Appendix B: Methodological and Technical Notes

Mortality Data

County Definitions

We used Federal Information Processing Standard (FIPS)¹ codes to link county definitions across multiple data sets in this atlas. To ensure accurate linking of counties across the data sets, the following modifications were made:

Independent Cities
The following independent cities were retained in the geographic database as discrete entities separate from adjacent counties.
 

Alaska

Arizona

Hawaii

Virginia

Virginia has 34 independent cities. We used the 1996 Area Resource File database² to incorporate data from these cities into their adjacent counties, which is standard practice.

Yellowstone National Park

Data Sources

Heart Disease and Stroke Mortality Data
We obtained death certificate data through the National Center for Health Statistics’ National Vital Statistics System, which is a compilation of statistics from all death certificates filed in the 50 states and the District of Columbia.³ Heart disease deaths were defined as those for which the underlying cause of death listed on the death certificate was diseases of the heart, defined according to the International Classification of Diseases (ICD-9 codes 390–398, 402, and 404–429; ICD-10 codes I00–I09, I11, I13, I20–I51).^4,5 Stroke deaths were defined as those for which the underlying cause of death listed on the death certificate was cerebrovascular disease (ICD–9–CM codes 430–438).⁴ For each decedent, underlying cause of death, age, race, ethnicity, gender, and county of residence at the time of death were abstracted from computerized death certificate files.

Population Data
For heart disease mortality rates during 1996–2000, we used postcensal population estimates for 1996–1999 and a special "bridged–race" version of the 2000 census population estimates that allowed us to aggregate the data across 1996–2000. CDC's National Center for Health Statistics has produced bridged-race versions of 2000 census data to allow comparisons between these data and earlier reports, which used fewer race/ethnicity categories (see the Definition of American Indians and Alaska Natives section on pages 64–65 of this appendix for a discussion of race/ethnicity categories used for federal data collection).⁶ For stroke mortality rates during 1991–1998, we used postcensal estimates calculated by the U.S. Bureau of the Census through extrapolation of linear trends in population growth and intercounty migration patterns between the 1980 and 1990 censuses.

Map Projections

We used several different map projections to produce the county-level maps in this publication. For the contiguous United States, an Albers Conic Equal Area projection was used. For Alaska, the Miller's Cylindrical projection was used. For the Hawaii map, we used geographic coordinates (latitude and longitude). Neither Alaska nor Hawaii is in proper geographic scale relative to the continental United States on the national maps. The combination of different projections and scales allowed for presentation of a relatively familiar orientation of these geographic features.

The coordinate information for the contiguous 48 states was projected using the Albers Conic Equal Area projection with the following parameters:

The coordinate information for Alaska used the Miller's Cylindrical projection with the following parameters:

Definition of American Indians and Alaska Natives

The definition for American Indian and Alaska Native (AI/AN) people used in this publication is based on the definition established in 1977 by Directive 15 of the Office of Management and Budget (OMB), which is the federal agency that defines standards for government publications.⁷ The categories are not based on biological or anthropological concepts. OMB developed categories for racial and ethnic groups in response to the need for standardized data for record keeping and data collection and presentation by federal agencies (e.g., to conduct federal surveys, collect decennial census data, and monitor civil rights laws).

In 1997, OMB issued new race and ethnicity categories following criticism that the categories did not reflect the country's increasing diversity. All federal agencies were instructed to begin collecting and analyzing data using the new categories no later than January 1, 2003. However, the census and vital statistics data used in this publication were collected before the 1997 directive was implemented. Consequently, the racial and ethnic categories analyzed here are consistent with the 1977 directive.

The 1977 definition for American Indian or Alaska Native is as follows: A person having origins in any of the original peoples of North America and who maintains tribal affiliation or community attachment.

Spatial Geometry

The geographic database used for the county–level maps in this publication came from the Environmental Systems Research Institute's (ESRI) ArcUSA database, which includes spatial geometry and characteristics of all U.S. counties. ESRI modified the 1973 Digital Line Graph source data produced by the U.S. Geological Survey to update county boundaries through 1988. The geographic scale of the spatial geometry (i.e., linework) used is 1:2 million, which is sufficient to identify major county features. Mortality and population data were linked to county geography using FIPS codes.

