 |
 |
|
|
The Health of American Indians and Latinos Sissi P. Foster Susan Applegate Krouse Statistical Brief No. 16 June 2003
The Julian Samora
Research Institute is a unit of the Colleges of Social Sciences and
Agriculture & Natural Resources at Michigan State University.
Author Bios Susan Applegate Krouse The Health of American Indians
and Latinos Abstract This paper summarizes a study that examined the availability and quality of local data on the health of American Indians and Latinos in the City of Lansing, Mich., and compared these populations to Whites and Blacks. Comprehensive databases on health (specifically morbidity and morality), environment, demographics, and spatial distribution were gathered and analyzed for these populations. Our study found that current local data is inadequate for proper population assessment, specifically in the areas of data quality, data scale, and data storage. Using the existing data with its limitations, we completed descriptions for these two populations and made suggestions for continued research. Introduction Nationally, American Indians and Latinos have higher rates of morbidity and mortality than do majority Whites. These health inequalities, along with national health concerns, have caused the federal government to highlight these populations in initiatives, such as Healthy People 2010 (www.health.gov/healthypeople), to reduce racial and ethnic disparities in health. Because the national health picture of these populations is bleak, we undertook this study to examine the local health status of American Indians and Latinos in Lansing, Michigan. Our goals were to create a comprehensive database of existing information, to compare these populations to Whites and Blacks, and to examine the quality of the existing data. This study examined the health and spatial distribution of American Indians and Latinos in Lansing and in the tri-county area of Clinton, Eaton, and Ingham Counties. Although our primary focus was on residents of the City of Lansing, we collected information on the larger tri-county area to allow for comparisons. This data included census demographics, health statistics and land use maps to assist us in making spatial connections between our populations and their environment. What we found was that the data regarding the health of these populations was inadequate in both the availability and the quality of the data found. We further attempted to present the best possible descriptions of American Indians and Latinos in Lansing, noting specific limitations and problems with the existing data. Our presentation of the data is both in statistical and in graphic form, to provide for more in-depth comparison and analysis. Since research has long confirmed a strong relationship between land uses and population health - whether beneficial or harmful - (Mutz, Bryner & Kenney, 2002), we added the spatial comparison section. Finally, we make some suggestions for future studies to improve the availability and quality of health statistics. A brief overview of our study populations as well as general demographic information follows to ascertain the population impacted and to understand their relationship to the majority population. The total population of the United States is 290,361,435. American Indians and Alaskan Native comprise less than 1% of the total U.S. population with 2,475,956 people. Latinos constitute a larger percentage and are the fastest growing population in the U.S. at 12.5% or 35,305,818 people. Whites make up 75.1% of the total with 211,460,626 people. Blacks represent 12.3% of the total with 34,658,190 people, and Asians are 3.6% of the total with 10,242,998 people. The population of Lansing, according to the 1990 Census, was 127,173. Our populations of concern, American Indians and Latinos, make up small percentages of that total. In 1990, American Indians numbered only 1,535 individuals, or just 1.2% of the total city population. Most of the Indian people in Lansing are Anishinabe, members of the three related tribes of Chippewa or Ojibwe, Odawa or Ottawa, and Potawatomi (“Native American and Hispanic Health Concerns in Ingham County, Michigan,” n.d.:6). Latinos contributed a larger percentage, 7.9%, with 10,061 individuals. Most Latinos in Lansing are of Mexican origin (Siles & Rochin, 1998). Data Sources In order to best examine the quality of local data for our study populations, we sought to identify all possible sources. What follows are our successes and failures in the data gathering process. We began by contacting and gathering data from a variety of governmental agencies at the federal, state, and municipal levels. Many federal statistics are available on-line, and we made use of census information from both the 1990 census and, as data became available, from the 2000 census. The Indian Health Service (IHS), the primary health provider for approximately half of the Native American population in the United States, does not offer services in the