Data/Methods: Annual data were collected on a population cohort of more than 27,000 persons for the full 9 years of the study. Demographic, diagnostic and enrollment data were obtained from state Medicaid files and mental health services use data were obtained from the Maryland Public Mental Health System. For the analysis we employed a clog-log binary regression model with an observation included for each person for each year before first uptake of SE services and for the year of uptake. (Observations for years following the uptake were excluded.) The 0-1 SE uptake variable was the dependent variable.
Results: Results of the study provided tentative evidence of the initiative’s positive impact on take-up (particularly during the recession downturn of 2007-2010). We also estimated a relatively large and negative marginal effect for the county unemployment rate (as a proxy for local labor market conditions). Geographic access (distance) from SE providers had a negative estimated marginal effect on take-up probability, but the magnitude was small and we observed only small decreases in this access barrier over time in our data.
Our analysis results also suggested that personal characteristics indicative of stronger prior job history or better labor market prospects were significant predictors of SE take-up; these included gender, age, prior work history, and absence of pre-baseline use of inpatient somatic care. Contrary to expectations, we also found that relative to other SMI diagnosis groupings, the probability of take-up was significantly higher for persons with schizophrenia. We conjecture that this latter result may be due to yet unexplored patterns of mental health provider referral practices by which patients with diagnoses other than schizophrenia are less likely to be referred since in the past these patients were also less likely to be served in SE programs.