Using baseline target moderation to guide decisions on adapting prevention programs

Peer Network Counseling Effects on Substance Use: an Individual Participant Data Meta-Analysis Integrating Three Randomized Controlled Trials

The current study describes an individual participant data meta-analysis (IPDMA) testing the efficacy of a peer-network counseling (PNC) intervention for preventing substance use escalation in adolescents and young adults. PNC has shown efficacy in reducing substance use among adolescents and young adults across small-scale randomized controlled trials (RCTs). Identifying expected large-scale effects and moderators is an important next step in guiding use of PNC in practice. To this end, we combine three small-scale RCTs to test PNC intervention effects on substance use change in a combined sample of 421 adolescents and young adults (50% intervention, 55% female, 69% Black/African-American, M age [SD] = 17.3 [2.2] years). Our approach combines latent change score modeling in a structural equation modeling (SEM) framework with study-level fixed effects to obtain (a) a more generalizable PNC effect than we could obtain with each constituent sample and (b) greater power and precision for individual-level moderation of treatment effects. We found that although PNC main effects on substance use outcomes (past 30-day cannabis, alcohol, tobacco, and drug use) were not significant, PNC effects were moderated by individual-level pre-intervention substance use frequency. PNC more strongly reduced drug use at the 1-month follow-up and cannabis use at the 3-month follow-up among participants who showed higher baseline use of these substances. Implications of our approach and findings for prevention researchers are discussed.

Estimating latent baseline-by-treatment interactions in statistical mediation analysis

Statistical mediation analysis is used to uncover intermediate variables, known as mediators [ M ], that explain how a treatment [ X ] changes an outcome [Y ]. Often, researchers examine whether baseline levels of M and Y moderate the effect of X on posttest M or Y . However, there is limited guidance on how to estimate baseline-by-treatment interaction (BTI) effects when M and Y are latent variables, which entails the estimation of latent interaction effects. In this paper, we discuss two general approaches for estimating latent BTI effects in mediation analysis: using structural models or scoring latent variables prior to estimating observed BTIs and correcting for unreliability. We present simulation results describing bias, power, type 1 error rates, and interval coverage of the latent BTIs and mediated effects estimated using these approaches. These methods are also illustrated with an applied example. R and Mplus syntax are provided to facilitate the implementation of these approaches.

Baseline Targeted Moderation in a Trial of the Family Check-Up 4 Health: Potential Explanations for Finding Few Practical Effects

Not all participants will benefit equally from even well-established, evidence-based prevention programs. For this reason, the field of prevention science is beginning to embrace individual tailoring of interventions. The Family Check-Up was among the first prevention programs to tailor at the family level as opposed to the more prevalent focus on adapting programs for different cultures, genders, and other immutable participant characteristics. Despite tailoring, families with lower levels of stress and parental mental health issues, children with lower baseline conduct problems, and families living in an extremely deprived neighborhood benefitted less from the Family Check-Up. This study examined baseline targeted moderation (BTM) within a trial of the Family Check-Up 4 Health (FCU4Health) program, an adaptation of the Family Check-Up for primary care delivery and explicit targeting of obesogenic behaviors. Ethnically diverse, low-income families (N = 240) with children ages 5.5 to 12 years identified in pediatric primary care with elevated body mass index (BMI) were enrolled and randomized to FCU4Health or usual care. Few BTM effects were found using single-variable-as-moderator and latent-class-as-moderator analytic approaches across the primary (child BMI, body composition) and secondary outcomes (family health routines; child eating behaviors, food choices, emotional problems, problem behaviors, quality of life; caregiver BMI and body composition), as well as hypothesized mediators (child self-regulation, parenting skills). The high-risk nature of the sample and the FCU4Health being individually tailored might have mitigated finding BTM effects. This trial was prospectively registered (NCT03013309 ClinicalTrials.gov).

Exploring the Mechanisms in Cognitive Behavioural Therapy for Anxious Children: Does Change in Emotion Regulation Explain Treatment Effect?

Cognitive behavioural therapy (CBT) for childhood anxiety has shown moderate effects. However, inconsistency in findings during the last decades of treatment research and lack of measurable treatment gains over time has led to a call for optimizing interventions by identifying the active mechanisms involved and for whom such interventions are effective. It has been suggested that the moderate effects may be explained by the fact that emotion regulation rarely is directly targeted in CBT-interventions and that interventions may be more effective for children with a certain level of problems with emotion regulation. Using data from a randomized controlled trial with 160 children and their mothers at baseline (t1) and posttreatment (t2), we examined whether being randomized to CBT predicted change in anxiety symptoms from t1 to t2 and whether this change was mediated by change in emotion regulation from t1 to t2. We also investigated whether the strength of this indirect pathway depended on the level of emotion regulation problems at baseline. Latent baseline target moderated mediation analyses within a structural equation modelling framework were conducted. Results showed a significant indirect pathway between receiving CBT to improved emotion regulation, which again was significantly associated with reductions in anxiety symptoms. The findings suggest that the effect of CBT is similar for children irrespective of initial levels of emotion regulation measured broadly, whereas there was some evidence of a baseline moderation effect of the subdomain emotional control. Emotion regulation, and especially emotional control, seems to be an underlying mechanism for positive effects of CBT for anxiety disorders in children, possibly indicating that a greater emphasis on emotion regulation may optimize the intervention.