Lansing area; however, a study conducted for the IHS did give some insights into the needs of the local urban Indian community (National Council of Urban Indian Health 2001). A complete list of sources we consulted and data we gathered are included in Appendix 1. At the state and municipal levels, we benefited from cooperative agreements maintained by the State of Michigan Department of Community Health and the Ingham County Health Department. We were able to access information about the three counties, Clinton, Eaton, and Ingham, that had already been synthesized by the state for the Ingham County Health Department. This data included live birth and death files for most of the 1990s, and files on communicable diseases. Some information is also summarized on the Capital Area Community Voices website (www.cacvoices.org), an outreach effort by the City of Lansing and the Ingham County Health Department. Specific information on demographic and environmental variables is maintained by a variety of municipal offices, both city and county. We accessed data ranging from land use to crimes in our effort to collect and synthesize information affecting health. We had hoped that local community organizations would be an additional source of information on American Indian and Latino health; however, we found that no local groups had attempted a systematic survey of health needs in their particular communities. The Mestizo – Anishinabe Health Project, connected to the Community Voices project and funded by the W. K. Kellogg Foundation, recently (2001) attempted a health survey in the American Indian and Latino communities in Lansing. This resulted in a booklet, Native American and Hispanic Health Concerns in Ingham County, Michigan (n.d.), with some general health statistics. Published information on American Indian and Latino health in Michigan is available only at the state level. Nothing has been published that provides information at the city or county level. Nan E. Johnson’s 1995 Health Profiles of Michigan Populations of Color is a good example of a state level synthesis, utilizing 1990 census data to provide a picture of minority health in Michigan. It does not, however, address more local communities or issues. A survey of a more limited population, Elizabeth Chapleski’s State of Michigan 1990 Survey of Native American Elders, again provides information on a state level, and does incorporate urban Indians in its survey population. Lansing was not among the urban locations included in this study. The Julian Samora Research Institute at Michigan State University conducts research on Latinos in the Midwest. JSRI lists 167 research reports, working papers, statistical briefs, and occasional papers on its website (www.jsri.msu.edu); two of those deal specifically with Latinos in Lansing. Lisa M. Topoleski (1997) looked at hospice underutilization by Mexican-Americans in Lansing, and Marcelo E. Siles and Refugio I. Rochin (1998) profiled the Latino community in North Lansing. Background to Data Problems As our study focused on the examination of existing data, we first looked at a number of studies that utilized this type of data (i.e., census records and vital statistics) to see what they concluded about the reliability of existing data. Analysis of census data and vital statistics for the investigation of racial and ethnic health is common in morbidity and mortality studies. Problems with data such as census records, state vital records, or other health statistics are widely documented. Two studies in particular summarize the issues involved in using statistical data for racial and ethnic groups. Morbidity statistics may come from data collected by county and state health departments, from the federal Centers for Disease Control, or from surveys such as the National Health Interview Survey. One study by Parker, Davis, Wingo, Ries, and Heath (1998) discusses the problems associated with using the racial and ethnic categories suggested by two specific surveys. These categories are African Americans, Asians and Pacific Islanders, American Indians, Hispanics, and Whites. Parker, et al. note that there are differences within these five populations, as well as between. For American Indians, the authors point out that this population represents 500 tribes, “each with unique cultural, genetic, and sociodemographic characteristics” and that “it is likely that cancer incidence rates vary considerably among tribes” (1998:43). The Hispanic category, the authors indicate, includes persons from several countries and “every racial group” (1998:45). In addition to these basic problems of categorization, Parker, et al. detail several other limitations of the available data, including the possible mis-coding of those racial and ethnic categories due to ambiguous definitions, generalizing from too specific data, and limited surveying among some populations, particularly American Indians (1998:47). Mortality statistics present additional problems.
