Tobacco Documents Online

OTA TESTIMON Y Statement of Roger Herdman, M.D. Acting Director Office of Technology Assessment and Maria Hewitt, Dr.P.H. Senior Analyst, Health Program Office of Technology Assessment and Mary Laschober, M.S. Analyst, Health Program Office of Technology Assessment On Smoking-Related Deaths and Financial Costs: Office of Technology Assessment Estimates for 1990 Before the Senate Special Committee on Aging Hearing On Preventive Health: An Ounce of Prevention Saves a Pound of Cure May 6, 1993 W O~ TA~ Congress d the United States ~~'~j` J/' Office of Tecfmobgy Assessment Washington, DC 20510-8025

SMOKING-RELATED DEATHS AND FINANCIAL COSTS: OFFICE OF TECHNOLOGY ASSESSMENT ESTIMATES FOR 1990 Cigarette smoking causes cancer, respiratory disease, and circulatory system disease, all conditions that contribute greatly to disability and death in the United States. In 1991, an estimated 46 million adults in the United States (26 percent) were current smokers and for the first time in nearly two decades smoking prevalence did not decline (MMWR, April 2, 1993;USDHHS, CDC, CCDPHP, 1989). Until many more U.S. residents stop or curb their smoking, smoking will continue to be the largest source of preventable death and disability and will burden the health care system with avoidable health care costs. At the request of the Senate Special Committee on Aging, OTA has assessed the extent of smoking-related deaths and overall financial costs for 1990 and has developed estimates of the smoking-related health care costs borne by the Federal government through the Medicare, Medicaid, and other government-financed programs. These estimates update earlier ones published bv OTA in 1985 (OTA, 1985). OTA Smoking-Attributable Mortality Estimates OTA estimates that smoking-related illness accounted for nearly one in five deaths in 1990, killing as many as 417,000 U.S. residents (table 1).'•Z These smoking-related deaths far exceed the combined number of deaths from 1 OTA's mortality estimate excludes those dying as a consequence of smokeless tobacco and passive smoking. In 1988, an estimated 3,825 U.S. residents died from passive smoking (MMWR, February 1, 1991) and subsequent estimates of deaths attributable to passive smoking have been higher (Steenland, K., 1992). 2 The number of smoking-attributable deaths has declined since 1988 (i.e., from an estimated 434,000 in 1988) primarily because of a general decline in cardiovascular deaths. Smoking-attributable cancer deaths have increased since 1988 (MMWR, February 1, 1991). 1

AIDS, automobile and other accidents, homicide, and suicide (173,000 deaths). In 1990, more than one-fourth of cancer deaths, nearly one-fifth of cardiovascular disease deaths, and one-half of respiratory disease deaths were attributable to smoking (table 1).3 The smoking-related mortality burden falls disproportionately on young-to-middle aged adults. More than one- quarter of all deaths among those age 35 to 64 are smoking-related (table 3). Because many deaths occur at relatively young ages, there are many years of potential life lost due to smoking. Each smoker who died as a consequence of his or her smoking would have, on average, lived at least 15 additional years had they not smoked.4 For the population at large, this premature mortality translates into more than 6 million years of potential life lost (table 4).= OTA's 1990 estimates of the number of smoking-attributable deaths and vears of potential life lost relied on three sources of information: .data on 1990 smoking prevalence (current, former, and never smokers) from the 1990 National Health Interview Survey (table 5) (CDCP, OSH. special tabulation, 1993); .numbers of 1990 deaths (by age, sex, and cause) as reported by the National Center for Health Statistics (USDHHS, PHS, NCHS, Arthur Horn, personal communications, April, 1993);6 and 3 Smoking-attributable deaths by more specific cause-of-death breakdowns are shown in table 2. 4 This assumes that individuals dying from smoking-related causes would have experienced the life expectancy of the total population (i.e., smokers and non-smokers) had they not died prematurely. S If only years of potential life lost until age 65 are considered, smoking- attributable deaths account for over 1 million years of potential life lost (table 4). This estimate assumes that each individual who died would have lived to age 65 had they not smoked. 6 Only deaths due to causes that have been causally linked to smoking were considered. 2

.estimates of the relative risks of dying from smoking-related causes of death, by smoking status, from an American Cancer Society health studv conducted in the early 1980s (table 6). As part of this study, a large group of middle-aged individuals was identified, their smoking status was recorded, and over the next 6 years the risk of death of current and former smokers relative to never smokers was assessed bv cause of death (USDHHS, PHS, CDC, Report of the Surgeon General, 1989).' OTA relied on a computer program called SAMMEC (Smoking Attributable Mortality, Morbidity, and Economic Costs), designed and distributed by the Centers for Disease Control and Prevention's Office on Smoking and Health, to estimate smoking-related mortality and economic impacts (USDHHS, PHS, CDC, OSH; Shultz, J.M., et al., 1991). For example, using epidemiologic data from the American Cancer Society and 1990 U.S. smoking prevalence rates, the SAMMEC program calculates that 91 percent of 1990 trachea, bronchus, and lung cancer deaths (ICD-9 code 162) among males 35 to 64 were caused by smoking (i.e., 28,173 of 31,076 respiratory cancer deaths were caused by smoking). Estimates of smoking-related years of potential life lost rely on estimates of the number of years that would have been lived had the smoking-related premature death not occurred. Estimates of years of potential life lost can be made to age 65 or to average life expectancy. 7 OTA mortality estimates are based on current and former smokers' risks of dying of smoking-related causes relative to never smokers. The relative risk estimates are based on 4 years of follow up (1980-84) of participants of the American Cancer Society Cancer Prevention Study (CPS-II) (USDHHS, PHS, CDC, Report of the Surgeon General, 1989). 3

OTA Smokin¢-Attributable Financial Cost Estimates The greatest "costs" of smoking are immeasurable insofar as they are related to dying prematurely and living with debilitating smoking-related chronic illness with attendant poor quality of life. Measuring the financial costs associated with smoking is an inexact science, but generally three cost components are included: the direct cost of providing personal health care services to those with smoking-related diseases; .the indirect morbidity costs associated with lost earnings from work or housekeeping because of smoking-related illness; and .the indirect mortality costs related to the loss of future earnings from premature death. OTA estimates the total financial cost of smoking in 1990 to be $68.0 billion or $2.59 per pack of cigarettes sold in the United States. The total cost of $68.0 billion includes $20.8 billion in direct health care costs, $6.9 billion in indirect morbidity costs, and $40.3 billion in indirect mortality costs (figure 1) (table 7).8 The total 1990 cost of smoking per smoker is $1,078, and per capita is $272 (table 8). A technical Appendix (attached) provides details on how direct and indirect smoking-related costs are calculated. Direct costs-Direct costs are measured as the expenditures for preventing, detecting, diagnosing, and treating smoking-related diseases and medical conditions (Rice, D.P., et al., 1986). In 1990, the United States 8 The indirect mortality estimate of $40.3 billion is based on a 4 percent rate to discount future lifetime earnings and excludes deaths of persons under age 35. Indirect mortality costs using a 2 and 6 percent discount rate are estimated at $49.4 and $36.6 billion, respectively. If smoking-related deaths of persons under age 35 are included, 1990 indirect mortality costs are estimated to be $41.9 billion (at a 4 percent discount rate). 4

