Cannabis Use Cessation and the Risk of Psychotic Disorders: A Case–Control Analysis from the First Episode Case–Control EU-GEI WP2 Study: L’arrêt de l’utilisation du cannabis et le risque de troubles psychotiques: Une analyse cas-témoins tirée de l’étude cas-témoins EU-GEI WP2 centrée sur les premiers épisodes psychotiques

Abstract

Objectives

To establish whether the risk of psychotic disorders in cannabis users changes with time following cannabis cessation using data from the European Network of National Networks studying Gene–Environment Interactions in Schizophrenia (EU-GEI) case–control study.

Methods

The EU-GEI case–control study collected data from first episode psychosis patients and population controls across sites in Europe and Brazil between May 2010 and April 2015. Adjusted logistic regressions were applied to examine whether the odd of psychosis case status changed: (1) with time following cannabis cessation and (2) across different cannabis use groups.

Results

Psychosis risk declined following cessation of cannabis use (β = −0.002; 95% CI −0.004 to 0.000; P = 0.067). When accounting for duration of use, this effect remained (β = −0.003; 95% CI −0.005 to −0.001; P = 0.013). However, in models adjusting for frequency and potency of use the result was not significant. Analysis of different cannabis use groups indicated that ex-users who stopped 1 to 4 weeks previously had the highest risk for psychotic disorder compared to never users (OR = 6.89; 95% CI 3.91–12.14; P < 0.001); risk declined for those who stopped 5 to 12 weeks previously (OR = 2.70; 95% CI 1.73–4.21; P < 0.001) and 13 to 36 weeks previously (OR = 1.53; 95% CI 1.00–2.33; P = 0.050). Ex-users who stopped 37 to 96 weeks (OR = 1.01; 95% CI 0.66–1.57; P = 0.949), 97 to 180 weeks (OR = 0.73; 95% CI 0.45–1.19; P = 0.204), and 181 weeks previously or more (OR = 1.18; 95% CI 0.76–1.83; P = 0.456) had similar psychosis risk to those who had never-used cannabis.

Conclusion

Risk of psychotic disorder appears to decline with time following cannabis cessation, receding to that of those who have never used cannabis after 37 weeks or more of abstinence. Although, preliminary results suggest that frequent users of high potency types of cannabis might maintain an elevated risk compared to never users even when abstaining for longer than 181 weeks.

Introduction

Epidemiological and experimental research has shown that cannabis use is associated with increased risk of psychotic disorders in a dose-dependent manner.^1–6 Moreover, individuals who develop psychosis in the context of daily and high potency cannabis use experience earlier illness onset and more severe positive symptoms than patients who do not use cannabis.^5,7,8 Recent studies have also demonstrated that differences in the frequency and potency of cannabis use can account for variation in psychosis incidence between demographic regions.^3,9,10

More recently, studies taking advantage of genetic data have re-addressed the question of direction of the causal association between cannabis use and psychotic disorders. Indeed, some evidence has suggested that underlying genetic risk for schizophrenia is associated with patterns of cannabis use to some extent, although some of this might be explained by the underlying risk for cannabis use disorder, which is known to be genetically correlated.^11,12 Moreover, the most recent Mendelian randomization study supports a bi-directional association, indicative of this complex relationship, but with a larger effect size for cannabis use causing psychosis. ¹¹

In their commentary, ¹³ Gillespie and Kendler appraise the evidence from epidemiological and genetics studies and suggest that the association between cannabis use and psychosis phenotypes is partially causal and partially confounded by familial factors as well as reverse causality. They discuss new data from studies investigating incidence rates of psychotic disorders in countries where rates of cannabis use have changed because of its legalization. Such studies are now also adding evidence to the causal link between cannabis use and psychotic disorders.^14–16

There is evidence of improvements in anxiety, depression, and sleep following reduction or cessation of cannabis use among the general population, as well as some limited evidence of improvements in quality of life.^17–19 In people with serious mental illness, however, the negative impact of heavy cannabis use is more extreme, ²⁰ and evidence shows that people who cease cannabis use after developing a psychotic disorder have better clinical and functional outcomes compared to those who continue use,^21–23 regardless of medication adherence. ²⁴ However, it remains unclear whether the risk of developing psychosis can be lowered through stopping cannabis use prior to psychosis onset.^25,26 This relationship is clear for other substances of abuse and their harm, e.g., lung and pharyngeal cancer risk declines following cessation of tobacco and alcohol respectively.^27–29

