A Methodology for Analyzing Healthcare Utilization Among College Students with Mental Health Disorders

Over the past decade, the prevalence and severity of mental health disorders have been growing among the college student population. Because psychological disorders and maladjustment can disrupt students’ academic, emotional and social lives, as well as the university community as a whole, most colleges and universities in the US have counseling centers that provide behavioral health services. However, surveys show that only a meager 18% of students with mental health problems are receiving proper treatment, whiles Student Health Centers (SHCs) struggled with offering adequate healthcare service to meet the treatment need from patients. Despite devoting significant resources to the provision of mental health services, there is no consensus about how these services should be integrated with medical health services.
This unique study investigates how information sharing between behavioral and medical services influences the utilization of healthcare service by mental health patients in American universities. Specifically, we used de-identified electronic health records (EHRs) data of 21 schools participating in the College Health Surveillance Network (CHSN) Project from January 1, 2011 through May 31, 2014, which include ICD-9 diagnostic codes and CPT procedure records of over 717,370 students with over 3.9 million visits. Then, we defined two levels of service integration: “standard” for universities with separate information systems and minimum clinical collaboration between medical and behavioral health services; “enhanced” for universities with an integrated EHR and consistent clinical collaboration between services. We proposed two measures of efficiency: frequency, defined as the total number of primary care visits per unit time, was compared using a negative binomial multilevel model; while complexity, defined as total visit length for primary care visits per unit time, was compared using a linear multilevel model. Statistical models showed that, when controlling for the variance in age, sex and total time in school, patients with a mental health diagnosis in standard systems had 15.58% (95% CI, 10.77% -20.44%) more primary care visits and spend 22.89% (95% CI, 21.42% -24.38%) more time than patients in the enhanced systems. This study demonstrates significantly lower utilization by college students with mental health disorders in an enhanced clinical integration service system, suggesting that patient care could be improved and cost reduced if institutions of higher education adopted the integrated health care. The contribution of this work is a novel framework to incorporate multisite electronic health data to study utilization of different health care models. Future work includes predictive modeling of utilization patterns in different care models and extending the study to specific patient subgroups.