Sweden’s first Take-Home Naloxone program: participant characteristics, dose endpoints and predictors for overdose reversals

The Stockholm NSP opened in April 2013 and consists of two clinics and one mobile unit. To date, more than 4,300 individuals have been registered in the program. The NSP offers sterile injection equipment such as needles, syringes and other paraphernalia (filters and cookers). In addition, services include THN and overdose prevention training, vaccination, counselling, wound care, treatment for hepatitis C (HCV) and HIV as well as referrals to other service providers for OAT and other substance use disorder treatments. In accordance with Swedish legislation, eligibility criteria for NSP enrolment include current injection drug use, being 18 years of age or older, and being able to prove your identity. Testing for hepatitis A, hepatitis B, HCV and HIV is offered on enrolment and continuously. The unique Swedish personal identity number is used for registration and those without such an ID are provided with a unique reserve number.

Study inclusion criteria were THN program enrolment in the Stockholm NSP and at least one additional NSP visit during the study period.

This was a prospective open inclusion cohort study conducted between January 24^th 2018 and March 31^st 2022. Clients in the Stockholm NSP were informed and recruited to the THN program through personal information from NSP staff, also promoted on an information screen in the waiting room. The study design relied on the existing NSP structure with passive follow-up when participants returned to the NSP for regular visits, there was no active following-up outside of the NSP. During the study period’s first four months, the only available naloxone on the Swedish market was a prefilled vial for intra muscular (i.m.) administration (Prenoxad, 0.4 mg/ml) which contained five doses of naloxone. From May 2018, i.m. naloxone was gradually replaced within a few months by an intranasal solution (Nyxoid, 1.8 mg/dose) which was given out in a kit containing two doses.

All THN participants individually completed a single training session, carried out by a NSP nurse or physician, before naloxone was handed out the first time. The sessions were brief (10–15 min) and conducted on the premises of the NSP clinics. A refresher training session was offered on-demand when returning for a refill of naloxone. The training protocol was based on the European Monitoring Centre for Drugs and Drug Addiction’s (EMCDDA) recommendations [31] and included information on how to identify an overdose, general overdose response, basic cardiopulmonary resuscitation (CPR) with focus on rescue breathing, and how to administer the naloxone (i.m. or nasal). Participants were informed of possible side effects, including the risk of withdrawal symptoms, the half-life of naloxone and the importance of calling for an ambulance in an overdose situation. At the end of the session, participants were given the opportunity to practice their skills on a CPR dummy and received a prefilled vial of naloxone for i.m. use, or two doses of nasal naloxone, free of charge. Participants were advised to return for a refill and debriefing if their current dose was used for an overdose reversal, lost, expired, given away or in any other way went missing. Participants did not receive any renumeration for participation in study.

THN participants were offered a pocket-sized brochure with key messages from the training. One year into the project, the National Board of Health and Welfare published standardised material concerning THN programs and overdose prevention, which was then distributed in the training sessions [30].

All study data was registered in the national quality register InfCare Needle Syringe Program (InfCare NSP) database, previously described in detail [32]. Six InfCare NSP questionnaires were used for this study: 1) the ‘NSP enrolment questionnaire’: basic socio-demographic data and drug history; 2) the ‘NSP standard visit questionnaire’: at every NSP visit, clients report which drug they last injected; 3) the ‘Three-to-six-month NSP follow up questionnaire’: updated information on employment and housing status; 4) the ‘Twelve-month NSP follow-up questionnaire’: updated information on employment, housing status and primary drug the past 12 months; 5) the ‘THN enrolment questionnaire’: primary drug and previous experiences of drug overdose; and 6) the ‘THN refill questionnaire’: information on what happened to the previous naloxone dose and, if naloxone was used to reverse an overdose, ten follow-up questions on the intervention.

THN questions were partly adapted from the Norwegian THN training curriculum [33].

