Depressed mood is one of the most common of all psychiatric symptoms occurring in children and adolescents. Population studies suggest a point prevalence of between 10 to 15% of children and adolescents having symptoms of depression. Further, depressed adolescents are also significantly more apt to demonstrate suicidal ideation accompanied by a concomitant sense of helplessness and hopelessness. The overall aim of the study was to identify and characterize profiles of depression and suicidal behavior within the adolescents of the USA in the past 10 years. This study utilized epidemiological, cross-sectional, data from the Youth Risk Behavior Surveillance System (YRBSS), a biennial census that monitors six types of health-risk behaviors that contribute to the leading causes of death and disability among youth. Latent classes of the indicators were generated utilizing latent class regression modeling. Predictors were then regressed on class membership in a multinomial logistic regression simultaneously to assess significance. Finally, a juxtaposition of the profiles and significant predictors followed to allow for observation of differences in number of profiles and other qualities (i.e., proportions of sample in each class, etc.) as well as to visualize and note “carryover” predictors across the past 10 years. Findings revealed a relatively stable pattern of profiles and predictors over the years with the exception of 2015. In the analysis of demographic variables, membership of the “low- risk non-depressed” class was consistently or more frequently associated with being male, older, not of an ethnic minority, and non-ethnically bi-or multiracial, across all time points. Three clusters of behaviors and factors emerged as significant predictors of depressed mood and suicidality. The first cluster consisted of typical adolescent risk behaviors that includes delinquent behaviors (i.e., fight, weapon carrying, or use of over-the-counter drugs), smoking, alcohol use, as well as consensual (non-violent) sexual activity. The second cluster of predictors that was significant consisted of experiences of traumatic events such as bullying, sexual assault, and intimate partner violence. Finally, a third cluster that showed significance consisted of self destructive behavior such as the use of illicit or hard drugs and maladaptive dieting, restricting or purging behavior. Several protective factors such as having sufficient physical activity and getting at least 8 hours of sleep daily also emerged as significant. Limitations to the YRBS and this study were discussed, and recommendations that tie to the implications of the findings were proposed. Future directions for research were also presented in light of the limitations of the study.

While the literature on Black-white differences in major depressive disorder (MDD) and depressive symptoms is robust, less robust is the literature on how these outcomes are patterned within the US Black population and why differences exist. Given increasing numbers of first-generation immigrants from the Caribbean, sub-Saharan Africa, Latin America, among other regions of the world, as well as increasing numbers of second- and third-generation immigrants, continued aggregation has the potential to mask intra-racial differences between these ethnic-immigrant groups and Black Americans with more distant ancestral ties to Africa (i.e., African Americans). Among these subgroups, the extremely limited data disaggregating the US Black population suggest the following patterns. First, foreign-born Black immigrants have lower levels of MDD and related symptoms relative to US-born Black Americans, a finding which is consistent with theories of foreign-born health advantage. Second, among the US-born, Caribbean adults have higher levels of MDD and related symptoms relative to all other Black Americans, a finding which is inconsistent with theories related to intergenerational declines in health toward convergence to native-born levels. Lastly, and contrary to results among adults, first- and second-generation Caribbeans have lower levels of depressive symptoms relative to all other Black youth. This dissertation sought to better understand how depression and its related symptoms are patterned within the US Black population, as well as how mechanisms causing these outcomes may vary across subgroups defined by domains related to immigration. Chapter 1 was a systematic review, which comprehensively synthesized depression and related symptoms within the US Black population across these domains, including a summary of mechanisms proposed toward explaining intra-racial variation. Using longitudinal data, Chapter 2 examined whether, and if so when, growth curve models of depressive symptoms varied by immigrant generation contrasts among a representative sample of Black youth followed into adulthood. And using representative data from the largest study of Black mental health, Chapter 3 examined whether the relationship between racial identity, a presumed protective factor against depression and related symptoms, and MDD varied between US-born Caribbeans and all other US-born Black Americans. The systematic review of Chapter 1 revealed substantial variation in the prevalence of depression and its related symptoms within the US Black population by nativity, region of birth, age at immigration, and Caribbean ethnic origin. Results additionally confirmed that much of what is known about intra-racial heterogeneity comes from a single data source, the National Study of American Life (NSAL). Using longitudinal data of youth followed into adulthood, Chapter 2 found evidence of diverging depressive symptoms trajectories among Black respondents by immigrant generation (first/second-generation compared with third and higher generations); notably, contrasts among Black respondents varied from those of other racial/ethnic groups (Asian, Hispanic/Latinx, non-Hispanic white). Lastly, results from Chapter 3 suggest aspects of racial identity may not be protective for US-born Caribbeans, pointing to variations in racialization experiences as a distal cause. Additional research using larger sample sizes, more diverse subgroups of Black ethnic immigrants, as well as longitudinal data, is needed to further understand patterns of and additional sources underlying heterogeneity of depression and its related symptoms within the US Black population.

