The impact of cannabis legalization and decriminalization on acute poisoning: A systematic review

Abstract Background and Aims Many countries have recently legalized medicinal and recreational cannabis. With increasing use and access come the potential for harms. We aimed to examine the effect of cannabis legalization/decriminalization on acute poisoning. Methods A systematic review and meta‐analysis registered with PROSPERO (CRD42022323437). We searched Embase, Medline, Scopus and Cochrane Central Register of Controlled Trials from inception to March 2022. No restrictions on language, age or geography were applied. Abstracts from three main clinical toxicology conferences were hand‐searched. Included studies had to report on poisonings before and after changes in cannabis legislation, including legalization and decriminalization of medicinal and recreational cannabis. Where possible, relative risk (RR) of poisoning after legalization (versus before) was calculated and pooled. Risk of bias was assessed with ROBINS‐I. Results Of the 1065 articles retrieved, 30 met inclusion criteria (including 10 conference abstracts). Studies used data from the United States, Canada and Thailand. Studies examined legalization of medicinal cannabis (n = 14) and decriminalization or legalization of recreational cannabis (n = 21). Common data sources included poisons centre records (n = 18) and hospital presentations/admissions (n = 15, individual studies could report multiple intervention types and multiple data sources). Most studies (n = 19) investigated paediatric poisoning. Most (n = 24) reported an increase in poisonings; however, the magnitude varied greatly. Twenty studies were included in quantitative analysis, with RRs ranging from 0.81 to 29.00. Our pooled estimate indicated an increase in poisoning after legalization [RR = 3.56, 95% confidence interval (CI) = 2.43–5.20], which was greater in studies that focused on paediatric patients (RR = 4.31, 95% CI = 2.30–8.07). Conclusions Most studies on the effect of medicinal or recreational cannabis legalization/decriminalization on acute poisoning reported a rise in cannabis poisoning after legalization/decriminalization. Most evidence is from US legalization, despite legalization and decriminalization in many countries.


INTRODUCTION
Many countries have moved to decriminalize or legalize medical and recreational use of cannabis in the past three decades [1].Decriminalization involves removal of criminal consequences for possession of small quantities of cannabis, while legalization involves removing penalties and allowing for sale and possession of cannabis [2].In 1996, California became the first state in the United States (US) to legalize cannabis for medical use and possession.This was later implemented in some other states [3].Currently, medicinal cannabis use is legal in the District of Colombia and 37 states in the US [3].From these 37 states, 22 have also legalized recreational cannabis [4].Access to medical marijuana became legal in Canada in 1999 [5].Recreational use of marijuana has been legal in Canada since 2018 [6].Luxembourg has recently legalized recreational cannabis, and cannabidiol (CBD) products containing low concentrations of tetrahydrocannabinol (THC) are available in many European countries [7].In Australia, the Therapeutic Goods Administration legalized the medical use of cannabis and THC in 2016 [8].Medicinal marijuana was legalized in Thailand in 2019 [9].
Medicinal cannabis legislation expands treatment options for conditions including refractory epilepsy, while recreational cannabis legislation aims to allow safe use while reducing criminal activity and use of other drugs including synthetic cannabinoid receptor agonists (SCRAs).However, these changes can also have negative public health impacts.For example, recreational cannabis legislation in the US has been associated with a rise in motor vehicle accidents, alcohol and opioid abuse [10].Overall hospitalizations for cannabis and cannabinoid hyperemesis syndrome also increased in Colorado after legalization [11].The prevalence of cannabis use has increased in US states where medicinal cannabis is legal [12].
With increasing use and availability of cannabis in the community comes the risk of harms, including overdose and poisoning.Cannabis has a wide safety margin in adults; however, overdose can cause central nervous system (CNS) excitation, hallucinations, psychosis, CNS depression, bradycardia and dysrhythmias [13,14].Even when used as intended, cannabis use increases the risk of acute coronary syndrome in the hours following consumption [15].It is more likely to cause severe toxicity in children, manifesting as CNS and respiratory depression, coma, seizures and apnoea [16].Among the individual cannabinoids, THC is considered more toxic than CBD; however, CBD can cause gastrointestinal, neurological and hepatic symptoms.