Overdose reversals were reported by the person whose prescribed naloxone dose had been used in the event. Consequently, the person experiencing the overdose could have been the study participant themselves or somebody else.

The latest available information was presented for variables concerning housing, education level, income type and primary drug. Static baseline variables included: age at inclusion in THN program, gender, country of birth, and experience of personal or witnessed overdose at inclusion in THN program. Housing status was re-coded into three main categories: homeless (which included sleeping in tents, in garages, in night shelters, on friends’ sofas and so on); unstable housing (temporary housing like hostels, rehabilitation homes etc.) and stable housing (longer-term rental or own home).

For intramuscular naloxone (which was only distributed in the first few months of the program), although the vial technically contained five doses, it was treated as a single dose in reporting.

Data on mortality were automatically reported from the Swedish population registry to the Stockholm NSP medical chart and quality register.

Continuous variables were presented both as mean (standard deviation) and median (inter quartile range) values, and categorical variables as counts and percentages. Between-group differences were presented using risk ratios (RR) for categorical variables and mean and median differences for continuous variables. RR were estimated using a log-linear model and mean and median differences using linear and quantile regression respectively.

The association of baseline information and number of naloxone doses used at overdose was estimated in Incidence Rate Ratios (IRR) using a zero-inflated Poisson regression model, offsetting for the time participants spent in the study (time-in-study). The offset was applied due to the high variance of time-in-study which directly affected the probability of reporting overdose reversal. Univariable models as well as a multivariable model with all covariates were estimated. All variables from the unadjusted model were included in the adjusted model (gender, age, country of birth, housing, primary drug, and overdose experience). The logistic zero-inflation part of the model only used the number of naloxone doses received as an independent variable, while the independent variables in the Poisson part of the model were changed depending on the variable of interest.

The association between background information and the endpoint of individual naloxone doses (i.e. was it used, lost, stolen etc.) was estimated in Relative Risk Ratios (RRR) using a multinomial logistic regression model. The clustered robust standard error estimator was used to account for repeated measures since a person could report different endpoints multiple times due to receiving more than one refill.

The difference in mortality between the group of participants that had reported at least one overdose reversal compared to the group that had not, was estimated in Hazard Ratios (HR) using an illness-death model based on the Cox proportional hazards model.

Due to very low levels of missing data, we restricted the analysis to subjects with complete data on the variables involved. Confidence interval (CI) level was set at 95%. All reported p-values are two-sided and a p-value of < 0.05 was considered as statistically significant.

Analysis was done in Stata, version 15 (StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC).

During the study period, 3,151 individuals visited the Stockholm NSP and 1,438 of them enrolled in the THN program. After excluding 143 people who did not make a return visit to the NSP during the study period, the final analysis contained 1,295 individuals. Naloxone was used in 1,625 separate overdose events, reported by 570 individuals, during the four-year study period (Fig. 1). A total of 11,440 doses of naloxone were distributed (2,590 at enrolment and 8,850 at subsequent refills).

The mean age of participants was 38, with the majority being male and born in Sweden, in line with the general population in the Stockholm NSP [32]. The majority stated opioids (mostly heroin) as their primary drug and had prior experience of personal as well as witnessed overdoses (Table 1).

By the end of the four-year study period, the majority of participants (74.3%) had returned to the NSP for at least one refill of naloxone and 44.0% had reported that their naloxone dose had been used in at least one overdose situation. Participants receiving a refill were more likely to: be younger, be homeless or in unstable housing, use benzodiazepines as their primary drug, have previous experience of personal overdose, or have a longer time-in-study. Reporting at least one overdose reversal with naloxone during the study period was associated with previous experience of personal or witnessed overdose, being male, and longer time-in-study. The comparison between participants receiving a refill or not, as well as participants reporting reversal or not, is depicted in Table 1.

Overall, 8.6% of the participants reporting reversals died during the study period compared to 4.1% within the group of participants not reporting reversals. The Cox regression analysis showed that the mortality was significantly higher among those who reported a reversal (HR 3.4; CI 2.2, 5.2).

