Guidelines For Using Rural-Urban Classification Systems for Public Health Assessment

Revision Date: February 5, 2009
Primary Contact: Zeynep Shorter, Ph.D., M.P.H., Rural Health Epidemiologist
Secondary Contact: Juliet VanEenwyk, Ph.D., State Epidemiologist for Non-Infectious Conditions

Purpose
What is new, and how does this affect public health assessment?
Why a guideline on rural-urban classification systems?
How rural is Washington State?
What systems are commonly used to classify rural-urban character?
County, Census Tract and ZIP Code definitions
Choosing the right classification system
The Rural-Urban Commuting Area (RUCA) system: a good choice
          A four-tier consolidation of the RUCA system at the sub-county level
Washington State classifications by county
Trend analysis
Other considerations when making rural-urban comparisons
Guidelines: A recap
List of Acronyms
References
Appendix 1: Comparison of public health indicators using 2001 and 2008 four-tiered consolidations of RUCA codes
Appendix 2: RUCA code definitions
Appendix 3: Comparison of the 1990 and 2000 census tract boundaries

View this document as PDF (PDF, 3.35 MB)

Purpose

The Assessment Operations Group in the Washington State Department of Health (DOH) coordinates the development of guidelines related to data development and use to promote good professional practice among staff involved in assessment activities within DOH and in Local Health Jurisdictions in Washington. While the guidelines are intended for an audience of differing levels of training related to data development and use, they assume a basic knowledge of epidemiology and biostatistics. They are not intended to recreate basic texts and other sources of information, but rather they focus on issues commonly encountered in public health practice and where applicable refer to issues unique to Washington State.

What is new, and how does this affect public health assessment?

The 2008 guidelines for rural-urban classification systems have two major changes from the 2001 version of these guidelines.

• The DOH Community Health Systems Office no longer recommends using the Dominant Rural-Urban Commuting Area (RUCA) classification system for Washington State counties. In 2003, several other county-based systems with methodologies compatible with those used to develop the Dominant RUCA Codes became available nationally (Table 4). As described below, DOH recommends using the Metropolitan, Micropolitan and Outside Core-Based Statistical Area classifications used by the US Office of Management and Budget (OMB) for county-level classification.
   
• As described below, the current guidelines adopt a four-tiered consolidation of RUCA codes derived from a seven-tiered consolidation by the Washington, Wyoming, Alaska, Montana, and Idaho (WWAMI) Rural Health Research Center. This four-tiered consolidation differs from that recommended in the 2001 guideline. This change transfers two census tracts in Chelan and Whatcom Counties from Small Town and Isolated Rural to Sub-Urban and one census tract in Whitman County from Large Rural Town to Small Town and Isolated Rural. Overall, the change affects less than 1% (2000 Census data) of Washington residents. Results from analysis of health indicators with old and new four-tiered RUCA consolidations did not show significant changes. (See Appendix 1)

Why a guideline on rural-urban classification systems?

The DOH Community Health Systems Office has documented differences in health status and related risk and protective factors between residents of rural and urban Washington. Analysts looking at rural health disparities must choose from several classification systems which change over time. Guidelines are useful for promoting consistency, comparability and best practice among statewide analyses that look at rural health. Local public health assessments and measurement of performance standards also benefit from consistent classification systems to compare local health data to areas with similar population and settlement patterns.

These guidelines do not cover the use of rural-urban classification schemes for determining eligibility for state or federal assistance programs. See How Many Agencies does it Take to Define Rural? for a summary of eligibility criteria for rurally-targeted programs. Also, see Washington State Office of Financial Management for references to population density in Washington law.

How rural is Washington State?

In 2008, between 13% and 27% of Washington’s residents lived in areas classified as rural depending on the classification method used. For example, nine Washington counties are classified as rural using the Washington State Office of Financial Management (OFM) designations and as metropolitan using nationally developed systems. This underscores that there is no gold standard for the defining features of rural areas or a single best theoretical basis for classifying rural areas. Some classification systems measure rurality on the basis of population density, others by economic or commuting connections. The choice of the geographic unit (county, census tract, or ZIP code) for analysis introduces further variation.

