INTRODUCTION
In 2013, illegally manufactured fentanyl began to emerge within the unregulated drug supply across Canada and the United States (US), quickly becoming a major driver of the current drug toxicity crisis [1, 2]. In the United States, there was a 100.7% increase in deaths involving synthetic opioids (predominantly fentanyl) between January 2020 and 2023 [3]. Similar trends have been observed in Canada, where over 80% of opioid‐related deaths in 2023 were attributed to fentanyl [4].
Despite substantial geographic variation in opioid‐related mortality across Canadian provinces and territories (from 41.9 to 4.5 per 100 000 in British Columbia and Prince Edward Island, respectively) [4], the unregulated drug supply continues to become increasingly volatile. Drug‐checking services across the country have detected a rising presence of nitazenes [5], xylazine [6, 7] and synthetic benzodiazepines [8] within fentanyl‐containing samples. Nitazenes, a class of synthetic opioids estimated to be several times more potent than fentanyl, have been increasingly detected in the drug supply since 2021 [9], while xylazine has become a common cutting agent added to fentanyl [10] and has been implicated in complicated infections since 2020 among people exposed through the unregulated drug supply [11].
There is limited national data on the composition and variation of substances circulating in Canada's unregulated opioid supply. Most existing data is obtained through the analysis of post‐mortem toxicology results following substance‐related toxicity deaths [12] (which represent a small fraction of people at risk of substance‐related harms) and drug checking services (which are under used and often restricted to large urban centres) [13, 14]. Samples from the unregulated supply are tested through Health Canada's Drug Analysis Service (DAS), which provides scientific and technical services to aid Canadian law enforcement agencies in their activities involving unregulated drugs [15].
The majority of samples analysed by DAS are obtained through drug seizures conducted by law enforcement agencies [15], with over 120 000 samples tested annually [15]. Although these data are not completely representative of all substances circulating within the unregulated drug supply, they can help provide important insights, with drug seizure data previously used to assess how the unregulated drug supply varies geographically across Canada [16, 17, 18, 19] and the United States [20, 21, 22]. Understanding current trends in the composition of the fentanyl‐dominated unregulated drug supply has important implications given the heightened risk of harm associated with the use of multiple sedating substances [23, 24].
Therefore, we sought to describe trends in the number and composition of fentanyl‐containing drug samples seized by law enforcement agencies across Canada.
METHODS
Design
We conducted a repeated cross‐sectional analysis [25] of all drug samples seized by law enforcement agencies and analysed by Health Canada's DAS that contained fentanyl between 1 February 2020 and 31 December 2024 across Canada. This study is reported as per the Reporting of Studies Conducted using Observational Routinely Collected Data (RECORD) checklist [26]. The study protocol and analysis plan were not pre‐registered via a publicly available registry, therefore, our findings should be considered exploratory.
Data sources and preparation
Our study used anonymized data distributed by DAS, who determine the chemical content of drug samples seized by Canadian law enforcement agencies and those voluntarily submitted by public health authorities. Samples seized by law enforcement agencies including police forces, the Canada Border Services Agency and the Correctional Service of Canada are routinely submitted to DAS for testing to obtain a Certificate of Analyst, which supports investigations and court evidence [15]. Health Canada scientists analyse all samples on a first‐in, first‐out basis, typically returning results within 45 days [15].
Testing is performed at three regional laboratories in Toronto, Montreal and Vancouver, following established detection thresholds for various substances [15]. Information regarding analysed drug samples is released on a monthly basis, which includes detected substances, location of the client who submitted the sample for testing, physical description, whether the sample was submitted by law enforcement or a public health authority and the date each was received by DAS and returned to the client. We compiled data for all samples received by DAS between February 2020 and December 2024, as before February 2020 DAS did not provide information regarding physical description and the date each sample was received.
We aggregated data based on the date drug samples were received by DAS (as opposed to analysis completion date) to more accurately depict time trends in the composition of Canada's unregulated drug supply, because of the potential lag between initial submission date and when results are returned.
We performed several data cleaning steps to classify each substance identified within the sample of interest using the chemical names provided by DAS. This included the classification of notable drug classes (i.e. non‐fentanyl opioid, stimulant, benzodiazepine, hallucinogen, cannabis, xylazine or unclassified), drug subclasses and drug names, where applicable. For example, among the opioid drug class, we further classified samples based on opioid type, defined as the opioid subclass (e.g. fentanyl, heroin, hydromorphone, etc.).
