Chapter 4. Medical Surveillance

• Medical surveillance is undertaken to identify changes in the distributions of diseases in order to prevent or control these conditions within a population.
• A comparison of incidence rates across populations can help to determine characteristics of populations at higher (and lower) risk.
• Surveillance of deaths is convenient because the information is virtually complete, standardized and inexpensive to obtain. Nevertheless, data collected from death certificates may be limited by omitted or inaccurate information.
•  Age adjustment is used to remove the influence of any age differences when comparing the disease frequencies of two populations.
•  Premature death measures the years of potential life lost to a particular disease, and therefore weighs most heavily deaths that occur at young ages.

The goal of this detection, or surveillance, is to identify changes in the distributions of diseases in order to prevent or control these diseases within a population. The term surveillance literally means “to watch over,” and traditionally medical surveillance activities were developed to monitor the spread of infectious diseases through a population. Today, however, surveillance programs have been applied to a wide variety of other conditions, such as congenital malformations, injuries, occupational health problems, and cancer, as well as other behaviors that affect health. Regardless of the type of outcome under consideration, medical surveillance activities involve the following key features:

• 1. Continuous data collection and evaluation
• 2. An identified target population (such as a community, a work force, or a group of patients)
• 3. A standard definition of the outcome of interest
• 4. Emphasis on timeliness of collection and dissemination of information
• 5. Use of data for purposes of investigation or disease control.

The goals of medical surveillance depend on the state of knowledge about the causes of the condition of interest and the extent to which effective preventive measures are known (Table 4–1). Surveillance activities can provide data about the distribution of a disease by person, place, and time. These patterns of occurrence can help shed light on possible causes of the disease. For example, if the time and place of disease occurrence are similar for two or more individuals, a shared source of illness, such as an infectious agent, may be involved. Other demographic information about affected individuals, such as age, race, and gender, typically are collected during surveillance and may provide further insight into the modes of disease acquisition. More detailed information on the personal characteristics of affected individuals can be collected through personal interviews.

In the following sections, various aspects of medical surveillance are described. By relating each of these activities to the diagnosis of lung cancer in the Patient Profile, the relationships among different types of surveillance are demonstrated.

In the United States, the incidence of cancer is monitored by the National Cancer Institute through population-based registries that collectively comprise the Surveillance, Epidemiology, and End Results (SEER) program. Population based means that the target group is the general population, usually identified as residents of a particular geographic area. The term is used to contrast with other types of registries, such as those based on patients who are treated at specific hospitals or those who belong to a particular health insurance plan.

The most recently reported SEER data on the incidence of cancer are derived from 12 areas, including five entire states (Utah, Iowa, Connecticut, New Mexico, and Hawaii), six metropolitan regions (Atlanta, Detroit, Los Angeles, San Jose/Monterey, Seattle, and San Francisco), and the native population of Alaska (Figure 4–1). Although about 14% of the population of the United States resides within these 12 areas combined, this is clearly not a random sample of the nation. The areas were selected largely on the basis of the ability to maintain ongoing population-based cancer reporting systems and epidemiologic interest in the population subgroups that reside there. Collectively these registries provide reasonably representative samples of different regions of the country, rural and urban populations, and most major racial and ethnic groups.

The SEER registries use a variety of methods to locate new diagnoses of cancer. The majority of diagnoses are identified from hospital admissions through the review of pathology reports and lists of discharge diagnoses. Additional identification sources include pathology laboratories outside of hospitals, office records of physicians, outpatient treatment facilities, and death certificates. The size of the population at risk of cancer is derived for each geographic area by extrapolation from census estimates.

The 2000 annual age-adjusted incidence rates for the five most common types of cancer among men and women of all races in the United States are shown in Figure 4–2. Note that the data on breast cancer are confined to women and the data on prostate cancer are confined to men. By convention, the incidence rates for cancer are expressed per 100,000 person-years. The incidence rate for lung cancer indicates that 62 individuals within a representative sample of 100,000 persons in the United States are expected to develop lung cancer in one year.