Regardless of the classification system, Washington is a predominantly urban state and is becoming more urbanized over time. Analysts comparing health indicators in rural and urban areas should be aware that rates may be worse in rural compared to urban areas, but with very few exceptions, the total numbers affected are much higher in urban areas.

What systems are commonly used to classify rural-urban character?

The first rural-urban classification system, developed in 1874 by the US Census Bureau, defined rural as the population of a county living outside of cities or towns with 8,000 or more inhabitants. That population threshold was changed to 2,500 in 1910. The US Census Bureau now defines ‘urban’ as all territory, population, and housing units in Urbanized Areas and Urban Clusters with 2,500 or more persons. This definition is based on a very fine level of geography, the census block group. Urbanized areas and urban clusters are contiguous built-up areas with a population density of more than 1,000 persons per square mile. Currently, an Urbanized Area is a contiguous set of block groups with a population of more than 50,000. An Urban Cluster is a contiguous set of block groups with populations of between 2,500 and 49,999. The remaining areas are Rural Areas. Most rural-urban classification systems are based on these definitions. The US Census Bureau’s most recent revision of these definitions in 2002 resulted in a larger enlargement of Urbanized Areas than would have happened by population increase alone. Figure 1 shows how the boundaries of Urbanized Areas changed in Washington.

Details about the methodological changes in 2002 and a list of urbanized areas are available by the Census Bureau. 
 
Early systems tended to be binary in nature, classifying counties as urban or rural. These binary systems gave way to more complex county coding systems such as Rural-Urban Continuum Codes (RUCC) that recognize differences in size and the importance of issues like economic influence. With advances in computing power and geographic information systems in the 1990s, sub-county classifications systems, most notably RUCA system, were developed. For more information see Ricketts et al. (1998) and the USDA Economic Research Service: Measuring Rurality Briefing Room.

Table 1: Commonly Used Rural-Urban Classification Systems

* Year published in the Federal Register

County, census tract and ZIP code definitions

Many classification systems (for example, Metropolitan, Micropolitan, and Outside Core-Based Statistical areas, Figure 2) use county-level geographic aggregation, based on population density and dominant economic commuting pattern.

Analysts should consider three main factors when choosing a rural-urban classification scheme.

For most analyses, the DOH Community Health Systems Office recommends the RUCA system. The US Departments of Agriculture and Health and Human Services and the WWAMI Rural Health Research Center, originally developed this system in the late 1990s. It is the only system available at the census tract or ZIP code levels. The RUCA system is currently considered the state of the art and is used widely. The RUCA system is more precise than county-based alternatives, since it uses smaller geographic units.

The RUCA system is a ten-tiered classification system that uses US Census Bureau definitions of Urbanized Areas and Urban Clusters and commuting relationships at the census tract level as the basic building blocks. The primary (largest) and secondary (second largest) commuting flows to core areas are identified using the most recently available commuting data, the 2000 US Census for the latest version of RUCA. Strongly linked tracts are those where the primary commuting flow to a core area is greater than 30% of commuting trips. Weakly linked tracts are those where the largest flow to a core area is 10-30%. Isolated rural areas are those with no town greater than 2,500 where the primary commuting flow is local. This yields the following scheme:

Table 2: Rural Urban Commuting Area (RUCA) Classification System: Ten Primary Tiers 

The ten codes are sub-divided into 33 secondary codes which can be consolidated or combined in several different ways. See Appendix 2 for a listing of the secondary codes.

Washington State RUCA codes using census tract geography are mapped in Figure 3.

A 2004 ZIP code approximation of the 2000 RUCA codes developed by the WWAMI Rural Health Research Center is mapped in Figure 4. The ZIP code approximation is less accurate than the census tract version. ZIP codes do not uniformly correspond with census blocks as do the census tracts. A listing of 2006 ZIP code approximations of the RUCA codes is available at DOH Health Care Access Page.

More information on the RUCA system is available at the WWAMI Rural Health Research Center.

A four-tier consolidation of the RUCA system at the sub-county level

Many datasets do not have sufficient sample size to support analysis using a ten-tiered classification system. For general descriptive analyses where sub-county data are available, we suggest a four-tiered consolidation based on secondary codes that identify the general character of an area.

