This Article Abstract Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Permissions Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Lightwood, J. M. Articles by Glantz, S. A. Search for Related Content PubMed PubMed Citation Articles by Lightwood, J. M. Articles by Glantz, S. A. Pubmed/NCBI databases Medline Plus Health Information Heart Attack Smoking Cessation

(Circulation. 1997;96:1089-1096.)
© 1997 American Heart Association, Inc.

Articles

Short-term Economic and Health Benefits of Smoking Cessation

Myocardial Infarction and Stroke

James M. Lightwood, PhD; ; Stanton A. Glantz, PhD

From the Institute for Health Policy Studies, Department of Medicine, University of California, San Francisco.

	Abstract

Top Abstract Introduction Methods Results Discussion References

 
Background Most analyses of the economic benefits of smoking cessation consider long-term effects, which are often not of interest to public and private policy makers. These analyses fail to account for the time course of the short-run cost savings from the rapid decline in risk of acute myocardial infarction (AMI) and stroke.

Methods and Results We estimate the time course of the fall in risk of AMI and stroke after smoking cessation and simulate the impact of a 1% absolute reduction in smoking prevalence on the number of and short-term direct medical costs associated with the prevented AMIs and strokes. In the first year, there would be 924±679 (mean±SD) fewer hospitalizations for AMI and 538±508 for stroke, resulting in an immediate savings of $44±26 million. A 7-year program that reduced smoking prevalence by 1% per year would result in a total of 63 840±15 521 fewer hospitalizations for AMI and 34 261±9133 fewer for stroke, resulting in a total savings of $3.20±0.59 billion in costs, and would prevent 13 100 deaths resulting from AMI that occur before people reach the hospital. Creating a new nonsmoker reduces anticipated medical costs associated with AMI and stroke by $47 in the first year and by $853 during the next 7 years (discounting 2.5% per year).

Conclusions Although primary prevention of smoking among teenagers is important, reducing adult smoking pays more immediate dividends, both in terms of health improvements and cost savings.

Key Words: smoking • prevention • cost-benefit analysis

	Introduction

Top Abstract Introduction Methods Results Discussion References

 
Smoking is the leading preventable cause of death, killing 420 000 smokers¹ and 53 000 nonsmokers^{2 3} annually. The total cost of medical services for smokers amounts to $50 billion annually, with another $50 billion in lost wages due to morbidity and mortality associated with smoking.⁴ Given these huge costs to society, the question arises as to why more resources are not expended in programs to reduce smoking and promote cessation. Tobacco control advocates are often confronted by the fact that policy makers take a short-run approach to decision making in which it is difficult to justify current expenditures to save money in the long run. Most studies provide little guidance for the short run; they focus on the long run.^{4 5 6 7} Actual public and private decisions generally depend on specifically defined costs and benefits over a relatively short time horizon. Examples are a cost-benefit analysis for an HMO or medical care provider with a large annual turnover rate or government funding allocations among competing projects made under severe budget constraints. Investments in reducing smoking prevalence often are not viewed as productive because of the length of time it takes to reduce treatment costs by preventing cancers and chronic lung disease.⁸ In contrast to cancer and emphysema, the impacts of smoking cessation accrue rapidly when heart disease and stroke are considered.⁸ The excess risk of a myocardial infarction or stroke falls by 50% within the first 2 years after stopping smoking. Treatment for these heart attacks and strokes is expensive, and their prevention provides immediate short-term financial returns for both private and public health insurers.

In the present study, we estimate the short-run absolute reduction in the number of hospitalizations due to AMI and stroke avoided by smoking cessation and the resulting savings in direct medical expenditures and short-run rehabilitation. We do not include the indirect costs of treatment or lost wages; we only consider the direct medical costs. We estimate the immediate savings (within 1 year) associated with the reduction in the number of heart attacks and strokes in 35- to 64-year-old adults due to a 1% absolute reduction in smoking prevalence (which corresponds to 3% to 4% of smokers quitting) and the cumulative effect of a 7-year annual 1% absolute reduction in prevalence, similar to that produced by California's large Proposition 99 anti-tobacco education program^{9 10 11} (Fig 1).

View larger version (16K): [in this window] [in a new window]   Figure 1. Proposition 99 accelerated the historical decline in smoking prevalence in California by 1% per year. Data source: California Department of Health Services.

	Methods

Top Abstract Introduction Methods Results Discussion References

 
Separate Monte Carlo simulations were conducted for AMI and stroke.

Step 1: Estimation of Fall in RR for Ex-smokers Over Time
Selection of Studies
Studies used to estimate RR(t) after smoking cessation had to meet four criteria. First, they must have reported results as specific clinical events corresponding to ICD-9 diagnosis codes for hospitalization statistics. Studies on heart attacks had to report the RR for AMI rather than all forms of sudden cardiac death or broad categories of nonfatal ischemic heart disease. This requirement excluded several studies^{12 13 14} that included angina and related ischemic heart diseases as end points. Second, the study must have reported age-adjusted RRs. Third, the study must have been either a case-control study that reported RRs adjusted for a reasonable number of cofactors and effect modifiers or a population-based study that could reasonably be applied to the United States. Fourth, the study must either have screened out or adjusted for preexisting cardiovascular and cerebrovascular diseases to reduce the influence of the "quitting sick effect" on the RR estimates and to reduce the possibility of confounding.

Five studies were used for the estimation of RR for AMI: Willett et al¹⁵ (1981), Rosenberg et al¹⁶ (1985), Rosenberg et al¹⁷ (1990), Dobson et al¹⁸ (1991), and Kawachi et al¹⁹ (1994). Each of these studies reports statistics adjusted for age and other risk factors for heart disease, except Dobson et al, which is a population-based study that reports statistics adjusted for age and history of cardiovascular disease.

