oversight

CDC's April 2002 Report On Smoking: Estimates of Selected Health Consequences of Cigarette Smoking Were Reasonable

Published by the Government Accountability Office on 2003-07-17.

Below is a raw (and likely hideous) rendition of the original report. (PDF)

United States General Accounting Office
Washington, DC 20548



          July 17, 2003

          The Honorable Richard Burr
          House of Representatives

          Subject: CDC’s April 2002 Report On Smoking: Estimates of Selected Health
                   Consequences of Cigarette Smoking Were Reasonable

          Dear Mr. Burr:

          Despite a recent decline in the population that smokes, smoking is considered the
          leading cause of preventable death in this country. According to the Centers for
          Disease Control and Prevention (CDC), over 2 million deaths in the 5-year period
          from 1995 through 1999 were attributable to cigarette smoking. CDC, part of the
          Department of Health and Human Services (HHS), is a primary source of information
          on the health consequences of smoking tobacco. CDC reported its most recent
          estimates of selected health consequences of cigarette smoking in an April 2002 issue
          of its publication Morbidity and Mortality Weekly Report.1 CDC reported that, on
          average, over 440,000 deaths, 5.6 million years of potential life lost, $82 billion in
          mortality-related productivity losses, and $76 billion in medical expenditures were
          attributable to cigarette smoking each year from 1995 through 1999. (See enclosures I
          and II.)

          CDC and others tasked with making such estimates face challenges. They build
          estimates on a set of assumptions and make choices about the data sources and
          methods used, each of which may have limitations that must be weighed against its
          advantages. Policymakers at both the state and federal levels have relied on estimates
          like these in considering bans on smoking in public places, taxes on cigarettes,
          litigation to recoup medical expenditures, and other matters concerning tobacco.
          Thus it is essential that the estimates CDC provides are sound and that their
          limitations are clear. In recognition of this, you asked us to review CDC’s April 2002
          report and determine whether its estimates of selected health consequences of
          cigarette smoking were reasonable. Specifically, we examined CDC’s estimates of
          (1) deaths and years of potential life lost and (2) mortality-related productivity losses
          and medical expenditures attributable to cigarette smoking.



          1
           Centers for Disease Control and Prevention (CDC), “Annual Smoking-Attributable Mortality, Years of
          Potential Life Lost, and Economic Costs – United States, 1995-1999,” Morbidity and Mortality Weekly
          Report, vol. 51, no. 14 (2002): 300-303. Morbidity and Mortality Weekly Report is a CDC publication
          for dissemination of information about the public health issues in which CDC is involved.


                                         GAO-03-942R CDC’s 2002 Report on Health Consequences of Smoking
To determine whether CDC’s estimates were reasonable, we reviewed CDC’s
approach and alternative approaches to developing them. Specifically, we reviewed
CDC’s assumptions, methods, and data sources; the choices CDC made about how to
best estimate the number of deaths, years of potential life lost, productivity losses,
and medical expenditures attributable to cigarette smoking; and CDC’s attempts to
deal with the limitations inherent in analyses of this kind. We examined CDC’s
choices in the context of the alternatives available and determined whether the
alternatives would have resulted in more reasonable estimates. In reviewing CDC’s
approach and the available alternatives, we searched the scientific literature using
the electronic databases MEDLINE and EconLit and reviewed over 200 studies on the
consequences of tobacco and approaches to estimating them. In addition, we
reviewed CDC’s documentation of its methods and interviewed CDC officials
involved in the report about their approach and their rationale for choices made in
deriving these estimates. We conducted our work from December 2002 through July
2003 in accordance with generally accepted government auditing standards.

In summary, CDC’s estimates of the average number of deaths and years of potential
life lost each year due to cigarette smoking were reasonable. The estimates were
based on the increases in deaths from 23 causes that were linked to cigarette
smoking. The linkages of cigarette smoking to increased mortality due to the included
causes, such as lung cancer or cardiovascular disease, had been well established by
the Surgeon General. CDC used the method generally accepted among
epidemiologists for estimating the increased deaths attributable to cigarette smoking.
The data sources CDC used were the best available and included: the largest study of
smoking behavior and health status available for data on the risk of death in smokers
relative to nonsmokers; the National Health Interview Survey (NHIS) of over 97,000
persons for data on the prevalence of smoking; and death certificates compiled from
all states for mortality data. CDC recognized and handled appropriately the
limitations in the data from these sources.

CDC’s estimates of the annual mortality-related productivity losses and medical
expenditures due to cigarette smoking also were reasonable. CDC estimated
productivity losses associated with the years of potential life lost using assumptions
about employment and earnings that are generally accepted among economists, well-
established methods for extrapolating from present earnings to earnings that would
be made in the future, and large federal data sources on earnings. The assumptions
that CDC made and the methods it used to estimate medical expenditures were also
generally accepted among health care economists. CDC relied on the most
comprehensive data available on medical expenditures, the federally sponsored
National Medical Expenditure Survey (NMES) of over 38,000 persons. For both
productivity losses and medical expenditures, CDC recognized and handled
appropriately the limitations in the data.

In its comments on a draft of this report, CDC said that this report, in general,
accurately represents the intent, methods, and decision-making processes of its April
2002 report.



