Accessing Take-Home Naloxone in British Columbia and the role of community pharmacies: Results from the analysis of administrative data

Introduction British Columbia’s (BC) Take-Home Naloxone (THN) program provides naloxone to bystanders for use in cases of suspected opioid overdose. This study seeks to provide trends and analysis from the provincial BC THN program since inception in 2012 to the end of 2018. Materials and methods BC THN shipment and distribution records from 2012–2018 were retrieved. Frequency distributions were used to describe characteristics of individuals accessing the program. To evaluate correlates of distribution after the addition of hundreds of pharmacy distribution sites, an analytic sample was limited to records from 2018, and multivariate logistic regression was used to evaluate correlates of collecting naloxone at a pharmacy site. Results Since program inception to the end of 2018, there were 398,167 naloxone kits shipped to distribution sites, 149,999 kits reported distributed, and 40,903 kits reported used to reverse an overdose in BC. There was a significant increasing trend in the number of naloxone kits used to reverse an overdose over time (p<0.01), and more than 90% of kits that were reported used were distributed to persons at risk of an overdose. Individuals not personally at risk of overdose had higher odds of collecting naloxone at a pharmacy site, compared to other community sites (including harm reduction supply distribution sites, peer led organizations, drop-in centers, and supportive housing sites) (Adjusted Odds Ratio (AOR): 2.69; 95% CI: 2.50–2.90). Conclusions This study documents thousands of opioid overdose reversals facilitated through the BC THN program. While those at highest risk of overdose may preferentially access naloxone through community sites, naloxone distribution through pharmacies has allowed the BC THN program to expand dramatically, increasing naloxone availability through longer opening hours on evenings and weekends. and in rural and remote regions. A diversity of naloxone distribution sites and strategies is crucial to prevent rising opioid overdose deaths.


Introduction
College of Pharmacists of BC made it available as a Schedule II drug (behind the counter) [12]. In September 2016, the College of Pharmacists of BC changed the status of naloxone for emergency use to 'unscheduled', making it available outside of pharmacies [14]. Furthermore, the BC THN program removed the requirement to report individual naloxone kit IDs and/or patient names in October 2016 [15].
In December 2017, the BC THN program officially expanded to select community pharmacies to improve BC THN access across the province. Pharmacies are available widely, including in remote and hard-to-reach communities, with longer hours (including weekends and evenings), and they may be more acceptable to clients less likely to frequent harm reduction supply distribution sites [16]. The program partnered with pharmacy wholesalers, banners and chains to onboard pharmacies through centralized distribution models initially, later reaching stand-alone pharmacies in remote areas with limited access to harm reduction supply distribution sites. By the end of 2018, there were 1,448 active take-home naloxone distribution sites across BC [17], including 562 community pharmacy sites [18]-representing 40.5% of all pharmacies [19].
Understanding correlates of take-home naloxone distribution and access is crucial for planning and targeting services aimed at curbing opioid overdoses. While analyses from the BCCDC shows that the BC THN program, combined with other harm reduction interventions, has averted thousands of opioid overdose deaths [20,21], there is still relatively little data in the published literature examining community reach and barriers to access of naloxone distribution programs in Canada. Furthermore, preliminary evidence from the BC Coroners Service suggests a spike in drug toxicity deaths corresponding with COVID-19 physical distancing measures and changes to the illicit drug supply; the Coroners Service reported a 130% increase in drug toxicity deaths in June 2020 compared to June 2019 [22]. As overdose deaths continue to rise and organizations critical in the training and distribution of naloxone face reduced hours and temporary closures due to COVID-19 [23], there is a need to understand barriers and access associated with different avenues of distribution.
The current paper seeks to first provide an overview of uptake of the BC THN program since inception in August 2012 to the end of 2018, as well as correlates of naloxone distribution through pharmacy sites after the BC THN's pharmacy expansion in late 2017. Findings may help inform optimization of take-home naloxone programs across jurisdictions.

Ethics
This study analysed anonymous administrative data. Ethics approval was obtained from the University of British Columbia Behavioural Ethics Board (H12-02557).