The incidence of lung cancer is 60% higher among males (80 per 100,000 person-years) than females (50 per 100,000 person-years) in the United States. Moreover, as illustrated in Figure 4–3, the incidence of lung cancer is not constant across age groups. This disease is extremely rare in persons under 40 years of age. After age 40, the incidence of lung cancer rises sharply, reaching a peak among persons in their 70s.

The striking relationship between age and incidence of lung cancer (and most other types of cancer) creates a potential complication when comparing the incidence rates of population groups with different age distributions. In other words, simply because of their age, a younger group of individuals will tend to have fewer occurrences of lung cancer than an older group of people. Failure to account for this age discrepancy would result in a distorted comparison of incidence rates for lung cancer between the two groups. To allow a comparison of incidence rates that is not influenced by age differences in the underlying populations, an age-adjustment procedure must be performed. The usual approach is referred to as direct age adjustment (or direct age standardization), in which a single standard age structure is applied to the age-specific incidence rates for the groups being compared, resulting in summary rates for the groups that are not distorted by differences in age. For direct age adjustment of cancer incidence rates in the United States, the standard age distribution currently in use by the SEER program is that of the entire population of the country in 2000.

As shown later in this chapter, blacks in the United States have an age distribution that is younger than the corresponding age distribution for whites. Because lung cancer tends to occur primarily in older adults, a comparison of overall lung cancer incidence rates for blacks and whites would be misleading, unless an adjustment is made for the underlying difference in age distribution between the races. Summary lung cancer incidence rates for whites and blacks can be compared fairly by determining the lung cancer incidence rate that would have occurred in each racial group if it had the same age distribution as the 2000 U.S. population. The choice of the standard age distribution is arbitrary—any distribution can be used as long as it is applied equally to the groups being compared. We will focus on the topic of age adjustment in greater depth later in this chapter; a discussion of the calculation of adjusted rates can also be found in Dawson and Trapp (2004, Basic and Clinical Biostatistics).

The age-adjusted incidence rates for leading forms of cancer in the United States are shown by race in Table 4–2. From these data, it can be seen that if the effect of age is held constant, blacks tend to have higher rates of occurrence of lung, colon, and prostate cancers than do whites. In contrast, breast and bladder cancers tend to occur with greater incidence among whites than blacks.

By dividing the incidence rate among blacks by the incidence rate among whites, a summary measure of disparity in rates of occurrence is obtained. An index of the racial disparity in cancer incidence is the ratio of black-to-white incidence rates, or rate ratio (RR). If blacks and whites have the same rate of disease occurrence, the rate ratio would have a value of unity (RR = 1). When blacks have an elevated incidence rate compared with whites, the black-to-white rate ratio is greater than one (RR > 1). In contrast, when blacks have a lower incidence rate than whites, the rate ratio is less than one (RR < 1). The further the rate ratio is from unity, the greater the disparity in incidence rate between the races.

The black-to-white rate ratio of 1.25 for lung cancer indicates that the incidence of this cancer in black persons is about 25% greater than it is in white persons (Figure 4–4). In contrast, the rate ratio of 0.57 for bladder cancer indicates that the incidence rate of this cancer in blacks is more than 40% lower than it is in whites. These patterns suggest that the factors that influence the development of lung cancer and the factors that influence the development of bladder cancer are distributed differently between the races. These predisposing conditions, termed risk factors, could include genetic susceptibility to the cancers in question as well as exposure to environmental agents. The most common epidemiologic approaches to evaluating risk factors are discussed in Chapters 8 (Cohort Studies) and 9 (Case-Control Studies). Approaches to the study of genetic susceptibility are presented in Chapter 11: Epidemiologic Studies of Genetics.