spent an estimated $20.8 billion on health care for smoking-related diseases. representing 3.5 percent of total U.S. 1990 personal health care expenditures. This amounts to about $329 per smoker, $83 per capita, and 79 cents for each pack of cigarettes sold in the United States in 1990 (table 8). OTA estimates that in 1990, Federal, state, and local governments together funded approximately 43 percent, or $8.9 billion, of smoking- attributable direct costs. The 1990 Federal government share was an estimated $6.3 billion or about 24 cents for each pack of cigarettes sold (table 9). Estimated Medicare costs were $3.5 billion, Medicaid costs were $2.7 billion, and spending for other government-funded health programs was $2.7 billion in 1990 (table 9).9 Total government smoking-related direct costs were fairly evenly split between the population under age 65 ($4.5 billion) and the population age 65 and over ($4.3 billion) (table 9). OTA estimated the direct costs of smoking-related illness using the SAMMEC program (US DHHS, SAMMEC;Shultz, J.M., et al., 1991). Data used to estimate direct costs include: 1990 national estimates of personal health care spending broken down by hospital services, physician services, nursing homes, medications and other medical nondurables, and other professional services as published by the Health Care Financing Administration (Lazenby, H.C., ec al., 1992); 9 Other Federal government smoking-attributable direct medical expenditures include those of the following programs and agencies: Workers' Compensation; Department of Defense; Maternal and Child Health; Vocational Rehabilitation; Alcohol, Drug Abuse, and Mental Health Administration; Indian Health Service; and miscellaneous general hospital and medical programs. Other State and local expenditures include those of the Temporary Disability Program, Workers' Compensation, General Assistance, Maternal and Child Health, Vocational Rehabilitation, hospital subsidies, and school health (Levit, K.R., et al, 1991; USDHHS, HCFA, ORD, 1990; Waldo, D.R., et al., 1989). 5

.estimates of the proportion of personal health care expenditures used for the treatment of cancer, cardiovascular disease, and respiratory disease, derived from a 1980 study of health expenditures by disease category (Hodgson, T. and Kopstein, A.N., 1984): and .estimates of utilization of short-stay hospital days and physician visits by smoking status for cancer, cardiovascular disease, and respiratory disease by adults 35 and older, from the 1987 and 1989 National Health Interview surveys. These relative rates of service use were applied to the 1990 personal health care spending estimates for cancer, cardiovascular disease, and respiratory disease to obtain smoking-attributable costs (Schultz, J.M., et al., 1991). Indirect morbidity costs-Smoking-related disease results in productivity losses to the economy through lost time at work (e.g., sick leave) and lost housekeeping services by homemakers. OTA estimates indirect morbidity costs at $6.9 billion or $109 per smoker, $28 per capita, and 26 cents per pack of cigarettes sold in 1990 (table 8).l0 OTA's estimate for smoking-related indirect morbidity costs relies on the SAMMEC program (USDHHS, SAMMEC;Shultz, J.M., et al., 1991). Data used in this estimate include: .1990 population daily earnings;" .average disability days per year for people with smoking-related diseases (work-loss days for employed persons or bed disability days for persons keeping house) estimated from the 1987 National Health Interview Survey; and 10 Methods used to calculate population daily earnings in the SAMMEC program likely overestimate indirect morbidity costs (see technical appendix). 11 Population daily earnings are computed using U.S. population estimates, labor force participation rates, mean annual income, and imputed values for housekeeping services for unemployed and employed men and women (Rice, D.P., et al., 1986). 6

.rates of work-loss and bed-disability days among people with smoking- related diseases for current and former smokers relative to never smokers estimated from the 1987 National Health Interview Survey (Shultz, J.M., et al., 1991). Relative rates of work-loss and bed-disability days of smokers to never smokers (by age and sex) were applied to average disability days to estimate smoking-related lost productivity days for people with cancer, cardiovascular disease, and respiratory disease. These, in turn, were multiplied by age- and sex-specific population daily earnings data. Indirect mortality costs-The foregone earnings of those dying prematurely in 1990 from smoking-related causes amount to $40.3 billion or $639 per smoker, $162 per capita, and $1.54 per pack of cigarettes sold in 1990 (table 8).12 Using the SAMMEC program, OTA calculated indirect mortality costs using two sources of data: the number of smoking-related deaths by age and sex; and .the value of future earnings according to age at death (by sex), discounted by 4 percent to 1990 present-valued dollars.13 Differences Between OTA's 1985 and 1990 Estimates of SmokinQ-Related Financial Costs In 1985, OTA estimated that between $12 to $35 billion were spent to provide medical care to those with smoking-related illness and $39 tc $96 billion were lost in terms of productivity (combined indirect morbidity and mortality costs) because of smoking-related sickness and early death. The middle estimates of these costs sum to $65 billion or $2.17 per pack of 12 The indirect mortality estimate excludes those dying before age 35. 13 Indirect mortality costs discounted by 2 and 6 percent are estimated at $49.4 and $36.6 billion, respectively. 7

cigarettes sold in the United States in 1985. The 1985 OTA cost estimates are surprisingly similar to those made in 1990 ($21 billion for direct and $47 billion for indirect costs). The methods used by OTA in 1985 differ from those used in the current 1990 estimate, so the two estimates are not strictlv comparable. The most important difference is that for the 1985 estimate, in the absence of other data, OTA used smoking-attributable mortality data to estimate smoking-related direct health care costs (table 10). For example, 1985 OTA estimated that approximately one-half of cancer deaths among males under age 65 were attributable to smoking and so assumed that this fraction in of expenditures for cancer care was due to smoking. The service utilization data from the National Health Interview Survey indicate that this estimate was too high. Since OTA published its 1985 estimates, more precise methods have been developed to estimate the portion of disease-specific expenditures attributable to smoking. For example, instead of mortality data, the SAMMEC program uses information on health care service use of smokers (current and former) versus never smokers as reported by those with cancer, cardiovascular disease, or respiratory disease in the National Health Interview Survey. If OTA had used these data, the 1985 estimate for smoking-attributable direct costs would have been lower. The Centers for Disease Control and Prevention is further refining methods used to estimate smoking-related costs and will publish a 1990 estimate by the fall of 1993 (Nelson, D., personal communication, April 1993). The improved direct cost estimation will relv on analvses, bv smoking status, of the 1987 National Medical Expenditure Survev (Rice, D.P., personal communication, April 1993) and will adjust for differences in sociodemographic characteristics that exist between smokers and nonsmokers (Novotny, T.E., personal communication, April 28, 1993). 8