Recent findings from multiple localities in the United States suggest that liberalization of cannabis legislation has been followed by higher rates of cannabis use, cannabis use disorder, and psychosis.^10,30,31 Given these findings and the global trend to more permissive cannabis laws, a greater understanding of the effects of cannabis cessation prior to psychosis onset can contribute to better informed public health policies, clinical care, and personal decision-making. Consequently, we aimed to establish whether the risk of psychotic disorders in cannabis users changes with time since cannabis cessation using data from the European Network of National Networks studying Gene–Environment Interactions in Schizophrenia (EU-GEI) study.

Methods

Study Design and Participants

The EU-GEI case–control study is a multi-centre study of first episode psychosis (FEP) patients and population controls recruited between May 2010 and April 2015 across 17 catchment areas in England, France, the Netherlands, Italy, Spain, and Brazil. ³² Ethical approval was provided by the relevant committees at each of the study sites, and all participants provided written informed consent. For our analysis, we only included the 11 EU-GEI study sites with complete data on cannabis use (Supplementary Figure 1). ³

Patients presenting with FEP were identified by trained researchers who carried out regular checks across each catchment area's mental health services. Patients were included if they: (1) were aged between 18 and 64 years; (2) resided within the catchment areas at FEP presentation; (3) and received an International Classification of Diseases, 10th Revision (ICD-10 F20-29, F30-33) diagnosis of psychosis. The Operational Criteria Checklist algorithm was used to ensure that a research-based diagnosis was received by every FEP patient. ³³ Exclusion occurred if an individual: (1) had previously been treated for psychosis; (2) met criteria for organic psychosis (ICD-10: F09); or (3) was diagnosed with transient psychotic symptoms resulting from acute intoxication (ICD-10: F1X.5). For a more detailed breakdown of diagnostic sub-categories see Rodriguez et al., 2021. ³⁴ FEP patients were assessed within 6 weeks of first contact with psychiatric services both in community and in patients’ settings. If the first presentation to Mental Health Services for FEP coincided with admission to hospital, patients were asked about their cannabis use up to the point of admission (e.g., where you a current user before coming to hospital?).

Control subjects were recruited using random and quota sampling strategies at each of the sites. The quotas were set based on local census data, ensuring that the controls were representative of each catchment area's population at risk. Exclusion of controls occurred if they had received a diagnosis of, and/or treatment for, a psychotic disorder. ³²

Measures

Sociodemographic data was collected by administering the Medical Research Council Sociodemographic Schedule. ³⁵ Data pertaining to patterns of cannabis use was collected using the modified Cannabis Experience Questionnaire. ³ To minimize recall bias, neither cannabis nor its role as a psychosis risk factor was referred to in any of the materials used for recruitment. Participants were initially asked if they had ever used cannabis. Those answering yes were then asked about their pattern of use, including the type of cannabis. This enabled participants to report the colloquial name.

We used the following measures of cannabis use in our analysis: (1) lifetime cannabis use, (2) current cannabis use, (3) age first tried cannabis, (4) duration of cannabis use, (5) frequency of cannabis use, (6) potency of cannabis use, and (7) time since cessation.

The variable estimating time since cannabis cessation was continuous (in weeks). Participants reporting both current cannabis use and a time since cessation value were excluded from our analysis. As this variable was only available for lifetime cannabis users, we created a separate variable termed “cannabis use groups” which categorized participants according to their lifetime cannabis use and duration of cessation. This was done to establish the approximate duration taken for the psychosis risk among those who ceased use (“ex-users”) to become equivalent to those who never used cannabis (“never users”). Due to the novel nature of this analysis, we had no indication of the appropriate groupings for ex-users. We first divided the controls into quartile groups and used these as thresholds for the overall sample. Based on the existing evidence reporting cannabis withdrawal symptoms that can precipitate psychosis and observation of the data distribution (Figure 1), we further divided the lowest quartile (most recent cannabis cessation) into 3 groups: ex-users who stopped 1 to 4 weeks ago, 5 to 12 weeks ago, and 13 to 16 weeks ago. Our cannabis use groups were as follows: never users = 0 (reference category); ex-users who stopped 181 weeks or longer previously = 1; ex-users who stopped 97 to 180 weeks previously = 2; ex-users who stopped 37 to 96 weeks previously = 3; ex-users who stopped 13 to 36 weeks previously = 4; ex-users who stopped 5 to 12 weeks previously = 5; ex-users who stopped 1 to 4 weeks previously = 6; and those who currently use cannabis (“current users”) = 7.

