2
Potential to Reduce the Cancer Burden Through Cancer Prevention and Early Detection1

Several attempts have been made to estimate the potential reduction in cancer incidence and mortality rates that could be achieved with major favorable shifts in the distribution of modifiable risk factors in the U.S. population. The four analyses described in this chapter apply different methods and underlying assumptions in making their projections but arrive at a similar conclusion: that major reductions in the cancer burden are achievable by sharply reducing rates of tobacco use, increasing levels of physical activity, decreasing the prevalence of obesity, improving dietary practices, keeping alcohol consumption at low to moderate levels, and getting screened for cancer at recommended intervals. How rapidly such changes can occur will largely depend upon social investments and political will.

The methods, key assumptions, and strengths and limitations of each of the selected analyses are described in the following sections. Many of the assumptions used to develop the predictive models vary and are uncertain (Rockhill et al., 1998). There is uncertainty, for example, regarding biological latency, the strength of causal associations, and the prevalence of risk factors. Nevertheless, such models provide a useful mechanism that can be used to gauge what is achievable under various social conditions.

1  

This chapter is based on the background paper prepared by Graham A. Colditz, Catherine Tomeo Ryan, Charles H. Dart III, Geetanjali Datta, Laurie Fisher, and Beverly Rockhill (www.iom.edu/ncpb).



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 30
Fulfilling the Potential of Cancer Prevention and Early Detection 2 Potential to Reduce the Cancer Burden Through Cancer Prevention and Early Detection1 Several attempts have been made to estimate the potential reduction in cancer incidence and mortality rates that could be achieved with major favorable shifts in the distribution of modifiable risk factors in the U.S. population. The four analyses described in this chapter apply different methods and underlying assumptions in making their projections but arrive at a similar conclusion: that major reductions in the cancer burden are achievable by sharply reducing rates of tobacco use, increasing levels of physical activity, decreasing the prevalence of obesity, improving dietary practices, keeping alcohol consumption at low to moderate levels, and getting screened for cancer at recommended intervals. How rapidly such changes can occur will largely depend upon social investments and political will. The methods, key assumptions, and strengths and limitations of each of the selected analyses are described in the following sections. Many of the assumptions used to develop the predictive models vary and are uncertain (Rockhill et al., 1998). There is uncertainty, for example, regarding biological latency, the strength of causal associations, and the prevalence of risk factors. Nevertheless, such models provide a useful mechanism that can be used to gauge what is achievable under various social conditions. 1   This chapter is based on the background paper prepared by Graham A. Colditz, Catherine Tomeo Ryan, Charles H. Dart III, Geetanjali Datta, Laurie Fisher, and Beverly Rockhill (www.iom.edu/ncpb).

OCR for page 30
Fulfilling the Potential of Cancer Prevention and Early Detection DESCRIPTIONS OF THE FOUR ANALYSES Doll and Peto, 1981 In 1981 Doll and Peto examined the degree to which cancer incidence and mortality rates could be reduced in the United States. Their approach did not attempt to project the level of reduction in the rate of mortality from cancer within a defined time frame. Rather, Doll and Peto estimated the reduction theoretically possible by comparing the rates in the United States with those in other countries. Their choice of method stemmed from the observation that the cancer incidence rate among migrants tends to be that found in the country to which they migrated, indicating that differential cancer incidence rates are due in part to environmental factors such as diet, exercise, occupational exposures, and smoking habits, and that cancer does not arise exclusively because of genetic factors. Doll and Peto included in their estimates only information for people who were younger than age 65 because data on the incidence of cancer among older individuals were considered unreliable. Furthermore, they omitted cases of nonmelanoma skin cancers from their analysis because data on the incidence of these cancers were unreliable and also because these cancers are easily treated and are rarely fatal. Doll and Peto’s analysis was conducted by comparing the site-specific cancer incidence rates among male and female residents in Connecticut (the most complete U.S. cancer registry at that time) with the lowest reliable international site-specific cancer incidence rates available. Age-adjusted rates were collected from 1968 to 1972 from registries selected by the International Agency for Research on Cancer (IARC) (International Agency for Research on Cancer, 1976). The analysis used data from the United Kingdom, New Mexico, Japan, East Germany, Norway, Israel, Nigeria, Iowa, Puerto Rico, Finland, and New Zealand. The results of the analysis suggested that in 1970, 75 to 80 percent of all cancers in the United States could have theoretically been avoided if the population of the United States could be like that of the countries with the lowest incidences. This figure of 75 to 80 percent can thus be thought of as a crude estimate of the proportion of cancer in the United States that was (in 1970) due to “environmental” factors that made the U.S. population different from low-risk populations. The “environmental” factors that differ between the United States and low-risk populations are many and diverse and include birth weight; age at puberty; and lifelong patterns of tobacco use, diet, physical activity, alcohol consumption, use of pharmacological agents, and reproduction. Of course, the more traditional and limited definition of “environmental” exposure fits in here, as the U.S. population and the populations of other countries also experience different levels of exposure to contaminants in air, water, and food. Occupational and environmental exposures are