• Urban Core: contiguous built up areas of 50,000 persons or more. These areas correspond to US Census Bureau's Urbanized Areas.
   
• Sub-Urban: areas, often in Metropolitan Counties, with high commuting flows to Urban Cores (for example, Eatonville in Pierce County). These areas also include all other areas where 30%-49% of the population commutes to Urban Cores for work.
   
• Large Rural Town: towns with populations between 10,000 and 49,999 and surrounding rural areas with 10% or more primary commuting flows to these towns, as well as secondary commuting flows of 10% or more to Urban Cores.
   
• Small Town and Isolated Rural Areas: towns with populations below 10,000 and their surrounding commuter areas and other isolated rural areas with more than one hour driving distance to a nearest city.

Table 3 provides the four-level consolidation recommended by the DOH Community Health Systems Office. This consolidation is based on the WWAMI Rural Health Research Center’s seven-tiered classification system. WWAMI’s own four-tiered classification collapses all of urban into one urban category, and splits rural among three additional categories. (See http://depts.washington.edu/uwruca/maps.html.) We recommend the consolidation in Table 3, because large proportions of Washington residents live in urban and sub-urban areas, making analyses with these designations informative. This Guideline defines the ‘urban core’ and ‘other urban’ found in the seven-tiered consolidation as Urban Core and Sub-Urban, respectively. Large Rural Town combines ‘large rural core’ and ‘other large rural’ from the seven-tiered classification. Small Town and Isolated Rural Areas comprise the remaining three rural categories from WWAMI’s seven-tiered classification: small rural core, other small rural, and isolated rural.

Table 3: Four-Tier Consolidation of Secondary RUCA Codes

Class	Tier	Secondary RUCA Codes
Urban Core	1	1.0, 1.1
Sub-Urban	2	2.0, 2.1, 3.0, 4.1, 5.1, 7.1, 8.1, 10.1
Large Rural Town	3	4.0, 4.2, 5.0, 5.2, 6.0, 6.1
Small Town/Isolated Rural	4	7.0, 7.2, 7.3, 7.4, 8.0, 8.2, 8.3, 8.4, 9.0, 9.1, 9.2, 10.0, 10.2, 10.3, 10.4, 10.5, 10.6

As with the full set of RUCA codes, the census tract version of the four-tiered consolidation (Figure 5) allows more precise classification than a similar consolidation done with the ZIP code RUCA approximations (Figure 6), since the ZIP code boundaries do not always coincide with census tract or county boundaries. But the ZIP code approximation of RUCA codes can be used with several public health datasets that use ZIP code as the smallest geographic level. Data analysts need to be aware that ZIP codes change over time. These changes can result in misclassification.

 

Washington State classifications by county

Table 4 shows classifications of Washington’s counties based on the county-level classification systems described below. The DOH Community Health Systems Office recommends using Metropolitan, Micropolitan, and Outside Core-Based Statistical Area classification used by the OMB, because this is the most widely used classification system nationally. It is more precise than two-level classification systems while having the advantage of a small number of categories allowing collapsing over sufficient numbers for analysis.

Metropolitan, Micropolitan and Outside Core-Based Statistical Areas: The OMB has used this national classification system since the 1940s for statistical reporting and allocating funds. Until recently, this system classified counties as Metropolitan or Non-Metropolitan. In 2003, the US Census Bureau that develops the system, divided Non-Metropolitan counties into Micropolitan and Outside Core-Based-Statistical Areas. At the same time, the US Census Bureau increased the commuting threshold for tying in outlying areas from 15% to 25%. Currently, counties with cities or urbanized areas over 50,000 are classified as Metropolitan. Counties with urban clusters of 10,000 to 49,999 persons are classified as Micropolitan. In addition, any county in which at least 50% of the population resides in an Urbanized Area are designated as Metropolitan or Micropolitan. Outlying counties meeting a complex set of conditions based on commuting patterns and population density are also designated either as Metropolitan or Micropolitan. All other areas are designated Outside a Core-Based Statistical Area. The US Census Bureau maintains the most recent national list and technical documentation. Figure 2 shows a map of Washington counties classified by this system.

