Robert Hest
MPP , Senior Research Fellow

p 612-624-5507
e hestx005@umn.edu

Changing Population Estimates: Implications for Data Users

.

The National Center for Health Statistics (NCHS) at the Centers for Disease Control (CDC) recently changed the population estimates they use to calculate population-level rates in their National Vital Statistics System (NVSS), such as birth rates and mortality rates that are commonly accessed through tools such as CDC WONDER. These population estimates allow for the production of rates by providing a population denominator against which to compare the number of births or deaths.

Beginning with the 2021 data year, CDC transitioned from using "bridged-race" population estimates based on the race and ethnicity categories specified in the 1977 OMB Race and Ethnic Standards for Federal Statistics and Administrative Reporting to using "single-race" population estimates from the U.S. Census Bureau (“Census”) based on the race and Hispanic ethnicity categories specified in the 1997 OMB standards for the collection of data on race and ethnicity. The bridged-race estimates had been created to better align the population estimates from Census that used the 1997 OMB categories with the birth and death records from vital statistics systems that in many cases still used the 1977 OMB categories. This transition to the single-race population estimates appears to be precipitated by Census' recent improvement to their collection of race and ethnicity data beginning with Census 2020 data collection and continuing with subsequent years of the American Community Survey.

This move to single-race population estimates has several implications for data users:

• Statistics based on the single-race population estimates are only made available via CDC WONDER for data years 2018 forward, and statistics based on bridged-race population estimates are only available through data year 2020. Data users should not make comparisons between statistics based on bridged-race population estimates and statistics based on single-race population estimates. Further, because of changes made in 2020 to how Census measures race and ethnicity, caution should be used when comparing statistics by race and ethnicity from 2020 and later to data from 2019 and earlier.
   
• CDC WONDER has expanded the number of race and ethnicity categories available to data users. Previously, only five race categories and two Hispanic ethnicity categories were available. This has been expanded to up to 31 race categories and two Hispanic ethnicity categories.
   
• The single-race population estimates produced by Census suppress the number of persons less than five years of age at the county level to comply with Census' privacy policies. This prevents users from calculating age-adjusted rates at the county level or by level of urbanization using the single-race data, though crude rates can still be calculated at these levels of geography.

This change also affects SHADAC’s Suicide Deaths measure on State Health Compare. That measure’s Race and Ethnicity breakdown is now available in separate series for data years 1999–2020 and 2018–2021; estimates should not be compared between these series. Furthermore, SHADAC is unable to update the Metropolitan Status breakdown for this measure past data year 2020 due to the previously explained county-level suppressions.

Going forward, data users should keep changes in the denominator in mind when comparing annual estimates based on CDC vital statistics data. SHADAC strives to account for these types of considerations on State Health Compare by making clear when data years aren’t comparable and providing data users with relevant context around changes in methodology and data collection.