|Home | About | Journals | Submit | Contact Us | Français|
Inclusion of minorities is an important but challenging aspect of epidemiologic studies in the US. One aspect of this challenge that has received little attention is the actual number of minorities affected with specific cancers. We aimed to understand how population characteristics affect the numbers of minority cancer cases in Surveillance, Epidemiology and End Results (SEER) regions.
Using SEER data, we identified 6 cancers with higher incidence in racial and ethnic minorities and reviewed the annual number of cases of those cancers in SEER areas where there are large numbers of blacks, Hispanics, and Asians. We examined the age characteristics of the populations in SEER areas using data from the U.S. Census.
For blacks, while there are substantial numbers of cases for the most common cancers with higher incidence in this group, numbers are quite small for other cancers: <150 cases, and in many areas, <100 per year. Few registries have substantial numbers of Hispanics or Asians. As expected, the proportion of the minority populations is lower in older age groups, while the proportion of non-Hispanics whites is larger.
Because of the sharp decline in minority populations associated with age, and the high age-specific incidence of most cancers, the actual number of minority cases is quite small for several cancers. This is a further challenge to including minority groups in studies of any but the most common cancers.
Epidemiologists attempt to include racial and ethnic minorities in their studies in order to obtain a study population that reflects the source population. We are particularly concerned with including minorities in studies of those diseases that have higher incidence in minorities in order to understand the sources of health disparities. Motivated by an effort to understand the reasons for having relatively small numbers of blacks enrolled in a study of pancreatic cancer in New York City, we investigated the actual numbers of black cases in the area, which we found to be surprisingly small in light of the large minority population in this area and the higher incidence of pancreatic cancer in blacks. This investigation was expanded to other cancers that have higher age-adjusted incidence in blacks and other minorities and to Surveillance, Epidemiology and End Results (SEER) Registry areas in order to gain an understanding of how age-adjustment and age distribution affect the practical aspects of including minority groups in epidemiologic studies. We focused on SEER areas because epidemiologic studies are frequently conducted in these geographically-defined areas and population data are available.
Using data from the American Cancer Society1 we determined the most common cancers in the U. S. in 2004, choosing this year as the midpoint of the period for which recent data on numbers of cases were available (2002–2006) from the SEER Cancer Statistics Review 1975–2006.2 We limited our analysis to 14 cancers with at least 20,000 cases in total. These cancers (and the numbers diagnosed in 2004, in thousands) were: prostate , lung/bronchus , breast , colorectal , bladder , non-Hodgkins lymphoma , melanoma , kidney/renal pelvis , uterine corpus , pancreas , oral cavity and pharynx , ovary , thyroid , stomach . We then considered cancers with higher age-adjusted incidence3 in minorities. Compared to non-Hispanic whites, age-adjusted incidence in the 17 SEER registries was higher in blacks for 6 of these 14 cancers: prostate, lung/bronchus, colorectal, kidney/renal pelvis, pancreas, and stomach. Stomach cancer was the only one with higher incidence in Hispanics and Asians. For American Indian/Alaska Natives (AI/AN), the incidence of both stomach and kidney/renal pelvis cancer was higher than in non-Hispanic whites.