Another index used to measure the population distribution of a disease is the mortality rate, which characterizes the rapidity with which deaths from the disease occur over time. The mortality rate is determined by the combined forces of the rate of new diagnoses (incidence rate) and the likelihood of death following diagnosis (case fatality). For diseases such as lung cancer, which have a high case fatality (ie, a low rate of cure or recovery), mortality rates give a reasonable approximation of incidence rates. As shown in Figure 4–5, the overall age-adjusted mortality rate for lung cancer is about 90% as large as the corresponding incidence rate. For a cancer with a more favorable prognosis, such as thyroid cancer, there will be a greater disparity between mortality and incidence rates (Figure 4–6). The age-adjusted mortality rate for thyroid cancer is only about 7% as large as the corresponding incidence rate, because most persons who develop this disease do not die from it.

• 1. Widely collected, virtually complete data: registration of deaths is compulsory in most industrialized countries, and few deaths are not reported
• 2. Standardized nomenclature: the International Classification of Diseases is used to promote uniformity in reporting of causes of death
• 3. Modest cost: recording of deaths is relatively inexpensive.

Statistics on mortality thus serve as a convenient tool for epidemiologic surveillance, particularly when incidence data are not available. For example, as already noted, the SEER program covers only about 14% of the population of the United States, but death registration is compulsory throughout the nation. Accordingly, statistics on mortality can provide a more complete picture of the geographic distribution of cancer than can be determined from incidence data alone. A map of age-adjusted mortality rates for lung cancer for males and females of all races in the United States illustrates this point (Figure 4–7).

The process of collecting information on deaths in the United States begins with completion of a death certificate. Background demographic data (eg, age, birthdate, birthplace, race, sex, marital status, place of residence, occupation, and education) usually are recorded by the funeral director. A physician is required to certify the conditions responsible, in whole or in part, for the patient’s death. For registration purposes, a distinction is made among

• 1. Immediate cause of death
• 2. Conditions that led to the immediate cause of death
• 3.Underlying cause of death.

Only the underlying cause of death is tabulated in official statistics. The underlying cause of death is defined as (1) the disease or injury that initiated the train of morbid events leading directly to death, or (2) the circumstances of the accident or violence that resulted in the fatal injury. For each of the causes of death listed on the certificate, the physician records the length of time between onset and death. Other required information includes whether an autopsy was performed, the place of death, and the manner of death (ie, natural, unintentional injury, homicide, suicide, or unknown).

Overall, only about 10% of death certificates indicate that an autopsy was performed on the deceased individual (the decedent). For people who die from cancer, stroke, chronic obstructive pulmonary disease, or diabetes mellitus, an autopsy is performed for fewer than 1 person in 20. The causes of death with the highest percentages of subsequent autopsies are homicides (97%), suicides (56%), and unintentional injuries (49%). Medical examiners and coroners are responsible for investigating sudden or unanticipated deaths—such as those resulting from homicide, suicide, or unintentional injuries—as well as deaths arising unexpectedly from natural causes. The investigation includes collecting information on the circumstances surrounding the death, and where appropriate, measuring the decedent’s alcohol and drug levels. The information reported by medical examiners and coroners usually is not available at the time of initial certification of death, and therefore is added later as an amendment.

Death certificates are filed with a local registrar, who checks for completeness of the information provided and then forwards the certificates to the state vital records office. The aggregated certificates are coded, numbered, and stored at the state level, and certificates for nonresidents are forwarded to their state of residence. The composite death information from each state is transferred to the National Center for Health Statistics for compilation into a national data base. The process of collecting information on deaths in the United States is summarized in Figure 4–8.