2                         GAO-03-942R CDC’s 2002 Report on Health Consequences of Smoking
Background

The Surgeon General’s first report on smoking and health was published in 1964. This
report was the first of many to describe the links between tobacco smoking and
health. Since then, several federal agencies have issued reports on tobacco and
       2
health. In the last four decades, the Office of the Surgeon General has published
dozens of reports on the health consequences of smoking. CDC’s Office on Smoking
and Health originated as the National Clearinghouse for Smoking and Health in the
Office of the Surgeon General and became part of CDC in 1986. Through this office,
CDC has become a chief source of information on the health consequences of
smoking.

Although the health consequences of cigarette smoking are numerous, CDC’s April
2002 report provided four estimates—number of deaths, years of potential life lost,
mortality-related productivity losses, and annual medical expenditures attributable to
cigarette smoking. The estimate of number of deaths is the foundation for both the
years of potential life lost and mortality-related productivity loss estimates. Number
of deaths and years lost are two different ways of measuring mortality attributable to
cigarette smoking, and mortality-related productivity loss is a way of measuring lives
and years of life lost in economic terms. All three of these estimates are limited to
mortality and do not measure morbidity attributable to cigarette smoking, such as
disability, diminished quality of life, and reduced productivity associated with
diseases linked to cigarette smoking. Unlike the other three estimates, CDC’s
estimate of annual medical expenditures attributable to cigarette smoking includes
the additional medical expenses attributable to cigarette smoking of all smokers in a
given year, not those who died in that year. Thus, CDC’s estimate is of annual medical
expenditures for morbidity attributable to cigarette smoking. CDC labels as economic
costs attributable to cigarette smoking the sum of its estimates of mortality-related
productivity losses and medical expenditures. However, this summary estimate of
economic costs does not include such costs as time lost in the workplace due to sick
leave and disability.

CDC’s Estimates of Number of Deaths and Years of Potential Life Lost Due
to Cigarette Smoking Were Reasonable

CDC’s estimates of an average of over 440,000 premature deaths and 5.6 million years
of potential life lost each year attributable to cigarette smoking were reasonable.
CDC relied on well-established criteria for the causes of death to include and used

2
 See for example, U.S. Department of Health, Education, and Welfare, National Institutes of Health,
National Cancer Institute, and National Heart, Lung, and Blood Institute, Smoking and Health: A
Program to Reduce the Risk of Diseases in Smokers, Status Report (Bethesda, Md.: December 1978);
U.S. Department of Agriculture, Economic Research Service, Tobacco: Situation and Outlook
(Washington, D.C.: April 1995); U.S. Department of the Treasury, The Economic Costs of Smoking in
the U.S. and the Benefits of Comprehensive Tobacco Legislation (Washington, D.C.: March 1998); and
U.S. Department of Health and Human Services, National Institutes of Health, National Cancer
Institute, Strategies to Control Tobacco Use In the United States: A Blueprint for Public Health
Action In the 1990's, Smoking and Tobacco Control Monograph 1 (Bethesda, Md.: December 1991).


3                             GAO-03-942R CDC’s 2002 Report on Health Consequences of Smoking
the standard method for attributing deaths to cigarette smoking. In addition, it used
the best data sources available and recognized and handled appropriately the
limitations in the data on which the estimates are based.

Total Number of Deaths Attributable to Cigarette Smoking

CDC’s estimate of the average total number of deaths attributable to cigarette
smoking annually is the sum of deaths in four categories that reflect differences in
how the estimates are obtained: adult deaths from diseases causally linked to
cigarette smoking, infant deaths from conditions causally linked to maternal cigarette
smoking during pregnancy, adult deaths from diseases causally linked to exposure to
secondhand cigarette smoke, and deaths from residential fires caused by smoking.
(See table 1.) CDC generated the estimates of adult deaths from diseases linked to
cigarette smoking and infant deaths from conditions linked to maternal cigarette
smoking and relied on the estimates of others for secondhand smoke and fire deaths.




4                         GAO-03-942R CDC’s 2002 Report on Health Consequences of Smoking
Table 1: CDC’s Estimates of Average Annual Deaths Attributable to Cigarette Smoking and Years of
Potential Life Lost (1995 to 1999)

