Distribution of intranasal naloxone to potential opioid overdose bystanders in Sweden: effects on overdose mortality in a full region-wide study

Introduction

Mortality from opioid use remains high.1 For decades, take-home naloxone has been suggested, and increasingly implemented, as a strategy against overdose.2 3 Take-home naloxone programmes have been increasingly established in many parts of the world.4 5

When a large-scale distribution of naloxone has been achieved, naloxone usage by overdose bystanders can become relatively common, and many opioid overdoses may thereby be reversed.6 7 In the reporting of events experienced by bystanders who have been included in naloxone programmes, the vast majority of overdose victims have been reported to survive the overdose event after bystander administration of naloxone.6 8 Although a certain fear of an increase in substance use due to naloxone provision programmes has been mentioned as a potential challenge, such a trend was not confirmed in a systematic review.9 However, there also has been study data indicating an increase in opioid-related emergency visits and crimes after the legal introduction of naloxone provision in the USA.10

However, few large-scale population studies have demonstrated whether the distribution of naloxone decreases overdose mortality in the population. Among four studies assessing overdose incidence in a naloxone programme compared with controls, only one study demonstrated a significant decrease, whereas the remaining studies demonstrated no change.9 In an observational study in a primary care setting in the USA, a reduction in emergency room visits for overdoses was seen among primary care patients who received a naloxone prescription along with their opioids for pain, in comparison to their counterparts who did not receive naloxone.11 Three studies with relatively limited study samples have failed to find favourable results on overdose incidence. These studies addressed clients in addiction programmes in Australia,12 in people who inject drugs in the USA,13 and in overdose survivors from emergency departments in the USA.14 On the other hand, indirect evidence from uncontrolled observational research has pointed to potential effects of large-scale naloxone provision.8 After the introduction of the Scottish prison naloxone programme, the proportion of overdoses occurring after prison release decreased substantially.15 Likewise, Walley et al in the USA demonstrated that the reduction in overdose deaths was the greatest in areas with a higher rate of people trained in naloxone intervention (above 100 individuals trained per 100 000 inhabitants).16 Altogether, more evidence is needed to outline the effects of naloxone distribution programmes.17

In Sweden, naloxone distribution for the prevention of overdose death was implemented in 2018, after legislative changes allowed for naloxone to be distributed to individuals at risk of overdose. The study aimed to assess, for a time period stretching 3.5 years from the start of the programme, whether rates of overdose deaths in the community decreased in comparison to a historic control period when naloxone was unavailable. To assess the potential role of changes in the at-risk population in the community, healthcare data and social services data were collected, to account for such potential changes in the overall problem burden of problematic opioid use.

Methods

Setting

The naloxone programme of Region Skåne started in 2018 and has been described in more detail elsewhere.7 18 Region Skåne constitutes the southernmost region of Sweden, with a total population in the region of 1.4 million inhabitants. The largest city in the region is Malmö (340 000 inhabitants), situated 30 min away from the Danish capital Copenhagen, and followed in size by the cities of Helsingborg, Lund and Kristianstad. In contrast to the rest of Sweden, needle and syringe programmes (NSPs) were introduced in Skåne already in the mid-1980s. Legislation did not allow NSP in the rest of the country until 2006. Even so, NSP access in Skåne is still the highest in national comparison, with well-established NSP facilities in all the four largest cities of the region.

In this setting, opioids represent by far the most common drug category involved in fatal drug overdoses. Overdose mortality increased substantially during the 2000s and 2010s, and in 2017, 87% of fatal drug intoxications were related to opioids, with heroin/morphine (27%), methadone (15%) and buprenorphine (15%) being the most common ones.19 In the Skåne region assessed here, a study from the 2010s reported that 44% of fatal opioid intoxications involved methadone, 21% involved heroin, 16% buprenorphine, 15% fentayl and 6% other opioids such as oxycodone or morphine. Sixteen per cent of overdose victims were in opioid maintenance treatment (OMT) at the time of death, and in total, at least 84% of deaths involved either benzodiazepines or other GABA-ergic sedatives.20 Access to OMT has also increased significantly during the past decade. Access to OMT historically was low in Sweden, as well as in the present region. However, research projects increasing treatment entry were introduced in the region in the early 2010s, in the cities of Malmö21 and Lund,22 and were also followed by a patient choice reform which further expanded access to treatment in large parts of the region.23 The region is, therefore, well equipped to reach the target population and the regional naloxone programme was the first in the country.

The first doses of naloxone in Region Skåne were distributed on 11 June 2018. Before that, the potential feasibility and acceptability of naloxone use in opioid-dependent patients had been demonstrated,24 as well as the need to implement further overdose-preventing measures, primarily in a harm reduction setting such as syringe exchange facilities,21 but until 2018, handing out naloxone to laypersons was not supported by Swedish legislation.