Increased prevalence of cannabis poisoning has been observed
following legalization [17,18].Studies on poisoning typically use data from poisons centres.Poisons centre exposure calls can be triggered for a variety of reasons, including paediatric exposures (e.g. a child accessing edible marijuana products), a person experiencing signs/ symptoms of toxicity following therapeutic or recreational use, dosing errors with medicinal marijuana and as an agent in deliberate selfharm overdoses.Poisons centre coding of substances involved in exposures is typically extremely accurate, as it is coded by the poisons specialist handling the call.However, poisons centres underestimate the rates of exposures, as calls are voluntary.Other studies use hospital presentation or admission data, with reasons for presentation likely to be similar to reasons for calling a poisons centre.There are limitations in how well hospital data collects specific substances involved in poisonings, particularly with recreational drugs [19].A previous study identified a 34% increase in poison centre calls for cannabis exposure per year following legalization in Colorado [20].
Similarly, emergency department presentations due to cannabis toxicity increased from 1.2 to 2.3 per 100 000 population after legalizing recreational cannabis [20].
We conducted a systematic review to investigate the effect of changing cannabis legalization and decriminalization on poisoning.Specifically, we aimed to evaluate the effect of the legalization of medicinal cannabis, and legalization or decriminalization of recreational cannabis, on acute poisoning with cannabis and cannabinoids.This review was registered with PROSPERO (CRD42022323437); a brief protocol is accessible on the PROSPERO website.One amendment has been made to the pre-registered protocol: we removed SCRAs from the inclusion criteria, as they are not subject to similar legislative changes as cannabis, THC and CBD.

Inclusion and exclusion criteria
We included studies that investigated the changes in cannabinoid poisoning/exposure frequency or rates following cannabis legalization/ decriminalization.Cohort studies, observational studies, repeated cross-sectional and interrupted time-series (ITS) studies with pre-and post-intervention data sets were included.We focused on acute poisoning, and included studies that reported acute psychosis and acute cardiovascular events.There were no restrictions on age.
We excluded studies that reported on trends in SCRAs only, animal poisoning, chronic psychotic illness, hyperemesis syndrome, chronic cardiovascular disease, cancer and in-utero exposures/ developmental toxicity.Studies that discussed the legalization of cannabis but did not report trends in cannabinoid poisoning and those that reported change in cannabinoid poisoning without cannabinoid legalization/decriminalization were excluded.

Selection of studies
We exported studies into Covidence software (Veritas Health Innovation, Melbourne, Australia) and removed duplicates.Two authors (S.A. and J.L.) independently screened the titles and abstracts for eligibility, and irrelevant articles were removed.Conflicts were discussed with a third author (R.C.).Articles that passed the title and abstract screening were reviewed for inclusion criteria by full-text screening by two independent authors.A third author (R.C.) resolved conflicts between reviewers.

Data extraction
Data extraction was performed by two independent authors (S.A. and J.L.) with a structured data collection form.Data extracted included study geographic location, population characteristics, study design, time-period examined, year and details of the intervention, data sources, type of cannabinoid poisoning (CBD, THC, both), trends in cannabinoid poisoning (poisoning cases/incidence before and after the intervention), median age, clinical effects, clinical outcomes and any other impacts of the legalization or decriminalization.

Synthesis methods
Studies were stratified according to intervention type (decriminalization versus legalization; medicinal versus recreational use), study population (children/adolescents versus adults), data source (poisons centre versus hospital records) and literature type (grey literature versus peer-reviewed original research).Studies were eligible for quantitative synthesis if they provided raw counts or population rates.Raw counts were converted to population rates using publicly available population data.Relative risk (RR) was calculated as the rate of cannabis poisoning after the legalization/decriminalization divided by the rate of cannabis poisoning before the legalization/decriminalization.We calculated RR, confidence intervals (CIs) and heterogeneity (I 2 ) in Microsoft Excel using the MetaXL tool version 5.3 (www.epigear.com), using the random effects model.
We performed subgroup analysis based on characteristics of studies that we hypothesized may impact on estimates and contribute to heterogeneity.We examined legislation relating to medicinal use versus recreational use, as recreational legislation may result in more widespread use and therefore poisonings.We compared age groups (paediatric versus all ages), as children are at most risk of severe poisoning from cannabinoids.We stratified by data source (poisons centre versus hospital presentations/admissions data), as hospital data are likely to represent a more severe subset of exposures.We also looked for differences by study type comparing the more robust peerreviewed literature with grey literature (conference abstracts).Where possible, we also performed stratification analysis, e.g.where one paper included data from poisons centre and hospital events, these 'substudies' were stratified accordingly.GraphPad Prism 9.4 (GraphPad Software, San Diego, CA, USA) was used for data visualization.

