Life expectancy in the United States continues to lengthen.

Reduced Mammographic Screening May Explain Declines in Breast Carcinoma in Older Women Robert M. Kaplan, PhD and Sidney L. Saltzstein, MD, MPH wz OBJECTIVES: To examine whether declines in breast cancer in the oldest-old women correspond with declines in the use of cancer testing. DESIGN: Cross-sectional evaluation of three databases. SETTING: Public access data. PARTICIPANTS: Cases recorded in the California Registry and the Surveillance, Epidemiology, and End Results Program between 1988 and 1997. The study also included respondents to the 2002 Centers for Disease Control and Prevention Behavioral Risk Factor Surveillance System. MEASUREMENTS: Cancer incidence and self-reported mammography within the previous 2 years. RESULTS: Most previous analyses have combined all individuals aged 75, 80, or 85 and older. Creating separate categories for age ranges 85 to 89, 90 to 94, 95 to 99, and 100 and older suggests different incidence patterns for a variety of cancers, including mammary carcinoma in situ (CIS). Between the ages of 40 and 74, there is a significant rise in CIS. Beginning at age 75, there is a significant decline in CIS through the highest age categories. The use of mammographic screening increases between the ages of 40 and 60. Beginning at age 75, there is a significant decline in the use of mammography that parallels the decline in incident cases of CIS. CONCLUSION: There may be a substantial reservoir of undiagnosed CIS in the population. Surveillance bias might explain the decreasing incidence of CIS with advancing age in the oldest age groups. Autopsy studies are needed to estimate the true prevalence of CIS in older women. JAm Geriatr Soc 53:862 866, 2005. Key words: mammary carcinoma in situ (CIS); mammography; oldest old; screening; cancer epidemiology From the Department of Health Services, University of California at Los Angeles, Los Angeles, California; and Departments of w Family and Preventive Medicine and z Pathology, University of California at San Diego, San Diego, California. Address correspondence to Robert Kaplan, PhD, Professor, Chair, Department of Health Services, UCLA School of Public Health, PO Box 951772, Room 31 293C CHS, Los Angeles, CA 90095. E-mail: rmkaplan@ucla.edu DOI: 10.1111/j.1532-5415.2005.53263.x Life expectancy in the United States continues to lengthen. 1,2 Although there are many explanations for this increasing life expectancy, evidence suggests that an increasing portion of the population lives to be older than 85. 3,4 Some estimates suggest that, by 2050, about one person in 43 will be aged 90 and older. 5 As more people live to older ages, there is also an increasing incidence of chronic disease, although much of the disease in the oldest adults is occult or subclinical. 6 The term pseudodisease has been used to refer to the subset of subclinical disease that will never become clinically relevant. 7 Specifically, pseudodisease will not affect life expectancy or quality of life. When subclinical or occult diseases have strong potential for future poor health outcomes, screening may be beneficial, but a framework for screening older adults has been offered with the argument that current life expectancy and consequences of screening must be considered, because screening older adults is often unlikely to offer improved health outcomes. 6 The American Geriatrics Society currently recommends annual or biennial mammography until age 75 with continued testing every 3 years for women with an estimated life expectancy of four or more years. 8 Considerable evidence suggests that the incidence of cancer systematically increases with age 9 up to a point, but most analyses lump together all age groups greater than 65, 75, or 85. It has been rare to include separate categories for those aged 85 to 89, 90 to 94, or 95 to 99. Recently, cancer incidence in the California and U.S. populations was described, and separate categories for the oldest age groups were included. 10 The study found that the number of cancers declines for adults aged 80 and older. It is unclear whether the incidence of cancer actually declines with increasing age or whether cases are missed because there is less surveillance in the oldest age groups. The purpose of this study was to examine whether declines in cancer in the oldest old correspond with declines in the use of cancer testing. Invasive carcinoma and carcinoma in situ (CIS) of the breast were used as an example. METHODS Cross-sectional data from three sources were evaluated: the California Cancer Registry (CCR); the Surveillance, Epidemiology, and End Results (SEER) Program; and the Behavioral Risk Factor Surveillance System (BRFSS). JAGS 53:862 866, 2005 r 2005 by the American Geriatrics Society 0002-8614/05/$15.00

JAGS MAY 2005 VOL. 53, NO. 5 MAMMOGRAPHY IN OLDER WOMEN 863 CCR and SEER Cancer became a reportable disease in California in 1988. Since that time, all cases of cancer have been reported to the CCR. This analysis uses data reported between 1988 and 1997. The file contains 1,346,859 cases. Of these, 1,204,960 were invasive and 141,888 were CIS. The system contains demographic information, including age, race/ethnicity, sex, marital status, place of birth, county of residence, and date of diagnosis. It also includes stage of disease and pathologic data, including cell differentiation and in situ versus invasive status. In addition to the California data, national U.S. data (California excluded) from the SEER Program sites of the National Cancer Institute for the same 10 parallel years were used for comparison. 11 The data have been described in detail elsewhere. 