The RR values for the final open-ended lengths of cessation (eg, 15 years) were excluded from the estimation because they might have large effects on the estimates. One outlying observation from Dobson et al¹⁸ (RR >9 after 9 months' cessation for females) was also excluded because it was based on a very small sample. The omission of this outlier did not noticeably affect the estimates.

Two studies of the fall in risk of stroke after smoking cessation met all the selection criteria outlined above.^{20 21} Other studies met the criteria but either focused on only one type of stroke or did not provide enough information to estimate an RR over time since cessation. Both the studies we used were prospective cohort designs and also adjusted the estimated RRs for confounders and modifiers. Wannamathee et al²⁰ (1995) considered all strokes (ICD-9 codes 430 through 438) for British males. Kawachi et al²¹ (1993) considered codes 430 through 434 and code 436 for females in the United States and Canada. Kawachi et al omitted some ICD-9 codes, such as transient ischemic events, and ill-defined events. However, they included the principal types of events: subarachnoid hemorrhage, intracerebral hemorrhage, and ischemic stroke.

Estimation of Fall in RR(t) After Cessation
The five studies of AMI each used different time intervals of smoking cessation. The RRs were assigned to the midpoint of the reported observation intervals. Meta-analytic pooling methods were not feasible for the estimation of a continuous decline curve because there were seldom two statistics from different studies that estimated RR across the same time interval. We combined all of the reported RRs and estimated a function for the decline in RR from the combined data as a function of time (Fig 2A). Several functional forms were fit to the data; all showed a significant reduction in RR(t). We found that an exponential decay in the ln RR produced the most parsimonious description of the data:

(1)

where RR(t) is the RR of AMI t months after cessation, RR₀ and RR are the estimated RRs for AMI at time 0 (just before smoking cessation) and at steady state, and is the time constant for the fall in risk after cessation. F is an indicator variable set to 0 for males and 1 for females, and RR_0F and RR_F quantify the difference between the parameters for male and female subjects. The difference in between sexes was very small, so it was not modeled.

View larger version (13K): [in this window] [in a new window]   Figure 2. Estimated decline in RR for AMI (A; , male; , female) and stroke (B) over time after smoking cessation.

The regression errors were assumed to be independent within each study, which resulted in a diagonal covariance matrix for the regression errors (). The data were fit to Equation 1 by use of nonlinear weighted least-squares regression with each point weighted by the inverse of its variance. Estimates were calculated by use of the nonlinear regression facility in Statistica release 4.5B²² with the quasi-Newton algorithm, using 0.0001 as the convergence criterion.

Fig 2A and Table 1 show the estimated decline curve for AMI after smoking cessation, together with the uncertainty in the parameter estimates. The t value associated with RR_F is only 1.36 (less than usually considered statistically significant), but this term was retained because it produced residuals that more closely approximated the normal distribution.

View this table: [in this window] [in a new window]   Table 1. Fall in Risk of AMI and Stroke Over Time After Smoking Cessation

Data for the fall in risk of stroke after smoking cessation for both sexes were pooled because there did not appear to be significant differences between male and female RRs, given the data available. The data are shown in Fig 2B. As with AMI, an exponential decay in the ln RR produced the best description of the data, dropping the terms that coded for sex differences (F), ie,

(2)

Fig 2B and Table 1 show the estimated results for stroke. The estimate for for stroke is not statistically significant by conventional standards; however, retaining this parameter is required to predict RR(t). The reduction in stroke risk in ex-smokers is well established,²³ so the actual is certainly not zero.

Simulation of Uncertainty in RR(t)
The regression estimates are used to estimate the fall in RR over time, including a random component reflected in the uncertainty in the parameter estimates in Table 1. A standard linear approximation was used to estimate the probability distribution of ln RR(t) over time:

(3)

where RR(t) is the point estimate of the RR t months after cessation, t is Student's t with degrees of freedom, d(lnRR)/dp is the derivative of ln RR(t) with respect to the vector of parameters in Equations 1 or 2 evaluated at t using the point estimates of the parameters, and cov(p) is the covariance matrix of the parameter estimates. We assume that the hospitalization rate is uniformly distributed across time within each year, so t is set to the midpoint of each year. Estimates for the distribution of ln RR(t) for men and women with AMI and stroke were computed separately.

Step 2: Calculation of the Average Hospitalization Rates for Never-Smokers
AMI hospitalization rates by smoking status are needed to turn the estimated RRs into absolute incidences among current and ex-smokers. We estimated the never-smoker hospitalization rate using published data on smoking prevalence, RRs, and observed AMI hospitalizations for the entire population (including both smokers and nonsmokers). We calculated the never-smoker hospitalization rate for males and females (r_n) separately using

(4)

where r_o is the observed incidence rate for AMI hospitalizations for the population, RR₀ is the sex-specific RR of AMI for current smokers, RR is the sex-specific average RR for all ex-smokers, p_s is the sex-specific proportion of the population 35 to 64 years old who are current smokers, and p_x is the sex-specific proportion of the population 35 to 64 years old who are ex-smokers. The values for p_s and p_x are derived by adjusting the age-specific smoking rates reported by the Centers for Disease Control and Prevention²⁴ to the age distribution of the 1994 US resident population found in the Statistical Abstract of the United States,²⁵ Table 14. The statistics p_s, p_x, and r_o are all assumed to be normally distributed. The AMI hospital discharge rate was used as a proxy for incidence (r_o) of admissions. We used the CDC Wonder computer system to retrieve these data from the 1988 to 1990 National Hospital Discharge Survey. The specification was ICD-9 codes 410.0 through 410.91, all initial hospitalizations for first diagnosis for every discharge status for people aged 35 to 64 years. The regression estimate for the lower limit of the RR for an ex-smoker (RR) was used for the average RR for all ex-smokers. The regression estimate of RR(0)=RR₀ is used to estimate the RR for a current smoker. Both estimates are very close to the pooled average RR for current smokers from the studies.