Calculation of Spatially Smoothed and Age–Adjusted Death Rates

Rationale for Spatial Smoothing
Although county death rates provide a high degree of spatial specificity, rates in counties with small populations and few heart disease or stroke deaths can be unstable. This problem is particularly relevant when examining geographic disparities among AI/AN populations because many counties have small or nonexistent numbers of this population. We used two approaches to reduce the statistical instability of county death rates for heart disease and stroke: 1) temporal aggregation of the data (1996–2000 for heart disease, 1991–1998 for stroke) and 2) application of a statistical procedure known as spatial smoothing.

We chose to spatially smooth heart disease and stroke death rates using a spatial moving average. The number of deaths (numerators) and population counts (person–year denominators) for each county were combined with the deaths and population counts of the immediate neighboring counties (i.e., contiguous counties), and then divided to produce an average rate. We used the contiguity matrix for all U.S. counties from the 1996 Area Resource File database to identify contiguous counties and to perform spatial smoothing. Thus, a single county's heart disease or stroke mortality rate actually represents an average of the rates of that county and all of its contiguous neighbors.

Calculation of Death Rates
Spatially smoothed and age–adjusted death rates were calculated at the county level for all AI/AN people and again for AI/AN women and men separately. Heart disease and stroke deaths were obtained from the National Vital Statistics System and included all deaths for which the underlying cause of death reported on the death certificates was diseases of the heart (ICD-9-CM codes 390–398, 402, or 404–429: ICD-10 codes I00–I09, I11, I13, or I20–I51) or cerebrovascular disease (ICD-9-CM codes 430–438).^4,5 Population data were obtained from the U.S. Bureau of the Census.

For each county, deaths (numerators) and population counts (denominators) for 10–year age groups (i.e., ages 35–44, 45–54, 55–64, 65–74, 75–84, and ≥85 years) were summed across the years. County numerators and denominators were then combined with numerators and denominators of all neighboring counties. Neighboring counties were defined solely by contiguity (as opposed to distance). The combined numerators were divided by the combined denominators to produce spatially smoothed, age–specific (i.e., by 10–year age group) death rates. These spatially smoothed rates were then directly age–adjusted to the 2000 U.S. population for 10–year age groups starting at 35. These calculations were repeated separately by gender.

Two constraints were applied to the calculation of county death rates. A stroke death rate was not calculated for any county for which the total number of stroke deaths in that county plus its neighbors was fewer than 20 during 1991–1998.8 A heart disease death rate was not calculated for any county for which the total number of heart disease deaths in that county plus its neighbors was fewer than 20 during 1996–2000. To avoid calculating rates for counties that had no AI/AN population but whose neighboring counties had significant populations, rates were calculated only for counties with a population count of 5 or more (i.e., person–years were ≥5) during 1996–2000 for heart disease and 1991–1998 for stroke.

Unfortunately, death rates could not be adjusted to account for misreporting of AI/AN people as "white" on death certificates (see the Introduction for a discussion of this issue). Although the Indian Health Service (IHS) has established a series of weights that can be used to estimate more accurate death rates for AI/AN populations, these weights are designed to be applied to IHS areas, not U.S. counties.⁹ Because the weights were calculated on the basis of deaths from all causes combined, even the adjusted heart disease and stroke death rates for AI/AN people may still be less than the true rates for this population.¹⁰

Standard Population Weights
Because we calculated directly age–adjusted heart disease and stroke death rates for people ages 35 years and older, but not for the entire age range of the population, we had to recalculate the standard weights for the 2000 U.S. standard population. New weights for age groups 35–44 through ≥85 years were calculated using a two–step procedure. First, we calculated the sum of the original 2000 standard weights for 10year age groups 35–44 through ≥85 years. Second, for each age group, we divided the original weight by the sum of the weights for ages ≥35 years. The resulting quotients are the new standard population weights. The weights were rounded to two decimal places and used to calculate directly age–adjusted death rates for people ages >35 years.

Contiguity Matrix for Alaska
We used the contiguity matrix for all U.S. counties from the 1996 Area Resource File database to perform spatial smoothing of heart disease and stroke mortality rates for this publication. However, this database did not include information for counties in Alaska, because Alaska was considered to be a single geographic unit. Because we are interested in the geographic patterns of heart disease and stroke mortality within the state, we created the following contiguity matrix for the counties in Alaska:

* Each county can be bordered by as few as one or as many as eight neighboring counties.

