TAKE-HOME DRUG CHECKING AS A NOVEL HARM REDUCTION STRATEGY IN BRITISH COLUMBIA, CANADA

Abstract

Background

Drug checking is a harm reduction strategy used to identify components of illicitly obtained drugs, including adulterants, to prevent overdose. This study evaluated the distribution of take-home fentanyl test strips to people who use drugs (PWUD) in British Columbia, Canada. The primary aim was to assess if the detection of fentanyl in opioid samples was concordant between a take-home model and testing by trained drug checking staff.

Methods

Take-home fentanyl test strips were distributed at ten sites providing drug checking services from April to July 2019. The fentanyl positivity of the aggregate take-home and on-site drug checking groups were compared by class of substance tested. An administered survey assessed acceptability and behaviour change.

Results

1680 take-home results were obtained from 218 unique participants; 68% of samples (n=1142) were identified as opioids and 23% (n=382) were stimulant samples. During this period, 852 samples were tested using on-site drug checking. The fentanyl positivity of opioid samples was 90.0% for take-home samples and 89.1% for on-site samples (Difference 0.8% (95% CI −2.3% to 3.9%)). These results were not affected by previous experience with test strips. Fentanyl positivity of stimulants in the take-home group was higher than on-site (24.7% vs. 3.2%), but the study was underpowered to conduct statistical analysis on this sub-group. When fentanyl was detected, 27% of individuals reported behaviour change that was considered safer/positive. Greater than 95% of participants stated they would use fentanyl test strips again.

Conclusions

Take-home fentanyl test strips used by PWUD on opioid samples can provide similar results to formal drug checking services and are a viable addition to existing overdose prevention strategies. Use of this strategy for detection of fentanyl in stimulant samples requires further evaluation. This intervention was well accepted and in some participants was associated with positive behaviour change.

BACKGROUND

The opioid overdose epidemic in North America has claimed thousands of lives in a short time span and remains an ongoing concern. Between May 2020 and April 2021, there were 75,673 deaths attributed to opioid overdose in the United States while between January 2016 and September 2020, there were 19,355 deaths in Canada (Centers for Disease Control and Prevention, 2021a, 2021b; Special Advisory Committee on the Epidemic of Opioid Overdoses, 2021). A contributing factor to mortality related to overdose is a rapidly changing drug supply with the increased presence of fentanyl and other potent synthetic opioids (Ciccarone, 2017; National Institute on Drug Abuse, August 2018; Special Advisory Committee on the Epidemic of Opioid Overdoses, 2020). British Columbia (BC), the westernmost province of Canada, has seen the highest rates of deaths due to opioid overdose in the country (Special Advisory Committee on the Epidemic of Opioid Overdoses, 2020). In response, there has been growth of harm reduction services including expanded naloxone distribution and an increase in the number and capacity of supervised consumption services (BC Centre for Disease Control, 2021; Strike & Watson, 2019). In BC, two forms of supervised consumption services exist: supervised consumption sites (SCS) staffed by nurses that exist under a federal exemption of the Controlled Drugs and Substances Act and overdose prevention services (OPS) that provide similar services but exist under a temporary exemption granted by the declaration of a public health emergency in 2016 and are supported by community organizations (Wallace et al., 2019). Despite rebound of attendance at harm reduction services following an initial decline when COVID-19 emerged, deaths continue to increase (BC Centre for Disease Control, 2021; Kuehn, 2021).