When Is It Time to Revise or Adapt Our Prevention Programs? Introduction to Special Issue on Using Baseline Target Moderation to Assess Variation in Prevention Impact

Adaptation of preventive interventions to increase their impact can be advanced by identifying subgroups or specific contexts where program effects appear stronger or weaker. But how do we know where to look for effect heterogeneity in ways that will inform successful adaptation? This paper introduces a special issue that brings together research across prevention science sub-disciplines that adopted baseline target moderated mediation (BTMM) designs to search for effect heterogeneity and guide adaptation of established prevention programs. For this special issue intervention  scientists analyzed data from randomized trials using BTM and BTMM models, evaluating evidence for variation in intervention impact for trials spanning different health outcomes, different developmental periods, and different social units. This introduction provides a brief summary of the various patterns of effect reported in these papers, noting that the most common pattern involved compensatory effects (those beginning the trial with greater risk or fewer protective factors benefit the most), but other patterns including rich-get-richer and partially iatrogenic effects were also detected. This paper ends with a discussion of methodological and substantive implications of these findings for future prevention research, including next-generation prevention trials.

Familias Unidas Prevents Youth Internalizing Symptoms: a Baseline Target Moderated Mediation (BTMM) Study

Family-based preventive interventions have been found to prevent youth internalizing symptoms, yet they operate through diverse mechanisms with heterogeneous effects for different youth. To better target preventive interventions, this study examines the effects of the Familias Unidas preventive intervention on reducing internalizing symptoms with a universal sample of Hispanic youth in a real-world school setting (i.e., effectiveness trial). The study utilizes emerging methods in baseline target moderated mediation (BTMM) to determine whether the intervention reduces internalizing symptoms through its impact on three distinct mechanisms: family functioning, parent stress, and social support for parents. Data are from a randomized controlled effectiveness trial of 746 Hispanic eighth graders and their parents assessed at baseline, 6-, 18-, and 30-month post-baseline. BTMM models examined three moderated mechanisms through which the intervention might influence 30-month adolescent internalizing symptoms. The intervention decreased youth internalizing symptoms through improvements in family functioning in some models, but there was no evidence of moderation by baseline level of family functioning. There was some evidence of mediation through increasing social support for parents for those intervention parents presenting with lower baseline support. However, there was no evidence of mediation through parent stress. Post hoc analyses suggest a possible cascading of effects where improvements in support for parents strengthened parental monitoring of youth and ultimately reduced youth internalizing symptoms. Findings support the intervention's effects on internalizing symptoms in a universal, real-world setting, and the value of BTMM methods to improve the targeting of preventive interventions. ClinicalTrials.gov Identifier: NCTO1038206, First Posted: December 23, 2009.

Family Functioning in Hispanic Parents of Adolescents: Who Benefits Most from a Family-Based HIV and Substance Use Preventive Intervention?

To understand which families are likely to benefit most from resource-intensive family-based, evidence-based interventions (EBIs), we must examine the key, modifiable determinant of family functioning. The purpose of this study was to (1) identify whether there are subgroups of Hispanic parents that differ meaningfully based on their family functioning at baseline, (2) test whether the Familias Unidas preventive intervention was differentially effective across the baseline family functioning subgroups, and (3) understand the mechanisms of intervention effectiveness within each baseline family functioning subgroup. On a pooled data set of 4 completed efficacy and effectiveness trials of Familias Unidas (n = 1445 low-income, Hispanic immigrant origin, parents and their adolescent between the ages of 12-17), we conducted a series of secondary data analyses. Latent profile analyses revealed four significantly different profiles: (1) low family functioning (n = 210, 14.55%), (2) low-to-moderate family functioning (n = 554, 38.39%), (3) moderate-to-high family functioning (n = 490, 33.96%), and (4) high family functioning (n = 189, 13.10%). A structural equation modeling approach found there were significant differences in intervention effectiveness between the subgroups. The low family functioning subgroup experienced gains in family functioning, and in turn, lower levels of adolescent substance use, internalizing, and externalizing symptoms. The high family functioning subgroup showed significant direct effects of the intervention on adolescent substance use, internalizing, and externalizing symptoms, but no indirect effects through improvements in family functioning. Implications for screening, targeting, and adapting interventions are discussed.