A full list of all drug classes and subclasses can be found in Table S1. We did not classify all substances into drug classes, and therefore, we also looked at the unique chemical substances (including fillers and buffers) within each sample. Finally, we created count variables corresponding to the number of unique drug classes, subclasses and individual chemical substances contained within each sample.
For counts of chemical substances, we considered all reported substances (regardless of whether classified or unclassified). All data cleaning was completed independently by two members of the study team (S.L. and R.G.), with an inter‐rater agreement of 91.7%. All discrepancies were reviewed and resolved through consensus with an external pharmacist consulted as needed.
Our study was restricted to seizures (1) that contained at least one substance classified as fentanyl or a fentanyl analogue; (2) described as powder and/or grainy substance, tablets, crystalline substance or a rock‐like solid; and (3) submitted by law enforcement. Drug samples submitted by public health authorities were excluded because of their limited volume (<4000 over the study period) and inconsistent submission practices across regions. These samples are more likely to be associated with toxicity events and, therefore, may not be reflective of the broader drug supply.
Finally, to calculate population‐adjusted rates, we identified the annual population size of each province and territory using de‐identified, aggregated, publicly available data from Statistics Canada [27]. Because of the secondary use of anonymous data in this study, no review by a research ethics board was required as per the Canadian Tri‐Council Policy Statement (TCPS) Article 2.4 [28].
Study measures and analysis
Our primary study measures were annual counts and crude rates of fentanyl‐containing drug seizures per 100 000 population, stratified by province and territory. We also reported the quarterly proportion of seizures that contained only fentanyl as compared to fentanyl in combination with at least one other drug class, from the second quarter of 2020 onward (because of incomplete data in January 2020). For each fentanyl‐containing seizure, we reported the number and proportion that contained each drug class, the top five most frequently detected drug subclasses (for opioids, benzodiazepines and stimulants), the number of chemical substances and drug classes detected in each sample (grouped as: 1, 2, 3, 4, ≥5) and the physical description of each sample, in each year over the study period.
We used the Cochrane‐Armitage test for trend with Monte‐Carlo estimates to assess for significant changes in the proportion of fentanyl‐containing seizures within each category between 2020 and 2024. We applied the Bonferroni correction to limit the likelihood of false positive error rate because of multiple comparisons and, therefore, used a type I error rate of P ≤ 0.001 (P = 0.05/41).
We reported all measures overall and stratified by year and each region across Canada. Annual data for 2020 covers the period from February to December. Because of small counts, we followed Statistics Canada's Standard Geographical Classification [29] to group together the four provinces of Atlantic Canada (New Brunswick, Nova Scotia, Newfoundland and Labrador and Prince Edward Island) and the three Northern Territories of Canada (Northwest Territories, Nunavut and Yukon).
All analyses were conducted using SAS version 9.4 (SAS Institute).
Involvement of people with lived and living experience
The Ontario Drug Policy Research Network's (ODPRN) opioid‐related research is informed by a Lived Experience Advisory Group, comprised of individuals with lived and living experience using opioids. We discussed this project at an ODPRN Lived Experience Advisory Group meeting, answered questions and asked which members were interested in participating. Because the full group was interested, we discussed this at regularly scheduled meeting where six members attended.
Study results were shared with the group who provided insights that informed our interpretation of findings.
RESULTS
Trends over time
Between February 2020 and December 2024, we identified 545 319 drug seizures (see Table S2 for overall seizures for 2020 and 2024); of which 3722 were submitted by Public Health Authorities, 86 861 were described as a residue, syringe, controlled‐release beads or derma patches and 343 693 did not contain fentanyl or a fentanyl analogue, resulting in 71 996 fentanyl‐containing drug seizures over the study period. These seizures were submitted from law enforcement agencies located in 522 cities across Canada (Table S3). The quarterly number of fentanyl‐containing seizures rose 24.4% (from 2640 to 3284), with annual population‐adjusted rates declining slightly (34.5 to 33.1 per 100 000) across Canada (Table 1).
In the second quarter of 2020, most fentanyl‐containing seizures did not contain another drug class (66.2%; 1748/2640 seizures), however, this changed over time, with 47.3% of fentanyl‐containing seizures containing at least one other drug class by the final quarter of 2024 (Figure 1).
