Patterns of cancer mortality among Native Americans

Authors

  • Nathaniel Cobb M.D.,

    Corresponding author
    1. Indian Health Service, Cancer Prevention and Control Program, Albuquerque, New Mexico
    • Indian Health Service, Cancer Prevention and Control Program, 5300 Homestead Road, NE, Albuquerque, NM 87110
    Search for more papers by this author
  • Roberta E. Paisano M.H.S.A.

    1. Indian Health Service, Cancer Prevention and Control Program, Albuquerque, New Mexico
    Search for more papers by this author

Abstract

BACKGROUND

Native Americans have been reported to have lower cancer incidence and mortality than other racial groups in the U.S., although some have questioned whether this was due to racial misclassification. This study provides improved estimates of cancer mortality, determined from a sampling of people who live on Indian reservations.

METHODS

The authors reviewed death certificates from U.S. counties that contain Indian lands, excluding certain areas with known problems of racial misclassification. Age-adjusted mortality rates for specific types of cancer were calculated using U.S. Census population figures, and these rates were compared with rates for all races in the U.S.

RESULTS

This sample included 38% of the American Indian and Alaska Native populations. The age-adjusted annual mortality rate for all cancers combined was 148.2 per 100,000 for both genders, 133.1 for females, and 167.2 for males. The rates for males and for both genders combined, but not for females, were significantly lower than the U.S. rates for all races (P < 0.05). Females had significantly lower rates of death from carcinoma of the lung and breast and significantly higher rates of death from carcinoma of the cervix and gallbladder (P < 0.05). Males had significantly lower rates of death from carcinoma of the lung, colon, and prostate, and significantly higher rates of liver carcinoma. Both genders combined had significantly lower rates of death from lung and colon carcinoma and significantly higher rates of death from stomach, liver, kidney, and gallbladder carcinoma. Geographic differences were substantial, with the Northern and Plains regions experiencing much higher mortality from lung, colon, and breast carcinoma than the Southwest region.

CONCLUSIONS

Compared with the general U.S. population, Native Americans experience quite different patterns of cancer mortality. Cancer prevention and control programs should be designed specifically for this minority population. [See editorial on pages 2247-50, this issue.] Cancer 1998;83:2377-2383. © 1998 American Cancer Society.

During the last decade, cancer was the third leading cause of death for American Indians and Alaska Natives, following heart disease and injuries. In 1993, for the first time, the number of cancer deaths exceeded accidental deaths, moving into second place.1 Published reports over the past 40 years have noted that Native Americans have lower overall cancer incidence2-4 and mortality5-7 than other racial groups in the U.S.. At the same time, rates of mortality from certain cancers, such as those of the cervix, liver, stomach, and gallbladder, are dramatically higher.7 Accurate estimates of cancer mortality are important in planning for cancer prevention and control efforts, and they may serve to generate hypotheses about risk factors and causes of various types of cancer. In this study, we update and refine estimates of cancer mortality among Native American people in the U.S., particularly those residing in areas served by the Indian Health Service.

Native Americans are a diverse group, with many distinct cultures and languages. They live in environments ranging from the deserts of the Southwest to the Alaskan tundra. Subsistence farming and hunting are still common. Over the past 3 decades, an accelerated rate of acculturation has led to major changes in diet8, 9 and an increase in habits such as cigarette smoking and a sedentary lifestyle. Cigarette smoking rates vary considerably; Alaska Natives and Northern Plains tribes smoke heavily, whereas the large Indian populations in the Southwest still smoke very little.

The Indian population is young, with a median age of 24.2 years compared with 32.9 years for all races in the U.S.1 Many are also poor, with 31.6% of Indians below the poverty level, in contrast to 13.1% for the U.S. population (all races).1 Although 44% of Native Americans live in nonurban areas, which are subject to less environmental pollution than urban areas,10 many do not have running water, telephones, electricity, or refrigeration, all of which may affect their health status.

