Racial/Ethnic Differences in the Role of Childhood Adversities for Mental Disorders Among a Nationally Representative Sample of Adolescents:

Application of targeted maximum likelihood estimation in public health and epidemiological studies: a systematic review

PURPOSE

The targeted maximum likelihood estimation (TMLE) statistical data analysis framework integrates machine learning, statistical theory, and statistical inference to provide a least biased, efficient, and robust strategy for estimation and inference of a variety of statistical and causal parameters. We describe and evaluate the epidemiological applications that have benefited from recent methodological developments.

METHODS

We conducted a systematic literature review in PubMed for articles that applied any form of TMLE in observational studies. We summarized the epidemiological discipline, geographical location, expertize of the authors, and TMLE methods over time. We used the Roadmap of Targeted Learning and Causal Inference to extract key methodological aspects of the publications. We showcase the contributions to the literature of these TMLE results.

RESULTS

Of the 89 publications included, 33% originated from the University of California at Berkeley, where the framework was first developed by Professor Mark van der Laan. By 2022, 59% of the publications originated from outside the United States and explored up to seven different epidemiological disciplines in 2021-2022. Double-robustness, bias reduction, and model misspecification were the main motivations that drew researchers toward the TMLE framework. Through time, a wide variety of methodological, tutorial, and software-specific articles were cited, owing to the constant growth of methodological developments around TMLE.

CONCLUSIONS

There is a clear dissemination trend of the TMLE framework to various epidemiological disciplines and to increasing numbers of geographical areas. The availability of R packages, publication of tutorial papers, and involvement of methodological experts in applied publications have contributed to an exponential increase in the number of studies that understood the benefits and adoption of TMLE.

Social Determinants Associated With Exposure to Childhood Parental Bereavement and Subsequent Risk for Psychiatric Disorders

IMPORTANCE

Prevalence of childhood parental death varies by race and ethnicity and socioeconomic status, yet whether similar variation persists in the association with lifetime psychiatric disorder is unknown.

OBJECTIVE

To assess whether race and ethnicity and parental educational attainment are associated with the risk of death of a parent; to determine whether the risk for lifetime psychiatric disorder associated with death of a parent was moderated by race and ethnicity and highest parental educational attainment; and to examine a potential intersection of race and ethnicity with parental educational attainment in the risk of lifetime psychiatric disorder associated with death of a parent.

DESIGN, SETTING, AND PARTICIPANTS

This retrospective cohort study used data from the National Comorbidity Study: Adolescent Supplement (NCS-A), 2001 to 2004. Participants included youth aged 13 to 18 years, restricted to Black, Hispanic, and White youth due to power limitations. Data were analyzed from February 26, 2021, to April 21, 2022.

EXPOSURE

Death of a parent during childhood.

MAIN OUTCOMES AND MEASURES

The primary study outcome was any lifetime Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition) psychiatric disorder, assess via assessed via structured clinical interviews.

RESULTS

Among 9501 youth (mean [SD] age, 15.2 [1.5] years; 50.9% female), including 511 youth who had experienced parental death and 8990 youth who had not, the cumulative hazard of parental death by age 18 years was approximately doubled for Hispanic (10.1%; 95% CI, 6.9%-14.7%) and Black (14.0%; 95% CI, 10.6%-18.4%) youth compared with White youth (6.0%; 95% CI, 4.7%-7.8%). Similar patterns were noted by parental educational attainment: the cumulative hazard of parental death for youth of parents with less educational attainment was nearly double (10.1%; 95% CI, 8.1%-12.6%) compared with youth of parents with more education (6.6%; 95% CI, 5.2%-8.4%). Death of a parent was positively and significantly associated with risk of any lifetime psychiatric disorder (aOR, 1.34; 95% CI, 1.03-1.75) compared with youth who had not experienced death of a parent. However, this association was not moderated by race and ethnicity (aOR, 1.05; 95% CI, 0.58-1.92) or parental educational attainment (aOR, 1.19; 95%, 0.70-2.04), although power analyses suggest that larger sample sizes are needed.