Rosenberg, Maurer, Sorlie, Johnson, MacDorman, Hoyert, Spitler, and Scott (1999)
report on the quality and
reliability of death rates, by race and Hispanic origin, in mortality statistics
produced by the National Center for Health Statistics and other agencies, including
the U.S. census. While the reporting of race for Whites and Blacks on death
certificates is deemed “highly reliable” (Rosenberg, et al. 1999:8), for other
racial and ethnic groups the rate of misclassification on death certificates
results in significant undercounting. For American Indians, the rate of undercounting
is 20.6 %, for Asians and Pacific Islanders, it is 10.7 %, and for Hispanics,
1.6 %. In addition to the problems already found in existing data, the authors
point out different data disparities that will result from new racial and ethnic
categorizations used in the 2000 census. Population Morbidity and Mortality Morbidity statistics proved difficult to obtain and to analyze. We were not able to obtain information on sexually transmitted diseases (including HIV and AIDS) due to confidentiality considerations and, in addition, the communicable disease data did not include statistics on chicken pox or tuberculosis. We reviewed information on some communicable diseases (Figure 1), but it was available only at the tri-county level (Clinton, Eaton, and Ingham Counties). As our data were so limited we were not able to make comparisons at the national level. Figure 1. 1990 census block group boundaries for Lansing and corresponding distributions of major populations (U.S. Census Bureau 1990 - spatial data from Michigan DNR Spatial Data Library and MSU Center for Remote Sensing). The quality of the communicable disease data for our purposes was poor, with 58% of cases reporting “unknown race or ethnicity.” Only 0.1% of cases were attributed to American Indians (a total of four cases over a 10-year period). With a population of 1,535, Indians in Lansing are either extremely healthy, or their ethnicity is not being reported in the communicable disease statistics. We were able to obtain and to analyze more data at both local and national levels on mortality. Figure 2 graphs American Indian causes of death, comparing the local populations to all American Indians nationally. For heart disease, cancer, cerebrovascular disease, and COPD, American Indians in Lansing and the tri-county area have higher rates than do Indians nationally. Considering the high rates for all Indians, these are very real areas of concern.
Figure 2. Communicable Diseases in Tri-County area by ethnicity (Michigan Department of Community Health, Division of Communicable Diseases and Immunization) Latino causes of death are indicated in Figure 3. The trends are more complex than for Indians, but Latinos in the tri-county area have a comparatively high rate of cancer and Latinos in Lansing have high rates of diabetes and cerebrovascular disease.
Locally, mortality statistics indicate some disheartening numbers (Figures 4 and 5). American Indians have the highest percentage of deaths of any population due to cancer and cerebrovascular disease, while Latinos have the highest percentage of diabetes and accidents. Mortality data was only available for the years 1995 through 1998.
The actual numbers, however, may be worse than these figures suggest, due to two factors. The first is the very real possibility of underreporting of American Indian and Latino ethnicity on death certificates, which are often completed by physicians or funeral directors unfamiliar with the individual or the family. The second is that statistics on leading causes of death do not provide information on the accompanying conditions or secondary causes of death that may have contributed to an individual’s death; for example, death may be due to a heart attack, but the attack was precipitated by diabetes. Information on age at death (Figures 6 and 7) indicates that for both American Indians and Latinos in Lansing (shown as Urban) there is a spike in death rates between ages 50 and 60. For Latinos in the surrounding tri-county area (shown as Rural), the spike does not appear until ages 60 to 70.
Figures 8 through 10 present age-specific death rates by race and ethnicity for Ingham County. The high death rate for American Indians at every age is statistically problematic, again due to the very low total numbers for this population. For example, in 1990 (U.S. Census Bureau) the male American Indian population ages 0 to 4 was 91. In order to calculate death rates per 1,000 population for American Indians, we expanded the numbers. This results in a graphical and statistical distortion of the number of deaths. For other populations, the numbers are condensed, creating a problem with data scale, resulting in invalid population comparisons.
Figure 8. White and Black population distributions (1990 census block group compared to 2000 census tracts Michigan DNR Spatial Data Library).
Figure 9. City parks in relation to American Indian, Latino, and White population distribution (U.S. Census Bureau 1990- spatial data from Michigan DNR Spatial Data Library).
Figure 10. Industrial sites (health, schools, government) in relation to American Indian, Latino, and White population distribution (U.S. Census Bureau 1990-spatial data from Michigan DNR Spatial Data Library). We had hoped to be able to do inferential statistical analysis of both morbidity and mortality data for American Indians and Latinos in Lansing. The very low numbers of cases of communicable diseases, particularly for American Indians, and the non-availability of data on sexually transmitted diseases, have limited our ability to do more than descriptive statistics for morbidity. For mortality, we are able to look at some patterns, and make some comparisons between our two populations in Lansing, in the surrounding tri-county area, and nationally. However, these figures must be used with caution, due to the very small total number of American Indians, in particular. Spatial Comparisons Figure 11 demarcates the census blocks groups used in 1990 within the boundaries of the City of Lansing with major highways, roads, and rivers shown for orientation. Figure 12 provides a look at the population distribution of American Indians, Latinos, Whites, and Blacks in the City of Lansing, by census block groups. We chose to use census blocks as our unit of analysis, as privacy considerations restrict more specific information being made available from the Census Bureau. All data displayed in the graphics were classified using the Jenk’s optimization method of natural breaks. This method identifies breakpoints in the data by placing it into categories that minimize the sum of the variance within each group. Jenk’s method finds groupings and patterns inherent in the data (www.dartmouth.edu/~dbkarnes/jenks/jenks.html).