                                                       Percentage of     Years of potential    Percentage of
                                               Deaths    total deaths      life lost (YPLL)       total YPLL
    Adult deaths from diseases causally linked to cigarette smokinga
    Cancer of lip, oral cavity, pharynx         5,137             1.2                85,521               1.5
    Cancer of esophagus                         7,893             1.8               120,045               2.1
    Cancer of pancreas                          6,480             1.5                98,593               1.8
    Cancer of larynx                            3,127             0.7                48,616               0.9
    Cancer of trachea, lung, bronchus         124,813            28.2             1,869,786              33.3
    Cancer of cervix uteri                        522             0.1                13,606               0.2
    Cancer of urinary bladder                   4,752             1.1                53,498               1.0
    Cancer of kidney, other urinary             3,035             0.7                46,039               0.8
    Hypertension                                6,060             1.4                87,577               1.6
    Ischemic heart disease                     81,976            18.5             1,172,699              20.9
    Other heart diseases                       29,368             6.6               371,083               6.6
    Cerebrovascular disease                    17,445             3.9               280,728               5.0
    Atherosclerosis                             2,527             0.6                22,802               0.4
    Aortic aneurysm                             9,624             2.2               116,223               2.1
    Other arterial disease                      1,605             0.4                20,894               0.4
    Pneumonia, influenza                       15,576             3.5               156,133               2.8
    Bronchitis, emphysema                      17,696             4.0               216,376               3.9
    Chronic airways obstruction                64,735            14.6               732,189              13.0
    Adult smoker deaths from disease          402,373            91.0             5,512,405              98.1
                                                                                                  b
    Infant deaths from conditions causally linked to maternal cigarette smoking during pregnancy
    Short gestation/low birthweight               402            0.09                30,556              0.54
    Respiratory distress syndrome                 109            0.02                  8,198             0.15
    Other respiratory—newborn                     117            0.03                  8,793             0.16
    Sudden infant death syndrome                  377            0.09                28,677              0.51
    Infant deaths from maternal smoking         1,007            0.23                76,224              1.36
                                                                                             a
    Adult deaths from diseases causally linked to exposure to secondhand cigarette smoke
    Lung cancer                                 3,000             0.7                      -                -
    Ischemic heart disease                     35,053             7.9                      -                -
    Adult deaths from secondhand smoke         38,053             8.6                      -                -
    Deaths from residential fires caused
    by smoking                                    966             0.2                27,756               0.5
    Total deaths attributable to cigarette
    smoking                                   442,398           100.0             5,616,385             100.0

Source: CDC, “Annual Smoking-Attributable Mortality, Years of Potential Life Lost, and Economic Costs – United States, 1995-1999.”



Note: Individual entries may not sum to totals because of rounding.
a
For adults 35 years old and older.
b
For infants 1 year old and younger.




5                                                 GAO-03-942R CDC’s 2002 Report on Health Consequences of Smoking
In deciding which causes of death to include in its analysis, CDC relied on the
                                                                                      3
Surgeon General’s determination of the causes of death linked to cigarette smoking.
These determinations are based on extensive reviews of scientific literature and are
widely regarded as valid. When new data become available, the Surgeon General’s
determination changes accordingly. An alternative method of estimating deaths
attributable to cigarette smoking that does not depend on decisions about which
causes of death to include has been employed by some researchers. Rather than
including only those deaths due to diseases or conditions that the Surgeon General
considers linked to cigarette smoking, this method estimates deaths attributable to
cigarette smoking regardless of the specific cause of death. CDC officials told us that
they explored this approach, which yielded an estimate of over 540,000 deaths
attributable to cigarette smoking during 1999, but chose not to use it because it would
                                    4
have resulted in inflated estimates.

The method that CDC used to estimate adult and infant deaths and that others used
to estimate secondhand smoke deaths is generally accepted among epidemiologists
                                                             5
as appropriate for attributing deaths to cigarette smoking. Use of this method is
necessary because it is not possible to definitively attribute an individual case of
disease to smoking—deaths from a disease can only be attributed to smoking on a
population basis.6 For example, in the case of lung cancer, not all cigarette smokers
develop lung cancer, and not all people who develop lung cancer are cigarette
smokers. Thus, counting the lung cancer deaths in cigarette smokers and attributing
them to cigarette smoking would not be accurate because some of those deaths
would have occurred even in the absence of cigarette smoking. Instead, the generally
accepted approach attributes to cigarette smoking only the lung cancer deaths among




3
 The Surgeon General’s determinations are based on the application of standard criteria for
establishing causality to information from comprehensive reviews of the scientific literature. For
standard causality criteria, see A. B. Hill, “The Environment and Disease: Association or Causation?”
Proceedings of the Royal Society of Medicine, vol. 58, no. 5 (1965): 295-300.
4
 This method was also applied to 1993 data and produced an estimate of 569,000 deaths attributable to
cigarette smoking in 1993. D.M. Burns, L. Garfinkel, and J.M. Samet, “Introduction, Summary, and
Conclusions,” Changes in Cigarette-Related Disease Risks and Their Implication for Prevention and
Control, Smoking and Tobacco Control Monograph 8 (Bethesda, Md.: U.S. Department of Health and
Human Services, 1997).
5
See, for example, P. Bruzzi et al., “Estimating the Population Attributable Risk for Multiple Risk
Factors Using Case-control Data,” American Journal of Epidemiology, vol. 122, no. 5 (1985): 904-914.
6
It was not necessary to use this method to estimate deaths from fires because, unlike deaths from
disease, an individual death in a fire can be definitively attributed to smoking if the fire department
determines that smoking was the cause of the fire.


6                               GAO-03-942R CDC’s 2002 Report on Health Consequences of Smoking
smokers that are in excess of those expected among nonsmokers. Estimates of
deaths attributable to cigarette smoking using this approach are based on three
components: (1) estimates of the risk for smokers7 relative to nonsmokers of dying
from each specific disease or condition linked to cigarette smoking, (2) estimates of
the prevalence of cigarette smoking, and (3) the number of deaths from each disease
and condition. The estimate of deaths attributable to cigarette smoking derived from
these three components for lung cancer, for example, represents the excess number
of lung cancer deaths that occurred because of cigarette smoking.

