Series 2, Number 166 April NCHS Urban Rural Classification Scheme for Counties

Transcription

1 Series 2, Number 166 April NCHS Urban Rural Classification Scheme for Counties

2 Copyright information All material appearing in this report is in the public domain and may be reproduced or copied without permission; citation as to source, however, is appreciated. Suggested citation Ingram DD, Franco SJ NCHS urban rural classification scheme for counties. National Center for Health Statistics. Vital Health Stat 2(166) Library of Congress Catalog Number dc23 For sale by the U.S. Government Printing Office Superintendent of Documents Mail Stop: SSOP Washington, DC Printed on acid-free paper.

3 Series 2, Number NCHS Urban Rural Classification Scheme for Counties Data Evaluation and Methods Research U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES Centers for Disease Control and Prevention National Center for Health Statistics Hyattsville, Maryland April 2014 DHHS Publication No

4 National Center for Health Statistics Charles J. Rothwell, M.S., M.B.A., Director Jennifer H. Madans, Ph.D., Associate Director for Science Office of Analysis and Epidemiology Irma E. Arispe, Ph.D., Director

5 Contents Acknowledgments... Abstract.... Introduction NCHS Urban Rural Classification Scheme for Counties OMB Metropolitan Nonmetropolitan Classification of Counties... 2 County Urbanization Levels Defined... 2 Assignment of Counties to the Six Levels of the 2013 NCHS Scheme Changes in County Urbanization Category: Comparing 2013 and 2006 Schemes... 4 Health Measures by Urbanization Level... Considerations When Analyzing Trends Using Urbanization Classification Schemes.... Summary.... References.... Appendix I. Assignment of Large Metropolitan Counties Appendix II. List of U.S. Counties and County-equivalent Entities and Their Urbanization-level Assignments List of Detailed Tables 1. Categories and classification rules of the 2013 NCHS Urban Rural Classification Scheme for Counties Number of counties and percentage of the U.S. population in each urbanization level of the 2013 and 2006 NCHS Urban Rural Classification Schemes for Counties Median for selected county characteristics, by urbanization level: 2013 NCHS Urban Rural Classification Scheme for Counties Classification of counties according to the 2013 and 2006 NCHS Urban Rural Classification Schemes for Counties Age-adjusted death rates for all causes, by age, sex, and NCHS Urban Rural Classification Scheme for Counties: United States, average annual Age-adjusted death rates for selected causes, by sex and NCHS Urban Rural Classification Scheme for Counties: United States, average annual Selected health measures for persons aged 18 64, by NCHS Urban Rural Classification Scheme for Counties: National Health Interview Survey, List of Appendix Tables I. Comparison of classification rule and discriminant model assignments of large metro counties to large central and large fringe metro categories: 2013 NCHS Urban Rural Classification Scheme for Counties II. Classification rule and discriminant model assignments of the large metro counties and county-equivalent entities with discordant assignments and their final assignments: 2013 and 2006 NCHS Urban Rural Classification Schemes for Counties III. First quartile, median, and third quartile of selected characteristics of large central and large fringe metro counties: IV NCHS Urban Rural Classification Scheme for Counties Counties in large metropolitan statistical areas (1 million or more population) and their 2013 and 2006 NCHS Urban Rural Classification Scheme assignments, by February 2013 Office of Management and Budget-designated metropolitan statistical area V. List of U.S. counties and county-equivalent entities and their urbanization levels: 2013, 2006, and 1990 census-based NCHS Urban Rural Classification Scheme for Counties iv iii

6 Acknowledgments The authors gratefully acknowledge the contributions made by Jeffrey Pearcy who created the map of the 2013 NCHS Urban Rural Classification Scheme for Counties and Shilpa Bengeri who did the computer programming for the National Health Interview Survey data examples. The authors also gratefully acknowledge the assistance of the following people at NCHS who reviewed this report: Jeffrey Berko, Jennifer D. Parker, Julia S. Holmes, and Jennifer H. Madans. The report was edited and produced by NCHS Office of Information Services, Information Design and Publishing Staff: Jen Hurlburt edited the report, typesetting was done by Annette F. Holman, and graphics were produced by Sarah Hinkle. iv