CONCLUSIONS AND RELEVANCE

In this cross-sectional study, Black and Hispanic youth experienced elevated parental death compared with White youth, yet the risk for any lifetime psychiatric disorder after parental death was not significantly moderated by race and ethnicity or parental education. Both individual- and population-level interventions may be needed to address the increased risk of psychiatric disorders, although additional studies with larger sample sizes are needed.

The influence of childhood and early adult adversities on substance use behaviours in racial/ethnically diverse young adult women: a latent class analysis

Childhood and adult adversities occur more frequently among women and persons of colour, possibly influencing racial/ethnic disparities in substance use behaviours. This study investigates how childhood and adult adversities cluster together by race/ethnicity and how these clusters predict binge drinking, tobacco, e-cigarette, and marijuana use. Latent class analysis (LCA) was used in a combined sample from the 2015 to 2018 Minnesota College Student Health Survey to identify clusters of childhood and adult adversities among Asian, Black, Latina, and White women aged 18-25. Each substance use outcome was regressed on each adversity cluster across each race/ethnicity group. Across all racial/ethnic groups and substance use outcomes, the high adversity cluster exhibited the greatest risk. Significant racial/ethnic disparities were observed across several substance use behaviours; these were attenuated among women with fewer adversities. The reduced substance use disparities found among those with lower adversities suggest that prevention of adversities may advance health equity.

Sensitive period-regulating genetic pathways and exposure to adversity shape risk for depression

Animal and human studies have documented the existence of developmental windows (or sensitive periods) when experience can have lasting effects on brain structure or function, behavior, and disease. Although sensitive periods for depression likely arise through a complex interplay of genes and experience, this possibility has not yet been explored in humans. We examined the effect of genetic pathways regulating sensitive periods, alone and in interaction with common childhood adversities, on depression risk. Guided by a translational approach, we: (1) performed association analyses of three gene sets (60 genes) shown in animal studies to regulate sensitive periods using summary data from a genome-wide association study of depression (n = 807,553); (2) evaluated the developmental expression patterns of these genes using data from BrainSpan (n = 31), a transcriptional atlas of postmortem brain samples; and (3) tested gene-by-development interplay (dGxE) by analyzing the combined effect of common variants in sensitive period genes and time-varying exposure to two types of childhood adversity within a population-based birth cohort (n = 6254). The gene set regulating sensitive period opening associated with increased depression risk. Notably, 6 of the 15 genes in this set showed developmentally regulated gene-level expression. We also identified a statistical interaction between caregiver physical or emotional abuse during ages 1-5 years and genetic risk for depression conferred by the opening genes. Genes involved in regulating sensitive periods are differentially expressed across the life course and may be implicated in depression vulnerability. Our findings about gene-by-development interplay motivate further research in large, more diverse samples to further unravel the complexity of depression etiology through a sensitive period lens.

Racial Differences in Early Parental Death, Midlife Life Problems, and Relationship Strain With Adult Children

OBJECTIVES

Black Americans typically experience the death of a parent earlier in the life course than do non-Hispanic Whites, and early parental death is known to hinder subsequent relationship outcomes. Whether early parental death may contribute to racial differences in midlife family relationships and the role midlife adults' current life problems play remain unexplored.

METHOD

Using multilevel modeling, we examined how timing of parental death is associated with relationship strain with adult children and whether the association differs by midlife adults' life problems in Black (n = 166) and non-Hispanic White (n = 467) families from the Family Exchanges Study.

RESULTS

Losing a parent in childhood was associated with more relationship strain with adult children for Black midlife adults, but not for their non-Hispanic White counterparts. Among the bereaved, earlier timing of parental death was associated with more relationship strain with adult children only for Black midlife adults. In both bereaved and nonbereaved sample, participants' recent physical-emotional problems exacerbated the link between timing of parental death and relationship strain with adult children for Black midlife adults.

DISCUSSION

Experiencing the death of a parent in the early life course can be an added structural disadvantage that imposes unique challenges for Black Americans in midlife. Policies and programs aimed at supporting bereaved children may benefit relationships with their own children later in life, and addressing physical-emotional problems in midlife may be a viable intervention point for those midlife adults who experienced the death of a parent in the early life course.