Risk of bias
Two authors (S.A. and J.L.) assessed each study for quality and potential bias using the ROBINS-I (risk of bias in non-randomized studies of interventions) tool [21].For the 'risk of bias due to confounding' domain, simple pre-post studies were classified as high risk of bias.
Studies that presented simple pre-post data with some quantification of pre-intervention trends were classified as moderate risk.Studies employing interrupted time-series analysis, difference-in-difference analysis or comparison with a control (e.g. a state/province without legalization) were classified as low risk.The detailed outcomes of the assessment tool are summarized in Supporting information, Table S1.
Canadian recreational cannabis legalization has occurred in several stages and legalization of edibles has differed by province.Sale of dried cannabis flower and oils for non-medical use has been legal since October 2018, implemented by all provinces.Edible products (e.g.gummies) were legalized in October 2019 but each province could choose whether they approved sales of edibles.Alberta, British Columbia and Ontario approved sales of edibles in January 2020 [16,34,41].We found three studies from Canada reporting on the legalization of flower-based products and oils [16,34,41].Studies focusing on the 2018 legislation found no statistically significant increase in overall cannabis-related poisoning, but reported some increases for young adults [34] and paediatric exposures [41].However, Cohen et al. found increased severity of symptoms, with higher rates of critical care unit admission, respiratory involvement and altered mental status after 2018 legalization [16].
Canadian studies on legislation affecting edibles showed different results.A repeat cross-sectional study comparing both phases of legalization in Ontario, Canada found a significant increase in paediatric ED visits and hospitalizations, particularly following the introduction of edibles [47].In addition, a recent study compared Canadian provinces, where edibles are legal, to a control province which allows recreational marijuana but not edibles (Quebec).Provinces where edibles are legal had much higher rates of hospital admissions for paediatric cannabis exposures compared to the control province [48].Bennet et al. also compared states by legalization status in the US and found no differences when comparing states with legalized recreational marijuana to states with legalized medicinal marijuana as well as states without legalization [29].However, there was a 13-fold increase in total marijuana-related hospital presentations in children from 2004 to 2018 in the 27 states analyzed [29].
Only two studies specifically examined the effect of decriminalization of recreational cannabis (as opposed to legalization) [39,45].were not reported separately, so differences between these interventions cannot be specifically evaluated [45].
Shi & Liang examined recreational cannabis legalization and commercialization separately, using a quasi-experimental design with control states [33].Following legalization, they did not observe a significant increase when all exposures were included; however, there was a significant increase in unintentional exposures [33].Recreational cannabis commercialization resulted in a significant and consistent increase in poisons centre exposures [33].
This study included dried plant cannabis only (i.e.no edibles or concentrates), which may account for the lack of increase observed overall poisonings [33].Indeed, other studies have indicated that increases in recent years are not being driven by smokeable marijuana [30].