Factors Excluded From OTA's Estimate of Smoking-Related Financial Costs Neither the 1985 nor the 1990 OTA estimate of smoking-related financial costs includes all of the effects that smoking has on the economy or on all government programs. Only the mortality toll of smoking and its effects on direct medical care spending and the indirec` costs of lost productivity and lost earnings were estimated. Smoking currently leads to a substantial loss of life and significant health care spending. Reduction or elimination of smoking would improve health and extend longevity, but may not lead to savings in health care costs. In fact, significant reductions in smoking prevalence and the attendant increase in life expectancy could lead to future increases in total medical spending, in Medicare program outlays, and in the budgets of the Social Security and other government programs (Warner, K.E., 1987). OTA has not estimated what these hypothetical effects might be. Others have assessed these "off-setting" costs in their estimates of smoking-related costs (Manning, W.G., et al., 1991). Conclusions OTA estimates that as many as 417,000 United States residents died in 1990 as a consequence of smoking and that the total financial cost of smoking was $68.0 billion or $2.59 per pack of cigarettes sold in the United States in 1990. Reductions in smoking prevalence would lead to marked improvements in health and gains in years of life for thousands currently dying of smoking- related disease (USDHHS, Report of the Surgeon General, 1990). Health education and smoking cessation programs, especially those targeted to children, adolescents, and young adults might lead to large improvements in longevity and thus represent significant ways to improve health and prevent 9

premature death. Other policies that might discourage smoking include raising taxes on tobacco products, enforcing minor-access laws, restricting smoking in public places, and restricting tobacco advertising and promotion (MMWR, April 2, 1993). 10

Figure 1--Components of 1990 Smoking-Related Cost Estimates Direct Costs + Indirect Costs = Total Costs $ 21 Billion (Costs of providing + $ 47 Billion _ $ 68 Billion health care to persons with smoking-related illnesses) Indirect Morbidity Costs $ 7 Billion (Costs of lost productivity for persons disabled by smoking-attributable diseases) Indirect Mortality Costs $ 40 Billion (Estimates of forfeited earnings of those dying premature deaths from smoking-attributable diseases) Source: Office of Technology Assessmentas calculated using the SAMMEC 2.1 program, 1993. VBes494g

Table 1--Total Deaths and Deaths Attributable to Smoking by Cause of Death, United States. 1990 Total Smoking-attributable deaths Cause of death deaths' Number Percent All causes 2,148,463 416,829 19.4% Neoplasms 505,322 148,224 29.3 Cardiovascular disease 916,007 179,436 19.6 Respiratory disease 168,203 84,872 50.5 Perinatal disease 15,237 2,215 14.5 Burns° 4,175 2,082 49.9 'Total neoplasm deaths include ICD-9 codes 140-208, total cardiovascular diseases include ICD-9 codes 390-448, total respiratory diseases include ICD- 9 codes 10-12, 466, 480-87, 490-96, total perinatal conditions include ICD-9 codes 765, 769, 770, 798.0, and total burn deaths include ICD-9 codes E-890- 899. e0ne-half of all burn deaths are assumed to be cigarette-related (DHHS, CDC. Office on Smoking and Health, 1990). SOURCES: Office of Technology Assessment, as calculated using the SAMMEC 2.1 program, 1993; USDHHS, PHS, CDC, NCHS, Advance Report of Final Mortality Statistics. 1990 41(7) Supplement, January 7, 1993.

Table 2--Total Deaths and Deaths Attributable to Smoking by Detailed Cause of Death, United States, 1990 Cause of death All causes Neoplasms Lip, oral cavity, pharynx Esophagus Pancreas Larynx Trachea, lung, bronchus Cervix, uterus Urinary bladder Kidney, other urinary Cardiovascular diseases Rheumatic heart disease Hypertension Ischemic heart disease Pulmonary heart disease Other heart disease Cerebrovascular disease Atherosclerosis Aortic aneurysm Other arterial diseases Respiratory diseases Respiratory TB Pneumonia, influenza Bronchitis, emphysema Asthma Chronic airways obstruction Pediatric diseases Short gestation, low birth weight Respiratory distress syndrome Respiratory conditions of newborn Sudden infant death syndrome Burns Total Smokinz-attributable deaths deaths Number Percent 2,148,463 416,829 19.4% 8,311 6,470 77.8 9,698 7,277 75.0 25,006 6,109 24.4 3,702 2,988 80.7 140,947 116,848 82.9 4,303 1,292 30.0 10,316 4,024 39.0 10,153 3,217 31.7 5,864 922 15.7 32,351 5,436 16.8 487,900 98,707 20.2 11,185 1,977 17.7 186,166 32,342 17.4 142,638 23,231 16.3 18,027 6,408 35.5 16,275 7,271 44.7 8,075 3,143 38.9 1,267 359 28.3 77,574 19,158 24.7 19,196 15.295 79.7 4,284 1,093 25.5 61,556 48,967 79.5 4,013 654 16.3 2.850 464 16.3 2,957 482 16.3 5,417 615 11.4 4,175 2,082 49.9 SOURCES: Office of Technology Assessment, as calculated using the SAMMEC 2.1 program, 1993; USDHHS, PHS, CDC, NCHS, Mortality Branch, Arthur Horn, personal communication, 1993.

Table 3--Total Deaths and Deaths Attributable to Smoking by Age and Sex, United States. 1990 Total Smokinr-attributable deaths deaths Number Percent Total < 1-34 150,542 3,083 2.0% 35-64 454,866 121,275 26.7 >_ 65 1,542,493 292,471 19.0 All ages' 2,148,463 416,829 19.4 Male < 1-34 102,882 1,855 1.8 35-64 286,762 84,804 29.6 > 65 723,370 188,937 26.1 All ages' 1,113,417 275,597 24.8 Female < 1-34 47,660 1,227 2.6 35-64 168,104 36,470 21.7 > 65 819,123 103,534 12.6 All ages' 1,035,046 141,232 13.6 'Age-specific numbers of deaths do not add to the total because of a small number of deaths with unknown age of death. SOURCES: Office of Technology Assessment as calculated using the SAMMEC 2.1 program, 1993; USDHHS, PHS, CDC, NCHS, Advance Report of Final Mortalitv Statistics, 1990 41(7) Supplement, January 7, 1993.