7 Objectives This report details development of the 2013 National Center for Health Statistics (NCHS) Urban Rural Classification Scheme for Counties (update of the 2006 NCHS scheme) and applies it to health measures to demonstrate urban-rural health differences. Methods The methodology used to construct the 2013 NCHS scheme was the same as that used for the 2006 NCHS scheme, but 2010 census-based data were used rather than 2000 census-based data. All U.S. counties and county-equivalent entities are assigned to one of six levels (four metropolitan and two nonmetropolitan) based on: 1) their February 2013 Office of Management and Budget designation as metropolitan, micropolitan, or noncore; 2) for metropolitan counties, the population size of the metropolitan statistical area (MSA) to which they belong; and 3) for counties in MSAs of 1 million or more, the location of principal city populations within the MSA. The 2013 and 2006 NCHS schemes were applied to data from the National Vital Statistics System (NVSS) and National Health Interview Survey (NHIS) to illustrate differences in selected health measures by urbanization level and to assess the magnitude of differences between estimates from the two schemes. Results and Conclusions County urban-rural assignments under the 2013 NCHS scheme are very similar to those under the 2006 NCHS scheme. Application of the updated scheme to NVSS and NHIS data demonstrated the continued usefulness of the six categories for assessing and monitoring health differences among communities across the full urbanization spectrum. Residents of large central and large fringe metro counties differed substantially on many health measures, illustrating the importance of continuing to separate these counties. Residents of large fringe metro counties generally fared better than residents of less urban counties. Estimates obtained from the 2013 and 2006 schemes were similar. Keywords: urbanization scheme metropolitan nonmetropolitan micropolitan 2013 NCHS Urban Rural Classification Scheme for Counties by Deborah D. Ingram, Ph.D., and Sheila J. Franco, Office of Analysis and Epidemiology Introduction Urban rural differences in health measures have long been recognized. The National Center for Health Statistics (NCHS) Urban Rural Classification Scheme for Counties was developed for use in studying associations between urbanization level of residence and health and for monitoring the health of urban and rural residents. The scheme groups U.S. counties and county-equivalent entities into six urbanization levels (four metropolitan and two nonmetropolitan), on a continuum ranging from most urban to most rural. (Note: In this report, the term counties will be used to refer to counties and county-equivalent entities.) The first NCHS urban rural scheme, developed in 2001 and referred to as the 1990 census-based NCHS urban rural scheme, was a six-level classification scheme for counties. This scheme was based on the 1990 Office of Management and Budget (OMB) standards for defining metropolitan statistical areas (MSAs) and on data from the 1990 census. It was updated in 2006 (referred to as the 2006 NCHS Urban Rural Classification Scheme for Counties), following release of MSAs and micropolitan statistical area delineations based on application of the 2000 OMB standards for defining MSAs and micropolitan statistical areas to 2000 census data. A detailed description of the construction of these two schemes is provided in an earlier report (1). Release of 2010 census-based population data, and of MSA and micropolitan statistical area delineations based on the application of the 2010 OMB standards for defining these areas to 2010 census data, prompted this update of the 2006 NCHS scheme. The updated scheme will be referred to as the 2013 NCHS Urban Rural Classification Scheme for Counties. The NCHS urban rural schemes are county-based because counties generally are the primary political units of local government and have programmatic importance at the federal and state levels, and their boundaries are relatively stable. Further, county-level measures of health and of demographic, economic, and environmental characteristics are widely available, in contrast to the paucity of data available at the subcounty level. A distinguishing feature of the NCHS urban rural schemes is that they separate counties in the largest metropolitan areas (MSAs with populations of 1 million or more) into two groups, referred to as large central metro and large fringe metro. The large central metro group includes counties that contain all or part of the area s principal city (akin to inner cities); the large fringe metro group includes the surrounding counties of the MSA. NCHS has found that there are significant health differences between residents of these two types of counties and that urban rural health differences often are largest between residents of large fringe metro counties and residents of the most rural counties. Therefore, Page 1

8 Page 2 [ Series 2, No. 166 accurate assessment of urban rural health differences requires a classification scheme that differentiates between large central metro and large fringe metro counties. This report describes the construction of the 2013 NCHS Urban Rural Classification Scheme for Counties and compares it with the 2006 NCHS scheme. The 2013 NCHS scheme was constructed using the same methodology used for the 2006 scheme, but it is based on 2010 census data rather than 2000 census data (1). To illustrate differences in selected health measures by urbanization level and to compare estimates from the two schemes, selected health measures are analyzed according to both the 2013 and 2006 schemes NCHS Urban Rural Classification Scheme for Counties The 2013 NCHS Urban Rural Classification Scheme for Counties, like the 2006 NCHS urban rural scheme, is a county-level scheme with six levels: four metropolitan (large central metro, large fringe metro, medium metro, and small metro) and two nonmetropolitan (micropolitan and noncore) (Table 1 and Figure 1). A detailed description of the construction of the 2006 NCHS scheme is provided in an earlier report (1). All counties in the United States were assigned to one of the six levels based on: 1) their status under the OMB delineation of MSAs and micropolitan statistical areas, 2) the population size of MSAs, and 3) the location of principal city populations within the largest MSAs (1 million or more population) OMB Metropolitan Nonmetropolitan Classification of Counties MSAs and micropolitan statistical areas, delineated in accordance with the 2010 OMB standards for defining metropolitan and micropolitan statistical areas, formed the basis for the 2013 NCHS scheme. The 2010 OMB standards are nearly identical to the 2000 standards (2). They include criteria for differentiating metropolitan and nonmetropolitan counties and for organizing metropolitan counties into MSAs and some of the nonmetropolitan counties into micropolitan statistical areas. The term core-based statistical area (CBSA) refers collectively to MSAs and micropolitan statistical areas. Nonmetropolitan counties not designated as micropolitan are referred to as counties outside of CBSAs or noncore counties. According to the 2010 OMB standards, the basic concept of an MSA remained that of an area containing a large population nucleus together with adjacent communities having a high degree of economic and social integration with that core. One or more census-defined urbanized areas form the core of an MSA and the county (or counties) that contain them are referred to as central counties of the MSA. (An urbanized area has a population of at least 50,000 and consists of an urban nucleus with a population density of 1,000 persons per square mile together with adjoining territory with at least 500 persons per square mile) (3). In addition to the central counties, an MSA may contain outlying counties that are economically and socially tied to the central counties as measured by the percentage commuting to or from the central counties for work. Generally, all or most of the counties in MSAs qualify as central counties. The concept of a micropolitan statistical area closely parallels that of the MSA, but a micropolitan statistical area comprises nonmetropolitan counties and has a smaller nucleus. One or more urban clusters form the core of a micropolitan statistical area, and the county (or counties) that contain them are referred to as central counties (an urban cluster is a small version of an urbanized area with 2,500 49,999 inhabitants) (3). In addition to the central counties, a micropolitan statistical area may contain outlying counties that meet specified requirements of commuting to or from the central counties. Nonmetropolitan counties not defined as micropolitan are considered to be noncore and may be thought of as the most rural areas. The largest incorporated city in an MSA or micropolitan statistical area is designated as a principal city. Additional cities qualify as principal cities if specified population size and commuting criteria are met. County Urbanization Levels Defined From the most urban to the most rural, the six levels of the 2013 NCHS scheme are defined according to the following classification rules (Table 1 and Figure 1): Metropolitan categories Large central metro Counties in MSAs of 1 million or more population that: 1. Contain the entire population of the largest principal city of the MSA, or 2. Have their entire population contained in the largest principal city of the MSA, or 3. Contain at least 250,000 inhabitants of any principal city of the MSA. Large fringe metro Counties in MSAs of 1 million or more population that did not qualify as large central metro counties. Medium metro Counties in MSAs of populations of 250,000 to 999,999. Small metro Counties in MSAs of populations less than 250,000. Nonmetropolitan categories Micropolitan Counties in micropolitan statistical areas. Noncore Nonmetropolitan counties that did not qualify as micropolitan. Assignment of Counties to the Six Levels of the 2013 NCHS Scheme Based on the February 2013 delineation of MSAs and micropolitan statistical areas, all counties in the