Rural-Urban Continuum Codes (RUCC): The US Department of Agriculture (USDA) developed the RUCC system, also known as the Beale code system, in the mid-1970s. It was a forerunner of the Urban Influence Codes and present RUCA system. The system uses Metropolitan, Micropolitan and Outside Core-Based Statistical Area classifications as a starting point. Metropolitan counties are classified into three population categories. Non-Metropolitan counties are classified into six categories based on total population in US Census Bureau’s Urbanized Areas. Non-Metropolitan communities are further classified by adjacency to Metropolitan counties and commuting patterns. This system better differentiates between central and fringe metropolitan areas than the OMB’s three-level system. The most recent update was in 2003. For more information and to download codes see http://www.ers.usda.gov/Data/RuralUrbanContinuumCodes/.

Urban Influence Codes: The USDA developed a 12-level classification scheme in the mid-1990s to emphasize the tendency of economic systems to centralize around very large metropolitan counties. Metropolitan counties are classified as Large Metropolitan (population of at least one million) or Small Metropolitan (population less than one million). Micropolitan counties are classified by adjacency to Large or Small Metropolitan counties. Counties in Outside a Core-Based Statistical Area are classified by their adjacency to Metropolitan and Micropolitan counties and whether they contain a town of 2,500. This system was most recently updated in 2003. For more information and to download codes see http://www.ers.usda.gov/Briefing/Rurality/.

Other Variations: Several other classification systems that apply to subsets of areas, activities, or populations may be useful for rural public health assessment. In 1999, the Washington State Legislature passed a law defining ‘rural distressed’ counties as those with population densities of fewer than one hundred persons per square mile. Other classification systems include multiple definitions of frontier areas and county-based typologies of primary economic activity. For additional information related to county classification, see http://www.ers.usda.gov/Briefing/Rurality/.

