Written and maintained by CASRAI Editorial Board
Published · Last updated
On this page
Definition in one paragraph
CDC PLACES (Population Level Analysis and Community Estimates) is an open-access public health data initiative developed by the Centers for Disease Control and Prevention (CDC) Division of Population Health in partnership with the Robert Wood Johnson Foundation and the CDC Foundation. The project generates model-based small-area estimates (SAEs) for 40 standardized chronic disease measures across four geographic tiers covering the United States: all 3,143 counties, ~28,000 incorporated places and census-designated places, ~84,000 census tracts, and ~32,000 ZIP Code Tabulation Areas (ZCTAs). By linking individual-level Behavioral Risk Factor Surveillance System (BRFSS) survey microdata with U.S. Census Bureau population distributions and American Community Survey (ACS) contextual poverty metrics through Multilevel Regression and Poststratification (MRP), CDC PLACES delivers hyper-local prevalence estimates where direct survey sample sizes are unavailable.
Formula with every symbol explained
The CDC PLACES estimation framework uses the two-stage Multilevel Regression and Poststratification (MRP) methodology developed by Zhang et al. (2014, 2015). Stage 1 models individual survey outcomes via multilevel logistic regression, and Stage 2 weights predicted cell probabilities by Census population counts.
logit(pijcs) = log(pijcs / (1 − pijcs)) = β0 + Xjβ + Wcγ + uc + vs
- pijcs: Probability of the health condition for individual i in demographic cell j, county c, and state s.
- β0: Baseline fixed intercept representing the reference demographic group log-odds.
- Xj: Vector of demographic indicators (age group, sex, race/ethnicity, and education) from BRFSS.
- β: Vector of demographic fixed-effect regression coefficients.
- Wc: County-level contextual covariate (% of county residents below 150% of the federal poverty level from ACS).
- γ: Contextual poverty fixed-effect coefficient.
- uc: County random intercept, where uc ~ N(0, σcounty2).
- vs: State random intercept, where vs ~ N(0, σstate2).
In Stage 2, the small-area prevalence estimate P̂A for target area A is aggregated using inverse-logit cell predictions (p̂jcs) and Census population counts (NA,j):
P̂A = [∑j=1J (NA,j × p̂jcs)] / [∑j=1J NA,j]
Where p̂jcs = 1 / [1 + exp(−(β̂0 + Xjβ̂ + Wcγ̂ + ûc + v̂s))], and ∑j=1J NA,j = NA represents total adult population in area A.
Worked example with real numbers and the arithmetic shown
Consider estimating adult asthma prevalence in a census tract of N = 4,000 adults classified across four demographic strata. Assume model parameters: baseline intercept β0 = −2.40, older age (45+) βage = +0.25, no college degree βeduc = +0.40, county poverty term Wcγ = 0.18 × 0.50 = +0.09, county effect ûc = −0.05, and state effect v̂s = +0.06. The baseline linear predictor is ηbase = −2.40 + 0.09 − 0.05 + 0.06 = −2.30.
- Stratum 1 (Age 18–44, College; N1 = 1,200):
η1 = −2.30; p̂1 = 1 / [1 + exp(2.30)] = 0.09112 (9.11%)
Expected cases = 1,200 × 0.09112 = 109.34 - Stratum 2 (Age 18–44, No College; N2 = 1,000):
η2 = −2.30 + 0.40 = −1.90; p̂2 = 1 / [1 + exp(1.90)] = 0.13011 (13.01%)
Expected cases = 1,000 × 0.13011 = 130.11 - Stratum 3 (Age 45+, College; N3 = 1,000):
η3 = −2.30 + 0.25 = −2.05; p̂3 = 1 / [1 + exp(2.05)] = 0.11405 (11.41%)
Expected cases = 1,000 × 0.11405 = 114.05 - Stratum 4 (Age 45+, No College; N4 = 800):
η4 = −2.30 + 0.25 + 0.40 = −1.65; p̂4 = 1 / [1 + exp(1.65)] = 0.16111 (16.11%)
Expected cases = 800 × 0.16111 = 128.89
Summing expected cases yields: 109.34 + 130.11 + 114.05 + 128.89 = 482.39 cases.