To be eligible for Indian Health Service care, an American Indian or Alaska Native must be an enrolled member of a federally recognized tribe or Alaska Native village, and must reside on or near a reservation. The Indian Health Service is responsible for providing health care to 1.2 million eligible Native Americans directly, through contracts with tribes, through contracts with private providers, or through self-governance funding agreements with tribes. The 1990 U.S. Census enumerated approximately 2 million Native Americans; thus, there are approximately 800,000 people in the U.S. who identify themselves as Native Americans but who are not served by the Indian Health Service. Many in this group live in urban areas where the Indian Health Service does not maintain facilities, and health statistics for this widely dispersed urban population are difficult to obtain.

For this study we used death certificates and U.S. Census data to determine cancer mortality rates for Native Americans. Because of concerns about misclassification of race in areas where American Indians are a small minority, we used as our study sample those American Indians and Alaska Natives who live in the service-area counties of the Indian Health Service, where the concentration of Native Americans is high. In so doing, we hoped to increase the accuracy of our estimates of mortality and population, at the risk of underrepresenting the urban group.

METHODS

By law, every death in the U.S. is recorded on a certificate, which is usually filled out by a coroner, an attending physician, or a funeral director. Death certificates are compiled in each state and sent to the National Center for Health Statistics, where they are stripped of identifiers, edited for consistency, and combined onto a data tape for public use. The information collected on death certificates includes county of residence, race, and cause of death. Those death records identified as American Indian or Alaska Native are then compiled onto a separate tape, which is the primary source of data for this report. To reduce the likelihood of racial misclassification in both death certificates and census counts, we included in our sample only death certificates of Native Americans who resided at the time of death within the Indian Health Service "service area." This area includes counties that contain federally recognized reservation lands, and all adjacent counties.11 Because a large part of the population of these counties is American Indian or Alaska Native, we felt that racial coding on death certificates in those counties would accurately reflect the actual race of the deceased. Cancer patients may die in urban referral hospitals far from their homes, but because Native people from reservation areas generally have their care paid for by the Indian Health Service, the hospital record will have a clear indication of race, which will be reflected on the death certificate.

Three of the 12 administrative areas of the Indian Health Service are known to have underreporting of race on death certificates: the Portland area12 (Washington, Oregon, and Idaho), the Oklahoma City area13 (Oklahoma, Texas, and Kansas), and the California area14 (California only). To provide as reliable a sample as possible, we excluded all counties in these three areas from this study. It should be noted that two of these areas (California and Oklahoma) contain large numbers of American Indians, who comprise almost 30% of the reported U.S. Native population.

Because American Indians from the southwestern U.S. are known to have low rates of some cancers compared with other Native groups, we divided our sample into two geographic regions: Southwest (Albuquerque, Tucson, Phoenix, and Navajo areas) and Northern/Plains (Alaska, Billings, Aberdeen, Bemidji, and Nashville areas). We calculated rates separately for the most common cancers in these two regions.

We obtained population denominators from U.S. Census figures for American Indians and Alaska Natives residing in the same counties, using intercensal estimates as revised in February of 1994. Race on the U.S. Census is self-identified by the individual.

For cancer deaths, we used death records with the underlying cause of death ICD-9 (International Classification of Diseases, 9th revision) codes 140.0-208.9 for the years 1989-1993. In accordance with popular convention, we combined sites for these common cancer types: colon, rectosigmoid, and rectum; liver and intrahepatic duct; lung and bronchus; and kidney and renal pelvis. Some cancer deaths were not specified as to site. We combined these records, including ICD-9 codes 159.1, 195.0-195.8, 196.1-196.9, 199.0-199.1, 202.3, and 202.5-202.6 into the category of "Unspecified."

Death rates were age-adjusted by the direct method to the 1970 U.S. population. We used comparison rates for cancer mortality in the U.S. all-races population from National Cancer Institute15 for the years 1988-1992.