Data Source

We obtained data for eight important risk factors for heart disease and stroke from the Behavioral Risk Factor Surveillance System (BRFSS). BRFSS data are collected monthly by state departments of health through telephone interviews of noninstitutionalized adults aged 18 years or older. The states use a multistage design for stratified random sampling of the telephone numbers dialed. Complete details of the BRFSS methodology have been published elsewhere.^11–13

The BRFSS includes a set of core questions that are asked every year in all states, as well as a set of rotating core questions that are asked every other year. This publication presents prevalence data for the following risk factors included in the annual core questions: diabetes, cigarette smoking, obesity, physical inactivity, and poor health. From the rotating questions that are asked in odd–numbered years, it presents data on high blood pressure, high blood cholesterol, and cholesterol screening.

BRFSS core questions are available in English and Spanish. If the interviewer determines that the respondent is not proficient in either language, the interviewer does not administer the survey and notes that the interview was ended because of a language barrier.

Once the monthly state data are collected, they are sent to CDC to be edited and checked for accuracy. CDC staff members aggregate the monthly data files for each state to create annual totals. These totals are then weighted according to the respondents’ probability of being sampled, given the race, age, and gender of the population from which they were selected. Weighting is based on the most current census data for each state. The prevalence of each risk factor for each state is calculated from the weighted data.

Because of the small number of AI/AN respondents in the BRFSS, we combined data for 2001–2003 to increase the precision of our estimates. Prevalence estimates for states that reported fewer than 50 AI/AN respondents were considered unreliable and are not presented in this publication.¹⁴

Telephone Coverage

A recent study indicates that about 17% of AI/AN people do not have telephones in their homes.¹⁵ This percentage is higher than that of any other U.S. racial/ethnic group. The percentages within this population vary sharply depending on where people live; only 47% of AI/AN people living on reservations have telephones compared with 75% of those who live in rural areas and 88% of those who live in urban areas.^15,16

Other studies have found that AI/AN people who live in households without telephones are more likely to be physically inactive and to smoke cigarettes.^17–19 Therefore, the findings reported in this atlas are more likely to represent AI/AN people who live in urban areas and not on reservations, and they likely underestimate the prevalence of some risk factors for heart disease and stroke.

Definition of Risk Factors

For this publication, we defined eight risk factors for heart disease and stroke on the basis of specific questions from the BRFSS during 2001– 2003. As of 1996, state health departments also can ask about regular aspirin use, prior history of heart disease, and prior history of stroke on their BRFSS questionaires. However, only a few states do so, and many of these states do not have large enough AI/AN populations to generate stable estimates. Therefore, data for these heart disease and stroke risk factors are not included in this atlas.

Map Projection

We combined two map projections to produce the risk factor maps in this publication. For the contiguous United States, an Albers Conic Equal Area projection was used. For Alaska, the Miller's Cylindrical projection was used. Neither Alaska nor Hawaii is in proper geographic scale relative to the continental United States on the risk factor maps. The combination of different projections and scales allowed for presentation of a relatively familiar orientation of these geographic features.

Definition of American Indians and Alaska Natives

Respondents to the BRFSS were asked to select a race of origin from the following list: White, Black or African American, Asian, Native Hawaiian or Other Pacific Islander, American Indian/Alaska Native, or Other (Specify). Only those respondents selecting American Indian/Alaska Native were included in this atlas.

Spatial Geometry

The geographic database used for the risk factor maps in this publication came from the Environmental Systems Research Institute's (ESRI) ArcUSA database, which includes spatial geometry and characteristics of all U.S. counties. The geographic scale of the spatial geometry used is 1:42,874,983, which is sufficient to identify state features.

Calculation of Prevalence Estimates

Because of the complex survey methodology used to produce prevalence estimates for this publication, we used SUDAAN statistical software to calculate standard errors and 95% confidence intervals. The prevalences reported in this atlas are weighted according to the respondents' probability of being sampled, given the race, age, and gender of the state population from which they were selected. No statistical tests were performed for comparison, so the findings of this publication should be considered descriptive in nature.