Heterogeneity in the Effects of Interventions to Prevent Depression in Couples Facing Job Loss: Studying Baseline Target Moderation of Impact

Couples' communication styles are associated with depression following job loss for both job seekers and their partners. The Couples Employment Program (CEP), an eight-session program for couples facing job loss, was developed to integrate job search strategies sessions from the JOBS program with couple communication sessions, targeting job search behavior, motivation, mastery, and couple communication. We hypothesized that CEP would have compensatory effects, such that those who began the program with lower job search behavior, lower motivation, less mastery, and more negative or less positive couple communication would make more gains on these targets, and this would mediate impact on reducing risk for depression. We conducted a randomized field trial of CEP with 1477 heterosexual couples facing recent unemployment. Baseline levels of job search behavior and motivation, but not mastery or depression, moderated the impact of intervention on job seeker depression slopes over 12 months; job seekers reporting less job search behavior and motivation at baseline benefited more. Male partners with higher baseline depression also benefited. Opposite to our hypothesis, baseline levels of couple's communication moderated the impact of intervention such that partners in couples with more negative and less positive communication showed iatrogenic effects. There was no evidence that baseline target levels moderated the impact of the intervention on any of these targets. We speculate that more intensive communication training may be necessary for positive impact.

Commentary for Special Issue on Using Baseline Target Moderation to Assess Variation in Prevention Impact: When (and How) to Revise Our Programs

Developing a better understanding of sources and mechanisms of heterogeneity is a key route to improving outcomes and targeting of preventive interventions. This commentary attempts to draw together findings from eight intervention trials in this special issue, each exploring baseline target moderation (BTM) or baseline target moderated mediation (BTMM). It considers their implications for prevention research and program design, particularly the question of whether they can help us to revise or adapt interventions. The studies cover a range of interventions, targets, and contexts, including parenting, couple, and CBT interventions, for depression, anxiety, conduct problems, or obesity. Some important findings stand out. Where studies found moderator effects, they tended to operate in a "compensatory" fashion, such that greater benefit was found in higher risk groups, suggesting that closer targeting might be warranted. It was rare for harmful effects to be detected for any subgroups. In other respects, patterns of BTM/BTMM findings were quite mixed across studies, suggesting it would be premature to change our interventions based on these trials. Implications of the findings for equity, for "slimming down" and scaling up interventions, and for research are discussed, including the need to combine BTMM with intervention component research, and to accumulate a more robust body of evidence by pooling data across trials.

Secondary Analysis to Inform the Development of Adaptive Preventive Interventions

For the past 30 years, scholars across the fields of epidemiology, education, psychology, and numerous other fields have worked to develop interventions designed to reduce risk and enhance protection to prevent mental, emotional, and behavioral problems across the lifespan. This article presents a series of next steps that leverage this foundational science to inform the development of adaptive preventive interventions. Adaptive preventive interventions (APIs) tailor the intervention to fit the diverse, sometimes changing, needs of participants with the goal of better prevention outcomes for more individuals. Secondary analyses of data from preventive intervention trials to identify moderators, mediators, and antecedents of attrition and intervention failure can be useful for designing effective APIs. Moderators that identify intervention effect heterogeneity can be used within an API to tailor the intervention to meet the unique needs of important participant subgroups. Mediators and predictors of disengagement and attrition can be helpful tailoring variables in an API to trigger change to the intervention. Preventive intervention trials that incorporate frequent assessment of potential mediators, moderators, and antecedents of attrition during the intervention period are needed. Secondary analyses of data from preventive intervention trials provide an important foundation for next-generation APIs.

Moderators of Psychosocial Intervention Response for Children and Adolescents with Conduct Problems

Objective: The purpose of this Brief Report is to synthesize the current evidence base examining moderators of psychosocial intervention response for children and adolescents with conduct problems (CP). We also provide directions for future research.Method: We focused on four categories of psychosocial interventions for the prevention and/or treatment of CP: (1) parent management training (PMT) for children, (2) other family-based interventions for adolescents, (3) youth skills training, and (4) multicomponent interventions (i.e., family-based intervention plus skills training). Emphasis is placed on findings from meta-analyses.Results: Moderation analyses have occurred more frequently for PMT than for other types of interventions. Variables for which there was consistent evidence for positive moderation included higher initial severity of CP, father engagement, higher maternal depressive symptoms, individual administration (vs. group), and treatment/targeted prevention approaches (vs. universal prevention). Variables where there was evidence for no moderation (demonstrating generalizability) included child diagnostic status and family risk in PMT, and diagnostic status and intervention setting for skills training. However, for some variables, evidence of moderation was dependent on intervention type.Conclusions: Future research should examine multiple moderators in combination; incorporate innovative techniques such as integrative data analyses, individual participant data, and class-based modeling, which may identify moderator effects that are undetected by more traditional variable-oriented moderation analyses; and conduct moderated mediation models for informing developmental theory on the interplay of risk and protective factors.