1

Epidemiology and
Prevention of
Cancer
Otis W. Brawley, MD

LEARNING OBJECTIVES
After reviewing this chapter, the reader should be able to

Explain the epidemiology of cancer as a disease process

Summarize screening and prevention techniques for various types of cancer

for American men and women from 1996 to 2000 have been
compiled (Table 1-1).1
Survival is defined as the time from diagnosis to death.
A commonly used measure is the proportion of people
alive at five years after diagnosis (Table 1-2).
The American Cancer Society publishes an annual estimate of the absolute number of new cancer cases and
deaths. These numbers are widely quoted, especially by the
lay press. It should always be remembered that the numbers
are crude estimates based on rates measured in past years.2

Overview
Epidemiology is the study of disease in populations. It
is the analysis of trends over time, as well as the characteristics of individuals in whom a disease develops and in
whom it does not. Other aspects of epidemiology include
assessment of the effects of treatment, prevention, or
screening interventions on a group of people.
Descriptive epidemiology involves incidence and mortality rates, survival, and the number of cases of a disease diagnosed in a certain population. Incidence and mortality rates
are commonly expressed as the number of cases per 100,000
in the group at risk. These rates are frequently age adjusted,
meaning they are mathematically adjusted to a standard population to remove the effect of a populations age distribution
changing over time. Cancer is primarily a disease of older
people. With the extensive increase over the past 30 years in
the number of individuals in the United States who are 70
years and older, the absolute number and the crude rate per
100,000 of patients with cancer has also increased. Adjusting
disease rates for age removes the effects of aging of the population. Increases in age-adjusted incidence and mortality
rates are the result of causes other than the aging of the population. Average annual cancer incidence and mortality rates

Risk
Much of epidemiology involves assessment of cancer
risk. A person can be at increased risk of cancer due to

KEY POINT
Epidemiology is the study of disease in populations. It can include assessment of treatment
outcomes, disease prevention, and disease
screening.

Table 1-1. Cancer Incidence and Mortality, Annualized Per 100,000 Using Surveillance, Epidemiology, and End Results (SEER) Data from 1996 to 20001`
Women

Incidence
White
Non-Hispanic

Cancer
Oral cavity and pharynx
Esophagus
Stomach
Colon and rectum
Liver and intrahepatic
Pancreas
Lung and bronchus
Melanoma of the skin
Breast
Cervix uteri
Corpus and uterus
Ovary
Urinary bladder
Kidney and renal pelvis
Brain and other nervous system
Thyroid
Hodgkins lymphoma
Non-Hodgkins lymphoma
Myeloma
Leukemia

6.8
2.1
4.3
47.5
2.4
9.3
53.9
18.6
148.3
7.6
27.0
18.3
10.2
7.6
6.2
10.5
2.9
16.6
4.0
9.6

Men

Black
6.4
4.2
9.9
56.2
3.9
14.3
54.8
0.8
121.7
12.4
17.9
11.9
7.6
9.4
3.4
5.6
2.0
11.6
10.2
7.8

Hispanic
3.8
1.2
10.0
32.9
5.6
9.3
24.4
4.4
89.8
16.8
16.5
14.0
4.9
8.0
4.5
9.6
1.9
13.2
4.2
7.6

White
Non-Hispanic

Black

Hispanic

1.7
1.7
2.7
7.6
2.6
8.9
43.1
2.1
27.4
2.6
3.8
9.3
2.3
2.9
4.2
0.4
0.4
7.4
2.9
6.1

2.0
3.4
6.5
24.6
3.7
12.9
40.0
0.5
35.9
5.9
7.0
7.4
3.0
2.8
2.3
0.5
0.3
4.6
6.6
5.5

0.8
1.0
5.3
11.4
5.0
7.4
15.1
0.6
17.9
3.7
3.1
6.1
1.2
2.5
2.5
0.7
0.4
5.5
2.9
4.40

Incidence
White
Non-Hispanic

Cancer
Oral cavity and pharynx
Esophagus
Stomach
Colon and rectum
Liver and intrahepatic
Pancreas
Larynx
Lung and bronchus
Melanoma of the skin
Prostate
Testes
Urinary bladder
Kidney and renal pelvis
Brain and other nervous system
Thyroid
Hodgkins lymphoma
Non-Hodgkins lymphoma
Myeloma
Leukemia

16.6
7.6
10.0
64.6
6.2
12.4
6.8
80.5
27.5
163.3
6.8
40.9
15.6
9.0
4.0
3.5
25.1
6.7
16.7

Black
20.5
11.4
19.9
72.4
11.0
18.0
12.0
120.4
1.5
272.1
1.4
20.4
17.7
4.8
2.0
2.8
18.6
12.9
13.0

Extrinsic influences, such as environmental pollutants,

cultural habits, and diet
Intrinsic influences, such as genetics
Risk is often reported relative to another population.
For example, between 1996 and 2000, the breast cancer
Table 1-2. Definition of Terms Related to Survival
Survival
Disease-free survival
Five-year survival rate
Disease-specific survival rate

Overall survival rate

Mortality

Time from the initial diagnosis of cancer

to death
Time from complete remission to relapse
of disease
Proportion of patients who are alive five
years after the time of diagnosis
Proportion of patients who have died of
the specific disease (does not take into
account deaths unrelated to the disease)
Proportion of patients who are alive at a
specific time after the diagnosis (takes
into account all causes of death)

| Epidemiology and Prevention

Mortality

Hispanic
9.9
6.0
18.1
49.8
13.8
10.6
5.4
46.1
4.3
137.2
3.5
18.3
13.9
5.5
2.6
3.1
19.0
6.4
11.6

White
Non-Hispanic
4.1
7.5
5.7
25.4
5.7
12.0
2.4
80.8
4.7
30.5
0.3
8.1
6.3
6.3
0.4
0.6
11.2
4.5
10.7

Black
7.9
12.2
14.0
34.6
9.3
16.4
5.7
107.0
0.5
73.0
0.2
5.8
6.2
3.3
0.3
0.6
7.6
9.2
9.3

Hispanic
3.1
4.5
9.9
18.4
10.5
9.4
2.3
40.7
1.2
24.1
0.2
4.2
5.4
3.5
0.5
0.8
8.4
3.7
6.8

mortality rate for black women was 35.9 per 100,000, and
the rate for non-Hispanic white women was 27.4 per
100,000. During that period, the relative risk of death for
black women was 1.31 times that of white women (35.9
divided by 27.4).
A number of genes, polymorphisms, and genetic
mutations have been identified as predisposing a patient
for a disease. Many more will be discovered in the near

KEY POINTS
Risk can be increased due to extrinsic and
intrinsic influences.
A population risk is often reported relative to
another population.

future. Testing for these genes can define a high-risk

population.
The study of mutations of the BRCA genes provide some
relevant lessons about genetic mutations and polymorphism
and cancer risk. BRCA1 and BRCA2 are genes with
well-defined DNA sequences. Some BRCA1 and BRCA2
mutations are known to increase the risk of breast and ovarian cancer and possibly other malignant diseases.3 Other
BRCA mutations have an unknown effect on breast cancer
risk. A BRCA mutation of unknown significance in a
woman with an extensive family history of breast cancer
likely means the mutation is related to the increased risk.
Such a mutation in a woman without a family history may
have no effect.
The ability to predict development of a particular
cancer can help select present therapeutic options as well
as present ethical dilemmas. It may allow for early intervention that could prevent disease or limit its severity.
Chemoprevention and screening should be carefully
investigated for high-risk populations to determine efficacy. To date, individuals at high risk of cancer can
engage in intensive screening for the cancer in question.
Although such screening may be clinically prudent, it
may be less effective for individuals at very high risk. In
theory, a screening test might benefit those at risk for
sporadic cancers and not benefit those at genetic high
risk for a cancer.
Population categorization is important in epidemiology.
Populations can be delineated by sex, nationality, culture,
ethnicity, area of geographic origin, socioeconomic status,
or race. Race and ethnicity are the most common ways of
dividing populations in the United States. It should be
remembered that race is a sociopolitical categorization.4
The definitions used by the United States when generating
population statistics are not formulated on the basis of science. Indeed, some groups such as native Hawaiians and
natives of the Indian subcontinent of Asia have been
placed in different categories over the past 30 years.
Whereas the genetic basis of race has been a sociopolitical topic, a tenet of anthropology is that there is no genetic
basis for racial categorization. The area of geographic origin
can, however, be used to define a group with a higher prevalence of a specific genetic mutation or polymorphism. The
sickle-cell mutation is prevalent among individuals originating in north and central Africa and in the northern and
eastern Mediterranean region, but not in those who originate in southern Africa. Hence, there is one population
commonly referred to as white (in Italy, Greece, and
Lebanon) in which the sickle-cell mutation is found and
another commonly referred to as black (southern Africa)
in which the mutation is absent. A malaria epidemic sev-

eral thousand years ago in the contiguous area of north and

central Africa and the European Mediterranean region is
thought to be associated with the sickle-cell mutation.
The 185delAG and 5382insC mutations in the
BRCA1 gene and the 6174delT mutation in the BRCA2
gene have been found in approximately 2.5% of Askenazi
Jews.5 Population scientists believe the prevalence of this
mutation in this specific group of people is the result of just
a few individuals having this marker approximately 1,200
years ago. This so-called founder effect is an example of
both the effect of sociopolitical pressures on human genetics and the importance of family on genetics. There are
family clusters of breast cancer that have not been linked
to defined genetic mutations.6,7
Socioeconomic status and education can also be
related to the risk of disease and death. A higher rate of
breast cancer among white women in the San Francisco
Bay area and on Long Island was linked to a higher prevalence of professional women in those areas who, as a
cohort, are less likely to have a full-term pregnancy by age
30, a known risk factor for breast cancer.8
Race and ethnicity can correlate with other methods
of categorization, such as poverty or prosperity, both
of which are capable of changing the incidence of cancer
and its related mortality.