Depression is a major public health problem. Decades of research have been conducted to create a classification system aligned with the complex phenomenological features of depression. The dominant classification system for depression is the latent paradigm, which conceptualizes observable symptoms of depression as effects of an underlying disorder. There is increasing evidence, however, that the latent model is inadequate to inform the prognosis and treatment of depression. Specifically, evidence is accumulating that symptoms of depression do not necessarily arise due to an underlying condition, but that symptoms occur as a network in which each one is causally related to a previous symptom. This dissertation critically evaluated the underlying assumptions of this “network paradigm,” one of the frameworks which had been proposed as an alternative to the traditional latent paradigm, as an appropriate model for studying depression. The first chapter systematically evaluated empirical depression network studies regarding whether the study design included an examination of the paradigm’s assumptions. In the second chapter, I investigated the relationships among depressive symptoms and determined whether causal relationships among depressive symptoms, a key assumption underlying this paradigm, could be a plausible explanation. The last chapter investigated a central controversy within the network literature regarding consistent findings and measurement error. The first chapter found that the majority of depression network studies published in the literature were not capable of providing empirical support of symptom causal relationships and often neglected to investigate the impact of measurement error. The second chapter estimated a significant relationship between two depressive symptoms - sadness and anhedonia, using an inverse probability treatment-weighted regression estimation approach in the context of longitudinal data. Causal relationships among symptoms, a key assumption underlying the network paradigm, may be a plausible explanation for the depressive symptom relationships. The third chapter found that statistical network models are not robust to measurement error through a series of simulation studies. Measurement error remained a general threat against the network paradigm, and existing network findings should be interpreted with caution. Overall, the network paradigm may be appropriate for study depression, but existing findings should be interpreted with caution. There is a need to explore the fundamental assumptions of paradigms prior to widespread application.

Background: Common mental illnesses are a major public health challenge. Two common mental illnesses, depression and anxiety, were respectively ranked the second and eighth major causes of disability in 2019. However, the mental health treatment gap in low- and middle-income countries (LMICs) is higher than 90%. Systematic reviews suggest that integrated models delivered by primary health or lay providers are effective at reducing symptoms and improving quality of life among those with mental illnesses in LMICs. However, integrated models have not been widely implemented in routine primary care and community settings, beyond researcher-controlled pilot studies in LMICs. This integrated learning experience (ILE) contributes to key gaps in global mental health and implementation research by outlining implementation strategies (the ‘how’) and components of integrated models (the ‘what’) for people living with common mental illnesses in Latin America, a region composed of LMICs and selected high income countries (HICs) widely known for being early adopters of integrated models. Given current literature gaps, this study also provides an applied example of the assessment of contextual implementation factors and the selection of implementation strategies for integrated models for common mental illnesses in Guatemala, a LMIC in Central America where the burden due to common mental illnesses is high and the implementation of integrated models is low. Methods: First, we conducted a scoping review to map and summarize the existing literature on integrated service models for common mental illnesses in primary care and community settings in Latin America. Second, we conducted a multi-methods assessment of the local context prior to selecting the implementation strategies for a collaborative care program for Maya T’zutujil young adults living with common mental illnesses in a rural municipal health district in Sololá, a rural department in Guatemala. We used data collected through the public health system to develop behavior-over-time (BOT) graphs outlining the number of primary care visits for common mental illnesses over time (2018-2022). We followed the Practical, Robust Implementation and Sustainability Model (PRISM) framework to conduct qualitative semi-structured interviews. Participants represented Ministry of Health coordinators