Quantitative synthesis
Twenty studies were suitable for quantitative synthesis, six studies of all ages [24,28,30,35,37,46], one of adults [34] and the remainder focusing on the paediatric population [16, 17, 20, 22, 25, 27, 39-43, 47, 48].Due to partial data overlap results from the following studies were pooled: Wang 2017 [46] and Wang 2020 [30]; Wang 2016 [20] and Wang 2019 [40]; and Wang 2012 [25] and Wang 2013 [17].One study used data from Thailand, five used Canadian data and the remainder were from the US.The rates of cannabis poisoning generally increased after interventions with a pooled RR estimate of 3.56 (95% CI = 2.43-5.20,random-effects model, Figure 2).However, results from legalizing/decriminalizing medicinal or recreational cannabis were highly heterogeneous (I 2 = 97%).The RR varied greatly, from 0.81 to 29.00 [16,17].There was one negative study, one where the 95% CI crossed one, the remainder had RR above one, and 95% CI not crossing one.We conducted several subgroup analyses, and RR increased after the interventions for all the study subgroups we explored (Table 2).RR appeared to be lower for studies that only used hospital data (RR = 3.42, 95% CI = 1.60-7.31)and studies that included all ages (RR = 2.63, 95% CI = 1.78-3.88),and higher for I G U R E 2 Forest plot displaying the relative risk (RR) of cannabis poisoning following the intervention from all studies where raw numbers were available for quantitative synthesis.A random effects model was used.RRs have been plotted on a logarithmic (base 10) scale.Studies have been ordered by duration of time between the main intervention and the final data point (shortest on the bottom).those focused on paediatrics (RR = 4.31, 95% CI = 2.30-8.07).Studies that focused on the medical use of cannabis reported a higher RR and were less heterogeneous (RR = 3.67, 95% CI = 2.36-5.71,I 2 = 72%) than those that included the recreational use of cannabis (RR = 3.49, 95% CI = 2.22-5.48,I 2 = 97%).

Clinical effects, outcomes and disposition
Reported clinical effects of the poisonings were described by many studies, including lethargy, drowsiness, dizziness, hypertension, palpitations, tachycardia, nausea, vomiting, irritability, agitation, coma and CNS depression (Supporting information, Table S2).CNS depression was most commonly reported in children (Figure 3).There was a total of 72 reports of coma from all studies; however, there may be some overlap between data sources used.The majority of paediatric exposures were managed in a health-care facility (range = 55-100%), intensive care unit (ICU) admissions were also common (Figure 4).'Minor effect' and 'moderate effect' were the most frequently reported outcomes in children (Figure 5).Studies that reported on all age groups reported more patients as asymptomatic ('no effect') than those focusing on children.One death was reported after legalization of recreational marijuana in Colorado (Wang 2016 and Wang 2020 appear to report the same paediatric case, Supporting information, Table S2) [20,30].

Quality assessment
We assessed the quality of each study using the ROBINS-I tool (Supporting information, Table S1).Most studies received a score of low to medium risk of bias.Main sources of bias were not accounting for time-varying confounders (also displayed in Table 1), missing data, lack of detail about the intervention and imprecision in the intervention time (e.g. in the US where legislation varies across states and time).Almost half (n = 14) the studies simply presented pre-and post-intervention counts/rates, without accounting for broader trends [17, 24, 25, 27, 28, 34-36, 38-40, 42, 44, 45], and some presented some quantification of pre-intervention trends [16,30,31,37,43].Only a few stronger quasi-experimental study designs were used, including an interrupted time-series analysis [41], a repeat cross-sectional study with segmented regression T A B L E 2 Relative risk (RR) of cannabis poisoning following legalization by subgroups.analysis [48] and quasi-experimental difference-in-differences design, with comparison to control states [29,33].Several studies adjusted for confounding by using control states/provinces where cannabis laws were unchanged or compared to overall national trends [20,22,23,26,32,40,47].Two studies using a more robust approach were negative overall (but showed significant increases for some subgroups) [33,41].