9 Series 2, No. 166 [ Page 3 U.S. counties and county equivalents Metropolitan Nonmetropolitan Large metro MSA population 1 million or more Medium metro MSA population 250, ,999 Small metro MSA population less than 250,000 Micropolitan Urban cluster population 10,000 49,999 Noncore Large central metro Large fringe metro NOTE: MSA is metropolitan statistical area. Figure 1. Structure of the 2013 NCHS Urban Rural Classification Scheme for Counties United States were identified as either metropolitan or nonmetropolitan, and nonmetropolitan counties were further identified as either micropolitan or noncore (4). MSA populations were derived from the Vintage 2012 postcensal estimates of the July 1, 2012, resident population of counties and were used to divide metropolitan counties into three categories based on the population size of their MSA: large (MSA population of 1 million or more), medium (MSA population of 250, ,999), and small (MSA population less than 250,000) (5). Counties in the large metro category (MSA population of 1 million or more) were further subdivided into the large central metro and large fringe metro categories of the NCHS scheme using the classification rules outlined above that involve the location of the MSA principal cities. The set of principal cities used to subdivide the large MSAs was based on city designations in the February 2013 OMB delineation of MSAs (6). Principal city populations were derived from the Vintage 2012 postcensal estimates of the July 1, 2012, resident population of places (7). The classification rule assignments of the large metro counties (counties in MSAs of 1 million or more population) to the large central metro and large fringe metro categories were confirmed with a discriminant analysis. The discriminant model included settlement density, socioeconomic, and demographic variables considered key in differentiating central and fringe counties (see Appendix I for details). The discriminant analysis confirmed that large central metro and large fringe metro counties continue to differ on key characteristics and that the classification rules continue to work well. As for the 2006 scheme, the discriminant analysis also identified counties whose assignments warranted review (those with discordant classification rule and discriminant model assignments). For the 2013 scheme, only 12 of the 436 large metro counties had discordant assignments according to the two methods. After examining various characteristics of these counties, the classification rule assignments of five were changed. The classification rule assignments of the remaining seven discordant counties were not changed (Appendix I). The number of counties and percentage of the U.S. population in each level of the 2013 NCHS scheme are presented in Table 2. Thirty-seven percent (1,167 counties) of the 3,143 counties were assigned to a metropolitan category and 63% were assigned to a nonmetropolitan category (20%, 641 counties, were assigned to the micropolitan category and the remaining 42%, 1,335 counties, were assigned to the noncore category). Although a majority of counties are nonmetropolitan, the vast majority of the U.S. population resides in MSAs (85%). All counties and their classifications under the 2013 NCHS scheme are shown in the map (Figure 2). The map illustrates that the eastern half of the United States is more densely settled than the western half and that counties east of the Mississippi River are smaller than those west of the river. A listing of county assignments according to the 2013, 2006, and 1990 census-based NCHS schemes is provided in Appendix II. Table 3 presents medians of selected county characteristics for each of the six urbanization levels. Median county population, population density, and housing densities decrease with decreasing urbanization. The differences between large central and large fringe metro counties for these characteristics

10 Page 4 [ Series 2, No Large central metro 2 Large fringe metro 3 Medium metro 4 Small metro 5 Micropolitan 6 Noncore Figure 2. Distribution of counties according to the 2013 NCHS Urban Rural Classification Scheme for Counties are especially large. The median percentage of residents commuting outside of the county to work and the median percentage of the population that is non-hispanic white are lower for large central counties than for counties at the other urbanization levels. Large fringe metro counties have the highest percentage commuting outside of the county to work, the highest median household income, and the lowest percentage of families below the poverty level. Noncore counties have substantially lower median population, population and housing densities, a lower median household income, and a somewhat less racially and ethnically diverse population than counties at the other urbanization levels. Changes in County Urbanization Category: Comparing 2013 and 2006 Schemes Comparison of the 2013 and 2006 NCHS scheme assignments for each county showed that 286 (9%) had different category assignments in the two schemes (Table 4). Most of these 286 counties moved to a more urban category, with about 17% shifting from a less urban metropolitan category to a more urban one, about 40% shifting from a nonmetropolitan category to a metropolitan category, and 15% shifting from the noncore category to the micropolitan category. Many of the category shifts involved MSAs and micropolitan statistical areas shifting from one level to another because of population growth. Some shifts involved changes in the composition of specific MSAs (counties being added or deleted). Few shifts occurred among the large central and large fringe metro counties; all 63 of the counties in the large central metro category for the 2006 scheme also were large central for the 2013 scheme; 338 of the 354 large fringe counties in the 2006 scheme also were large fringe counties for the 2013 scheme. Twenty-eight percent (81 counties) of the 286 counties that changed category moved from a more urban category to a less urban one. One-half of these shifts occurred because the county was identified as micropolitan under the December 2005 OMB delineation, but as noncore under the February 2013 OMB delineation.