The programme involves the distribution of intranasal naloxone, as well as a brief emergency rescue training programme, for individuals perceived to be at risk of overdose and who attend either a syringe exchange programme, an OMT programme, an addiction specialised in-patient hospital ward or other addiction outpatient and emergency units. Up to the end of the present follow-up (December 2021), a total of 37 units in the county had initiated naloxone training and distribution (two of which had later closed down), and a total of 2097 patients had received take-home naloxone doses from any of those units until then.

As described previously,7 participants in the naloxone distribution programme primarily have a recent use of opioids (77% reporting a past-month use of opioids in a subsample of naloxone recipients enrolled in a clinical follow-up study from the project), while sedative drugs are the second most common category used during the past month (42%). Sixty-eight per cent of participants in that study are men.7

Study procedures

Study outcome was the number of fatal cases of overdoses, defined as the total number of annual deaths diagnosed as either F11.0 (acute opioid intoxication), X42 (accidental narcotics intoxication) and Y12 (narcotics intoxication with unknown intent), in the Skåne region, during the intervention period (2019–2021) compared with the historic control period (2013–2017). For further specification, we reported the subproportion of deaths with an X42 diagnosis, the subproportion of women among those deceased with fatal overdose, the proportion of deceased patients with a prior history of any substance use disorder (diagnostic group F1), and the total number of overdose deaths calculated per 100 000 inhabitants. The rate of overdoses was calculated both in total for the region of Skåne, Sweden and in a subanalysis among patients with a substance use disorder history, defined as any F1 diagnosis during the past 5 years, either in outpatient or inpatient care, either as primary or secondary diagnosis. These outcome data were derived from the patient registry which is held by the Swedish National Board of Health and Welfare.25 This is a well-established source of information describing specialised healthcare contacts and diagnoses, and is described to have a nearly full coverage for specialist diagnoses in Swedish healthcare, that is, in all healthcare except general practitioner facilities.25

The latter was carried out in order to separate—to the largest extent possible—overdoses related to problematic drug use from accidental overdoses related to licit opioid analgesic treatment. Here, a definition was chosen to include all substance use disorders, as it has been detected that even among opioid-related overdose deaths, a substantial majority have had polysubstance use at the time of death,20 and a large majority of opioid overdose victims have been reported to have had prior hospital contacts during the last year prior to death, and not necessarily for opioid use disorder treatment.26

To account for whether the total opioid-related problem in the region may have changed during the study period, hospital-based overdose data were derived from both the healthcare system, and treatment uptake data were derived from the social services system. Hospital-based overdose data were defined as the total of treatment occasions involving any of the diagnostic codes F11.0, X42 and Y12, either as primary or secondary diagnosis. These data were derived from the patient registry of the Swedish National Board of Health and Welfare.25 Patients reported in the treatment data here had to be at least 15 years of age, and data are reported per calendar year as unique numbers for that year, regardless of whether the patient was treated during previous years or not. Specialised outpatient healthcare and inpatient care are included.

Treatment uptake for opioid use disorders in the social services was recorded for a majority of the municipalities in the county, for the whole time period 2013–2021, year by year. Swedish social services (municipalities) and healthcare (regions) share the responsibility for the assessment and treatment of addictive disorders. In social services, many units use the Addiction Severity Index27 for a baseline assessment of clients seeking treatment. Data describing treatment uptake in the county were derived from the municipalities from whom a consent could be obtained from chief representatives in these municipalities. From 8 out of 33 municipalities, the researchers never heard back from the representatives contacted, and these were, therefore, excluded. Thus, finally, data were obtained from the remaining 25 out of 33 municipalities (including the four major cities of the county). From these, a number of treatment applications with a primary opioid problem were calculated for each year, as well as the mean number of past-30-day heroin use, methadone use, buprenorphine use and other opioid use. These comparisons thereby serve as a measure of whether the total opioid-related problem in the region can be suspected to have changed, as a decrease in opioid availability or in the number of opioid users could theoretically decrease opioid overdoses and mimic a treatment effect from naloxone provision. Thus, although methadone and buprenorphine use may represent either a regular treatment for opioid use disorders or illicit intake of the substances, these measures aim to describe the total number of clients seen by the social services for opioid use disorder issues.

Statistical methods

Incidence ratios with exact 95% CIs were calculated, comparing incidence before and after implementation. All analyses were carried out in Stata/SE V.17.0.

Patient and public involvement

Peer patient representatives have been involved in the planning of the study, and in parts of the patient training in the study.