Figure 1.

The distribution of time since cessation in FEP cases and population controls in our EU-GEI sample.

The duration of cannabis use was a continuous measurement, measuring the years of cannabis use. This was derived through the following steps for non-current cannabis users:

1. Dividing weeks since cannabis cessation by 52.143 to establish the number of years since cannabis cessation;

2. Establishing the age cannabis use was ceased by subtracting a participant's years since cannabis cessation value from their age at the time of the study;

3. Subtracting a participant's age at cessation from the age they tried cannabis.

For those currently using cannabis, duration of use was established by subtracting the age they first tried cannabis from their age at the time of the study. Individuals with negative duration of use values whose current age matched the age they tried cannabis had their values recoded to 0 as we were unable to establish the exact difference between the age they tried cannabis and their current age. Individuals with negative duration of use variables that were greater than −1 were excluded entirely from our analysis due to the incorrect reporting of either their time since cessation, age tried cannabis, or current age variables. Individuals who had never used cannabis were assigned 0 as their duration of use value.

Statistical Analysis

Analyses were conducted using RStudio version 4.3.1. Descriptive analyses were conducted using t-tests and chi-squared tests. Adjusted linear and logistic regression models established if time since cessation predicted psychosis risk and whether psychosis risk differed across our cannabis use groups. Our regression models were adjusted for key sociodemographic variables (age, site, sex, and ethnicity) and lifetime tobacco use (0 = never smoked or ≤10 cigarettes per day; 1 = 11 cigarettes or more per day 3) as most cannabis users consumed cannabis with tobacco. The addition of each covariate led to an improvement of our models as assessed by Akaike information criterion (Supplementary Table 1). Additional models were adjusted for duration of cannabis use, frequency of cannabis use, and potency of cannabis use, which were added in a stepwise fashion to consider other cannabis use patterns known to impact case status.^1,3

Table 1.

Sample Sociodemographic Characteristics and Measures of Cannabis Use.

		FEP (N = 875)	Controls (N = 1,230)	Test-statistic (df)	P
Age	Mean (SD)	30.89 (10.59)	36.20 (13.39)	t = −10.15 (2079)	<0.001
Sex	Male (%)	539 (61.60)	576 (46.83)	χ2(1) = 43.95	<0.001
Site	Amsterdam	96	100	χ2(10) = 47.24	<0.001
	Barcelona	31	36
	Bologne	66	65
	Cambridgeshire	43	106
	Gouda and Voorhout	97	108
	Madrid	39	38
	Palermo	55	99
	Paris	45	100
	Puy de Dome	15	47
	Ribeirao Preto	189	302
	Southeast London	199	229
Ethnicity	White	516	923	χ2(5) = 70.63	<0.001
	Black	166	117
	Mixed	101	113
	Asian	31	33
	North African	38	23
	Other	22	19
Highest educational attainment	School, no qualifications	157	65	χ2(5) = 236.44	<0.001
	School, with qualifications	227	159
	Tertiary	194	333
	Vocational	150	175
	Higher (undergraduate)	101	313
	Higher (postgraduate)	36	177
Current employment status	Unemployed	353	708	χ2(1) = 22.77	<0.001
	Employed	399	511
Lifetime tobacco use	Lifetime users (%)	251 (28.69)	133 (10.81)	χ2(1) = 113.89	<0.001
Lifetime cannabis use	Lifetime users (%)	559 (63.89)	569 (46.26)	χ2(1) = 65.84	<0.001
Current cannabis use	Current users (%)	169 (19.31)	126 (10.24)	χ2(1) = 35.13	<0.001
Frequency of cannabis use	Rare	171	346	χ2(3) = 206.98	<0.001
	Weekly	99	86
	Daily	238	81
Potency of cannabis use	Low <10% THC	226	293	χ2(2) = 81.95	<0.001
	High ≥10% THC	283	221
Weeks since cessation	Mean (SD)	64.75 (106.12)	129.14 (125.88)	t = −7.50 (721.07)	<0.001
Duration of cannabis use (years)	Mean (SD)	6.69 (7.56)	6.18 (9.53)	t = 1.33 (1931.80)	0.185

df = degrees of freedom; SD = standard deviation.