Adult Deaths from Diseases Causally Linked to Cigarette Smoking

CDC’s estimate of deaths in adult smokers due to diseases causally linked to cigarette
smoking accounted for about 91 percent of its estimate of total deaths attributable to
cigarette smoking. The data sources that CDC used for each of the three components
of the estimate of the number of deaths attributable to cigarette smoking all had
limitations that potentially could have affected the estimate. However, each was the
best data source available for each particular purpose, and CDC recognized and dealt
appropriately with the limitations so that their effects on the estimate were minimal.

For the first component of its estimate of adult deaths—estimating the risk of death
for smokers relative to nonsmokers—CDC used the American Cancer Society’s
                                            8
second Cancer Prevention Study (CPS-II). This study gathered data on individuals’
demographic traits, medical history, and behavior (such as alcohol use) and reported
on the relationship between cigarette smoking and death. CPS-II, with a sample of
about 1.2 million individuals, had a size advantage over other studies that have similar
information. Smaller samples are not sufficient to produce estimates of cigarette
smoking risks that have margins of error as small as those obtained using CPS-II.

Although the CPS-II sample was not representative of the national population—for
example, nonwhites were underrepresented—adjustments can be made for the
nonrepresentativeness of the overall sample by estimating cigarette smoking risks
taking account of other factors, such as race. The size of the CPS-II sample enabled
CDC to isolate the increase in risk that was directly attributable to cigarette smoking
and adjust for the effect that multiple factors can have on a person’s risk of death.




7
 Smokers are generally classified as either current smokers or former smokers, and separate estimates
are derived for each group.
8
 M.J. Thun et al., “Trends in Tobacco Smoking and Mortality from Cigarette Use in Cancer Prevention
Studies I (1959 through 1965) and II (1982 through 1988),” Changes in Cigarette-Related Disease Risks
and Their Implication for Prevention and Control, Smoking and Tobacco Control Monograph 8
(Bethesda, Md.: U.S. Department of Health and Human Services, 1997). Study participants self-
reported information on their medical history, current health status, and a series of lifestyle factors
including smoking behaviors. During the 6-year follow-up period, deaths among participants were
recorded along with the cause of death as recorded on the death certificate. Death certificates were
obtained for approximately 97 percent of all study participants known to have died.


7                              GAO-03-942R CDC’s 2002 Report on Health Consequences of Smoking
For example, although the proportion of nonwhites in the sample was less than the
proportion in the general population, the sample still contained enough nonwhites to
analyze the effect of race on the relative risks. CDC used data from CPS-II and
additional studies to evaluate the importance of race and other factors—such as
education, alcohol use, and diabetes—and concluded that only age and sex needed to
be taken into account in estimating the relative risks.9

CPS-II was almost 20 years old at the time of CDC’s report. It was initiated in 1982
and follow-up of individuals in the study is ongoing. The relative risk estimates that
CDC used were based on follow-up through 1988. Thus, if relative risks had changed
over time, those estimated from CPS-II might not have been accurate for estimating
deaths during 1995 through 1999. However, CDC and others reviewed studies at
different points in time and determined that the relative risks were likely to have
remained stable and were still applicable.

For the second component of the estimate of deaths attributable to cigarette
smoking—estimates of the prevalence of cigarette smoking among adults—CDC used
the National Health Interview Survey (NHIS), which has detailed data on cigarette
                                                    10
smoking for the years included in CDC’s analysis. CDC chose to use data that
capture cigarette smoking prevalence during the same years that the deaths of
interest occurred. Using prevalence data from the same years that the deaths
occurred underestimates the number of deaths attributable to cigarette smoking
because, for example, deaths in 1999 are the result of exposure to cigarette smoke
during previous decades and the prevalence of cigarette smoking declined by 25
percent during the 1990s. In addition, former smokers in 1999 may have been
different from former smokers in the year that relative risks were estimated—that is,
having quit relatively recently, their risk may resemble that of current smokers more
closely than that of former smokers. CDC officials said that they accepted this
limitation since its result was a lower estimate of the number of deaths attributable to
cigarette smoking.

CDC’s source for the data needed for the last component of the estimate of deaths
attributable to cigarette smoking—the total number of deaths due to each disease
each year—was death certificates. CDC obtained these data from the National Center
for Health Statistics (NCHS), which is the national repository for information from
birth and death certificates. NCHS has determined that death certificates accurately
capture the cause of death about 97 percent of the time.


9
M.J. Thun, L.F. Apicella, and S.J. Henley, “Smoking vs Other Risk Factors as the Cause of Smoking-
Attributable Deaths: Confounding in the Courtroom,” JAMA, vol. 284, no. 6 (2000): 706-712 and A.M.
Malarcher et al., “Methodological Issues in Estimating Smoking-Attributable Mortality in the United
States,” American Journal of Epidemiology, vol. 152, no. 6 (2000): 573-584.
10
 The National Health Interview Survey (NHIS) is a nationally representative survey of health trends in
the civilian population. The survey collects basic health and demographic information every year and
frequently includes questions on smoking. The 1999 NHIS sample consisted of 37,573 households,
which yielded 97,059 persons in 38,171 families. For the adult component, 30,801 persons 18 years or
older were interviewed.