To compute 95% confidence intervals for each rate, we used the method described by Armitage.16 If the U.S. all-races rate was not included in the calculated 95% confidence interval, we considered the rates significantly different at the P < 0.05 level.

RESULTS

The average annual population for the sample of this study was 735,049. This represented 59% of the 1990 Indian Health Service "service" population (1,238,937) and 38% of the 1990 U.S. Census American Indian/Alaska Native population (1,959,234). The Northern/Plains region had a population of 331,849, and the Southwest region had a population of 403,200.

The number of cancer deaths, age-adjusted mortality rates, and comparison rates for the entire U.S. are shown in Tables 1-3. During the study period, there were 3043 deaths from cancer in the study sample; 1535 males and 1508 females died. The age-adjusted annual mortality rate for all sites was 148.2 per 100,000 for both genders combined, 133.1 for females, and 167.2 for males. The rates for males and for both genders combined, but not for females, were significantly lower than the U.S. all-races rates (P < 0.05).

Table 1. Native American Cancer Deaths, Both Genders, 1989-1993
Cancer sitenRateaU.S. rate (all races)
  • a

    95% confidence interval in parentheses. Rate per 100,000, age-adjusted to the U.S. 1970 population.

All sites3043148.2 (136.3-160.2)172.8
Lung63332.3 (26.6-37.9)49.6
Colon26313.1 (9.5-16.7)18.7
Breast1848.2 (5.5-10.9)15.2
Prostate1558.2 (5.3-11.1)9.9
Stomach1677.9 (5.2-10.7)4.7
Pancreas1507.7 (4.9-10.4)8.4
Kidney1356.7 (4.2-9.3)3.5
Liver1316.5 (4.0-9.1)2.9
Cervix894.0 (2.1-5.9)1.6
Non-Hodgkin's813.8 (1.9-5.7)6.3
Gallbladder733.8 (1.8-5.7)0.7
Multiple myeloma723.6 (1.7-5.4)3.0
Ovary673.1 (1.4-4.9)4.4
Unspecified32616.2 (12.2-20.2)11.9

The five most common causes of cancer death for females were carcinoma of the lung, breast, colon, pancreas, and cervix; for males, the top five were carcinoma of the lung, prostate, colon, stomach, and liver. For both genders combined, the top five were carcinoma of the lung, colon, breast, prostate, and stomach.

Compared with the all-races U.S. mortality rates, Native American females experienced significantly (P < 0.05) lower rates of death from carcinoma of the lung and breast, and significantly higher rates of death from carcinoma of the cervix and gallbladder. Males had significantly lower rates of death from carcinoma of the lung, colon, and prostate, and significantly higher rates of death from liver carcinoma. Both genders combined had significantly lower rates of death from lung and colon carcinoma, and significantly higher rates of death from stomach, liver, kidney, and gallbladder carcinoma. The ratio of Native American rates to U.S. all-races rates are shown in Figures 1-3.

Figure 1.

The cancer mortality ratio of Native American males to U.S. males of all races is shown. Cancers are listed in order of their prevalence in the Native American population.

Figure 2.

The cancer mortality ratio of Native American females to U.S. females of all races is shown. Cancers are listed in order of their prevalence in the Native American population.

Figure 3.

The cancer mortality ratio of Native Americans of both genders to both genders of all races in the U.S. is shown. Cancers are listed in order of their prevalence in the Native American population.

Deaths coded as "Unspecified" accounted for 326 (10.7%) of the total, 173 (11.5%) among females and 153 (9.9%) among males. For comparison, 6.9% of all recorded cancer deaths in the U.S. were coded as "Unspecified." This difference was statistically significant for females and for both genders combined.