Figure 11. Neighborhood business in relation to American Indian, Latino, and White population distribution (U.S. Census Bureau 1990- spatial data from Michigan DNR Spatial Data Library). Figure 12. (inset) Major transportation routes (highways, freeways, railways) in relation to American Indian, Latino, and White population distributions (U.S. Census Bureau 1990 - spatial data from MI DNR Spatial Data Library). Major highways, roads, and rivers are shown for orientation (U.S. Census Bureau 1990 – spatial data from MI DNR Spatial Data Library and MSU Center for Remote Sensing). Once we had gathered all the data we were able to find, we began to analyze it. We combined different data sets relating to population trends and distribution, and also to factors affecting the health of American Indians and Latinos in Lansing. We hope that by mixing these various data sets and showing them spatially, we might be able to understand the relationship of these populations to the local environment. Previous studies have shown that environmental injustice is prevalent in minority populations and have gone so far as to indicate that race alone is the single most significant predictor of locating hazardous waste sites (Mutz, Bryner, & Kenney, 2000). Figures 13 and 14 illustrate population trends for American Indians, Latinos, Whites, and Blacks in Lansing for 1990 and 2000. Declines in population can be seen for American Indians and Whites, while increases in population show up in both the Latino and the Black populations. American Indians, Latinos and Black appear to concentrate in the urban core, while Whites are more evenly distributed throughout the city. Data for 2000 was not available at the census block level during this study, but only at the census tract level. This may create some misrepresentations of population densities in these figures. For example, population concentrations that are apparent in 1990 in census blocks appear less concentrated in 2000, possibly due to the larger size of the census tracts.
Figure 13. Industrial areas in relation to American Indian, Latino, and White population distributions (U.S. Census Bureau 1990 - spatial data from MI DNR Spatial Data Library).
Figure 14. Polluted sites (hazardous waste, solid waste, and environmental contamination) in relation to American Indian, Latino, and White population distributions (U.S. Census Bureau 1990 - spatial data from Michigan DNR Spatial Data Library). In assessing the impact of the environment on health, we combined population densities with specific land uses. We looked first at the distribution of American Indians and Latinos in relation to what we grouped as factors having a positive impact on health and urban life, specifically parks, institutions (health, educational, religious, and governmental facilities), and neighborhood businesses (non-industrial). Figures 15, 16, and 17 show population distributions in relation to buffers of one-quarter mile surrounding these positive factors. These figures indicate that Whites enjoy greater proximity to all these factors, with the exception of neighborhood businesses, where Latinos live in greater densities.
Figure 15. Population densities within _ of a mile buffer surrounding industry and polluted sites.
Figure 16. Drug-related crimes in relation to American Indian, Latino, and White population distributions (City of Lansing Police Department - spatial data from Michigan DNR Spatial Data Library).
Figure 17. Weapon-related crimes in relation to American Indian, Latino, and White population distributions (City of Lansing Police Department - spatial data from Michigan DNR Spatial Data Library). More negative aspects of the urban setting include living near industries and polluted sites. Figures 18 and 19 map these sites and the distribution of American Indians and Latinos in relation to them. Figure 18. Murder in relation to American Indian, Latino, and White population distributions (City of Lansing Police Department - spatial data from Michigan DNR Spatial Data Library).
Figure 19. Assault in relation to American Indian, Latino, and White population distributions (City of Lansing Police Department - spatial data from Michigan DNR Spatial Data Library). Figure 20 provides a table comparing census block areas with population numbers of American Indians, Latinos, and Whites who live within one-quarter mile buffers of industrial sites or polluted sites (hazardous waste management zones, solid waste management sites, and sites of environmental contamination). While this spatial analysis does not indicate statistical significance and uses population counts rather than densities, it clearly shows a pattern. Latinos are more concentrated around industrial and polluted sites than Whites, and both the Latino and White populations are more concentrated around these sites than the American Indian population. The Indian population may be too small to establish a trend. Further analysis is needed to establish the statistical significance of this data.
