Socioeconomic status and reported health, 2003-2007

Problem: The elimination of health disparities by race and socioeconomic status is one of the main public health goals. Progress toward this goal implies that health status differences among subgroups of the population will have to decrease over time. The purpose of this presentation is to examine whether the association between two socioeconomic variables and reported health status has diminished in the last 4 years.

Data: The March Supplement of the Current Population Survey consistently asks respondents about their general health, family income, educational attainment and age among other items. Respondents' report their general health as either excellent, very good, good, fair, or poor. The scale is recoded into two broader classes (fair-poor health vs. good health) to form the binary dependent variable. Family income is measured in reference to the poverty line and is expressed in 6 classes ranging from below poverty line to more than 5 times of the poverty line. Educational attainment is recoded into four groups: less than high school, high school, up to two years of college, and college degree or higher degrees. To control for population ageing, the analysis is separately run for five age groups. The analysis sample includes adults 25 years of age and older for survey years 2003 and 2007. The analysis sample includes persons who uniquely identified themselves as Asian, African American, or Caucasian (n=249,516, years 2003 and 2007 combined).

Method: For each age group logistic regression models relate the log-odds ratio of poor/good health to 6 income classes, 4 educational classes and the year of survey. Interaction terms involving year of survey measure change in the SES-health gradients. Models are hierarchically tested starting with models including interaction terms. Parsimonious models are the basis for the interpretation of results.

Preliminary Findings: (1) Age-specific models without interaction terms fit to the observed data well. That is, the SES-reported health gradients appear stable over the last 4 years in all age groups. (2) The association of SES and self-reported health is most pronounced in the pre-retirement years. Due to the large sample size, all relationships reported are statistically significant. The final analysis will also report on changes in the reported health gaps between Asian, African American and Caucasian subgroups.

Conclusion: The preliminary analysis suggests little progress in reducing the reported health status disparities between high SES and low SES respondents in the last four years.