Results

The numbers of overdose deaths in the region, for each of the years 2013–2021, can be seen in table 1. The annual average number of overdose deaths (table 2) decreased from 3.9 per 100 000 inhabitants in the control period (2013–2017), to 2.8 per 100 000 inhabitants in the intervention period (2019–2021). The corresponding change in men was from 6.7 to 4.3 per 100 000 inhabitants (table 3), and for women, from 1.2 to 1.3 per 100 000 inhabitants (table 4). In statistical analysis, this was shown to represent a significantly decreased incidence ratio of overdose deaths in the whole population and in men specifically, but not in women. When restricting the analysis only to individuals with an SUD history, the decrease in incidence ratio was also significant for the whole population, and for men specifically, but not for women (table 5).

Table 1

Numbers of overdose deaths in 2013–2021 i Skåne county (age 15+, underlying cause of death either X42 or Y12, International Classification of Diseases, 10th revision, ICD-10)

Table 2

Deceased from overdose before/during naloxone programme

Table 3

Deceased from overdose before/during naloxone programme, men

Table 4

Deceased from overdose before/during naloxone programme, women

Table 5

Risk of death from overdose before and during the naloxone programme

As seen in online supplemental table 1 (for the whole population), 2 (for men) and 3 (for women), there was no indication of a decreased clinical problem burden in the intervention period, compared with the control period, as described by the treatment uptake seen in the hospital system. Likewise, in social service treatment data, no indication was seen of a lowered problem burden during the intervention period (online supplemental figures 1–5). The annual mean number of treatment-seeking clients with opioids as their predominating problem was 56.4 for the historic control period, and 75.7 for the intervention period (for men 38.8 for the control period and 53.3 for the intervention period, and for women 17.6 and 22.3, respectively). Past-30-day heroin use was reported by an average of 56.6 for the control period and 66.3 for the intervention period (43.8 and 54.3 for men, and 12.8 and 12.0 for women). Past-30-day methadone use was reported by an average of 45.4 for the control period and 38.3 for the intervention period (36.0 and 30.0 for men, and 9.4 and 8.3 for women). Past-30-day use of buprenorphine was reported by an average of 81.8 for the control period and 81.3 for the intervention period (62.2 and 61.7 for men, 19.6 and 19.7 for women). Past-30-day use of other opioids was reported by an average of 131.4 for the control period and 141.7 for the intervention period (90.0 and 101.0 for men, and 41.4 and 40.7 for women).

Supplemental material

Discussion

The present study is, to the best of our knowledge, one of the first ones to demonstrate a decrease in opioid overdose death in the community, after the introduction of a take-home naloxone programme. While this study does not have access to any control population which may be at risk of overdose but which is not exposed to take-home naloxone, it used a historic control period to assess overdose risk with and without a naloxone programme. Importantly, the present study assessed potential effects of the naloxone programme on overdose mortality in the population as a whole, that is, not restricted to participants of a formal study.

Also, and importantly, treatment uptake data did not indicate that the decrease in overdose death would be due to a decrease in opioid use in society; acute overdose data from hospitals and data describing social service treatment contacts (online supplemental tables 1–5) for opioid use disorders, indicated that problems related to opioids in the region had been stable or increased over time, rather than the opposite. This includes the fact that a patient choice reform expanded the number of patients in OMT from 2014,23 and this overall increase in medical treatment may also explain the increase in opioid-related treatment uptake in social services from the control period to the intervention period. Overall, there appears to be no reason to believe that the number of patients with an opioid use disorder decreased in the region, and that the decrease in overdose deaths was related to that and not to the introduction of naloxone.

In gender-specific analyses, however, a significant decline in overdose death could only be seen in men. The reported use of opioids was consistently higher among men, regardless of the type of opioid used. This is in accordance with previous reports from facilities in the region, both regarding participants in OMT and in NSP.21 28 Here, the absolute number of women in the analyses is too low for any certain conclusions to be drawn. Low absolute numbers make it difficult to demonstrate significant changes in any direction from an intervention, and a floor effect also cannot be excluded, making any decrease difficult to demonstrate. Thus, hitherto no decrease in overdose deaths in women was seen and it cannot be concluded whether effects of a naloxone programme are less likely in women. This merits further research, in the light of the likely gender differences in drug use practices. In the setting of OMT, including in the present region, it has been shown that women may have a markedly higher psychiatric problem burden than men,28 and women with injecting drug use may have a higher degree of risky injection practices than men.29 Altogether, the inconclusive findings of the present study within the female groups of patients call for specific studies addressing women’s participation in naloxone programmes and their involvement in the peer management of opioid overdoses.