Table 4--Smoking-Attributable Mortality by Age and Cause of Death, United States, 1990 Niunher of deaths Years of nott•nt ial life lost all ages < 35 35-64 65 + to age 65 to life expectancy Cause of death Neoplasms 148,224 Cardiovascular diseases 179,436 Respiratory diseases 84,872 Peritiatal diseases 2,215 0 0 0 2,215 53,139 55,258 12,305 0 95,085 124,179 72,567 0 429,010 495,777 89,321 142,857 2,421,891 2,559,615 1,029,642 165,408 Burns Total 2,082 416,829 868 3,083 573 121,275 641 292,471 51,166 1,208,130 78,057 6,254,612 SOURCE: Office of Technology Assessment, as calculated using the SAMMEC 2.1 program, 1993. 98664949

Table 5--Smoking Prevalence, United States, 1990 Males Females Age Currently Formerly Currently Formerly smoke smoked smoke smoked 35-64 31.6% 35.2% 24.8% 22.7% 65+ 14.6 55.2 11.5 23.2 SOURCE: National Centers for Disease Control and Prevention, Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health, special tabulation of the 1990 National Health Interview Survey, April 1993.

Table 6--Relative Risk of Death for Current and Former Smokers by Cause of Death and Sex, United States (Page 1 of 2) Men Women Cause of death IC[)-9 code Currently smoke Formerly smoked Currently smoke Formerly smoked Neoplasms Lip, oral cavity, pharynx 140-149 27.48 8.80 5.59 2.88 Esophagus 150 7.60 5.83 10.25 3.16 reas P 157 14 2 1 12 33 2 78 1 anc . . . . Laryiix 161 10.48 5,24 17.78 11.88 Trachea, lung, bronchus 162 22.36 9.36 11.94 4.69 Cervix, uterus 180 -- -- 2.14 1.94 Urinary bladder 188 2.86 1.90 2.58 1.85 Kidney, other urinary 189 2.95 1.95 1.41 1.16 Cardiovascular diseases Rheumatic heart disease 390-398 1.85 1.32 1.69 1.16 Hypertension 401-404 1.85 1.32 1.69 1.16 Ischemic heart disease 410-414 35-64 years 2.81 1.75 3.00 1.43 65 and over 1.62 1.29 1.60 1.29 Pulmonary heart disease 415-417 1.85 1.32 1.69 1.16 Other heart disease 420-429 1.85 1.32 1.69 1.16 Cerebrovascular disease 430-438 35-64 years 3.67 1.38 4.80 1.41 65 and over 1.94 1.27 1.47 1.01 Atherosclerosis 440 4.06 2.33 3.00 1.34 Aortic aneurysm 441 4.06 2.33 3.00 1.34 Other arterial diseases 442-448 11.06 2.33 3.00 1.34 Respiratory diseases Respiratory TB 010-012 1.99 1.56 2.18 1.38 Pneumonia, influenza 480-487 1.99 1.56 2.18 1.38 Bronchitis, emphysema 490-492 9.65 8.75 10.47 7.04 Asthma 493 1.99 1.56 2.18 1.38 Chronic airways obstruction 496 9.65 8.75 10.41 7.04 0666G9L9

Table 6--Relative Risk of Death for Current and Former Smokers by Cause of Death and Sex, United States (Page 2 of 2) Men Women Cause of death ICD-9 code Currently smoke Formerly smoked Currently smoke Formerly smoked Pediatric diseases Short gestation, low birth weight 765 1.76 1.76 Respiratory distress syndrome 769 1.76 1.76 Respiratory conditions of newborn 770 1.76 1.76 Sudden infant death syndrome 798.0 1.50 1.50 SOURCE: US DIIIIS, PHS, CDC, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Ilealth, SAMMEC 2.1 Comuuter Software and Docwnenration. T6E6494e

Table 7--Smoking-Attributable Direct and Indirect Financial Costs by Age and Sex, United States. 1990 Uirect costs Indirect morbidity costs lndirect mortality costs' (millions of $) a ge (millions of $) age (millions of $) age 35-64 65 + Total 35-64 65 4 Total 35-64 65 t Total Mcile $11,315 $3,395 $14,710 $3,507 $1,171 $4,678 $25,088 $4,411 $29,499 Female 3,077 2,988 6,065 2,019 187 2,207 8,250 2,548 10,798 Total 14,392 6,383 20,775 5,527 1,358 6,885 33,339 6,959 40,298 dThe indirect mortality cost estimates are based on a 4 percent rate to discount future lifetime earnings and exclude deaths of persons under age 35. SOURCE: Office of Technology Assessment, as calculated using the SAMMEC 2.1 program, 1993. zsES~s~,e

Table 8--Cost of Smoking by Type of Cost and Sex, United States, 1990 (Page 1 of 2) TypW ot cost by sex Cost (millions of $) PercenL distribution Per capitab Per smoker` Per packd 1'otal $67,958 100.0X $272 $1,078 $2.59 Direct cost 20,775 30.6 83 329 .79 ilospital 14,419 69.4 58 229 .55 i'hysician 2,689 12.9 11 43 .10 Nursing home 2,332 11.2 9 37 .09 Medication 1,208 5.8 5 19 .05 Other professional 127 0.6 1 2 .01 Indirect cost 47,183 69.4 189 748 1.80 Morbidity 6,885 14.6 28 109 .26 Mortality' 40,298 85.4 162 639 1.54 Men, total $48,887 100.0% $196 $1,354 $1.86 Direct cost 14,710 30.1 59 407 .56 Hospital 11,533 78.4 46 319 .44 Physician 1,365 9.3 5 38 .05 Nursing home 1,137 7.7 5 31 .04 Medication 597 4.1 2 17 .02 Other professional 78 0.5 0 2 .00 Itidirect cost 34,177 69.9 137 947 1.30 Morbidity 4,678 13.7 19 130 .18 Mortality' 29,499 86.3 118 817 1.12 Women, total $19,071 100.0X $76 $707 $.73 Direct cost 6,065 31.8 24 225 23 ilospital 2,887 47.6 12 107 .11 Physician 1,324 21.8 5 49 .05 Nursing home 1,195 19.7 5 44 .05 Medication 611 10.1 2 23 .02 Other professional 49 0.8 0 2 .00 Indirect cost 13,005 68.2 52 482 .50 Morbidity 2,207 17.0 9 82 .08 Mortality' 10,198 83.0 43 401 .41 C6E6G9Z•8

Table 9--Cost of Smoking by Type of Cost and Sex, United States, 1990 (Page 2 of 2) Nuto: Ntunhers may nul ac1d tn total (Iuw to rounding. figIlres apply tu tht. hopulation age 35 anci over. "Uiscount.ed at 4 percent. t"I'utal Uttited States resident population as of July 1, 1990 (U.S. Bureau of the Census, Current Population Reports, U.S. Population Estimates, by Age, Sex, Race, and Hispanic 0rigin: 1980 to 1991, Table 1, pg. 4). `Smokers include both current and former smokers as of 1990 (smoking prevalence rates: 1990 National Health lnterview Survey). Per smoker estimates for males include only male smokers; estimates for females include female smokers. dTotal United States consumption of cigarettes, 1990 (U.S. Department of Agriculture, Tobacco Situation and Outlook Report, September 1992, Table 1, pg. 4). SOURCE: Office of Technology Assessment, as calculated using the SAMMEC 2.1 program, 1993. V6E6G91-8 only %I