and providers; community youth leaders with lived experience; and community providers. We performed matrix-based thematic analysis of interview transcripts. Third, we used group model building (GMB), a participatory systems thinking approach to inform the selection of implementation strategies for a primary care, community-based collaborative care program for common mental illnesses in rural Guatemala. Results: First, our scoping review included 33 publications conducted in 6 countries (Belize, Brazil, Chile, Colombia, Mexico, Peru) about 18 programs commonly addressing depression (N=14, 77.78%). Four studies were experimental. The most and least common components were ‘team-based care’ (N=14, 77.78%) and ‘family/user engagement’ (N=1, 5.55%). The most and least common Expert Recommendations for Implementing Change (ERIC) categories were ‘supporting clinicians’ (N=17, 94.44%), mainly through task-sharing, and ‘changing infrastructure’ (N=4, 22.22%). We found wide heterogeneity across studies about combinations of components and implementation strategies. Second, our multi-methods assessment showed that less than 1% of the total number of public health visits corresponded to common mental illnesses in the study health district. A collaborative care program could help to increase the number of visits. To enhance fit to the study health district, the program would need to ensure the users’ right to privacy and engage community providers (e.g. Maya providers, religious leaders) and Maya explanatory models of mental health. Infrastructural elements at the municipal health district, such as the availability of psychotropic medications, would need to be met to ensure the program’s implementation and sustainability. Third, we identified two health-district subsystems influencing the implementation of public primary mental health services. At the community-level, we identified four subsystems. We identified 32 distinct implementation strategies representing the nine ERIC categories. Conclusion: This ILE indicates the need for additional studies focused on the participatory design and evaluation of implementation strategies that go beyond the provider-level (supply side of implementation) and focus on the community-level (demand side of implementation). Our results and methodologies may be utilized by researchers and implementers seeking to integrate mental health services in Guatemala and other LMICs.

A groundbreaking narrative, which would come to be known as the theory of “deaths of despair”, emerged in 2015 from a study by Case and Deaton analyzing mortality rates in the United States between 1999 and 2013. They found an increasing trend in all-cause mortality rates due to drug poisonings, alcohol-related liver disease, and suicides, which they called “deaths of despair”, among non-Hispanic (NH) white Americans aged 45 to 54—this age group was called the midlife. Case and Deaton’s findings and their narrative about the hypothetical causes of their findings garnered significant attention. The authors of this narrative hypothesized that the observed increases in mortality rates were due to white individuals in midlife increasingly suffering from “despair” and proposed a causal link between increasing “despair” rates and increased mortality rates only among white Americans in midlife. Case and Deaton did not provide a clear definition of “despair”; they presumed that white Americans in midlife were hopeless about their prospects for the future compared to what their parents had attained. This provocative narrative persisted and gained momentum because it functioned as an explanation of recent events like the 2016 U.S. presidential election, rise in white nationalism, and far right extremism. These white-related events were thought to be expressions of an agonizing, poor, under-educated generation of white Americans increasingly suffering from hypothetical feelings of "despair”, which have led them to self-destructive behaviors and premature death. However, no study has investigated the central claim of this theory: whether there is evidence of an association between increased “despair” rates and increased mortality rates only among white individuals in midlife, particularly for those with low education. Moreover, there is little evidence of their hypothesis of an increasing epidemic of “despair” affecting only white Americans in midlife, particularly those with low education. The theory of “deaths of despair” can be understood through Geoffrey Rose’s framework of causes of incidence and causes of cases, which highlights the difference between between-population and inter-individual causes of disease. Rose’s argues that causes of incidence explain the changes in outcome rates between populations, and may be uniform and imperceptible within populations. On the other hand, the causes of cases explain why some individuals within a population are susceptible or at high risk of the outcome. Like Rose’s causes