**3. Results**

#### *3.1. Sociodemographic and Clinical Characteristics*

The study sample consisted of 1356 individuals with AUD or CUD screened for lifetime DD with the DDSI-IV. This study sample represented 48.0% of the total number of individuals who started treatment due to AUD or CUD in the four CAS during the study period. The prevalence of individuals screening positive for lifetime DD was 74.0% (*n* = 1000). Among these, the lifetime comorbid mental disorders were depression (76.4%), dysthymia (27.2%), mania (13.1%), panic disorder (37.5%), generalised anxiety disorder (26.5%), specific phobia (13.4%), social phobia (17.9%), agoraphobia (13.2%), psychosis (30.1%), post-traumatic stress disorder (23.5%) and attention deficit hyperactivity disorder (19.3%). A total of 71.4% (*n* = 971) were individuals with AUD and 77.5% (*n* = 386) were individuals with CUD (data not shown in Table 1).


**Alcohol or cocaine + cannabis** 53 (6.0%) 22 (41.5%) 20,823 12 (3.0%) 6 (50.0%) 4295


interquartile range; 9, \*, indicates that differences between individuals with and without a dual disorder are statistically significant (*p*-Value < 0.05).

**Table 1.** *Cont.*

Table 1 details the individuals' sociodemographic and clinical characteristics. Compared with individuals screening negative for lifetime DD, those screening positive at baseline were more frequently women (30.0% vs. 17.0%, *p*-value < 0.001), younger (56.0% vs. 49.0%, *p*-value = 0.049), unemployed (39.5% vs. 27.3%, *p*-value < 0.001) and reported higher polysubstance use (13.0% vs. 9.0% of alcohol and stimulants, respectively, and 6.0% vs. 3.0% of alcohol/cocaine and cannabis, respectively, *p*-value = 0.006). Moreover, a higher proportion had received previous treatment for an SUD (57.3% vs. 42.1%, *p*-value < 0.001), previous treatment for a psychiatric disorder (45.1% vs. 23.9%, *p*-value < 0.001), and more frequently reported a history of organic and psychiatric problems (35.6% vs. 21.4%, *p*-value < 0.001), a family history of substance use (44.5% vs. 34.3%, *p*-value = 0.002) and poorer self-perceived health (43.6% vs. 28.9%, *p*-value < 0.001). The median number of medical or psychiatric treatment visits (8 [IR: 4–13] vs. 6 [IR: 4–10], social care visits (2 [IR: 1–5] vs. 1.5 [IR: 1–3]) and follow-up time (423 vs. 369 days) were relatively higher and were significant in those screening positive for lifetime DD (data not shown in Table 1).

#### *3.2. Characteristics of Treatment Retention*

At one year of follow-up (Figure 1), treatment retention was more than 75% in both groups. Moreover, more than 50% of individuals remained in treatment for the entire follow-up period (38 months). Treatment retention decreased similarly in both groups during the study period, and the difference was not statistically significant (Wilcoxon *p*-value = 0.659; Log-Rank test *p*-value = 0.769). The proportion of dropouts in individuals screening positive for lifetime DD was 29.5% and was 28.4% in those screening negative. There were 458,941 person-days of follow-up among individuals screening positive for lifetime DD and 151,543 person-days of follow-up among those screening negative (Table 1).

11

#### *3.3. Multivariable Explanatory Models of Treatment Dropout*

Table 2 shows the different multivariable Cox regression models estimated. After adjustment for different sociodemographic, clinical and follow-up covariates, individuals screening positive for lifetime DD had a 26% increased risk of treatment dropout (HR = 1.26; 95% CI = 1.00–1.60) than those with SUD alone (no DD). According to the substance of use, those who used alcohol only and those who used alcohol or cocaine with cannabis had a 35% (HR= 1.35; 95% CI 1.04–1.77) and a 60% (HR = 1.60; 95% CI = 1.03–2.49) higher risk of treatment dropout, respectively, than those using cocaine only. Individuals who lived alone had a 34% (HR = 1.34; 95% CI = 1.04–1.72) increased risk of treatment dropout than those living with a partner and/or children. The risk of treatment dropout was reduced by 22% with one additional medical visit (HR = 0.78; 95% CI = 0.75–0.80), by 4% with one additional psychologist visit (HR = 0.96; 95% CI = 0.94–0.97) and by 3% with one additional visit with a social worker (HR = 0.97; 95% CI = 0.95–1.00). The Cox proportional hazard assumption (*p*-Value > 0.05) was observed for all variables included in the final model (model 4), except for the variables of previous psychiatric treatment and number of visits with a physician/psychiatrist and with a psychologist (Appendix A, Table A2).