10 In women aged 90 and older, 94% of breast CIS is ductal CIS (DCIS); in women younger than 70, 70% of the CIS is DCIS, the remainder being mostly lobular CIS, mixed DCIS plus lobular CIS, or in a small percentage of cases, unspecified type (unpublished data). Behavioral Risk Factor Surveillance System The BRFSS is a collaborative project between the Centers for Disease Control and Prevention and the various U.S. territories and states. The BRFSS samples the adult population aged 18 and older using telephone surveys. The study uses a random sample in each state of all households that have a telephone. Approximately 95% of all households have telephone service. Interviews are conducted using Computer-Assisted Telephone Interviewing. BRFSS includes a set of core questions that are asked in all states. These questions cover health status, health insurance, diabetes mellitus, tobacco use, and selected cancer screening, including mammography. The core questions also include demographic characteristics of the sample. The 2002 BRFSS, which included 99,262 male and 148,708 female respondents, was used. The analysis considered whether a woman had had a mammogram within the previous 2 years. In the 2002 BRFFS, women were selected if they were aged 40 and older. There were 98,809 female respondents aged 40 and older. Two questions were used to determine whether a mammogram had been performed in the previous 2 years. The first question asked whether the woman had ever had a mammogram, and the second asked how long ago the test had been given. Two thousand four hundred sixty-four of the responses were missing, leaving 96,345 cases for analysis. RESULTS Between 1988 and 1997, breast CIS accounted for about 20% of all female cancers for women younger than 70 (19,747 cases). The breast was the most common site of CIS for women aged 70 to 89, accounting for 41.5% of all female CIS (6,775 cases), but in women aged 90 and older, mammary CIS was the third-most-common cause of CIS, with only 108 cases. Age was broken down into 12 categories in 5-year blocks, ranging from aged 40 to 44 to aged 95 to 99. Table 1 shows the number of women responding to BRFFS in each age category and the number in each category receiving Table 1. Behavioral Risk Factors Surveillance System Participation and Mammography in Previous 2 Years by Age Age n % Mammogram in Previous 2 Years Receiving Mammogram (%) 40 44 15,276 15.7 9,928 65 45 49 14,607 15.0 10,780 73.8 50 54 13,731 14.1 10,878 79.2 55 59 11,517 11.8 9,273 80.5 60 64 9,505 9.8 7,596 79.9 65 69 8,945 9.2 7,105 79.4 70 74 8,478 8.7 6,663 78.6 75 79 7,349 7.6 5,532 75.3 80 84 5,077 5.2 3,442 67.8 85 89 2,112 2.2 1,191 56.4 90 94 669 0.7 282 42.2 95 94 127 0.1 43 33.9 Total 97,393 100.0 72,713 mammography in the previous 2 years. In addition, the percentage of women in each age category who completed a mammogram in the previous 2 years is listed. Figure 1 shows the rate of breast CIS per 100,000 women broken down by age. The figure shows that cases peak at age 70 to 74 and then systematically decline. Figure 1 also shows the percentage of women who reported receiving a mammogram in the previous 2 years on the BRFSS. Mammography increases with age until age 50 to 54 and then plateaus. Then, beginning with age 75 to 79, mammography systematically declines, clearly paralleling the decline in breast CIS. Using 5-year age groupings as the unit of analysis, the correlation between mammography and breast CIS was highly significant (correlation coefficient 5 0.86, Po.01). This analysis uses data on CIS from California and other SEER registries but uses national data for mammography. National data were used because there are more cases and because reliability should be higher; whether mammography patterns in key SEER states (California, Connecticut, Hawaii, Iowa, New Mexico, and Utah) were different than the national patterns was also checked. Figure 2 shows that the pattern was similar. There was some minor divergence after age 90, but this was likely due to the lower reliability in some SEER states because of the smaller sample size. Discussion Evidence from the CCR, SEER, and the BRFFS demonstrates that the incidence of breast CIS increases with age until about age 75. After age 75, mammary CIS and breast cancer testing decline. These results are consistent with earlier reports in the literature. For example, it has been shown that there is a decreasing risk of incident breast cancer cases beyond age 75. It was suggested that the lower risk at older ages might be an artifact of screening patterns. 12 These results are important for two reasons. First, they illustrate the importance of using separate age categories for

864 KAPLAN AND SALTZSTEIN MAY 2005 VOL. 53, NO. 5 JAGS Figure 1. Breast carcinoma in situ (CIS) and mammography by age. Based on CIS data 10 and mammography from the Behavioral Risk Factor Surveillance System. DCIS 5 ductal CIS. individuals aged 75 and older. Previous analysis that combined all participants in the old age categories missed these trends toward declining breast cancer in older women. 10 The second reason these results are important is that they provide evidence relevant to the disease reservoir hypothesis. The hypothesis suggests that disease is common, particularly in the oldest members of society. If disease is present, the more we look for it, the more likely it is to be found, yet much of detected disease may actually be pseudodisease, a subcategory of subclinical disease that has little or no clinical importance. 10 Pseudodisease will go undetected unless there is surveillance or autopsy. Evidence from this study suggests that testing for mammary CIS increases and then decreases with advancing age. Thus, one Figure 2. Mammography rates: Total United States versus Surveillance, Epidemiology, and End Results (SEER) states.