A scoping review on the use of machine learning in research on social determinants of health: Trends and research prospects

Background

Machine learning (ML) has spread rapidly from computer science to several disciplines. Given the predictive capacity of ML, it offers new opportunities for health, behavioral, and social scientists. However, it remains unclear how and to what extent ML is being used in studies of social determinants of health (SDH).

Methods

Using four search engines, we conducted a scoping review of studies that used ML to study SDH (published before May 1, 2020). Two independent reviewers analyzed the relevant studies. For each study, we identified the research questions, Results, data, and algorithms. We synthesized our findings in a narrative report.

Results

Of the initial 8097 hits, we identified 82 relevant studies. The number of publications has risen during the past decade. More than half of the studies (n = 46) used US data. About 80% (n = 66) utilized surveys, and 70% (n = 57) employed ML for common prediction tasks. Although the number of studies in ML and SDH is growing rapidly, only a few studies used ML to improve causal inference, curate data, or identify social bias in predictions (i.e., algorithmic fairness).

Conclusions

While ML equips researchers with new ways to measure health outcomes and their determinants from non-conventional sources such as text, audio, and image data, most studies still rely on traditional surveys. Although there are no guarantees that ML will lead to better social epidemiological research, the potential for innovation in SDH research is evident as a result of harnessing the predictive power of ML for causality, data curation, or algorithmic fairness.

What is Machine Learning? A Primer for the Epidemiologist

Machine learning is a branch of computer science that has the potential to transform epidemiologic sciences. Amid a growing focus on "Big Data," it offers epidemiologists new tools to tackle problems for which classical methods are not well-suited. In order to critically evaluate the value of integrating machine learning algorithms and existing methods, however, it is essential to address language and technical barriers between the two fields that can make it difficult for epidemiologists to read and assess machine learning studies. Here, we provide an overview of the concepts and terminology used in machine learning literature, which encompasses a diverse set of tools with goals ranging from prediction to classification to clustering. We provide a brief introduction to 5 common machine learning algorithms and 4 ensemble-based approaches. We then summarize epidemiologic applications of machine learning techniques in the published literature. We recommend approaches to incorporate machine learning in epidemiologic research and discuss opportunities and challenges for integrating machine learning and existing epidemiologic research methods.

Distress level and daily functioning problems attributed to firearm victimization: sociodemographic-specific responses

PURPOSE

The purpose of this study was to estimate the effect of firearm involvement during violent victimization on the level of distress experienced and daily functioning within sociodemographic subgroups.

METHODS

We used cross-sectional data from the National Crime Victimization Survey (n = 5698) and Targeted Maximum Likelihood Estimation. Sociodemographic subgroups were defined by age, race, sex, and socioeconomic position. Outcomes included experiencing the victimization as severely distressing and problems in the workplace or at school, or with peers or family.

RESULTS

Among people victimized with a firearm, nearly 40% experienced the victimization as severely distressing and 28% reported daily functioning problems as a result of the victimization, compared with 25% and 27% of those victimized without a firearm. In most of the subgroups examined, a greater proportion of people described the event as severely distressing when a firearm was involved in the victimization, ranging up to 19 percentage points higher among women and among black respondents (95% CI for women = 10%-28%; for blacks = 6%-31%).

CONCLUSIONS

Our findings suggest an almost universal negative response to firearm involvement during a violent victimization as compared with violent victimizations involving other or no weapons. These findings highlight the need for efforts by medical and mental health practitioners to address the potential sequelae of experiencing severe distress during a firearm victimization.

An educational intervention to improve knowledge about prevention against occupational asthma and allergies using targeted maximum likelihood estimation

PURPOSE

Occupational asthma and allergies are potentially preventable diseases affecting 5-15% of the working population. However, the use of preventive measures is often insufficient. The aim of this study was to estimate the average treatment effect of an educational intervention designed to improve the knowledge of preventive measures against asthma and allergies in farm apprentices from Bavaria (Southern Germany).

METHODS

Farm apprentices at Bavarian farm schools were asked to complete a questionnaire evaluating their knowledge about preventive measures against occupational asthma and allergies (use of personal protective equipment, personal and workplace hygiene measures). Eligible apprentices were randomized by school site to either a control or an intervention group. The intervention consisted of a short educational video about use of preventive measures. Six months after the intervention, subjects were asked to complete a post-intervention questionnaire. Of the 116 apprentices (70 intervention group, 46 control group) who answered the baseline questionnaire, only 47 subjects (41%; 17 intervention group, 30 control group) also completed the follow-up questionnaire. We, therefore, estimated the causal effect of the intervention using targeted maximum likelihood estimation. Models were controlled for potential confounders.