Figure 20. American Indian leading causes of death. Tri-County area excludes Lansing (County Vital Statistics 1995-1998. National Vital Statistics Report 2001). In order to examine the relation between population and land use, we ran a regression analysis for American Indians and Latinos. We used 100 meter buffers from identified environmental factors and population densities surrounding these factors. Table 1 provides the results. Although these preliminary numbers do not indicate specific relationships, they do indicate that there is an overall significance in the relationship between population distributions and environmental factors. The lower the probability (‘P’), the higher the significance of the relationship between the population and the land use.
Because American Indians and Latinos live closer to the inner city, they are also more heavily impacted by crime, which tends to cluster toward the center of town. Drug-related crimes (Figure 21) and weapons-related crimes (Figure 22) seem to impact American Indian and Latino population concentrations more than White population, while assaults (Figure 23) are distributed more evenly across the entire Lansing area. Murders are not included here for analysis due to their relatively rare occurrence in Lansing.
Figure 21. Latino leading causes of death. Tri-County area excludes Lansing (County Vital Statistics 1995-1998. National Vital Statistics Report 2001).
Figure 22. Educational attainment at death by ethnicity (County vital statistics, 1995-1998).
Figure 23. Average age-specific death rate for ages 0-24 (County vital stats, 1995-1998, U.S. Census Bureau 2000). Figure 24 explores the question of environmental justice for these populations. We graphed the percentage of American Indians and Latinos who died from heart disease and cancer and who were also located within one-quarter mile buffer of industrial and polluted sites, and of parks and institutions (schools, hospitals, government buildings). Near industrial and polluted sites, heart disease is far more prevalent for Latinos and cancer for American Indians. While this increase may not be attributable to industry or pollution, it does raise questions and calls for further study, including analysis for statistical significance. Near parks and institutions, the differences in rates for the two populations are not nearly so dramatic. The small population numbers, especially for American Indians, may also impact these potential relationships and should be taken into account.
Figure 24. Major diseases in industrial areas, polluted areas, parks, and institutional areas. Discussion American Indians and Latinos represent small portions of the total Lansing population, and data on their health is not always accurate. For morbidity, the data are so inadequate, we cannot make good predictions about the health status of these two populations. For mortality, the data are more complete and we can see that they are disproportionally affected by cancer, cerebrovasuclar heart disease, diabetes, and accidents. It is clear from our collection of existing data that there are numerous problems with the data and its utility for making predictions concerning the health status of American Indians and Latinos. The problems can be broken into three categories: data shortages, data scale, and data quality. Data Shortages For many kinds of information, the data are simply not available, or have such restricted access that we did not attempt to incorporate them into our database. Some of the 2000 census statistics are not yet available on health, and are not expected to be released until 2003. This limited our ability to make comparisons over time, and particularly limited our ability to discuss current health status. Local community organizations have not collected health statistics, nor conducted community surveys on the health of their members. We had hoped that more locally controlled data would provide a counterpoint to state vital records and federal census data, and allow us to evaluate the quality of existing data. We did not attempt to access hospitalization data, as we were advised this would be a challenge, due to patient confidentiality concerns. Time limitations also precluded a lengthy attempt to access this data. Concerns with patient / client confidentiality also limited access to specifics on sexually transmitted diseases as well as mental health. Data Scale Statistics concerning the health of minority populations are collected by numerous agencies. Our first problem with data scale came with trying to find data specific to the City of Lansing, or the tri-counties of Clinton, Eaton, and Ingham. Census data and much of the health data is collected and analyzed at the state and federal levels. While the local counties collect data, much of it becomes synthesized within the State of Michigan’s Department of Community Health. Our second problem with data scale was a result of the populations we had chosen to study. Because these two populations, particularly American Indians, are such small portions of the total population of Lansing, making statistical inferences based on these numbers is not always valid. Compounded with the problem of misreporting or underreporting of American Indians and Latinos in vital statistics, the data become even more problematic. Qualitative studies