As noted, there are a number of advantages to the use of death registrations for purposes of medical surveillance. At the same time, it must be recognized that there are limitations to the information obtained in this manner:

• 1. Most physicians receive little formal instruction about how to fill out death certificates. As a result, the medical information reported is often incomplete or inaccurate.
• 2. In many situations, the certifying physician may have little particular knowledge of the decedent, leading to further errors in reporting or omissions of important conditions.
• 3. The use of only one underlying cause of death in official tabulations can lead to a very incomplete picture of the contributions of various conditions to mortality. For example, in instances in which hypertension is reported on a death certificate, it is listed as the underlying cause of death only 10% of the time.
• 4. Concern for the confidentiality of the decedent may cause some physicians to omit sensitive diagnoses. There is evidence, for instance, that acquired immune deficiency syndrome (AIDS) has been underreported as an underlying cause of death.
• 5. When standard classifications of disease are periodically updated, the rules about assigning underlying causes of death can change, resulting in abrupt, artificial changes in the mortality trends for certain diseases.
• 6. With the many steps involved in collecting, editing, coding, and processing of death certificates, summary mortality data for a given year are not reported, even in preliminary form, until about 2 years later.

For any particular application, the appropriateness of using death registrations for purposes of medical surveillance is determined by balancing the advantages of universal coverage, convenience, and accessibility against the limitations of inaccurate and incomplete information, as well as delays in availability. Circumstances in which this type of information is likely to prove most useful include:

• 1. Monitoring historical trends in the burden of a disease and forecasting future expectations
• 2. Identifying population subgroups with disproportionate burdens of disease
• 3. Generating hypotheses about risk factors for a disease
• 4. Prioritizing the allocation of health care resources
• 5. Monitoring progress toward meeting health status objectives for a population.

Mortality rates for all causes of death in the United States are shown by age and race in Figure 4–9. For both whites and blacks, mortality rates are high during the first year of life, are low during childhood, adolescence, and young adulthood, and then rise rapidly with increasing age. At every age depicted, however, the death rates for blacks exceed those of whites.

Therefore it might be surprising that the crude death rate (total deaths per total person-years) for blacks (774 deaths/100,000 person-years) is lower than the corresponding rate for whites (895 deaths/100,000 person-years). As indicated in Figure 4–10, the black-to-white ratio of crude mortality rates is 0.86, suggesting that blacks have a lower risk of death than do whites. This apparent paradox is explained by differences in the underlying age distributions of blacks and whites. On average, black persons in the United States tend to be younger than white persons (Figure 4–11). For example, only 7.8% of blacks in the United States are 65 years or older, compared with 13.4% of whites. Thus a smaller proportion of the black population experiences the high mortality associated with advanced age. To obtain an undistorted summary comparison of mortality for blacks and whites, the age differential between the races must be eliminated.

• 1. Select a standard age structure. By convention, the standard distribution used for age adjustment of mortality rates in the United States is the age distribution of the total population of the country in 2000.
• 2. Multiply the age-specific mortality rates for each group being compared by the corresponding age-specific numbers of persons in the standard population. The result is the expected number of deaths for that age category.
• 3. Sum the expected numbers of deaths across all age categories to yield a total number of expected deaths for each group being compared.
• 4. Divide the total number of expected deaths in each group by the total size of the standard population to yield the summary age-adjusted mortality rate.

When this direct age-adjustment procedure is performed on the age-specific death rates for blacks and whites in the United States for 2001, the summary mortality rates shown in Figure 4–12 are obtained. Note that the age-adjusted rates are higher than the crude rates for blacks (Figure 4–10), since the standard population has an older age distribution than does the black population. For whites, the reverse phenomenon is observed; the age-adjusted rate is lower than the crude rate because the white population has an older distribution of ages than the standard population.

The numerical values of the age-adjusted rates are not particularly meaningful by themselves, because the values will vary according to the standard age distribution used. The utility of the age-adjusted rates is that they allow comparisons across groups, such as the black-to-white rate ratio, by removing any age disparity between the groups being compared. It can be seen from Figure 4–12 that the age-adjusted mortality rate for blacks is almost 30% greater than it is for whites (rate ratio = 1.32). Thus the age-adjusted mortality rate ratio provides a summary measure that is consistent with the increase in black mortality shown in Figure 4–9. Through the use of an adjustment technique, the distorting effect of differing underlying age distributions has been removed from the contrast of summary mortality rates.

