Statistical Power of Alternative Structural Models for Comparative Effectiveness Research: Advantages of Modeling Unreliability

Association of adverse childhood experiences (ACEs) and substance use disorders (SUDs) in a multi-site safety net healthcare setting

•

Adverse childhood experiences are highly correlated with substance use disorders.

•

Each individual ACE significantly predicted an SUD.

•

In a large sample with high ACEs, no dose-response relationship was found.

•

The majority of female patients had an ACE score ≥ 4.

•

The prevalence of different ACEs varied significantly by race/ethnicity.

Background

Adverse childhood experiences (ACEs) and substance use disorders (SUDs) are highly prevalent public health challenges that have been shown to be strongly correlated. Although previous research has suggested a dose-response relationship between ACEs and SUDs, less is known about this phenomenon and the prevalence of ACEs in lower income, racially/ethnically diverse populations. This study sought to examine these relationships in a population treated at a multi-site safety net provider.

Methods

The ACEs survey was delivered as a standard assessment to all behavioral health patients seen at a large Federally Qualified Health Center (FQHC) in Connecticut. 4378 patients completed the questionnaire. Both total score and individual ACE questions were correlated with diagnostic history, according to chi-square and multiple-group structural equation modeling tests.

Results

84.8% of patients reported at least one ACE and 49.1% had an ACE score ≥ 4. Experiencing 1 or more ACEs predicted having any SUD, after controlling for race/ethnicity and gender. Parent substance use, physical abuse, and sexual abuse in particular were the strongest predictors of developing any SUD. Men and non-white individuals were more likely to develop an SUD with lower ACE scores than women and white individuals.

Conclusions

While ACEs predict an increased likelihood of developing any SUD, the nature of this relationship differs by both gender and race/ethnicity. In this FQHC patient population there is no obvious dose-response relationship between ACEs and SUDs. Additional research is required to help understand why the relationship between ACEs and SUDs observed here differs from other populations.

African-American, Puerto-Rican, and Other Hispanic Fathers' Differential Responses to a Parenting Intervention

Despite the recognized importance of fathers to children's well-being, there is a lack of research exploring the impact of parenting interventions on young fathers. Further, little work has been done to identify whether fatherhood interventions differentially benefit specific subgroups of fathers, including Hispanic subgroups.This research examines a 15-week fatherhood intervention for African American, Puerto Rican, and non-Puerto Rican Hispanic young fathers. Data were collected from 312 fathers ages 15-24 at baseline, post-intervention (15 weeks), and at 8 months, 12 months, and 16 months post-baseline follow-up. Latent growth models were used to examine differential shapes of change for the three groups. Results suggest different intervention responses across the three groups on risky parenting attitudes, African American fathers in the study indicated more risky parenting attitudes than Hispanic and Puerto Rican fathers. The practice and research implications of disaggregating data for minority fathers, particularly for Latino subgroups, are discussed.

A non-randomised pragmatic trial of a school-based group cognitive-behavioural programme for preventing depression in girls

The aim of this study was to investigate the effectiveness of the DISA-programme in preventing depressive symptoms (DS) in adolescent girls, as implemented in a real-world school setting, accounting for baseline socioeconomic and psychosocial factors, and to investigate whether the effects of these baseline variables on DS differed between intervention participants and non-participants. In this non-randomised pragmatic trial, an electronic questionnaire was disseminated in 2011 (baseline) and 2012 (follow-up) in schools in one municipality in northern Sweden. Pupils (total n=275; intervention participants identified in the questionnaire: n=53; non-participants: n=222) were 14–15 years old at baseline. The groups were compared by means of SEM. DISA could not predict differences in DS at follow-up in this real-life setting. In the overall sample, sexual harassment victimisation (SH) at baseline was associated with DS at follow-up and the estimate for SH increased in the DISA-participants compared to the overall sample.

A Review of Graphical Approaches to Common Statistical Analyses: The Omnipresence of Latent Variables in Statistics

We provide a comprehensive review of simple and advanced statistical analyses using an intuitive visual approach explicitly modeling Latent Variables (LV). This method can better illuminate what is assumed in each analytical method and what is actually estimated, by translating the causal relationships embedded in the graphical models in equation form. We recommend the graphical display rooted in the century old path analysis, that details all parameters of each statistical model, and suggest labeling that clarifies what is given vs. what is estimated. We link in the process classical and modern analyses under the encompassing broader umbrella of Generalized Latent Variable Modeling, and demonstrate that LVs are omnipresent in all statistical approaches, yet until directly 'seeing' them in visual graphical displays, they are unnecessarily overlooked. The advantages of directly modeling LVs are shown with examples of analyses from the ActiveS intervention designed to increase physical activity.