The parameters of stroke were calculated in the same way. The estimates for proportions of current smokers and ex-smokers were calculated by combining data for both sexes. The hospital discharge rate for stroke was obtained using this value from the National Discharge Survey data for 1988 to 1990 in CDC Wonder for ICD-9 codes 430 through 437.9, all stroke hospitalizations for every discharge status for people aged 35 to 64 years. This rate includes all ICD-9 codes for stroke except 438, late effects of cerebrovascular disease, which is only significant for those more than 65 years of age. The average RRs for all current smokers and ex-smokers were calculated from the regression estimates of RR(0) and RR(), respectively.

Table 2 lists the parameter values used in the Monte Carlo simulations.

View this table: [in this window] [in a new window]   Table 2. Parameters Used in Simulations for Events Prevented After Smoking Cessation

Step 3: Calculation of the Absolute Event Rate and Costs
The sex-specific incidence rate for AMI hospitalizations for an ex-smoker who quit t years ago [r(t)] is

(5)

The absolute number of AMI hospitalizations avoided [h_A(s)] in year s for a cohort of individuals who have quit smoking t years ago is simply

(6)

where q is the proportion of quitters in the cohort, N_cs is the number of individuals in the cohort of current smokers remaining in year s, N_qs is the number of individuals in the cohort of quitters remaining in year s, and r(t) is the rate of hospitalization given by Equation 5.

Smokers and ex-smokers who have never had an event are assumed to have the annual survival probability of the average current smoker aged 35 to 64 years. We used sex-specific probabilities reported by Hodgson.²⁶ There are no estimates of short-run changes in total mortality for ex-smokers; therefore, it was not clear how to adjust the current smoker mortality to reflect the experience of ex-smokers immediately after cessation.

Those people who suffer a hospitalization in year t are assumed to have the average annual survival probability for the first year after a hospitalization. Those who have had a hospitalization before year t are assumed to have the average survival probability for the second and subsequent years after a hospitalization. The rates for AMI were taken from McGovern et al.²⁷ Survival rates after stroke were calculated from age-specific rates reported by Taylor et al²⁸ for all stroke patients younger than 65 years old. The values for male and female cohort sizes are the midyear 1995 resident populations from the US Census Bureau monthly postcensus estimates (obtained on the Internet at http://www.census.gov/population under "additional detail files").

Total direct medical costs represent the number of events multiplied by the mean cost of an AMI in 1995 dollars for the first through third years after an event (Table 2). There are three parts to these amounts. The first part is the average costs associated with initial treatment for an AMI (from Kuntz et al²⁹ ). The second part is the expected cost of major surgical procedures such as angioplasty and coronary artery bypass surgery. Hlatky et al³⁰ reported the average costs of these procedures, and Nelson et al³¹ reported their frequencies in the first year after an AMI. The third part consists of follow-up and rehabilitation costs after the initial hospitalization, which are discussed below.

About half the expected cost of an admission for AMI arises from revascularization procedures. According to Nelson et al,³¹ 58% of the admissions for AMI include a revascularization procedure. Langa and Sussman³² report that between 34% and 45% of HMO and fee-for-service admissions for all ischemic heart disease included a revascularization procedure in California in the late 1980s. The rate of revascularization was increasing during the 1980s, so these data probably underestimate the current revascularization rate. The rate of revascularization reported by Langa and Sussman³² also may be low compared with the rest of the US because of the market share of HMOs in California. The revascularization rate Nelson et al³¹ report seems comparable given the greater severity of AMI versus all ischemic disease and the national market share of HMOs compared with that in California, and it is used for calculation of average cost. The cost of thrombolytic therapy was omitted because of insufficient data.

The third part of the cost consists of annual direct medical follow-up, and rehabilitation costs are included for 3 years after an AMI. The follow-up costs are calculated with the use of data from Hemenway et al³³ on the ratio of annual follow-up resource usage to initial hospitalization admission usage for angina. This formula has been used for cost-effectiveness analysis of cardiac procedures, such as post-AMI angioplasty.²⁹ Documented rehabilitation costs and usage are quite small.^{34 35}

There are three components to the cost of stroke: the direct medical short-term care cost of treatment, rehabilitation costs, and cost of care in nursing facilities.²⁸ Taylor et al²⁸ provide cost data by stroke subtype (subarachnoid hemorrhage, ICD-9 code 430; intracerebral hemorrhage, ICD-9 codes 431 and 432; and ischemic stroke, ICD-9 codes 434 and 436), age group, and race for direct medical and rehabilitation costs for the first 2 years; we used these data to calculate costs for these types of strokes for people aged 35 to 64 years. The estimates were adjusted for the inclusion of transient ischemic events and unclassified strokes with the use of the relative charge data for ICD-9 codes 430 through 436 reported by Holloway et al³⁶ and the relative frequencies of the relevant ICD-9 codes from the National Hospital Discharge Surveys for 1984, 1985, 1987, 1988, and 1990 through 1993.^{37 38 39 40 41 42 43 44} We assumed that the unclassified strokes included in ICD-9 code 437 had the same average cost as the classified strokes and that the cost-to-charge ratio was the same across all stroke subtypes.

The expected costs of rehabilitation in the third and subsequent years after a stroke or AMI were calculated from the results in Taylor et al²⁸ and Adelman⁴⁵ and by allocating the present discounted value of these costs reported by Taylor et al²⁸ to each year after the annual service usage pattern reported by Adelman.⁴⁵ The cost of care in skilled nursing facilities after hospitalization is omitted because only 5% of stroke patients younger than 65 years of age are discharged to this type of institution.⁴⁶

The medical costs used in the present study included the average hospital room and service charges, physician and other health professionals' fees, and ancillary charges. These cost estimates were based on very large numbers of patients, so the uncertainty of estimated mean costs was very low, and we therefore treated them as constants. The cost estimates were derived in a variety of ways by the various studies that we used. They typically included charge data adjusted for Medicare Cost Report cost-to-charge ratios, Medicare allowable charge ratios, and actual average patient and third-party reimbursements for services. These cost data are best interpreted as approximations of what is actually paid for the services, as opposed to the charges generated, or variable resource usage costs as defined by various cost-accounting methods.