In addition to variation by age and race, mortality in the United Sates also varies by other characteristics. Age-adjusted mortality rates for whites and blacks for calendar years 1980 through 2001 are shown in Figure 4–13. Within each race, the age-adjusted mortality rates tended to decline, and the percentage decline was similar for whites (17%) and blacks (16%) over this time period. The fall in mortality rates was fairly continuous and progressive among whites. For blacks, however, there was a substantial decline between 1980 and 1982, with relatively level rates between 1982 and 1995, followed by a progressive decline thereafter.

As shown in Figure 4–14, black males have the highest age-adjusted mortality rate, followed in succession by white males and black females, with the lowest death rates observed among white females. Within both racial groups, males have higher mortality rates than females. For both genders, blacks have higher mortality rates than whites.

Table 4–3 lists the annual age-adjusted mortality rates for all persons in the United States according to the 10 leading causes of death. Four of the five most common causes of death in this country result from long-term, chronic processes: heart diseases, malignantneoplasms (cancer), cerebrovascular disease (stroke), and chronic respiratory diseases.

Age-adjusted mortality rates for the individual causes of death vary by race and sex. Blacks have higher death rates than whites for 7 of the 10 leading causes of death (Figure 4–15). The relative increase in mortality among blacks is greatest for nephritis/nephrosis (black-to-white RR = 2.4), followed by septicemia (RR = 2.3), diabetes mellitus (RR = 2.1), stroke (RR = 1.4), heart disease and cancer (RR = 1.3), and influenza and pneumonia (RR = 1.1). Unintentional injuries occur with similar age-adjusted death rates among blacks and whites (RR = 1.0). Only chronic respiratory diseases and Alzheimer’s disease (RR = 0.7) are responsible for lower age-adjusted rates of death among blacks than whites.

As depicted in Figure 4–16, males have higher age-adjusted mortality rates for 8 of the 10 leading causes of death in the United States. The relative increase in mortality among males is greatest for unintentional injuries (male-to-female RR = 2.2); followed by homicide, diseases of the heart, malignantneoplasms, and nephritis/nephrosis (RR = 1.5); chronic respiratory diseases, influenza, and pneumonia (RR = 1.4); and diabetes mellitus and septicemia (RR = 1.2). No gender difference in age-adjusted death rates was observed for stroke (RR = 1.0). Only for Alzheimer’s disease did males have a lower age-adjusted mortality rate than females (RR = 0.8).

Trends over time in the overall age-adjusted mortality rates for the 10 leading causes of death between 1999 and 2001 are displayed schematically in Figure 4–17. Over this relatively brief period, declines in death rates occurred for diseases of the heart (–7%); stroke, influenza, and pneumonia (–6%); chronic respiratory diseases (–4%); and malignantneoplasms (–2%). Concurrent percentage increases in age-adjusted; mortality rates occurred for Alzheimer’s disease (+16%); nephritis/nephrosis (+8%); and unintentional injuries, diabetes mellitus, and septicemia (+1%).

Returning to lung cancer, the focus of the Patient Profile, the age-adjusted mortality rate from this disease is the highest for any form of cancer in the United States (Figure 4–18). Lung cancer accounts for more than one fourth of all deaths from malignantneoplasms in the United States. Between 1975 and 1990 the age-adjusted mortality rate from lung cancer increased steadily in the United States (Figure 4–19). Between 1990 and 1993, lung cancer mortality remained level, with a decline thereafter. Even with the recent declines in mortality, over the period 1975 through 1995 the rate of deaths from this disease increased by almost 40%. Although males account for almost 60% of all deaths from lung cancer, the percentage increase in age-adjusted mortality between 1973 and 1995 was much greater for females (129%) than for males (10%). Among males, the age-adjusted death rates for lung cancer rose from 1975 to 1984, reached a plateau from 1984 to 1991, and declined after 1991. For females, the age-adjusted death rates climbed persistently from 1975 through 1995 and were relatively level thereafter.