Patterns of Care and

Outcomes Research
The assessment of treatment outcomes has become an
important aspect of epidemiology. A phase II clinical trial
can demonstrate that a treatment intervention works (i.e.,
shrinks the tumor). A phase III study can compare two
interventions to determine which is superior. Prevention
trials usually require phase III studies to show efficacy.
Clinical trials demonstrate efficacy of a treatment. How
well the intervention works in the population as a whole
in routine practice is termed effectiveness.
The study of patterns of care or treatments used is
an aspect of outcomes research. Numerous studies
have demonstrated regional differences in the preferred

KEY POINT
Most trials are designed to determine efficacy,
meaning how well the treatment works in a
selected environment. Some larger trials and
outcomes studies are designed to show effectiveness, meaning how well the treatment works in
the population as a whole.

Epidemiology and Prevention |

treatment of several cancers.9,10 For example, for women

with localized breast cancer, the decision to treat with
lumpectomy and radiation therapy or with mastectomy
varies depending on the patients geographic location.
Similar regional differences have been noted for prostate
cancer screening and the types of treatment used for localized prostate cancer.11
Many of the disparities in outcomes among groups as
defined by race and socioeconomic status have been linked
to differences in patterns of care. For example, treatment
is less than optimal for a substantial proportion of patients
with cancer who are poor or of certain ethnic backgrounds.12 The reasons for these variances in care are complex. Some are the result of cultural differences in the
acceptance of therapy. In other cases, poverty can make
accessible care less convenient because of transportation
difficulties. Patients with severe comorbid disease may justifiably not be offered risky aggressive cancer treatments as
they are at higher risk of a treatment-related morbidity.
Unfortunately, racism and socioeconomic discrimination
do play some part.

Cancer Prevention
Primary cancer prevention is best defined as the use of
interventions to lower cancer risk. Important to prevention
is the fact that carcinogenesis is not a distinct event but
rather a process that occurs over time. It is a cumulative
continuum of discreet cellular changes resulting in uncontrolled growth. Primary prevention involves interventions
or manipulations of the genetic, biologic, and environmental factors in the causal pathway of carcinogenesis.
Smoking cessation, sun avoidance, diet modification, and
chemoprevention are primary-prevention activities.
Screening for asymptomatic cancers is considered secondary prevention. For some cancers, intraepithelial neoplasia

KEY POINTS
Avoidance of carcinogens is the easiest way to
prevent cancer. Smoking is the cause of nearly
one-third of all cancers in the United States.
Other environmental influences such as sun
overexposure, certain chemicals, and certain
viruses are associated with cancer causation.
Drugs and vitamins previously thought to be
harmless have been associated with increased
risk of harm, demonstrating the need to
rigorously assess a putative cancer preventative
agent before advocating its use.

| Epidemiology and Prevention

is an intermediate step in carcinogenesis, and treatment of

this condition is a form of cancer prevention.13

Smoking Cessation
Tobacco use is the most avoidable risk factor for
cardiovascular disease, pulmonary disorders, and cancer.
Smoking cessation and avoidance have the potential to
save and extend more lives than any other public health
activity. A smoker has a one in three lifetime risk of dying
prematurely of a smoking-related disease. More human
lives are lost because of cardiovascular disease caused by
smoking than from smoking-related cancer. In addition to
lung cancer, cigarette smoking has been linked to cancer
of the larynx, oropharynx, esophagus, kidney, bladder, and
possibly the pancreas.
The risk from tobacco smoke is not necessarily limited
to the smoker. Epidemiologic studies strongly suggest that
environmental tobacco smoke, often called secondhand or
passive smoke, may cause lung cancer and other pulmonary diseases in nonsmokers. The amount of smoke
exposure, as well as the degree of inhalation of cigarette
smoke, is correlated with the risk of mortality associated
with lung cancer. The number of cigarettes smoked per day
is an approximate surrogate for cigarette exposure; however, socioeconomically disadvantaged people smoke cigarettes more efficiently, inhaling more smoke per cigarette
than middle-class individuals. Twenty pack-years (meaning one pack a day for 20 years or two packs a day for
10 years) means more smoke inhaled if the smoker is
poor.14
Light and low-tar cigarettes are not safer, as smokers
tend to inhale them more frequently and deeply.
Compared with nonfiltered cigarettes, filtered ones allow
smaller particles to get into the peripheral parts of the lung
and cause different histologic subtypes of cancer.15,16
Those who stop smoking almost immediately lower the
risk of cancer, despite the fact that some carcinogeninduced gene mutations persist for years.
More than 80% of adult American smokers begin
smoking before the age of 18.17 Communicating health
messages to the pediatric and adolescent population is
therefore a major public health challenge. Studies show
that a physicians simple advice to avoid or quit smoking
can positively influence patients.
Smoking is an addiction. It is easier for light smokers,
the less addicted, to quit. Experts believe heavy smokers
generally need an intensive, broad-based cessation program that includes counseling, behavioral strategies, and
drug therapy, such as nicotine-replacement therapy and
bupropion. Most Americans who successfully quit smoking
do so on their own without participation in an organized

cessation program, but such strategies can be helpful for

some. Smokers who stop completely are more likely to be
successful than smokers who gradually reduce the number
of cigarettes smoked or change to cigarettes with lower
amounts of tar or nicotine. The smoker who is quitting
goes through a process with identifiable stages that include
contemplation of quitting, an action phase during which
the smoker quits, and a maintenance phase.
Much of the literature focuses on the risks of cigarette
smoking. Cigar smokers do not inhale, but the health
risks associated with cigars are similar to those of cigarettes, especially the risks of oropharyngeal cancers.18
Smokeless tobacco is the fastest growing segment of the
tobacco industry and represents a serious health risk.
Chewing tobacco has been linked to dental caries, gingivitis, oral leukoplakia, and oral cancer. The systemic
effects of smokeless tobacco may increase the risks of lung
cancer. The nitrosamines found in this product cause
lung cancer in animal studies. Esophageal cancer is
linked to the carcinogens in tobacco that dissolve in
saliva, are swallowed, and then come into contact with
the esophagus.

Diet Modification
Rates of cancer of the breast, colon, endometrium, and
prostate are higher in the Western cultures than in cultures in the Far East or even in the old Eastern bloc. This
observation is the basis for the belief that dietary modification can significantly lower cancer risk for individuals in
the United States.19
Despite correlative data, the dietary fat-cancer hypothesis has not been definitively demonstrated. Case-control
and cohort epidemiologic studies yield conflicting results.
In addition, diet is a highly complex exposure to many
nutrients and chemicals. Low-fat diets may render some
protection through anticarcinogens found in vegetables,
fruits, legumes, nuts, and grains. Potentially protective
substances found in foods include phenols, sulfur-containing compounds, flavones, and fiber.
In epidemiologic (observational) studies, dietary fiber
intake is inversely associated with the risk of colonic
polyps and invasive cancer of the colon.20,21 Although
there is a correlation, the causal mechanism is unknown.
It has been suggested that fiber binds oxidized bile acids.
High-fiber diets may also protect against breast and
prostate cancers by absorbing and inactivating dietary
compounds that promote carcinogenesis through estrogenic and androgenic activity. In no prospective clinical
trial has it been demonstrated that cancer can be prevented through lowering dietary fat or increasing fiber
intake.

A simple way to decrease the proportion of dietary fat

and increase fiber is for an individual to consume at least
five to nine servings of fruits and vegetables per day.
Although the cancer-prevention benefits are theoretical
and not fully demonstrated, such a diet does lower the risk
of cardiac disease. Vitamin, mineral, or nutritional supplements in amounts greater than that provided by a good
diet have not been demonstrated to be of value.

