[Psychosis and addiction: The evidence cemetery]

Objectives The comorbidity between psychosis and substance use has attracted wide attention over the years, and a vast literature is now available for meta-analytic treatment. In the field, a majority of authors assume that cannabis smoking is a risk factor for psychosis, that substance abuse is highly prevalent in schizophrenia, that substance abuse worsens the prognosis of schizophrenia, and that integrated treatments have greater efficacy than treatment-as-usual for this complex population. The objective of the current article is to review the meta-analyses that have been published in the comorbidity field in order to determine if the above-mentioned assumptions are substantiated by evidence or not. Methods A search of the literature was performed using PubMed, PsycINFO and EMBASE. The literature search retrieved a total of 25 systematic quantitative reviews, addressing the following issues: etiology, age at onset, prevalence rates, cognition, treatment, as well as psychiatric, neurologic and functional outcomes. Results Evidence shows that the prevalence of tobacco smoking, cannabis smoking and alcohol use is elevated in psychosis. However, this prevalence is likely to be over-estimated since studies have been performed in clinical settings rather than the general population. Reliable evidence also suggests that cannabis smoking is a risk factor for psychosis outcomes. However, the association is rather small and it remains difficult to draw an unequivocal public health message from this literature. In the same vein, evidence suggests that cannabis smoking is associated with an earlier age at onset of psychosis. However, this observation is derived from cross-sectional studies, not longitudinal ones; thus, no undisputable claims on causality can be made from them. On clinical grounds, some evidence also suggests that substance use is associated with self-harm, increased positive and depressive symptoms in psychosis patients, but this evidence is derived from cross-sectional studies, not longitudinal ones. Cocaine may exacerbate antipsychotic-induced extrapyramidal symptoms in schizophrenia, but this observation is based on a small number of studies. In the case of violence, the aggregation of studies involving very large samples of patients has shown a strong association with substance abuse in psychosis patients. However, this association is based on statistics that are not adjusted for potential confounds, and the role of cluster-B personality disorders in the substance abuse-violence association has yet to be determined from an evidence-based perspective. The effects of psychoactive substances on cognition in psychosis patients are inconsistent and contradictory. In terms of treatment, evidence shows that bupropion and varenicline increase tobacco smoking cessation rates in psychosis. However, this observation is based on a small number of studies. Finally, there is no evidence that integrated psychosocial interventions are superior to treatment as usual in this population. This lack of efficacy may due to a real lack of efficacy or to methodological problems making the comparison of intervention studies difficult. Discussion The evidence supporting the main assumptions of the comorbidity field is not as strong as it may seem. Moreover, important gaps in our understanding of the psychosis-addiction comorbidity remain. Due to lack of interest or lack of data, no meta-analysis has been performed, in the dual-diagnosis population, on injectable antipsychotics, subjective reasons for use, treatment compliance, medical comorbidities, the social context of use, the neurobiological links between substance use and psychosis, as well as the comparative efficacy of nicotine replacement therapy.