8                              GAO-03-942R CDC’s 2002 Report on Health Consequences of Smoking
Infant Deaths from Conditions Causally Linked to Maternal Cigarette Smoking
during Pregnancy

Infant deaths from conditions causally linked to maternal cigarette smoking
accounted for less than one half of 1 percent of the total deaths attributable to
cigarette smoking. CDC’s estimate of infant deaths was based on the same three
components as for adults, but the data sources were necessarily different because,
for example, CPS-II was a study of only adults. CDC’s source of data for the first
component—estimates of the risk of dying for infants whose mothers smoked
cigarettes during pregnancy relative to those whose mothers did not smoke—was a
                                                                11
review of studies of the effects of maternal cigarette smoking. For the second
component, CDC used data compiled from birth certificates and surveys of new
mothers to obtain estimates of cigarette smoking prevalence among pregnant women.
As for adult deaths, the source for the last component of the estimate was NCHS data
on the number of infant deaths each year from each condition.

Adult Deaths from Diseases Causally Linked to Exposure to Secondhand Cigarette
Smoke

Deaths associated with secondhand cigarette smoke accounted for about 9 percent of
CDC’s estimated total number of deaths attributable to cigarette smoking. CDC
obtained its estimates of deaths attributable to secondhand cigarette smoke from a
National Cancer Institute (NCI) report.12 CDC used the NCI report’s estimate of 3,000
                                                                         13
annual lung cancer deaths associated with secondhand cigarette smoke. The NCI
report presented a range of estimates (35,000-62,000) for deaths from ischemic heart
disease. CDC used the estimate of 35,000 because it was the lowest of the range and
relied on the same data source CDC used to develop estimates for adult deaths due to
cigarette smoking (CPS-II) and thus would be consistent with those estimates.




11
 N.I. Gavin, C. Wiesen, and C. Layton, Review and Meta-Analysis of the Evidence on the Impact of
Smoking on Perinatal Conditions Built into SAMMEC II, (Washington, D.C.: Centers for Disease
Control and Prevention, September 2000).
12
 U.S. Department of Health and Human Services, National Institutes of Health, National Cancer
Institute, Health Effects of Exposure to Environmental Tobacco Smoke: The Report of the California
Environmental Protection Agency, Smoking and Tobacco Control Monograph 10 (Bethesda, Md.:
1999).
13
 The NCI report cited the Environmental Protection Agency’s (EPA) estimate of 3,000 annual lung
cancer deaths associated with secondhand smoke (see U.S. Environmental Protection Agency, Office
of Research and Development, Office of Health and Environmental Assessment, Respiratory Health
Effects of Passive Smoking: Lung Cancer and Other Disorders (Washington, D.C.: December 1992)).
After the report was published, several tobacco companies filed a lawsuit seeking to have the report
withdrawn, claiming that EPA had violated procedural requirements in developing the report. In 1998,
a district court invalidated certain chapters of the report, including those on lung cancer. In December
2002, the U.S. Court of Appeals overturned the district court’s decision and ordered that the suit be
dismissed, concluding that the district court had lacked jurisdiction to hear the suit. (See Flue-Cured
Tobacco Cooperative Stabilization Corporation v. United States EPA, 313 F.3d 852 (4th Cir. 2002).)


9                              GAO-03-942R CDC’s 2002 Report on Health Consequences of Smoking
Deaths from Residential Fires Caused by Smoking

Deaths from residential fires accounted for less than one half of 1 percent of CDC’s
estimated total number of deaths attributable to cigarette smoking.14 CDC obtained its
estimates of residential fire deaths from the National Fire Protection Association
(NFPA). NFPA national estimates are of the average annual number of deaths due to
fires caused by smoking and are based on data reported to the U.S. Fire
Administration and NFPA’s annual survey of fire departments.15

Years of Potential Life Lost

CDC’s estimate of the years of potential life lost was built on the estimate of number
of deaths attributable to cigarette smoking and provided another perspective on
mortality attributable to cigarette smoking. CDC reported that, on average, men and
women who died from cigarette smoking-related illness each lost about 13 and 15
years of life, respectively. When mortality attributable to cigarette smoking is
measured in terms of the number of deaths, each death contributes equally to the
total. In contrast, when mortality is measured in terms of years of potential life lost,
each death contributes to the total depending on how premature the death was. This
measure takes life expectancy into account, and thus death at a younger age results
in a greater loss of potential years of life than death at an older age. For example, an
infant who died as a result of maternal cigarette smoking would likely have had a
greater life expectancy than an elderly lifetime smoker who died of lung cancer and
so would contribute more years of potential life lost to the total. Thus, although
infant mortality accounts for .23 percent of the total number of deaths, it accounts for
almost six times that percentage (1.36 percent) of the total number of years lost. In
contrast, adult lung cancer mortality accounts for about the same proportion of both
the total number of deaths and the total number of years lost.

CDC used national life expectancy data published by NCHS to estimate the expected
years of life remaining for those who died from cigarette smoking. The expected life




14
 The National Fire Protection Association (NFPA) estimate that CDC cited includes deaths from the 1
to 2 percent of fires caused by cigars and pipes. The estimate does not include deaths from
nonresidential and auto-related fires.
15
 The United States Fire Association (USFA) is part of the Federal Emergency Management Agency.
The fire reports sent to USFA’s voluntary fire reporting system account for about half of the fires each
year, and representation of certain regions of the country and communities may not be uniform. To
address these issues, NFPA supplements USFA’s data with its own annual survey of a sample of fire
departments. NFPA assumes that fires with unknown or unreported causes have the same
proportional distribution as fires for which the cause is known and reported.