As shown in Figure 4–20, the age-adjusted mortality from lung cancer varies considerably by race and sex in the United States. Among both whites and blacks, there is a considerably greater mortality rate from this disease among males. Black males have a more than 30% higher death rate from lung cancer than white males. In contrast, black females have about a 5% lower age-adjusted mortality rate from lung cancer than do white females.

Mortality rate is a convenient and easily understood measure of the burden of a disease on a population. However, by weighing the impact of all deaths equally, mortality rate does not convey the extent to which persons are dying prematurely. In this context, premature death means a death that occurs earlier than would have been expected in the absence of the disease. Clearly, causes of death that tend to occur among young people result in a much greater loss of life expectancy than do causes of death that tend to occur among the elderly. For example, in 2001 the life expectancy at birth was 77 years, compared with only 12 years for persons aged 75 years or older. It may seem paradoxical that persons born in 2001 were expected to live only to age 77, whereas persons born in 1926 were expected to live to age 89 (77 + 12 years). Remember, however, that the older persons had already survived to age 75 years by 2001, and thus had escaped all of the risks of death between birth and 75 years that the newborns still faced.

One approach to estimating the impact of a cause of death on premature loss of life is the person-years of life lost, which is calculated in the following manner:

• 1. Identify the age at death for each decedent
• 2. Given the decedent’s age, estimate the years of life expectancy that were lost because of the death
• 3. Sum the years of life expectancy lost over all decedents.

Selected leading contributors to person-years of life lost in the United States are summarized in Figure 4–21. In 2000, there were 2.4 million deaths, which resulted in a total of about 35 million person-years of life lost. Heart disease, which tends to occur among the elderly, accounted for 30% of all deaths, but only 22% of the person-years of life lost. In contrast, injuries, which tend to occur among the young, accounted for only 4% of all deaths, but almost 9% of person-years of life lost.

The person-years of life lost from selected leading causes of death from cancer are shown in Figure 4–22. The leading contributor is lung cancer, which accounts for almost one third of all of the person-years of life lost to cancer. Breast cancer and large bowel malignancies each account for more than 10% of the person-years of life lost to cancer.

Another measure of premature loss of life is the years of potential life lost before age 75 years (YPLL < 75), which is calculated in the following manner:

• 1. The difference between age 75 years and the age of death is calculated for each decedent.
• 2. These differences are summed over all decedents.
• 3. This summation is divided by the number of persons under age 75 years in the source population of persons.
• 4. The resulting estimate is multiplied by 1000 or 100,000, depending on the units desired.

The overall YPLL < 75 in the United States during 2001 was 7564 per 100,000 person-years. The YPLL < 75 for selected underlying causes of death are shown in Figure 4–23. The leading contributor to YPLL < 75 was cancer, which accounted for 22% of the total and a similar percentage of all deaths (23%). The proportionate contribution of unintentional injuries was much greater to YPLL < 75 (14%) than to deaths (4%). Similarly, suicide and homicide also accounted for larger proportionate shares of total YPLL < 75 (5% and 4%, respectively) than of all deaths (about 1% each). The larger contributions of unintentional injuries, suicide, and homicide to YPLL < 75 resulted from the comparatively young ages at which these fatalities tend to occur.

In the preceding section, attention was focused on surveillance of medical events, such as new diagnoses or deaths from specific diseases. Surveillance techniques also can be used to characterize patterns of risk factor distribution by person, place, and time. The Behavioral Risk Factor Surveillance System (BRFSS), which is supported by the Centers for Disease Control and Prevention, collects this type of information. The BRFSS was initiated in 1984 to provide statewide data on life-style characteristics that could affect health status. Data are collected by the health departments of most states using a standard protocol. Adult respondents are sampled randomly and are briefly interviewed over the telephone.