OCR for page 30
Fulfilling the Potential of Cancer Prevention and Early Detection estimated to be less important than lifestyle factors as contributors to cancer in the United States today (Monson and Christiani, 1997). Given the major shift toward the regulation of occupational exposures to known carcinogens and the active surveillance of workers previously exposed to many of these carcinogens, the most modifiable component of cancer risk remains lifestyle factors. Doll and Peto acknowledged that although their estimates were theoretical maximums, it would take a great amount of time to effect social change and that the large amount of change necessary to decrease the cancer incidence rate 75 to 80 percent was unlikely. Although Doll and Peto’s methods might seem rudimentary by today’s methodological standards, they provided a foundation for later, similar work. Willett et al., 1996 In 1996 Willett and colleagues used an international comparison approach at the ecological level similar to that used by Doll and Peto (1981) to assess the degree to which cancer mortality could be reduced in the United States. They limited their data on rates of mortality from cancer to those for the United States, Japan, and China because they were considered to be the most reliable data available. They used age-adjusted data from the Surveillance, Epidemiology, and End Results (SEER) Program and from Cancer Incidence in Five Continents, an IARC publication (Parkin et al., 1992). The country with the lowest site-specific cancer incidence rate was considered to have the “baseline” rate. The difference between the highest and the baseline incidence rates was calculated to indicate the maximum degree of cancer reduction possible. To provide information on behavior-specific reductions in the rate of mortality from cancer, the investigators considered data for unique populations in the United States such as those from the Adventist Health Study, which indicated the magnitude of change possible within the context of the current U.S. culture. Neither Willett and colleagues nor Doll and Peto provided a time frame for the reduction of the cancer mortality rate estimated in their analyses; that is, they did not specify the latency of effects. Willett and colleagues reported the degree of reduction in the incidence of cancer possible by making specific behavioral changes, considering not only the theoretically greatest possible reduction but also more realistic reductions based on the amount of behavioral change that would likely occur in the U.S. population. Considering the downward trend in cigarette smoking, they suggested that a two-thirds reduction in the number of individuals who smoked and, thus, a similar eventual reduction in tobacco-related cancer mortality rates would be possible. If a percentage of those who consumed more than two drinks per day reduced their intakes, they estimated that alcohol-related cancer mortality rates could be reduced by

OCR for page 30
Fulfilling the Potential of Cancer Prevention and Early Detection a third. Another 10 percent decrease in cancer mortality rates could be achieved if a majority of the U.S. population made important diet- and exercise-related changes, such as exercising vigorously for 20 minutes a day, eating one additional serving of leafy vegetables each day, or consuming no more than one serving of red meat a week. The investigators acknowledged that not everyone could or would make behavioral changes and thus stated that although the comparisons of international rates of cancer incidence would suggest that cancer risk could theoretically be reduced by 60 percent, a more realistic estimate of future reduction would be closer to 33 percent. National Cancer Institute, 1986 Using another approach, the National Cancer Institute (NCI) estimated in 1986 the likely consequences of reductions in specific cancer-related risk factors from 1980 to 2000 in an effort to set national cancer control objectives. NCI considered prevention, screening, and treatment in modeling future cancer mortality rates and in setting objectives for 2000. NCI set a goal of a 50 percent reduction in the total cancer mortality rate between 1980 and 2000. The corresponding risk factor reduction goals determined by NCI to allow achievement of this goal are listed in Table 2.1. The NCI goal of a 50 percent reduction in the rate of mortality was, in retrospect, overly optimistic. Cancer mortality rates in the United States (for both sexes combined) actually rose through the 1980s and reached their highest levels ever from 1990 to 1992. Since the early 1990s, however, mortality rates have declined somewhat, and mortality rates in 1997 were approximately 7 percent lower than they were in 1980 (Ries et al., 2000b). Although NCI modeled the effects of reductions in various risk factors with a sophisticated computer program (called Can*trol), to arrive at their goal of a 50 percent reduction in total cancer mortality rates, they underestimated the latencies of the social and political changes that would be needed to bring about large changes in the behavior of the population. Byers et al., 1999 In 1996, the American Cancer Society (ACS) set an ambitious “challenge goal” for a 25 percent reduction in cancer incidence and a 50 percent reduction in the rate of mortality from cancer by 2015. Thereafter, Byers and colleagues (1999) examined the feasibility of reaching the ACS challenge goals by 2015 on the basis of possible reductions in selected major risk factors. For that analysis, the researchers used the 1990 cancer incidence and mortality rates as the baseline rates. Byers and colleagues used data on cancer incidence and survival rates from the SEER Program, cancer mortality statistics from the National Cen-