Results

Participant Characteristics

A summary of demographic characteristics and cannabis use measures across our sample is presented in Table 1. The working sample for our analysis consisted of 875 FEP patients and 1,230 population-based controls. In comparison to controls, FEP patients were more likely to be current cannabis users (χ²(1) = 35.13; P < 0.001) and have had fewer weeks of cannabis cessation: FEP mean = 64.75, SD = 106.12; controls mean = 129.14, SD = 125.88 (t = −7.50 (721.07); P < 0.001). FEP patients were also more likely to use cannabis frequently (χ²(3) = 206.98; P < 0.001) and higher potency forms (χ²(2) = 81.95; P < 0.001). There was no significant difference in the duration of cannabis use between groups: FEP mean = 6.69, SD = 7.56; controls mean = 6.18, SD = 9.53 (t = 1.33 (1931.80); P = 0.185).

Does the Risk of Psychosis Change With Time Following Cannabis Cessation?

Linear regression adjusted for age, site, sex, ethnicity, country, and lifetime tobacco use showed a trend of declining psychosis risk each week following the cessation of cannabis use (Model 1: β = −0.002, 95% CI −0.004 to 0.000, P = 0.067). This effect remained similar when additionally adjusting in a stepwise manner for the duration of cannabis use (Model 2: β = −0.003, 95% CI −0.005 to −0.001, P = 0.013), frequency of cannabis use (Model 3: β = −0.002, 95% CI −0.005 to 0.001, P = 0.129), and potency of cannabis use (Model 4: β = −0.002, 95% CI −0.005 to 0.001, P = 0.133). The full results of each of our adjusted linear regression models are displayed in Table 2 and Supplementary Table 2. Figure 2 plots the decrease in psychosis risk over time for users based on the frequency and potency of their cannabis use, demonstrating that risk is highest and takes the longest time to decline for regular, high potency users.

Table 2.

Adjusted Logistic Regression Models Examining Whether Psychosis Risk Declines with Time Following Cessation and Across Different Cannabis Use Groups.

	Model 1: adjusted for age, sex, site, ethnicity, tobacco use			Model 2: adjusted for age, sex, site, ethnicity, tobacco use, and duration of cannabis use			Model 3: adjusted for age, sex, site, ethnicity, tobacco use, duration of cannabis use, and frequency of cannabis use			Model 4: adjusted for age, sex, site, ethnicity, tobacco use, duration of cannabis use, frequency of cannabis use, and potency of cannabis use
Predictors	β	CI	P	β	CI	P	β	CI	P	β	CI	P
Weeks since cessation	−0.002	−0.004 to 0.000	0.067	−0.003	−0.005 to −0.001	0.013	−0.002	−0.005 to 0.001	0.129	−0.002	−0.005 to 0.001	0.133
	Observations		720	Observations		719	Observations		652	Observations		652
	R2 Tjur		0.239	R2 Tjur		0.245	R2 Tjur		0.322	R2 Tjur		0.324

Figure 2.

Fully adjusted logistic regression model showing the probability of developing psychosis as a function of time for 6 different categories of ex-users.

Predictors	OR	CI	P	OR	CI	P	OR	CI	P	OR	CI	P
Current-users	1.60	1.16–2.20	0.005	1.94	1.26–2.97	0.003	2.42	1.42–4.11	0.001	2.03	1.17–3.52	0.012
Ex-users 1–4 weeks	6.89	3.91–12.14	<0.001	8.34	4.49–15.47	<0.001	10.10	4.87–20.96	<0.001	8.61	4.09–18.11	<0.001
Ex-users 5–12 weeks	2.70	1.73–4.21	<0.001	3.22	1.95–5.30	<0.001	4.46	2.35–8.44	<0.001	3.71	1.92–7.17	<0.001
Ex-users 13–36 weeks	1.53	1.00–2.33	0.050	1.86	1.14–3.04	0.014	3.05	1.61–5.78	0.001	2.61	1.36–5.03	0.004
Ex-users 37–96 weeks	1.01	0.66–1.57	0.949	1.25	0.75–2.08	0.392	2.02	1.05–3.91	0.036	1.72	0.87–3.38	0.117
Ex-users 97–180 weeks	0.73	0.45–1.19	0.204	0.91	0.52–1.59	0.742	1.90	0.95–3.77	0.068	1.70	0.84–3.41	0.138
Ex-users 181+ weeks	1.18	0.76–1.83	0.456	1.60	0.87–2.92	0.130	3.35	1.54–7.28	0.002	2.88	1.31–6.35	0.009
	Observations		1,914	Observations		1,909	Observations		1,838	Observations		1,838
	R2 Tjur		0.185	R2 Tjur		0.185	R2 Tjur		0.216	R2 Tjur		0.218

OR = odds ratio; CI = 95% confidence interval.