In addition to the above services, there has been a focus on the implementation of drug checking to detect the presence of fentanyl, a potent synthetic opioid (Tupper et al., 2018). Drug checking as a harm reduction tool aims to identify components of illicitly obtained drugs, including potential adulterants. It encompasses various technologies, from those that require trained technicians such as gas chromatography-mass spectrometry (GCMS) and Fourier-transform infrared spectroscopy (FTIR) to simpler methods such as substance-specific immunoassay test strips (Kerr & Tupper, 2017). The existing body of drug-checking literature has demonstrated benefits from a surveillance perspective, specifically to allow for identification of changing components of the unregulated drug supply and specific novel contaminants associated with harm (Kerr & Tupper, 2017; Ontario Agency for Health Protection and Promotion (Public Health Ontario), 2017). The use of drug checking services also has benefits on an individual level, with an increase in positive behavioural changes when unexpected substances are identified, which can reduce mortality from inadvertent exposure to adulterants (Giulini et al., 2022; Maghsoudi et al., 2021). Detection of fentanyl specifically can lead to reduction in overdose through disposal of a substance, use of a smaller amount, or use in a supervised setting (Goldman et al., 2019). Benefits may extend beyond users of a drug checking service to others through communication of test results (Measham & Turnbull, 2021). Studies examining these effects are often situated in settings of “party-drug” use (MDMA at music festivals, amphetamines and GHB in “party and play”) with individuals using drugs recreationally (Hungerbuehler et al., 2011; Maghsoudi et al., 2021; Mema et al., 2018). Evaluation of these models of drug checking service delivery suggests a potential lack of generalizability due to self-selection bias (Palamar et al., 2021). The effects on behaviour change and health outcomes are less established for those who are using substances such as heroin or crystal methamphetamine, owing partially to inconsistent methods of service delivery as well as barriers to utilization (Bardwell et al., 2019; Maghsoudi et al., 2021). Limitations may also stem from physical dependence to substances amongst sub-populations of people who use drugs (PWUD), which may affect the decision-making process for an individual who is trying to address symptoms of withdrawal rather than engaging in purely recreational use of a substance (Bardwell et al., 2019). More recent drug checking literature involving individuals using fentanyl has demonstrated a high willingness to use immunoassay test strips and the feasibility of integrating them into existing harm reduction services (Karamouzian et al., 2018; Krieger et al., 2018a,b; Sherman et al., 2019; Tupper et al., 2018).

The availability of fentanyl test strips for commercial purchase in BC has called into question whether they should be offered as a harm reduction tool for personal, off-label use rather than needing to locate a site offering testing. A large proportion of the opioid overdose deaths in the province have occurred in private residences when individuals were using drugs alone (British Columbia Coroners Service, 2019). In this setting, naloxone may not be effective given the inability to self-administer this medication. However, knowledge of the presence of fentanyl may lead to safer drug use. In controlled comparisons, fentanyl test strips have a high degree of sensitivity and specificity compared to FTIR, but it is unknown whether this extends to “real-world” settings (Ti et al., 2020). Questions remain about whether enough is known about the effectiveness and acceptability of drug checking to promote its use outside of SCS. Specific concerns include whether the accuracy of testing is retained outside of controlled settings by trained individuals, whether these interventions are acceptable to PWUD, and whether their implementation can occur in a feasible manner that can adapt to evolving contaminants (McGowan et al., 2018). To address these important concerns, a pilot program of distributed fentanyl test strips was conducted, referred to as “take-home drug checking” (Interior Health, 2019). The primary aim of the study was to determine if the fentanyl positivity rate of samples tested by PWUD using take-home drug checking was concordant with the fentanyl positivity of samples tested using on-site drug checking, utilizing aggregate data collected over a four-month period. Secondary outcomes were to assess behaviour change following use of take-home drug checking and gauge the acceptability of the intervention to PWUD. We hypothesized that fentanyl positivity of opioid samples tested using take-home drug checking would be equivalent to fentanyl positivity of opioids tested on-site using drug checking services provided by trained peer support workers and technicians.

METHODS

Study design

This is a cross-sectional observational study that utilized data collected from a pilot program involving the distribution of fentanyl test strips for personal, “take-home” use. The pilot was conducted from April to July 2019 at ten sites that serviced a large geographical area including the Downtown Eastside neighbourhood of Vancouver, the epicentre of overdose in the province, as well as several smaller urban and rural communities. Sites were selected if they were already providing on-site drug checking using fentanyl test strips to enable comparison of aggregate data. Including only sites providing both take-home and on-site drug checking during the same period was done to minimize the potential for differing drug supplies between the groups. Prior to this pilot program, test strips were only available on-site and were not distributed for take-home use.