DISCUSSION
This systematic review found that cannabis legislation or decriminalization was associated with an increased incidence of cannabis poisoning.The probable explanation is that the legislation increased use which also increases poisoning.The modification of cannabis laws could increase perceived acceptability of cannabis use in the community, increasing use and dependence [49].Increased medical and nonmedical use can increase the risk of poisoning due to availability and access.Increased use and availability of edibles appears to be an important driver of the increase in poisonings, particularly in the paediatric population.Despite legalization and decriminalization in many other countries, the vast majority of evidence was from the US and Canada.
Most studies included in this review reported increases in poisoning exposures following legalization/decriminalization of either medicinal or recreational cannabis.Studies which did not report a statistically significant change for the primary outcome of overall poisoning often found increases among subgroups, e.g.children [34,41] or ICU admissions [16].The three studies in Canada which reported no significant change in cannabis exposure after recreational cannabis legalization were in the earlier phase of legalization, where flower products were legalized before edibles [16,34,41].There was also an additional lag between legalization of edibles and their commercial availability for purchase [41].Changes in paediatric poisonings due to edibles have probably not yet been covered in these study periods.
The only study with sufficient data post-edible availability showed a    [29].However, these comparisons may be confounded by people purchasing marijuana from neighbouring states with more liberal laws [29].
In contrast to North America, Uruguay's recreational cannabis legalization is heavily controlled by the government.Uruguay has a mandatory registry and sets a limit to grams of cannabis used by each individual [50].They do not allow advertising, selling to tourists or edibles [50].The different legalization models would be expected to have a different impact on outcomes.However, we could not find any studies assessing the impact of the Uruguayan model on poisonings.
The assumption that increased poisonings are a result of increased use requires examination of the impact of legislation on overall use patterns.Cannabis use in the last year increased 70% in Canada post-legalization [51], with a particularly large increase in use of edibles following legalization of their retail sale [52].In the US, Cerdá and colleagues examined the impact of recreational cannabis legalization on use [53].Adjusting for trends observed in control states, they found that cannabis use did not increase in adolescents and young adults following legalization of recreational cannabis.However, frequent use increased in adults.Similar results were observed following medicinal marijuana legalization [54].Increasing use in older adults could increase poisoning across all age groups by increasing availability in households.A comparison of prevalent use in states where cannabis is legal versus illegal shows higher rates of use in legal states, particularly with edibles [52].Taken together, this indicates that patterns of use are influenced by legalization, although it can be difficult to separate this effect from baseline trends.This underscores the importance of accounting for time-varying confounding (discussed further in 'Strengths and limitations', below).
Cannabis exposures in children are of particular concern, as they have the potential for severe toxicity, including coma, neurological and cardiovascular effects.While exact circumstances surrounding exposures was not the focus of this review, adverse reactions or dosing errors with medicinal cannabis products used in children were rare [27].In support of this, one paediatric study found low (< 10) counts of diagnoses of conditions for which medical cannabis is commonly used (e.g.epilepsy, cancer) [40].Rather, exposures in young children tend to be unintentional exposures to cannabis products owned by parents, older siblings or care-givers [27].These findings are potentially due to the significant rise in the availability of edible products such as cookies, gummies and brownies, which are attractive to children [49,55].Harm minimization approaches need to be applied, including the sale of edible cannabis products in childresistant containers, opaque packaging and enhancing parents' awareness of the risk of edible marijuana for children [56].The increase in paediatric cannabis exposures may also be driven by increasing recreational use of cannabis among older children and adolescents [27].
We found that data on the impact of cannabis decriminalization are lacking, with most studies focusing on legalization.Decriminalization was often followed by legalization, and studies tended to investigate the effect of legalization together with decriminalization [39].
Publication bias is another potential reason for the limited data on decriminalization (which would have been expected to have much less of an impact on use and misuse).There may be some unpublished studies that did not report a significant change in poisoning after the decriminalization [57].
A minority of studies reported other impacts of the intervention in addition to poisoning.Data on the co-ingestion of other substances conflicted between studies [32,41].One study reported an increase in alcohol intake after cannabis legislation [32].A prior study reported that cannabis use is significantly associated with a rise in the ingestion of alcohol [58].In contrast, one study reported a decrease in cocaine, alcohol and opioid consumption in adolescents [41].Reduced consumption of opioids is consistent with a previous systematic review, which reported that the legalization of medical cannabis led to a decrease in prescription opioids dispensing [59].That research found a decrease in opioid overdose mortality by 21% in states with medicinal cannabis laws, 1999-2010 [59].However, this result was not replicated when extended to 2017, with a 23% increase in opioid mortality seen [60].
SCRAs are more dangerous than naturally occurring cannabinoids, with a different pharmacological profile including psychostimulant-like properties [61].These products are widely used in the community and present a law enforcement challenge [62].Most users prefer natural cannabis to SCRAs and cite perceived legality and the fact that SCRAs are not detectable on routine drug screening as reasons for use [63,64].One potential impact of the legalization of recreational cannabis is a reduction in harm from SCRAs.A study has shown a reduction in SCRA exposures following medical and recreational cannabis legalization [65].Further research is needed in this area.