11 Series 2, No. 166 [ Page 5 As a result of the 286 county category shifts, the four metropolitan levels of the 2013 NCHS scheme have more counties than the corresponding levels of the 2006 scheme, and the two nonmetropolitan levels have fewer (Table 2). The largest differences in the category counts occur for the medium metro level (373 counties in the 2013 scheme compared with 332 in the 2006 scheme) and the micropolitan level (641 counties in the 2013 scheme compared with 694 in the 2006 scheme). The percentage of the U.S. population in each of the six levels was similar for the two schemes. Health Measures by Urbanization Level The 2013 and 2006 NCHS Urban Rural Classification Schemes for Counties were merged with National Vital Statistics System (NVSS) mortality records and National Health Interview Survey (NHIS) data using restricted-use files to illustrate the ability of the NCHS scheme to identify health differences across urbanization levels. Sex-specific death rates for selected age groups and causes of death during were examined across the urbanization levels (Tables 5 and 6). Estimates of a health status measure (report of fair or poor health), a health access measure (lack of health insurance), and a health-related behavior (current cigarette smoking) were examined using data from the NHIS (Table 7). NHIS collects data on a broad range of health topics through personal household interviews. For all of the health measures, a test of linear trend across the urbanization levels was performed. For all but 4 of the 10 health measures (death rates for adults aged 65 and over, motor vehicle and cerebrovascular death rates, and percentage who currently smoke), the large central metro category value was higher than the large fringe metro category value, so the large central metro category was omitted from the trend analysis. For each of the health measures, the statistical significance of pairwise comparisons of the urban rural categories was assessed using a z test and a p value of 0.05 without adjusting for multiple comparisons. Note that while the large central metro category was omitted from some of the trend analyses, it was included in all of the pairwise comparisons. For all of the health measures examined, the estimates for each urbanization level under the 2013 and 2006 NCHS schemes were identical or very similar. Therefore, statements about the patterns observed will be restricted to the 2013 estimates. Infant mortality Infant mortality rates during increased for both sexes with decreasing urbanization from a low in large fringe metro counties. For both males and females, infant mortality was 11% 23% lower in fringe counties than in any of the other urbanization levels. Mortality for children and young adults For males aged 1 24 years, the age-adjusted death rate was lowest in large fringe metro counties and highest in noncore counties (the most rural counties). Comparing rates for males in fringe counties with those in the other urbanization levels shows that the rates in large central, medium, and small metro counties were all moderately higher (5% 8%), the rate in micropolitan counties was 22% higher, and the rate in noncore counties was 57% higher. For females, the age-adjusted death rates in large central and large fringe metro counties were similarly low, those in medium and small metro counties were about 10% higher, and those in micropolitan and noncore counties were 34% and 68% higher, respectively. Mortality for adults aged In , the death rate for adults aged was lowest in fringe counties of large metro areas and increased steadily as counties become more rural. The age-adjusted death rate in the noncore counties was 44% higher for males and 47% higher for females. For both males and females in this age group, the death rate was higher in large central metro counties than in large fringe metro counties. Mortality for adults aged 65 and over Among adults aged 65 and over, the age-adjusted death rate was lowest in large central metro counties and increased with decreasing urbanization. For both sexes, the death rate in noncore counties was 14% higher than the rate in large central metro counties. Homicide The ageadjusted homicide rate for males in large central metro counties was about one and one-half times the rate in medium metro counties and more than double the rate in any of the other urbanization levels. There was no linear trend across the urbanization levels. For females, the homicide rate was highest in large central metro and noncore counties. Homicide rates for females are much lower than those for males, and the relative differences among the rates in the urbanization levels were smaller than those among males. Motor vehicle accident mortality Death rates for motor vehicle accidents increase markedly as counties become less urban. For males, the age-adjusted motor vehicle death rates in the most rural noncore counties were nearly three times as high as those in large central metro counties; for females, they were more than three times as high. Cerebrovascular disease mortality For males, the age-adjusted cerebrovascular disease death rate was lowest in large fringe metro counties and increased with decreasing urbanization level (the rates in micropolitan and noncore counties were about 23% higher than in large fringe counties). For females, the age-adjusted cerebrovascular disease death rates also increased with decreasing urbanization level, but the rate was similarly lowest in the large central and large fringe metro counties. The rate in noncore counties was about 29% higher than the rates in the large central and fringe counties. Health status The percentage of NHIS respondents aged reporting fair or poor health (out of a five-level scale of excellent, very good, good, fair, or poor)