Single-use urine fentanyl test strips purchased from BTNX Inc. were utilized, having already been employed for on-site drug checking (Tupper et al., 2018). In the drug checking context, these are used to test a small portion of a substance diluted in water rather than the original intended use on urine samples. This method of using fentanyl test strips is off-label, and thus instructions for use were created and provided by study staff, rather than the manufacturer. While a novel utilization, the use of test strips in this way has been previously described (Krieger et al., 2018b; Tupper et al., 2018). Their detection limit for fentanyl is 130ng/ml and they are able to detect various fentanyl analogues (McCrae et al., 2020; Sherman & Green, 2018). Recent data suggests the sensitivity of these immunoassay strips for detecting fentanyl is 87.5%, while the specificity is 95.2% (Ti et al., 2020). While fentanyl test strips can detect the presence or absence of the substance, they are not able to quantify the amount present in a sample.

Recruitment and training of participants

Individuals were approached to participate in the pilot program while using services at one of the included harm reduction sites. A peer support worker employed by these sites assessed their interest in being involved in a pilot project. If interested, they were provided a testing kit and training consisting of a review of the contents of the take-home drug checking kit as well as a brief explanation about how to conduct a test and interpret the result. The potential for false negatives and the inability of the strips to detect all potential adulterants was emphasized. Once the training was completed, the participant’s confidence was assessed, and they were asked to confirm how to interpret the result. During this time, peers were able to provide other harm reduction and overdose prevention messages including encouraging clients not to use alone, to use at an SCS, and to use small, “test”, doses. Naloxone kits and training were also available for distribution. Participants could contribute results to both the on-site drug checking and take-home drug checking groups as they were not limited from accessing existing drug checking services.

Survey tools and testing kits

Each of the kits contained five fentanyl test strips, a set of instructions detailing dilution and preparation of a sample, a result interpretation card, a pencil, a small bottle of water, and small cups. A results sheet was also included in the kits to record the outcome of each test strip (positive, negative, unclear) and the expected substance tested. The results sheet specified “down” (a slang term for unregulated opioids) as an option to refer to heroin/fentanyl to differentiate these opioids from prescription opioid pills. If the participants returned the sheet without the results completed, they were prompted to recall the results of the tests, and these were recorded.

At the time of returning the results, a secondary survey was administered that collected self-reported demographic information, details of test strip use including location and method, and questions regarding behaviour change and acceptability of the intervention. Each participant was asked about actions undertaken in response to both a positive and negative fentanyl test result. The list of presented options included no change in action taken, using a different quantity of substance, using alone or with others, and using in a different setting. For most questions, participants were allowed to provide multiple answers to reflect the potential for varied use of the five strips that were included in each kit. The questions in the secondary survey were not based on the results of the returned test strips but were instead standardized questions asked of each participant about their response to potential testing results. Participants received a $5 CAD honorarium or $5 CAD gift card for completing both the results sheet and the secondary survey. Demographic information collected included age, city of residence, gender identity, and self-identification as First Nations. Participants were able to return results on an ongoing basis and in keeping with a pragmatic study design, no specific time for follow-up was arranged. As no demographic or contact information was collected when the testing kits were distributed, there was no follow-up for kits that were distributed but not returned.

Statistical analysis

A cross-sectional observational study design was used to compare take-home and on-site drug checking results. A pre-defined limit of equivalence was within (+/−) five percent, specifically that the fentanyl positivity rate for take-home drug checking is within five percentage points of the fentanyl positivity rate obtained through on-site drug checking. A web-based power calculator was used to determine the required sample size (available at https://www.sealedenvelope.com/power/binary-equivalence). It was determined that 617 samples would be required in each group for a total sample size of 1234. The parameters used for this calculation were an alpha of five percent, a power of 80%, and an equivalence limit of five percent. Drug checking data in Vancouver, which includes both immunoassay test strips and technologies such as GCMS and FTIR, between 2017 and 2018 demonstrated a fentanyl positivity rate of 90% amongst samples identified as opioids (Vancouver Coastal Health, 2018). Based on this, the expected fentanyl positivity in both the take-home and on-site drug checking groups was set at 90%. Surveillance data in Vancouver suggests 3% of stimulant samples are contaminated with fentanyl (Vancouver Coastal Health, 2018). The pilot program was operated for four months to test enough opioid samples. This timeframe did not allow for the collection of sufficient stimulant samples. Take-home drug checking results were self-reported, and due to the study design there were no confirmatory analyses.

Data were entered into an electronic REDCap survey tool and then exported to R version 3.6.2 for descriptive and statistical analysis (R Core Team, 2020). For the primary outcome, a two-one-sided t-test procedure was used to assess for equivalence between take-home drug checking and on-site drug checking. Chi-squared test of independence was used to assess for potential confounding variables.

