Skip to main content

Table 1 Covariates for regression models

From: Social determinants of health and disparities in prenatal care utilization during the Great Recession period 2005-2010

Covariate Level Covariate Name/Description
Individual • Race/ethnicity: non-Hispanic White (White), Hispanic White (Hispanic), non-Hispanic Black (Black)a
• Maternal age
• Marital status (Married/Unmarried)
• Mother foreign-born (Yes/No)
• Maternal education (Less than HS; HS Diploma or GED; some college not assessed (age < 20 years))
• WIC (maternal WIC enrollment) (Yes/No)
• Maternal insurance status (e.g., Medicaid or private insurance).
(at the LHJ level unless otherwise indicated)
• Core Based Statistical Area (CBSA) (metropolitan, micropolitan, or rural)
• Community poverty (binary variable, 1 for LHJs with highest percentage (top 1/3) of residents age 0–17 in poverty in each state, 2 for lower number of residents age 0–17 in poverty (non-poor LHJs)
• Partisan Voting Patterns: Percent of voters voting Republican (vs. Democrat or Independent) in the 2004 and 2008 presidential elections c
• Gini coefficient (2000 census; measure of income distribution/inequality (0–1), larger number > inequality), measuring levels of income inequality
• Per Capita General and Family Practitioner MDs/LHJs (for years 2005, 2008, 2010)
• Per capita LHJ unemployment rated
Expendituree • Total LHD expenditures
• WIC expenditures
• Family Planning (FP) expenditures
• Maternal/Infant/Child/Adolescent (MICA) services expenditures
• 2MCH--Combined expenditures for 2 MCH services (FP and MICA)f
State • State-level dummy variables were created for WA and FL to capture any state-level differences.
  1. aRace/ethnicity groups were defined using data from two separate variables (maternal race and maternal ethnicity) to create a 3-category combined race/ethnicity variable
  2. bCommunity level covariates were selected based on previous research or for which social determinants of health theories suggest a plausible association to maternal and child health (MCH) outcomes in the context of the Recession [16, 17, 24,25,26,27,28,29,30,31,32]
  3. cThe partisan voting patterns measure was intended to act as a proxy for differences in political orientation at the community level as previous research has identified Republican voters as less likely than Democrats to perceive that there are people in the United States who encounter access to care issues and are less likely to support public health reform [27]
  4. dIndividual unemployment data were not available
  5. eLHD-specific per capita expenditure data were included in the preliminary model as the Recession yielded widespread reports of budget cuts to LHDs [7]. Per capita rates were calculated using total LHJ population as a denominator. Differences in fiscal years between WA and FL were reconciled by assigning FL’s FY to the earlier year (e.g., FL FY 2005–2006 associated with WA FY 2005)
  6. fMICA [25, 31] represents a composite of similar expenditure categories for WA and FL LHDs that includes comparable intervention activities among LHDs in both states—e.g., home visiting, prenatal health programs