Technical Notes for Subcounty Areas

The New York State (NYS) Environmental Public Health Tracking (EPHT) Program created sub-county areas to display health outcome measures at finer geographic resolution than counties in a way that retains rate stability and protects confidentiality.

Visualization of rates by sub-county areas is intended to provide transparency about the burden of health outcomes among New York State residents. There are many reasons that a particular area could have a higher rate of certain health outcomes than other areas, including differences in the distribution of demographic factors, access to health care, modifiable risk factors (e.g., smoking, diet), and environmental factors (including those which may occur inside and outside the home). It is not possible to determine what factors may be contributing to health outcomes in a particular area simply by viewing the maps and charts that are provided. Additionally, the rates reported in sub-county maps may be influenced by access to health care services across state borders and errors in available population estimates.

Creation of Aggregated Areas

The Geographical Aggregation Tool was developed and used to automate the process of creating aggregated areas at the finest geographic resolution practicable. EPHT 7.5k areas were created by aggregating census tracts to meet a minimum population threshold of 7,500, using US Census Bureau boundary and population files. While many urban census tracts may have sufficient population to provide stable rates while maintaining confidentiality, many rural census tracts do not and therefore need to be aggregated. The EPHT 7.5k areas were developed as follows:

Areas of New York State outside of New York City

Tracts with zero population and tracts with 50% or greater of the population living in group quarters (mostly colleges, prisons, and nursing homes) were excluded from the aggregation.

  • Tracts with zero population were excluded because they are not expected to have populations that would contribute to health outcomes. Including them in nearby aggregated areas would misrepresent the geographic area where people are living and experiencing health outcomes.
  • Tracts with 50% or greater of the population living in group quarters were excluded because group quarters represent very specific populations that may distort health outcomes. For example, tracts dominated by men’s prisons would have unreliable fertility rates due to few women and tracts dominated by colleges would have unreliable heart attack rates due to few people over age 35. Rates are not provided separately for tracts with group quarters because the populations are too small to calculate stable rates and protect confidentiality.

Remaining tracts were pre-processed to assign tracts to Minor Civil Divisions (MCDs).

  • Using the ArcGIS Pro spatial join tool, 2010 census tracts were overlaid on 2010 MCDs to derive the proportions of each census tract’s population that fell within each MCD.
  • Tracts located within one MCD were assigned to that MCD (or vice-versa).
  • Where a tract overlapped multiple MCDs, the MCD was assigned based on the largest overlap.
  • Using the Geographic Aggregation Tool, EPHT 7.5k areas were created to nest within MCDs (i.e., cities, towns, and villages) as much as possible. Aggregation sought to maintain MCD boundaries as much as possible under the assumption that users are more likely to associate with a town or village than they are a census tract.
  • The minimum population for EPHT 7.5k areas was set at 7,500 to provide rate stability, and the ceiling was set at 22,500 (3 times the minimum population) to facilitate creation of areas with relatively comparable population sizes.

To create EPHT 7.5k areas for NYS EPHT indicators, census tracts were first separated into two datasets: one for MCDs with populations of at least 15,000, and the other for MCDs with populations less than 15,000. The separation of census tracts based on MCD population prior to aggregation allowed for the creation of areas across MCDs while preventing the aggregation of more urban areas with more rural ones. Each dataset underwent the following aggregation steps.

  • Aggregate Within MCDs: In the first aggregation, census tracts were aggregated to an adjacent area within the same MCD based on the minimum population. This produced two groups: aggregated areas with populations of at least 7,500 and aggregated areas with populations below 7,500.
  • Aggregate Within County: The results of aggregation within MCDs were aggregated to adjacent areas by closest geographic centroid within the same county. This produced aggregated areas within MCDs/counties with populations of at least 7,500. Aggregated areas that still had populations below 7,500 after this step are specified in the section below and were handled through manual aggregation or left as-is.

