Table 1 presents the main sample characteristics and cannabis use patterns, including sociodemographic variables and academic outcomes, which are further disaggregated into dropout and repetition, as well as consumption behaviors. All variables were comparable across sex, with no statistically significant differences observed; detailed sex-stratified data are not shown. For categorical variables, chi-square values ranged from χ² = 0.01 to χ² = 1.17, with corresponding p-values between 0.921 and 0.279. For continuous variables, F-statistics ranged from F = 0.26 to F = 1.12, with p-values between 0.608 and 0.291.

Information on cannabis use was collected, including age of initiation (M = 13.14, SD = 1.45) and age of regular use onset (M = 13.80, SD = 1.53), as well as weekday consumption patterns. This pattern was measured by asking participants about their consumption habits before attending school (Yes = 173, 63.4%), during school hours (Yes = 121, 44.3%), and after school (Yes = 158, 57.9%). Most participants were male (n = 178, 58.7%) and had a dual diagnosis (n = 248, 82.7%).

Regarding educational outcomes, 61.7% of participants (n = 183) showed school impairment, defined as either grade repetition or school dropout. Specifically, 57.7% (n = 173) had repeated at least one academic year, while 27.3% (n = 82) had dropped out of school. To avoid multicollinearity issues, we included the time between initiation and regular cannabis use in the model (latency, in years).

Of note, the logistic regression was conducted on the complete-case sample of n = 300 participants, following the application of listwise deletion to the initial sample (N ≈ 360). Approximately 17% of cases had at least one missing value across core variables. As noted in the Statistical Analysis section, non-random missingness cannot be formally excluded, which may have introduced some bias and should be considered when interpreting the results.

A multivariate logistic regression analysis was performed to determine the association between dual diagnosis and school impairment among adolescents with CUD. Psychiatric comorbidity status was included in the model, along with age of initiation, latency from first cannabis use to regular cannabis use, and sex as covariates. The Hosmer & Lemeshow test (χ² = 8.94; df = 8, p = 0.348) showed acceptable model fit (Nagelkerke pseudo R² = 8.7%).

A significant association was found between co-occurring internalizing and/or externalizing disorders and school impairment among adolescents with CUD (OR: 3.73; 95% CI: [1.06–13.12]; p = 0.04). Regarding the main effects, psychiatric comorbidity was associated with approximately twice the likelihood of repeating one grade or dropping out (OR: 2.05; 95% CI: [1.10–3.84]; p = 0.03), whereas the CUD-only group was not independently associated with higher odds of school impairment. Sex, age of initiation, and latency to regular cannabis use were not statistically significant predictors in the adjusted model. Therefore, these variables should not be interpreted as independent protective or risk factors in the present analysis.

Although female sex showed a higher point estimate, and older age of initiation and longer latency showed lower point estimates, none of these associations reached statistical significance. Consequently, the percentage changes derived from these odds ratios are presented only descriptively and should not be interpreted as statistically supported effects. Figure 1 summarizes the results of the model. Each point represents a single participant.

Predicted probabilities for academic failure for each person (point) were derived from the multivariable logistic regression model, with their values of age of initiation, latency (years from first use to problematic use), sex, cannabis use disorder and externalizing (Ext) and internalizing (Int) disorders.

The main finding of this study is that adolescents with DD showed poorer school performance. Specifically, co-occurring internalizing and/or externalizing disorders among adolescents with CUD were associated with higher odds of school impairment, suggesting that the combined presence of cannabis use disorder and psychiatric comorbidity may be particularly relevant for academic functioning. The proportion of school dropout or grade repetition was higher among adolescents with CUD and co-occurring psychiatric disorders than among those with CUD only, supporting the clinical relevance of considering psychiatric comorbidity when evaluating school impairment in adolescents with cannabis use disorder.

Previous studies have reported a relationship between early onset of cannabis use and lower high school graduation rates (Burke et al. 2021). Although early cannabis use has also been associated with poorer educational and socioeconomic outcomes in adulthood (Chan et al. 2024), age of initiation and latency to regular use were not statistically significant predictors of school impairment in the adjusted model of the present study. Therefore, these variables should be interpreted as clinically relevant background factors, but not as independent predictors of school impairment in our data.