OCR for page 30
Fulfilling the Potential of Cancer Prevention and Early Detection TABLE 2.1 Risk Factors, Goals, and Assumptions Used by NCI Working Group and ACS in Predictions for the United States Risk Factor NCI Goals for Reduction Tobacco By 1990, rate of decline of smoking should be 4% or more per year; by 2000, proportion of all persons 21 years and older who smoke should be =15%; by 2000, proportion of youth who begin to smoke should be =15% (proportion of youth ages 12–18 years who begin to smoke should be <3%). Diet By 1990, per capita consumption of fiber from grains, fruits, and vegetables should be increased to =15 g/day; per capita consumption of fat should be below 35% of total daily calories. By 2000, per capita consumption of fiber should increase to 20–30 g/day; per capita consumption of fat should be 30% of total daily calories. Alcohol No goals Antiestrogen use (e.g., Tamoxifen) No goals Screening By 1990, 70% of women aged 50–70 years should have clinical breast exam each year; 45% of these women should have a mammogram every year; 85% of women aged 20–39 years should have a Papanicolaou (Pap) smear every 3 years; 70% of women aged 40–70 years should have a Pap smear every 3 years. By 2000, 80% of women aged 50–70 years should have clinical breast exam each year; 80% of these women should have a mammogram every year; 90% of women aged 20–39 years should have a Pap smear every 3 years; 80% of women aged 40–70 years should have a Pap smear every 3 years. Treatment By 2000, increase the rate of adoption of state-of-the-art treatment; continue to advance treatment, as reflected in increasing cancer survival rates; cancer-specific objectives defined in terms of desired increase in 5-year survival rates. ter for Health Statistics, the prevalence of cancer risk factors from the Behavioral Risk Factor Surveillance System of the Centers for Disease Control and Prevention, and information on cancer treatment from the American College of Surgeons’ National Cancer Database. Byers and colleagues made projections for cancer incidence and mortality rates for each cancer site separately and for all cancers combined, con-

OCR for page 30
Fulfilling the Potential of Cancer Prevention and Early Detection ACS Risk Factor Reduction Assumptions Biological Latency Assumptions NCI ACS Smoking prevalence reduced to: 16–20% by 2005 15 years 10 years (18–22% among men, 14–18% among women)     Prevalence of low levels of fruit and vegetable consumption reduced to 47–58% by 2010; Prevalence of high fat intake reduced to 9–15% by 2010 not explicitly stated 5 years Prevalence of frequent alcohol intake reduced among women to 12–14% by 2010; Prevalence of frequent and heavy alcohol intake reduced to 1–4% by 2010 — 5 years Prevalence of nonuse reduced to 23–35% by 2010 — 5 years Prevalence of failure to get sigmoidoscopy every 5 years reduced to 45–55% by 2010; Prevalence of failure to get mammography every 2 years reduced to 10–15% by 2010; Prevalence of failure to get PSA test every 2 years reduced to 20–30% by 2010 not explicitly stated 5 years Prevalence of failure to get best therapy reduced to 14–16% by 2010 not explicitly stated 5 years sidering major cancer risk factors that had been changing over time (see Table 2.1). They assumed a 10-year latency for the effects of tobacco and a 5-year latency for the effects of other factors. They reported two sets of calculations: (1) calculations in which the current trends in the reductions of risk factor prevalence continued, and (2) calculations in which the trend in the reduction of risk factor prevalence accelerated because of increased

