A Multilevel Model of Health Status Change among Welfare Recipients Following Welfare Reform
Christina R. Studts, MSW, University of Louisville, Ramona Stone, PhD, University of Louisville, and Gerard M. Barber, PhD, University of Louisville.
Patterns of change in the health status of welfare recipients following welfare reform have implications for policy-makers and direct service practitioners working with this population. Several studies have reported negative outcomes regarding physical and mental health status in welfare recipients in the years following welfare reform. This longitudinal panel study followed 507 participants from 1998-2001 to identify predictors of changes in health status over time. Telephone interview data were merged with state administrative welfare records. A health problems index score served as the dependent variable. Differences in baseline health status and rates of change of health status were explored among a group of welfare recipients affected by welfare reform. The study design allowed comparisons based on a range of sociodemographic variables, including time-varying predictors such as welfare status, marital status, and employment status. It was hypothesized that sociodemographic factors (e.g., age, education, marital status), as well as health and welfare characteristics (e.g., welfare status, access to health care services), would be associated with initial health status and changes in health status over time. Descriptive statistics were obtained for the sample at each wave of data collection. Bivariate analyses were conducted to investigate possible associations between hypothesized predictors and the health problems index score. Finally, a mixed multilevel model for change was utilized to capture both the individual growth patterns and the systematic interindividual differences in change in health status over time. This analysis approach allowed for use of unbalanced data (i.e., variation in number of respondents at each time wave) to glean the most information from the dataset. Results suggested that most participants were Caucasian, single, female, unemployed, and lacking a high school diploma. Several sample characteristics remained stable across all four years, including education level, number of children, and region of residence. However, over time, decreased proportions of participants were noted to (1) receive welfare cash assistance; (2) be unemployed; (3) have incomes below the poverty line; and (4) be single. In contrast, increasing proportions of participants reported problems with (1) access to health care, (2) lacking health insurance, and (3) having at least one child with physical and/or mental health problems. Predictors of negative change in health status over time included lacking a high school diploma; being unemployed; having a child with physical or mental health problems; having problems with access to health care services; being at least five years older than the mean age of the sample; living in an Appalachian county; and receiving welfare cash assistance. Being unemployed and living in an Appalachian county were two predictors that interacted significantly with time: both were associated with steeper declines in health status as time progressed. “Best case” and “worst case” individual growth curve graphs were generated to illustrate results visually. Results highlight the differences in health status initial starting points and trajectories among a sample of participants affected by welfare reform. Factors associated with negative changes in health status should be considered by policy-makers and practitioners in responding to the needs of this high-risk population.