Does the Risk of Psychosis Differ Across Cannabis Use Groups?

Logistic regression adjusted for age, site, sex, ethnicity, and lifetime tobacco use indicated that current cannabis users had elevated psychosis risk compared to never-users (OR = 1.60, 95% CI 1.16–2.20, P = 0.005). In comparison to never users, those with the greatest psychosis risk were ex-users who stopped 1 to 4 weeks previously (OR = 6.89, 95% CI 3.91–12.14, P < 0.001); risk declined for ex-users who had stopped 5 to 12 weeks previously (OR = 2.70, 95% CI 1.73–4.21, P < 0.001), and declined further for ex-users who had stopped for 13 to 36 weeks (OR = 1.53, 95% CI 1.00–2.33, P = 0.062). Ex-users who stopped for 37 to 96 weeks were no different in their psychosis risk to never users, (OR = 1.01, 95% CI 0.66–1.57, P = 0.949), as were ex-users who stopped for 97 to 180 weeks (OR = 0.73; 95% CI 0.45–1.19, P = 0.204) and 181 or more weeks (OR = 1.18; 95% CI 0.76–1.83; P = 0.456).

These effects increased slightly across all use groups when additionally adjusting in a stepwise manner for duration of cannabis use (Model 2), frequency of cannabis use (Model 3), and potency of cannabis use (Model 4). Furthermore, in Model 3, which included frequency of use and Model 4, which also included potency of cannabis use, ex-users who stopped using cannabis 181 weeks or longer previously had an elevated risk of psychosis compared to never-users (Model 3: OR = 3.35; 95% CI 1.54–7.28; P = 0.002; Model 4: OR = 2.88; 95% CI 1.31–6.35; P = 0.009), although the wide confidence intervals for this group indicate that this result should be interpreted with caution.

The full results of each of our adjusted logistic regression models are displayed in Table 2 and Supplementary Table 3. Figure 3 provides a visual representation of the differences in the odds ratios and 95% confidence intervals of our cannabis use groups in Model 1.

Figure 3.

Forest plot displaying the odds ratio and 95% confidence intervals of the cannabis use groups in our adjusted logistic regression models.

Discussion

This is the first study to examine whether the risk for psychotic disorders associated with cannabis use changes following cessation of use in a sample of FEP patients and population controls from the EU-GEI study. Our findings suggest that psychosis risk declines following the cessation of cannabis but not independently of pattern of cannabis use. This is aligned with literature examining cessation of other substances of abuse and their harm.^27–29

We investigated whether the risk of psychosis changed across different time-windows of cannabis abstinence. In this analysis, our models unadjusted for frequency and potency of cannabis use, showed that ex-users who had stopped between 1 and 4 weeks, had a 7-fold increase in the probability of psychosis compared to never-users. Interestingly, this effect persisted slightly after 5 to 12 weeks of abstinence, with these individuals having a 3-fold increase in the probability of psychosis compared to never-users, which was greater than current cannabis users. However, in those who had stopped cannabis for 37 weeks or more, the risk of psychosis was no different to never-users, suggestive of a reduction of the previously acquired cannabis-associated risk. Of note, the lung cancer risk associated with smoking reduces annually following cessation but never fully attenuates, ²⁸ while other evidence suggests that the alcohol-associated pharyngeal cancer risk takes approximately 39 years of abstinence to reverse. ²⁹ However, there is evidence that normalization of biomarkers and risk factors associated with vascular dysfunction and disease begins soon after tobacco abstinence.^36–38