OCR for page 30
Fulfilling the Potential of Cancer Prevention and Early Detection SOURCE: PhotoDisc, Inc. public health efforts. The future rates of cancer were calculated by the following formula: (1990 population attributable risk2) × (projected reduction after the period of latency in the prevalence of the risk factor). The total reduction in cancer incidence and mortality rates was calculated as a weighted average of the rates of cancer of the lung, other tobacco-related sites, colon-rectum, breast, prostate, and other sites. Byers and colleagues found that past and future reductions in rates of tobacco use were the single largest contributor to the projected future declines in overall cancer incidence and mortality rates. Other risk factors for which declines in prevalence were projected to be important contributors to declines in cancer incidence and mortality rates were poor dietary practices (low levels of consumption of fruits and vegetables), especially for colorectal and lung cancers; levels of alcohol intake; and failure to be screened, especially for colorectal cancer and, to a lesser extent, breast and prostate cancer. Byers and colleagues estimated that if the reductions in the prevalence of risk factors continued at the 1990 rate, the total reduction in the cancer incidence rate would be 13 percent from 1990 to 2015 and the reduction in the cancer mortality rate would be 21 percent from 1990 to 2015. If the declines in the prevalence of these risk factors accelerated, however, the cancer incidence rate could decline by 19 percent and the cancer mortality rate could decline by 29 percent. When specific cancers are considered, some of the projections based on accelerated declines in risk factor prevalence were 2   The population attributable risk is the proportion of all cancers in the population due to a particular risk factor. Its calculation is derived from estimates of the relative risk for each risk factor (the ratio of the cancer incidence rate among those exposed to a risk factor divided by the rate among those not exposed) and the proportion of the population exposed to that risk factor (Byers et al., 1999).

OCR for page 30
Fulfilling the Potential of Cancer Prevention and Early Detection striking: they calculated that up to 33 percent of deaths from colorectal and breast cancer could be eliminated and that up to 50 percent of the occurrences of and deaths from tobacco-related cancers could be eliminated by 2015 if the prevalence of each of the specified risk factors was to decline at an accelerated pace. Although these projections apply to the entire population, if the rates of decline in risk factors vary by race, ethnic group, or income, disparities in cancer incidence and mortality rates may increase. It is possible that the estimates of Byers and colleagues are overly optimistic. They assume in their model that current rates of decline in risk factor prevalence will continue or even accelerate, but such declines may not occur. It is possible that persons who have not adopted behavioral changes to date (e.g., those who continue to smoke, avoid dietary change, or avoid screening recommendations) may be more resistant to change, so that it may be difficult to sustain the pace of change over time as “resistors” become a larger proportion of the remaining “unchanged” population (Rogers, 1993). In addition, Byers and colleagues did not estimate future increases in obesity-related cancers (e.g., postmenopausal breast cancer and colon cancer) due to the adverse trends in obesity in the United States. They stated that they assumed the obesity epidemic would be turned around in the coming years, thus negating the past adverse trends, but this turnaround is far from certain. If the obesity trends continue, adverse effects will be seen on breast and colon cancer by the year 2015. Presentation and analysis of historical trend data would be useful in an evaluation of the feasibility of achieving estimated cancer mortality rate reductions as a result of increased screening. For instance, since the late 1980s a decline in the rate of mortality from breast cancer has been observed in the United States. There is debate about how much, if any, of this decline is attributable to increased rates of screening and earlier detection of invasive disease as compared to improved treatment. This debate is especially active for prostate cancer, for which the certainty of screening benefit is less, but which has shown a sharp mortality decline since 1990. Despite the lack of scientific consensus on the explanations for cancer mortality declines, it seems reasonable to state that the detectable favorable changes in incidence or mortality rates that occur as a result of population-wide shifts in risk factors usually take at least several years to become manifest. Byers and colleagues discuss the value of well-reasoned predictions with respect to achievable reductions in cancer incidence and mortality rates, noting that such predictions are far better than guesses. They state that although all causes of cancer are not known, many of the major population-level determinants have been identified and that over the next decade expanded efforts to reduce the prevalence of these known, major risk factors could have a substantial impact on the population burden of cancer.

