Abstract: Configuration and Spatial Validation of a Neighborhood Risk Index of the Social Determinants of Health (Society for Social Work and Research 25th Annual Conference - Social Work Science for Social Change)

All live presentations are in Eastern time zone.

276P Configuration and Spatial Validation of a Neighborhood Risk Index of the Social Determinants of Health

Schedule:
Tuesday, January 19, 2021
* noted as presenting author
Kristen A. Berg, PhD, Postdoctoral Research Scholar, Case Western Reserve University, Cleveland, OH
Jarrod E. Dalton, PhD, Assistant Professor, Cleveland Clinic Lerner College of Medicine, Cleveland, OH
Douglas D. Gunzler, PhD, Associate Professor of Medicine, Case Western Reserve University, Cleveland, OH
Claudia J. Coulton, PhD, Lillian F. Harris Professor of Urban Research & Social Change, Case Western Reserve University, Cleveland, OH
Darcy A. Freedman, PhD, Associate Professor, Case Western Reserve University, Cleveland, OH
Nikolas I. Krieger, MS, MA, Data Scientist, Cleveland Clinic Lerner Research Institute, Cleveland, OH
Neal V. Dawson, MD, Professor, Case Western Reserve University, Cleveland, OH
Adam T. Perzynski, PhD, Associate Professor, Case Western Reserve University, Cleveland, OH
Background and Purpose: Neighborhood (area-based) socioeconomic position has been shown to affect a range of physical, behavioral, and social health outcomes. Further, measures of neighborhood social determinants of health (SDH) have shown prognostic value for informing preventive screening decisions, care utilization, and treatment decisions by medical and social service providers. Given growing interest in using SDH measures, it is critically important to evaluate construct validity. We examined the configuration of a commonly-used index of neighborhood area deprivation (the Area Deprivation Index; ADI) as well as its spatial invariance.

Method: Data for all United States Census Tracts (N=73,056) from the 2017 American Community Survey were compiled into constituent items of the ADI: percentage of owner-occupied housing, households with no vehicle, families in poverty, people living below 150% of the federal poverty level, children in single-parent households, people unemployed, people with white-collar jobs, people with < 9 years of education, people with at least a high-school education, and households with over one person per room; ratio of those making under $10,000 to those making over $50,000; and median mortgage, house value, rent, and household income (Singh, 1993; Kind, 2018). Exploratory factor analysis (EFA) models were estimated to evaluate the ADI’s (i) configuration (whether the index is unidimensional) and multigroup confirmatory factor analysis (CFA) models were estimated to evaluate the ADI’s (ii) spatial invariance (measurement equivalence for heterogeneous states in the U.S.).

Results: EFA results indicated a three-dimensional (Comparative Fit Index; CFI=.91) factor structure fit the data better compared to the unidimensional model (CFI=.53), with financial strength, economic inequality, and education forming the three factors. Evaluation of spatial noninvariance between New York and Minnesota indicated partial invariance; median rent is a stronger correlate of the financial strength of neighborhoods in New York, but median house value and median mortgage are more strongly associated with neighborhood financial strength in Minnesota. Moreover, the percent of households with no vehicle is a stronger correlate of economic inequality in Minnesota than in New York state.

Conclusions: A commonly-used SDH index fails standard tests of construct validity. Models incorporating neighborhood SDH concepts should include more robust multidimensional measures, and investigate the influence of distinct dimensions of SDH on health and social outcomes. Further, clinical and social care decisions may be best informed by integrating regionally-relevant SDH measures.