RESULTS

During the study period, 336 result sheets were returned to the ten participating sites. As each result sheet included five drug checking results, a total of 1680 results were received. Substances were classified as opioids, crystal methamphetamine, cocaine, other, or unknown by the owner. To be included as an interpretable result, both the substance tested, and the test strip result needed to be recorded. 1506/1680 (89.6%) strips had both questions answered, the remaining 174 either did not indicate the substance tested (111), or had not provided a response for the test result (63). A large proportion of this missing data came from the substance tested category, where 111 responses (6.6%) were “did not answer”. Only a small proportion (3.3%) of the total number of strips tested yielded an unclear or invalid response for presence of fentanyl.

Participants were able to participate in the pilot multiple times throughout the study period. Table 1 includes the demographics of study participants who reported completing a results sheet for the first time (n=218). Of these, 134 (61.5%) identified as male, 66 (30.3%) identified as First Nations, with a mean age of 36 years (SD 10.6).

Table 1.

Demographics of unique take-home drug checking study participants (n=218).

Characteristic	N	Mean (Range)	%	SD
Gender
Male	134		61.5%
Female	84		38.5%
Two-Spirit	2*		0.9%
First Nations
Yes	66		30.3%
No	138		63.3%
Prefer Not to Answer	4		1.8%
No Response	10		4.6%
Age	214	36 (19–70)		10.6

^*Both participants also identified as male

Most samples tested using take-home drug checking were classified as opioids (68.0%) with stimulants including crystal methamphetamine and cocaine making up 16.6% and 6.1%, respectively. A minority (2.8%) of tested samples were classified as “other”, and these were predominantly a mix of stimulants and opioids, substances such as MDMA, or contained no substance other than water.

During the study period, a total of 852 samples that were believed to be either an opioid (n=663) or a stimulant (n=189) in isolation were tested using on-site drug checking. Demographic information was not collected during on-site drug checking as the existing intervention was meant to be anonymous and confidential. For each of these substance categories, the proportion of fentanyl positive samples was compared between those tested using take-home drug checking and on-site drug checking (fentanyl positivity). Sample results that were unclear/invalid were excluded. The fentanyl positivity of opioid samples was 89.9% for take-home drug checking and 89.1% for on-site drug checking. Based on the pre-specified equivalence limit, these results were equivalent, with a difference of 0.81% between testing groups (95% CI −2.3% to 3.9%, p=0.00275). Full results of testing by substance category are presented in Table 2. Fentanyl positivity of opioids tested using take-home drug checking was compared across key demographic characteristics to assess for potential confounding factors (Table 3). Gender, First Nations ancestry, and prior experience with fentanyl test strips were not associated with a difference in fentanyl positivity. Furthermore, there was no difference between those who recalled test strip results at the time of returning their results sheet (n=122) and those who returned completed sheets (n=206) (Table 3).

Table 2.

Results of take-home drug checking and on-site drug checking by substance category, April to July 2019.

		Take-Home Drug Checking (n=1680)		On-Site Drug Checking (n=852)		Difference
Substance Tested	Fentanyl Test Strip Result	N	%	n	%	%	95% CI	P-Value
Opioids	Positive	994	90.0%	591	89.1%	0.8%	−2.3 to 3.9	0.59
	Negative	111	10.0%	72	10.9%
	Total	1105		663
Crystal Methamphetamine	Positive	73	27.6%	5	5.2%	22.4%	14.8 to 30.1	<0.0001^
	Negative	192	72.5%	92	94.9%
	Total	265		97
Cocaine	Positive	16	17.2%	1	1.1%	16.1%	7.1 to 25.1	0.0048^
	Negative	77	82.8%	91	98.9%
	Total	93		92

^{^}The study was not sufficiently powered to evaluate the difference between groups for samples of stimulants (including crystal methamphetamine and cocaine)

Table 3.

Results of take-home drug checking of opioid samples across demographic variables and testing patterns.