Table 4: Commonly Used Rural Urban Classifications for Washington Counties

County	2003 Metropolitan, Micropolitan, Outside Core-Based Statistical Area (OMB)	2003 Rural Urban Continuum Codes (USDA)	2003 Urban Influence Codes (USDA)	2008 Rural Distressed (OFM)	2008 Population (OFM)
Adams	Outside CBSA	Small Rural Adjacent to Metro	Noncore adjacent to a small metro with town of at least 2,500 residents	Rural	17,800
Asotin	Metropolitan	Small Metropolitan	Small-in a metro area with fewer than 1 million residents	Rural	21,400
Benton	Metropolitan	Small Metropolitan	Small-in a metro area with fewer than 1 million residents	Rural	165,500
Chelan	Metropolitan	Small Metropolitan	Small-in a metro area with fewer than 1 million residents	Rural	72,100
Clallam	Micropolitan	Large Rural Not Adjacent	Micropolitan not adjacent to a metro area	Rural	69,200
Clark	Micropolitan	Large Metropolitan	Large-in a metro area with at least 1 million residents or more	Urban	424,200
Columbia	Outside CBSA	Small Rural Adjacent to Metro	Noncore adjacent to a small metro with town of at least 2,500 residents	Rural	4,100
Cowlitz	Metropolitan	Small Metropolitan	Small-in a metro area with fewer than 1 million residents	Rural	99,000
Douglas	Metropolitan	Small Metropolitan	Small-in a metro area with fewer than 1 million residents	Rural	37,000
Ferry	Outside CBSA	Isolated Rural Not Adjacent	Noncore not adjacent to a metro/micro area and does not contain a town of at least 2,500 residents	Rural	7,700
Franklin	Metropolitan	Small Metropolitan	Small-in a metro area with fewer than 1 million residents	Rural	70,200
Garfield	Outside CBSA	Isolated Rural Adjacent to Metro	Noncore adjacent to a small metro and does not contain a town of at least 2,500 residents	Rural	2,300
Grant	Micropolitan	Large Rural Adjacent to Metro	Micropolitan area adjacent to a small metro area	Rural	84,600
Grays Harbor	Micropolitan	Large Rural Adjacent to Metro	Micropolitan area adjacent to a small metro area	Rural	70,900
Island	Micropolitan	Large Rural Adjacent to Metro	Micropolitan adjacent to a large metro area	Urban	79,300
Jefferson	Outside CBSA	Small Rural Adjacent to Metro	Noncore adjacent to a small metro with town of at least 2,500 residents	Rural	28,800
King	Metropolitan	Large Metropolitan	Large-in a metro area with at least 1 million residents or more	Urban	1,884,200
Kitsap	Metropolitan	Small Metropolitan	Small-in a metro area with fewer than 1 million residents	Urban	246,800
Kittitas	Micropolitan	Small Rural Adjacent to Metro	Micropolitan adjacent to a large metro area	Rural	39,400
Klickitat	Outside CBSA	Small Rural Adjacent to Metro	Noncore adjacent to a small metro with town of at least 2,500 residents	Rural	20,100
Lewis	Micropolitan	Large Rural Adjacent to Metro	Micropolitan area adjacent to a small metro area	Rural	74,700
Lincoln	Outside CBSA	Isolated Rural Adjacent to Metro	Noncore adjacent to a small metro and does not contain a town of at least 2,500 residents	Rural	10,400
Mason	Micropolitan	Small Rural Adjacent to Metro	Micropolitan area adjacent to a small metro area	Rural	56,300
Okanogan	Outside CBSA	Small Rural Adjacent to Metro	Noncore adjacent to a small metro with town of at least 2,500 residents	Rural	40,100
Pacific	Outside CBSA	Small Rural Not Adjacent to Metro	Noncore adjacent to micro area and contains a town of at least 2,500 residents	Rural	21,800
Pend Oreille	Outside CBSA	Isolated Rural Adjacent to Metro	Noncore adjacent to a small metro and does not contain a town of at least 2,500 residents	Rural	12,800
Pierce	Metropolitan	Large Metropolitan	Large-in a metro area with at least 1 million residents or more	Urban	805,400
San Juan	Outside CBSA	Isolated Rural Not Adjacent	Noncore not adjacent to a metro/micro area and does not contain a town of at least 2,500 residents	Rural	16,100
Skagit	Metropolitan	Small Metropolitan	Small-in a metro area with fewer than 1 million residents	Rural	117,500
Skamania	Metropolitan	Large Metropolitan	Large-in a metro area with at least 1 million residents or more	Rural	10,700
Snohomish	Metropolitan	Large Metropolitan	Large-in a metro area with at least 1 million residents or more	Urban	696,600
Spokane	Micropolitan	Mid-size Metropolitan	Small-in a metro area with fewer than 1 million residents	Urban	459,000
Stevens	Outside CBSA	Small Rural Adjacent to Metro	Noncore adjacent to a small metro with town of at least 2,500 residents	Rural	43,700
Thurston	Metropolitan	Small Metropolitan	Small-in a metro area with fewer than 1 million residents	Urban	245,300
Wahkiakum	Outside CBSA	Isolated Rural Adjacent to Metro	Noncore adjacent to a small metro and does not contain a town of at least 2,500 residents	Rural	4,100
Walla Walla	Micropolitan	Large Rural Adjacent to Metro	Micropolitan area adjacent to a small metro area	Rural	58,600
Whatcom	Metropolitan	Small Metropolitan	Small-in a metro area with fewer than 1 million residents	Rural	191,000
Whitman	Micropolitan	Large Rural Adjacent to Metro	Micropolitan area adjacent to a small metro area	Rural	43,000
Yakima	Metropolitan	Small Metropolitan	Small-in a metro area with fewer than 1 million residents	Rural	235,900

Trend Analysis

Four major changes occurred between the 1990 and 2000 US Censuses which complicate attempts to track trends in rural-urban disparities.

1)  Census tract boundaries were realigned. An overlay of 1990 and 2000 Washington Census Tracts (Appendix 3) shows that the realignments are particularly noticeable in the rapidly growing areas surrounding major population centers. 