Sun Avoidance
Results of epidemiologic studies show a correlation
between the risk of nonmelanoma skin cancers (basal and
squamous cell) and cumulative exposure to ultraviolet radiation. Possible risk factors for melanoma include a propensity to sunburn, a large number of benign melanocytic nevi,
and atypical nevi. A history of severe sunburns, especially
in childhood and adolescence, is associated with increased
risk of melanoma in adulthood. Reduction of sun exposure
through the use of protective clothing and changing ones
pattern of outdoor activities to avoid the most intense and
direct sunlight have been advocated as ways to reduce the
risk of skin cancer. The use of sunscreens is controversial.
Whereas sunscreens decrease the risk of actinic keratoses,
the precursor to squamous cell skin cancer, findings from
epidemiologic studies indicate that an increased risk of
melanoma is associated with sunscreen use. Sunscreen use
may encourage prolonged exposure to the melanomainducing sun rays by protecting against sunburn.22

Chemoprevention
Cancers are prevented through chemoprevention or, in
certain cases, through surgical removal of the organ at risk.
Cancer chemoprevention is the use of natural or synthetic
chemical agents to reverse, suppress, or prevent carcinogenesis before the development of an invasive malignant
process.23 Although the concept that pharmacologic
agents can prevent a cancer is relatively new, the idea that
a compound can prevent chronic disease is not.
Antihypertensive agents are used to prevent heart disease,
kidney disease, and stroke. Lipid-lowering drugs are
prescribed to prevent coronary artery disease.
The initial genetic changes of carcinogenesis are
termed initiation. This alteration can be inherited or
acquired. Acquired genetic damage is the result of physical, infectious, or chemical carcinogens (Table 1-3). The
influences that cause the initiated cell to change phenotypically are called promoters. Known promoters include
androgens linked to prostate cancer and estrogen linked to
breast and endometrial cancers. The distinction between
the initiator and promoter can sometimes blur; some
Epidemiology and Prevention |

Table 1-3. Inhibitors and Promotors of Cancer

Carcinogen
Alkylating agents
Androgens
Aromatic amines (dyes)
Arsenic
Asbestos
Benzene
Chromium
Diethylstilbestrol (prenatal)
Epstein-Barr virus
Estrogens
Ethyl alcohol
Helicobacter pylori
Hepatitis B virus
Hepatitis C virus
Human T-cell leukemia
(HTLV)-1 virus
Human immunodeficiency
virus (HIV)
Immunosuppressive agents
(azathioprine, cyclosporine,
corticosteroids)
Nitrogen mustard gas
Nickel dust
Phenacetin
Polycyclic hydrocarbons
Schistosomiasis
Alkylating agents
Sunlight (ultraviolet)
Tobacco (including smokeless)
Vinyl chloride

Associated Cancer or Neoplasm

Acute myeloid leukemia, bladder
Prostate
Bladder
Lung, skin
Lung, pleura, peritoneum
Acute myelocytic leukemia
Lung
Vagina (clear cell)
Burkitts lymphoma, nasopharynx
Endometrium, liver, breast
Liver, esophagus, head and neck
Gastric
Liver
Adult T-cell leukemia, lymphoma
Non-Hodgkins lymphoma, Kaposis
sarcoma, squamous cell cancer
Non-Hodgkins lymphoma

Lung, head and neck, nasal sinuses

Lung, nasal sinuses
Renal pelvis and bladder
Lung, skin (especially squamous cell
of the scrotum)
Bladder (squamous cell)
Acute myeloid leukemia, bladder
Skin (squamous and melanoma)
Upper aerodigestive tract, bladder
Liver (angiosarcoma)

These agents are thought to act as cancer initiators or promoters for the
cancers with which they have been associated.

components of cigarette smoke are referred to as complete

carcinogens and serve as both initiator and promoter.
Cancer can be prevented or controlled through interference with the factors causing disease initiation, promotion, or progression. Many compounds are of interest in
chemoprevention. Most are classified as anti-inflammatory
agents, antioxidants, or hormone antagonists.
A long-term, randomized, placebo-controlled clinical
trial is generally necessary to establish the efficacy of a
chemopreventive agent, and several large clinical trials
have been completed.13,24,25 Almost all current chemopreventive interventions now under study are being assessed
to determine if they decrease the so-called period prevalence of the disease, that is, to learn whether the intervention lowers risk while the intervention is used. It is difficult
to design a study that shows that intervention decreases
lifetime risk of disease.

Cancers of the Lung, Head, and Neck

Tobacco smoking is the major cause of squamous
cancers of the lung, head, and neck. The risk of a second
cancer of the lung, head, or neck is high, as much as 5% per
year of smoking, for patients cured of these diseases. This is
due to field cancerization, meaning the carcinogens in
tobacco smoke affect all tissues exposed to them. Even after
6

| Epidemiology and Prevention

smoking cessation, the tissues that have come in contact

with smoke have some molecular damage. These so-called
cured patients are candidates for chemoprevention trials.26
In small, randomized, placebo-controlled trials of short
duration, isotretinoin (13-cis retinoic acid) caused regression of oral leukoplakia, a premalignant lesion. The lesions
recur when the therapy is withdrawn, and the findings of a
recent large trial suggested that isotretinoin might even be
harmful for current smokers.27,28 Premalignant lesions in
the oropharyngeal area have also responded to betacarotene, retinol (vitamin A), alpha-tocopherol (vitamin
E), and selenium in randomized trials. In other clinical trials, adjuvant isotretinoin reduced the number of second
primary tumors for patients cured of localized head and
neck cancer, but it did not improve overall survival.
Several large-scale studies have been launched to assess
potential chemopreventive agents for patients at high risk
for lung cancer. The Alpha-Tocopherol, Beta-Carotene
Cancer (ATBC) Prevention Trial 29 and the Beta-Carotene
and Retinol Efficacy Trial (CARET)30 were prevention trials
that showed the importance of testing even seemingly harmless chemoprevention agents, such as vitamins, before widespread use. The results of both trials are in contrast to
numerous observational studies. The ATBC Prevention Trial
enrolled Finnish male smokers between the ages of 50 and
69 years, with an average of 36 pack-years of smoking.
Participants received either alpha-tocopherol, beta-carotene,
both, or placebo in a randomized, two-by-two factorial
design. After a median follow-up of six years, there was a
significant increase in lung cancer incidence and mortality
for the subjects who received beta-carotene. There was no
evidence that alpha-tocopherol had any impact on lung cancer mortality. CARET enrolled 17,000 smokers and workers
exposed to asbestos. Entrants were randomly assigned to four
arms and received either beta-carotene, retinol, both, or
placebo in a two-by-two factorial design. The results of the
trial demonstrated a 28% increase in lung cancers and 17%
more deaths in the subjects taking beta-carotene. The study
was closed early due to this finding. A third study, the
Physicians Health Trial, which treated nonsmoking physicians with beta-carotene, did not show a harmful effect.31
An important serendipitous finding occurred in the
ATBC Prevention Trial. The period prevalence of prostate
cancer was reduced by one-third for the participants taking
alpha-tocopherol compared with participants not taking
the agent.

Gastric Cancers
Excessive salt intake, limited consumption of fresh
fruits and vegetables, and infection with Helicobacter pylori
are associated with an increased risk of gastric cancer.

Experimental evidence of causality is scarce. A diet high

in whole-grain cereals, carotenoids, vitamin C, allium
compounds, and green tea are associated with a reduced
risk of this type of cancer. The results of a chemoprevention trial in China showed a significant reduction in
mortality related to gastric cancer after supplementation
with beta-carotene, vitamin E, and selenium.32,33 There is
insufficient evidence that treatment of H. pylori infection
prevents either gastric cancer or precancerous lesions.34

Colon Cancer
Findings from epidemiologic studies suggest that nonsteroidal anti-inflammatory agents, such as piroxicam,
sulindac, and aspirin, have protective effects against adenomatous polyps and invasive cancer. The results of
prospective intervention trials have demonstrated some
positive effects on the prevention of polyps but have yet to
show that these agents prevent colon cancer. In a placebocontrolled trial, high-dose celecoxib, a cyclo-oxygenase-2
(COX-2) inhibitor, was found to reduce the occurrence of
colorectal polyps in patients with familial adenomatous
polyposis. Trials to assess COX-2 inhibitors and other nonsteroidal anti-inflammatory agents for the prevention of
sporadic colorectal cancers are underway.20
The results of epidemiologic studies indicate that diets
high in calcium are associated with a lower risk of colon
cancer. Evidence from prospective randomized studies
shows that calcium supplementation decreases the risk of
recurrence of adenomatous polyps by about 20%. Calcium
binds bile and fatty acids, reducing intraluminal exposure
to compounds that cause hyperproliferation of the colonic
epithelium.
The Womens Health Initiative was a prospective, randomized study involving postmenopausal women randomly assigned to either combination therapy with
estrogen plus progestin or placebo. The rate of colorectal
cancer was lower for women taking the study drug compared with women taking placebo. The effect is, however,
offset by the life-threatening cardiovascular and breast
cancer risks associated with treatment with estrogen plus
progestin.35
Colectomy is used as a preventive measure for individuals at extremely high risk of colon cancer as a result of a
history of ulcerative colitis or a genetic predisposition to
the disease, such as familial adenomatous polyposis.36