Cochrane Armitage test for trend
Overall, we observed significant increases in fentanyl‐containing seizures where at least one benzodiazepine (13.4%–40.2%; P < 0.001) or xylazine (1.5%–18.9%; P < 0.001) was detected, with seizures containing either cannabis or hallucinogens remaining extremely rare. Although the proportion of fentanyl‐containing seizures where at least one stimulant decreased over the study period (7.7%–6.2%; P < 0.001), we observed a peak in 2023 (16.4%). Among types of fentanyl, fentanyl itself was the most commonly detected in 2020, but declined over the study period (94.3%–64.7%; P < 0.001), whereas para‐fluorofentanyl (0.0%–46.3%; P < 0.001) and methylfentanyl (0.0%–28.6%; P < 0.001) increased correspondingly.
Among benzodiazepines, in 2020 the most common type was etizolam, which declined over the study period (11.7%–1.2%; P < 0.001) and was replaced by bromazolam (0.0%–33.7%; P < 0.001), which rapidly grew over the study period. Among stimulants, the presence of both methamphetamines (5.3%–3.6%; P < 0.001) and cocaine (2.5%–2.3%; P < 0.001) declined slightly overall, although both peaked in 2023 at 9.2% and 10.3%, respectively. In contrast, the presence of non‐fentanyl opioids declined over the study period (14.5%–4.2%; P < 0.001), with decreases in heroin (11.9%–2.6%; P < 0.001), with the exception of a small rise in nitazenes in 2021 (5.2%) before declining thereafter (0.9%–0.8%; P < 0.001).
The majority of fentanyl‐containing seizures were qualitatively described as a powder over the study period (96.8%–97.5%; P < 0.001). In parallel, we observed minor changes in the proportion of seizures described as tablets (2.5%–1.3%; P < 0.001) or a crystalline substance (0.4%–0.6%; P < 0.001). Overall, less than one‐third of all seizures contained only one (4.2%) or two (26.1%) chemical substances, with the proportion of seizures containing three or more chemical substances increasing over time.
In particular, seizures containing four (12.8%–24.7%; P < 0.001) and five or more (5.7%–25.2%; P < 0.001) grew significantly over the study period. When assessing the presence of drug classes (Table S4), we found that over half (50.6%) of all seizures contained fentanyl with at least one other drug class.
Trends by province and territory
Trends in fentanyl‐containing seizures stratified by province and territory are presented in Figure 2, Figures S1–S2 and Table S5. Over the study period, most provinces saw increases with the largest rise occurring in Manitoba (13.7–19.1 per 100 000) and Atlantic Canada (0.9–10.7 per 100 000). By 2024, Quebec (2.9 per 100 000) and the Northern Territories (5.2 per 100 000) had the lowest rates, although Quebec (1.8 –3.0 per 100 000) and Atlantic Canada (0.9–10.7 per 100 000) experienced the largest relative increases.
In contrast, we observed declines in British Columbia (88.2 to 60.4 per 100 000) and stable rates in Ontario (41.7 to 41.4 per 100 000) over the study period, however, both provinces still had among the highest seizures rates overall.
In 2024, less than one‐quarter of seizures contained only fentanyl in Manitoba (15.5%) and Quebec (20.7%) rising to approximately half in British Columbia (48.9%) and Ontario (52.6%). The composition of seizures varied widely, for example, in 2024, benzodiazepine detection ranged from 14.2% in the Northern Territories to 83.1% in Manitoba. Similarly, only 1% of seizures contained a stimulant in Manitoba (1.4%) and Saskatchewan (0.7%), to over 25% in Atlantic Canada (26.1%) and the Northern Territories (28.6%).
DISCUSSION
We observed a significant increase in the rate of fentanyl‐containing drug seizures where fentanyl was detected in combination with at least one other drug class, with over 50% containing fentanyl alongside at least one other drug class by the end of the study period. The rising prevalence of benzodiazepines and xylazine, combined with the evolving composition of benzodiazepine types (most not licensed for use in Canada) and growing detection of potent fentanyl analogues, is particularly concerning because of the heightened risks associated with the combined use of multiple sedating substances [23, 24] particularly among people not intending to use these in combination. Although DAS data represent only a subset of the unregulated drug supply, we generally found that provinces with the highest rates of fentanyl‐containing seizures also had high rates of accidental opioid toxicity death [4].
