Session: Pooling Data Across Multiple Randomised Trials to Investigate Equity Effects of Youth Interventions (Society for Prevention Research 24th Annual Meeting)

3-060 Pooling Data Across Multiple Randomised Trials to Investigate Equity Effects of Youth Interventions

Thursday, June 2, 2016: 3:00 PM-4:30 PM
Grand Ballroom B (Hyatt Regency San Francisco)
Theme: Promoting Health Equity Among Populations at Risk
Symposium Organizer:
Frances Gardner
Health inequity can be defined as disparities in health or health care outcomes that are systemic and avoidable and, therefore, considered unfair. Issues of equity are important in prevention science because of the possibility that our effective interventions, which show average benefit to a population, may at the same time have the unintended consequence of increasing social disparities in outcome.  This can arise if disadvantaged subgroups benefit differentially less well from interventions.  A key approach for examining equity effects is to investigate how different forms of social disadvantage moderate intervention outcomes. However, there are considerable challenges in studying moderator effects. Firstly, many randomized trials, although well-powered for testing main effects, lack adequate power for secondary moderator analyses. Secondly, aggregating studies via systematic review and meta-analysis is a common solution for identifying moderator effects, but by relying solely on trial level moderators (e.g. mean SES level in a sample), conventional meta-analysis fails to make use of rich within-trial variability in participant characteristics. In effect, the sample size is only as large as the number of trials. Hence, there is a pressing need to study moderators in samples with sufficient power, which can be achieved by pooling individual-level data across trials (Hendricks Brown et al., 2013).  

This symposium brings together data from a total of 38 randomized trials in 7 countries, of three different youth interventions. For each intervention type, investigators pooled individual-level data from multiple trials in order to i) investigate whether there are differential effects of ethnicity, or SES and other social disadvantages, on intervention outcome; ii) to explore multiple methods for enhancing the quality of harmonised data, and reducing bias, while examining the invariance (or otherwise) of different models across SES and ethnicity. The 3 distinct interventions include: preventive interventions to reduce risk for adolescent depression; multidimensional family therapy for adolescent drug use; and a parenting intervention for preventing child conduct disorder.

This symposium presents new moderator findings from uniquely large data sets, potentially generalizable across multiple service settings and countries. By bringing together cutting edge methodologies, the findings enhance our insight into the extent to which preventive interventions promote equity, as well as bringing average benefits to the population. At the same time, these studies address major methodological challenges in data harmonisation, essential for well-powered investigation of differential effects.

2723 characters

* noted as presenting author
To What Extent Are Parenting Interventions for Conduct Problems Likely to Affect Social Disparities in Youth Outcomes? Pooling Data from Multiple Trials of Parenting Interventions
Frances Gardner, DPhil, University of Oxford; Patty Leijten, PhD, University of Amsterdam; Stephen Scott, PhD, Institute of Psychiatry; Sabine Landau, PhD, Institute of Psychiatry; Joanna Mann, PhD, Oxford University; Victoria Harris, PhD, Institute of Psychiatry; Judy Hutchings, PhD, Bangor University; Jennifer Beecham, PhD, London School of Economics
Bias Correction in Integrative Data Analysis and Application in Evaluating Differentiated Effects on Subpopulations of an Adolescent Substance Treatment Program
Wei Wang, PhD, University of South Florida; Paul Ellis Greenbaum, PhD, University of South Florida; Craig E. Henderson, PhD, Sam Houston State University; Chen Henian, PhD, University of South Florida
Comparing IRT, Latent Variable, and Effect-Size Methods for Harmonizing Adolescent Depression Measures in Synthesis Datasets, and Testing Invariance Across SES and Ethnicity
George W. Howe, PhD, George Washington University; Tatiana Perrino, PsyD, University of Miami Miller School of Medicine; Hilda M. Pantin, PhD, University of Miami Miller School of Medicine; Ahnalee Brincks, PhD, University of Miami; C. Hendricks Brown, PhD, Northwestern University; Getachew A. Dagne, PhD, University of South Florida