The recent reductions in the incidence of hepatoma

suggest some success.37

Breast Cancer
Tamoxifen has mixed estrogenic and antiestrogen
activities. It acts as an estrogen agonist in the
endometrium and bone and as an estrogen antagonist
in breast tissue. It also upregulates transforming growth
factor-beta, which decreases breast cell proliferation. In
randomized, placebo-controlled trials to assess tamoxifen
as an adjuvant treatment for patients with early-stage
breast cancer, women receiving this drug were found to
have one-third fewer new cancers in the contralateral
breast than women receiving placebo. The Breast Cancer
Prevention Trial was a randomized, placebo-controlled
study of more than 13,000 women at high risk of breast
cancer. After a median treatment of 69 months, tamoxifen
was found to decrease the period risk of breast cancer by
49%. It was also associated with a reduction in bone fractures and a small increase in risk of endometrial cancer,
stroke, pulmonary emboli, and deep vein thrombosis.24
A trial to compare tamoxifen with another selective
estrogen-receptor modulator, raloxifene, is underway.
Raloxifene may have similar efficacy as a breast cancer
preventive agent with less risk of endometrial cancer.
Prophylactic bilateral mastectomy to prevent breast
cancer has not been well assessed. In a prospective series
of 139 women with BRCA1 and BRCA2 mutations, 76
chose prophylactic bilateral mastectomy and 63 chose
close surveillance. At three years, there was no breast cancer diagnosed in those who chose surgery; eight women in
the surveillance group had been diagnosed with breast
cancer. This study is small, of short duration, and, by
design, prone to selection biases. It is fair to say that the
short-term risk of breast cancer appears to be lower for
women with certain BRCA1 and BRCA2 mutations who
choose prophylactic mastectomy. Because this surgery
leaves some breast tissue behind, a patients risk is not
reduced to zero. Retrospective analysis of mastectomies
for women at high risk because of family history suggests
that prophylactic mastectomy can lead to a 90% reduction in risk. The impact of this procedure on mortality for
women at high risk of breast cancer is not completely
known.38

Prostate Cancer
Liver Cancer
Hepatitis B-induced hepatoma is one of the most common cancers diagnosed in Asia. The hepatitis B vaccine
has been advocated for its ability to prevent the disease.

Androgens stimulate prostate-cell proliferation and,

in laboratory animals, cause prostate carcinogenesis.
Finasteride decreases androgenic stimulation of the
prostate by inhibiting the production of 5-alpha-reductase.
Epidemiology and Prevention |

This enzyme, which is found in high amounts in the

prostate, converts testosterone to the more potent dihydrotestosterone. Finasteride was tested as a preventive
agent for prostate cancer in the Prostate Cancer
Prevention Trial, a 10-year, randomized, placebo-controlled study involving 18,000 men 55 years and older.
Results of the study showed that this drug was associated
with a 25% reduction in the risk of prostate cancer during the treatment period. There are some concerns
regarding the histologic grade and aggressiveness of the
cancers that developed while patients were treated with
finasteride. The results of this trial are still being
analyzed.24
Findings from epidemiologic studies indicate a correlation between high intake of antioxidants, such as selenium
and vitamin E, and lower risk of prostate cancer. The results
of a small, randomized skin cancer prevention trial of selenium compared with placebo show a significant decrease in
the number of prostate cancers in men treated with selenium
compared with men receiving placebo. Among the 843 men
who began the study with a serum prostate-specific antigen
(PSA) level of less than 4 ng/mL, 16 prostate cancers were
diagnosed in the placebo group compared with four in the
treatment group.39 Eight years into the ATBC Prevention
Trial , which enrolled 29,000 men, there were 99 cases of
prostate cancer among men receiving vitamin E and 151
cases among men taking placebo. The cancers diagnosed
were almost all detected as a result of the work-up of symptoms as there is no prostate cancer screening in Finland.40
The prostate cancer findings in both of these trials were
incidental results of a secondary analysis. A prospective,
randomized, placebo-controlled trial to properly assess
these drugs is ongoing. The trial will involve more than
32,000 men.

Gynecologic Cancer
There are limited epidemiologic data to suggest that
the use of oral contraceptives by women in their 20s and
30s decreases the risk of ovarian cancer in later life.41
Laser ablation, conization, or hysterectomy are used to
treat cervical dysplasia or intraepithelial neoplasia, precursor to cervical cancer. In the future, vaccines for human
papillomavirus may prevent cervical cancer.

Cancer Screening
Screening is an attempt to detect disease early in
asymptomatic individuals, with the goal of intervening
and decreasing morbidity and mortality. A screening test
is not generally diagnostic for cancer. It is usually per8

| Epidemiology and Prevention

KEY POINTS
Cancer screening is far more complicated than
just doing a test and finding a localized cancer.
The biases of screening are selection, lead time,
length, and overdiagnosis. These biases can make
a screening test appear beneficial when there is
actually net harm.
A number of expert organizations have rigorously
accessed the scientific data supporting screening
methods and made recommendations.

formed to determine whether cancer might be present and

if additional testing, including a biopsy and staging, is
necessary. To be of benefit, screening must lead to earlier
treatment that must offer a better outcome compared with
treatment at the onset of symptoms. The ideal evaluation
of a screening technology is through the assessment of
disease-specific and overall mortality in a randomized
clinical trial.
Early detection of an apparently localized cancer does
not in and of itself confer benefit. There are screening tests
for some diseases that have been found to be of no benefit,
such as chest x-ray screening for lung cancer, which was
used in the 1960s,42,43 or urine screening for vanillylmandelic acid to detect neuroblastoma.44 A number of common screening tests used in the United States are of
undetermined benefit.
Screening tests and the appropriate use of such tools
should be carefully evaluated before they are widely
adopted. Survival time after diagnosis is a poor and misleading measure of the efficacy of a test outside of a randomized clinical trial, in which the survival of individuals
screened is compared with the survival of individuals
randomly assigned to a nonscreening arm.

Assessment of Screening Tests

A screening intervention is best evaluated in a population-based, randomized, controlled screening trial with
cause-specific mortality as the endpoint. Studies showing a
reduction in the incidence of advanced-stage disease, an
improved survival, or a shift in the stage of disease are
weaker and offer potentially misleading evidence of benefit. These latter criteria by themselves are not sufficient to
establish the value of a screening test.45
Because the gold-standard randomized screening trials
are large (often involving thousands of people) and last for
years, less definitive study designs are often used to estimate the efficacy and effectiveness of screening practices.

In order of strength of evidence, efficacy can by assessed

from
Findings of internally controlled trials in which intervention-allocation methods other than randomization are
used, such as allocation determined by birth date or date
of clinic visit
Results of cohort or case-control analytic observational
studies
Findings of multiple time-series study with or without the
intervention
Opinions of respected authorities based on clinical experience, descriptive studies, or consensus reports of experts
The latter form of evidence is the weakest, as even experts
can easily be misled by the biases described.

Potential Biases
Screening is subject to several biases, including lead
time, length, and selection bias, the influences of which
are lessened in a randomized trial. All can lead one to
believe that there is a benefit to a screening test when, in
truth, there is none; there can even be a net harm.
Screening, whether beneficial or not, will increase the
number of specific cancers diagnosed. It can also produce
a shift in stage that will improve survival statistics without reducing mortality (i.e., the number of deaths of a
given cancer per number of people at risk of the disease).
In such a case, apparent duration of survival, measured
from the date of diagnosis, would increase without lives
being saved or life expectancy being changed.
When pure lead-time bias occurs, survivalthe time
from diagnosis to deathis increased, but treatment does
not prolong life. Patients do not live longer, they are
merely diagnosed at an earlier date. The screening test
only prolongs the time the individual is aware of the
disease and the time as a patient.
Length bias occurs when slow-growing, less aggressive
cancers are detected during screening. Cancers diagnosed

as the result of the onset of symptoms between scheduled

screenings are, on average, more aggressive, and treatment
outcomes are not as favorable. An extreme form of length
bias is termed overdiagnosis, or detection of pseudodisease.
Some undetected slow-growing tumors fulfill the histologic
criteria for cancer but will never be clinically significant or
cause death. This phenomenon is compounded by the fact
that the most common cancers are most frequent among
older people. Other competing causes of death, such as
heart disease, become more relevant.
Selection bias must be considered when assessing the
results of any clinical trial. The group most likely to seek
entry in the study may differ from the general population to
which the study result might be applied. In a screening trial,
the individuals may have volunteered because of a particular risk factor not found in the larger population, such as a
strong family history. In general, volunteers are more health
conscious and are likely to have a better prognosis or lower
mortality rate irrespective of actually being screened; this
trend is referred to as the healthy volunteer effect.

Potential Harmful Effects

Subjects can be harmed as a result of screening. A harmful effect can be associated with the test itself, the work-up
of positive results of screening tests (both true-positive and
false-positive results), and injuries from the treatment of
true-positive results. Screening can detect some cancers that
would never have caused medical problems. The unnecessary treatment of these cancers can be harmful.

Accuracy
The accuracy of any medical test is described using
four indices: sensitivity, specificity, positive predictive
value, and negative predictive value, and the results of
screening tests can be classified into four categories
(Tables 1-4 and 1-5). Sensitivity and specificity are relatively independent of the underlying prevalence or risk of
the population being screened, but the positive and

Table 1-4. Indices for Describing the Accuracy of Screening Tests

Term
Sensitivity
Specificity
Positive
predictive value
Negative
predictive value

Definition
Proportion of people with the disease who
have a positive result on a screening test
Proportion of people who do not have the disease and
who have a negative result on a screening test
Proportion of people with a positive result on a
screening test and actually have the disease
Proportion of people who have a negative results on
a screening test and truly do not have the disease

Ability of Test

Equation*

To detect disease when it is present

A / (A + C)

To correctly identify the absence of disease

D / (B + D)

To accurately predict the presence of disease

A / (A + B)

To accurately predict the absence of disease

D / (C + D)

*A = true-positive result, B = false-positive result, C = false-negative result, D = true-negative result. Refer to Table 1-5 for definitions of these types of test results.