In models additionally adjusted for frequency and potency of cannabis use, we see a notable increase in the odds of psychosis across all groups, suggesting a modifying effect of patterns of cannabis use on the effect of cannabis cessation. These results follow the same overall pattern as the minimally adjusted models, where odds for psychosis declined until 181 weeks and then peaked slightly. This increase at 181 weeks is only significant when we adjust for cannabis frequency and potency. Although the wide confidence intervals for this group warrant a cautious interpretation and do not allow for formal interaction testing, it may be that the long-term risk for psychosis remains elevated compared to never users if we account for cannabis frequency and potency among former users. This further highlights the importance of considering cannabis potency and frequency of use, which have been previously shown to be highly relevant to clinical outcomes.^21,39,40 The difference in the findings from the model using time since cessation as a continuous variable and the models looking at categorical time intervals since cessation, suggests that the association between time following cessation and risk of psychosis may not be linear. Additional studies using longitudinal design will be needed to fully understand the extent to which psychosis risk may decline over time following cannabis cessation.

It should be noted that FEP patients were recruited within 6 weeks of first contact with psychiatric services. Previous research has demonstrated that as many as 40% of FEP patients will cease cannabis use within 12 months of onset ⁴¹ ; therefore, it is possible that many ex-users who stopped 1 to 12 weeks ago had stopped using cannabis within the 6-week window post-initial contact with psychiatric services and recruitment to the study. This may partially explain the increase in odds ratio for this group compared to never users. Once again, prospectively designed longitudinal studies are required to understand this association. A recent case series suggests that some people who use cannabis experience psychosis when stopping cannabis as part of cannabis withdrawal syndrome (CWS). ⁴² CWS typically occurs in heavy cannabis users after abrupt cessation, with up to half of people describing the symptoms as severe.^43,44 Heavy cannabis users are known to have a delayed onset due to the time-lag between abstinence and normalization of CB1 receptor expression.^45,46 This could, partially, explain the increased odds ratio of ex-users who stopped between 1 and 12 weeks previously; the withdrawal symptoms which include disrupted sleep and increase anxiety might contribute to precipitate psychosis onset. Further studies are required to fully understand CWS and its potential role in precipitating psychosis.

Our research shows evidence that the time since cessation findings are not explained by duration of use, which, in our sample, does not differ between cases and controls. In contrast, adjusting for frequency and potency of use does impact the effect of cannabis cessation. This is in line with our previous evidence that shows that regular use of highly potent (THC concentrations > 10%) has a significant impact on psychotic disorder risk. ³ Indeed, stratifying users into groups based on pattern of use (combined frequency and potency) shows that risk of psychosis may remain elevated in daily users for several years.

Limitations

The main limitation of this study is the case–control design leading us to rely on retrospective data which is prone to bias. ⁴⁷ Ideally longitudinal research should use prospective data to allow more accurate assessment of whether cessation of cannabis use leads to a decline in risk for psychotic disorders. Secondly, both our measures of cannabis use relied on self-report from our study participants, rendering responses liable to bias.^48,49 However, it should be acknowledged that biological measures are also prone to error due to the time-lag between cannabis cessation and THC elimination, particularly from hair samples and in formerly heavy users.^50–52 Therefore, thoughtful consideration of which metabolites and which sample type to examine should occur.

Implications

The findings of this study represent an important step in our understanding of the effects of cannabis cessation prior to the onset of psychotic disorder. Research in this area is of crucial importance for informing public health policy, clinical care, and personal decision-making, particularly given the recent global trend towards liberalization of cannabis legislation which appears to be increasing rates of cannabis use and CUD.^30,31,53 It seems that individuals should not only be made aware that cannabis use will increase their psychosis risk but that this risk can be attenuated through cessation, the latter approached cautiously to prevent cannabis withdrawal which in turn might precipitate psychosis onset.

In primary care settings, interventions targeting cannabis use reduction should be routinely offered to help-seeking individuals who wish to reduce/stop their cannabis use. The current evidence-based provides some support for the use of psychosocial interventions to this end.^54,55 More recent research has suggested that cannabinoids may effectively facilitate cannabis cessation and attenuate withdrawal symptoms.^56,57

Conclusion

This is the first study to suggest that the psychosis risk acquired from cannabis use declines following use cessation, especially after 37 or more weeks. It remains unclear if such decline is independent of the frequency and potency of the cannabis used.

While further research addressing the limitations in this study design is essential to fully elucidate the potential impact of cannabis cessation on risk for psychotic disorder, our findings could have important implications for public health messaging, primary prevention and personal decision-making.