Data sources
The BC THN program draws on two key sources of data to monitor and evaluate naloxone distribution and use across BC. Firstly, 'supply order forms' were used internally at the BCCDC to track the number of kits shipped, to participating distribution sites [9].
Second, 'distribution records' were used by naloxone distribution sites to track characteristics of individuals who receive a naloxone kit [10]. No personal identifiers were collected, and sites were asked to return completed distribution records to the BCCDC on a monthly basis. Since program inception, distribution records have included site-specific information (site's unique ID, city, and the date the kit was distributed to an individual) completed by on-site staff, as well as a 'site type' identifier (corrections facilities, pharmacy, post-secondary, and other (housing sites, treatment centres, non-governmental and peer-led organizations).
Individuals collecting a kit were asked to self-report their gender, their age range, and whether they are collecting a first or replacement kit (due to the previous kit being used, lost, stolen, expired or confiscated).
Following the unscheduling of naloxone and THN program expansion in 2017 to individuals not personally at risk of an overdose, distribution record forms were amended to include self-reported risk of overdose i.e. personally at risk of overdose or at risk of witnessing an overdose (i.e. friends and family of someone at risk of overdose). Before this date, all kit recipients were assumed to be personally at risk of overdose.
Distribution records often experience a delay in return and entry due to two reasons: 1) a delay and general decrease in submission of records from distribution sites and, 2) a lag in manual entry of records at the BCCDC. Together, we estimate the return window to be on average between three to six months.

Study variables
Supply order forms were geographically coded by health authority (Vancouver Coastal Health, Fraser Health, Island Health, Interior Health, and Northern Health) and used to provide the number of naloxone kits shipped to distribution sites over time.
Distribution records were the primary data source used in this study. From these records, site-specific geographical information was used to classify observations by Health Authority (Vancouver Coastal Health, Fraser Health, Island Health, Interior Health, and Northern Health). Records were also used to classify records by age group (<19, 19-30, 31-60, >60, Unknown), gender (Male, Female, Other (Trans, gender non-conforming), Unknown), reason for collecting a kit (1 st kit, replacement kit-used, replacement kit-other (previous kit lost, stolen, confiscated or expired), site type (corrections facility, pharmacy, post-secondary institution, other (housing sites, treatment centres, non-governmental and peer-led organization), and 'eligibility for collecting a naloxone kit (not personally at risk of overdose, peronlly at risk of overdose, unknown).

Analysis plan
This study includes findings from distribution records and supply order forms across BC from program inception on August 31 st , 2012 to December 31 st , 2018. To evaluate the effect of the THN program's expansion into community pharmacies, a separate analytic sample was created which limits distribution records to those returned between January 1 st , 2018 and December 31 st , 2018. Data for this study were retrieved and analyses conducted in January 2020. All analyses were conducted using R version 3.5.3 [24].
Frequency distributions were used to describe characteristics of individuals accessing the BC THN program from 2012 to the end of 2018, as well as to compare shipment and distribution data. A chi-squared test for trend (Cochran-Armitage test) was used to evaluate differences in the proportion of naloxone kits used to respond to an opioid overdose over time. Frequency distributions were subsequently used to examine the separate analytic sample, limited to 2018 distribution records. In this sample, all non-pharmacy sites were grouped as 'other' sites to compare with pharmacy distribution. Multivariate logistic regression was used to examine factors associated with distribution site type.
The model building strategy described by Hosmer and Lemeshow [25], and further operationalized by Zhang [26], was employed. All available covariates were included for bivariable analysis to explore trends and examine distributions by the outcome variable. Covariates were cross-tabulated with the outcome variable and those with a p-value < 0.25 and were deemed to be conceptually relevant as indicated in the literature, were included for further multivariate analysis. As per the purposeful selection model building strategy, variables which did not meet the p-value cut-off but were known to be conceptually relevant were included in further analyses. Adjusted odds ratios (AOR), 95% confidence intervals (CI), and p-values are reported. Pvalues less than 0.05 were considered statistically significant.  Table 1 provide a summary of naloxone kits shipped, reported distributed, and reported used to reverse an overdose from the BC THN program, from August 31 st , 2012 to December 31 st , 2018. During this time period, a total of 398,167 kits were shipped and 149,999 naloxone kits were reported distributed, according to records returned. Data show a consistent increase in kits shipped, however the number of kits reported distributed increased through 2017 but dropped in 2018 (Fig 1).