Epidemiology and Prevention |

Table 1-5. Types of Results on Screening Tests

Term

Test
Result

Disease
Present

True positive (A)

False positive (B)
False negative (C)
True negative (D)

Positive
Positive
Negative
Negative

Yes
No
Yes
No

negative predictive values are highly dependent on prevalence (Table 1-6). In other words, screening is most beneficial, efficient, and economical when targeting a cancer
common to the general population or groups with a high
prevalence (or high risk) of the specific disease being
screened. For a screening test to be valuable, high specificity is needed. Sensitivity need not be extremely high
(Table 1-6).46

Screening for Specific Cancers

Results from studies are convincing that screening
for cervical, colon, and breast cancers is beneficial at
certain ages for people at normal risk. Although special
surveillance of individuals at high risk for some specific
cancers because of family history or genetic risk may be
prudent, few studies have been carried out to assess its
true worth.
A number of organizations have evaluated certain
screening tests and considered whether to endorse routine
use of such measures. The U.S. Preventive Services Task
Force (www.ahcpr.gov/clinic/uspstfix.htm) and the
Canadian Task Force on Preventive Health Care
(www.ctfphc.org) published screening recommendations
after a rigorous review process. Each recommendation is
made with a thorough, structured evaluation of the literature by screening experts. The American Cancer Society
(www.cancer.org) publishes the most commonly quoted
screening guidelines (Table 1-7),47 and the National

Cancer Institute (NCI) publishes extensive reviews of the

screening literature (http://cancernet.nci.nih.gov).
Breast Cancer
Studies of breast self-examination have not shown that
this practice decreases mortality.48,49 The results of the
largest randomized, controlled study of breast self-examination reported to date showed both an increased rate of
biopsy and enhanced detection of benign lesions but little
or no stage shift and no reduction in breast cancer mortality.
Findings from several studies indicate that screening
women with normal risk over the age of 50 using mammography alone or mammography and clinical breast
examination every one to two years decreases mortality by
20% to 30%. Each trial has been criticized for a certain
aspect of its design but there is some power in the consistency of the observations.50
Experts disagree on whether women of average risk
between the ages of 40 and 49 benefit from screening
(Table 1-7). A meta-analysis of eight large randomized trials showed no benefit from mammography screening for
women in this age group, when assessed five to seven years
after trial entry.51 There was a small benefit for women 10
to 12 years after entry, which may have been the result of
screening these women after they turned 50. In randomized screening studies of women 50 to 69 years old,
the decline in benefits begins about five years after the
initiation of screening.
The results from outcome studies show that there is
substantial variation among American radiologists with
regard to recommendations for additional testing or biopsy.
This disparity is especially variable among younger
women. In large cohorts, nearly half of all women between
the ages of 40 and 49 screened annually for 10 years will
have false-positive mammograms necessitating repeat
mammography, ultrasound examination, or biopsy.

Table 1-6. Influence of Prevalence on Predictive Value

Positive Predictive Value* =

(Sensitivity)(Prevalence)
(Sensitivity)(Prevalence)  (1-Specificity)(1-Prevalence)

Positive Predictive Value

for a Disease with Prevalence of 5 Affected
Individuals per 1,000 Population

Positive Predictive Value

for a Disease with Prevalence of 1 Affected
Individual per 10,000 Population

Sensitivity
0.8
Specificity
0.95
0.999

7%
80%

Sensitivity
0.95
9%
83%

Specificity
0.95
0.999

0.8

0.95

0.2%
7%

0.2%
9%

*The positive predictive value is expressed as a percentage. It is influenced by the sensitivity and specificity of the screening test and the prevalence of the
disease being screened for. The positive predictive value is particularly influenced by the specificity of the screening test at a given prevalence for relatively
uncommon diseases, such as cancer.

10

| Epidemiology and Prevention

Table 1-7. Screening Recommendations for Asymptomatic Subjects with Normal Risk*
Test or Procedure

U.S. Preventive Services Task Force

Sigmoidoscopy

Periodically for individuals older

than 50 years; not recommended for
those younger than 50 years
Yearly for individuals 50 years old and more
No recommendation

Fecal occult-blood testing

Digital rectal examination

Canadian Task Force

on Preventive Health Care

American Cancer Society

Insufficient evidence

Every 3-5 yrs. for individuals

50 years old and more

Insufficient evidence
Poor evidence to include or
exclude for men older than 50 years
Recommendation against
Fair evidence to include in the
examination of sexually active
women

Yearly for individuals 50 years old and more

Yearly for individuals 40 years old and more

Prostate-specific antigen
Pap test

Insufficient evidence to recommend

Every 1-3 yrs. for women
between the ages of 18 and 65

Pelvic examination

Not considered

Endometrial tissue sampling

Do not recommend; advise adnexal

palpation during examination for other
reasons
Not considered

Breast self-examination

No recommendation

Breast clinical examination

Yearly for women 50 years and older

Insufficient evidence to make

a recommendation
Yearly for women age 50 and older

Mammography

Every 1-2 yrs. for women between

the ages of 40 and 75 (labeled as
fair evidence)

Not considered

Yearly for women between the ages

of 50 and 69

Yearly for individuals 50 years old and more

For women with a uterine cervix, starting
3 yrs. after beginning intercourse or by
age 21. Yearly for standard Pap test; every
2 yrs. for liquid-based cytologic tests
Every 1-3 yrs. for women between
the ages of 18 and 40 years; yearly for those
over 40
At menopause, if obese, or has a history of
unopposed estrogen use
Monthly for individuals 20 years old and older
Every 3 yrs. for women 20 to 40; yearly
for women older than 40
Yearly for women 40-49 years of age; every
1-2 yrs. for women 50 years
and older

*These recommendations were made for the general population; that is, asymptomatic people who have no risk factors, other than age or gender, for the targeted
condition.

Mammography may not be as sensitive for detecting

breast cancers among women with BRCA1 or BRCA2
mutations. This weakness may be because such women
tend to have cancers at a younger age, when mammography is less sensitive.

results for the last 10 years may choose to stop cervical

cancer screening.
Colorectal Cancer
Several methods have been considered for colorectal
cancer screening

Cervical Cancer
No randomized clinical trial has been completed to
determine whether cervical cancer screening with the Pap
test reduces mortality, but findings from several cohort and
case-control studies have shown the utility of the Pap test.
Regular Pap testing is recommended for women who are
sexually active or over the age of 18. The recommended
interval for Pap screening ranges from one to three years.
An upper age limit at which screening ceases to be
effective is not known.
The American Cancer Society recently revised its
screening guidelines to recommend that screening start
about three years after a woman begins having vaginal
intercourse, but no later than age 21.52 The society recommends that cervical screening be performed annually in
the case of regular Pap tests or every two years in the case
of liquid-based cytologic tests. Women who have had normal results on three consecutive tests may be screened
every two to three years. Women with certain risk factors,
such as infection with the human immunodeficiency virus
(HIV) or a weakened immune system, might be screened
more frequently. Women 70 years and older who have had
normal results on three or more Pap tests and no abnormal

Fecal occult-blood testing

Rigid and flexible sigmoidoscopy
Colonoscopy
Digital rectal examination
Radiographic barium contrast studies

The results of randomized studies indicate that annual

fecal occult-blood testing using rehydrated stool
specimens can reduce mortality from colorectal cancer by
one-third.53 Rehydration increases sensitivity but lowers
specificity. The rate of false-positive results for rehydrated fecal occult-blood testing is 1% to 5%. Less than
10% of individuals with occult blood in stool have
cancer, and approximately one-fifth to one-third have
adenomas.
Findings from two case-control studies show that
screening sigmoidoscopy is associated with a decrease in
mortality among subjects 50 years and older.54 The results
from other studies show that approximately one-half of all
polyps are found with the 35-centimeter flexible scope and
two-thirds to three-quarters are found with a 60-centimeter scope. Diagnosis of polyposis by sigmoidoscopy should
lead to evaluation of the entire colon with colonoscopy.
Epidemiology and Prevention |