10                               GAO-03-942R CDC’s 2002 Report on Health Consequences of Smoking
span differs for women and men, by age group, and by year assessed.16 For example,
in 1995, a 65-year old woman was expected to live another 19 years to age 84 and a
65-year old man was expected to live another 16 years to age 81. A 75-year old woman
in that year was expected to live another 12 years to age 87. In contrast to life
expectancies in 1995, a 65-year old woman in 1950 had an expected life span of 80
years. CDC calculated years of potential life lost by multiplying the estimated
remaining life expectancy for each sex and age group in each year from 1995 through
1999 by the number of cigarette smoking-attributable deaths in that group in each
     17
year. CDC did not estimate the years lost from secondhand cigarette smoke deaths
because the NCI report from which CDC obtained the estimate of secondhand smoke
deaths did not have sufficient age-specific data. Thus CDC’s estimate of the total
number of potential years of life lost did not include the lost years associated with
about 9 percent of the total estimated deaths.

CDC’s Estimates of Mortality-Related Productivity Losses and Medical
Expenditures Due to Cigarette Smoking Were Reasonable

CDC’s estimates of $82 billion annually in productivity losses from mortality
attributable to cigarette smoking and $76 billion in additional medical expenditures
for all smokers annually were reasonable. CDC arrived at these estimates using
approaches that were generally accepted among economists and relied on large
federal data sources. CDC recognized and handled appropriately the limitations in the
data on which the estimates are based.

Mortality-Related Productivity Losses

CDC’s estimate of mortality-related productivity losses built on its estimates of death
and years of potential life lost and measured mortality in economic terms. CDC
valued the years of potential life lost in terms of the productivity lost as a result of
those lost years. CDC used expected future earnings, calculated in current dollars, to
represent mortality-related productivity losses. An alternative approach to estimating
mortality-related productivity losses attempts to capture the broader impact on
productivity of death by accounting for such factors as time and costs to replace
                                            18
workers and restore productivity levels. CDC did not take this approach because it
was not widely accepted.


16
  Life expectancy also differs by race, with blacks of both sexes and of all ages generally having lower
life expectancies than whites for all years. CDC did not estimate cigarette-smoking attributable deaths
separately by race and thus did not estimate years of potential life lost by race.
17
 An alternative method for estimating years of potential life lost is to calculate the years of life
remaining using life expectancy at birth rather than at the age of death. This method would likely have
resulted in a lower estimate of the total years of productive life lost; however, it is not the method
generally accepted among public health experts.
18
 M.A. Koopmanschap, “Estimating the Indirect Costs of Smoking Using the Friction Cost Method,” ed.
C. Jeanrenaud and N. Soguel, Valuing the Cost of Smoking: Assessment Methods, Risk Perception
and Policy Options (Boston: Kluwer Academic Publishers, 1999).


11                               GAO-03-942R CDC’s 2002 Report on Health Consequences of Smoking
In estimating mortality-related productivity losses, CDC used estimates of expected
         19
earnings derived from the Bureau of Labor Statistics (BLS), U.S. Census Bureau, and
other national sources. They take into account both changes in earnings and the
value of money over time. They also include the estimated value of household work
that accounts for the productivity losses among individuals who do not earn wages
for household services. CDC updated the published earnings estimates using an
adjustment factor from BLS so that the estimates would reflect 1995-99 earnings.
CDC used a single average estimate of future lifetime earnings for men and women.
Because of men’s higher average earnings and higher incidence of cigarette smoking-
related death compared to women, the productivity loss estimate was likely to be
lower than if separate average earnings had been used for men and women.

CDC’s estimate of mortality-related productivity losses did not include the expected
lost earnings associated with infant or secondhand cigarette smoke deaths (about 9
percent of the total deaths). CDC said that it did not develop an estimate of
productivity losses for infants because of a lack of consensus among economists
about the best method for estimating the potential future earnings of infants.
Similarly, the NCI report from which CDC obtained the estimate of secondhand
smoke deaths lacked specific data on the age at which those deaths occurred—
information needed to estimate expected lost earnings. CDC informed us that it is
working on including these two categories in future estimates of productivity losses
when more reliable data become available.

Medical Expenditures

CDC’s estimate of $76 billion annually in additional medical expenditures attributable
to cigarette smokers was not built on the other three estimates, and its approach to
developing this estimate was different from its approach to the others. CDC
examined the use of health care services and the cost of those services for smokers
compared to nonsmokers independent of the reason—that is, the disease or
condition—for the services. Thus, this estimate is not limited to medical expenditures
associated with a set of diseases and conditions causally linked to cigarette smoking.
CDC’s estimate of total medical expenditures was the sum of five estimates by type of
health care service for adults—ambulatory care, hospital care, prescription drugs,
nursing home,20 and other (including home health care, nonprescription drugs, and
nondurable medical equipment)—and an estimate of expenditures for neonatal health
care services. (See table 2.) CDC estimated these expenditures on an annual basis.