	Fentanyl Positive (%)	Fentanyl Negative (%)	Total	P-Value
Gender				0.66
Male (n=202)	659 (89.7%)	76 (10.3%)	735
Female (n=128)	334 (90.5%)	35 (9.5%)	369
Completed/Recalled				0.36
Completed (n=122)	352 (91.4%)	33 (8.6%)	385
Recall (n=206)	618 (89.7%)	71 (10.3%)	689
First Nations				0.68
FN (n=111)	341 (90.0%)	38 (10.0%)	379
Not FN (n=208)	608 (90.7%)	62 (9.3%)	670
Experienced with Test Strips				0.24
No, First Time (n=173)	508 (90.7%)	52 (9.3%)	560
Yes, Experienced (n=153)	116 (78.4%)	32 (21.6%)	148

Results from samples expected to be stimulants were divergent between testing groups. Crystal methamphetamine samples tested using take-home drug checking were reported as fentanyl positive more often than on-site samples (27.6% vs. 5.2%). The same pattern was seen for cocaine samples tested using take-home drug checking (17.2% vs. 1.1%). However, the study was underpowered to evaluate equivalence between these testing groups. A small portion of the test strips (3.8%) yielded an unclear or illegible response. It is unclear what the participants did in these cases, but the inclusion of multiple test strips would have allowed for repeat testing.

Notably, when the results of take-home drug checking were stratified based on previous experience with fentanyl test strips, there was a trend towards a smaller difference between the results of take-home drug checking and on-site drug checking. For opioids, when only results from those who were using fentanyl test strips for the first time were included, there was a difference of 1.6% between take-home drug checking and on-site drug checking. This difference was reduced to 0.6% when only including samples from those who had self-reported prior experience with using fentanyl test strips. Similar results were seen for crystal methamphetamine (28.9% to 15.2%) and cocaine (28.3% to 7.8%).

Take-home strips were utilized in a variety of settings that reflected the numerous and varied locations of drug use. A total of 184 (54.8%) completed results sheets reported testing of drugs in settings identified as “home” including private residences, tents, and single-room occupancy hotels. Another 74 (22.0%) reported testing in alleys and streets. Notably, 97 (28.9%) indicated using test strips at OPS/SCS locations. Testing was predominantly done before drug use, with 261 (77.2%) reporting this when asked about timing of use. The majority, 305 (90.8%), reported that a sample of the drug itself was tested while another 40 (11.9%) reported testing residue such as a wash (i.e., residual substance left in a cooker, bag, or syringe). One individual tested their urine.

When asked about their response if a drug tested positive for fentanyl, 217 (64.6%) participants reported no behaviour change. When behaviours were grouped as positive/safer or negative/riskier, 86 participants (25.6%) reported at least one positive behaviour change. The most common response was to use less/use more slowly (n=45), use with someone else (n=26), use at an OPS/SCS (n=9), not use at all (n=7), or have someone check on them (n=4). Conversely, only six participants (1.8%) reported a negative/riskier behaviour. Of these, one individual reported using more/more quickly and five individuals reporting using alone after a fentanyl positive result.

Regarding acceptability of the intervention, 323/336 (96.1%) respondents indicated that the instructions included with the test strips were easy to follow, 319 (94.9%) felt confident in their ability to read the results and 322 (95.8%) indicated they would use the test strips again.

DISCUSSION

Our study demonstrates that for opioid samples, there was concordance of rates of fentanyl detection between samples tested using distributed fentanyl test strips and samples tested on-site by trained individuals. This suggests that PWUD can accurately use fentanyl test strips with minimal training in their chosen location. Specifically, results from take-home drug checking yielded similar results to on-site drug checking conducted by trained staff. Based on secondary survey results, the intervention was well accepted amongst PWUD, and the vast majority were interested in using fentanyl test strips in the future. Previous work has called for evaluation of the effectiveness and acceptability of the fentanyl test strips outside of a harm reduction environment, which was addressed by our study (McGowan et al., 2018). In a changing drug market, test strips can provide increased agency amongst PWUD (Weicker et al., 2020). While underpowered for this comparison, the results suggest there may be a difference in reliability between take-home drug checking and on-site drug checking when testing stimulants. Notably, stimulant samples tested on-site had a lower rate of fentanyl positivity compared to those tested in the take-home group. This evaluation existed within real world testing conditions, maintaining low barrier access to harm reduction services. The control group consisted of current drug checking environments, and the intervention was applied pragmatically to allow individuals to utilize the technology as they normally would. Restrictions on location or method of use were not applied.