11

The most efficient interval for screening individuals with a

history of polyps is unknown.
Although no study results have clearly demonstrated
benefit, it is regarded as prudent to perform colonoscopy
on subjects at extremely high risk, such as those with a
genetic predisposition to colorectal cancer and those with
inflammatory bowel disease. Little information is available
on the utility of the digital rectal examination or barium
enema as screening tools. The barium enema is insensitive
for flat lesions and small polyps.55,56
Lung Cancer
Screening for lung cancer with chest x-ray and sputum
cytologic testing was evaluated in four randomized lung
cancer screening trials in the 1960s and 1970s. No reduction in lung cancer mortality was seen in these studies. An
insignificant increase in mortality was found for the
screened arms in these studies.42
Studies have shown that spiral computerized tomography (CT) can diagnose lung cancers at early stages, but it
is unclear whether it will save lives. This technology is
being evaluated in a large, randomized clinical trial. Spiral
CT can also detect many benign processes that cause noncalcified lung radiodensities; these are false-positive findings. Although it is not known whether spiral CT saves
lives, it increases the number of lesions diagnosed and,
thus, will increase the number of diagnostic and therapeutic procedures performed.57,58
Ovarian Cancer
Adnexal palpation, transvaginal ultrasound, and measurement of serum CA-125 have been considered for ovarian cancer screening. No randomized prospective trial of
screening for ovarian cancer has been completed. The
results of such studies could lead to futile invasive diagnostic testing that might include laparotomy. In one nonrandomized clinical trial, 4% of 915 adult women had an
abnormal serum CA-125 level (greater than 35 U/mL).59
According to this finding, the prevalence of an abnormal
CA-125 level would be more than 3,900 in 100,000
women. The prevalence of ovarian cancer in the female
adult population is approximately 20 per 100,000 cases. If
100,000 women are screened, the test will identify 3,900
individuals who would have additional, usually invasive,
evaluation in order to identify 20 cases of ovarian cancer.
Some of these 3,900 women would have an ultrasound and
others would have an exploratory laparotomy. A substantial proportion of the 20 women diagnosed with ovarian
cancer will still have incurable advanced disease and, thus,
not benefit from screening. A National Institutes of
Health consensus conference in 1994 concluded that rou12

| Epidemiology and Prevention

tine screening for ovarian cancer was not indicated

for women with normal risk but might be worthwhile for
families with genetic ovarian cancer syndromes.60
Prostate Cancer
The digital rectal examination and measurement of
serum PSA are commonly used in the United States, despite
most professional organizations advising caution in the use of
such screening tools. No well-designed trial has been completed to test the true benefit of screening and treatment of
prostate cancer, but trials are in progress. Prostate cancer is
prone to lead-time bias, length bias, and overdiagnosis.
Whereas screening with PSA and digital rectal examination
clearly detects many asymptomatic cancers, its ability to reliably distinguish tumors that could be lethal but are still curable from those that are of little or no threat to health is
limited. It has been estimated that more than 30% of localized prostate cancers diagnosed during screening are indolent
and clinically insignificant. Treatment of screen-detected
cancers may cause morbidity, such as impotence and urinary
incontinence, and carries a small risk of death.61
Most expert organizations actually recommend against
screening for prostate cancer. In late 2002, the U.S.
Preventive Services Task Force examined the evidence in
support of screening and found there was insufficient evidence to recommend it. The American Cancer Society
and the American Urologic Association recommend that
men over the age of 50 with normal risk be offered screening and be allowed to make a choice after being informed
of its potential risks and benefits (Table 1-7).
Skin Cancer
No randomized study has been conducted to assess
whether screening for skin cancer decreases mortality.
Screening programs in Scotland and Australia may have
caused the stage shift in diagnosed melanomas.62 These
programs also may reinforce sun avoidance and other
prevention behaviors.

REFERENCES
1. Ries LAG, Eisner MP, Kosary CL, et al, eds. SEER
Cancer Statistics Review, 1975-2000. Bethesda, Md:
National Cancer Institute; 2003. Available at: http://seer.
cancer.gov/csr/1975_2000. Accessed April 26, 2004.
2. Jemal A, Murray T, Samuels A, et al. Cancer statistics,
2003. CA Cancer J Clin. 2003;53:5-26.
3. Welcsh PL, King MC. BRCA1 and BRCA2 and the
genetics of breast and ovarian cancer. Hum Mol Genet.
2001;10:705-713.
4. Brawley OW. Population categorization and cancer
statistics. Cancer Metastasis Rev. 2003;22:11-19.

5. Piura B, Rabinovich A, Yanai-Inbar I. Three primary

malignancies related to BRCA mutation successively
occurring in a BRCA1 185delAG mutation carrier. Eur J
Obstet Gynecol Reprod Biol. 2001;97:241-244.
6. Ganguly A, Leahy K, Marshall AM, et al. Genetic testing for breast cancer susceptibility: frequency of BRCA1
and BRCA2 mutations. Genet Test. 1997;1:85-90.
7. Robson M, Gilewski T, Haas B, et al. BRCA-associated
breast cancer in young women. J Clin Oncol. 1998;16:
1642-1649.
8. Kulldorff M, Feuer EJ, Miller BA, Freedman LS. Breast
cancer clusters in the northeast United States: a geographic analysis. Am J Epidemiol. 1997;146:161-170.
9. Gilligan MA, Kneusel RT, Hoffmann RG, et al.
Persistent differences in sociodemographic determinants
of breast conserving treatment despite overall increased
adoption. Med Care. 2002;40:181-189.
10. Carney ME, Lancaster JM, Ford C, et al. A populationbased study of patterns of care for ovarian cancer: who is
seen by a gynecologic oncologist and who is not? Gynecol
Oncol. 2002;84:36-42.
11. Lu-Yao G, Albertsen PC, Stanford JL, et al. Natural
experiment examining impact of aggressive screening and
treatment on prostate cancer mortality in two fixed
cohorts from Seattle area and Connecticut. BMJ. 2002;
325:740.
12. Shavers VL, Brown ML. Racial and ethnic disparities
in the receipt of cancer treatment. J Natl Cancer Inst.
2002;94:334-357.
13. Greenwald P. Cancer prevention clinical trials. J Clin
Oncol. 2002;20:14S-22S.
14. Shields PG. Tobacco smoking, harm reduction, and
biomarkers. J Natl Cancer Inst. 2002;94:1435-1444.
15. Wynder EL, Muscat JE. The changing epidemiology of
smoking and lung cancer histology. Environ Health
Perspect. 1995;103:143-148.
16. Carpenter CL, Jarvik ME, Morgenstern H, et al.
Mentholated cigarette smoking and lung-cancer risk. Ann
Epidemiol. 1999;9:114-120.
17. Bowen DJ, Orlandi MA, Lichtenstein E, et al.
Intervention effects on youth tobacco use in the Community
Intervention Trial (COMMIT). Tob Control. 2002;11:382.
18. Baker F, Ainsworth SR, Dye JT, et al. Health risks associated with cigar smoking. JAMA. 2000;284:735-740.
19. Greenwald P, Clifford CK, Milner JA. Diet and cancer
prevention. Eur J Cancer. 2001;37:948-965.
20. Giovannucci E. Modifiable risk factors for colon
cancer. Gastroenterol Clin North Am. 2002;31:925-943.
21. Schatzkin A, Lanza E. Polyps and vegetables (and fat,
fibre): the polyp prevention trial. IARC Sci Publ. 2002;
156:463-466.

22. Autier P, Severi G, Boyle P, Dore JF. RESPONSE more

about: sunscreen use and duration of sun exposure: a
double-blind, randomized trial. J Natl Cancer Inst. 2000;
92:1532-1533.
23. Greenwald P. Cancer chemoprevention. BMJ. 2002;
324:714-718.
24. Thompson IM, Goodman PJ, Tangen CM, et al. The
influence of finasteride on the development of prostate
cancer. N Engl J Med. 2003;349:215-224.
25. Fisher B, Costantino JP, Wickerham DL, et al.
Tamoxifen for prevention of breast cancer: report of the
National Surgical Adjuvant Breast and Bowel Project P-1
Study. J Natl Cancer Inst. 1998;90:1371-1388.
26. Cohen V, Khuri FR. Progress in lung cancer chemoprevention. Cancer Control. 2003;10:315-324.
27. Lippman SM, Spitz MR. Lung cancer chemoprevention:
an integrated approach. J Clin Oncol. 2001;19:74S-82S.
28. Lippman SM, Lee JJ, Karp DD, et al. Randomized
phase III intergroup trial of isoretinoin to prevent second
primary tumors in stage I non-small-cell lung cancer. J Natl
Cancer Inst. 2001;93:605-618.
29. Virtamo J, Pietinen P, Huttunen JK, et al. Incidence
of cancer and mortality following alpha-tocopherol and
beta-carotene supplementation: a postintervention followup. JAMA. 2003;290:476-485.
30. Omenn GS, Goodman GE, Thornquist MD, et al. Risk
factors for lung cancer and for intervention effects in
CARET, the Beta-Carotene and Retinol Efficacy Trial. J
Natl Cancer Inst. 1996;88:1550-1559.
31. Cook NR, Le IM, Manson JE, et al. Effects of betacarotene supplementation on cancer incidence by baseline
characteristics in the Physicians Health Study (United
States). Cancer Causes Control. 2000;11:617-626.
32. Taylor PR, Qiao YL, Abnet CC, et al. Prospective
study of serum vitamin E levels and esophageal and gastric
cancers. J Natl Cancer Inst. 2003;95:1414-1416.
33. Taylor PR, Wang GQ, Dawsey SM, et al. Effect of
nutrition intervention on intermediate endpoints in
esophageal and gastric carcinogenesis. Am J Clin Nutr.
1995;62:1420S-1423S.
34. Alexander GA, Brawley OW. Association of
Helicobacter pylori infection with gastric cancer. Mil Med.
2000;165:21-27.
35. Nelson HD, Humphrey LL, Nygren P, et al.
Postmenopausal hormone replacement therapy: scientific
review. JAMA. 2002;288:872-881.
36. Lynch HT, Lynch JF. Hereditary cancer: family history,
diagnosis, molecular genetics, ecogenetics, and management strategies. Biochimie. 2002;84:3-17.
37. Lee CL, Hsieh KS, Ko YC. Trends in the incidence of
hepatocellular carcinoma in boys and girls in Taiwan after
Epidemiology and Prevention |