of incidence, the authors of the theory of “deaths of despair” argue that “despair” increased between the midlife white American population in 1999 and in 2014, which led to increased mortality rates. Conversely, this theory does not claim that some individuals are at higher risk of death due to “despair”, which would be analogous to causes of cases. Therefore, the contrast of interest to test the central claim of Case and Deaton’s theory of “deaths of despair” is a between-population contrast (causes of incidence). As such, this dissertation aims to test the claims of the theory of “deaths of despair” proposed by Case and Deaton at the right level (causes of incidence). I began by conducting a scoping review of the current literature providing empirical support to the different elements of this theory: 1) socioeconomic causes as causes of “despair”, “diseases of despair”, “deaths of despair”, and all-cause mortality, and 2) “despair” as the cause of “diseases of despair”, “deaths of despair”, and all-cause mortality. I found 43 studies that I organized and displayed in two graphs according to Rose’s causes of cases (individual-level causes of “deaths of despair”) and causes of incidence (between-population level causes of “deaths of despair” rates). In each graph, I showed the number of studies that provided evidence for the individual- or population-level elements of the theory of “deaths of despair”. Of these 43 studies, I found that only 13 studies explicitly stated that they tested this theory. Three studies provided different definitions of “despair”, which did not align with the previous vague definition provided by Case and Deaton about white individuals’ hopeless about their prospects for the future. Most studies provided individual-level evidence for “despair” increasing the likelihood of death and despair-related outcomes, which is analogous to a type III error—a mismatch between the research question and the level at which the studies’ design and analyses were conducted to answer that question. Further, no study addressed at the right level—between populations—the central claim of the theory of “deaths of despair”. This led me to review the literature around concepts similar to “despair” and propose a suitable indicator to test the claims of the theory of “deaths of despair”. I leveraged data from the National Health Interview Survey and the Centers for Disease Control mortality data to test whether increases in the prevalence of “despair” were associated with increases in all-cause mortality rates only among white individuals in midlife and whether this effect was bigger among low educated white individuals. To obtain a valid estimate of this association, I adapted econometric methods to develop a valid estimator of the association between increasing “despair” prevalence and increased all-cause mortality rates. After adjusting for potential confounders at the between-population level, I found that the trends in the prevalence of “despair” were negligible across all race and ethnic groups and that an increasing trend could not be identified. Further, I found no evidence that increasing prevalences of “despair” were associated with increased all-cause mortality rates among NH white individuals in midlife, or that this association was more pronounced for those with low education. Lastly, I conducted a similar analysis looking at the association between increased prevalences of “despair” and increased rates of “deaths of despair”. I replicated Case and Deaton’s observed increased rates of “deaths of despair” among white individuals in midlife. However, I found no evidence that increased prevalences of “despair” were associated with increased “deaths of despair” rates among white individuals in midlife or that this association was higher for those with low education. Together, these findings suggest that the claims about the causes of increased mortality rates among white Americans in midlife are at best, questionable, and at worst, false. My aim with this work is to challenge and provide a critical examination of the theory of "deaths of despair", which has fueled the narrative of a suffering white generation and justified recent problematic events as white individuals lashing out for being forgotten to despair and die. While Case and Deaton’s observed rise in mortality rates among whites is a reproducible fact, their narrative ignores other evidence of white racial resentment as the cause of rise in mortality among white individuals. With this work, I intend to help stopping the perpetuation of narratives that favor structural whiteness by promoting an unsubstantiated narrative of psychosocial harm experienced by white Americans. Ultimately, I hope this work helps shift the focus in public health away from Case and Deaton's findings, which may overshadow and detract from the stark reality that mortality rates for Black individuals significantly exceed those for white individuals.

