Methodology
The methodology described here is only for the small area estimation (SAE) approach that is used to generate estimates using the Behavioral Risk Factor Surveillance System (BRFSS) data, the U.S. Census Bureau’s American Community Survey (ACS) data, and decennial population counts or annual county population estimates. The SDOH measures were derived directly from ACS data and are not included in this section. See SDOH measure definitions for details.
- PLACES uses a multilevel regression and poststratification (MRP) method to generate estimates of each measure at the county, place (incorporated and census designated), ZIP Code Tabulation Area (ZCTA), and census tract levels for adults ≥18 years in the US.
- A multilevel logistic regression model is constructed for each measure. It includes some or all of the following variables based on model performance and final prediction: individual-level age, sex, race/ethnicity, and education level from CDC’s Behavioral Risk Factor Surveillance System (BRFSS); county-level percentage of adults below 150% of the federal poverty level from the 5-year American Community Survey (ACS); and state- and county-level random effects. The model is applied to annual county-level census population estimates to compute a predicted probability of having each outcome. The county-level estimates are obtained by multiplying the probability by the total adult population of each county.
- The model is applied to decennial 2020 (2010 for releases 2023 and before) census block-level population counts to compute a predicted probability as well. The estimated prevalence can be obtained by multiplying the probability by the total adult population for each block, which can be aggregated to place, census tract, and ZCTA levels.
- Monte Carlo simulation is used to generate 1,000 simulated datasets for the point estimate, the final estimates are reported as the mean and 95% confidence interval (the 2.5th, 97.5th percentiles) over 1,000 draws.
- The MRP approach is flexible by modeling nationally and predicting locally and can be used to provide modeled estimates at any geography above the census blocks (the smallest census geography).
- Both internal and external validation studies showed strong/moderate correlations between model-based estimates and direct survey estimates at state, county, and place levels.
- The primary data sources for PLACES are CDC’s BRFSS, decennial census 2020/2010 population counts, annual (intercensal) county-level census population estimates, and 5-year ACS data.
Further information on the PLACES approach and small area estimation methodology can be obtained from:
- Multilevel Regression and Postratification for Small-Area Estimation of Population Health Outcomes: A Case Study of Chronic Obstructive Pulmonary Disease Prevalence Using the Behavioral Risk Factor Surveillance System.[PDF-5.53MB]
- Validation of Multilevel Regression and Postratification Methodology for Small Area Estimation of Health Indicators From the Behavioral Risk Factor Surveillance System.
- Comparison of Methods for Estimating Prevalence of Chronic Diseases and Health Behaviors for Small Geographic Areas: Boston Validation Study, 2013
- Using 3 Health Surveys to Compare Multilevel Models for Small Area Estimation for Chronic Diseases and Health Behaviors
- PLACES: Local Data for Better Health
- Constructing Statistical Intervals for Small Area Estimates Based on Generalized Linear Mixed Model in Health Surveys
Measures
- The 40 measures in the 2024 release (36 in the 2023 release, 29 in the 2022 and 2021 releases; 27 in the 2020 release) include 12 health outcomes, 7 prevention practices, 4 health risk behaviors, 7 disability measures, 3 health status measures, and 7 health-related social needs (the last category only for 39 states and District of Columbia).
- The measures include major risk behaviors that lead to illness, suffering, and early death related to chronic diseases and conditions, as well as the conditions and diseases that are the most common, costly, and preventable of all health problems.
- Each measure has a comprehensive definition that includes the background, significance, limitations of the indicator, data source, and limitations of the data resources.
- Measures complement existing sets of surveillance indicators that report state, metropolitan area, and county data, including County Health Rankings and Chronic Disease Indicators.
List of Measures