“I have no disease and weed just relaxes me!”: The therapeutic challenge in young patients with psychosis and cannabis abuse
P. Mota
European Psychiatry April 1, 2021 DOI: 10.1192/j.eurpsy.2021.456 (opens in new tab)
Study at a glance
AI-extracted from the abstract| Characteristics | Narrative review Peer reviewed |
|---|---|
| Topics | Cannabis |
| Key findings | The authors argue that cannabis use is a potentially preventable risk factor for poor treatment adherence and relapse in young people with psychosis, noting that non-adherence to therapy exceeds 50 percent during the inaugural phases of psychotic disorders and is reportedly higher in younger patients. They propose that interventions to improve adherence should specifically target reducing cannabis consumption. |
Abstract
Introduction: Substance use disorders (SUDs) are estimated to affect around 30 million people worldwide, and are characterized by repeated use of a substance that leads to clinically significant impairment or suffering, making it a serious health problem, with high associated costs.
Objectives: Understand and evaluate the impact of cannabis use on adherence to treatment in young patients with psychosis.
Methods: Narrative literature review by performing a search on MedLine for English-written articles. The query used was “(Cannabis) AND (Schizophrenia OR Psychosis) AND (Adherence)”.
Results: About 70 to 80% of young people with SUDs have at least one concomitant psychiatric disorder and cannabis is involved in approximately 50% of psychosis or schizophrenia of those cases, so there is a growing concern about the deleterious medical and psychiatric consequences of the increase and early initiation of consumption of this substance. It is estimated that about 26% of patients with psychotic conditions do not adhere to the treatment plan established by the psychiatrist; however, especially during the inaugural phases of psychotic disorders, rates of non-adherence to therapy are high (above 50%), and are said to be higher in younger patients.
Conclusions: The risk of relapse after a first psychotic episode is high. As the use of cannabis is a potentially preventable risk factor, interventions aimed at improving therapeutic adherence in psychotic conditions must specifically target the use of this substance, since reducing its consumption can lead to a more favorable course of the disease and at less expensive costs in addressing these pathologies. Disclosure No significant relationships.
Comparable studies
Other narrative reviews on cannabis, most cited first.
| Study | Year | Design | Participants |
|---|---|---|---|
| Causal association between cannabis and psychosis: examination of the evidence | 2004 | Review | |
| Cannabis sativa: A comprehensive ethnopharmacological review of a medicinal plant with a long history. General (historical and pharmacological review) | 2018 | Review | |
| Psychiatric effects of cannabis | 2001 | Review | |
| Beyond THC: The New Generation of Cannabinoid Designer Drugs | 2011 | Review | |
| Cannabidiol, a Cannabis sativa constituent, as an antipsychotic drug | 2006 | Review |