Step 4: Simulation of the Number of AMIs and Strokes Avoided
The simulations are designed to estimate the distribution of the reduction of AMI events in the cohort of 35- to 64-year-old quitters as opposed to an identical cohort that continued to smoke. Five thousand trials were generated for each of male AMI, female AMI, and stroke. The results were within 2% across multiple simulations. The simulations were run on Minitab software, release 10.2.⁴⁷

We simulated two patterns of cessation: (1) A one-time, 1% reduction in smoking prevalence (equivalent to 3% to 4% of smokers quitting). Prevalence is reduced by 1% at the beginning of year zero and then remains at the new lower level. (2) A continuing annual reduction in smoking prevalence of 1% per year for 7 years. Absolute smoking prevalence is reduced by 1% at the beginning of year zero, and then is reduced an additional 1% per year. This second case is based on the observed effect of a program like the California Proposition 99 antismoking campaign,⁹ which increased the rate of decline in absolute smoking prevalence by 1% per year over the historical trend (Fig 1).

	Results

Top Abstract Introduction Methods Results Discussion References

 
The results for a one-time, 1% reduction in the absolute smoking prevalence rate appear in Table 3. In the first year, we estimate 924 fewer hospitalizations for AMIs and 538 fewer strokes. (See Table 3 for the associated SDs.) Because 17% of people suffering AMIs die before reaching the hospital,⁴⁸ we estimate that an additional 190 deaths would be prevented. After 7 years, the annual reduction in the incidence of AMIs and strokes approaches a steady state of 3200 hospitalizations for AMI and 1700 for stroke each year. (Approximately 660 deaths from AMI that occur before reaching the hospital would also be prevented.) The medical costs avoided are $44 million in the first year (in 1995 dollars), increasing to $168 million per year at steady state. The cumulative (undiscounted) savings in medical costs avoided amount to $933 million over 7 years.

View this table: [in this window] [in a new window]   Table 3. Effects of Reductions in Smoking Prevalence1

The effect of a program designed to reduce prevalence by 1% per year are much larger because the benefits grow rapidly as the number of people who have quit for at least 2 years increases. The annual reduction in AMI and strokes reaches 3022 and 1684 cases during the second year and increases to 18 356 AMIs and 9729 strokes avoided in the seventh year (Table 3). Over the course of 7 years, such a program would prevent a total of 63 840 hospitalizations for AMI and 34 261 for strokes. Approximately 13 100 myocardial infarction fatalities that occur before the patient reaches the hospital would also be prevented. The cumulative (undiscounted) savings reach $191 million in the second year and $3.2 billion by the seventh year.

We also estimated the discounted present value of short-term medical costs associated with AMI and stroke avoided across various time horizons to estimate the value to an organization such as an HMO of creating a new nonsmoker. The average ex-smoker will reduce these medical costs by $47 in the first year. The discounted present value of this new nonsmoker grows the longer the time horizon that is used (Table 4); over a 7-year period, the expected reduction in direct medical costs for each ex-smoker is $990 (undiscounted) or $853 and $738 (using 2.5% and 5% discount rates, respectively).

View this table: [in this window] [in a new window]   Table 4. Cumulative Expected Savings per Quitter at Time of Cessation in US Dollars

	Discussion

Top Abstract Introduction Methods Results Discussion References

 
This analysis shows that there are substantial short-term benefits to assisting adult smokers in stopping smoking and that these benefits accrue rapidly as one extends the time horizon used to assess the benefits of creating a new nonsmoker. In particular, a national program that would reduce smoking prevalence by 1% per year, similar to California's Proposition 99, would prevent 98 100 hospitalizations for AMIs and strokes (and 1300 deaths of acute myocardial infarction outside the hospital) and eliminate the need to spend $3.2 billion (undiscounted) on the treatment of myocardial infarctions and strokes over a 7-year period.

For California, the program used as the model for these calculations, this effect corresponds to prevention of 12 000 myocardial infarctions and strokes with an associated cost savings of $390 million in direct medical costs. For comparison, during the first 7 years of the Proposition 99 program, $411 million (in 1995 dollars) was spent on antismoking media and community-based programs.⁴⁹ (This amount excludes expenditures on school-based programs, which would not affect adult smoking.) Thus, the avoided short-term medical costs for AMI and stroke alone would almost pay for the entire program.

The data in Table 3 can be used to calculate the savings if all smokers aged 35 to 64 years quit at once. There would be 25 000 AMI and 15 000 stroke hospitalizations prevented in the first year; these effects would grow to 88 000 AMIs and 46 000 strokes prevented per year by year 7. The total annual savings would be much greater: $1.2, $4.4, and $4.6 billion in the first, fifth, and seventh years, respectively. The cumulative undiscounted savings over a 7-year period would be $25.4 billion.

The value of a quitter in terms of AMI and stroke costs avoided varies by sex (Table 4). Women have a much lower risk of AMI than men until after menopause, so the savings from avoided AMIs are much less for women aged 35 to 64 years. At the time of cessation, a female quitter can expect to avoid $376 in undiscounted AMI costs and $350 in undiscounted stroke costs, for a total of $726 over a 7-year period. The cumulative 7-year undiscounted savings for a male are $903 for AMI and $350 for stroke, for a total of $1253.