Dividing by the total tract population yields the final point estimate:
P̂tract = 482.39 / 4,000 = 0.1206, or 12.1% (95% CI [10.8%, 13.5%]).
Interpretation
Interpreting CDC PLACES data requires distinguishing between crude prevalence and age-adjusted prevalence. Crude rates measure the absolute disease burden within a community, making them essential for local healthcare budgeting, staffing, and clinical facility planning. Age-adjusted rates remove demographic age confounding to permit unbiased comparisons across communities with divergent age distributions. Researchers frequently evaluate PLACES metrics alongside the County Health Rankings and the CDC Social Vulnerability Index (SVI) in social epidemiology investigations.
| Geographic Level | Units | Preferred Metric | Primary Use Case | Reading Guidance |
|---|---|---|---|---|
| County | 3,143 | Crude / Age-Adjusted | County health department planning | Broad surveillance; obscures intra-county variation |
| Place (City / CDP) | ~28,000+ | Crude (needs); Age-Adjusted (benchmarks) | Municipal policy and community health | Aligns with city administrative boundaries |
| Census Tract | ~84,000+ | Crude (local burden); Age-Adjusted (disparities) | Hyper-local intervention targeting | Model-based synthetic estimate; verify overlapping CIs |
| ZCTA | ~32,000+ | Crude | Hospital community benefit reporting | Facilitates linkage with patient postal addresses |
Visual analysis using the CDC PLACES Interactive Map or GIS layers allows health geographers to map spatial autocorrelation and identify high-burden neighborhood clusters while evaluating uncertainty through published 95% confidence intervals.
Assumptions and common mistakes
Applying CDC PLACES estimates appropriately requires recognizing core statistical assumptions and avoiding analytical errors:
- Transportability of Demographic Risk: MRP assumes that individual demographic risk profiles estimated from state BRFSS samples generalize across census tracts within that state, conditioned on county poverty and random intercepts.
- The Intervention Evaluation Trap: The CDC specifically cautions that PLACES estimates cannot evaluate the success of local programs or municipal health policies. Because the MRP model predicts outcomes from state BRFSS relationships and Census demographics, a localized clinical or policy intervention will not be reflected in tract-level estimates unless underlying Census demographics or state survey coefficients shift.
- The Ecological Fallacy: Applying an aggregate tract prevalence rate to an individual patient residing in that area. PLACES characterizes neighborhood population burden, not individual diagnostic status.
- Conflating Crude and Age-Adjusted Metrics: Comparing crude rates between communities with disparate age distributions (such as university districts versus retirement enclaves) produces misleading conclusions regarding social disparities.
- Longitudinal Comparability Boundaries: Comparing PLACES releases across different years as continuous time-series can be invalid due to rotating BRFSS survey modules, question revisions, and updated Census decennial baselines. Review sample size determination formulas when designing local survey validations.
How to report it (APA example)
Scholarly reporting of CDC PLACES data follows APA 7th edition standards, specifying the release edition, data sources, geographic scale, metric type, and confidence intervals.
Centers for Disease Control and Prevention. (2025). PLACES: Local Data for Better Health, County Data, 2025 release [Data set]. U.S. Department of Health and Human Services. https://data.cdc.gov/api/views/swc5-untb.json
An illustrative manuscript narrative reporting passage:
“Census tract health estimates were obtained from the CDC PLACES 2025 release, which applies multilevel regression and poststratification (MRP) to 2023 BRFSS microdata and U.S. Census Bureau population counts (CDC, 2025). Across study tracts, crude adult asthma prevalence averaged 10.4% (95% CI [9.1, 11.8]), varying from 6.2% (95% CI [5.4, 7.1]) in suburban tracts to 15.8% (95% CI [14.1, 17.6]) in core urban tracts. Age-adjusted rates confirmed persistent spatial disparities independent of demographic age distributions.”
How to calculate it in Excel, R or Python
These snippets illustrate how to compute Stage 2 poststratification prevalence from stratum population counts and model logits.
