First‐episode psychosis with co‐occurring cannabis use disorder (FEP&CUD) was associated with worse clinical trajectory compared to FEP without SUD and CIP. Individuals in the FEP&CUD cohort experienced a greater burden of inpatient care, including higher rates of hospitalizations due to psychotic relapse and longer durations of inpatient treatment. These findings highlight the importance of clinicians recognizing the impact of co‐occurring CUD on the prognosis of FEP at the time of first service presentation to optimize clinical outcomes. However, FEP&CUD and incident CIP cohorts shared similar demographic profiles, clinical comorbidities, and mortality risk, underscoring the need for comprehensive clinical approaches to address the disease burden associated with both conditions. These similarities also suggest that cannabis‐related psychoses may exist along a continuum rather than representing fully distinct diagnostic entities.

This is the first study to leverage nationwide datasets to examine the clinical and demographic correlates of FEP&CUD, CIP and FEP without SUD. Our findings have important clinical implications, shedding light on the similarities and distinctions between cannabis‐related psychotic conditions and their prognoses. As this is the first study utilizing nationwide registry data, direct comparisons with previous studies based on clinical samples are limited. Consistent with prior research, we observed a predominance of males in both groups [21, 26, 27, 32]. Some earlier clinical studies have reported a younger age at admission for psychosis in FEP with substance use compared to SIP [21, 32], that we were not able to replicate in this population‐based sample. This discrepancy may be attributable to our larger dataset including all visits to specialized healthcare with inclusion of both inpatient and outpatient visits, as well as our focus on recorded CUD diagnosis rather than substance or cannabis use.

Hospitalization for psychosis during the first year after cohort entry was more common in the FEP &CUD (36%) and CIP cohorts (33%) than in the FEP without SUD cohort (24%) (Table 2), and most hospitalizations in all cohorts occurred within the first three years (Figure 4). The FEP&CUD cohort received more intensive inpatient care, characterized by longer hospital stays and a higher risk of hospitalization for psychosis compared to the CIP and FEP without SUD cohorts. Mortality was high and comparable between the FEP&CUD and CIP cohorts, but higher than in the FEP without SUD cohort, suggesting a worse overall prognosis in cannabis‐related psychoses.

Notably, individuals in the FEP without SUD cohort more often received a subsequent schizophrenia diagnosis within 2 years of the incident psychosis (14.41%–10.88%). In contrast, such diagnosis remained uncommon in cannabis‐related psychoses, recorded in only 6%–7% of FEP&CUD and 2%–3% of CIP cases during the first and second years after cohort entry. This suggests potential underdiagnosis and delay of schizophrenia diagnosis in cannabis‐related populations despite evidence of worse clinical trajectory. Clinicians may initially attribute symptoms to cannabis use, particularly when intoxication or heavy use is evident, potentially delaying the assignment of a schizophrenia diagnosis even when the underlying trajectory aligns with primary psychotic disorders. Our findings emphasize earlier identification and sustained treatment of cannabis‐related psychotic disorders to mitigate the clinical trajectories of these individuals.

Individuals in FEP without SUD cohort had higher rates of any antipsychotic use 30 days after cohort entry (57%) compared to the FEP&CUD and CIP cohorts where approximately 48% and 45% received antipsychotic treatment. This suggests that, in the context of pharmacotherapy, undertreatment of cannabis‐related psychoses may occur after the first service presentation despite their poorer clinical trajectory. Despite the lack of guidelines on pharmacotherapy for SIP, individuals experiencing FEP in the context of problematic cannabis use received comparable antipsychotic treatment during the initial phase of care, regardless of whether their psychosis is classified as substance‐induced (F1x.5) or primary (F2x). However, the FEP&CUD cohort was more likely to receive LAIs within 30 days, which may reflect a clinical perception that SIP often resolves with abstinence. Additionally, individuals with FEP&CUD were more likely to have records of comorbid polysubstance use and stimulant use disorder diagnoses and worse occupational functioning, which may be associated with more frequent nonadherence and greater clinical severity, potentially contributing to more frequent use of LAIs as well as higher risk of hospitalization for psychosis. In general, use of LAIs within 30 days after cohort entry was strikingly low in all cohorts (0.34%–2.79%). Our recent studies have demonstrated that LAIs, alongside clozapine, are associated with favorable real‐world effectiveness in relapse prevention for both CIP and FEP&CUD [28, 29]. Given the findings of this study regarding more severe clinical trajectories in cannabis‐related psychoses, prescribers should consider increasing the use of LAIs to improve treatment adherence and outcomes in these populations.

The construct validity of SIP has been recently challenged [20]. Evidence suggests similar rates of subsequent schizophrenia diagnoses after a first episode of brief, atypical, or not otherwise specified psychosis compared to CIP [19], as well as the similar severity of psychotic symptoms, relapse risk and remission rates [23, 24] and treatment response observed in FEP&CUD [28, 29]. This complexity calls for a nuanced reconsideration of the conceptual distinctions between primary psychosis and SIP and especially with CIP. Rather than treating these conditions as discrete diagnostic entities, it may be more appropriate to conceptualize them along a continuum of psychotic symptomatology, reflecting the considerable overlap and heterogeneity in clinical presentation. Within this context, we argue that CIP should be seen as a subtype of primary psychosis in individuals with cannabis use, rather than a disorder solely secondary to substance use.

Our study has notable strengths and some important limitations. To our knowledge, this is the first study to examine comparative clinical trajectories of FEP&CUD, CIP and FEP without SUD using nationwide register‐based data with a large sample size. The comprehensive coverage of the Swedish national registers, which include all residents, enables accurate identification of first‐episode cases with long‐term follow‐up. However, certain limitations must be acknowledged. We lacked objective information on continued cannabis use beyond what was recorded in the registers, which is relevant as ongoing use is associated with more frequent relapses and higher burden from inpatient care in psychosis [6]. Nevertheless, we did not observe differential rates of CUD diagnoses between CIP and FEP&CUD cohorts after cohort entry, suggesting this limitation is unlikely to introduce major bias. The reliance on register‐based diagnoses may introduce misclassification due to variability in clinical practices, and incomplete outpatient data prior to 2001 may have affected case identification. Furthermore, to our knowledge, the validity of SIP diagnoses within population‐based registers has not been rigorously established. Additionally, the registers do not capture detailed clinical information such as symptom severity or psychosocial factors. Finally, our findings may not be fully generalizable to populations outside Sweden. Thus, subsequent clinical studies in diverse populations and treatment settings are warranted to examine the clinical trajectories of cannabis‐related psychoses. Despite these limitations, our study provides valuable evidence to inform future clinical practice and research on cannabis‐related psychotic disorders.