Our computations do not include the value of the lives saved or the suffering avoided or the many other medical costs associated with smoking. Our estimates were designed to be low because we omitted the indirect costs from lost work and productivity, which may roughly equal the medical costs.⁴ We omitted many other outcomes that are also rapidly affected by smoking cessation, such as the impact of smoking during pregnancy on the risk of having a low-birth-weight infant⁵⁰ and exposure of children to secondhand smoke, which accounts for thousands of hospitalizations and physician office visits annually.^{51 52 53}

The results emphasize the importance of encouraging smoking reduction across a broad range of ages. Although primary prevention is important,^{54 55} much of the cost of smoking in terms of cardiovascular disease is not incurred until early middle age. This study shows that there are quick and substantial benefits to implementing programs that reduce the number of adult smokers. In addition, it is through interventions that reduce adult smoking that one can demonstrate the kind of short-term changes in cigarette consumption that appear necessary to convince policy makers to make the necessary investments in tobacco use prevention.

Our estimate that there is some residual risk for AMI long after cessation contradicts the hypothesis that excess risk for AMI disappears entirely 2 or 3 years after cessation.¹⁸ The existence of both a relatively rapid decline in RR and a small long-term excess risk >1.0 (Fig 2A) is consistent with the view that smoking damages the cardiovascular system in at least two ways. One is a short-term effect produced by agents that have an immediate effect on the circulatory system, probably related to the thrombotic effects of smoking and vasoconstriction caused by nicotine.⁵⁶ This effect is a function of current consumption and would disappear soon after exposure to the relevant toxins ends. The other is a long-term effect mostly determined by cumulative consumption, probably involving an increased rate of arteriosclerosis induced by smoking that may be irreversible.^{13 14 56}

This simulation counts both first and subsequent events. The inclusion of subsequent events is important. The RR for subsequent versus first AMI is between 3 and 10 depending on age group,²⁹ and 15% of admissions for AMI are for subsequent events.⁵⁷ The RR for a subsequent versus a first stroke is between 10 and 20, and 10% to 20% of admissions are for subsequent events for people younger than 65 years old.⁵⁸ It would be preferable to model first and subsequent events separately; however, we could not find data that were detailed enough to do so. Therefore, we assumed that the RR of events for smokers versus nonsmokers was the same for first and subsequent events. This assumption will lead to an underestimate of the actual number of events.

Bradley et al⁵⁹ showed that continuing smokers have a higher incidence of subsequent events than nonsmokers or quitters after an AMI and that they incur more costs. The statistics reported by Bradley et al⁵⁹ do not contain enough detail to calculate the time course of excess direct medical costs associated with smoking. However, they report the figures for cumulative events and costs for 10 years after an initial AMI. These estimates are consistent with the assumption that the cost difference between smokers and ex-smokers after the first AMI is approximately the same as before the first event. We have not found similar data for stroke. If the RRs for excess costs for stroke survivors who continue to smoke are similar to those for AMI, then our assumption results in an underestimate. Our assumption provides a reasonable approximation until more detailed data are available.

Study Limitations
The main limitation of the present study is its treatment of first and subsequent AMI and strokes. We assume that the RR between smokers and ex-smokers does not change after a first event. The omission of stroke almost certainly results in an underestimate of savings, but this effect is small compared with the overall uncertainty associated with the other parameters.

There are some potential limitations in the studies we used to estimate the time course of the reduction in AMI after smoking cessation. Three were retrospective case-control studies, which have tended to report a greater decline in events than other study designs.^{13 19} However, the results of the five selected for use in the present study are much more consistent with each other than in the literature for broader categories of disease and mortality. The study populations for both Willett et al¹⁵ (1965 through 1976) and Kawachi et al¹⁹ (1976 through 1988) are from the same longitudinal study of US and Canadian nurses. However, the sample periods for the two do not overlap significantly, so they may be considered independent samples from the same population for distinct periods. One of the studies¹⁸ reported combined RRs for nonfatal AMI and sudden cardiac death attributable to AMI. Kawachi et al¹⁹ reported these rates separately, and the RR for sudden cardiac death was similar to that for hospitalization. The number of deaths was also a small fraction of the hospitalizations.

We do not account for the people who grow older than 64 after the first year of the program. This failure is not a serious problem. It is known that the RR for AMI and stroke decreases with age, and it is thought that the decline in RR after smoking cessation becomes slower for people aged 65 years. We know of no studies that estimate the time course of RR after smoking cessation for the elderly, but there is evidence that the shape of the decline curve shifts slowly with age. LaCroix et al,⁶⁰ Paganini-Hill and Hsu,⁶¹ and Seeman et al⁶² present some evidence of the decline in RR for older smokers but do not present enough data to estimate a short-run curve. Their evidence indicates a gradual reduction in RR for current smokers and a shift to a more gradual decline curve. However, the shift may not occur until very old age; LaCroix et al⁶⁰ report results from a prospective study of people aged 65 years that indicate a return to never-smoker rates for cardiovascular mortality within 5 years after quitting.

Conclusions
This analysis reveals that a one-time, 1% reduction in absolute prevalence of smoking produces substantial short-run savings, both in terms of events avoided and dollars saved. In the first year, we estimate 924±679 (mean±SD) heart attacks and 538±508 strokes are avoided, resulting in an immediate savings of $44±26 million (in 1995 dollars). A 7-year program that reduced smoking prevalence by 1% per year would result in a total of 63 840±15 521 fewer hospitalizations for heart attacks and 34 366±9261 fewer hospitalizations for strokes, resulting in a total savings of $3.20±0.59 billion in short-term medical costs. In addition, this program would prevent 13 100 deaths from AMI that occur before people reach the hospital. (Data are not available to estimate the number of prehospital stroke deaths.) At an individual level, creating a new nonsmoker will reduce anticipated medical costs associated with myocardial infarction and stroke by $47 in the first year, with a discounted present value of $853 during a 7-year period (using a 2.5% discount rate). These figures justify significant investment in programs designed to reduce adult smoking.