12 Page 6 [ Series 2, No. 166 during was lowest among those residing in large fringe metro counties and increased as counties became less urban. Compared with large fringe metro counties, the prevalence of fair or poor health was 21% higher in large central metro counties and twice as high in noncore counties. Health insurance The percentage of NHIS respondents aged reporting no health insurance during was lowest in large fringe metro counties and increased with decreasing urbanization. Lack of health insurance was 34% 53% higher for large central metro and nonmetropolitan (micropolitan and noncore) residents than for residents of large fringe metro counties. For medium and small metro counties, the percentage reporting no health insurance was 22% 24% higher than that for large fringe metro counties. Current smokers The percentage of NHIS respondents aged reporting during that they smoked cigarettes increased steadily with decreasing urbanization. The percentage was 16.9% among residents of large central metro counties and 29.1% 29.4% among residents of nonmetropolitan counties. Application of the 2013 NCHS urban rural scheme to the selected mortality and health measures demonstrated that the six levels of the scheme capture important health differentials across the entire urbanization spectrum. The examples show that residents of large fringe metro counties fare better on most of the mortality and health measures than residents of other urbanization levels (including residents of large central metro counties). In addition, significant differences for several measures were observed between medium and small metro counties and between micropolitan and noncore counties. Application of the 2006 NCHS scheme to the same data to which the 2013 scheme was applied produced identical or very similar estimates. Considerations When Analyzing Trends Using Urbanization Classification Schemes Changes in the classification of counties that result when urbanization schemes are updated ensure that current settlement patterns are represented, but they can make comparison of statistics computed using different versions of the scheme problematic. When studying urban rural differentials in health over a time period that spans more than one version of an urban rural classification scheme, the analyst must decide whether to compute the health measures using the schemes in effect at each point in time, or to compute them using only one version of the scheme. For some purposes, using different versions of an urbanization scheme in an analysis may be appropriate; for other purposes, it may be desirable to compute all of the statistics using the same version so that a stable set of counties is used. Certainly, whenever statistics are computed for several years of data combined, data for the aggregated years should be classified using the same version of the urban rural scheme. It may be useful to assess the comparability of different versions of an urbanization scheme by computing statistics using the different schemes. As described in Changes in County Urbanization Category: Comparing 2013 and 2006 Schemes, the county assignments under the 2013 and 2006 NCHS schemes are similar: 91% of counties were assigned to the same category for both schemes, and most category shifts occurred between the two nonmetropolitan levels (where they are less noticeable because of the large numbers of counties in those levels). As a result, statistics computed using the 2013 and 2006 schemes were quite similar. This was illustrated in Health Measures by Urbanization Level. There were many more differences between the county assignments under the 2006 and 1990 census-based NCHS schemes; thus, statistics computed using these two schemes are more likely to differ (1). Summary This report describes the 2013 NCHS Urban Rural Classification Scheme for Counties and demonstrates its use in describing urban rural health differences. The 2013 NCHS scheme, which was constructed using 2010 census-based data, is an update of the 2006 NCHS scheme. Like the 2006 scheme, it groups U.S. counties into six urbanization levels reflecting their position on a continuum ranging from most urban to most rural: large central metro, large fringe metro, medium metro, small metro, micropolitan, and noncore. Because the 2010 OMB standards for defining MSAs and micropolitan statistical areas did not change materially from the 2000 OMB standards, and the rules used to assign counties to the six urban rural levels of the scheme did not change, any category shifts between the 2006 and 2013 NCHS schemes are simply a reflection of population or commuting gains or losses. As a result, the county assignments of the 2013 and 2006 NCHS schemes are similar. Application of the two schemes to selected health measures from NVSS mortality files and NHIS data produced very similar estimates. Significant health differences between residents of the six urbanization levels were observed. For some measures, residents of large fringe metro counties fared better than residents of other urbanization levels, demonstrating the importance of separating large metro counties into the large central and large fringe metro categories. Assessing the magnitude and types of health problems confronting communities at different urbanization levels contributes to effective health programs and policies. The NCHS scheme is a useful tool for studying associations between urbanization level of residence and health and for monitoring the health of urban and rural residents. A key feature of the NCHS scheme is its separation of MSAs of

13 Series 2, No. 166 [ Page 7 1 million or more population into the large central metro and large fringe metro categories. This feature is important because residents of central and fringe counties often differ substantially on health measures, and residents of fringe counties often fare better than residents of other urbanization levels. This feature means that the NCHS scheme provides a more accurate characterization of differences in health measures across the complete range of the urban rural continuum than other urban rural schemes. The NCHS scheme is available on some NCHS data files (both public-use and restricted-use) and can be linked to many other NCHS restricted-use data files as well as to other county-level data files. References 1. Ingram DD, Franco SJ. NCHS urban rural classification scheme for counties. National Center for Health Statistics. Vital Health Stat 2(154) Office of Management and Budget standards for delineating metropolitan and micropolitan statistical areas. Fed Regist 75(123): Available from: files/omb/assets/fedreg_2010/ _metro_standards-complete.pdf. 3. U.S. Census Bureau census urban and rural classification and urban area criteria. Washington, DC. Available from: reference/ua/urban-rural-2010.html. 4. Office of Management and Budget. Revised delineations of metropolitan statistical areas, micropolitan statistical areas, and combined statistical areas, and guidance on uses of the delineations of these areas. OMB bulletin no Washington, DC Available from: whitehouse.gov/sites/default/files/omb/ bulletins/2013/b pdf. 5. National Center for Health Statistics. Vintage 2012 postcensal estimates of the resident population of the United States (April 1, 2010, July 1, 2010 July 1, 2012), by year, county, single-year of age (0, 1, 2,..., 85 years and over), bridged race, Hispanic origin, and sex. Prepared under a collaborative arrangement with the U.S. Census Bureau. Available from: cdc.gov/nchs/nvss/bridged_race.htm as of June 13, 2013, following release by the U.S. Census Bureau of the unbridged Vintage 2012 postcensal estimates by 5-year age group on June 13, U.S. Census Bureau. Principal cities of metropolitan and micropolitan statistical areas. Washington, DC. Available from: metro/data/def.html. 7. U.S. Census Bureau. Annual estimates of the resident population for incorporated places: April 1, 2010 to July 1, Washington, DC. Available from: popest/data/cities/totals/2012/sub EST html. 8. SAS Institute. SAS (Version 9.3) [computer software] Massey DS, Denton NA. The dimensions of residential segregation. Soc Forces 67(2): Iceland J, Weinberg DH, Steinmetz E. Racial and ethnic residential segregation in the United States: Appendix B. Measures of residential segregation. Washington, DC: U.S. Census Bureau Available from: hhes/www/housing/housing_patterns/ app_b.html.