This pattern reinforces the association between the presence of fentanyl within regional drug supplies across Canada and the United States and opioid‐related harms reported in post‐mortem toxicology data for several years [4, 30, 31, 32]. In 2024, the highest rates of fentanyl‐containing seizures were observed in British Columbia, Alberta and Ontario, which also had the highest national rates of opioid toxicity deaths in the same year [4]. The increasing presence of multiple drug classes in fentanyl samples highlights the value of drug seizure data as part of ongoing drug surveillance, providing important insights into the evolving composition of the unregulated drug supply.
These findings support the continued need for targeted harm reduction and evidence‐based treatment, including the expansion and standardization of drug checking services, greater access to supervised consumption sites, prescribed safer supply and a focus on improving retention on opioid agonist treatment (OAT), especially in regions where the drug supply appears to be increasingly volatile.
The regional variation in fentanyl‐containing seizures we observed further highlights the shifting landscape of Canada's unregulated drug supply as trends in fentanyl seizures likely result from a combination of dynamics including changes in the unregulated drug supply, enforcement practices and procedural differences in sample submission to DAS. Although Alberta and Ontario had consistently high fentanyl‐containing seizures rates, we observed a decline in seizures in British Columbia over the study period, despite being the province with the highest overall rate in 2020. This decrease may be linked to changes in law enforcement practices following the implementation of the province's decriminalization policy in January 2023 [33], which removed criminal penalties for possession of small quantities of certain substances [33] and may have reduced the number of seizures submitted by law enforcement agencies for testing.
However, despite a reduction in arrests, preliminary evidence suggests that police departments may continue to seize drugs intended for personal use, requiring people who use drugs to more frequently engage with the unregulated market [34, 35]. In contrast, provinces with historically lower rates of fentanyl‐related harms such as those in Atlantic Canada and Quebec [4], demonstrated some of the largest relative increases in fentanyl‐containing seizures. This may reflect shifts in the unregulated drug supply across eastern Canada, where fentanyl has only more recently begun to emerge [4].
We also found that the composition and physical characteristics of fentanyl‐containing seizures varied nationally, highlighting the often unpredictable formulation of substances. In addition to fentanyl, seizures often contained stimulants and sedating substances such as benzodiazepines, xylazine and nitazenes, contributing to a highly toxic unregulated drug supply. For example, in 2024, benzodiazepine positivity rates ranged from approximately one‐third in Ontario and British Columbia, to over 65% in Saskatchewan and Manitoba.
The proportion of fentanyl‐containing seizures where a stimulant was detected also varied, from 5% in provinces such as British Columbia and Alberta to over 25% in Atlantic Canada and the Northern Territories. This variability highlights the significant regional variation in the drug supply in different areas of Canada, including in potency, pharmacological effects and potential drug–drug interactions, which pose serious challenges for people who use drugs [36, 37, 38]. We also observed notable regional differences in the physical descriptions of seizures, particularly those described as tablets.
In 2020, across Atlantic Canada, nearly one‐third of fentanyl‐related seizures were tablets, compared to just 2.2% nationally. Because of the small number of samples this should be interpreted with caution. However, in markets that have historically high rates of use of prescription opioids [39, 40], the potential increase in counterfeit pressed pharmaceutical tablets with unknown compositions in the unregulated market presents a significant risk for toxicities.
Together, these findings emphasize the urgent need to reduce harms from the increasingly unpredictable drug formulations by expanding access to a range of services for people who use drugs, including safer supply programs, which can provide regulated alternatives to the unregulated drug supply through prescriber [41, 42] and non‐prescriber‐based [43, 44] models.
Our finding that almost half of all drug samples submitted to DAS that contained fentanyl in 2024 also contained a benzodiazepine (an increase from just 13.4% in 2020) is important, as harms related to benzodiazepines have risen across Canada in recent years [38]. In British Columbia, benzodiazepines that are not licensed for use in Canada were detected in over half of all substance toxicity deaths in January 2021, a three‐fold increase from 6 months prior [45]. In Ontario, 45.6% of opioid‐related deaths in the first 10 months of the coronavirus disease 2019 (COVID‐19) pandemic also had a benzodiazepine detected [46].