We used a zero-inflated Poisson regression to evaluate predictors of number of reported overdose reversals among participants who received at least one refill. In the adjusted model, we noted a greater average number of reported reversals by homeless participants (aIRR 1.35; CI 1.06, 1.73) compared to those with other housing situations. Those with benzodiazepines (aIRR 1.75; CI 1.1, 2.78) or stimulants (aIRR 1.26; CI 1.01, 1.58) as their primary drug reported a greater average number of reversals than those who primarily used other drugs. Additionally, participants who were born outside of Europe (aIRR 1.37; CI 1.06, 1.76) also reported a higher average number of reversals than those born in Europe (including Sweden). The result of the unadjusted and adjusted model is summarized in Table 2.

In the case of overdose reversals, the majority of refills (63.2%) were requested within four weeks of the incident (Table 3). In most cases (67%), overdose reversals were reported to have been carried out on an acquaintance. Just over half (51.8%) of the overdoses took place in a private space (normally the participant’s or somebody else’s home) while 45.3% occurred in a public space (most frequently “outdoors”). Homeless participants had more than twice the risk of reporting that the reversal took place in a public space than participants who were not homeless (RR 2.42; CI 1.60, 3.67).

Apart from administering naloxone, overdoses were responded to with actions promoted in the overdose training such as rescue breathing and/or heart compressions (35%) or calling an ambulance (46.6%) (Table 3). The vast majority of participants reported that opioids (on their own or in combination with another drug) were believed to have been used prior to the overdose (Table 3). Apart from opioids, benzodiazepines were the most common additional drugs involved, reported in 34.8% of cases.

Participants reported that the person who experienced an overdose was known to have survived in 95.6% of incidents. There were eight cases (0.5%) where the person who experienced an overdose could not be resuscitated and subsequently died. The outcome of the remaining cases was unknown. The vast majority (93.1%) of the participants responded that they felt very comfortable, or quite comfortable, using naloxone in overdose situations.

THN participants made a total of 4,884 refill reports, stating that the previous dose was used in an overdose situation in 33.3% of these reports. Other reasons for refill were: dose lost (32.9%), dose given away (15.5%), dose stolen (4.9%) or “other reason” (13.3%, most commonly the previous dose had expired or been confiscated by police or security guards). We used multinomial logistic regression to explore participant predictors for different naloxone dose endpoints (Supplementary Tables S1a-c). The risk of giving away the naloxone dose was higher among participants with stimulants as their primary drug (RRR 1.46; CI 1.07, 1.98) compared to those who primarily used other substances, and lower amongst those with prior experience of personal overdose (RRR 0.60; CI 0.50, 0.88) or witnessed overdose (RRR 0.66; CI 0.41, 0,85) compared to participants with no such experiences. Having the dose stolen was more likely amongst participants who were homeless (RRR 3.73; CI 2.06, 6.75) or in unstable housing (RRR 1.9; CI 1.03, 3.42) compared to people with other housing situations; those whose primary drug was stimulants (RRR 1.77; CI 1.14, 2.74) compared to those who mainly used other substances; women (RRR 1.70; CI 1.17, 2.47) compared to men; and participants with prior experience of witnessed overdose (RRR 2.18; CI 1.12, 4.23) compared to those without such experience. Participants with experience of personal overdose on the other hand, were less likely to report this outcome (RRR 0.51; CI 0.34, 0.76). Lastly, the risk of losing the naloxone dose was higher among those with unstable housing (RRR 1.32; CI 1.04, 1.68) and homelessness (RRR 1.91; CI 1.47, 2.48) compared to participants with other housing situations and lower amongst those with personal (RRR 0.64; CI 0.50, 0.81) and witnessed overdose experience (RRR 0.65; CI 0.45, 0.94) compared to those without such experiences.