Based on our findings, distributed fentanyl test strips would be reliable for the testing of samples identified as opioids and should be more widely distributed. There is growing evidence that fentanyl test strips may help prevent overdose when included with other evidence-based strategies (Peiper, 2019). Other informal techniques such as visual inspection of a substance have been applied by PWUD, but may not be effective in substances that contain traces of fentanyl (Peiper et al., 2019). Our study situated fentanyl test strips within sites that provide naloxone kits, drug use supplies such as syringes, supervised consumption of substances, and drug checking using both test strips and more sophisticated technologies. In contrast to the potential for behaviour change from drug checking results, analysis of several cohort studies within Vancouver during the period of increasing fentanyl contamination in late 2016 showed that a majority of PWUD did not change their drug use behaviours nor translate the knowledge of a changing drug supply to an increased risk of overdose (Brar et al., 2020; Moallef et al., 2019). These findings indicate the need for targeted education and harm reduction interventions for those at risk. Distribution of testing supplies provides an opportunity for further engagement. In BC, an expansion of this pilot program, including continuation at sites included in this evaluation, has occurred to distribute fentanyl test strips labelled with instructions for use. Notably, the described positive behaviour changes rely on an individual possessing knowledge around safer ways to use substances, including knowledge around using a small amount (“test dosing”) and using with others or not alone to avoid overdose and allow for naloxone administration. Furthermore, participants identified using at an OPS/SCS as a potential behaviour, which necessitates that these services exist. Our findings around behaviour change in response to a positive fentanyl result underscore the need for comprehensive harm reduction services and education.

An unresolved question is whether fentanyl test strips can be reliably utilized by PWUD for the testing of stimulants. The number of samples identified as crystal methamphetamine and cocaine in our study were significantly lower than opioids and our study was therefore underpowered to perform an analysis. However, the trend to divergence in fentanyl positivity for these substances between take-home and on-site samples was apparent. While this raises the possibility that the drug supplies between these two groups were differently contaminated with fentanyl, a more likely explanation is related to the method of testing. While fentanyl test strips are quite sensitive and specific for fentanyl, there have been reported concerns of false positives with stimulants when sufficient dilution is not used (Harm Reduction Coalition; Lockwood et al., 2021). Included in each of our kits were instructions to dilute the sample in 30mL of water and a small water bottle was provided. It is possible that these instructions were not followed. The trend towards equivalent results seen when participants were experienced with fentanyl test strips suggests that this is a plausible explanation. Further education regarding the risk of false positives with stimulants may be warranted. As well, differing instructions based on the expected substance being tested may be necessary to improve accuracy. Among people who use stimulants, there appears to be a desire to use fentanyl test strips (Reed et al., 2021). However, a consistent concern is the lack of alternatives or options available should a substance test positive for fentanyl (Bardwell et al., 2019; Long et al., 2020).

Based on the overwhelming interest in using test strips in the future, it appears that this technology is meeting a patient-centred desire for knowledge about the components of substances. This is in keeping with previous research evaluating the acceptability of this intervention to PWUD (Goldman et al., 2019; Sherman et al., 2019; Weicker et al., 2020). Further research will be required to determine whether the use of drug checking services can lead to clinically meaningful outcomes such as decreased overdose rate. The feasibility of this intervention was established as take-home drug checking was easily integrated into existing harm reduction services. Ensuring confidentiality and protection from stigmatization of drug use is paramount (Wallace et al., 2020). Because the fentanyl test strips and trained individuals were already present at each site, the primary step needed to initiate the pilot was training peer support workers on how to provide education about the use of the strips and their limitations.

In our setting, the presence of fentanyl in the illicit opioid supply is quite high; however, in other jurisdictions it remains less common. A recent analysis from Australia detected fentanyl in low rates but with evidence of unintentional exposure, which may pose a greater risk of overdose (Lam et al., 2022). In locations where fentanyl may be intermittently present, or where the risk of fentanyl entering the drug market exists, this intervention may have even greater value. In addition, novel contaminants including benzodiazepines have been shown to rapidly enter an unregulated drug supply (Laing et al., 2021). Immunoassay test strips for these other substances of concern should be introduced and evaluated quickly.