JAGS MAY 2005 VOL. 53, NO. 5 MAMMOGRAPHY IN OLDER WOMEN 865 explanation for the decline in mammary CIS with advancing age is that physicians simply are not ordering the tests that would detect it. In one study, autopsy data were used to estimate the reservoir of DCIS. 13 Using hospital-based and forensic autopsy data, the prevalence of occult invasive breast cancer or DCIS was estimated over a series of autopsy evaluations. In seven autopsy series for women who were not known to have breast cancer during their lifetimes, they found a median prevalence of DCIS of 8.5%. In addition, the autopsy studies differed in the number of slides per breast that were examined. There was a relationship between number of slides examined and probability of detecting mammary DCIS. Although all of the series had potential selection biases, these results suggest that there is a substantial reservoir of undetected mammary DCIS and that the greater the scrutiny, the higher the probability that DCIS will be detected. The current analysis did not separate CIS into subcategories of ductal, lobular, and mixed, but 94% of the observed cases in women 90 and older were DCIS. The results of the current study are not surprising because mammography is the most effective means of detecting DCIS. It might be argued that clinically significant is defined as being detectable and that disease in its predetectable phase may be of less importance. Thus, autopsy studies will identify true pseudodisease, whereas screening studies will identity disease that has a higher probability of becoming clinically important. The relationship between age, tumor characteristics, sojourn time, and screening is complex. Some will argue that screening identifies treatable cases and allows the reservoir to be depleted. As age increases, sojourn time also becomes longer. Thus, screening women in their 50s and 60s results in reductions of incident cases later in life by removing emerging cases from the reservoir. Although this explanation is attractive, it cannot account for the results. Differences in breast cancer between women screened and not screened in randomized, clinical trials are not large enough to account for the large reductions in incidence observed in the oldest women. 14 Although less-frequent mammography provides a greater opportunity for in situ disease to progress to invasive cancer, 15 the rate of transition from CIS to invasive cancer is not known, although based on autopsy studies, it appears that most CIS in the oldest women does not become invasive or metastatic. Many will take issue with the argument that screening mammography in women aged 75 and older is not necessary. Some argue that most of the reduction in mortality associated with mammographic screening results because of early diagnosis of invasive tumors. The detection of DCIS contributes little to mortality reduction. 16 Nevertheless, the same study showed that DCIS is likely to become invasive cancer. Another study used data from the Swedish twocountry trial and from screening programs in England, the Netherlands, Australia, and the United States to model transition of DCIS to invasive cancer. It suggested that twothirds of DCIS is of the variety that can progress to invasive cancer. It estimated that the interval between DCIS and invasive cancer is about 4 years. Whether or not these models justify continued screening remains uncertain. 15 There are other compelling reasons for the decline in breast cancer with older age. Older women with undetected breast cancer may be more likely to die from other causes. Women with other serious chronic illnesses may also be less likely to be tested for breast cancer. In addition, why the rate of ordering mammography declines with age was not discovered. It is possible that physicians intuitively know that breast cancer mortality declines with age and therefore do not order the tests. There are at least two reasons for ordering mammography: screening and clinical evaluation. Because mammography is often ordered in response to lump identification, declining breast cancer may explain the decline in mammography. Some autopsy evaluations suggest that the true incidence of CIS declines with age. 17 However, one report suggests that there is a substantial reservoir of undiagnosed CIS that greater surveillance would identify. 