Table 9--Smoking-Related Government Spending for Providing Personal Health Care, 1990 Breakdown of Amount expenditures (millions of $) Share of total Total,government spending $8,878 Level of government Federal 6,257 70% State/Local 2,621 30 Government trogram Medicare 3,478 39 Medicaid 2,678 30 Other' 2,722 30 Age roup 0-64 4,544 51 65 and over 4,334 49 Note: Numbers may not add to total due to rounding. 'Other Federal Government smoking-attributable direct medical expenditures include those of the following programs and agencies: Workers' Compensation: Department of Defense; Maternal and Child Health; Vocational Rehabilitation; Alcohol, Drug Abuse, and Mental Health Administration; Indian Health Service; and miscellaneous general hospital and medical programs. Other State and local expenditures include those of the Temporary Disability Program, Workers' Compensation, General Assistance, Maternal and Child Health, Vocational Rehabilitation, hospital subsidies, and school health. Sources: Levit, K.R., Lazenby, H.C., Cowan, C.A., et al., "National Health Expenditures, 1990," Health Care Financing Review, 13(1):29-54, Fall 1991, Table 12; USDHHS, HCFA, ORD, Program Statistics: Medicare and Medicaid Data Book. 1990, HCFA Pub. No. 03314 (Baltimore, MD:1990), Table 4.23; Waldo, D.R., Sonnefeld, S.T., McKusick, D.R., et al., "Health Expenditures by Age Group, 1977 and 1987," Health Care Financing Review, 10(4):111-120, Summer 1989, Table 3.

Table 10--Comparison of Smoking-Attributable Fractions for Direct Costs, OTA Estimates 1985 vs. OTA Estimates 1990a Service comhottent I)isease category Ilospitals Pltysicians Nursing homes Medic:ations Other professionals Males under aEe 65 Neoplasms .50/.39 .50/.09 .50/.39 .50/.()9 .50/.39 Circulatory diseases .30/.39 .30/.09 .30/.39 .30/.09 .30/.39 Respiratory diseases .23/.39 .23/.09 .23/.39 .23/.09 .23/.39 Males age 65 and over Neoplasms .41/.13 .41/.05 .41/.13 41 05 /. .41 /.13 Circulatory diseases .09/.13 .09/.05 .09/.13 .09/.05 .09/.13 Respiratory diseases .28/.13 .28/.05 .28/.13 .28/.05 .28/.13 Females under age 65 Neoplasms .23/.05 .23/.08 .23/.05 .23/.08 .23/.05 Circulatory diseases .25/.05 .25/.08 .25/.05 .25/.08 .25/.05 Respiratory diseases .25/.05 .25/.08 .25/.05 .25/.08 .25/.05 Females age 65 and over Neoplasms .15/.07 .15/.03 .15/.07 .15/.03 .15/.07 Circulatory diseases .05/.07 .05/.03 .05/.07 .05/.03 .05/.07 Respiratory diseases .19/.07 .19/.03 .19/.07 .19/.03 .19/.07 °The first number in each cell is OTA's 1985 middle estimate; the second number iii each cell is OTA's 1990 estimate. SO(1RCE: Office of Technology Assessment Staff Memorancltim, 1985; OTA, as c•alcttlatecl ttsing tlie SAMMEC 2.1 hroeiam, 1993 9666G9L8

SMOKING-RELATED DEATHS AND FINANCIAL COSTS: OFFICE OF TECHNOLOGY ASSESSMENT ESTIMATES FOR 1990 TECHNICAL APPENDIX The technical appendix provides detailed formulas from the SAMMEC 2.1 program, used for OTA's estimation of the 1990 costs of smoking. The appendix also documents sources of data supplied either within the SAMMEC program or by OTA, and provides examples of the various calculations of smoking-attributable direct health care costs, indirect morbidity costs, and indirect mortality costs. Direct Costs Smoking-attributable direct costs are expenditures for the prevention, detection, diagnosis, and treatment of smoking-related diseases and medical conditions (Rice, D.P., et al., 1986). The direct cost calculations are based on three factors: .1990 personal health expenditures (PHE) disaggregated by type of service:' .the share of each PHE service category accounted for by three smoking- related diseases (neoplasms, circulatory diseases, and respiratory diseases (ncr)); and 1 Categories of spending included in the analysis are hospital services, office-based physician services, nursing home care, drugs and other medical nondurables, and other professional services (those of licensed health practitioners other than physicians and dentists such as private duty nurses, chiropractors, podiatrists, psychologists and optometrists, as well as services delivered in freestanding outpatient clinics). Excluded from the cost calculations are personal health expenditures for dental services, home health care, vision products and other medical durables, and other personal health care. In 1990, these excluded components together accounted for approximately 11 percent of total personal health expenditures.

.the portion of these expenditures attributable to smoking (the smoking- attributable fraction or SAF), which is estimated from diagnosis- specific relative risks and smoking prevalence rates. These three factors are multiplied together to obtain smoking-attributable direct expenditures for the three smoking-related disease categories analyzed. As an example, the specific calculation for 1990 smoking-attributable hospital expenditures for males under age 65 is presented. Smoking-attributable hospital costs - (total hospital expenditures) * (the proportion of hospital expenditures for ncr, by age and sex) * (the SAF, by age and sex) where: total hospital expenditures - total spending on hospital care for all ages and both sexes, ($256 billion in 1990), calculated by HCFA, Office of the Actuary (Lazenby, H.C., et al., 1992, table 1). hospital coefficient for ncr - the share of 1990 total hospital expenditures accounted for by persons with the three smoking-related diseases, by age (under 65, and 65 and over) and sex, according to a 1980 expenditure-by- disease study (Hodgson, T. and Kopstein, A.N., 1984, tables 1 and 4). SAMMEC assumes that the share of hospital expenditures accounted for by the relevant group for the three disease categories in 1990 is the same as in 1980.` The share of 1990 total hospital expenditures accounted for by males under age 65 with ncr is .0943 ($9,474\$100,461). 2 The assumption of a constant share of expenditures under- or over-estimates smoking-attributable hospital costs if hospital expenditures for the three disease categories have increased or decreased faster than total hospital spending, perhaps due to changes in the technology employed to treat these diseases. Appendix-2