References

1. National Institute of Standards and Technology. Federal Information Processing Standards Publication 55–3: Codes for Named Populated Places, Primary County Divisions, and Other Locational Entities of the United States, Puerto Rico, and the Outlying Areas. Gaithersburg, MD: U.S. Department of Commerce; 1994. Available at http://www.itl.nist.gov/fipspubs/fip55-3.htm.
2. Bureau of Health Professions. Area Resource File. Rockville, MD: U.S. Department of Health and Human Services, Health Resources and Services Administration; 1996. Available at http://www.arfsys.com.
3. CDC. National Vital Statistics System Web site. Available at http://www.cdc.gov/nchs/nvss.htm.
4. U.S. Department of Health and Human Services. The International Classification of Diseases, 9th Revision, Clinical Modification. Washington, DC: Public Health Service, Health Care Financing Administration; 1980. HHS publication no. (PHS) 80–1260.
5. World Health Organization. International Classification of Diseases and Related Health Problems, 10th Revision, Clinical Modification. Geneva: World Health Organization; 1992.
6. CDC. National Center for Health Statistics Web site. U.S. Census Populations with Bridged Race Categories: Bridged–Race Population Estimates for April 1, 2000. Available at http://www.cdc.gov/nchs/about/major/dvs/popbridge/popbridge.htm.
7. Wallman KK, Hodgdon J. Race and ethnic standards for federal statistics and administrative reporting. Statistical Reporter 1977;77(10):450–454.
8. Hoyert DL, Arias E, Smith BL, Murphy SL, Kochanek KD. Deaths: final data for 1999. National Vital Statistics Reports 2001;49(8):110.
9. Indian Health Service. Adjusting for Miscoding of Indian Race on State Death Certificates. Rockville, MD: U.S. Department of Health and Human Services, Public Health Service; 1996.
10. Rhoades DA. Racial misclassification and disparities in cardiovascular disease among American Indians and Alaska Natives. Circulation 2005;111(10):1250–6.
11. CDC. Technical Information and Data: BRFSS User's Guide. Atlanta: U.S. Department of Health and Human Services; 1998. Available at http://www.cdc.gov/brfss/pubs/index.htm.
12. Holtzman D. Analysis and interpretation of data from the U.S. Behavioral Risk Factor Surveillance System (BRFSS). In: McQueen DV, Puska P, et al. Global Behavioral Risk Factor Surveillance. New York: Kluwer Academic/Plenum Publishers; 2003:35–46.
13. Mokdad AH, Stroup DF, Giles WH. Public health surveillance for behavioral risk factors in a changing environment: recommendations from the Behavioral Risk Factor Surveillance Team. Morbidity and Mortality Weekly Report 2003;52(RR–9).
14. CDC. 2000 BRFSS Summary Prevalence Report. Atlanta: U.S. Department of Health and Human Services; May 2001:9. Available at http://www.cdc.gov/brfss/pdf/2000prvrpt.pdf.(PDF 1832K)
15. National Telecommunications and Information Administration. Falling Through the Net: Defining the Digital Divide. Washington, DC: U.S. Department of Commerce; 1999.
16. Bureau of the Census. Housing of American Indians on reservations—equipment and fuels. Bureau of the Census Statistical Brief 1995. Publication no. SB/95–11.
17. Pearson D, Cheadle A, Wagner E, Tonsberg R, Psaty BM. Differences in sociodemographic, health status, and lifestyle characteristics among American Indians by telephone coverage. Preventive Medicine 1994;23(4):461–4.
18. Cheadle A, Pearson D, Wagner E, Psaty BM, Diehr P, Koepsell T. Relationship between socioeconomic status, health status, and lifestyle practices of American Indians: evidence from a Plains reservation population. Public Health Reports 1994;109(3):405–13.
19. Peterson DE, Remington PL, Kuykendall MA, Kanarek MS, Diedrich JM, Anderson HA. Behavioral risk factors of Chippewa Indians living on Wisconsin reservations. Public Health Reports 1994;109(6):820–3.

Date last reviewed: 05/12/2006
Content source: Division for Heart Disease and Stroke Prevention, National Center for Chronic Disease Prevention and Health Promotion