Strengths and limitations
This study had several strengths.A comprehensive search strategy was used, with no language restrictions.Grey literature conference abstracts were also included, which provided more data.Screening and data extraction was performed by two independent authors, with conflicts resolved by discussion with a third author, reducing the risk of bias in the selection of studies.
Despite these strengths, this review had some limitations.We focused on acute poisoning, and do not report on harms from longerterm exposure such as hyperemesis syndrome, chronic psychosis and chronic cardiovascular illness.Hyperemesis syndrome can be severe and life-threatening, and thus we may be underestimating harms from subacute toxicity by excluding this.In addition, there was often a delay between law enactment and implementation of the intervention in the community, including the opening of cannabis dispensaries [32,42].This lag can influence the interpretation of any changes in health outcomes or limit the findings from studies with a short post-intervention period.Our quantitative analysis is limited by the high heterogeneity observed, which was able to be slightly reduced by subgroup analysis.We examined exposures to cannabis, as well as THC and CBD specifically.However, we found no studies that specifically evaluated different isomers (i.e.delta-9-THC versus the newly emerging delta-8-THC and delta-10-THC) [66].
The main quality issue of the identified studies was high risk of bias due to time-varying confounding.Many studies simply presented pre-and post-intervention counts/rates, rather than using more robust quasi-experimental designs.Increases observed by those studies could be simply a result of longer-term trends.This limitation also applies to our quantitative synthesis.Other factors, besides legalization, could influence trends in adult cannabis use, which are not accounted for with a simple pre-post study design.This includes changing community attitudes to cannabis use (reductions in perceptions of risk since 2000) [67] and the increasing potency of the illicit cannabis supply [68].However, the fact that increases seem to be more influenced by the availability of edible forms of cannabis supports the conclusion that the potency of flower/plant products is not a major driver of the increase.Future studies should employ robust methods, including difference-in-difference analysis, interrupted timeseries analysis and/or use of control jurisdictions to increase confidence in findings.

CONCLUSION
Medicinal or recreational cannabis legalization is associated with increased cannabis poisoning in adults, adolescents and children.However, there was much unexplained heterogeneity in the outcomes measured.The majority of evidence is from North America, although medicinal cannabis is legal in many other countries, and recreational cannabis in some countries [2][3][4][5][6][7][8].It is important that public health agencies consider applying harm minimization approaches to limit the impact of cannabis legislation on acute poisonings, especially as legalization or recreational cannabis continues to be debated.With changing attitudes and perception of risk, there is a need for greater public awareness of the risks of cannabis poisoning, particularly to young people.
We searched Embase (11 March 2022), Medline, Scopus and Cochrane Central Register of Controlled Trials (CENTRAL) (14 March 2022) without language restrictions from inception until March 2022.Our search strategy focused on three main concepts: cannabinoids, legalization/decriminalization (intervention) and poisonings (outcome).The full search strategy is available in Supporting Information.We performed forwards and backwards citation chaining of included papers to search for additional relevant papers, with the final citation chaining completed in August 2022.We examined grey literature by hand-searching conference abstracts from the annual conferences of the three main international clinical toxicology societies: Asia Pacific Association of Medical Toxicology (APAMT), the European Association of Poisons Centres and Clinical Toxicologists (EAPCCT) and the American Academy of Clinical Toxicology (AACT) for the past 5 years.

F
I G U R E 1 Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) diagram of study selection.*Additional sources include grey literature and citation chaining.T A B L E 1 Main characteristics of included studies.

F
I G U R E 3 Heat-map displaying the percentages of clinical effects of cannabis exposures.Unless otherwise indicated, studies were of the paediatric population.Cells shaded grey indicate the study did not report that parameter.a Study focused on all ages; *study only examined cannabidiol exposures.

4
Heat-map displaying the percentages of people managed at a health-care facility, admitted, and admitted to critical care/intensive care.Studies only reporting on hospitalized patients are indicated with the square bracket.Unless otherwise indicated, studies were of the paediatric population.Cells shaded grey indicate the study did not report that parameter.a Study focused on all ages; # study of adolescents only; *study only examined cannabidiol exposures.