SAF - l(P0 + P1(RR1) + P2(RR2)) - 11 \(PO + P1(RR1) + P2(RR2)) .PO, P1, and P2 are the proportions of never, current, and former smokers in the population, respectively (Office Health, personal communication, April 1993).3 on Smoking and Four separate SAFs are calculated, one for each age (35-64, and 65 and over) and sex group: and .RR1 and RR2 are the relative risks, measured in this case by the rates of utilization of short-term hospital days for current and former smokers with ncr, respectively, relative to never smokers with ncr, by age (35-64, and 65 and over) and sex (1987 National Health Interview Survey). For males 35-64, P0 is .3313. P1 is .3164, and P2 is .3523 (i.e., the 1990 proportion of males 35-64 who were never, current, and former smokers, respectivelv). For the same group, RR1 is 1.98 and RR2 is 1.93 (i.e., the 1987 relative rates of short-term hospital days per 100 persons for current and former smokers, respectively, compared to never smokers for males 35-64 with conditions in the combined three disease categories).` Using these values and the SAF formula, the SAF for males age 35-64 with ncr is .3894. 3 Smoking prevalence data provided by the Office on Smoking and Health is based on the 1990 National Health Interview Survey. 4 In the SAMMEC program, when smokers' utilization rates were less than nonsmokers' utilization rates, relative rates were set to 1. The program assumes that smoking cannot depress service use. For example, 1987 NH:S daza show that women smokers over age 65 used 39 percent less hospital days than nonsmokers in that group, but the relative rate was set to 1. Appendix-3

The 1990 smoking-attributable hospital costs for the three smokine- related diseases for males under 65 equals:5 . ($256 billion)*(.0943)*(.3894) - $9.4 billion Comparable direct cost estimates were calculated for males 65 and over. for females under 65, and for females 65 and over. The proportion of physician expenditures included in the smoking- attributable direct costs is calculated similarly to hospital costs, except that relative risks are measured as the rates of physician visits per 100 persons for current and former smokers with ncr relative to never smokers with ncr, by broad age intervals and by sex. The calculation of SAFs for nursing home care and other professional services use the relative rates for hospital services. The SAFs for drug costs use the relative rates of physician visits.° Smoking-attributable direct costs are summed over each categorv of service and each age and sex group to arrive at total direct costs. Indirect Morbidity Costs Smoking-attributable indirect morbidity costs are the estimated costs of lost income and productivity of persons who are unable to work or keep house because of illness and disability caused by smoking-related disease (Rice, D.P., et al., 1986). Estimation of 1990 smoking-attributable indirect morbiditv costs were made by multiplying the following three factors: 5 Note that while the SAFs are calculated for the age group 35-64, since the hospital coefficient applies to all males under 65, the smoking-attributable cost figure also represents spending for all males under age 65. 6 Smoking-attributable cost estimates may be higher or lower depending on the direction and the extent to which utilization rates for these other categories of spending deviate from utilization rates of hospital and physician services. Appendix-4

.population mean daily earnings; .the average number of disability days for persons with the three smoking-related diseases; and .the fraction of these disability days attributable to smoking. The calculations of indirect morbidity costs for males age 35-64 are presented as an example. Smoking-attributable indirect morbidity costs - (estimated daily earnings of the U.S. population, by age and sex)*(average disability days for people with for ncr, by age and sex)*(the SAF, by age and sex) where: population mean daily earnings - population*([LFPR*(mean daily earnings of employed persons) + (the mean daily value of housekeeping services of employed persons)] +[(1-LFPR)*(the mean daily value of housekeeping services of unemployed persons);1 where: population - the total number of people in the midyear 1990 U.S, population, by 5-year age intervals and by sex (e.g., there were 9.9 million males age 35-39 residing in the U.S. in July, 1990 (U.S. Dept. of H&SS, 1993, table 2)).' 7 The estimate of population mean daily earnings should have been made using the population with smoking-related diseases. Instead, the SAMMEC program uses the total U.S. population (by age and sex). This potentially substantiall•; overestimates the indirect morbidity costs of smoking.

LFPR - the 1990 labor force participation rate for each 5-year age interval, by sex (U.S. Dept. of Labor, Jan. 1991, table 3) (e.g., the 1990 LFPR for males 35-39 was 94.9 percent).8 The labor force participation rate is the proportion of the noninstitutional population that is in the labor force. The labor force consists of all employed or unemployed civilians. Persons not in the labor force are those not classified as employed or unemployed; this group includes persons who are retired, those engaged in their own housework, those not working while attending school, those unable to work because of long-term illness, those discouraged from seeking work because of personal or job-market factors, and those who are voluntarily idle (U.S. Dept. of Labor, Nov. 1991).9 The SAMMEC program uses (1-LFPR) for each 5-year age interval in place of the 1990 proportion of people who are not in the labor force and keeping house. The SAMMEC program assumes that anyone who is not in the labor force is keeping house.lo mean dailv earnings of employed persons - the 1990 mean annual income per year-round, full-time worker, by sex and by 5-year age intervals (U.S. Department of Commerce, 1991, table 30). Annual mean income is adjusted upward to include 1990 supplements to wages and salaries, consisting mainly of employer contributions to social insurance, private pensions, and welfare 8 The labor force participation rate applies to the civilian noninstitutional population. Published data exist for the age group 75 and over although SAMMEC requires breakdowns by 5-year age groups up through age 85. As these data are statistically unreliable, they are unavailable from the BLS and must be imputed. 9 The SAMMEC program assumes that both employed and unemployed civilians looking for work have the same average daily earnings. This assumption may lead to an overestimate of lost earnings. 10 Use of (1-LFPR) leads to an overestimate of lost earnings, especially for the older age categories in which many people are no longer in the labor force but may be unable to keep house. Appendix-6

funds (U.S. Department of Commerce, 1991). Annual mean earnings are divided by 250 - the number of working days in a year assuming 10 days, or 2 weeks, of annual leave - to arrive at mean daily earnings per person (e.g., 1990 mean daily earnings for males 35-39 were $171.84 per person).11 mean daily value of housekeeping services of emploved and unemploved persons - the 1990 imputed values of housekeeping services, by 5-year age intervals and by sex, for both employed persons and unemployed persons (Peskin, J., 1984).12 The annual mean value is divided by 365 to arrive at the mean daily value (e.g., the 1990 mean daily value of housekeeping services per employed male age 35-39 was $12.40 and per unemployed male was $21.05). The three earnings figures (mean daily income per employed person, mean dailv value of housekeeping services per employed person, and mean daily value of housekeeping services per unemployed person) are weighted by the relative proportions of the population, by age and sex, who are in and out of the labor force (e.g., the 1990 weighted mean daily income per male 35-39 was $176.00). The resulting figure is multiplied by the relevant population to arrive at population mean daily earnings (e.g., $176*9.9 million - $1,742 million popula`ion mean dailv earnings for males 35-39 in 1990). 11 This represents an underestimate of average daily earnings if the number of working days is less than 250 because people take more than two weeks of paid annual leave. 12 These values are updated annually using the change in average weekly earnings of private nonfarm workers, found in Employment and EarninQs.

