Heat and socioeconomic deprivation compound to drive coronary heart disease in Los Angeles.

Coronary heart disease hot spots geographically overlapped with areas of high socioeconomic deprivation and high heat exposure in Los Angeles County.

• Analysis covered 2,513 census tracts in Los Angeles County.
• Overlapping CHD, deprivation, and heat hot spots were concentrated in the south and east of Los Angeles County.
• Hot spots were identified using Getis-Ord Gi* statistics.
• CHD prevalence data were obtained from CDC Population Level Analysis and Community Estimates (2021).

Higher socioeconomic deprivation was significantly associated with higher coronary heart disease prevalence in standardized OLS regression models.

• A 1-SD increase in the Social Deprivation Index (SDI) was associated with a 0.163-SD higher CHD prevalence.
• This association was statistically significant (p < 0.001).
• Models used z-score-standardized OLS regression with tract-level sociodemographic controls.
• Socioeconomic deprivation was measured using the Social Deprivation Index (SDI).

Higher land surface temperature was significantly associated with higher coronary heart disease prevalence in standardized OLS regression models.

• A 1-SD increase in land surface temperature (LST) was associated with a 0.070-SD higher CHD prevalence.
• This association was statistically significant (p < 0.001).
• LST was retrieved from the ECOsystem Spaceborne Thermal Radiometer Experiment on Space Station (ECOSTRESS).
• The LST-CHD association was smaller in magnitude than the SDI-CHD association (0.070-SD vs. 0.163-SD).

Geographically weighted regression revealed substantial spatial heterogeneity in the associations between deprivation, heat, and coronary heart disease across census tracts.

• The SDI-CHD association was strongest in central and southern tracts.
• The LST-CHD association was strongest in central and eastern tracts.
• GWR was applied with adaptive bandwidths to characterize spatial heterogeneity.
• These findings suggest that dominant risk drivers and intervention priorities differ by neighborhood.

Spatial dependence was assessed in the regression models, and geographically weighted regression was applied to account for it.

• The study assessed spatial dependence in the OLS residuals.
• Geographically weighted regression (GWR) with adaptive bandwidths was applied as a response to detected spatial dependence.
• The ecological geospatial analysis used 2,513 census tracts as the unit of analysis.
• Tract-level sociodemographic controls were included in the regression models.