A total of 150 patients beginning psychosocial treatment for chemical substance use-related problems participated in the study. The second column of Table 1 outlines the distribution of patients according to their social and economic characteristics, previous psychosocial treatments, severity of alcohol dependence at the beginning of treatment, and use of illicit drugs in the previous 30 days. Most participants were men (87.3%; n = 131/150) and 12.7% were women (n = 19/150). Their age ranged from 19 to 68 years, including 34.7% (n = 52/150) of participants who were 19 to 30 years old, 40.6% (n = 61/150) from 31 to 40 years old, and 24.7% (n = 37/150) from 40 to 68 years old. Most participants (55.4%; n = 83/150) had only a primary school education, 35.3% (n = 53/150) had a secondary education, and 9.3% (n = 14/150) had attended college. The majority (73.3%; n = 110/150) of participants had no source of income. Social support outside the clinical environment was considered high (higher than the median) for 50.7% (n = 76/150) of participants. Approximately half (47.3%; n = 71/150) lived alone, whereas the remaining 52.7% of participants lived with other people. Most participants (64.7%; n = 7/150) had previously taken part in some type of treatment for alcohol or drug use-related problems; the rest (35.3; n = 53/150) were starting treatment for those problems for the first time. Nearly all patients (98.0%; n = 147/150) had severe alcohol dependence at the beginning of treatment, whereas the other 2.0% (n = 3), had mild or moderate dependence. Most patients (89.3%; n = 134/150) reported the use of illicit drugs in the last month.

Most patients (88.7%; n = 133/150) dropped out of the current psychosocial treatment at the CAPS-AD during the first two months of treatment. The third column of Table 1 outlines the number and percentage of participants who dropped out of treatment, according to the exposure variables. The dropout rate decreased with education (chi-squared for linear trend, p < 0.001) from 96.4% (n = 80/83) for primary education, to 71.4% (n = 10/14) for patients with higher education. Having some source of income was also associated with a lower treatment dropout rate (chi-squared, p = 0.009); 77.5% (n = 31/40) of patients reporting some income dropped out, compared to 92.7% (n = 102/133) of those with no source of income. No evidence was found for associations between the psychosocial treatment dropout rate and the following variables: gender (chi-squared, p = 0.454), age group (p = 0.301), living alone or with other people (p = 0.589), social support outside the clinical setting (chi-squared, p = 0.475), participation in previous treatment for alcohol or drug-related problems (p = 0.297), severity of alcohol dependence (p = 0.305), and illicit drug use in the last month (p = 0.696).

At the beginning of treatment, the score for self-efficacy expectations of changing addictive behaviors ranged from 11 to 100, and the score for self-efficacy expectations of remaining in treatment in the first two months ranged from 1 to 5. Table 2 shows that the self-efficacy expectations of changing addictive behaviors averaged 63.60 for patients who remained in treatment and 39.96 for those who dropped out in the first two months of treatment. Conversely, the self-efficacy expectations of remaining in treatment averaged 4.82 for the group of patients who remained in treatment, and 3.56 for those who dropped out of treatment in that period.

The Cox regression model outlined in Table 3 showed evidence for associations between the two types of self-efficacy expectations and treatment dropout. The treatment dropout rate increased by 3% with every one-point increase in the score of self-efficacy expectations of changing addictive behaviors (HR = 1.03; p = 0.021 after adjusting for the variables outlined in Table 1 and self-efficacy expectations of remaining in treatment). In this model, the dropout rate increased approximately two-fold with every one-point decrease in the score of self-efficacy expectations of remaining in treatment (HR = 2.35; p = 0.034, after adjusting for the variables outlined in Table 1 and self-efficacy expectations of changing addictive behaviors).

The present study aimed to examine the association between motivation to continue psychosocial treatment for chemical dependence-related problems and adherence to this treatment. One hundred and fifty patients were followed up at a CAPS-AD for the first two months of treatment, 88.7% of whom dropped out of treatment. The following variables showed a significant association with the dropout risk: education, income, self-efficacy expectations of changing addictive behaviors, and self-efficacy expectations of remaining in treatment.