14 Page 8 [ Series 2, No. 166 Table 1. Categories and classification rules of the 2013 NCHS Urban Rural Classification Scheme for Counties Urbanization level Classification rules 1 Metropolitan counties 1 Large central metro 2... Counties in MSAs of 1 million or more population that: 1) Contain the entire population of the largest principal city of the MSA, or 2) Have their entire population contained in the largest principal city of the MSA, or 3) Contain at least 250,000 inhabitants of any principal city of the MSA Large fringe metro 2... Counties in MSAs of 1 million or more population that did not qualify as large central metro counties Medium metro 2... Counties in MSAs of populations of 250, ,999 Small metro 2... Counties in MSAs of populations less than 250,000 Nonmetropolitan counties 1 Micropolitan 1... Counties in micropolitan statistical areas Noncore 1... Nonmetropolitan counties that did not qualify as micropolitan 1 Status determined from the February 2013 Office of Management and Budget s delineation of metropolitan and micropolitan statistical areas. 2 MSA and principal city populations derived using Vintage 2012, postcensal estimates of the July 1, 2012, resident population of counties and places. NOTES: NCHS is National Center for Health Statistics; metro is metropolitan; MSA is metropolitan statistical area. The urbanization levels and classification rules used for the 2013 NCHS scheme are the same as those used for the 2006 NCHS scheme. Table 2. Number of counties and percentage of the U.S. population in each urbanization level of the 2013 and 2006 NCHS Urban Rural Classification Schemes for Counties 2013 NCHS scheme 2006 NCHS scheme Urbanization level Number of counties 1 Distribution of U.S. resident population 2 (percent) Number of counties 1 Distribution of U.S. resident population 3 (percent) All counties.... 3, , Metropolitan counties 4... Large central metro... Large fringe metro... Medium metro.... Small metro... 1, , Nonmetropolitan counties... Micropolitan 4... Noncore , , , , For comparability, U.S. Census 2010 county geography is used for the 2013 and 2006 NCHS schemes. 2 July 1, 2012, estimate of the resident population of U.S. counties from the Vintage 2012 series of postcensal estimates. 3 July 1, 2006, estimate of the resident population of U.S. counties from the intercensal series of estimates. 4 For the 2013 NCHS scheme, the metropolitan, micropolitan, and noncore status of counties was determined from the February 2013 Office of Management and Budget s delineation of metropolitan and micropolitan statistical areas. For the 2006 NCHS scheme, it was determined from the December 2005 delineation of areas. NOTES: NCHS is National Center for Health Statistics; metro is metropolitan.

15 Table 3. Median for selected county characteristics, by urbanization level: 2013 NCHS Urban Rural Classification Scheme for Counties Housing density, Residents County Population urban (housing commuting Median Families population, density Housing density units per square outside of Jobs to household below Non-Hispanic July 1, 2012 (persons per (housing units per mile in urban county to workers in income poverty white population Urbanization level (number) square mile) square mile) block groups) 1 work (percent) county (ratio) (dollars) level (percent) (percent) Metropolitan counties Large central metro ,458 2, , Large fringe metro... 93, , Medium metro... 92, , Small metro... 76, , Nonmetropolitan counties Micropolitan... 38, , Noncore... 11, , Urban block groups have 640 or more housing units per square mile. NOTES: NCHS is National Center for Health Statistics; metro is metropolitan. Table 4. Classification of counties according to the 2013 and 2006 NCHS Urban Rural Classification Schemes for Counties 2006 NCHS scheme urbanization level Large Large Medium Small 2013 NCHS scheme urbanization level central metro fringe metro metro metro Micropolitan Noncore Large central metro Large fringe metro Medium metro Small metro Micropolitan Noncore ,269 NOTES: NCHS is National Center for Health Statistics; metro is metropolitan. Series 2, No. 166 [ Page 9

16 Table 5. Age-adjusted death rates for all causes, by age, sex, and NCHS Urban Rural Classification Scheme for Counties: United States, average annual Infants 1 Children and young adults 2,3 Adults aged years 3 Adults aged 65 years and over NCHS scheme 2006 NCHS scheme 2013 NCHS scheme 2006 NCHS scheme 2013 NCHS scheme 2006 NCHS scheme 2013 NCHS scheme 2006 NCHS scheme Urbanization level Rate SE Rate SE Rate SE Rate SE Rate SE Rate SE Rate SE Rate SE Metropolitan counties Large central metro , , Large fringe metro , , Medium metro , , Small metro , , Males Page 10 [ Series 2, No. 166 Nonmetropolitan counties Micropolitan , , Noncore , , Metropolitan counties Large central metro , , Large fringe metro , , Medium metro , , Small metro , , Nonmetropolitan counties Micropolitan , , Noncore , , Under age 1 year. Infant mortality rates are infant deaths per 1,000 live births. 2 Aged 1 24 years. 3 Rates are deaths per 100,000 population and are age-adjusted. NOTES: NCHS is National Center for Health Statistics; SE is standard error; metro is metropolitan. Cause of death was coded according to the International Classification of Diseases, Tenth Revision (ICD 10). Females

17 Series 2, No. 166 [ Page 11 Table 6. Age-adjusted death rates for selected causes, by sex and NCHS Urban Rural Classification Scheme for Counties: United States, average annual Homicide 1 Motor vehicle accidents 2 Cerebrovascular diseases NCHS scheme 2006 NCHS scheme 2013 NCHS scheme 2006 NCHS scheme 2013 NCHS scheme 2006 NCHS scheme Urbanization level Rate SE Rate SE Rate SE Rate SE Rate SE Rate SE Metropolitan counties Males Large central metro... Large fringe metro... Medium metro... Small metro Nonmetropolitan counties Micropolitan.... Noncore Metropolitan counties Females Large central metro... Large fringe metro... Medium metro... Small metro Nonmetropolitan counties Micropolitan.... Noncore Includes ICD 10 codes *U01 *U02, X85 Y09, and Y Includes ICD 10 codes V02 V04, V09.0, V09.2, V12 V14, VV19.0 V19.2, V19.4 V19.6, V20 V79, V80.3 V80.5, V81.0 V81.1, V82.0 V82.1, V83 V86, V87.0 V87.8, V88.0 V88.8, V89.0, and V Includes ICD 10 codes I60 I69. NOTES: NCHS is National Center for Health Statistics; SE is standard error. Rates are deaths per 100,000 population and are age-adjusted. Cause of death was coded according to the International Classification of Diseases, Tenth Revision (ICD 10). Table 7. Selected health measures for persons aged 18 64, by NCHS Urban Rural Classification Scheme for Counties: National Health Interview Survey, Fair or poor respondent-assessed health status 1 No health insurance coverage 2 Current smoker NCHS scheme 2006 NCHS scheme 2013 NCHS scheme 2006 NCHS scheme 2013 NCHS scheme 2006 NCHS scheme Urbanization level Percent SE Percent SE Percent SE Percent SE Percent SE Percent SE Metropolitan counties Large central metro Large fringe metro Medium metro Small metro Nonmetropolitan counties Micropolitan Noncore Based on response to the question, Would you say your health in general is excellent, very good, good, fair, or poor? 2 At the time of interview. Persons not covered by private insurance, Medicaid, Children s Health Insurance Program, state-sponsored or other government-sponsored health plans, Medicare, or military plans are considered to have no health insurance coverage. Persons with only Indian Health Service coverage are considered to have no health insurance coverage. 3 Defined as ever smoking 100 cigarettes in his or her lifetime and smoking now every day or some days. NOTES: NCHS is National Center for Health Statistics; SE is standard error; metro is metropolitan.