Similarly, data from Toronto's Drug Checking Service found that between 2023 and 2024, approximately half of all samples that were expected to be fentanyl also contained a benzodiazepine [47]. The high proportion of fentanyl‐containing seizures with benzodiazepines we observed is alarming, as toxicities resulting from opioid‐benzodiazepine combinations can present with atypical and more complex symptoms [48], making these events harder to recognize and treat. Although medications including flumazenil [49] may be used in clinical settings to manage benzodiazepine sedation, their use remains limited with no safe, widely accessible medications available for lay responders at the community level to reverse these effects [50].
Inadvertent exposure to benzodiazepines in the unregulated drug supply can lead to dependence and complicate the treatment of opioid withdrawal and opioid use disorder [45]. Although the shift to fentanyl in the unregulated drug supply and its health impacts are well recognized [13], the harms of unintended benzodiazepine exposure remain less acknowledged, even though many people who use opioids from the unregulated supply are involuntarily exposed to benzodiazepines with little ability to avoid them [51]. Given that novel benzodiazepines are not always identifiable through drug testing services [52], the composition of unregulated substances is often unknown, underscoring the value of safer supply programs that enable individuals to make informed choices regarding the substances they consume.
Several limitations of our study warrant discussion. First, DAS data may not completely represent drug seizures in Canada or substances circulating on the market because they rely on samples submitted by law enforcement agencies. Submission decisions may vary geographically and be influenced by policy changes and structural differences between agencies, limiting our ability to assess changes in total seizures, including those not submitted to DAS.
Regions with higher per capita seizure rates may reflect more active enforcement or greater DAS engagement, rather than differences in the unregulated drug supply. Without validation against other data, our findings should be interpreted with caution. Additionally, DAS does not provide information related to substance‐related toxicities, and only reports substances classified under the Controlled Drugs and Substances Act, therefore, it is possible that not all non‐controlled substances (i.e. adulterants, cutting agents, emerging substances) are reported.
Second, the origin of substances (pharmaceutical vs. unregulated) submitted to DAS is unknown and qualitative physical descriptions reported by DAS may vary by laboratory. We excluded seizures described as residue, syringes, controlled‐release beads and derma patches (~8% of fentanyl‐containing seizures) as the substances physical description cannot be determined for residue and syringes, and controlled‐release beads and derma patches would likely be of pharmaceutical origin (potentially diverted slow‐release oral morphine and fentanyl patches). Third, cross‐contamination of different substances during collection and testing may affect the substance combinations in samples.
We limited our analysis to powders, tablets, crystalline and rock‐like substances to help minimize this. Our data also lacks substance concentration information. Fourth, we were unable to include data from January 2020 and did not apply a correction to our rate calculations, therefore, this annual data may be under‐reported.
Additionally, we could not compare trends directly to pre‐COVID‐19 periods. Finally, our findings may differ from other publicly available data from DAS, because those data may be presented and analysed using different methodologies, definitions or aggregation approaches. These differences could affect direct comparisons or reproducibility across datasets.
However, we anticipate such discrepancies to be minor.
CONCLUSIONS
Across Canada, there has been a significant rise in fentanyl‐containing drug seizures where multiple substances and potent fentanyl analogues are detected. By 2024, half of fentanyl‐containing seizures submitted to DAS contained at least one additional drug class—the majority of which were benzodiazepines, xylazine and stimulants—highlighting both the growing complexity and volatility of the unregulated drug supply and the value of seizure data in ongoing drug surveillance. These evolving trends highlight the urgent need for regionally tailored harm reduction and treatment responses including improved access to drug checking services, safer supply programs, supervised consumption sites and low‐barrier OAT, especially in areas experiencing an increasingly unpredictable drug supply.
AUTHOR CONTRIBUTIONS
Shaleesa Ledlie: Conceptualization; methodology; formal analysis; writing—original draft; writing—review and editing. Ria Garg: Conceptualization; methodology; formal analysis; writing—review and editing. Pamela Leece: Conceptualization; methodology; writing—review and editing.
Gillian Kolla: Conceptualization; methodology; writing—review and editing. Elaine Hyshka: Conceptualization; methodology; writing—review and editing. Mina Tadrous: Conceptualization; methodology; writing—review and editing.
Tara Gomes: Conceptualization; methodology; writing—review and editing; project administration; funding acquisition.
DECLARATION OF INTERESTS
T.G. reports funding from the Ontario Ministry of Health and Indigenous Services Canada unrelated to this work.
Supporting information