19
 CDC’s estimates were drawn from A.C. Haddix et al., eds., Prevention Effectiveness: A Guide to
Decision Analysis and Economic Evaluation (New York: Oxford University Press, 1996).
20
  CDC’s nursing home estimate accounts for differences between smokers and nonsmokers in the
likelihood of admission to a nursing home but not differences in readmission or length of stay.


12                             GAO-03-942R CDC’s 2002 Report on Health Consequences of Smoking
Table 2: CDC’s Estimates of Annual Mortality-Related Productivity Losses and Medical Expenditures
Attributable to Cigarette Smoking

    Dollars in millions
                                                                                                                                       Total
    Mortality-related productivity losses
    Men                                                                                                                              $55,389
    Women                                                                                                                             26,483
    Total mortality-related productivity losses                                                                                      $81,872
                          a
    Medical expenditures
    Ambulatory care                                                                                                                  $27,182
    Hospital care                                                                                                                     17,140
    Prescription drugs                                                                                                                 6,364
    Nursing home                                                                                                                      19,383
    Other care                                                                                                                         5,419
    Neonatal                                                                                                                             366
    Total medical expenditures                                                                                                       $75,854

Source: CDC, “Annual Smoking-Attributable Mortality, Years of Potential Life Lost, and Economic Costs – United States, 1995-1999.”
a
 CDC’s estimate of annual personal medical expenditures for adults attributable to cigarette smoking was derived using 1998
data obtained from the Health Care Financing Administration and is in 1998 dollars. Its estimate of annual neonatal medical
expenditures attributable to maternal cigarette smoking was based on 1996 data and is in 1996 dollars.


The data source that CDC used to determine medical expenditures for smokers
compared to nonsmokers allowed CDC to adjust for many factors—including certain
risk-taking behaviors (e.g., not wearing a seat belt)—that may affect health care
                                              21
expenditures independent of smoking status. However, although the data source
was the most comprehensive available, it did not include information on alcohol
consumption. CDC used data from another study to assess the importance of drinking
alcohol with respect to expenditures attributable to cigarette smoking and concluded
that adjusting the data for drinking would not have had an appreciable effect on the
results.22 Expenditures for dental care and mental health care and certain costs




21
 CDC’s primary data source for determining medical expenditures for smokers compared to
nonsmokers was the 1987 National Medical Expenditure Survey, a population-based survey of over
38,000 individuals in about 14,000 households.
22
 This study assessed utilization of health care services using data from a nationally representative
survey of adults that included information on utilization of medical care, smoking, and alcohol
consumption. CDC based its conclusion on findings from this study, after an expert panel determined
that these findings were applicable to CDC’s analysis. V.P. Miller, C. Ernst, and F. Collin, “Smoking-
Attributable Medical Care Costs in the USA,” Social Science & Medicine, vol. 48, no. 3 (1999): 375-391.


13                                                  GAO-03-942R CDC’s 2002 Report on Health Consequences of Smoking
associated with the care of infants of cigarette smoking mothers were not included in
                23
CDC’s estimate. In addition, certain expenditures for health services associated with
secondhand cigarette smoke (e.g., care for lung cancer due to secondhand cigarette
smoke in a nonsmoker) and care for nonsmokers injured in residential fires caused
by smoking are not accounted for in the estimate.

By estimating medical expenditures on an annual basis, CDC avoided limitations
associated with the alternative of estimating expenditures over an individual’s
lifetime. The lifetime approach is based on a series of assumptions and predictions
about disease course and duration, survival rates, patterns of medical care, and
impact of disease on employment, among other factors. Results using a lifetime
approach have varied widely—some studies have concluded that smokers have more
medical expenditures than nonsmokers over their lifetimes and other studies have
                                  24
come to the opposite conclusion. Changes in the assumptions underlying the annual
approach have less of an effect on the results. CDC’s estimates are consistent with
                                                                           25
other annual estimates of medical expenditures published in the literature.

Agency Comments

In its comments on a draft of this report (see enclosure III), CDC said that this report,
in general, accurately represents the intent, methods, and decision-making processes
of its April 2002 report. With respect to our discussion of the relative risks obtained
from CPS-II, CDC noted that while the overall prevalence of smoking may have
decreased since CPS-II, the relative risks for smokers compared to nonsmokers
would not have decreased because smoking behavior was similar. We have
incorporated CDC’s technical comments as appropriate.

                                                -----




23
 CDC’s estimate of costs associated with smoking during pregnancy includes only neonatal hospital
expenditures and excludes costs of care throughout infancy (for example, those associated with
hospital readmissions in the first year of life) and expenditures associated with treating secondhand
smoke-related conditions arising after birth. The estimates were based on data from CDC’s Pregnancy
Risk Assessment Monitoring System and 1996 private sector claims data from the Medstat
MarketScan™ database.
24
 For an example of a study that found greater lifetime expenditures for smokers, see T.A. Hodgson,
“Cigarette Smoking and Lifetime Medical Expenditures,” Milbank Quarterly, vol. 70, no. 1 (1992): 81-
125. For an example of a study that found fewer lifetime expenditures for smokers, see B.C. Lippiatt,
“Measuring Medical Cost and Life Expectancy Impacts of Changes in Cigarette Sales,” Preventive
Medicine, vol.19, no. 5 (1990): 515-532.
25
 CDC’s estimate of annual personal medical expenditures attributable to smoking, $76 billion,
represents approximately 8 percent of total personal medical expenditures, an estimate within the
range of other annual estimates (from about 6 to about 9 percent). See W. Max, “The Financial Impact
of Smoking on Health-related Costs: A Review of the Literature,” American Journal of Health
Promotion, vol. 15, no. 5 (2001): 321-333.