18 Page 12 [ Series 2, No. 166 Appendix I. Assignment of Large Metropolitan Counties Large metropolitian (metro) counties [counties in metropolitan statistical areas (MSAs) of 1 million or more population] are assigned either to the large central metro category or the large fringe metro category of the National Center for Health Statistics (NCHS) Urban-Rural Classification Scheme for Counties. Classification rules based on the location of principal city populations are used to make the assignments. The same rules used to make these assignments for the 2006 NCHS scheme were used to make assignments for the 2013 NCHS scheme (Table 1). Whether the rules achieved a reasonable separation of the large metro counties into the categories for the 2013 scheme was assessed by examining the agreement between each county s classification rule assignment and its assignment according to a discriminant model that included selected settlement density, socioeconomic, and demographic variables. Discriminant Analysis The settlement density, socioeconomic, and demographic variables considered for inclusion in the discriminant model were derived from the 2010 census, the American Community Survey, and the Vintage 2012 postcensal estimates of the resident population of counties. A stepwise discriminant procedure (performed using SAS PROC STEPDISC) identified 12 variables as significant predictors of urbanization category. These variables were included in the discriminant model fit using SAS PROC DISCRIM (8): + Population of the MSA as of July 1, 2012 (log) + Population of the county as of July 1, 2012 (log) + Population density (number of people residing per square mile) + Mean housing density of urban block groups (number of housing units per square mile for block groups with 640 or more housing units per square mile) + Mean housing density of exurban or rural block groups (number of housing units per square mile for block groups with fewer than 64 housing units per square mile) + Percentage of county area in urban block groups (block groups with 640 or more housing units per square mile) + Percentage of county area in exurban or rural block groups (block groups with fewer than 64 housing units per square mile) + Percentage of housing units that are owner occupied + Percentage of county residents commuting outside of the county for work + Ratio of jobs to workers in the county + Median household income in the county + Percentage of families below the poverty level The classification rule and discriminant model county assignments were in close agreement they made the same assignment for all but 12 of the 436 large metro counties (Table I). Four of the 12 counties for which there was disagreement were categorized as central by the classification rules and as fringe by the discriminant model; the remaining eight were categorized as fringe by the classification rules and as central by the discriminant model. Thus, as for the 2006 NCHS scheme, the separation of large metro counties achieved using simple classification rules based on the location of large principal city populations within the MSAs closely parallels the separation achieved using a discriminant model with settlement density, socioeconomic, and demographic variables. Resolution of Discordant Assignments The 12 counties with discordant assignments in 2013 and the 10 counties with discordant assignments in 2006 are shown in Table II. Six of the 12 counties that were discordant for the 2013 scheme were similarly discordant for the 2006 scheme; the other six were classified as fringe in 2006 by both methods. The remaining four counties that had discordant assignments in 2006 had concordant assignments in For three of these, the 2013 assignment agreed with the final 2006 assignment, but for the fourth, Norfolk city, VA, it did not. In 2006, Norfolk city was given a final assignment of large central metro, while in 2013 both the classification rules and discriminant model assigned it to large fringe metro. Table I. Comparison of classification rule and discriminant model assignments of large metro counties to large central and large fringe metro categories: 2013 NCHS Urban Rural Classification Scheme for Counties Discriminant model assignment of large metro counties Classification rule assignment Large Large All large of large metro counties central metro fringe metro metro counties Large central metro Large fringe metro All large metro counties Counties for which category assignment by the classification rules agrees with assignment by the discriminant model. 2 Counties for which category assignment by the classification rules disagrees with assignment by the discriminant model. NOTES: NCHS is National Center for Health Statistics; metro is metropolitan.

