaluation of long-term efi'ecL1.33t4 Recent efforts at standardization will remedy some of the arbitrari- ness,2r27 but fundamental problems with the meth- od still remain. Finally, with respect to public health policy, how important are the costs of smoking? Society clearly has an interest in this matter, now that several states are trying to recoup Medicaid expenditures from to- bacco firms and the tobacco companies have agreed to a settlement. Yet w'e believe that in formulating public health policy, whether or not smokers impose a net financial burden ought to be of very limited importance. Public health policy is concerned with health. Smoking is a major health hazard, so the ob- jective of a policy on smoking should be simple and dear: smoking should be discouraged. Since we as a society are clearly willing to spend money on added years oflife and on healthier years, the method of choice in evaluating medical inter- ventions is cost-effectiveness analysis, which ~ields costs per year of life gained. Decision makers then implement the interventions that vield the highest return in health for the budgct?a 4Ve have no doubt that an efl'ective antismoking policy fits the bill. Supptxted by the Dutch Ministry of Health. REFERENCES 1. Ma.•Kenzie TD. 8artecchi CE, Schricr RW, The human cmn of tobacco uss•. N Engl 131cd 1994;330:975-50. 2. Lippian SC. \kawring medical cost and lik erpectancy impacts of ch.nges in cigarette wks. PmMcd I990J9:513-32. P97/10 -044 ~ 3. Manningw'G.KakrEB-4xMwsclPSbxEa.tvasscmunl.Theux- j a of un: do snwken and drinkers pal• the•tr u•av? L1AN 1989;26I:1601-9. ! 4. Leu RE, Schaub T. Docs smoking inaease medical carc cqpenditute? i Suuc Sci 3ted 1983;17:1907•14. i 5. Schnen R. M.dcling muttigmup populations. %'ets' Yor1: Plenum t4eas, 19$8. ~ 8. Barcndregt 11. tan Ourtmarssen G1. san Hout M, san den Bosrh 1M, ~ Busurcux L Coping with multipk morbidip• in a lifs table. S4th lFtpul Stud (in pn:ss). 7. Sotmcux 4 Barcndregt 11, bh<ter t:, Bamcl Gl, s,an der Maas t1. Eai• mating s5nkal mrxbidiq due to hc•hemic hcart dnease and can8ntist heart Gilum: the Emue aise of heart faihue. Am I ILbfic Health 1994:g4:tb8. B. Nicsecn UV, Basendrcgt 11, Bonncua 4 Koudstaal t7. 5ttotc taendt in an aging population. Suukt 1993:24:931•9. B. Eindhatn Cancer Rtgirtrc. Cancer incidence and wnn•al 1975•1987 in southeastern Nethcrlands. Eindhosxn, the ~lethedands: Eindhmien - Gncer Rcgbtrv,1991. . 10. an /Veel C, s•an den Bosih WIHM, ran den Hougen HIM, Smita AIA. tfnelqmem of rapirat9r•r i8ttex in childhood - a longitudinal study in 8eneral practice. I R Co0 Gen Praa 198fi37.404•8. 11. Ossrkdenen n»r doodsoortaak kcRpd an 8erlacht in her Jar t989: scrie 61. Voarburg, the \etheAands: CeneraY Buteau seur de Statistiek, 1989. 12. hant:nlag 1992. The Hague. the Netherhak: Srichting Votluge- zondheid en Rohen,1993.. 13. ean de Mheen M. Gunnin$-Sdtepen U. Dilkrenca berwmYn seudks Ie reported nlath+e tids associated .ith rmoking: an oaenieu•. Public Health Rep 1996:111:/20•6. 14. Koopnanschap M, van Roiien 4 Bonneu: 4 Bansel G, Rutten F nn dsr Mau P Coua of diseases in an intematiaul penptttive. Eur I rLblic Health 1994;4:238-64. 15. Gunning-Schepen U, Ban:ndregt 1/. Van DsY Mass PI. rbpu6tion in• tentntions trassessed. Lancet 19a9;1:479•81. 18. lkummoM MF Stoddan GL Torrancr GW. Mathods for the eco- i nomic e+aluation of health care programmes. OtfunL England; Oifad L'nhcrsity Peas, 1990. 17. Afurrar C1L. Rethinking DALTs. In: Muray C14 Lopez AD. eds 7Ue Blobol burden of discase: a campochensise assessment of mortality and dis• ability fiom d'ueases, injuries, and risk factors in 1990 and projected to 2010. Vd. I of Gkibal burden of discase and inlury serke B9aton: Hanard School of Public Health, 1996:1A6. 18. Do8 R Mo R, wfieatlcy T., Gray R. SutherWd 1. Monalit•r in teb• sion to smoking: 40 )eati obsenatbm on malc Brithh damts. BM1 . 1994e309:901•11. ~ 18. Kkinbaum DG. Kupper L4 Morgenstem H. Epkkmiologic rcaardc i principles apd quantiuthY methods. Belntont. Calif.; Liktime Leauning, 1992. i 20. Shultz 131. Nosutrty TE, Rice DP flttantifiing the diseasc impact of I cigarette smoking rith SLNMEC 11 atftwarc. tLNic Health Rep 1991; j 106:326-33. . . j 21. Htdgmn TA. Cigatrne smoking and ihtime medical expenditura. ~ Milbank Q 1992:70:81•125. 22. %-nsvsi tVK. Cigarette taxatkm and the social cunsryuenca st(smok• ( Ing:1tiU[Ljntt paprr ml. 4891. r'JmbfY3ge. 311N.: NatMlnil Bureau of Lb- ~ nntnic Reuarch, 1994. ~ 23. xrahn M, Gahti A. Dix.mnting in the e.•utuunic a•aluatinn of hcahh care intenenitum. Msd t:an: 1993:31:403-18. 1 24. tVat RR. Discounting the finure: influence of the e.iuxxnic model. `` I Epidcmh>I Gamtuniq• Health 1996:50:239-44. 25. hfasnn I. Drummwwd M. Reporting guidelines fir eaawxnw studies. i Health Ecun 199S,y:85-94. ; 28. Drummond l1F.1c0irrm TD. Guidclina 8s authors and p.Yr «- - ~wven of ecotxenw subtniuions to the RMI. BMI 1996:313:273-$3. ~ 27. (;dd MR. Siegel /E, Russell LB. w'cinstein MC. eds. Cost•arTecthx- : nes in health and medicine. \m Ynrt: thhmf l'nnmin' Prcss, 1996. : 20. Murray Ch t:reuscr I. tVlyng tl'. Gpt-e0eahrneu anahvs and policy ; ch.das: im<stmg in health s)stems. Bull wiuW Health Organ 1994:72: 663-74. il Volume 337 Number 15 - 1057