Areas still not meeting the minimum population threshold of 7,500 were handled as follows:

  • Hamilton County: Total county population size is below 7,500, so no EPHT 7.5k areas were created. Hamilton County itself ranks as the aggregated area with the smallest population.
  • The Town of Shelby (Orleans County): Shelby is surrounded by an aggregated group with a population of just over 20,000. Shelby’s population of just over 5,000 is too great to aggregate in, so it has a population lower than 7,500 and cannot be aggregated within its county.
  • The Town/Village of East Rochester (Monroe County): East Rochester has a population of approximately 7,000. It was not aggregated into the surrounding areas because the neighboring MCDs are substantially larger in area and have over three times the total population of East Rochester.
  • Remaining 2 areas (Tuscarora Nation, Niagara County and Shelter Island Town, Suffolk County): These were manually aggregated to either the closest area or most similar area based on either demographics or land use.

New York City

  • Pre-defined Neighborhood Tabulation Areas (NTAs), developed by the NYC Department of City Planning, were used.
  • NTAs are based 2010 census tracts and have a minimum population of about 15,000. Although the minimum population for these tracts is double the minimum population for non-NYC aggregated areas, the geographic size of these areas tends to be smaller due to high population density in NYC.
  • To be consistent with the rest of the state, group quarters, zero-population tracts, and joint interest areas were removed from aggregation. Joint interest areas are defined by the NYC Office of Planning as areas where people are not expected to live, such as parks, government installations, and airports.
  • The remaining census tracts were aggregated to their pre-defined NTAs.

Data Dictionary

Variable Type Description Possible Values
FID object ID Unique ID Any positive numeric value
EPHT7p5kId string Unique ID created by NYS EPHT to identify the EPHT 7.5k area A string in the format A000000000
EPHT_mcds string MCDs comprising aggregated area Any valid town name(s) separated by commas
tract_ids1 string 2010 census tracts within aggregated area Any valid 2010 census tract(s) code(s) separated by commas
tract_ids2 string 2010 census tracts within aggregated area Any valid 2010 census tract(s) code(s) separated by commas

Notes:

1. To manage "tract_ids1" that exceeds 254 characters, the tract_ids are divided into two columns: tract_ids1 and tract_ids2.

2. EPHT 7.5k IDs are created by the EPHT program to label aggregated areas.

  • Each EPHT 7.5k ID begins with a letter defined as follows:“C” refers to areas made up of less densely populated cities and towns (individual populations below 15,000) aggregated within county borders, “M” refers to areas aggregated within the borders of more populous cities/towns with populations of at least 15,000, “N” refers to New York City defined Neighborhood Tabulation Areas (NTA), “H” refers to Hamilton County (whose population is too low to allow for EPHT 7.5k areas), “G” refers to census tracts with over 50% of their population living in group quarters, “Z” refers to census tracts with zero population, “P” refers to census tracts with very unreliable populations, typically those that are largely parks or ‘joint interest’ areas.
  • The three digits after the letter are the county FIPS code.
  • For “G”, “Z”, and “P” areas, the remaining six digits are the census tract FIPS code.
  • For “N” areas, the last four digits are from the NTA IDs.
  • For the remaining areas, the last six digits are values assigned in the aggregation process.

Resources

  • 2010 Census Tract Population and Demographics: 2010 Census Summary File 1 (New York State) prepared by the U.S. Census Bureau, 2011.
  • 2010 Census Tract boundaries: 2010 TIGER/Line Shapefiles (machine-readable data files) / prepared by the U.S. Census Bureau, 2011.
  • 2010 Minor Civil Division Areas: 2016 TIGER/Line Shapefiles (machine-readable data files) / prepared by the U.S. Census Bureau, 2016.
  • New York City Neighborhood Tabulation Areas
  • Gwen LaSelva D. Babcock (2017). Assign Zone Tool. MapInfo. Bureau of Environmental and Occupational Epidemiology, Environmental Health Surveillance Section, New York State Department of Health.
  • Abigail Stamm and Gwen LaSelva D. Babcock (2021). gatpkg: Geographic Aggregation Tool (GAT). Rpackage version 1.61.0. https://github.com/ajstamm/gatpkg
  • Thomas Talbot and Gwen LaSelva D. Babcock (2013). GIS Tools for Sharing Health Data and Protecting Patient Confidentiality. NYGeoCon November 12-13, 2013, Saratoga Springs, NY https://www.nysgis.net/Docs/NYGeoCon2013/Slides/Wed_0245_0415/Sharing4_HealthData.pdf