Given that the rapid spread of fentanyl in BC has been widely publicized, it was unclear how much behaviour change could be expected following its detection, particularly as it has become the predominant unregulated opioid in the province, replacing heroin (Vancouver Coastal Health, 2018). A systematic review of drug checking services suggests that disposal of a tested substance may be more frequent when the results are unexpected (Maghsoudi et al., 2021). Similarly, it is difficult to interpret what should be considered a meaningful level of change. Rates of disposing of substances that contain unexpected adulterants have been reported to be up to 66% within certain populations, significantly higher than our findings (Giulini et al., 2022). However, for a population that may experience physical dependence and withdrawal, or be structurally vulnerable, discarding may not be a realistic option. An important finding from an evaluation of on-site drug checking services in Vancouver was a stated intent to use a smaller amount initially (“test dose”) if fentanyl was detected, and a lower rate overdose and naloxone administration in this group was observed (Karamouzian et al., 2018). Despite these described factors that likely contributed to a lower rate of behaviour change, positive harm reduction strategies were still employed commonly. Due to the design of the study, with each results sheet being linked to five testing results, it was not possible to assess behaviour change across substance use patterns. As well, the survey questions concerning behaviour change were open to interpretation by the participant and they could have reported either a hypothetical change or a specific action they took following an actual result. Previous studies evaluating the effect of drug checking on behaviour change have identified an influence on behavioural intent but a lesser effect on enacted behaviour (Maghsoudi et al., 2021). As with other studies, the risk of selection bias remains as our sample was not randomly selected.

An additional limitation of our study is the lack of demographic information available for those utilizing on-site drug checking. This information is not routinely collected by the sites included in our study to protect the anonymity of clients utilizing harm reduction services. This prevents us from making comparisons between the populations utilizing take-home drug checking and on-site drug checking. An assumption of the study was that there was no systemic difference in the drug samples tested by the two groups. Ideal research conditions would be to have a direct control group in which the same sample was tested by an individual and by a drug-checking expert. However, this approach would have required increased research support that was not available and would have dampened the external validity. As each distributed kit contained five fentanyl test strips, it is possible that the same sample was tested repeatedly. This was not specifically discouraged in the included instructions. On the contrary, it was suggested a test be repeated if the result was unclear.

CONCLUSIONS

Our study examined the concordance, acceptability, and impact on behaviour change of take-home fentanyl test strips within a study design that was limited by self-reported data and a lack of confirmatory analysis. The results demonstrated value and reliability of this novel harm reduction approach for testing opioid samples for the presence of fentanyl, as well as the potential for false positives when testing stimulants. Distributed test strips were an acceptable and desired harm reduction tool that may be a valuable addition to existing efforts to prevent overdose and other harms of a toxic and variable drug supply. Positive or safer behaviour change in response to test results tended to incorporate other overdose prevention strategies such as using in a supervised setting or consuming a smaller quantity, which emphasizes the need for comprehensive harm reduction services and education. Future work should examine the cost-effectiveness of fentanyl test strip distribution as well as health outcomes associated with their use. Expansion of drug checking services should be considered to settings with lower rates of fentanyl contamination and should be responsive to other adulterants that may enter the unregulated drug supply.

HIGHLIGHTS.

• Fentanyl test strips are a form of drug checking used to prevent overdose.

• Distribution of fentanyl test strips for “take-home” use has not been evaluated.

• Take-home testing of opioid samples provided similar results to on-site testing.

• Take-home drug checking was associated with positive harm reduction behaviours.

Abbreviations:

BC

British Columbia

PWUD

people who use drugs

SCS

supervised consumption sites

OPS

overdose prevention services

FTIR

Fourier-transform infrared spectroscopy

GCMS

gas chromatography-mass spectrometry

Availability of data and materials

The data that support the finding of this study are available on request from the corresponding author.

Funding

Fentanyl test strips, kit materials, and data entry staff were provided by the BC Centre for Disease Control. Honoraria were provided by individual health authorities. SK is supported by an International Collaborative Addiction Medicine Research Fellowship (US National Institute on Drug Abuse, R25-DA037756), and the Research in Addiction Medicine Scholars Program (US National Institute on Drug Abuse, R25-DA033211).