The present study is one of few which can compare overdose death rates in a region during time periods with and without access to naloxone distribution, and where the impact on the population as a whole—not only specifically in individuals enrolled in a research study—can be measured.8 9 For comparison, however, in a community-based Massachusetts study, it was shown that in geographical entities where a higher level of naloxone distribution was achieved, mortality from overdose death decreased significantly, compared with areas where no naloxone distribution was carried out.16 Correspondingly, the present study instead demonstrated lower rates of deaths compared with historic control years. In addition, previous case series and observational studies have shown an overall high success rate in events where community-distributed naloxone has been used on overdose victims, that is, with low numbers of deaths in cases where naloxone is administered. Altogether, the current evidence hitherto relies on few controlled studies but on many uncontrolled and favourable follow-up studies,8 and in that perspective, the present data add to the body of evidence from controlled data. These findings should, however, be seen in the context of study data showing a lack of effects on overdoses but a simultaneous increase in other drug-related variables after naloxone introduction, demonstrating the likely complexity of causal and non-causal associations occurring in parallel with this harm reduction intervention.10 Altogether, this points to the need for more controlled, and longitudinal, research studies in this area.

Theoretically, a head-to-head comparison between regions with and without naloxone programmes could provide treatment evidence for this type of intervention. For example, the Massachusetts study used a strategy similar to this, although regions with a higher or lower level of distribution were compared. In the present setting, naloxone programmes were gradually introduced, starting from 2018, in a number of regions, although the region studied here is one of the regions with the most extensive and structured programmes, following a large top-down implementation procedure.7 A head-to-head comparison with a region without such implementation, or with a more limited naloxone distribution, could be possible, but would be strongly affected by the fact that overdose deaths are unevenly distributed across the country. National reports of overdose deaths have confirmed this, indicating a stronger concentration of overdose deaths to urban areas, in comparison, for example, to suicide cases.30

Important challenges remain for naloxone programmes in reducing opioid-related overdose death. It has been reported that a large majority of overdoses occur in private, without witnesses, such that the potential of take-home naloxone is substantially reduced.31 Likewise, less is known about how naloxone programmes counteract the overdosing of less typical substance use patterns than heroin; research has been called for which can shed light on the effects and feasibility of take-home naloxone where other opioids than heroin are predominating in overdoses.17

Also, there is reason to believe that individuals in active drug use are more likely to use naloxone. In a Norwegian study, individuals in active drug use, or who had experienced overdoses to a larger extent at baseline, were more likely to report having administered naloxone to overdose victims,32 and in Sweden, syringe exchange clients were more likely than individuals in opioid dependence treatment to have administered naloxone.7

The present study has limitations. These are mainly since overdose occurrence is measured through national registers, which is both a strength as they provide a full picture of mortality and causes of death in society, but also do not provide any exact, individual assessment of the fatal cases for the present study specifically. Thus, data rely on the evaluation made by the physician responsible for the cause-of-death certificate, but based on the causes of death assessed here, and based on Swedish legislation, a large number of these cases are likely to have been registered by physicians in specialised forensic medicine, after a full forensic-oriented autopsy including police reports and other circumstantial information about the death and events preceding it. Thus, although a number of deaths also may have been registered only after a clinical assessment in hospital, causes of death data are likely to be as grounded as possible. In addition, in a study addressing community deaths rates from overdose, the actual degree of penetration of naloxone distribution to people at risk cannot be accounted for. Again, no individual assessment was done in the present study, and whether or not people had been reached by—or were theoretically subject to—the naloxone programme in the region cannot be demonstrated here, but will be subject to future studies within the project, aiming to study the individual course in substance use and health among project participants specifically.18 Likewise, the number of affected individuals who were themselves in OMT is also unknown. Also, the present study is limited by the fact that we cannot measure overdose deaths as a proportion of the total number of opioid users in the region, as this total number is unknown. Therefore, instead, the number of deaths was studied among the number of individuals needing hospital care for substance use disorders, but cannot account for the unknown trends in the actual opioid use.

Also, one further limitation is the fact that the role of concurrent non-opioid use in overdose deaths is difficult to account for. Here, benzodiazepines and other sedatives are likely to have contributed to death in a number of overdose cases, as well as other life circumstances preceding death. As the present study is an analysis of overdose deaths in the community, rather than a study following identified individuals, such contributing factors cannot be accounted for in the present study. This includes the limitation that the proportion of overdoses involving non-opioid sedatives cannot be reported here. Finally, the intervention period in the present study coincided with the SARS-CoV2 pandemic. Some limitations in access to overall healthcare therefore may have been inevitable. The Swedish COVID-19 policy, however, did not involve formal lockdown procedures and did not limit access to needle exchange or OMT, and the naloxone distribution was continuously active throughout the pandemic.

In conclusion, based on the study design and with potential study limitations in mind, the present study lends support to the hypothesis that a large-scale naloxone programme in a region may decrease rates of overdose death in the population, in comparison to a historic control period when naloxone distribution was unavailable. The potential effect hitherto can be shown for men only, and further, larger studies are needed to fully outline the effects of this intervention on overdose mortality in women.

This post was originally published on https://bmjopen.bmj.com