A comparison between mortality rates in the Southwest region and the Northern/Plains region is shown in Table 4 and Figures 4 (89K) and 5 (85K). There were 1836 cancer deaths in the Northern/Plains region and 1207 cancer deaths in the Southwest region during the study period. For both males and females in the Northern/Plains region, lung carcinoma death rates were six times greater than in the Southwest, and colon carcinoma mortality was four times greater. Breast carcinoma mortality in the Northern/Plains region was double that in the Southwest region, whereas prostate carcinoma mortality was the same in the two regions.

Figure 4.

Cancer death rates for Native American females (American Indians and Alaska Natives) are shown by region. Rates are per 100,000, age-adjusted to the 1970 U.S. Census.

Figure 5.

Cancer death rates for Native American males (American Indians and Alaska Natives) are shown by region. Rates are per 100,000, age-adjusted to the 1970 U.S. Census.

Table 2. Native American Cancer Deaths, Males, 1989-1993
Cancer sitenRateaU.S. rate (all races)
  • a

    95% confidence interval in parentheses. Rate per 100,000, age-adjusted to the U.S. 1970 population.

All sites1535167.2 (148.2-186.2)219.6
Lung37642.6 (33.0-52.3)74.4
Prostate15518.6 (12.1-25.2)26.0
Colon13314.5 (8.9-20.1)23.1
Stomach10010.5 (5.8-15.2)6.8
Liver809.0 (4.5-13.4)4.2
Kidney758.3 (4.1-12.6)5.0
Pancreas657.3 (3.3-11.3)10.0
Non-Hodgkin's434.2 (1.2-7.1)7.8
Multiple myeloma424.6 (1.4-7.7)3.7
Esophagus353.8 (0.9-6.7)6.0
Brain272.3 (0.2-4.3)5.1
Nasopharynx242.4 (0.2-4.6)0.4
Gallbladder212.5 (0.1-4.9)0.5
Unspecified15316.8 (10.7-22.9)14.9
Table 3. Native American Cancer Deaths, Females, 1989-1993
Cancer sitenRateaU.S. rate (all races)
  • a

    95% confidence interval in parentheses. Rate per 100,000, age-adjusted to the U.S. 1970 population.

All sites1508133.1 (117.9-148.3)141.5
Lung25723.8 (17.3-30.4)31.4
Breast18415.2 (10.2-20.2)27.1
Colon13011.9 (7.3-16.5)15.6
Pancreas857.9 (4.1-11.7)7.2
Cervix897.4 (3.8-10.9)3.0
Ovary675.8 (2.6-9.0)7.8
Stomach675.8 (2.6-9.0)3.1
Kidney605.4 (2.3-8.4)2.3
Gallbladder524.8 (1.9-7.8)0.9
Liver514.5 (1.7-7.4)1.9
Non-Hodgkin's383.5 (1.0-5.9)5.1
Multiple myeloma302.7 (0.5-4.9)2.5
Biliary191.7 (0.0-3.5)0.5
Unspecified17315.7 (10.4-21.0)9.7
Table 4. Native American Cancer Mortality Ratesa by Region
Cancer siteSouthwest femalesNorthern/Plains femalesSouthwest malesNorthern/Plains males
  • a

    95% confidence interval in parentheses. Rate per 100,000, age-adjusted to the U.S. 1970 population.

All sites102 (84-120)170 (144-195)113 (92-134)230 (197-263)
Lung7.3 (2.3-12.2)43.2 (30.3-56.2)13.3 (5.9-20.6)77.1 (57.8-96.4)
Colon4.8 (0.9-8.8)20.3 (11.3-29.2)6.5 (1.4-11.6)23.9 (13.3-34.6)
Breast9.8 (4.4-15.3)21.5 (12.7-30.3)--
Prostate--17.3 (8.8-25.8)20.4 (10.0-30.7)