	Selected Abbreviations and Acronyms

 

AMI = acute myocardial infarction HMO = health maintenance organization ICD-9 = International Classification of Diseases, 9th revision RR = relative risk RR(t) = relative risk over time

	Acknowledgments

 
This work was supported by NCI grant CA-61021 (Dr. Glantz) and a Pew health policy fellowship grant (Dr Lightwood).

	Footnotes

 
Reprint requests to Stanton A. Glantz, PhD, Professor of Medicine, Division of Cardiology, University of California, San Francisco, CA 94143-0124.

Received December 31, 1996; revision received April 22, 1997; accepted April 28, 1997.

	References

Top Abstract Introduction Methods Results Discussion References

 

1. Reducing the Health Consequences of Smoking: 25 Years of Progress. Atlanta, Ga: Centers for Disease Control, Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health; 1989. US Dept of Health and Human Services publication CDC 89-8411.
   
2. Glantz SA, Parmley WW. Passive smoking and heart disease: epidemiology, physiology, and biochemistry. Circulation. 1991;83:1-12.[Abstract/Free Full Text]
   
3. Wells AJ. An estimate of adult mortality in the United States from passive smoking. Environ Int. 1988;14:249-265.
   
4. Bartlett J, Miller L, Rice D, Max W. Medical-care expenditures attributable to cigarette smoking: United States, 1993. MMWR Morb Mortal Wkly Rep. 1994;43:469-472.[Medline] [Order article via Infotrieve]
   
5. Ben-Shlomo Y, Smith GD, Shipley M, Marmot MG. What determines mortality risk in male former smokers? Am J Public Health. 1994;84:1235-1242.[Abstract/Free Full Text]
   
6. Ockmann GS, Anderson KW, Kronmal RA, Vlietstra RE. The temporal pattern of reduction of total mortality risk after smoking cessation. Am J Prev Health. 1990;6:251-257.
   
7. Kupersmith J, Holmes-Rovner M, Hogan A, Rovner D, Gardiner J. Cost-effectiveness analysis in heart disease, part II: preventive strategies. Prog Cardiovasc. 1995;37:243-271.
   
8. The Health Benefits of Smoking Cessation: A Report of the Surgeon General, 1990. Atlanta, Ga: Centers for Disease Control, Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health; 1990. US Dept of Health and Human Services.
   
9. Bal DG, Kiser KW, Felten PG, Mozar HN, Niemeyer D. Reducing tobacco consumption in California: development of a statewide anti–tobacco use campaign. JAMA. 1990;264:1570-1574.[Abstract]
   
10. Glantz SA. Changes in cigarette consumption, prices, and tobacco industry revenues associated with California's Proposition 99. Tob Control. 1993;2:311-314.
    
11. Glantz S. The ledger of tobacco control. JAMA. 1996;276:871-872. Letter.[Medline] [Order article via Infotrieve]
    
12. Doll R, Peto R. Mortality in relation to smoking: 20 years observation on male British doctors. BMJ. 1976;2:1525-1536.
    
13. Cook DG, Shaper AG, Pocock SJ, Kussick SJ. Giving up smoking and the risk of heart attack. Lancet. 1986;2:1376-1379.[Medline] [Order article via Infotrieve]
    
14. Cook DG, Shafer AG. Stopping smoking and the risk of ischaemic heart disease. Lancet. 1989;2:895.
    
15. Willett WC, Hennekens CH, Bain C, Rosner B, Speizer FE. Cigarette smoking and non-fatal myocardial infarction in women. Am J Epidemiol. 1981;113:575-582.[Abstract/Free Full Text]
    
16. Rosenberg L, Kaufman DW, Helmrich SP, Shapiro S. The risk of myocardial infarction after quitting smoking in men under 55 years of age. N Engl J Med. 1985;313:1511-1515.[Abstract]
    
17. Rosenberg L, Palmer JR, Shapiro S. Decline in the risk of myocardial infarction among women who stop smoking. N Engl J Med. 1990;322:213-217.[Abstract]
    
18. Dobson A, Alexander H, Heller R, Lloyd D. How soon after quitting smoking does risk of heart attack decline? J Clin Epidemiol. 1991;44:1247-1253.[Medline] [Order article via Infotrieve]
    
19. Kawachi I, Colditz G, Stampfer M, Willett W, Manson J, Rosner B, Speizer F, Hennekens C. Smoking cessation and time course of decreased risks of coronary heart disease in middle-aged women. Arch Intern Med. 1994;154:169-172.[Abstract]
    
20. Wannamathee SG, Shaper G, Whincup P, Walker M. Smoking cessation and the risk of stroke in middle-aged men. JAMA. 1995;274:155-160.[Abstract]
    
21. Kawachi I, Colditz GA, Stampfer MJ, Willett WC, Manson JE, Rosner B, Speizer FE, Hennekens CH. Smoking cessation and decreased risk of stroke in women. JAMA. 1993;269:232-236.[Abstract]
    
22. Statistica 4.1. Tulsa, Okla: Statsoft, Inc; 1991.
    
23. Shinton R, Beevers G. Meta-analysis of relation between cigarette smoking and stroke. BMJ. 1989;298:789-794.
    
24. Centers for Disease Control and Prevention. Cigarette smoking among adults: United States, 1994. MMWR Morb Mortal Wkly Rep. 1994;45:588-590.
    