Regarding education, the study indicates that patients with an education up to primary school had a higher risk of treatment dropout than did those who had attended college. This association corroborates findings of other studies on adherence to outpatient treatment for alcohol and drug use-related problems (Micheletti et al., 2016; Sousa, Ribeiro, Melo, Maciel, & Oliveira, 2013). These studies showed increased difficulties in treatment adherence with decreases in level of education.

Regarding income, the present study indicates that patients without any source of income have a higher risk of treatment dropout (albeit only 2%) compared with those who have a source of income. Difficulties in social inclusion, due to either low education and income or lack of social support, are a barrier to treatment adherence (Silva, Maftum, & Mazza, 2014; Varela, Sales, Silva, & Monteiro, 2016).

Regarding the self-efficacy expectations (or motivation) of changing addictive behaviors, the results showed that less motivated participants had a higher risk of treatment dropout compared to other participants. A similar association was also found regarding the motivation to remain in treatment; that is, less motivated patients had more than double the risk of treatment dropout (HR = 2.35) than did the other patients. These results are in line with other studies on adherence to treatment for problems related to alcohol and other chemical substances (Ferreira et al., 2015; Sousa et al., 2013). In these studies, motivation to adhere to treatment was assessed by applying the transtheoretical model of behavior change (Prochaska, DiClemente, & Norcross, 1992). Their findings showed that low levels of treatment adherence and persistence of addictive behaviors were more commonly found among participants allocated to the contemplation stage, and that significant increases in treatment adherence and changes in addictive behaviors were observed among the most motivated patients (typically allocated to the action stage).

Limitations during the data collection of the present study should be noted and these may have led to data bias. The lack of a venue to perform the interviews or to assess the medical records forced the researchers and staff members to intercalate the visits, thus lengthening the research study time. Another difficulty was the high frequency of incomplete medical records, forcing the researchers to search for such data in interviews with the different reference technicians, according to their shift.

Despite the high treatment dropout rates at the CAPS-AD, facilitating motivation to change addictive behaviors through treatment adherence remains the most effective strategy to promote general health and improve the quality of life for individuals and communities. This study suggests that research should focus particularly on the interview for admission to the treatment program as a possible approach to adjust patient expectations of treatment results and challenges. Health professionals following this approach are expected to contribute to improved adherence to the therapeutic process. In addition to the value of the welcome (initial approach) in promoting treatment adherence, the present study provides likely evidence for the impact of social inequality on access to healthcare, as shown by the effects of poor access to education and income on treatment adherence failure. Future studies with population-based designs will be able to further explore such evidence to provide data for public initiatives and policies promoting mental health.

Lastly, the role of motivation in health behavioral change processes is noteworthy. Specifically, self-efficacy and self-determination theories (despite their particularities) are based on a key area of common ground in the understanding of human actions (Ryan & Deci, 2017). In these theories, humans are regarded as active agents; that is, they have cognitive processes that enable them to make decisions regarding actions. Regarding health promotion, studies integrating the above theories aim to contribute to better understand the mechanisms involved in changing behaviors from risk to protection. For example, in adults with type 2 diabetes Mellitus, the intrinsic motivation to adhere to the practice of physical exercise was mediated by the self-efficacy expectations of performing such behavior (Varming, Hansen, Andrésdóttir, Husted, & Willaing, 2015). In individuals with heart disease, internal motivation and self-efficacy were identified as key predictors of changes in physical activity after cardiac rehabilitation (Mildestvedt, Meland, & Eide, 2008). Therefore, (a) studies focusing on the integration between self-efficacy and self-determination theories in motivational interventions and (b) the promotion of social inclusion policies through access to education and income generation may represent key advances in treatment effectiveness for those seeking psychosocial rehabilitation. In this healthcare context, rehabilitation can be understood as a process whereby individuals with limitations are given an enhanced opportunity to restore the best possible level of autonomy in their functioning in the community.