This page follows a source whose reuse terms permit translation, preserving its structure and line of argument in English.
About this resource
This is an English version of the EU-GEI study, which compared people presenting for medical care with a first psychotic disorder against community controls across research sites in Europe and Brazil. It examined how cannabis-use frequency and estimated potency related to onset of psychotic disorder and geographic variation. The principal individual-level analysis included 901 cases and 1,237 controls.
The paper is not a causal experiment proving that cannabis use alone caused psychosis in an individual. It was a case-control study that asked about past use after onset, not a randomized trial. The authors statistically adjusted for age, sex, ethnicity, other drugs, and additional factors, but unmeasured confounding and recall error remain. It is essential to distinguish “association,” “estimate,” and “population attributable fraction” from an established cause.
Study design
From 2010 to 2015, the study identified first-episode psychotic disorders within several geographically defined areas. Controls were recruited from the same communities. Participants were asked whether they had ever used cannabis, their age at first use, frequency, and the type of product they used most often. Product names and regional information were used to classify as high potency products estimated to contain at least 10% THC.
Statistical analyses used nonuse as the reference and calculated odds ratios for combinations of frequency and estimated potency. Researchers also examined relationships between local incidence of first-episode psychotic disorder and the prevalence of daily use and high-potency use among controls. Population attributable fractions estimated how the proportion of cases might change under the hypothetical absence of a particular exposure.
The high-potency classification was not a chemical analysis of each person’s sample; it was estimated from the reported product and existing information. This is a principal constraint alongside the study’s strengths of scale and multiple regions.
Content
Incidence of psychotic disorders varies substantially among regions. The research team asked whether cannabis-use patterns might explain part of that difference. A distinctive feature was separating daily use and products with estimated high THC potency rather than considering only whether someone had ever used cannabis.
Several explanations may underlie the association between cannabis and psychosis: a hypothesis that THC exposure contributes to symptom risk; reverse causation, in which people undergoing changes before onset select cannabis; shared genetic and environmental factors; and effects of other substances or social stress. A case-control study can show the strength and pattern of an association but cannot separate these explanations completely.
Because cases were people who reached medical care for a first episode, those who were undiagnosed or used different support pathways were not included. Geographic differences in access to care may also affect ascertainment of incidence.
Results
The analysis found an association between daily cannabis use and first-episode psychotic disorder, with a stronger association among people reporting daily use of high-potency products. In Amsterdam and London, where high-potency use was common, higher incidence of first-episode psychotic disorder was also observed. Both individual- and area-level results showed the need to consider frequency and estimated potency.
Using models, the authors estimated that if high-potency cannabis had not been available, a certain proportion of first-episode psychotic-disorder cases in Amsterdam and London might not have occurred. Population attributable fractions, however, depend on assumptions including that the association is causal and confounding is adequately controlled. They are not results from an intervention experiment that actually removed a product.
The findings do not mean every user will develop a psychotic disorder, and they do not calculate an individual’s probability of diagnosis. Conversely, one individual’s experience of not developing the disorder cannot negate the population association. Risk is probabilistic and is treated as the possibility that exposure, vulnerability, and environment combine.
Limitations
Exposure was self-reported and may be affected by memory and response tendencies after onset. THC concentration was estimated rather than analyzed, and CBD content, amount per occasion, depth of inhalation, and variation within products were not captured. Control participation rates and representativeness may also have differed by region.
Cases and controls may differ in education, income, urbanicity, childhood experiences, family history, other substance use, and additional characteristics. Residual confounding remains even after statistical adjustment. Geographic correlations with incidence include ecological relationships and cannot be transferred directly to causation in individuals.
The study concerned first-episode psychotic disorders in particular locations and years and cannot automatically be generalized to the current product market or the entire world. Definitions of high-concentration products also change over time.
Safety
If psychosis-like symptoms, intense suspiciousness, confusion, or impaired reality testing occur, securing safety and obtaining medical assessment take priority over self-interpretation of research information. A history or family history of psychotic disorder, younger age at initiation, frequent use, and high-THC exposure are factors that may be emphasized in individualized preventive consultation.
This page does not present a low-risk concentration or frequency. An observational study cannot establish a threshold below which use is safe, and it has not proved that adding CBD cancels risk. Driving, pregnancy, and combinations with other psychoactive substances or medications add separate risks.
It is also important not to discontinue prescribed medication or self-treat psychiatric symptoms on the basis of this study. Geographic analysis provides public-health information, not an individual diagnosis.
People may have used cannabis to manage worsening sleep or anxiety before onset, and prospective research is needed to track the temporal relationship between exposure and early symptoms in greater detail.
Source and rights
Original source: Di Forti M, et al. The contribution of cannabis use to variation in the incidence of psychotic disorder across Europe (EU-GEI): a multicentre case-control study. The Lancet Psychiatry. 2019;6(5):427–436. DOI: 10.1016/S2215-0366(19)30048-3.
The original article is available under Creative Commons Attribution 4.0 (CC BY 4.0). This page is a source-aligned English translation under that license, reconstructed while preserving the methods and argument. Figures and tables are not reproduced. The original source takes precedence for complete odds ratios, confidence intervals, regional estimates, and adjustment variables.