13

large-scale hepatitis B vaccination. Cancer Epidemiol

Biomarkers Prev. 2003;12:57-59.
38. Levine DA, Gemignani ML. Prophylactic surgery in
hereditary breast/ovarian cancer syndrome. Oncology.
2003;17:932-941.
39. Clark LC, Dalkin B, Krongrad A, et al. Decreased incidence of prostate cancer with selenium supplementation:
results of a double-blind cancer prevention trial. Br J Urol.
1998;81:730-734.
40. The effect of vitamin E and beta carotene on the incidence of lung cancer and other cancers in male smokers.
The Alpha-Tocopherol, Beta Carotene Cancer Prevention
Study Group. N Engl J Med. 1994;330:1029-1035.
41. Ozols RF, Daly MB, Klein-Szanto A, et al. Specific
keynote: chemoprevention of ovarian cancer: the journey
begins. Gynecol Oncol. 2003;88:S59-S66.
42. Marcus PM. Lung cancer screening: an update. J Clin
Oncol. 2001;19:83S-86S.
43. Marcus PM, Bergstralh EJ, Fagerstrom RM, et al.
Lung cancer mortality in the Mayo Lung Project:
impact of extended follow-up. J Natl Cancer Inst. 2000;
92:1308-1316.
44. Woods WG, Gao RN, Shuster JJ, et al. Screening of
infants and mortality due to neuroblastoma. N Engl J Med.
2002;346:1041-1046.
45. Kramer BS, Brown ML, Prorok PC, et al. Prostate cancer screening: what we know and what we need to know.
Ann Intern Med. 1993;119:914-923.
46. Kramer BS, Brawley OW. Cancer screening. Hematol
Oncol Clin North Am. 2000;14:831-848.
47. Smith RA, Cokkinides V, Eyre HJ. American Cancer
Society guidelines for the early detection of cancer, 2003.
CA Cancer J Clin. 2003;53:27-43.
48. Barry H. Breast self-examination does not reduce
mortality. Am Fam Physician. 2003;67:1784.
49. Thomas DB, Gao DL, Ray RM, et al. Randomized trial
of breast self-examination in Shanghai: final results. J Natl
Cancer Inst. 2002;94:1445-1457.
50. Green BB, Taplin SH. Breast cancer screening controversies. J Am Board Fam Pract. 2003;16:233-241.

14

| Epidemiology and Prevention

51. Fletcher SW, Black W, Harris R, et al. Report of the

International Workshop on Screening for Breast Cancer.
J Natl Cancer Inst. 1993;85:1644-1656.
52. Saslow D, Runowicz CD, Solomon D, et al. American
Cancer Society guideline for the early detection of
cervical neoplasia and cancer. CA Cancer J Clin. 2002;
52:342-362.
53. Mandel JS, Church TR, Bond JH, et al. The effect of
fecal occult-blood screening on the incidence of colorectal
cancer. N Engl J Med. 2000;343:1603-1607.
54. Ault MJ, Mandel SA. Screening for colorectal cancer.
N Engl J Med. 2000;343:1652-1654.
55. Pignone M, Rich M, Teutsch SM, et al. Screening for
colorectal cancer in adults at average risk: a summary of
the evidence for the U.S. Preventive Services Task Force.
Ann Intern Med. 2002;137:132-141.
56. McLeod RS. Screening strategies for colorectal cancer:
a systematic review of the evidence. Can J Gastroenterol.
2001;15:647-660.
57. Manser RL, Irving LB, Byrnes G, et al. Screening for
lung cancer: a systematic review and meta-analysis of
controlled trials. Thorax. 2003;58:784-789.
58. Pastorino U, Bellomi M, Landoni C, et al. Early lungcancer detection with spiral CT and positron emission
tomography in heavy smokers: 2-year results. Lancet.
2003;362:593-597.
59. Zurawski VR, Jr., Broderick SF, Pickens P, et al. Serum
CA 125 levels in a group of nonhospitalized women:
relevance for the early detection of ovarian cancer.
Obstet Gynecol. 1987;69:606-611.
60. NIH consensus conference. Ovarian cancer. Screening,
treatment, and follow-up. NIH Consensus Development
Panel on Ovarian Cancer. JAMA. 1995;273:491-497.
61. Harris R, Lohr KN. Screening for prostate cancer: an
update of the evidence for the U.S. Preventive Services
Task Force. Ann Intern Med. 2002;137:917-929.
62. MacKie RM, Hole D, Hunter JA, et al. Cutaneous
malignant melanoma in Scotland: incidence, survival, and
mortality, 1979-94. The Scottish Melanoma Group. BMJ.
1997;315:1117-1121.

Epidemiology and Prevention of Cancer

Questions
1. Which of the following describes the difference
between efficacy and effectiveness in the interpretation
of clinical trial results?
A. Efficacy considers the outcomes of a treatment under
carefully controlled conditions, whereas effectiveness
considers the outcomes when applying a treatment in the
routine (real-world) setting
B. Efficacy considers short-term and long-term effects,
whereas effectiveness considers long-term effects only
C. Efficacy considers only the effect of survival, whereas
effectiveness considers all outcomes
D. Efficacy is a clinical measure, whereas effectiveness is
statistical measure

2. A 60-year-old woman at high risk of breast cancer

because of an extensive family history and a personal
history of lobular carcinoma in situ asks what her
options are for decreasing her risk of breast cancer risk.
Which of the following measures has been demonstrated in clinical trials to decrease the five-year risk of
breast cancer for this woman?
A.
B.
C.
D.

Tamoxifen therapy
Raloxifene therapy
Annual mammography
Adoption of a low-fat diet

3. Which of the following provides the strongest evidence that a cancer screening test is beneficial?
A. Increased five-year survival of those screened compared
with historical controls
B. Earlier stage at time of diagnosis for those screened
compared with those not screened
C. Large numbers of so-called cured patients who die of
causes other than cancer
D. Decreased cause-specific mortality for those in the
screened arm compared with those in the control arm of a
randomized clinical trial

4. The introduction of a new large-scale screening test has

its greatest immediate effect on which of the following?
A.
B.
C.
D.

Mortality rate related to the disease

Incidence of the disease
Survival rate
Morbidity

5. Which of the following describes the cancer mortality

rate of a country?
A. Percentage of patients with cancer in a country who die
from cancer each year
B. Absolute number of patients with cancer in a country who
die of cancer in a given year
C. Number of people who die of cancer per 100,000 individuals living in the country in a given year
D. Number of people diagnosed with potentially fatal cancers
per 100,000 cases per year

6. Which of the following is true about age adjustment of

incidence and mortality rates?
A. It allows for more accurate comparison of the incidence or
mortality rate for two populations by removing the affect of
different age distributions in those populations
B. It makes it necessary to determine the absolute number of
cancers and deaths in a population
C. It allows for more accurate five-year survival statistics
D. It allows for more accurate assessment of cancer in
younger populations

7. A total of 100 patients are screened for cancer using

a newly developed screening test with the following
results:
8 results are truly positive
1 result is falsely positive
1 result is falsely negative
90 results are truly negative
Which of the following formulas is used to determine the
sensitivity of the test?
Epidemiology and Prevention |

15

A.
B.
C.
D.

90 divided by (1 + 90)
8 divided by (90 + 1)
90 divided by (90 + 8)
8 divided by (8 + 1)

8. Which of the following is necessary for a screening

test to be valuable?
A.
B.
C.
D.

16

Specificity should be high and sensitivity need not be

Sensitivity should be high and specificity need not be
Positive predictive value must be high
Negative predictive value must be high

| Epidemiology and Prevention

9. A postmenopausal woman is considering using estrogen and progesterone and asks you for information.
Which of the following do you tell you about the use of
these agents in postmenopausal women?
A. This use has been shown to lower the risk of cardiovascular
disease
B. This use has been shown to lower the risk of colorectal
cancer
C. This use has been shown to lower the risk of breast cancer
D. This use has been shown to improve cognition

Epidemiology and Prevention of Cancer

Explanatory Answers
1. A
Educational Objective: To understand the concepts of
efficacy and effectiveness as they relate to the interpretation of clinical trials results.
Efficacy is the outcome of a clinical trial. The population
enrolled in the study will differ from the population at
large. Participants in clinical trials generally have less
comorbid disease and, in prevention studies, there is often
a healthy volunteer effect. Volunteers are often more
interested in health than the average member of the population at large. How well the intervention works in the
group as a whole in routine practice is termed effectiveness.
There are a few trials with effectiveness endpoints. Such
studies are generally large and long-term.

as a preventive agent for breast cancer with less risk of

endometrial cancer. Mammography does not affect the
development of breast cancer, and a lower fat intake in
certain populations has been shown to correlate with
lower risk of breast cancer, but in no trial has it been
demonstrated that a change to a low-fat diet reduces risk.