19 Series 2, No. 166 [ Page 13 Table II. Classification rule and discriminant model assignments of large metro counties and county-equivalent entities with discordant assignments and their final assignments: 2013 and 2006 NCHS Urban Rural Classification Schemes for Counties 2013 NCHS scheme 2006 NCHS scheme Assignment Assignment Assignment Assignment according according to according to according to to classification discriminant Final classification discriminant Final County name rules model assignment rules model assignment Alexandria city, VA 1,2... Fringe Central Central Fringe Central Central Arapahoe County, CO 1... Central Fringe Fringe Fringe Fringe Fringe Arlington County, VA 1... Fringe Central Central Fringe Fringe Fringe Broward County, FL 1... Fringe Central Fringe Fringe Fringe Fringe Collin County, TX 1... Central Fringe Central Fringe Fringe Fringe DeKalb County, GA 1,2... Fringe Central Fringe Fringe Central Fringe Hudson County, NJ 2... Central Central Central Fringe Central Central Newport News city, VA 1... Fringe Central Fringe Fringe Fringe Fringe Norfolk city, VA 2,3... Fringe Fringe Central Fringe Central Central Pierce County, WA 2... Fringe Fringe Fringe Fringe Central Fringe Pinellas County, FL 1,2... Fringe Central Central Fringe Central Central Portsmouth city, VA 2... Fringe Fringe Fringe Fringe Central Fringe Providence County, RI 1,2... Central Fringe Central Central Fringe Central San Bernardino County, CA 1,2... Fringe Central Fringe Fringe Central Fringe Union County, NJ 1... Fringe Central Central Fringe Fringe Fringe Virginia Beach city, VA 1,2... Central Fringe Central Central Fringe Central 1 For the 2013 NCHS scheme, category assignments by classification rules and discriminant model disagreed. 2 For the 2006 NCHS scheme, category assignments by classification rules and discriminant model disagreed. 3 Although the classification rules and discriminant model both assigned Norfolk city to the fringe category for the 2013 NCHS scheme, it was assigned to the central category after examination of settlement, socioeconomic, and population variables. This assignment is consistent with its final assignment for the 2006 NCHS scheme. NOTES: NCHS is National Center for Health Statistics; metro is metropolitan. Final Assignment The final 2013 assignment for the 12 counties for which the classification rule and discriminant model assignments did not agree was made after comparing the settlement density, socioeconomic, and demographic characteristics of each county with those of the counties assigned to the central and fringe categories (data not shown). The classification rule assignments of 4 of the 12 counties (Alexandria city, VA; Arlington County, VA; Pinellas County, FL; and Union County, NJ) were changed from large fringe metro to large central metro because of their large populations and high housing densities, and because their socioeconomic and demographic characteristics were more like those of large central metro counties than those of large fringe metro counties (Table II). The classification rule assignment of Arapahoe County, CO, was changed from large central to large fringe metro because its settlement density values were more similar to those of large fringe counties than those of large central metro counties. The classification rule assignments of the seven remaining counties were not changed. For two of them, Providence County, RI, and Virginia Beach city, VA, their assignment by the classification rules to the large central metro category was retained because both counties contained all of the population of the largest principal city of their MSA. Although the classification rules and discriminant model assigned Norfolk city, VA, to the large fringe metro category, its assignment was changed to large central metro after examination of its settlement density, socioeconomic, and demographic variable values. This change made its 2013 assignment consistent with its 2006 assignment. The final assignment of the 436 large metro counties assigned 68 counties to the large central metro category and 368 to the large fringe metro category (Table 2). Comparison of Large Central and Large Fringe Metro Counties After making the final category assignments, the settlement density, socioeconomic, and demographic characteristics of the large central metro counties were compared with those of the large fringe metro counties to assess the magnitude of the differences in their characteristics. Table III shows the first quartile, median, and third quartile values for the examined characteristics (means are not shown because the distributions of many of the characteristics are highly skewed). There were substantial differences between the central and fringe counties for many of the variables. Generally, the interquartile portion of the fringe county distribution does not overlap that of the central county distribution. In summary: Settlement density: Large central counties tend to be more densely settled than large fringe counties. Median county population, population density, and housing density were about 9 10 times as high among large central counties compared with large fringe counties. In addition, large central counties had substantially higher housing densities within urban block groups, larger percentages of their areas in urban block groups, and smaller percentages of their areas in exurban and rural block groups.

20 Page 14 [ Series 2, No. 166 Table III. First quartile, median, and third quartile of selected characteristics of large central and large fringe metro counties: 2013 NCHS Urban Rural Classification Scheme for Counties Large fringe metro Large central metro Characteristic 1st quartile Median 3rd quartile 1st quartile Median 3rd quartile Settlement density County population (July 1, 2012)... 36,404 93, , , ,458 1,702,397 Population density (persons per square mile) ,216 2,037 4,822 Area in urban block groups (percent) Area in suburban block groups (percent) Area in exurban/rural block groups (percent) Housing density (housing units per square mile) ,903.8 Housing density, urban (housing units per square mile in urban block groups) , , , , ,244.5 Housing density, suburban (housing units per square mile in suburban block groups) Housing density, exurban/rural (housing units per square mile in exurban and rural block groups) Socioeconomic Commute outside of county to work (percent) Jobs to workers in county (ratio) Unemployed (percent) Owner-occupied housing units (percent) Median household income (dollars)... 48,977 56,861 68,416 45,305 50,950 58,265 Families below poverty level (percent) College education or more (percent) Less than high school education (percent) Demographic Population, foreign-born (percent) Population, white non-hispanic (percent) Isolation index for all other Urban block groups have 640 or more housing units per square mile. Suburban block groups have 64 to 639 housing units per square mile. Exurban and rural block groups have fewer than 64 housing units per square mile. 2 For all racial and ethnic groups other than non-hispanic white persons compared with non-hispanic white persons alone group. The index ranges from 0 (fully segregated) through 1 (fully integrated). NOTES: NCHS is National Center for Health Statistics; metro is metropolitan. Socioeconomic: The median percentage commuting to another county for work was 67% lower and percentage of owner-occupied housing units was almost 33% lower in large central counties than in large fringe counties. The median jobs-to-workers ratio is considerably higher in large central counties than in fringe counties. Despite this, the median unemployment rate is higher and the median household income is lower in central counties, though the central and fringe county distributions overlap considerably. Additionally, the median percentage of families with income below the poverty level was 58% higher in large central counties compared with fringe counties. Demographic: Large central counties tend to be much more racially and ethnically diverse than large fringe counties, as shown by comparing the percentage of the population that is non-hispanic white and the percentage that is foreign-born. The isolation index for all groups other than non-hispanic white persons (compared with non-hispanic white persons alone) is substantially higher in large central counties than in large fringe counties, indicating that the probability that someone who is not non-hispanic white will meet someone else who is non-hispanic white in their census tract is higher in central counties than in fringe counties (9,10). A list of all counties in large metro areas for the 2013 NCHS scheme is shown in Table IV, which is organized by MSA. The assignments of large metro counties to the large central and large fringe categories in the 2013 and 2006 NCHS schemes also are shown in this table.