AIDSImpact.com:: - Santa Fe, USA 2011

Abstract #283 - Trajectories of multiple risk behaviors in adolescents: A multivariate growth-mixture modeling approach
Authors: Presenting Author: Dr. Brian Mustanski - Northwestern University
Additional Authors: Dr. Gayle Byck, Dr. Allison Dymnicki, Dr. David Henry, Dr. John Bolland,
Aim: Behaviors that put a youth at risk for HIV tend to co-occur with other risk-taking behaviors (i.e. substance use, conduct problems). Some have referred to this as “problem behavior syndrome” or the related adverse outcomes as a “syndemic.” The current study focused on the clustering of sexual risk taking, substance use, and conduct problems among very low-income African American adolescents in the southern U.S. This particular population is at an increased risk for HIV infection compared to other ethnic, income, and geographic groups. We advance understanding of the clustering of these risk behaviors by making use of longitudinal data and advanced statistical approaches to identify trajectories of risk behaviors across development and implications of trajectories for negative outcomes.
Method / Issue: Data came from the Mobile Youth Study (MYS), which has assessed thousands of 9-18 year old African American youth living in high-poverty neighborhoods annually since 1998. For this analysis, we included adolescents age 12-18 who had at least three waves of data between 1998 and 2006 and had valid data at age 12 or 18; N = 1,406. We had information for 61% of the sample for 3-4 waves, 34% for 5-6 waves, and 5% for all seven waves. MPlus estimated all missing data values using the EM algorithm to obtain maximum likelihood estimates with robust standard errors. Composite variables were created from multiple items assessing sexual behavior, substance use, and conduct problems. Our model was similar in approach to parallel process growth mixture models, but here three outcomes were included.
Results / Comments: The four-class model provided the best fit (log likelihood=-26876.81; BIC=54514.70; entropy=0.85). The sexual risk, conduct problems, and substance use trajectories of the four classes can be characterized as follows: (1) increasing high risk in all three areas (11.9% of sample; “increasing high risk takers”); (2) increasing to high sexual risk, adolescent-limited conduct problems, and high adolescent-limited substance use (8.0%; “adolescent-limited”); (3) consistent moderate-high sex risk, steadily decreasing conduct problems, and high to low to increasing drug risk (5.8%; “early experimenters”); and (4) steadily increasing sexual risk with minimal conduct problems and minimal drug risk (74.3%; “normative low risk”). Significant differences (Ps< .001) in outcomes related to class memberships included arrest, pregnancy, and STD diagnosis. The “normative low risk” group was almost always lowest at each age for each of the outcomes, while the “increasing high risk takers” were almost always showed the highest proportion.
Discussion: Our analyses identified four distinct trajectories of multiple risk behaviors. Most youth fit into the “normative low risk” group, which demonstrates the resiliency of these youth growing up in disadvantaged neighborhoods. All groups showed an increase in sexual behavior across the 6 years of development, which is consistent with other developmental studies. Two groups started at higher levels of risk, but diverged into an “adolescent-limited” and an “increasing” group. Future research should help characterize the factors that lead to the divergence of these two groups as the group that continues to increase is at particularly need of intervention to prevent adverse outcomes, such as HIV infection.
Go Back