Universal Depression Screening in Adolescents in Primary Care: Evaluating Effectiveness and Preferences using Health Decision Modeling

Abstract

In an era of intensified social media connectivity, compounded by modern sociopolitical threats, such as school gun violence or online bullying, adolescent depression is increasing in the United States. About one-fifth of adolescents has experienced major depression. About half of adolescents with depression are undiagnosed. Adolescent depression screening rates are low. Universal depression screening in adolescents is recommended as part of routine primary care, but it is unclear on how to best implement it. To my knowledge, no study has reported the effectiveness of depression screening on improving adolescent health outcomes. The preferences of primary care physicians, an important and relevant clinical decision-maker, has not been assessed. The purpose of this dissertation was to provide evidence to inform medical decision-making regarding the implementation of universal depression screening in adolescents in primary care.

In Chapter 2, I built a health simulation model to project health outcomes along the natural history of major depression in U.S. adolescents. I used a state-transition model to estimate the depression-related outcomes without applying an intervention. I derived the transition inputs by using survival analysis on depression data from the National Longitudinal Study of Adolescent to Adult Health. I then calibrated the model outputs to match historical prevalence data from the National Survey on Drug Use and Health. The model simulated a hypothetical cohort of 1,000 adolescents starting at 12 years old. After a ten-year modeling period, approximately 65% of adolescents never developed depression, 12% were depressed, 23% were recovered, and 0.1% died from suicide. The findings by age and sex confirmed the literature, but findings by race-ethnicity were mixed, suggesting that future research should attempt to collect higher quality, disaggregated data in diverse racial-ethnic groups.

In Chapter 3, I incorporated the first model into a decision analytic model to measure the intervention effects of universal depression screening using a cost-effectiveness analysis. The primary outcome was incremental cost-effectiveness ratios. Compared to usual care, universal annual screening was preferred if policy decision-makers are willing to invest at least $61,000/quality-adjusted life year (QALY) from the societal perspective (costs borne by society). The universal annual screening was preferred for 98% of 10,000 simulations at a $100,000/QALY threshold. At lower thresholds, multiple screening strategies were preferred except for biennial screening.

In Chapter 4, I evaluated the preferences of primary care physicians regarding the implementation of universal adolescent depression screening. I developed a survey instrument, containing a discrete choice experiment, by using exploratory sequential mixed methods. The survey was distributed to a national sample of 50 physicians. Respondents indicated the top priority was reaching the lowest number of missed depression cases. Respondents also preferred the shortest allotted clinician’s time, briefest screening completion time, provision of a private area, and an electronic depression screener over paper. Respondents were willing to trade approximately 18 minutes to reduce missed depression cases.

This dissertation demonstrated the effectiveness of universal depression screening on adolescent health outcomes. It also characterized the epidemiology of depression in adolescents, clarified the optimal screening frequency and their associated cost-effectiveness outcomes, and revealed the preferences of physicians, a relevant decision-maker. This work also demonstrated the potential of simulation modeling and discrete choice techniques to support health policy decision-making around complex issues, such as major depression in adolescents, a population for whom has largely unmet mental health needs.