The most important risk factor for the development of lung cancer is cigarette smoking. Data collected by the BRFSS can be used to describe smoking patterns within the general population. The overall prevalence of cigarette smoking in the United States in 2002, as determined by the BRFSS, was 23%. As shown in Figure 4–24, the estimated prevalence of cigarette use varied by state of residence, with the highest reported level in Kentucky (33%). Utah had the lowest reported level of any state (13%). The low frequency of cigarette smoking in Utah is attributable to the high proportion of residents who are practicing Mormons, and whose religious beliefs include abstinence from tobacco and alcohol. Given the marked differential in cigarette smoking in Utah and Kentucky, it is not surprising that between 1996 and 2000, the age-adjusted lung cancer mortality rate in Utah (26 cases per 100,000 person-years) was only about one third as high as that in Kentucky (79 cases per 100,000 person-years).

Another system used to collect information on risk factors and health status in the United States is the National Health Interview Survey (NHIS). The NHIS is a continuous nationwide household survey that uses data collected through personal interviews. In 2001, data were collected on 100,000 persons. The overall prevalence of current cigarette smoking among adults (≥ 25 years of age) in the United States during that year as determined by the NHIS was 22%.

The NHIS prevalence estimates for cigarette smoking are shown by race and sex in Figure 4–25. These data indicate that cigarette use is most common among black males, followed by white males, black females, and finally white females. In general, this pattern is consistent with the race and sex distribution of lung cancer mortality rates depicted in Figure 4–20. Although the prevalence of cigarette smoking has been declining overall, smoking remains the single most preventable cause of death in the United States. Progress toward meeting the national objective of reducing the prevalence of this behavior can be monitored through risk factor surveillance systems such as the BRFSS and the NHIS.

In this chapter, lung cancer was used as a focus for consideration of the role of surveillance in epidemiology. Surveillance was defined as the detection of the occurrence of health-related events or exposures in a target population. Successful surveillance activities require continuity over time, standardized methodology, and timeliness of data collection and dissemination.

Surveillance data can be used in different ways, depending on the type of information collected. Persons newly diagnosed with a disease can yield information on incidence rates, deaths from a disease can be used to describe mortality rates, indices of premature death can be used to assess the impact of a disease on longevity, and prevalence of risk factors can be used to predict future disease occurrence or to assess the status of prevention initiatives.

Surveillance activities in the United States related to lung cancer include collection of data on newly diagnosed persons (through population-based cancer registries), deaths (through death certificates), and prevalence of cigarette smoking (through population-based personal interview surveys).

The utility of age adjustment was demonstrated for the control of differences in underlying age structure, when comparing summary rates of lung cancer incidence or mortality across population subgroups.

Key findings from surveillance related to lung cancer in the United States are as follows:

• 1. Lung cancer has the third highest incidence rate among all forms of cancer, trailing only breast cancer among women and prostate cancer among men. Malignancies of the lung are the leading cause of death from cancer.
• 2. The incidence of lung cancer increases sharply between 40 and 70 years of age.
• 3. The overall age-adjusted incidence rate of lung cancer is about 25% higher in blacks than in whites. Contrast of age-adjusted mortality rates reveals a 32% higher rate among blacks.
• 4. The age-adjusted mortality rate for lung cancer is about 90% of the corresponding incidence rate, indicating a high case fatality. The lung cancer age-adjusted mortality increased by over 40% between 1973 and 1995.
• 5. Age-adjusted mortality rates for lung cancer vary widely across states, as do the corresponding prevalences of cigarette smoking.
• 6. Lung cancer accounts for about one third of the premature mortality from cancer and about 7% of total premature loss of life.
• 7. The comparatively high age-adjusted mortality rate for lung cancer among black males is paralleled by a comparatively high prevalence of cigarette smoking.

Several features of the Patient Profile illustrate these descriptive patterns of lung cancer distribution: (1) the patient was 68 years of age at presentation, a comparatively high-risk age for lung cancer occurrence, (2) she was a long-term cigarette smoker, thereby substantially increasing her risk for developing this disease, and (3) the rapidity of her death following diagnosis is consistent with the high case fatality of this disease. With the generally poor treatment results for lung cancer, the greatest promise for controlling the impact of this disease is by preventing people from starting to smoke cigarettes and encouraging current smokers to quit.