Effects of School-Based Health Interventions on Student Mental Healthcare Utilization: Evidence from Bunching Estimation

• In Event: New Methodologies for Distributional Weighting and Program Evaluation

Abstract

Increases in federal resources from the Affordable Care Act and Advancing Wellness and Resiliency in Education (AWARE) grants from the Substance Abuse and Mental Health Services Administration for addressing student health needs have contributed to expansion of School-Based Health Centers (SBHCs). Previous investigations into SBHCs have provided no individualized understanding of student mental health diagnoses or the extent to which reductions in diagnosed conditions are the result of increased student access to mental health resources. In this study, we compare students with one mental health visit to those who have never utilized mental healthcare, yet likewise have a high need or propensity for treatment. By investigating the differences between students who are “on the bubble” of engaging in a mental healthcare visit, we can estimate the causal effect of mental healthcare utilization on diagnoses.

We address two primary research questions: (1) To what extent does establishment of SBHCs change the rate of diagnoses of mental and behavioral health disorders among school-aged children over time? (2) To what extent are there heterogeneous effects in locations with greater unmet need?

We utilize a longitudinal dataset of health and education outcomes for all low-income school-aged children in Tennessee linked from 2006 to 2022. Utilization and treatment data include Medicaid claims of emergency department (ED), inpatient, and outpatient visits for mental health conditions, substance use, and suicidal thoughts and behaviors. These data are linked with mental health diagnosis data encompassing conditions like ADHD, depression, anxiety, and bipolar disorder.

We utilize a novel bunching methodology allowing examination of the effect of confounding variables at a bunching point. This point is due to the non-negativity constraints and diagnosis variables and is independent of access to an SBHC. In essence, since one cannot receive negative diagnoses, by examining the variation in student and district characteristics of those “on the bubble” of diagnosis, we can extract the effect of SBHCs on a student receiving a diagnosis. This analysis is conducted utilizing a semiparametric Tobit correction, with k-means clustered controls.

Bunching estimator results will be examined in conjunction with other rigorous quasi-experimental approaches, namely event studies and difference-in-differences (DiD) estimation. We will conduct doubly-robust DiD estimation based on stabilized inverse probability weighting with wild bootstrap standard errors, comparing mental health diagnoses and utilization for students in schools with an SBHC and/or an AWARE grant with those in schools without additional funding sources.

Preliminary findings utilize the bunching estimation approach to extract exogenous variation in healthcare utilization. These results indicate that one additional mental healthcare visit is likely to decrease mental health diagnoses; however, this result is altered with the presence of an SBHC. In a district with an active SBHC/AWARE grant, diagnoses increase. Additionally, outpatient and residential mental healthcare engagement slightly increases students’ reading and math scores, while inpatient and ER visits have negative impacts on student achievement. Results on education outcomes within schools with SBHCs are mixed. These effects will continue to be rigorously examined and assumptions tested.

Authors

• Kathryn Enriquez, Vanderbilt University
  Presenting Author

• Carolyn J. Heinrich, Vanderbilt University
  Non-Presenting Co-Author

• Carrie Fry, Vanderbilt University
  Non-Presenting Co-Author