would be one way to mitigate the problems with low numbers of individuals. Studies at a larger scale, such as the state level, would include a much larger number of people and be another way that inferences could then be made to specific city or county populations. Data Quality The quality of the data is also affected by data scale, and particularly by misreporting and underreporting. Studies at the national level (Parker et al. 1998; Rosenberg et al. 1999) suggest that the reliability and veracity of existing data on morbidity and mortality of American Indians and Latinos is questionable; our research at the local level confirmed these problems. Existing data on communicable diseases over a 10-year period indicated that only four American Indians out of a population of more than 1,500 were impacted by these diseases. This represents a gross underreporting, and a consequent concern with that data set. Our objectives in this project were to gather and examine data on health of American Indians and Latinos in the City of Lansing, and to compare these populations to the larger populations. While we found numerous problems in the existing data, we did produce a number of statistical and graphical figures that attempt to describe our specific populations, in relation to other populations in Lansing, and to their natural and social environment. This provides a base for future research that will increase the quantity and quality of data concerning minority health. Future research should address the concerns we identified with existing data. There is a need for more locally specific studies of American Indians and Latinos in Lansing, perhaps conducted in cooperation with local community organizations. Future studies should include more comprehensive health data, including information on sexually transmitted diseases and on hospitalizations. Data reporting needs more rigorous controls and training for those persons responsible for reporting, particularly more awareness of specific populations and the problems with undercounting. In particular, qualitative studies would provide actual individual and community data to complement the existing aggregate data on American Indians and Latinos in Lansing. Acknowledgements Our thanks to the College of Social Sciences, the Michigan Agricultural Experiment Station, and the Julian Samora Research Institute for funding this project, and for the opportunity we had to work together as a team and to grow as individual researchers. In addition, we would like to thank Jon Burley, from the Landscape Architecture Program and George Sirbu, from the Statistical Consulting Services at MSU, for their help in the regression analysis found in this report. And finally, we could not have completed this project without the cheerful and highly competent help of our graduate research assistants, Robert Cook and Mira Hidajat. References Chapleski, Elizabeth. 1990. State of Michigan 1990 Survey of Native American Elders. Michigan State University Office of Services to the Aging. Johnson, Nan E. 1995. Health Profiles of Michigan Populations of Color. Michigan Department of Public Health. Mutz, K.M., Bryner, G.C. and D.S. Kenney (eds). 2002. Justice and Natural Resources: Concepts, Strategies, and Applications. Island Press: Washington, D.C. National Council of Urban Indian Health. 2001. A Feasibility Study of Three New Urban Indian Health Sites: In Fayetteville, Lansing and Philadelphia. Prepared by Kauffman and Associates, Inc., Spokane, WA, Rockville, MD.: U.S. Indian Health Service. Native American and Hispanic Health Concerns in Ingham County, Michigan. n.d. Lansing, MI.: Mestizo Anishnabe Health Alliance. Parker, S.L., K.J. Davis, P.A. Wingo, L.A. Ries and C.W. Heath. 1998. “Cancer Statistics by Race and Ethnicity.” CA – A Cancer Journal for Clinicians. 48,1:31-48. Rosenberg, Harry M., Jeffrey D. Maurer, Paul D. Sorlie, Norman J. Johnson, Marian F. MacDorman, Donna L. Hoyert, James F. Spitler and Chester Scott. 1999. Quality of Death Rates by Race and Hispanic Origin: A Summary of Current Research. National Center for Health Statistics. Vital and Health Statistics. 2,128:1-13. Siles, Marcelo E. and Refugio I. Rochin. 1998. “North Lansing, Michigan: Profile of an Inner City.” JSRI CIFRAS Statistical Brief No. 10, the Julian Samora Research Institute, Michigan State University, East Lansing, MI. Topoleski, Lisa M. 1997. “An Interpretive Analysis of Hospice Underutilization by Mexican-Americans in Lansing, Michigan: En Sus Propias Palabras (In Their Own Words).” JSRI Research Report No. 28, the Julian Samora Research Institute, Michigan State University, East Lansing, MI.
|
|
|
|
For more
information, contact: |
Figure 3.
Leading causes of death for American Indians of all ages compared to
Whites and
Blacks in Lansing (County vital statistics, 1995-1998).
Figure 4.
Leading causes of death for Latinos of all ages compared to Whites and
Blacks
in Lansing (County vital statistics, 1995-1998).
Figure
5. American Indian /Age at death as compared with Whites and Blacks
(County vital statistics, 1995-1998).
Figure 6.
Latino /Age at death as compared to Whites and Blacks (County vital statistics,
1995-1998. 
Figure 7.
American Indian and Latino population distributions (1990 census block
groups compared to 2000 census tracts - Michigan DNR Spatial Data Library).