PLOS ONE
A total of 41,197 kits (27.5% of those distributed) were reported to have been obtained to replace a kit used to respond to an opioid overdose. As a proportion of naloxone kits reported distributed, naloxone kits reported used to reverse an overdose increased steadily over time, from 8.7% in the pre-2015 period (168/1921), to 12 Table 2 presents summary characteristics of BC THN distribution records, from program inception to the end of 2018. For all variables, missing information was categorized as 'unknown'. Overall, 24,890 kits (16.6%) were distributed to replace a lost, stolen, confiscated or expired kit. The majority of THN distribution records were received in 2017 (42.2%) and 2018 (40.1%). The large majority (94.2%) of THN kits were distributed at 'Other' sites (including housing sites, health and treatment centres, non-governmental and peer-led organizations). The overall kit distribution rate 2012-2018 was 31.4 kits per 1000 BC population but varied between health authorities (based on average population estimates 2012-2018). The rate of kit distribution varied between health regions, from 22.4 to 50.4 kits/1,000 population, while the proportions distributed were similar between the four most populous health regions (20.8% -26.7%), with 5.2% of kits distributed in the least populous region-see the table in S1 Appendix for details.
Overall, most THN kits were distributed to individuals who identified as male (43.4%) compared to female (36.2%), other genders (1.2%), and unknown (19.3%), and to those between the ages of 19-30 (27.9%) or 31-60 (45.4%). Eligibility criteria was introduced in 2017, when the THN program was expanded to individuals not personally at risk of overdose. All individuals who received a kit before 2017 were considered personally at risk of overdose, and account for the majority of the 'unknown' category. For the age and gender variables, a high proportion of 'unknown' is likely due to the nature of ensuring a low-barrier THN program-some information is not always reported by THN distribution sites or by recipients. Table 3 presents the analytic sample characteristics after limiting naloxone distribution records to 2018, stratified by eligibility. By the end of 2018, there were 562 community pharmacy distribution sites. A breakdown of THN distributing pharmacy sites by Health Authority is provided in the table of S2 Appendix. A total of 57,524 records were included, of which 39,208 (68.2%) self-reported being personally at risk of overdose, compared to 18,316 (31.8%) of individuals who reported not being personally at risk of overdose. Of naloxone kits reported used to respond to an opioid overdose in 2018, 90.2% were kits obtained by individuals who were personally at risk of overdose. Even after the introduction of pharmacy sites across BC, a large majority (93.2%) of individuals continued to access naloxone at 'other' community site types, compared to pharmacy sites (6.8%). Furthermore, the majority of those who self-reported not being personally at risk of overdose accessed naloxone at a pharmacy (60.1%), while the majority of individuals personally at risk of overdose accessed naloxone at other sites (70.2%). Table 4 presents the analytic sample characteristics after limiting naloxone distribution records to 2018, as well as multivariate logistic regression results (adjusted odds ratios and 95% confidence intervals) for correlates of THN kit collection at a pharmacy site, compared to other sites. In bivariate analyses, all covariates were statistically significant and conceptually relevant, and were included in the analysis.

Correlates of collecting a THN kit through a pharmacy site
After adjustment, and consistent with descriptive statistics, individuals who were not personally at risk of overdose had more than twice the odds of collecting a THN kit at a pharmacy site, compared to other sites (AOR = 2.69 [95% CI 2.50-2.90]). Individuals collecting a replacement kit were less likely to collect those kits at a pharmacy site compared to first time kit

Discussion
The current study provides an overview of uptake of the BC THN program. While all Canadian provinces and territories have publicly funded THN programs, research and evaluation of these programs in the scholarly literature is limited. A recent study from Ontario describes uptake of the province's pharmacy-dispensed naloxone kits in Ontario [27], and preliminary analyses of provincial THN programs in Ontario [28], Alberta [29], Manitoba [30], and British Columbia [15] have been published. Data from other jurisdictions are not directly comparable, due to differences in program structure, eligibility criteria, and scope [9]. Nevertheless, analyses across provincial THN programs are valuable for examining program uptake across jurisdictions. An environmental scan of provincial programs and distribution has also been conducted [9], and our group is involved in efforts to combine aggregate jurisdictional level data nationally. The current study finds that the BC THN program expanded to a total of approximately 1,500 sites across BC by the end of December 2018, including 562 pharmacy sites (40.5% of all community pharmacies in BC) since December 2017. Overall, almost 400,000 naloxone kits were shipped and 150,000 reported distributed since the program was launched in 2012.
Our findings show that there is a significant increasing trend in the proportion of kits reported used to reverse an overdose over time. Most individuals collecting a naloxone kit selfreported being male and between the ages of 31-60. This corresponds with data from the BC