Traumatic events are a common part of the human experience. Posttraumatic stress disorder (PTSD) and major depressive disorder (MDD) are common sequelae of trauma that are both associated with poor physical health and mortality. The objective of this dissertation is to identify common and unique risk factors for each disorder in order to identify at-risk groups for PTSD and/or depression following trauma. his dissertation is organized into five parts: 1) an introduction, 2) a simulation study exploring the use of test equating methods to standardize the Hospital Anxiety and Depression Scale (HADS) to the Beck Depression Inventory II (BDI) in order to create common depression scale across studies in a pooled analysis, 3) an individual participant data meta-analysis on risk factors for PTSD and depression following incident trauma, 4) a Mendelian Randomization analysis of childhood abuse victimization and neuroticism on PTSD and depression, 5) a discussion of the findings and avenues for future research. The simulation study found that an Equated BDI diagnosis had higher specificity (range: 0.86 to 0.91) compared to using the HADS diagnosis (range: 0.80 to 0.82) when the correlation between the BDI and HADS was greater than 0.7, but had lower sensitivity (Equated BDI range: 0.67 to 0.72; HADS range: 0.84 to 0.92). The Equated BDI diagnosis was found to improve statistical power when the prevalence of depression was 20% or higher with greater improvements when the proportion of studies assessing the depression with the HADS was less than 50%. In the individual participant data meta-analysis, common risk factors for acute and persistent MDD and PTSD were found including increased risk for female sex and reduced risk for those who experienced an accident versus an assault or other traumatic event as the index trauma. Acute MDD symptom severity was associated with persistent PTSD and remained significant after inclusion of acute PTSD symptom severity. In an analysis of PTSD symptom clusters, only reexperiencing symptoms were associated with persistent PTSD along with MDD symptom severity. In models of persistent MDD, acute PTSD symptom severity was associated with persistence, but neither overall symptom severity nor cluster symptom severities were associated with persistence after inclusion of acute MDD symptom severity. In the Mendelian Randomization analysis, childhood abuse victimization was found to be associated with PTSD symptom severity but was not associated with an increased odds of a MDD diagnosis, while neuroticism was associated with an increased odds of a diagnosis of MDD, but was not associated with an increase in PTSD symptom severity. Findings from the meta-analysis that leveraged the use of item-response theory imply that while PTSD and MDD share many risk factors for onset of symptoms following the experience of a traumatic event, persistence of symptoms depends most strongly on initial symptoms. However, PTSD and MDD were also found to have different relationships with childhood abuse victimization and neuroticism, indicating that some risk factors are unique to each disorder. Future studies can build upon these results, especially when pooling data from different studies, to further explicate the associations between PTSD, MDD, and their causes.

Introduction: There is growing evidence for increased prevalence of poor adolescent mental health, including depression, in the United States. Increases in adolescent depression beginning around 2008-2010 coincided with the timing of the Great Recession and there are plausible mechanisms through which economic recessions may influence adolescent depression (e.g., caregiver job loss, household economic hardship). More research is needed to understand the potential relationship between the 2007-2009 Great Recession and long-term impacts on mental health by household poverty, as many mechanisms (e.g., cumulative familial stress) can impact adolescent mental health after the peak of a recession passes. The objective of this dissertation is to examine the associations between economic recessions and adolescent depression. This dissertation includes five chapters: first, an introduction; second, a literature review to examine evidence of time trends and birth cohort effects in depressive disorders and symptoms among adolescents in recent years; third, an empirical study to assess changes in adolescent depression and depression treatment, including differences by household poverty occurring at the beginning of the Great Recession; fourth, an empirical study to estimate potential longer-term impacts of the Great Recession by examining whether young adults from birth cohorts who were adolescents at the time of the Great Recession had higher risk of MDE and mental health treatment use as young adults compared with birth cohorts who were adolescents and surveyed prior to the Great Recession