OCR for page 30
Fulfilling the Potential of Cancer Prevention and Early Detection SUMMARY AND CONCLUSIONS The models reviewed in this chapter illustrate the promise of stepped up efforts to assist individuals and communities in reducing the prevalence of modifiable risk factors but also suggest the difficulties of achieving large reductions in risk factor prevalence. The earliest models estimated the potential of reducing cancer incidence and mortality rates in the best of all possible worlds. When attempts were made to translate the available evidence into national goals, overly optimistic projections of behavioral change were incorporated into the models, leading to unrealistic expectations. Later models have generated more realistic projections by incorporating historical trends in rates of change in risk factor modification and participation in screening programs. In the earliest of these analyses, Peto and Doll concluded that 75 to 80 percent of all cancers in the United States in 1970 could have theoretically been avoided if the population of the United States could be like those of the countries in which the incidence of cancer was the lowest. Those investigators did not distinguish the contributions of environmental and occupational exposures to carcinogens from individual risk behaviors and noted that realistic and practical reductions in the cancer burden would be substantially lower than 75 to 80 percent. The analysis by Willett and colleagues (1996) found that the cancer mortality rate could theoretically be reduced by 60 percent but that a more realistic reduction based on trends in risk behaviors would be 33 percent. The model developed by NCI in 1986 relied on overly optimistic assumptions regarding population-level changes in risk behaviors that resulted in setting an unrealistic national goal to halve the total cancer mortality rate between 1980 and 2000 (in fact, the mortality rate was reduced by only 7 percent for 1980 to 1997). In the last analysis reviewed, Byers and colleagues estimated that by 2015 there would be a 13 percent total reduction in the cancer incidence rate and a 21 percent reduction in the cancer mortality rate if the prevalence of the major modifiable risk factors continued to decline at the 1990 rate; but if there were accelerated declines in these risk factors, they estimated that the cancer incidence rate could decline by 19 percent and that the cancer mortality rate could decline by 29 percent Using very different methods, the estimates of Willett et al. and Byers et al. are quite similar—that about a one-third reduction in cancer mortality is feasible with the concerted application of current knowledge. Improvements in the ability to project the impacts of interventions such as primary prevention and screening on observed trends in cancer incidence and mortality rates are likely with further development of analytic modeling techniques. One important initiative is the NCI-sponsored Cancer Intervention and Surveillance Modeling Network, through which a network of investigators has evaluated the impacts of population-level changes in smok-

OCR for page 30
Fulfilling the Potential of Cancer Prevention and Early Detection ing, diet, physical activity, weight status, and the use of screening tests on the rates of cancer incidence and mortality (http://www-dccps.ims.nci.nih.gov/SRAB/cisnet.html). To date, scientific estimates of the proportion of the cancer burden that can be eliminated if the population distributions of major risk factors were to be shifted in a positive direction have focused on the important but relatively narrow issues of strength of risk factor-disease associations and biological latency. The issue of social or political latency to support behavioral change has gone unaddressed. More research needs to be done on these issues, perhaps through historical case studies of previous successes and failures in encouraging widespread changes in the health of the population. For example, a recent publication of population-level trends in the mortality rate over the 15 years after the tobacco tax was implemented in California shows a substantial lag from the passage of the proposition to the implementation of programs, to reductions in tobacco use, and finally to reductions in lung cancer mortality rates (Fichtenberg and Glantz, 2000). Likewise, the estimate of the rate of decline in tobacco smoking will reflect the level of government and community commitment, as indicated by the experience in California and Massachusetts, where the rate of decline after the implementation of comprehensive programs was faster than that observed in other states (Biener et al., 2000; Pierce et al., 1998; Siegel et al., 2000). Since the introduction of a comprehensive tobacco control program in Massachusetts in 1992 and 1993, the prevalence of smoking among adults has decreased annually by 0.43 percent, whereas in other states there has been an increase of 0.03 percent (Biener et al., 2000). Another important area of research is how observed disparities in cancer incidence and mortality rates might be affected by divergent patterns of risk behaviors observed within certain racial and ethnic and socioeconomic groups. Goal-setting philosophies vary across individuals and organizations. Goals usually incorporate an element of challenge to improve prevention practice and an element of hope that new research will lead to favorable outcomes. Although setting lofty goals may be intended to provide extra motivation to individuals and institutions, goals that are unrealistic can have the opposite effect. If year after year goals are largely unmet, credibility may begin to wane, affecting both individual and political motivation. Consistently unmet goals may be construed as organizational failures and may therefore threaten future cancer prevention funding and undermine the potential value of public health more broadly. Challenge goals that are set to motivate the application of current knowledge and the research for new knowledge must be accompanied by the attention and resources needed to move toward success. National cancer-related objectives set as part of the Healthy People 2010 initiative are described in Chapter 9. These objectives differ from the projections reviewed in this chapter, but they have important implications for public health programs.

OCR for page 30
Fulfilling the Potential of Cancer Prevention and Early Detection In reviewing the models of potential reductions in rates of cancer incidence and mortality examined in this chapter, it is important to recognize that the benefits of primary prevention strategies against cancer (as distinguished from secondary prevention, i.e., screening) will lead to improvements in public health that go well beyond cancer alone, and these improvements may occur quite rapidly (Colditz and Gortmaker, 1995). A person who stops smoking today or who is prevented from starting smoking instantaneously reduces his or her risk of respiratory and cardiovascular problems, although the effect on the lung cancer risk may not be seen for many years (US DHHS, 1990). Similarly, a person who adopts a healthier diet and a more active lifestyle will reduce his or her risk of heart disease, stroke, diabetes (Knowler et al., 2002) and many other health problems that may have shorter latency periods than cancer.