10 An important limitation of the current study was that, although cancer cases from California and other SEER registries were used, estimates of mammography from the entire U.S. population were used. One analysis demonstrated that the California pattern of declining CIS in the older adult population mirrors that observed at other SEER registries. 10 In addition, rates of mammography by age in SEER states are similar to those in other states (Figure 2). Another study, using a different survey of California residents (the California Health Interview Survey), also found a decline in the use of mammography in the oldest ages groups. 18 Thus, the conclusions based on the current study would not change if screening data from SEER states were used. In conclusion, to understand the incidence of breast cancer in older women, it is important to include separate categories of women aged 85 to 89, 90 to 94, 95 to 99, and 100 and older. Studies using these older age categories show a decline in CIS and invasive breast cancer with advancing age. Decreased surveillance in older women may explain these declines. Future studies of the rate of CIS in asymptomatic healthy women are needed to better characterize the influence of subclinical breast cancer and the risks and benefits of screening mammography. REFERENCES 1. Buttler G. [Increasing life expectancyfwhat are the promises of demography?]. Z Gerontol Geriatr 2003;36:90 94. 2. Lubitz J, Cai L, Kramarow E et al. Health, life expectancy, and health care spending among the elderly. N Engl J Med 2003;349:1048 1055. 3. Tuljapurkar S, Li N, Boe C. A universal pattern of mortality decline in the G7 countries. Nature 2000;405:789 792. 4. Health, United States, 2003 with Chartbook on trends in the Health of Americans. Atlanta: Centers for Disease Control and Prevention, 2003. 5. Day J. Population projections of the United States by age, sex, race and Hispanic origin, 1995 2050. Current Population Reports. Washington DC: U.S. Government Printing Office, U.S. Bureau of the Census, 1996. 6. Walter LC, Covinsky KE. Cancer screening in elderly patients: A framework for individualized decision making. JAMA 2001;285:2750 2756. 7. Black WC, Welch HG. Screening for disease. AJR Am J Roentgenol 1997;168:3 11. 8. AGS Clinical Practice Committee. American Geriatrics Society Position Statement: Breast Cancer Screening in Older Women [on-line]. Available at www.americangeriatrics.org/products/positionpapers/brstcncr.shtml Accessed January 12, 2005. 9. Cancer Facts and Figures. Atlanta, GA: American Cancer Society, 2003. 10. Saltzstein SL, Behling CA. Cancer in the Chronically Gifted. A Study of 45,000 Cases of Cancer in People 90 Years of Age and Older. San Diego, CA: San Diego and Imperial Organization for Cancer Control; Cancer Surveillance Programs of Orange County, 2003.

866 KAPLAN AND SALTZSTEIN MAY 2005 VOL. 53, NO. 5 JAGS 11. Surveillance, Epidemiology, and End Results (SEER) Program Public-Use CD- ROM (1973 97) Rockville, MD: National Cancer Institute, Cancer Statistics Branch, DCPC Surveillance Program, 2000. 12. Kessler LG. The relationship between age and incidence of breast cancer. Population and screening program data. Cancer 1992;69 7 Suppl:1896 1903. 13. Welch HG, Black WC. Using autopsy series to estimate the disease reservoir for ductal carcinoma in situ of the breast: How much more breast cancer can we find? Ann Intern Med 1997;127:1023 1028. 14. Miller AB, To T, Baines CJ et al. Canadian National Breast Screening StudyF2: 13-year results of a randomized trial in women aged 50 to 59 years. J Natl Cancer Inst 2000;92:1490 1499. 15. Yen MF, Tabar L, Vitak B et al. Quantifying the potential problem of overdiagnosis of ductal carcinoma in situ in breast cancer screening. Eur J Cancer 2003;39:1746 1754. 16. Duffy SW, Tabar L, Vitak B et al. The relative contributions of screen-detected in situ and invasive breast carcinomas in reducing mortality from the disease. Eur J Cancer 2003;39:1755 1760. 17. Stanta G, Campagner L, Cavallieri F et al. Cancer of the oldest old. What we have learned from autopsy studies. Clin Geriatr Med 1997;13: 55 68. 18. Walter LC, Eng C, Covinsky KE. Screening mammography for frail older women: What are the burdens? J Gen Intern Med 2001;16:779 784.