Self-Reported Barriers to Substance Use Treatment and Recovery Support Services: A Latent Class Analysis

Background and Purpose: People at risk of an opioid overdose can experience multiple barriers to engaging in recovery support services. This study explores heterogeneity in self-reported treatment barriers among adults who use substances. Barriers were inclusive of motivational, logistical, service-related, and health-related barriers that participants cited for missing at least one treatment session in the past 30 days. Behavioral health and sociodemographic correlates of class membership are identified.

Methods: Baseline data were collected from adults (N=291) enrolled in recovery support services in the Midwestern United States. Eleven indicators related to missing treatment were rated by participants, including fear of relapse, feeling unmotivated, perceiving a lack of benefit from treatment, treatment staff and facility-related barriers, health issues, work or childcare responsibilities, or social support issues. We tested two to seven class solutions, which were evaluated based on Bayesian Information Criteria (BIC), sample size-adjusted BIC (SABIC), Akaike’s Information Criteria (AIC), Consistent AIC (CAIC). Entropy values, with higher values indicating greater differentiation between classes, smallest class size, and interpretability of class solutions were evaluated. Next, all variables that showed significant associations with the class membership were initially included in the logistic regression model, based on the membership identified through LCA. We hypothesized that levels of self-reported unmet social and economic needs would be differentially associated with class membership. Stepwise variable selection was then applied to obtain the final mode, with the inclusion and retention criteria set at p-values of 0.25 and 0.15, respectively.

Results: The two-class solution was optimal based on BIC and CAIC. The two-class solution had an entropy of 0.76, indicating adequate differentiation between classes, which we characterized as high barrier (79.7%, n=232) and low barrier (20.3%, n=59) subgroups. Larger class solutions had classes with 13 or fewer members (less than five per cent of the sample). Bivariate tests showed that the high barriers group significantly differed from the low barrier group by housing status, with unhoused individuals overrepresented, higher unmet basic needs, higher rates of past 30-day use of opioids, alcohol, or cannabis, higher rates of any lifetime psychoactive substance overdose, higher rates of depression, and higher rates of past 30-day suicidal ideation (all ps<.05). Class membership did not differ by race/ethnicity, gender, age, insurance, or employment status. The stepwise logistic regression model retained unmet basic needs, housing status, past 30-day suicidal ideation, past 30-day cannabis use, and lifetime overdose. Unmet basic needs (aOR=1.02, 95%CI:1.00, 1.04) and suicidal ideation (aOR=2.69, 95%CI: 1.15, 6.34) were statistically significantly associated with assignment to the high barrier class.

Conclusions and Implications: Prior research using national surveys has identified latent subgroups of barriers for treatment naïve patients. This study builds on prior research by identifying patterns of treatment barriers for participants who are enrolled into treatment and recovery support services. Understanding perceived treatment barriers can be useful for tailoring outreach and engagement strategies to enhance participants’ motivation to engage in treatment and for identifying strategies to mitigate the influence of social factors that potentially impact treatment participation.