Because the size of the minority populations in each SEER registry varies considerably, when reporting numbers of cases in minority groups, we included data from any given registry only for those minorities for which the total population of that minority was at least 100,000 in 2000.4 We reasoned that researchers concerned about including a particular minority would be most likely to use registries where there is a reasonable number of that minority group. Incidence rates, numbers of cases, and median age at diagnosis
Age-adjusted incidence rates and numbers of cases for minorities and non-Hispanic whites for the years 2002–2006 for the selected cancers in each registry were obtained from SEER*Stat.3 We included cases with earlier primary cancers, those without microscopic confirmation, and those diagnosed at autopsy or death certificate. The San Francisco/Oakland and San Jose/Monterey registries were combined to comprise the Greater Bay area, as defined by SEER. SEER*Stat was also used to estimate the median age at diagnosis3 for each racial and ethnic group by cancer site. We reported data for whites excluding Hispanics, whereas the category “Hispanic” is not mutually exclusive from black, Asian/Pacific Islander, and American Indian/Alaska Native. Identification of an individual as “Hispanic” is based on the North American Association of Central Cancer Registries (NAACCR) Hispanic Identification Algorithm (NHIA).5
We used intercensal and post-censal population estimates from the National Center for Health Statistics (NCHS) and the Census Bureau6 to describe the percentage of the population that is black, Hispanic, Asian, and AI/AN in 5-year age groups from 20–24 through >85 in 2004 in combined SEER registries. For the SEER registries covering areas other than states, we combined census data from the counties included (shown in footnote to Table 1). To quantify the changes in the proportion of each racial/ ethnic group, we calculated the ratio of the proportion in the oldest age group to the proportion in the youngest age group for the SEER registries as a whole and for the individual registries with the largest numbers of blacks, Hispanics, and Asians.
Table 1 shows age-adjusted incidence rates for the 6 cancers with higher overall incidence rates in blacks, Hispanics, Asians, or AI/AN than in non-Hispanic whites in the 17 SEER areas in total and, except for AI/AN, in the individual SEER areas. Individual areas are ordered by population size from largest to smallest. For several registries, particularly the smaller ones, statistics are based on <16 cases or cannot be calculated. Among blacks, the overall pattern of higher incidence compared to non-Hispanic whites is consistent for most of the cancers, with some exceptions, particularly in the smaller registries and for lung cancer and kidney/renal pelvis cancer. Among Hispanics and Asians, incidence of stomach cancer is higher than in non-Hispanic whites in all registries with adequate statistics.
Table 2 shows the average annual number of patients with the 6 cancers for which the age-adjusted incidence is higher in blacks in each of the 10 registries with black population ≥100,000. Those places where the number of cases is 100–149 or <100 are highlighted. For non-Hispanic whites, there are >150 cases of all cancers studied in these 10 registries, with the exception of pancreatic and stomach cancers in Atlanta. Among blacks, for the more common cancers (prostate, lung/bronchus, colorectal), there are >150 cases in most of the registries studied, with the exception of Seattle, and, for colorectal cancer, Connecticut. For the less common cancers (kidney/renal pelvis, pancreas, and stomach), there are <150 black cases in each of the 10 registries, with the exception of kidney/renal pelvis and pancreas cancers in Louisiana. In 7 of the 10 registries, there are <100 cases per year for some or all of these less common cancers. Table 3 shows the number of non-Hispanic whites, Hispanics, and Asians with stomach cancer in each registry studied. Only Greater California and Los Angeles have ≥150 Hispanics with stomach cancer and only the 3 California registries have ≥150 Asians with stomach cancer per year. For AI/AN, the total number of cases in SEER registries with these cancers is: prostate, 772; lung, 1,013; colorectal, 924; kidney, 409; pancreas, 216; stomach, 239 (not shown in tables).
Table 4 contains data on the median age at diagnosis across the racial/ethnic groups studied. Among non-Hispanic whites the median age at diagnosis ranges from 66 (for kidney/ renal pelvis cancer) to 73 (for pancreas and stomach cancers), while blacks are diagnosed at younger ages (range 61–68) for each of these cancers. For stomach cancer, Hispanics are diagnosed at a median age of 66, while Asians are diagnosed at a median age of 70. AI/AN are diagnosed with stomach and kidney cancer at younger ages than the other groups, 64 and 60, respectively.