When considering sex-related differences, female sex showed a higher point estimate in the adjusted model; however, this association did not reach statistical significance. Therefore, the present findings do not provide empirical support for sex differences in school impairment within this sample. Previous studies have reported sex-related differences in treatment trajectories, internalizing symptoms, cannabis-related motives, craving, and psychosocial outcomes among individuals with substance use disorders (Dahlberg et al. 2022; Foster et al. 2016; Gex et al. 2023; Fonseca et al. 2021; Tang and George 2024). In particular, adolescent girls with substance use disorders may present specific vulnerability profiles, including greater internalizing symptom burden, anxiety and depressive symptoms, suicide risk, and coping-related cannabis use motives (Foster et al. 2016; Gex et al. 2023). However, these mechanisms should be interpreted only as contextual hypotheses derived from previous literature and were not directly confirmed in the present study.

Another result of our study is that DD is associated with a higher likelihood of school impairment, suggesting that the presence of a DD is associated with poorer outcomes including lower academic performance (Adan and Torrens 2021). The bidirectional relationship between mental health and academic performance during adolescence (Lee et al. 2024) together with the importance of academic performance as a predictor of health in adulthood (Monzonís-Carda et al. 2024), highlight the importance of early interventions in adolescents with DD to improve their quality of life (Volkow and Wargo 2022).

Several limitations should be considered. First, the sample was predominantly male, which may limit the precision of sex-specific analyses and the generalizability of findings to female adolescents. Second, approximately 17% of the initial participants had at least one missing value among those used in the model (sex, age of initiation, years from problematic use, CUD, other diagnoses, and academic performance). The complete-case analysis was conducted under the assumption of missing completely at random (MCAR); however, missing at random (MAR) or missing not at random (MNAR) mechanisms cannot be formally excluded, and non-random missingness may have introduced bias in the estimates (Sterne et al. 2009). Third, the absence of a control group limits the ability to compare adolescents with CUD and DD with community adolescents or adolescents with psychiatric disorders but no cannabis use disorder. Fourth, clinical samples may differ from community samples in psychiatric severity, comorbidity burden, functional impairment, cannabis use patterns, and help-seeking pathways; therefore, caution is warranted when generalizing these findings to the broader adolescent population. Fifth, the study was conducted in an urban catchment area in Barcelona characterized by social, cultural, and ethnic diversity, but ethnicity was not systematically collected or analyzed. This limits our ability to examine whether cannabis use patterns, academic trajectories, or access to mental health services differed across ethnic or sociocultural groups. Sixth, the interpretation of school outcomes should be contextualized within the Spanish compulsory secondary education system (Educación Secundaria Obligatoria, ESO) and local educational and social support structures, which may differ substantially from those in other countries. Additional limitations include the absence of formal inter-rater reliability analyses for diagnostic assessments, the absence of detailed participant flow data, the lack of formal blinding of clinicians to academic outcome data during psychiatric assessments, and the observational clinical design, which limits the ability to establish causal relationships. Future research should include larger and more balanced samples, particularly with more female participants, longitudinal analyses of academic trajectories, and more detailed sociocultural and educational variables.

The present findings should be interpreted within the broader literature examining the relationship between adolescent mental health, substance use, and educational outcomes. Previous research has consistently shown that both substance use and psychiatric disorders are independently associated with poorer academic performance, including increased risk of school dropout, lower academic achievement, and functional impairment (Lee et al. 2024). However, relatively few studies have examined the combined impact of co-occurring psychiatric disorders and substance use on educational trajectories.

In this context, our results extend existing evidence by focusing on dual diagnosis and by examining the association between psychiatric comorbidity and school impairment among adolescents with cannabis use disorder. The findings suggest that their co-occurrence is associated with a substantially higher likelihood of school impairment compared to the presence of either condition alone, supporting the need for an integrated perspective on adolescent risk. Rather than reflecting additive effects only, this pattern may indicate a more complex vulnerability profile in which multiple clinical dimensions converge to affect academic functioning.

The results of the present study suggest that dual diagnosis, particularly the co-occurrence of CUD with internalizing and/or externalizing disorders, is associated with a higher likelihood of school impairment in adolescents. By contrast, sex, age of initiation, and latency to regular cannabis use were not statistically significant predictors in the adjusted model and should therefore be interpreted cautiously. These findings highlight the importance of early identification strategies targeting adolescents with cannabis use disorder and co-occurring internalizing and externalizing symptoms, particularly those showing broader clinical and functional vulnerability.

From a clinical and public health perspective, these results support the need for coordinated and integrated approaches across school systems, child and adolescent mental health services, and addiction services. Schools may play a key role in early detection through the identification of academic decline and behavioral changes, while mental health and addiction services may contribute to comprehensive assessment and intervention. Integrated care models that address both substance use and co-occurring psychiatric symptoms may be particularly relevant for adolescents with dual diagnosis, especially those showing broader clinical and functional vulnerability.