25. Statistical Abstract of the United States. Washington, DC: US Government Printing Office; 1995.
    
26. Hodgson TA. Cigarette smoking and lifetime medical expenditures. Milbank Q. 1992;70:81-125.[Medline] [Order article via Infotrieve]
    
27. McGovern P, Folsom A, Sprafka J, Burke G, Doliszny K, Demirovic J, Naylor J, Blackburn H. Trends in survival of hospitalized myocardial infarction patients between 1970 and 1985. Circulation. 1992;85:172-179.[Abstract/Free Full Text]
    
28. Taylor T, Davis P, Torner J, Holmes J, Meyer J, Jacobson M. Lifetime cost of stroke in the United States. Stroke. 1996;27:1459-1466.[Abstract/Free Full Text]
    
29. Kuntz K, Tsevat J, Goldman L, Weinstein M. Cost-effectiveness of routine coronary angioplasty after acute myocardial infarction. Circulation. 1996;94:957-965.[Abstract/Free Full Text]
    
30. Hlatky M, Rogers W, Johnstone I, Boothroyd D, Brooks M, Pitt B, Reeder G, Ryan T, Smith H, Whitlow P, Wiens R, Mark D. Medical care costs and quality of life after randomization to coronary angioplasty or coronary bypass surgery. N Engl J Med. 1997;336:92-99.[Abstract/Free Full Text]
    
31. Nelson EC, Greenfield S, Hays RD, Larson C, Leopold B, Batalden PB. Comparing outcomes and charges for patients with acute myocardial infarction in three community hospitals: an approach for assessing `value.' Int J Qual Health Care. 1995;7:95-108.[Abstract/Free Full Text]
    
32. Langa K, Sussman E. The effect of cost-containment policies on rates of coronary revascularization procedures in California. N Engl J Med. 1993;329:1784-1789.[Abstract/Free Full Text]
    
33. Hemenway D, Sherman H, Mudge J, Flately M, Lindsay N, Goldman L. Comparative costs versus symptomatic and employment benefits of medical and surgical treatment of stable angina pectoris. Med Care. 1985;23:133-141.[Medline] [Order article via Infotrieve]
    
34. Oldridge N, Furlong W, Feeny D, Torrence G, Guyatt G, Crowe J, Jones N. Economic evaluation of cardiac rehabilitation soon after acute myocardial infarction. Am J Cardiol. 1993;72:154-161.[Medline] [Order article via Infotrieve]
    
35. Pashkow F. Issues in contemporary cardiac rehabilitation: a historical perspective. J Am Coll Cardiol. 1993;21:822-834.[Abstract]
    
36. Holloway R, Witter DJ, Lawton K, Lipscomb J, Samsa G. Inpatient costs of specific cerebrovascular events at five academic medical centers. Neurology. 1996;46:854-860.[Medline] [Order article via Infotrieve]
    
37. Lawrence L. Detailed Diagnosis and Procedures for Patients Discharged From Short-Stay Hospitals, United States, 1984. Washington, DC: National Centers for Health Statistics; April 1986.
    
38. Pokras R. Detailed Diagnosis and Procedures for Patients Discharged From Short-Stay Hospitals, United States, 1985. Washington, DC: National Centers for Health Statistics; April 1987.
    
39. Graves EJ. Detailed Diagnosis and Procedures, National Hospital Discharge Survey, 1987. Washington, DC: National Centers for Health Statistics; March 1989.
    
40. Graves EJ. Detailed Diagnosis and Procedures, National Hospital Discharge Survey, 1988. Washington, DC: National Centers for Health Statistics; March 1991.
    
41. Graves EJ. Detailed Diagnosis and Procedures, National Hospital Discharge Survey, 1990. Washington, DC: National Centers for Health Statistics; June 1992.
    
42. Graves EJ. Detailed Diagnosis and Procedures, National Hospital Discharge Survey, 1991. Washington, DC: National Centers for Health Statistics; August 1994.
    
43. Graves EJ. Detailed Diagnosis and Procedures, National Hospital Discharge Survey, 1992. Washington, DC: National Centers for Health Statistics; February 1994.
    
44. Graves EJ. Detailed Diagnosis and Procedures, National Hospital Discharge Survey, 1993. Washington, DC: National Centers for Health Statistics; October 1995.
    
45. Adelman SM. The National Survey of Stroke: economic impact. Stroke. 1981;12(suppl):I-69-I-87.
    
46. Dombovy M, Bashford J, Whisnant J, Bergstralh E. Disability and use of rehabilitation services following stroke in Rochester, Minnesota 1975-1979. Stroke. 1987;18:830-836.[Abstract/Free Full Text]
    
47. Minitab. Release 10.2. State College, Pa: Minitab, Inc; 1996.
    
48. American Heart Association. Heart and Stroke Facts: 1996 Statistical Supplement. Dallas, Tex: American Heart Association; 1995.
    
49. Monardi F, Balbach E, Aguinaga S, Glantz S. Shifting Allegiances: Tobacco Industry Political Expenditures in California, January 1995-March 1996. San Francisco, Calif: UCSF Institute for Health Policy Studies; April 1996.
    
50. DiFranza J, Lew R. Effect of maternal cigarette smoking on pregnancy complications and sudden infant death syndrome. J Fam Pract. 1995;40:385-394.[Medline] [Order article via Infotrieve]
    
51. Respiratory Health Effects of Passive Smoking: Lung Cancer and Other Disorders. Washington, DC: US Environmental Protection Agency; 1992.
    
52. DiFranza J, Lew R. Morbidity and mortality in children associated with the use of tobacco products by other people. Pediatrics. 1996;97:560-568.[Abstract/Free Full Text]
    
53. Office of Environmental Health Hazard Assessment. Health Effects of Exposure to Environmental Tobacco Smoke. Berkeley, Calif: California Environmental Protection Agency; February 1996.
    
54. Institute of Medicine. Growing Up Tobacco Free: Preventing Nicotine Addiction in Children and Youths. Washington, DC: National Academy Press; 1994.
    
55. Preventing Tobacco Use Among Young People: A Report of the Surgeon General. Atlanta, Ga: Centers for Disease Control and Prevention; 1994. US Dept of Health and Human Services.
    