Estimated daily earnings for each age interval are summed to arrive at total 1990 population mean daily earnings, by sex and broad age group (e.g., total population mean daily earnings for males 35-64 in 1990 was $7,001 million). average disability days - According to Rice, et al., (1985), the average number of disability days lost per year per person with conditions in the three disease categories is a weighted average of the following three factors: the average number of work-loss days per year for employed persons with ncr; the average number of disability days per year for persons with ncr who keep house; and .the number of people with ncr who are too sick to be employed or keep house but would have done so otherwise (Rice, D.P, Hodgson, T.A., and Kopstein, A.N., 1985). However, the measure of the average number of disability days in the SAMMEC program only includes the average number of work-loss days per 100 employed persons and bed-disability days per 100 females keeping house (combined) with ncr, by sex and age, and excludes the number of persons unable to work or keep house due to disability (1987 NHIS).13•1` The 1987 estimated average work-loss days per male 35-64 is 2.8 days. 13 Average disability days were estimated only for people with ncr, and only for days lost due to smoking-related conditions (i.e., disabilitv days from conditions not related to smoking, such as a broken arm, were excluded from the estimates). 14 The exclusion of disability days for those too sick to work or keep house causes morbidity costs to be underestimated. Appendix-8

Average disability days per person with ncr are multiplied by population mean daily earnings for the relevant population to arrive at the total annual monetary value of lost productivity days (e.g., $1,742 million*2.8 -$4,8i6 million in lost productivity for males 35-39 in 1990, or $7,001 million*2.8 - $19,602 million in lost productivity for the total male population age 35-64). SAF - The same formula for calculating SAFs for direct costs is used to calculate morbidity cost SAFs, except that relative risks for current and former smokers are based on 1987 work-loss days per 100 employed persons and bed-disability days per 100 females keeping house (combined) for the three disease categories, by sex and broad age categories (35-64, and 65 and over) (1987 NHIS). These calculations include only persons 35 or older who reported illness due to any of the three smoking-related diseases. The 1990 smoking prevalence rates are the same as those used to calculate direct cost SAFs. For males age 35-64, the relative risks were 1.46 (current smokers) and 1.21 (former smokers), for an SAF of .1789. The smoking-attributable fraction is applied to the annual value of lost productivity to arrive at the proportion of lost productivity due to smoking- :elated illness (e.g., $19,602 million*.1789 - $3,507 million indirect morbidity costs for males 35-64 in 1990). The 1990 smoking-attributable morbidity costs for the three disease categories for males 35-64 equals:  ($7 billion)*(2.8)*(.1789) - $3.5 billion Appendix-9

Comparable indirect morbidity cost estimates were also calculated for males 65 and over, for females 35-64, and for females 65 and over. Indirect Mortality Costs Smoking-attributable indirect mortality costs are the estimated value lost earnings resulting from premature death due to smoking-related disease of and injury (i.e., burn deaths) (Rice, D.P., et al., 1986). Lost earnings represent the future earnings forfeited from the age at death to the age of average life expectancy. Because money received in future years is worth less than the same amount of money received in the present, these future expected lifetime earnings are discounted to their present value. The estimation of smoking-attributable mortality costs also relies on the multiplication of three factors: the total number of 1990 deaths in the three disease categories; .the proportion of these deaths attributable to smoking; and .the present value of future earnings (PVFE) lost due to smoking- attributable deaths. Mortality cost estimations for males age 35-64, with respiratory tuberculosis (TB) as cause of death, are presented as an example. Smoking-attributable indirect mortality costs - (number of deaths due to ncr) * (SAF) * (PVFE) where: Number of deaths - the number of 1990 deaths for 27 smoking-related causes of death for the ncr disease subcategories. The deaths are by 5-year age intervals, beginning with age 35, by sex. Perinatal smoking-related deaths Appendix-10

and burn deaths are provided for all ages (i.e., they include deaths under age 35) (NCHS, personal correspondence, April 1993). deaths due to TB in 1990 for males 35-39. For example, there were 46 SAF - Mortality cost SAFs are calculated using the same formula as direct cost SAFs, except that relative risks for current and former smokers are calculated as the risk of dying from smoking by current and former smokers relative to never smokers, according to an American Cancer Society study conducted in the early 1980s. The two relative risk rates (one for current smokers and one for former smokers) are available for males 35 and over for all but two disease subcategories.15 The relative risks for each disease subcategory are assumed to apply to both males 35-64 and to males 65 and over (e.g., the relative risk of dying from TB for males who were current smokers in 1990 (1.99) and the relative risk for males who were former smokers as of 1990 (1.56) apply to males in all age groups 35 and over). Cause-specific SAFs for each broad age interval are applied to the cause-specific number of deaths in each 5-year age interval (i.e., the SAMMEC program assumes that the fraction of smoking- attributable deaths in each 5-year age interval for males 35-64 is the same and for males 65 and over is the same: the SAF for TB for males 35-64 was .338 in 1990 and the SAF for males 65 and over was .312). In 1990, there were 16 smoking-attributable TB deaths for males age 35-39 (46*.338). PVFE - The present value of future earnings is based on four sources: .1990 cross-sectional profiles of mean earnings and supplemental income. by age and sex, adjusted for a 1 percent annual increase in labor productivity; 15 For ischemic heart disease and cerebrovascular diseases, avai:able for two age groups (35-64, and 65 and over). relative risks are Appendix-11

1990 cross-sectional profiles of the mean value of household services. by age and sex; .1990 labor force participation rates, by age and sex; and life expectancy tables, by age and sex (Rice, D.P, personal correspondence, 1993). The age at death in each 5-year age interval is calculated as the midpoint of the interval. OTA applied 2 percent, 4 percent, and 6 percent rates in its estimates of the total value of future earnings (e.g., the PVFE for males age 35-39 in 1990, discounted at a 4 percent rate, is $802,679).16 The 1990 smoking-attributable deaths in each 5-year age interval, by sex and disease subcategory, are multiplied by the 1990 present value of future earnings for each age interval, by sex (e.g., the total PVFE in 1990 for TB deaths for males age 35-39 is $12.5 million (16*$802,769)). Total PVFE are summed over age intervals and disease subcategories to arrive at total indirect mortality costs by broad age group, by sex. Since the formula for calculating smoking-attributable indirect mortality costs is more complicated than the other two formulas, the complete data are not presented. Comparable calculations were made for males 65 and over, females 35-64, and females 65 and over. Total Costs of SmokinQ Total 1990 smoking-attributable costs are the sum of direct health care costs, indirect morbidity costs, and indirect mortality costs. OTA's 1990 estimates use a 4 percent discount rate for indirect mortalitv costs and exclude indirect morbidity and mortality costs of people under age 35. 16 A 4 percent rate is the one most commonly used for cost of smoking calculations, and is the primary one used for OTA estimates. Appendix-12

