Search

Objective: To develop and test predictive models of discontinuation of behavioral health service use within 12 months in transitional age youth with recent behavioral health service use. Data sources: Administrative claims for Medicaid beneficiaries aged 15-26 years in Connecticut. Study design: We compared the performance of a decision tree, random forest, and gradient boosting machine learning algorithms to logistic regression in predicting service discontinuation within 12 months among beneficiaries using behavioral health services. Data extraction: We identified 33,532 transitional age youth with >1 claim for a primary behavioral health diagnosis in 2016 and Medicaid enrollment of >11 months in 2016 and >11 months in 2017. Principal findings: Classification accuracy for identifying youth who discontinued behavioral health service use was highest for gradient boosting (80%, AUC = 0.86), decision tree (79%, AUC = 0.84), and random forest (79%, AUC = 0.86), as compared with logistic regression (71%, AUC = 0.71). Conclusions: Predictive models based on Medicaid claims can assist in identifying transitional age youth who are at risk of discontinuing from behavioral health care within 12 months, thus allowing for proactive assessment and outreach to promote continuity of care for younger persons who have behavioral health needs. KEYWORDS emerging adults, machine learning, mental health services, What is known on this topic * Many youth and young adults with behavioral health needs drop out or experience significant disruption in continuity of care. * Discontinuation of behavioral [...]

1 | INTRODUCTION Alang et al.'s article in this issue of Health Services Research, 'Police Brutality and Unmet Need for Mental Health Care,' (1) identifies an important association between negative [...]

Objective: To examine the availability and facility-level predictors of LGBT-specific mental health and substance abuse treatment in the United States. Data Sources/Study Setting: 2016 National Survey of Substance Abuse Treatment Services, 2016 National Mental Health Service Survey, and 2015-2016 Gallup Daily tracking survey. Study Design: Logistic regression models and average marginal effects were used to identify characteristics of facilities that offer LGBT-specific programs. Linear regression models were used to estimate the association between the state-level proportion of LGBT people and the proportion of facilities that offer LGBT-specific programs. Data Collection/Extraction Methods: Secondary data analysis. Cases with missing values for any predictor were excluded. Principle Findings: 12.6 percent of mental health and 17.6 percent of substance abuse facilities reported LGBT-specific programs. Several facility characteristics were statistically associated with the likelihood of mental health and substance abuse facilities providing LGBT-specific programs, including offering outpatient or residential treatment, private ownership, religious affiliation, and payment type. The proportion of LGBT adults living within each state was statistically associated with state-level density of LGBT-specific mental health programs, but not substance abuse programs. Conclusions: Findings suggest limited availability of culturally competent mental health and substance abuse treatment, despite well-documented need. KEYWORDS mental health, sexual and gender minorities, substance abuse, treatment, 1 | INTRODUCTION In the United States (US), nearly 47 million people meet the criteria for a mental, behavioral, or emotional disorder (1) and approximately 20 million people meet the [...]

Objective. To estimate health care utilization and costs associated with the type of intimate partner violence (IPV) women experience by the timing of their abuse. Methods. A total of 3,333 women (ages 18-64) were randomly sampled from the membership files of a large health plan located in a metropolitan area and participated in a telephone survey to assess IPV history, including the type of IPV (physical IPV or nonphysical abuse only) and the timing of the abuse (ongoing; recent, not ongoing but occurring in the past 5 years; remote, ending at least 5 years prior). Automated annual health care utilization and costs were assembled over 7.4 years for women with physical IPV and nonphysical abuse only by the time period during which their abuse occurred (ongoing, recent, remote), and compared with those of never-abused women (reference group). Results. Mental health utilization was significantly higher for women with physical or nonphysical abuse only compared with never-abused women--with the highest use among women with ongoing abuse (relative risk for those with ongoing abuse: physical, 2.61; nonphysical, 2.18). Physically abused women also used more emergency department, hospital outpatient, primary care, pharmacy, and specialty services; for emergency department, pharmacy, and specialty care, utilization was the highest for women with ongoing abuse. Total annual health care costs were higher for physically abused women, with the highest costs for ongoing abuse (42 percent higher compared with nonabused women), followed by recent (24 percent higher) and remote abuse (19 percent higher). Women with recent nonphysical abuse only had annual costs that were 33 percent higher than nonabused women. Conclusion. Physical and nonphysical abuse contributed to higher health care utilization, particularly mental health services utilization. Key Words. Intimate partner violence, utilization, costs, abuse, Physical and nonphysical types of intimate partner violence (IPV) are common in women, with lifetime prevalence ranging from 17 to 34 percent for physical IPV, and 12 to 35 percent [...]

Objective. We examine whether ethnic differences in use of inpatient mental health services exist when the usually confounding effects of minority status and culture are minimized or controlled. Data Sources and Study Setting. Secondary analyses were conducted using a national insurance claims database for 1.2 million federal employees and their dependents insured by the Blue Cross/Blue Shield (BC/BS) Federal Employees Plan (FEP). Study Design. The Andersen-Newman model of health utilization was used to analyze predisposing, enabling, and need variables as predictors of inpatient mental health utilization during 1983. The study design was cross-sectional. Data Collection. The study database was made up of BC/BS insurance claims, Office of Personnel Management employee data, and Area Resource File data. Principal Findings. No significant differences were found among blacks, whites, and Hispanics in the probability of a psychiatric hospitalization or in the number of inpatient psychiatric days. Regression analyses revealed younger age and psychiatric treatment of other family members as significant predictors of a hospitalization; region of residence, younger age, hospital bed availability, and high option plan enrollment were significant predictors of the number of treatment days. Conclusions. Ethnic differences in use of inpatient mental health services were not significant in this generously insured population. Further research involving primary data collection among large and diverse samples of ethnic individuals is needed to fully examine the effects of cultural and socioeconomic differences on use of mental health services., Although the literature on ethnicity and mental health has expanded considerably over the past two decades, few studies have examined and compared the use of inpatient mental health services by [...]