Excel Formula
For Stratum Population in cells A2:A5 and Stratum Logits in cells B2:B5, calculate the population-weighted prevalence in a single array formula using SUMPRODUCT:=SUMPRODUCT(A2:A5, 1 / (1 + EXP(-B2:B5))) / SUM(A2:A5)
R Implementation
# Poststratification in R
strata_data <- data.frame(
pop_n = c(1200, 1000, 1000, 800),
logit_val = c(-2.30, -1.90, -2.05, -1.65)
)
strata_data$prob <- plogis(strata_data$logit_val)
tract_prev <- sum(strata_data$pop_n * strata_data$prob) / sum(strata_data$pop_n)
cat(sprintf("Estimated Tract Prevalence: %.2f%%n", tract_prev * 100))
# Output: Estimated Tract Prevalence: 12.06%
To generate formatted statistical tables of underlying survey logistic regression models, researchers frequently use gtsummary in R.
Python Implementation
import numpy as np
import pandas as pd
df = pd.DataFrame({
'pop_count': [1200, 1000, 1000, 800],
'linear_predictor': [-2.30, -1.90, -2.05, -1.65]
})
df['pred_prob'] = 1.0 / (1.0 + np.exp(-df['linear_predictor']))
tract_prevalence = (df['pop_count'] * df['pred_prob']).sum() / df['pop_count'].sum()
print(f"Model-Based Tract Prevalence: {tract_prevalence * 100:.2f}%")
# Output: Model-Based Tract Prevalence: 12.06%
FAQ
What is the difference between CDC PLACES and CDC WONDER?
CDC WONDER reports tabulated counts from direct public health registries—chiefly mortality statistics from death certificates and birth records—at national, state, and county levels. In contrast, CDC PLACES generates model-based synthetic estimates for chronic disease morbidity, preventive care, and risk behaviors down to census tracts and ZCTAs where direct survey data are absent.
How does the CDC PLACES dataset define asthma and chronic conditions?
Case definitions in PLACES mirror standard BRFSS survey questions. For example, current adult asthma prevalence represents the model-based percentage of respondents aged 18+ who report ever being diagnosed with asthma by a health professional and reporting they currently have asthma. Identical standardized definitions apply to diabetes, coronary heart disease, and depression.
How can analysts query the CDC PLACES API?
PLACES datasets on data.cdc.gov are accessible through the Socrata Open Data API (SODA). Developers can programmatically filter by state, county, measure identifier (such as MeasureId='CASTHMA'), and geographic level via HTTPS GET requests returning JSON or CSV data.
Can PLACES estimates be combined with the Social Vulnerability Index (SVI)?
Yes. Both CDC PLACES and the CDC/ATSDR Social Vulnerability Index share standardized Census FIPS codes. Researchers routinely link tract-level PLACES health outcomes with tract-level SVI themes to model the impact of socio-economic vulnerability on health disparities.
Sources
- Centers for Disease Control and Prevention. (2025). PLACES: Local Data for Better Health, County Data, 2025 release [Data set]. U.S. Department of Health and Human Services, Division of Population Health, Epidemiology and Surveillance Branch. Accessed November 10, 2026, from https://data.cdc.gov/api/views/swc5-untb.json
- Zhang, X., Holt, J. B., Lu, H., Keenan, N. L., Thatte, N., & Croft, J. B. (2014). Multilevel regression and poststratification for small-area estimation of population health outcomes: A case study of chronic obstructive pulmonary disease prevalence using the Behavioral Risk Factor Surveillance System. American Journal of Epidemiology, 179(8), 1025–1033. https://doi.org/10.1093/aje/kwu018 (PMID: 24598867). Read November 10, 2026.
- Zhang, X., Holt, J. B., Yun, S., Lu, H., Greenlund, K. J., & Croft, J. B. (2015). Validation of multilevel regression and poststratification methodology for small area estimation of health indicators from the Behavioral Risk Factor Surveillance System. American Journal of Epidemiology, 182(2), 127–137. https://doi.org/10.1093/aje/kwv002 (PMID: 25957312). Read November 10, 2026.
- Centers for Disease Control and Prevention. (2025). PLACES Overview and Methodology. U.S. Department of Health and Human Services. Read November 10, 2026.







