Re-scaling and small area estimation of behavioral risk survey guided by social vulnerability data

Area of Science:

• Public Health
• Biostatistics
• Geospatial Analysis

Background:

• Local governments require granular population health data for resource allocation and intervention planning.
• National surveys often lack the necessary subcounty-level detail for these public health activities.
• Existing data collection methods are insufficient for small area estimation of health behaviors.

Purpose of the Study:

• To develop and demonstrate a novel two-step method for creating small area estimates (SAEs) of smoking rates.
• To rescale health survey data to a finer geographic resolution (census tracts).
• To utilize ancillary data on social vulnerability for improved local health risk assessment.

Main Methods:

• A two-step approach involving spatial microsimulation and logistic linear mixed modeling was employed.
• Survey respondent locations were rescaled from zip codes to census tracts using demographic data.
• Area-level modeling incorporated census tract-specific social vulnerability data.

Main Results:

• The study successfully estimated the ever-smoking rate for census tracts in Allegheny County, Pennsylvania.
• High ever-smoking rates (over 70%) were identified in specific census tracts southeast of Pittsburgh.
• Several tracts within Pittsburgh also exhibited elevated smoking rates (over 65%).

Conclusions:

• The developed small area estimates (SAEs) can guide local public health efforts to reduce smoking.
• The methodology offers a scalable approach for estimating other health outcomes in different geographic areas.
• This novel technique enhances the utility of survey data for localized public health decision-making.