with potential differences by household poverty; and fifth, a conclusion to summarize results and discuss implications for future research. Methods: The integrative systematic literature review included 10 studies related to the United States, adolescent populations, birth year and time trends, and depressive symptoms or disorders. The two empirical aims utilized data from the National Survey on Drug Use and Health (NSDUH), a national survey assessing behavioral health among participants aged 12 and older. For the first empirical aim, I analyzed data for adolescents ages 12-17 participating in the 2004-2019 NSDUH (N = 256,572). For the second empirical aim, I included young adults ages 18-29 from the 2005-2019 NSDUH (N = 135,158). For this aim, the main exposure measure was belonging to birth cohorts (1990-1994) who were adolescents during the Great Recession and surveyed in 2008-2019 versus those from birth cohorts (1976-1989) that did not experience the Great Recession and were surveyed prior to the Great Recession in 2005-2007. For both empirical aims, I measured past year DSM-IV and DSM-5 major depressive episodes (MDE) from self-reported symptoms. MDE treatment was assessed among those with past year MDE, excluding those who were already successfully treated for MDE. For the first empirical aim, I tested how MDE and MDE treatment conditioned on MDE changed from pre-Great Recession (2004 to Fall 2007) to post-Great Recession (Winter 2007 to 2019) using interrupted time-series (ITS) segmented regression models accounting for seasonality (January-March, April-June, July-September, October-December) and autocorrelation. For the second empirical aim, regression models assessed the relationships between the birth cohort exposure measure and MDE and mental health treatment utilization adjusting for age, gender, race/ethnicity, educational attainment, and insurance status. Both empirical aims tested effect modification by household poverty. Results: The review of 10 studies found increases in depressive symptoms and disorders in adolescents across recent survey years with increases observed between 1991 and 2020. Of the 3 articles that assessed birth cohort trends, birth cohort trends were less prominent than time period trends. Proposed explanations for increases included social media, economic-related reasons, changes in mental health screening and diagnosis, changes to mental health stigma and treatment and, in more recent years, the COVID-19 pandemic. In the first empirical study, I illustrated that the Great Recession was not associated with an immediate change in MDE prevalence (β: -0.77, 95% CI: -2.23, 0.69). However, following the Great Recession, the increase in MDE prevalence accelerated (β: 0.29, 95% CI: 0.13, 0.44). The Great Recession was not associated with acute changes in adolescent MDE treatment (β: -2.87, 95% CI: -7.79, 2.04) nor longer-term slope effects (β: 0.03, 95% CI: -0.46, 0.51). Evidence of interaction by household poverty was not observed for either the MDE or MDE treatment outcome. In the second empirical aim, interaction between the birth cohort exposure and household poverty was observed for MDE (F=10.17, df=2, p=<0.0001), but not for mental health treatment use. Great Recession exposure effects were stronger among those at higher levels of household income. For example, among young adults who were living in households at two times the poverty threshold, those from birth cohorts who were exposed during adolescence to the Great Recession had higher odds of MDE compared with young adults from birth cohorts who were unexposed during adolescence to the Great Recession (adjusted odds ratio= 1.16, 95% CI= 1.04, 1.29). Conclusions: Multiple cross-sectional surveys and cohort studies documented rising prevalence of depressive symptoms and disorder among adolescents from 1991-2020. The Great Recession coincided with accelerated trends of increasing MDE, but not MDE treatment of these adolescents. Contrary to my hypothesis, the strength of changes in the rate of increase in MDE did not differ by household poverty and adolescents from households living in poverty, who likely experienced a greater financial burden during the recession, did not experience an increase in the rate of MDE. Birth cohort effects by household poverty were observed and exposure to the Great Recession during adolescence was associated with long-term effects on MDE, but not mental health treatment utilization, during young adulthood compared with those not exposed to the Great Recession. Young adults from higher income households who were exposed to the Great Recession had heightened likelihood of MDE. Future research should explore alternative drivers of MDE during the 2010s, as poverty-specific cohort analyses did not show that those living in poverty who likely experienced the greatest burden of a recession financially had increased risk of MDE.