2)  As discussed in What systems are commonly used to classify rural urban character?, the US Census Bureau revised the methodology for establishing Urbanized Areas and Urban Clusters in 2002. These changes expanded the boundaries of Urbanized Areas. Figure 1 compares 1990 and 2000 Census Urbanized Areas in Washington. It is unclear how much of the change between 1990 and 2000 is the result of the new method and how much is due to population growth, because the new method has not been applied to 1990 data. Because most rural classification schemes use the Urbanized Area definition as a starting point, this change has had broad ramifications. Chief among them is that comparisons over time are difficult to interpret.

3)  As described in Washington State classifications by county, the US Census Bureau revised the methods for establishing Metropolitan and Non-Metropolitan areas in 2003. This change also affected Urban Influence Codes and RUCC which are tied to the Metropolitan definitions.

4)   The rules for classifying census tracts for RUCA codes were revised in 2005. The primary change was raising the lower threshold for establishing commuting relationships from 5% to 10%. The primary effect of this change was to increase populations and areas classified as rural.

Rural-urban trends between 1990 and 2000 are obscured due to the magnitude and complexity of these changes and the absence of bridging methods built on 1990 and 2000 US Censuses.

Recommendations for trend analysis:

• Because classification systems built on the 1990 and 2000 US Censuses are not comparable, the DOH Community Health Systems Office recommends beginning trend analysis in 1995 and using classification systems based on methods that incorporate 2000 US Census data.
   
  • The author should note that the comparison is based on classifications at a point in time (2000). Some areas, particularly those with large population change, may be misclassified or would change classification if more complete data were available. An assessment of population change within classifications across the study years might help interpret findings.
     
• If the data analyst needs to assess trends prior to 1995 using systems based on 1990 and 2000 US Censuses, we recommend that the analyst
   
  • Explore the significance of classification changes before treating the data as a continuous series.
     
  • Clearly show on trend lines or charts where major methodological changes occurred.
     
  • Interpret trends with caution.

Other considerations when making rural-urban comparisons in Washington

Because the proportion of elderly residents in rural areas is higher than in urban areas, the analyst should consider age-adjustment for age-related public health indicators. (See Rates guideline for a discussion of age-adjustment.) Analysts should also keep in mind that, in general, the residents of rural Washington have lower incomes, have completed fewer years of formal education, and have differing racial and ethnic backgrounds than those in other areas. Differences in health status between rural and urban Washingtonians may reflect these underlying differences in demographics.

There are also regional variations in the demographic structure of rural Washington. The Hispanic population has a strong presence in Central Washington, and the tribal population has a strong presence in Northeast Washington. Northwest Washington (San Juan and Island Counties) are more Caucasian and affluent. 

Walla Walla, Whitman, and Kittitas Counties host universities which have significant influences on both age and poverty structure. Island County has a very large military presence. Sensitivity analyses excluding these “outlier” counties may be warranted in analyses examining health indicators or population demographics that may be inordinately influenced by these populations.

Guidelines: A recap

• When making rural-urban comparisons definitions matter. It is essential that analysts document the classification system, including explaining why the system was selected, and discussing its strengths, limitations and possible biases. This information needs to be easily accessible to users to help them interpret findings and compare across studies.
   
• If data are available at the census tract or ZIP code level, use the RUCA system, collapsing the ten RUCA codes into four categories as in Table 3.
   
• If data are only available at the county level, use Metropolitan, Micropolitan and Outside Core-Based Statistical Area classifications.
   
• Changes over time may be best assessed using 2000 rural-urban designations and 1995 as a starting point.
   
• Trend analyses using classification systems built on the 1990 and 2000 US Censuses are not recommended because of methodological changes and the absence of bridging data between census years. If long term comparisons are attempted,
   
  • County-level classifications are more stable than those based on smaller geographies.
     
  • Bias due to changes in methods and geographic boundaries needs to be discussed.
     
• Rural-urban health differences may reflect underlying differences in demographics. In general, rural-urban comparisons of public health indicators should be age-adjusted, as the proportion of elderly residents in rural areas is higher than in urban areas. Analysts should also keep in mind that, in general, the residents of rural Washington have lower incomes, have completed fewer years of formal education, and have differing racial and ethnic backgrounds than those in other areas.

Appendix 2

Appendix 3