Associations Between Social Determinants of Health and Emergency Department Visits for Myocardial Infarction, Florida, 2005-2012

BACKGROUND:  Myocardial infarction (MI) is an important public health problem in Florida, with 10% of Floridians reporting a history of acute MI or coronary heart disease in 2010. There are many known individual-level risk factors associated with MI; however, there are also area-level risk factors, known as social determinants of health (SDoH), that may increase risk of MI or worsen complications. Studies have found relationships between neighborhood-level income, minority status, unemployment, education, and housing variables and MI. Few studies have characterized these associations using statewide data. As part of a special project through the Environmental Public Health Tracking Program (EPHT), we sought to characterize the SDoH predictors of rates of MI emergency department (ED) visits in Florida from 2005 to 2012.

METHODS:  We obtained ED visit data from hospital facilities and 2010 Decennial Census and 2007-2011 American Community Survey data from the US Census Bureau. Census and health data were linked by matching patient ZIP code (smallest geographic unit available) with 2010 ZIP Code Tabulation Area (ZCTA) designations. Direct age-adjusted rates were calculated based on 2000 US standard population. Various spatial statistics tools available in ArcMap v.10.0 (ESRI; Redlands, CA) were used for analysis.

RESULTS:  The average age-adjusted rate of MI ED visits statewide was 204.9 per 100,000 (95% confidence interval [CI]: 171.1, 238.7). Clustering was noted (Moran’s I z-score: 4.0; p-value < 0.0001), with significantly higher rates occurring in areas of central and north central Florida and the Panhandle. At the ZCTA-level, percent married, percent with a high school diploma or higher, median household income and household value were inversely associated with rates of MI (p-values < 0.05). Percent rural and percent with less than a high school diploma had significant direct relationships with acute MI. Other SDoH variables were not significant in bivariate models. A final multivariable model was determined, with only percent rural and percent with a high school diploma remaining significant. The geographically weighted regression model accounts for 38.7% of the variability in rates of MI ED visits.

CONCLUSIONS: Clustering of MI rates and associations with SDoH factors provide important insights into new areas of intervention for improving public health in Florida. Additional work is ongoing and examines combined rates of ED visits and hospitalizations for MI that could be associated with area-level factors. The information obtained in these analyses will also be important contributions to another EPHT special project related to developing Community Health Profiles.