However, during follow-up the ICBT group had relatively greater increases in percent-network abstinent, and greater decreases in mean-network drug use and percent-network using drugs. Effects of group, time, and group x time were not statistically significant for self efficacy or the remaining social network variables. Overall the two treatment groups had few differences on process variables, with the exception of social network drug use. Finally, the effect of time was negative and statistically significant for controlled context , indicating an overall decrease over time in the proportion of time spent in controlled contexts, a trend that did not differ between treatment groups. We then examined whether self-efficacy and social network variables prospectively predicted post-treatment alcohol and drug use. Separate HLMS for each process variable tested associations with substance use outcomes , controlling for previously mentioned covariates. Each time-varying predictor was lagged and decomposed into two distinct variables to disaggregate between-person and within-person effects on PDD and PDDRG. Social network variables were substance-specific, with network drinking predicting PDD and network drug use predicting PDDRG. Results of these models are displayed in Table 10. With one exception , all between-person effects were statistically significant. Moreover, individuals with social networks characterized by lower mean use and a lower percentage of regular users had lower PDD and PDDRG during follow-up, and those with greater percent-network abstinent from drinking had lower PDD. For within-person effects, greater mean network drinking and greater percent-network drinking predicted PDD, indicating that fluctuations from the person- mean in these variables predicted drinking in the next 3 months. We then determined whether the amount of time spent in controlled contexts was associated with post-treatment substance use.
These HLMs tested the time-varying effect of controlled context on concurrent PDD and PDDRG, controlling for identical covariates as previous models. The results, displayed in Table 10,cannabis drying racks indicate that a greater proportion of time spent in controlled contexts was associated with significantly lower PDD and PDDRG during these same periods. At month 18 veterans with 0% days controlled had an estimated PDD of 18.19. Estimated PDD was lower for those at the mean of controlled context , and even lower for those 1 standard deviation above the mean . These results confirmed that the percentage of time spent in controlled contexts was significantly and negatively associated with frequency of current substance use. Final models included interactions between lagged process variables and controlled context, to test hypotheses that percent-days in controlled context would moderate the effects of lagged process variables on PDD and PDDRG. As displayed in Table 10, between-person effects for each social network drinking variable interacted significantly with controlled context to predict future PDD. Also, between-person effects for mean network drug use and percent-network regularly using drugs interacted with controlled context to predict future PDDRG. As displayed in Figure 1, social network effects on future PDD and PDDRG were weaker when time in controlled context was high. Controlled context did not significantly moderate self efficacy or within-person effects. Overall, these results indicate that social network characteristics of veterans with substance dependence and MDD predicted future drinking and drug use, but these predictive effects were weakened when veterans lived in controlled contexts for extended periods of time. Prior studies of determinants of post-treatment substance use have not typically focused on individuals with SUDs and co-occurring psychiatric disorders, despite high rates of these co-occurring disorders in many clinical settings . Co-occurring MDD is particularly common, and individuals with SUD and MDD typically have poorer outcomes following treatment and may receive less benefit from therapeutic mechanisms of change . Because they represent a large proportion of patients receiving substance dependence treatment and have a greater risk of poor outcomes, it is critically important to examine determinants of post-treatment substance use within this population.
In a sample of veterans who had received group outpatient psychotherapy for substance dependence and MDD, we examined self-efficacy and network substance use, the effects of controlled contexts, and their interaction in the prediction of post-treatment drinking and drug use. In this sample with substance dependence and MDD, individuals with greater self-efficacy had lower future drinking and drug use during the year following treatment. Self-efficacy has predicted post-treatment use across various substances and treatment settings , and our study extends these findings to patients with substance dependence and MDD. By separating the between-person and within-person components of self-efficacy, we demonstrated that generally maintaining higher self efficacy was protective against future substance use, while time-specific fluctuations in self-efficacy were less critical. Prior studies with this sample revealed that baseline self efficacy predicted time to relapse and self-efficacy increased significantly during treatment . Our results were encouraging in finding that during the year following treatment, self-efficacy did not decrease significantly and continued to predict substance use. Moreover, post-treatment self-efficacy was similar between ICBT and TSF, suggesting that self-efficacy can be sustained similarly following cognitive-behavioral or professionally-led, 12-step based interventions, replicating prior research . Our results support the value of abstinence self-efficacy for patients with substance dependence and MDD, and provide a strong rationale for investigating factors related to increasing self-efficacy within this population. Substance-specific social network variables were also predictive of post-treatment substance use. For both alcohol and drugs, future frequency of use was lower for individuals who generally had 1) lower average network use and 2) a lower proportion of regular users in their network. Furthermore, having a network with a greater density of non-drinking individuals predicted lower future drinking.
