Patterns of Personal Networks and Substance Use Among Substance-Using Women

Background/Rationale

The purpose of the present study is to apply a person-centered approach using Latent Class Analysis (LCA) to examine the patterns of personal networks and their relationships to substance use post treatment among low-income women with substance use disorders (SUD).  Previous studies have utilized variable-centered approaches (e.g., multiple regressions) to examine the relative contribution of each personal network characteristic after controlling for other personal network characteristics.  However, literature suggests personal network characteristics are often interrelated to each other, which may contribute the mixed findings regarding the effect of personal networks on treatment outcomes.  These mixed findings might be related to non-linear interaction among personal network characteristics   Therefore, the variable-centered approaches might be limited in assessing the complexity of personal networks.

Methods/Methodology

The present study included 272 substance-using women interviewed at 1 week and 1, 6, and 12 months post intake at three treatment programs.  Personal network characteristics were assessed using EgoNet software.  Women reported 25 network members and their characteristics such as how known (e.g., family, treatment-related people), quality of relationship (e.g., support, reciprocity), and structure (e.g., density) of relationships.  Substance use by 12 months post intake was utilized as a distal outcome of the study

LCA was employed to explore the underlying patterns of personal networks using eight personal network characteristics simultaneously (e.g., family, substance-using network members, emotional support, reciprocal relationships, critical people, and density).  The model fit was assessed by the Bayesian Information Criteria, the Bootstrap Likelihood Ratio Test, solution stability, and the interpretability for practical purposes.  LCA with a distal outcome was performed to examine the effect of underlying patterns of personal networks on substance use by12 months post intake.

Results

The majority of women were African American (63%), never married (67%), and received government assistance (75%), with a mean age of 36.5 (SD=10.4).  Approximately three fourth of respondents reported the presence of dual disorders (73%) and previous treatment history (73%).

LCA identified three underlying subgroups of women who had similar personal network characteristics: family-centered networks (36.7%), treatment-centered networks (26.2%), and negative networks (37.1%).  Compared to women with either family-centered networks or treatment-centered networks, those with negative networks reported more substance-using people, more critical people, fewer close people, and fewer supportive people.  Only women with family-centered networks reported significantly more reciprocal relationships and greater density compared to those with negative networks.

LCA with a distal outcome found that women with negative networks were more likely to relapse (66.6%) by 12 months post treatment intake than those with either family-centered networks (26.2%) or treatment-centered networks (21%). 

Conclusions

The findings highlight the underlying patterns of personal network characteristics among women with SUD.  The findings also indicate that compared to women with negative networks, those with either family-centered networks or treatment-centered networks were associated with decreased odds of substance use by 12 months post intake, suggesting multiple ways to improve treatment outcomes through personal network interventions.  Future research is needed to examine how these underlying patterns of personal networks are associated with other treatment outcomes in recovery.