PLOS ONE
Coroners Service, which has found that the group at highest risk of overdose death in BC are men between the ages of 40-59 years old, as well as those 19-39 years [31]. Importantly, our findings also indicate that the vast majority of kits reported used to reverse an overdose were distributed to persons themselves at risk of an overdose. While it is possible that individuals collect naloxone kits for others to use on them in the event they overdose, most individuals who use drugs do so around other people who use drugs. We believe our findings suggest then that people who use drugs are most likely to respond to overdoses in their communities, though further research to help elucidate this question is warranted. After evaluating data from 2018, our analysis also finds that individuals at risk of opioid overdose were significantly less likely to collect a naloxone kit at a pharmacy site compared to community-based sites. In developing the BC THN program, the priority was to ensure a low barrier environment for individuals receiving overdose response training, collecting naloxone kits, and seeking other supports. Consistent with our findings, evidence from the literature suggests that people with lived and/or living experience of substance use may be reluctant to access pharmacies and other medical services due to past experiences of stigma or discrimination [32,33]. Nevertheless, data from this study suggests that first time naloxone kit recipients may be more likely to collect kits from pharmacy sites, suggesting that these sites are important in reaching different cohort of clients who may not access traditional distribution sites aimed at higher-risk communities.
The authors believe that these findings offer support for increased access to naloxone through pharmacy sites in BC, and highlights the importance of having a variety of distribution site types. A previous evaluation from our group suggests pharmacists are willing to be involved in the BC THN initiative, despite some initial concerns about lack of remuneration for training and dispensing. There are mixed views regarding further program growth; some pharmacy managers feel that naloxone is now fairly accessible and that there is no need for further expansion, while others feel that expansion to more pharmacy banners and chains would be beneficial [16].
Due to the COVID-19 crisis, some harm reduction services and supply distribution sites have temporarily closed or have reduced hours [23], while pharmacies remain an essential service that continue to be widely available, open, and hence accessible to distribute THN. Online training is available for those who need refresher training or to be newly trained, allowing for physical distancing during the training process. The THN kits contain non-latex gloves which responders are advised to put on before responding to an overdose and to remove carefully after responding, as well as to practice hand hygiene.
With recent changes of temporary exemptions [34] and clinical opioid prescribing guidelines [35] in relation to COVID-19, individuals may be receiving new opioid agonist therapy (OAT) prescriptions, longer prescriptions, and/or delivery of medications by pharmacies. In many cases, pharmacists have been advised to co-dispense naloxone. Future research is needed to explore changes in distribution and the role of pharmacies in providing naloxone to individuals at risk of overdose during the dual public health emergency period.
While data suggests a decline in naloxone kits reported distributed compared to kits shipped in 2018, it is important to note that distribution numbers may not accurately reflect naloxone kits in public circulation. Uneven response rates across sites, delays in returning distribution forms, and delays in entering data means that the true count of naloxone kits distributed to the public likely exceeds the reported kits distributed. The authors note that the observed drop in reported kit distribution in 2018 is likely due to a lower return rate, and does not represent a true decrease in naloxone kits distributed to the public. Relatedly, the higher proportion of pharmacy naloxone kit distribution in the Vancouver Coastal Region in the 2018 analytic sample is likely due to low reporting in high volume, low-capacity community sites in the Vancouver region compared to other regions. Operationally, the program has confirmed with high volume sites, which have low reported distribution forms, that kits ordered are being distributed and not held in inventory or being thrown away due to naloxone expiry. The large majority of sites reorder kits as needed on a weekly or monthly basis. Thus, we believe that true distribution is likely much higher than the number of distribution records returned to the BCCDC.
As previously mentioned, careful interpretation of this data is required; the program is low barrier and to ensure anonymity identifiers are not collected from naloxone kit recipients. Therefore, we cannot identify how many unique individuals are accessing or using naloxone kits, or how many kits are distributed and used per individual in BC. Self-reported data is also susceptible to response bias. Stigma around drug use may discourage high-risk individuals from providing information or self-identifying as being at risk of experiencing an overdose, which may in turn bias results.

Conclusion
Between 2012 and 2018, more than 40,000 take-home naloxone kits have been reported used to reverse an opioid overdose in BC. Community distribution sites are crucial for providing naloxone to people at high risk of overdose, who may be most likely to respond to overdoses in their communities. Pharmacy sites appear to reach a cohort of first-time kit recipients, sustain longer opening and weekend hours, and support access in rural and hard-to-reach areas. Findings from this study highlight the importance of making naloxone available through a variety of distribution sites that are both widely accessible and responsive to the needs of people who require naloxone.