56. Benowitz N, Gourlay S. Cardiovascular toxicity of nicotine: implications for nicotine replacement therapy. J Am Coll Cardiol. 1997;29:1422-1431.[Abstract]
    
57. Johansson S, Nordin P, Wilhelmsen L, Tibblin G, Johansson BW, Hanson O, Ahlmark G, Jacobsson S, Michaelis E, Gillnas T. Geographical variation and time trends in the attack rate of coronary disease in five Swedish cities. J Intern Med. 1992;231:511-520.[Medline] [Order article via Infotrieve]
    
58. Robins M, Baum HM. The National Survey of Stroke: incidence. Stroke. 1981:12(suppl):I-45-I-57.
    
59. Bradley J, Rogers W, Fisher LD, Gersh BJ, Coggin CJ, Myers WO. Effects of smoking on survival and morbidity in patients randomized to medical and surgical therapy in the Coronary Artery Surgery Study (CASS): 10-year follow-up. J Am Coll Cardiol. 1992;20:287-294.[Abstract]
    
60. LaCroix AZ, Lang J, Scherr P, Wallace RB, Cornoni-Huntley J, Berkman L, Curb D, Evans D, Hennekens C. Smoking and mortality among older men and women in three communities. N Engl J Med. 1991;324:1619-1625.[Abstract]
    
61. Paganini-Hill A, Hsu G. Smoking and mortality among residents of a California retirement community. Am J Public Health. 1994;84:992-995.[Abstract/Free Full Text]
    
62. Seeman T, de Leon CM, Berkman L, Ostfeld A. Risk factors for coronary heart disease among older men and women: a prospective study of community-dwelling elderly. Am J Epidemiol. 1993;138:1037-1049.[Abstract/Free Full Text]
    

This article has been cited by other articles:

				B. Haglund, M. Eliasson, M. Stenbeck, and M. Rosen Is moist snuff use associated with excess risk of IHD or stroke? A longitudinal follow-up of snuff users in Sweden Scand J Public Health, December 1, 2007; 35(6): 618 - 622. [Abstract] [PDF]

				E. K. Tong and S. A. Glantz Tobacco Industry Efforts Undermining Evidence Linking Secondhand Smoke With Cardiovascular Disease Circulation, October 16, 2007; 116(16): 1845 - 1854. [Abstract] [Full Text] [PDF]

				M. J. Stark, K. Rohde, J. E. Maher, B. A. Pizacani, C. W. Dent, R. Bard, S. G. Carmella, A. R. Benoit, N. M. Thomson, and S. S. Hecht The Impact of Clean Indoor Air Exemptions and Preemption Policies on the Prevalence of a Tobacco-Specific Lung Carcinogen Among Nonsmoking Bar and Restaurant Workers Am J Public Health, August 1, 2007; 97(8): 1457 - 1463. [Abstract] [Full Text] [PDF]

				D. L. Brown, B. Boden-Albala, K. M. Langa, L. D. Lisabeth, M. Fair, M. A. Smith, R. L. Sacco, and L. B. Morgenstern Projected costs of ischemic stroke in the United States Neurology, October 24, 2006; 67(8): 1390 - 1395. [Abstract] [Full Text] [PDF]

				F. Barone-Adesi, L. Vizzini, F. Merletti, and L. Richiardi Short-term effects of Italian smoking regulation on rates of hospital admission for acute myocardial infarction Eur. Heart J., October 2, 2006; 27(20): 2468 - 2472. [Abstract] [Full Text] [PDF]

				R. L. Sacco, R. Adams, G. Albers, M. J. Alberts, O. Benavente, K. Furie, L. B. Goldstein, P. Gorelick, J. Halperin, R. Harbaugh, et al. Guidelines for Prevention of Stroke in Patients With Ischemic Stroke or Transient Ischemic Attack: A Statement for Healthcare Professionals From the American Heart Association/American Stroke Association Council on Stroke: Co-Sponsored by the Council on Cardiovascular Radiology and Intervention: The American Academy of Neurology affirms the value of this guideline. Circulation, March 14, 2006; 113(10): e409 - e449. [Abstract] [Full Text] [PDF]

				R. L. Sacco, R. Adams, G. Albers, M. J. Alberts, O. Benavente, K. Furie, L. B. Goldstein, P. Gorelick, J. Halperin, R. Harbaugh, et al. Guidelines for Prevention of Stroke in Patients With Ischemic Stroke or Transient Ischemic Attack: A Statement for Healthcare Professionals From the American Heart Association/American Stroke Association Council on Stroke: Co-Sponsored by the Council on Cardiovascular Radiology and Intervention: The American Academy of Neurology affirms the value of this guideline. Stroke, February 1, 2006; 37(2): 577 - 617. [Abstract] [Full Text] [PDF]

				P. M. Johansson, P. E. Tillgren, K. A. Guldbrandsson, and L. A. Lindholm A model for cost-effectiveness analyses of smoking cessation interventions applied to a Quit-and-Win contest for mothers of small children Scand J Public Health, October 1, 2005; 33(5): 343 - 352. [Abstract] [PDF]

				J. Barnoya, S. A. Bialous, and S. A. Glantz Effective Interventions to Reduce Smoking-Induced Heart Disease Around the World: Time to Act Circulation, July 26, 2005; 112(4): 456 - 458. [Full Text] [PDF]

				B. S. Coller Leukocytosis and Ischemic Vascular Disease Morbidity and Mortality: Is It Time to Intervene? Arterioscler. Thromb. Vasc. Biol., April 1, 2005; 25(4): 658 - 670. [Abstract] [Full Text] [PDF]

				H. Morita, H. Ikeda, N. Haramaki, H. Eguchi, and T. Imaizumi Only two-week smoking cessation improves platelet aggregability and intraplatelet redox imbalance of long-term smokers J. Am. Coll. Cardiol., February 15, 2005; 45(4): 589 - 594. [Abstract] [Full Text] [PDF]

Th. Dietrich and K. Hoffmann A Comprehensive Index for the Modeling of Smoking History in Periodontal Research J. Dent. Res., November 1, 2004; 83(11): 859 - 863. <