Suggested Reading
Kramer BS, Brawley OW. Cancer screening. Hematol
Oncol Clin North Am. 2000;14:831-848.

Suggested Reading
Fisher B, Costantino JP, Wickerham DL, et al. Tamoxifen
for prevention of breast cancer: report of the National
Surgical Adjuvant Breast and Bowel Project P-1 Study. J
Natl Cancer Inst. 1998;90:1371-1388.
Greenwald P, Clifford CK, Milner JA. Diet and cancer
prevention. Eur J Cancer. 2001;37:948-965.
Thomas DB, Gao DL, Ray RM, et al. Randomized trial of
breast self-examination in Shanghai: final results. J Natl
Cancer Inst. 2002;94:1445-1457.
Wickerham, DL. Tamoxifen versus raloxifene in the prevention of breast cancer. Eur J Cancer. 2002;38:S20-S21.

2. A

3. D

Educational Objective: To understand the known methods

of reducing the risk of breast cancer.

Educational Objective: To recognize why screening can be

controversial and even not recommended, and to appreciate the levels of proof of screening utility.

In randomized, placebo-controlled trials to assess tamoxifen as an adjuvant treatment for patients with early stage
breast cancer, women taking the drug were found to have
one-third fewer new cancers in the contralateral breast
than women taking placebo. The Breast Cancer
Prevention Trial was a randomized, placebo-controlled
study of more than 13,000 women at high risk of breast
cancer. After a median treatment of 69 months, tamoxifen was found to decrease the period risk of breast cancer
by 49%. Tamoxifen was also associated with a reduction
in bone fractures and a small increase in the risk of
endometrial cancer, stroke, pulmonary emboli, and deepvein thrombosis. A trial to compare tamoxifen with
another selective estrogen-receptor modulator, raloxifene, is underway. Raloxifene may have similar efficacy

The major reason to screen for cancer is to decrease mortality. A second reason would be to decrease morbidity
associated with treatment. Assessment of survival outside
of a randomized control trial is subject to lead-time bias.
It is possible to have survival after screening increase
while mortality also increases.
A screening intervention is best assessed in a randomized,
controlled screening trial with cause-specific mortality as
the endpoint. Studies showing a reduction in the incidence of advanced stage disease, improved survival, or a
shift in stage of the disease are weaker and offer potentially misleading evidence of benefit. These latter criteria
by themselves are not sufficient to establish the value of a
screening test.
Epidemiology and Prevention |

17

Because the gold-standard randomized screening trials are

large (often involving thousands of people) and last for
years, less definitive study designs are frequently used to estimate the efficacy and effectiveness of screening practices. In
order of strength of evidence, efficacy can be assessed by
Findings of internally controlled trials using interventionallocation methods other than randomization, such as
allocation determined by birth date or date of clinic visit
Findings of cohort or case-control analytic observational
studies
Results of multiple time-series study with or without the
intervention
Opinions of respected authorities on the basis of clinical
experience, descriptive studies, or consensus reports of
experts. (Even experts can easily be misled by biases.)

Suggested Reading
Harris R, Lohr KN. Screening for prostate cancer: an
update of the evidence for the U.S. Preventive Services
Task Force. Ann Intern Med. 2002;137:917-929.
Kramer BS, Brawley OW. Cancer screening. Hematol
Oncol Clin North Am. 2000;14:831-848.
Kramer BS, Brown ML, Prorok PC, et al. Prostate cancer
screening: what we know and what we need to know. Ann
Intern Med. 1993;119:914-923.
Mandel JS, Church TR, Bond JH, et al. The effect of fecal
occult-blood screening on the incidence of colorectal
cancer. N Engl J Med. 2000;343:1603-1607.
NIH consensus conference. Ovarian cancer. Screening,
treatment, and follow-up. NIH Consensus Development
Panel on Ovarian Cancer. JAMA. 1995;273:491-497.
5. C

Suggested Reading
Kramer BS, Brawley OW. Cancer screening. Hematol
Oncol Clin North Am. 2000;14:831-848.
Kramer BS, Brown ML, Prorok PC, et al. Prostate cancer
screening: what we know and what we need to know. Ann
Intern Med. 1993;119:914-923.
Smith RA, Cokkinides V, Eyre HJ. American Cancer
Society guidelines for the early detection of cancer, 2003.
CA Cancer J Clin. 2003;53:27-43.

4.

Educational Objective: To understand the impact of screening on cancer and population statistics.
Screening, whether beneficial or not, will increase the
number of specific cancers diagnosed in a population. Any
benefit, be it improvement in survival beyond the leadtime bias of screening or a decrease in morbidity or mortality, will only be seen much later. Some of the additional
cancers may be more indolent and less aggressive and
would not have been clinically diagnosed later in life. This
is the concept of length bias and, more specifically, the
more extreme form of length bias known as overdiagnosis.
Length bias occurs when slow-growing, less aggressive
cancers are detected during screening. Cancers diagnosed
as a result of the onset of symptoms between scheduled
screenings are, on average, more aggressive, and treatment outcomes are not as favorable. An extreme form of
length bias is referred to as overdiagnosis, or detection
of pseudodisease. Some undetected slow-growing tumors
fulfill the histologic criteria of cancer but will never be
clinically significant or cause death.
18

| Epidemiology and Prevention

Educational Objective: To understand concepts used in

descriptive epidemiology, such as cancer incidence rate,
mortality rate, survival rates, and age adjustment.
See explanatory answer for question 6.
6. A
Educational Objective: To understand concepts used in
descriptive epidemiology, such as cancer incidence rate,
mortality rate, survival rates, and age adjustment.
Descriptive epidemiology involves incidence and mortality
rates, survival, and number of cases diagnosed in a population. Incidence and mortality rates are commonly expressed
as the number of cases per 100,000 of the population at risk.
These rates are frequently age adjusted, meaning they are
mathematically adjusted to a standard group to remove the
effect of a population aging over time. Cancer is primarily a
disease of older people. As the number of Americans 70
years of age and older has increased dramatically over the
past 30 years, the absolute number and the crude rate per
100,000 of patients with cancer has increased. Age adjusting cancer rates removes the effect of aging on the population. Increases in age-adjusted incidence and mortality
rates are the result of causes other than aging. Comparison
of age-adjusted rates removes the fact that one population
might have a higher rate because it has more older individuals in age groups where cancer is more prevalent.
Suggested Reading
Weir HK, Thun MJ, Hankey BF, et al. Annual report to
the nation on the status of cancer, 1975-2000, featuring
the uses of surveillance data for cancer prevention and
control. J Natl Cancer Inst. 2003;95:1276-1299.

7. D
Educational Objective: To calculate and understand the
concept of sensitivity, specificity, positive predictive
value, and negative predictive value.
When screening for a particular condition, the following
outcomes are possible:
Condition
Condition
Present
Absent
Positive results True positive (A) False-positive (B)
Negative results False negative (C) True negative (D)
Sensitivity is the proportion of people with the condition
who have a positive test result: A divided by (A plus C).
Specificity is the proportion of people without the
condition who have a negative test result: D divided by
(B plus D).
Positive predictive value is the proportion of people with
a positive test result who have the condition: A divided
by (A plus B).
Negative predictive value is the proportion of people with
a negative test result who do not have the condition: D
divided by (C plus D).

valuable, high specificity is needed. Sensitivity need

not be extremely high.
Suggested Reading
Kramer BS, Brawley OW. Cancer screening. Hematol
Oncol Clin North Am. 2000;14:831-848.
9. B
Educational Objective: To recognize recent findings of the
Womens Health Initiative concerning hormone-replacement therapy for postmenopausal women.

Educational Objective: To calculate and understand the

concept of sensitivity, specificity, positive predictive
value, and negative predictive value.

The Womens Health Initiative was a prospective, randomized study involving postmenopausal women who
were randomly assigned to either estrogen plus progestin
or placebo. A lower rate of colorectal cancer was seen in
women taking the combination therapy than in women
taking placebo. This positive effect is, unfortunately, offset by the increased life-threatening cardiovascular and
breast cancer risks associated with treatment with estrogen plus progestin. Among postmenopausal women 65
years or older, treatment with this combination did not
improve cognitive function when compared with placebo.
Although most women taking estrogen plus progestin did
not have clinically relevant adverse effects on cognition
compared with women taking placebo, a small increased
risk of clinically meaningful cognitive decline did occur
in the treated group.

Sensitivity and specificity are relatively independent of

the underlying prevalence (or risk) of the population
screened, but the positive and negative predictive values are highly dependent on the prevalence of the disease being screened. In other words, screening is most
beneficial, efficient, and economical when targeting a
cancer common to the general population or groups
with a high prevalence (or high risk) for the specific
disease being screened. For a screening test to be

Suggested Reading
Nelson HD, Humphrey LL, Nygren P, et al. Postmenopausal hormone replacement therapy: scientific review.
JAMA. 2002;288:872-881.
Rapp, SR, Espeland, MA, Shumaker, SA, et al. Effect of
estrogen plus progestin on global cognitive function in
postmenopausal women: the Womens Health Initiative
Memory Study: a randomized controlled trial. JAMA.
2003;289:2663-2672.

8. A

Epidemiology and Prevention |

19