For the combined SEER registries, excluding Alaska, Figure 1 shows the number of people in each of the racial /ethnic groups in 5-year age categories from 20–24 through >85 years. It can be seen that the number of individuals is generally smaller in each successive age group, particularly at ages >50. The increase in non-Hispanic whites between the ages of 25 and 49 reflects births in the US from the late 1940s to the 1970s7, the period of the post-World War II baby boom. Figure 2 shows the percentage of each racial/ethnic group in the 20–24 through >85 age categories in the SEER areas. Comparing across age groups, there is a clear increase in the proportion of the population that is non-Hispanic white, while Hispanics show a decline. Blacks and Asians have similar proportions in the population and similar changes in relation to age group; the proportion who are AI/AN is small overall. To quantify the changes between age groups, we calculated the ratio of the proportion in the oldest age group to that in the youngest age group for each racial/ethnic group (Table 5). For non-Hispanic whites, the proportion in the oldest age group is considerably higher than in the youngest group, with a ratio of 1.59. For the other groups, the ratio is below 1: 0.67 for Asians, 0.46 for blacks, 0.26 for Hispanics and 0.25 for AI/AN. Examples of different population structures among registries are also illustrated in Table 5. For example, for blacks in Louisiana and Atlanta, the ratios are somewhat higher (0.62 and 0.53) than in New Jersey (0.37). For Hispanics, the ratio is slightly higher in Los Angeles than in Greater California and the Greater Bay Area. For Asians, the results from Los Angeles show relatively little change across age groups (ratio=0.86), while in Hawaii there is a higher proportion of Asians in the oldest group than in the youngest, with a ratio of 1.33 (data not shown in table).
Our examination of SEER and census data shows that, even with higher incidence in minorities, the actual number of minority patients in SEER registry areas for all but the most common cancers is usually quite small. This is related to the population structure of the U.S., reflected in the SEER registry areas, with minority groups having a relatively small proportion of the total population in older age groups likely to be affected by cancer. Comparing ratios of the proportions in oldest to youngest age groups, we noted the largest differences in population composition for Hispanics and AI/AN, intermediate changes for blacks, and smaller changes for Asians, and also noted differences among SEER areas for blacks and Asians. The reasons for differences in population structure vary for the ethnic groups: the relatively smaller numbers of blacks and AI/AN in older age groups are related at least in part to lower life expectancy8, while those noted for Hispanics are more likely related to their recent immigration to the U.S.9 The smaller differences for Asians, compared to blacks and Hispanics, may represent longer life expectancy.8 Age-standardization is necessary for comparison of incidence rates across populations with different age structures; however, since minorities represent relatively small proportions of the population particularly in older age groups, age-adjustment can distort expectations of the actual number of patients. Since we did not exclude cases with earlier primary cancers, without microscopic confirmation, or diagnosed at autopsy or by death certificate, the incidence rates and numbers of cases reported are maximized, and studies with these exclusions would have even smaller numbers of eligible minority cases. With the exception of kidney/ renal cell cancer, incidence of the cancers we studied has declined or remained stable in most racial / ethnic groups in recent years.10
Population-based approaches using cancer registries for case ascertainment can readily identify minority cases, although locating Hispanics may not be straightforward.11 Minority cases may be more difficult both to contact and to recruit into studies.11–13 Hospital-based studies at large centers allow for ready identification of and contact with cases, but there may be small minority patient populations because of the hospitals’ locations, insurance policies, or factors related to patients’ socioeconomic status.
It continues to be important to recruit minorities into epidemiologic studies to provide generalizable results and to explore differences in risk factors and outcomes. Investigators should make use of successful strategies that have been identified 12–19 and continue to develop and refine these strategies, while recognizing the limitations imposed, for some cancers, by absolute numbers of cases.
Financial Disclosures: There are no financial disclosures from any of the authors
Sara H. Olson, Department of Epidemiology & Biostatistics, Memorial Sloan Kettering Cancer Center.
Tracy M. Layne, Department of Epidemiology & Biostatistics, Memorial Sloan Kettering Cancer Center.
Jennifer A. Simon, Department of Epidemiology & Biostatistics, Memorial Sloan Kettering Cancer Center.
Emmy Ludwig, Department of Medicine, Memorial Sloan Kettering Cancer Center.
Eileen O’Reilly, Department of Medicine, Memorial Sloan Kettering Cancer Center.
Peter J. Allen, Department of Surgery, Memorial Sloan Kettering Cancer Center.
Robert C. Kurtz, Department of Medicine, Memorial Sloan Kettering Cancer Center.