RESULTS

Based on the targeted maximum likelihood estimation, the intervention would have increased the proportion of correct answers on all six preventive measures by 18.4% (95% confidence interval 7.3-29.6%) had all participants received the intervention vs. had they all been in the control group.

CONCLUSIONS

These findings indicate the improvement of knowledge by the educational intervention.

Adolescent Predictors of Incidence and Persistence of Suicide-Related Outcomes in Young Adulthood: A Longitudinal Study of Mexican Youth

In recent years, Mexico has seen one of the largest increases in suicide rates worldwide, especially among adolescents and young adults. This study uses data from the 1,071 respondents who participated in a two-wave longitudinal study when they were between 12 and 17 years of age, and again when they were between 19 and 26 years of age. The World Mental Health Composite International Diagnostic Interview assessed suicidal behavior and DSM-IV mental disorders. We used Cox regressions to evaluate which sociodemographic and psychiatric factors and life events predicted the incidence and remission of suicide ideation, plan, and attempt throughout the 8-year span. The 8-year incidence of suicide ideation, plan, and attempt was 13.3%, 4.8%, and 5.9%, respectively. We found that the number of traumatic life events during childhood, no longer being in school, and tobacco use predicted which adolescents developed suicide behaviors as they transitioned into young adulthood. Psychiatric disorders, particularly anxiety disorders, played a larger role in the persistence of those who already had suicidal behaviors, while behavioral disorders played a role in the transition from ideation to attempt. This distinction may be useful for clinicians to assess the risk of suicide.

Parenting style and mental disorders in a nationally representative sample of US adolescents

PURPOSE

We examined associations between parenting style and past-year mental disorders in a nationally representative cross-sectional survey of US adolescents and whether the associations differed by adolescent demographic characteristics.

METHODS

The sample included 6483 adolescents aged 13-18 years who were interviewed for a full range of DSM-IV mental disorders. Parenting style was assessed by adolescent-reported maternal and paternal care and control using items from the Parental Bonding Instrument. We controlled for socio-demographics, parental history of mental disorders, stressful life events, sexual violence, inter-parental conflict, and household composition. We also tested for two-way interactions between parental care and control and adolescent age, sex, and race/ethnicity.

RESULTS

In adjusted models, high maternal care was associated with lower odds of depressive, eating, and behavioral disorders, and high maternal control was associated with greater odds of depressive, anxiety, eating, and behavioral disorders. High paternal care was associated with lower odds of social phobia and alcohol abuse/dependence. High paternal control was associated with greater odds of agoraphobia and alcohol abuse/dependence but with lower odds of attention-deficit/hyperactivity disorder. Associations of maternal and paternal control with anxiety disorders and substance abuse/dependence differed by sex. High paternal care was associated with lower odds of anxiety disorders only among Hispanics and non-Hispanic blacks.

CONCLUSIONS

Perceived parental care and control were associated with adolescent mental disorders after controlling for multiple potential confounders. Differential patterns of association were found according to adolescent sex and race/ethnicity. Findings have implications for prevention and intervention programs that incorporate familial contextual factors.

Review of Recent Methodological Developments in Group-Randomized Trials: Part 2-Analysis

In 2004, Murray et al. reviewed methodological developments in the design and analysis of group-randomized trials (GRTs). We have updated that review with developments in analysis of the past 13 years, with a companion article to focus on developments in design. We discuss developments in the topics of the earlier review (e.g., methods for parallel-arm GRTs, individually randomized group-treatment trials, and missing data) and in new topics, including methods to account for multiple-level clustering and alternative estimation methods (e.g., augmented generalized estimating equations, targeted maximum likelihood, and quadratic inference functions). In addition, we describe developments in analysis of alternative group designs (including stepped-wedge GRTs, network-randomized trials, and pseudocluster randomized trials), which require clustering to be accounted for in their design and analysis.