14                              GAO-03-942R CDC’s 2002 Report on Health Consequences of Smoking
As agreed with your office, unless you publicly announce its contents earlier, we plan
no further distribution of this report until 30 days after its issue date. At that time, we
will send copies to the Director of CDC and other interested parties. We will also
make copies available to others upon request. In addition, this report will be
available at no charge on GAO’s Web site at http://www.gao.gov. If you have
questions or would like additional information, please call me at (202) 512-7119.
Another contact and contributors to this report are listed in enclosure IV.

Sincerely yours,




Janet Heinrich
Director, Health Care—Public Health Issues
Enclosures




15                          GAO-03-942R CDC’s 2002 Report on Health Consequences of Smoking
Enclosure I                                                                                                 Enclosure I

     CDC’s Table Presenting Its Estimates of Cigarette Smoking-Attributable
                   Mortality and Years of Potential Life Lost




Note: This table is taken from page 302 of CDC, “Annual Smoking-Attributable Mortality, Years of Potential Life Lost, and
Economic Costs – United States, 1995-1999.”




16                                     GAO-03-942R CDC’s 2002 Report on Health Consequences of Smoking
Enclosure II                                                                                                 Enclosure II

     CDC’s Table Presenting Its Estimates of Smoking-Attributable Mortality-
             Related Productivity Losses and Medical Expenditures




Note: This table is taken from page 303 of CDC, “Annual Smoking-Attributable Mortality, Years of Potential Life Lost, and
Economic Costs – United States, 1995-1999.”




17                                      GAO-03-942R CDC’s 2002 Report on Health Consequences of Smoking
Enclosure III                                                          Enclosure III



       Comments from the Centers for Disease Control and Prevention




18                     GAO-03-942R CDC’s 2002 Report on Health Consequences of Smoking
Enclosure IV                                                              Enclosure IV



                   GAO Contact and Staff Acknowledgments


GAO Contact

Michele Orza, (202) 512-6970

Acknowledgments

The following staff members made important contributions to this work: Angela
Choy, Chad Davenport, Maria Hewitt, Donald Keller, and Nkeruka Okonmah.




(290254)


19                        GAO-03-942R CDC’s 2002 Report on Health Consequences of Smoking
This is a work of the U.S. government and is not subject to copyright protection in the
United States. It may be reproduced and distributed in its entirety without further
permission from GAO. However, because this work may contain copyrighted images or
other material, permission from the copyright holder may be necessary if you wish to
reproduce this material separately.
                         The General Accounting Office, the audit, evaluation and investigative arm of
GAO’s Mission            Congress, exists to support Congress in meeting its constitutional
                         responsibilities and to help improve the performance and accountability of the
                         federal government for the American people. GAO examines the use of public
                         funds; evaluates federal programs and policies; and provides analyses,
                         recommendations, and other assistance to help Congress make informed
                         oversight, policy, and funding decisions. GAO’s commitment to good
                         government is reflected in its core values of accountability, integrity, and
                         reliability.


                         The fastest and easiest way to obtain copies of GAO documents at no cost is
Obtaining Copies of      through the Internet. GAO’s Web site (www.gao.gov) contains abstracts and full-
GAO Reports and          text files of current reports and testimony and an expanding archive of older
                         products. The Web site features a search engine to help you locate documents
Testimony                using key words and phrases. You can print these documents in their entirety,
                         including charts and other graphics.

                         Each day, GAO issues a list of newly released reports, testimony, and
                         correspondence. GAO posts this list, known as “Today’s Reports,” on its Web
                         site daily. The list contains links to the full-text document files. To have GAO e-
                         mail this list to you every afternoon, go to www.gao.gov and select “Subscribe to
                         e-mail alerts” under the “Order GAO Products” heading.


Order by Mail or Phone   The first copy of each printed report is free. Additional copies are $2 each. A
                         check or money order should be made out to the Superintendent of Documents.
                         GAO also accepts VISA and Mastercard. Orders for 100 or more copies mailed to
                         a single address are discounted 25 percent. Orders should be sent to:

                         U.S. General Accounting Office
                         441 G Street NW, Room LM
                         Washington, D.C. 20548

                         To order by Phone:    Voice:     (202) 512-6000
                                               TDD:       (202) 512-2537
                                               Fax:       (202) 512-6061


                         Contact:
To Report Fraud,
                         Web site: www.gao.gov/fraudnet/fraudnet.htm
Waste, and Abuse in      E-mail: fraudnet@gao.gov
Federal Programs         Automated answering system: (800) 424-5454 or (202) 512-7470


                         Jeff Nelligan, Managing Director, NelliganJ@gao.gov (202) 512-4800
Public Affairs           U.S. General Accounting Office, 441 G Street NW, Room 7149
                         Washington, D.C. 20548