DISCUSSION

The pattern of cancer death among Native Americans is quite distinct from that of the general U.S. population. The pattern of cancer mortality that we have observed is consistent with data from the previous 5 years7 and with other reports in the literature.2, 3 The overall low rate is primarily determined by low rates of lung, breast, prostate, and colon carcinoma. The low lung carcinoma rate is easily explained by the rarity of habitual cigarette smoking among tribes in the Southwest,17, 18 but reasons for the other low rates are not as evident. Breast carcinoma rates may be influenced by early pregnancy, breast-feeding, and large families, but this is unlikely to explain all of the deficit. Risk factors for prostate carcinoma are not well enough understood to allow speculation regarding the reason for such low rates. Subsistence farming and hunting cultures may have a low rate of colon carcinoma because of the large amount of fiber in their diet. Changes in diet19 and increased smoking among teens20, 21 suggest that lung and colon carcinoma rates may soon match the general U.S. rate. Clearly, there is a need for research into the factors that have protected some groups of Native Americans from cancer.

Similar to the patterns observed in many developing nations, the Native American population has high rates of death from carcinoma of the cervix, liver, and stomach. We have evidence that aggressive Papanicolaou screening has reduced the rate of invasive cervical carcinoma in several areas served by Indian Health Service.22, 23 Liver carcinoma is closely related to high rates of hepatitis B infection24 and may be expected to fall as the national program of hepatitis B immunization matures. A growing body of knowledge suggests that chronic Helicobacter pylori infection may be related to the high rates of stomach carcinoma in Alaska Native communities. The epidemiology of H. pylori infection and its link to stomach carcinoma need further study before preventive measures can be recommended.

The contribution of genetic factors to cancer in the Native American population is not well understood. The remarkably high rate of gallbladder carcinoma could be an indication of a cancer with genetic determinants,25 although diet may also be a factor.

Could the observed low cancer mortality rates be artifacts of improper coding of race on death certificates? Although several authors have demonstrated that racial ascertainment is poor in some areas,26-28 cancer registries with adequate procedures for ascertaining race, such as the New Mexico Tumor Registry, consistently find low rates for most cancers in the Native American population.3 By including only counties with known high percentages of American Indian or Alaska Native people, and excluding areas that have known problems with racial reporting, we believe that this study is the most accurate to date. We recalculated mortality rates for all 12 Indian Health Service areas, including the ones with known problems of racial misclassification, and found a 5-15% decrease in all death rates. This consistent result supports the idea that the American Indian/Alaska Native racial category is underreported on death certificates in those areas.

Cancers in the "Unspecified" group were significantly higher among Native Americans. The reasons for this are not well understood. Autopsy rates are low among Native Americans, probably for both economic and cultural reasons, so definitive diagnosis may not be available in some cases. It may be that Native Americans refuse diagnostic procedures more often than others. Presentation at late stages may make it more difficult to assign a primary site. This topic certainly deserves more focused research.

Is the sample in this study representative of all Native American people? Possibly not; as already noted, this sample includes a more rural and less urban group than the entire Native population. This sample was selected because it most closely represents the population served by the Indian Health Service, and because we do not have reliable health data on the urban Indian population. A careful survey of cancer mortality among urban Indians would be a valuable adjunct to this study.

There is enormous geographic variation in cancer mortality rates,29 with the Southwest showing lower rates of colon, lung, and breast carcinoma. The variation in lung carcinoma mortality can be attributed to the different smoking habits between the regions. The differences in colon and breast carcinoma are not so easily explained. Further research will have to be done to elucidate whether some protective factors have been at work in the Southwest. Such research could have important implications for cancer prevention in all racial and ethnic groups.

CONCLUSIONS

This study has demonstrated that Native Americans experience a unique pattern of cancer mortality. Prevention and screening programs should be specifically designed for this population. The relatively lower burden of cancer in this population seems to be real, not an artifact of ascertainment. The observed differences in mortality present a unique opportunity for research into both risk factors and protective factors in this population.

Acknowledgements

The source data for this article are being prepared in a different format for an Indian Health Service chart book.

The opinions expressed are the authors' and do not reflect the official policy of the Indian Health Service.

Ancillary