These social network characteristics have previously predicted drinking outcomes in alcohol-dependent samples , but our study is first to demonstrate these findings in patients with substance dependence and MDD. Social network effects at the within-person level were also predictive of substance use. More specifically, when individuals had increases from their own norm of average network drinking or proportion of regular drinkers, they drank more frequently in the future. It is likely that these interpersonal factors operate in multiple ways to influence future drinking . The presence of alcohol or other drinkers represent conditioned cues that elicit craving or other precipitants to drinking, and associating with drinkers may increase exposure to substances or undermine patients’ sobriety efforts via changes in outcome expectancies. Treatment protocols designed explicitly to change social networks have demonstrated long term efficacy in the treatment of alcohol dependence . These interventions have not been tested in patients with co-occurring psychiatric disoders, and our study suggests these interventions could be effective in patients with substance dependence and MDD. This study also found that post-treatment substance use depended on the level of constraint in the living environment. During follow-up over half of our sample had some time in “controlled contexts” at an average of 39% of days in these environments. As hypothesized, when patients spent a greater percentage of time in controlled contexts defined by limited access to substances or negative consequences for use, they had lower frequency of drinking and drug use. Moreover, controlled context acted as a moderator,cannabis grow tray such that social network influences were more predictive of future alcohol and drug use when the environment did not restricted. Social network effects were attenuated for veterans at the highest levels of controlled context, suggesting these contexts do assist in buffering against the maladaptive influence of substance use within the network. Research indicates that patients with substance dependence and MDD are especially likely to utilize inpatient/residential services . The role of environment is rarely examined in treatment outcome studies, which is surprising given theoretical interest in considering intrapersonal and interpersonal processes within environmental context . Our results suggest the level of constraint in the environment is an important construct to consider in future studies. While some 3 controlled contexts may have resulted from adverse events , the most common contexts utilized were residential sober living facilities. Our results suggest patients with substance dependence and MDD who cannot sustain more protective social networks may benefit most from these constrained environments. While this study has important research and clinical implications, its limitations should be noted. The immediate generalizability of these findings is limited due to the demographic characteristics of this veteran sample, which was heavily comprised of Caucasian males. While this is a common limitation of many clinical trials of treatments for addictive disorders , replication in more diverse samples is needed before broader generalizations can be made. Our social network measure did not capture important features of social support examined in prior studies , such as whether network members were supportive of our patients’ abstinence efforts. Furthermore, network members’ use patterns were reported as perceived by study participants, and may not be a completely accurate assessment of their substance use patterns. While the effects of controlled contexts were in the expected direction, context alone may not fully explain our findings as this variable could be confounded with other unmeasured characteristics that are also strong determinants of abstinence.
While this study identified variables that predict post treatment substance use in adults with co-occurring substance dependence and MDD, underlying mechanisms have not yet been identified, which would explain how patients sustained greater levels of self-efficacy and lower levels of network substance use. As such, it will be important for future studies to explore whether specific processes of treatment assist patients in maintaining greater levels of these proximal variables following the conclusion of formal intervention. Study 3 has been submitted for publication in Alcoholism: Clinical and Experimental Research, as following: Worley, M.J., Tate, S.R., & Brown, S.A. Self efficacy, social networks, and the moderating effects of context following treatment for co-occurring substance dependence and depression. The dissertation author was the primary author of this manuscript under review. Many individuals with alcohol or drug dependence have deficits in neurocognitive abilities such as executive functioning and problem solving , verbal or non-verbal memory , cognitive efficiency , and attention , whether these deficits are due to the neurotoxic effects of substance use, preexisting characteristics, or some combination of both . In clinical settings these deficits are especially prevalent as 30-80% of patients enter treatment with some neurocognitive impairment . Although modest recovery in neurocognition is possible, these changes are often not clinically significant and many deficits do not remit even after periods of extended abstinence . Because neurocognitive impairment is so common and persistent, there is a strong rationale for examining whether impairment relates to treatment outcomes within the substance-dependent population. Another factor that may compound these neurocognitive deficits is the presence of other psychiatric conditions. Comorbid psychiatric disorders are common in the substance-dependent population , especially among patients in treatment settings . Major depression is recognized as the most common Axis I condition, and patients with comorbid MDD typically have poorer outcomes from treatment for alcohol or drug dependence . Importantly, MDD is also associated with deficits in working memory, attention, verbal memory, verbal fluency, and processing speed . Consequently, the joint effects of substance dependence and MDD on cognition may be worse than in either disorder alone, as with comorbid substance dependence and bipolar disorder . Despite the prevalence of comorbid substance dependence and MDD and the deficits associated with both conditions, the effects of neurocognitive impairment in patients receiving treatment for substance dependence and MDD have not previously been examined. Psychological treatments for substance dependence often require patients to learn and retain new information, inhibit prepotent responses, engage in future-oriented behavior, and plan behavioral coping strategies. Theoretically, patients with neurocognitive deficits would be less adept at internalizing and engaging in these skills, leading to poorer outcomes from psychotherapy. Some studies support this hypothesis, as patients with poorer cognitive functioning had greater substance use or lower retention in cognitive-behavioral therapy . However, these effects have not been consistent across studies, as many investigations failed to find associations between neurocognition and substance use during or following treatment . While some methodological differences exist between these studies , it is equivocal whether cognitively-impaired patients have poorer outcomes from substance dependence treatment. An important distinction within this literature is whether neurocognitive impairment is best examined as a direct predictor of outcome, or whether different model specifications are more ideal for testing relations with substance use . This notion is supported by studies demonstrating that impairment does not predict outcomes directly, but impacts substance use indirectly through therapeutic mechanisms of change.