Potential Sources of Under- or Over-Estimates of SmokinR-Attributable Costs OTA has identified several sources of possible over- or under-estimation of smoking-related costs. However, OTA cannot assess the magnitude of over- or under-estimation with currently available data. Direct Costs: .Personal health expenditures for dental services, home health care. vision products and other medical durables, and other personal health care are excluded from the cost calculations, causing an underestimation of the direct costs of smoking. In 1990, spending on these services accounted for approximately 11 percent of total personal health expenditures. .If spending shares for individual service categories of personal health expenditures for the three diseases have increased or decreased faster from 1980 to 1990 than total spending in these categories, smoking-attributable direct costs are under- or over-estimated, respectively. .The assumption that'smoking cannot depress service utilization rates mav lead to an over-estimate of direct costs. Indirect Morbidity Costs: .The estimate of Fopulation mean daily earnings should have been made using the population with smoking-related diseases. Instead, the SAMMEC program uses the total U.S. population (by age and sex). This potentially substantially overestimates the indirect morbidity costs of smoking. .Smoking-attributable morbidity cost estimates may be higher or lower depending on the direction and the extent to which utilization rates of nursing home services, drugs, and other professional services by current and former smokers relative to never smokers deviate rates of hospital and physician services. m from the relative utilization *1 0 V O ~ Appendix-13

The use of labor force participation rates in place of employment rates assumes that unemployed people and employed people have the same daily earnings. This provides a potential overestimate of lost earnings and morbidity costs. The use of (1-LFPR) instead of the proportion of people who are not in the labor force and keeping house overestimates lost earnings and morbidity costs, especially for the older age categories in which many people are no longer in the labor force but may be unable to keep house. .SAMMEC's morbidity cost estimates are underestimated because they exclude lost earnings by people unable to work or keep house due to sickness caused by smoking-related diseases. Indirect Mortality Costs: Since the same set of data used for calculating earnings is applied to both the morbidity and mortality cost estimates, factors that lead to an over- or under-estimate of mean earnings also cause a comparable over- or under- estimate of lifetime earnings and smoking-attribLtable indirect mortality costs. .If future increases in labor productivity, and thus wage rates or earnings, are greater than 1 percent per year, lifetime earnings and indirect mortality costs are underestimated. Appendix-14

SMOKING-RELATED DEATHS AND FINANCIAL COSTS: OFFICE OF TECHNOLOGY ASSESSMENT ESTIMATES FOR 1990 REFERENCES Hodgson, T.A., and Kopstein, A.N., "Health Care Expenditures for Major Diseases in 1980," Health Care Financing Review, 5(4):1-12, Summer 1984. Lazenby, H.C., Levit, K.R., Waldo, D.R., et al., "National Health Accounts: Lessons from the U.S. Experience," Health Care Financing Review, 13(4):89- 103, Summer 1992. Levit, K.R.. Lazenby, H.C., Cowan, C.A., and Letsch, S.W., "National Health Expenditures, 1990," Health Care Financing Review, 13(1): 29-54, Fall 1991. Manning, W.G., Keeler, E.B., Newhouse, J.P., et al., The Costs of Poor Health Habits, (Cambridge, MA: Harvard University Press, 1991). Morbidity and Mortality Weekly Report, "Cigarette Smoking Among Adults--United States, 1991," MMWR 42(12):230-233, April 2, 1993. Morbidity and Mortality Weekly Report, "Smoking-Attributable Mortality and Years of Potential Life Lost -- United States, 1988," MMWR 40(4):62-71, Februarv 1, 1991. Nelson, D.E., Medical Epidemiologist, U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health, personal communication, April 2, 1993. Peskin. J., The Value of Household Work in the 1980s (Washington, DC: U.S. Congressional Budget Office, 1984). Rice. D.P., Professor. Institute for Health and Aging, Department of Social and Behavioral Sciences, Universitv of California, San Francisco, personal communication, April 7, 1993. Rice. D.P., Hodgson, T.A., Sinsheimer, P., et al., "The Economic Costs of the Health Effects of Smoking, 1984," The Milbank Ouarterlv 64(4):489-547, 1986. Rice, D.P., Hodgson, T.A., and Kopstein. A.h.. "The Economic Costs of Illness. A Replication and Update," Health Care Financing Review 7(l):61-80, Fall 1985. Shultz. J.M.. Novotnv. T.E., and Rice. D.P.. "Quantifying the Disease Impact of Cigarette Smoking with SA.MMEC II Software," Public Health Reports 106(3):326-333, May-June 1991. Steenland. K., "Passive Smoking and the Risk of Heart Disease," JAMA 267(l):94-99, Januarv 1, 1992. References-1

U.S. Congress, Office of Technology Assessment, "Smoking-Related Deaths and Financial Costs: Staff Memorandum," September 1985. U.S. Dept. of Commerce, Bureau of the Census, Money Income of Households. Families, and Persons in the United States: 1990, Current Population Reports, Consumer Income, Series P-60, No. 174 (Washington, D.C.: August 1991). U.S. Department of Commerce, Bureau of the Census, U.S. Population Estimates, by Age Sex Race and Hispanic Origin: 1980 to 1991, Current Population Reports, Series P25-1095 (Washington D.C.: 1993). U.S. Department of Health and Human Services, Health Care Financing Administration, Office of Research and Demonstrations, Program Statistics Medicare and Medicaid Data Book, 1990, HCFA Pub. No. 03314 (Baltimore, MD. 1990). U.S. Department of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention, Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health, SAMMEC 2.1: Computer Software & Documentation (Atlanta, GA). U.S. Department of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention, Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health, The Health Benefits of Smoking Cessation: A Report of the Surgeon General. 1990, DHHS Pub. No. (CDC) 90-8416 (Rockville, MD: 1990). U.S. Department of Health and Human Services, Public Health Service, Centers for Disease Control, Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health, Reducing the Health ConseQuences of Smoking, Years of Propress: A Report of the Surgeon General. 1989, DHHS Pub. No. (CDC) 89-8411 (Rockville, MD: 1989). U.S. Department of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention, National Center for Health Statistics, Mortality Branch, Arthur Horn, personal communication, April 12, 1993. I.S. Department of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention, National Center for Health Statistics, "Advance Report of Final Mortality Statistics, 1990," Monthly Vital Statistics Report 41 (7 Supplement), DHHS Pub. No. (PHS) 93-1120 (Hvattsville, MD: 1993). U.S. Department of Labor, Bureau of Labor Statistics, Emplovment and Earnings, January 1991. U.S. Department of Labor, Bureau of Labor Statistics, Monthly Labor Review, November 1991. Waldo, D.R., Sonnefeld, S.T., McKusick, D.R., and Arnett, R.H., "Health Expenditures by Age Group, 1977 and 1987," Health Care Financing Review, 10(4): 111-120, Summer 1989. References-2

Warner, K.E., "Health and Economic Implications of a Tobacco-Free Society," JAMA 258(15):2080-2086, October 16, 1987. References-3