An observational prospective cohort study of naloxone use at witnessed overdoses, Kazakhstan, Kyrgyzstan, Tajikistan, Ukraine

Abstract Objective To determine whether participation in the United Nations Office on Drugs and Crime (UNODC) and the World Health Organization’s (WHO) Stop Overdose Safely (S-O-S) take-home naloxone training project in Kazakhstan, Kyrgyzstan, Tajikistan and Ukraine resulted in naloxone use at witnessed opioid overdoses. Methods An observational prospective cohort study was performed by recruiting participants in the implementation of the S-O-S project, which was developed as part of the broader S-O-S initiative. Training included instruction on overdose responses and naloxone use. Study participants were followed for 6 months after completing training. The primary study outcome was participants’ naloxone use at witnessed overdoses, reported at follow-up. Findings Between 400 and 417 S-O-S project participants were recruited in each country. Overall, 84% (1388/1646) of participants were interviewed at 6-month follow-up. The percentage who reported witnessing an overdose between baseline and follow-up was 20% (71/356) in Tajikistan, 33% (113/349) in Kyrgyzstan, 37% (125/342) in Ukraine and 50% (170/341) in Kazakhstan. The percentage who reported using naloxone at their most recently witnessed overdose was 82% (103/125) in Ukraine, 89% (152/170) in Kazakhstan, 89% (101/113) in Kyrgyzstan and 100% (71/71) in Tajikistan. Conclusion Implementation of the UNODC–WHO S-O-S training project in four low- to middle-income countries resulted in the reported use of take-home naloxone at around 90% of witnessed opioid overdoses. The percentage varied between countries but was generally higher than found in previous studies. Take-home naloxone is particularly important in countries where emergency medical responses to opioid overdoses may be limited.


Introduction
Opioid overdose is a leading cause of drug-related death. 1 The risk of an overdose varies with the type of opioid consumed and the population group involved. Longitudinal studies suggest that 2-3% of people who use heroin die each year but higher rates have been observed. 2,3 Opioid overdoses are preventable through opioid agonist maintenance treatment, though treatment is not available everywhere and uptake can be low. 4,5 Consequently, responses to acute opioid overdoses are often required. More than 80% of overdoses are accidental and many are reversible through respiratory support and administration of an opioid antagonist such as naloxone. [6][7][8][9] Naloxone is available in medical facilities in many countries and, since the 1990s, efforts have been made to provide non-medically trained people with the drug through takehome naloxone programmes. 8,10 Such programmes are now included in the World Health Organization's (WHO) recommended response to opioid overdose. 8,10 Take-home naloxone programmes involve training lay people likely to witness an overdose, such as the friends or family of people at risk, in overdose recognition (e.g. signs such as cyanosis) and how to respond through, for example, rescue breathing and naloxone administration. 11 The evidence shows that these programmes increase participants' knowledge, confidence and skills in managing opioid overdoses. [11][12][13] Moreover, they appear to be cost-effective and to reduce overdose deaths. [14][15][16] Importantly, there is no evidence that take-home naloxone leads to riskier drug use behaviour. 17 Although take-home naloxone is now used around the world, there are few publications from low-or middle-income countries, where little is known about opioid use or overdoses and where there may be limited access to emergency medical services. [18][19][20] Studies of opioid overdose prevention have been carried out in Kyrgyzstan and Tajikistan and overdose prevention has been investigated in Kazakhstan as part of a broader evaluation of an intervention to reduce the risk of human immunodeficiency virus and hepatitis C virus infections. 21 The take-home naloxone pilot programmes in Kyrgyzstan and Tajikistan involved only people who inject drugs: they were trained in overdose responses and given either vouchers for naloxone (Kyrgyzstan) or naloxone itself (Tajikistan). Subsequently, 83% (109/131) of programme participants who returned for additional naloxone in Kyrgyzstan and 30/59 (51%) in Tajikistan reported they had used naloxone at the last overdose witnessed. 20 However, naloxone use by participants who did not request more was unknown. In Kazakhstan, Objective To determine whether participation in the United Nations Office on Drugs and Crime (UNODC) and the World Health Organization's (WHO) Stop Overdose Safely (S-O-S) take-home naloxone training project in Kazakhstan, Kyrgyzstan, Tajikistan and Ukraine resulted in naloxone use at witnessed opioid overdoses. Methods An observational prospective cohort study was performed by recruiting participants in the implementation of the S-O-S project, which was developed as part of the broader S-O-S initiative. Training included instruction on overdose responses and naloxone use. Study participants were followed for 6 months after completing training. The primary study outcome was participants' naloxone use at witnessed overdoses, reported at follow-up. Findings Between 400 and 417 S-O-S project participants were recruited in each country. Overall, 84% (1388/1646) of participants were interviewed at 6-month follow-up. The percentage who reported witnessing an overdose between baseline and follow-up was 20% (71/356) in Tajikistan, 33% (113/349) in Kyrgyzstan, 37% (125/342) in Ukraine and 50% (170/341) in Kazakhstan. The percentage who reported using naloxone at their most recently witnessed overdose was 82% (103/125) in Ukraine, 89% (152/170) in Kazakhstan, 89% (101/113) in Kyrgyzstan and 100% (71/71) in Tajikistan. Conclusion Implementation of the UNODC-WHO S-O-S training project in four low-to middle-income countries resulted in the reported use of take-home naloxone at around 90% of witnessed opioid overdoses. The percentage varied between countries but was generally higher than found in previous studies. Take-home naloxone is particularly important in countries where emergency medical responses to opioid overdoses may be limited.
the programme focused on couples, at least one of whom reported injecting heroin. 21 All participants were trained in overdose responses and naloxone use and given vouchers for take-home naloxone. Although only 36% (148/414) of participants redeemed their vouchers, 71% (105/148) of those reported using naloxone on themselves or others, indicating that most would use the drug if available. 21 In 2017, at the United Nations Commission on Narcotic Drugs meeting in Vienna, the United Nations Office on Drugs and Crime (UNODC) and WHO launched the Stop Overdose Safely (S-O-S) initiative within the framework of the UNODC-WHO Programme on Drug Dependence Treatment and Care. 22,23 This initiative, which targets opioid overdose, was developed in response to recommendations of the 2016 Special Session of the United Nations General Assembly on the World Drug Problem and of the United Nations Commission on Narcotic Drugs resolution 55/7 (2012). 24 The initiative's aims are aligned with WHO's guidelines on the community management of opioid overdose, which state that, "people likely to witness an opioid overdose should have access to naloxone and be instructed in its administration to enable them to use it for the emergency management of suspected opioid overdose." 8 As part of the S-O-S initiative, a training package was developed and implemented in a project conducted in Kazakhstan, Kyrgyzstan, Tajikistan and Ukraine -low-income to upper-middleincome countries that have widely varying policies and practices on drug law enforcement and treatment. The S-O-S training project in these countries began with stakeholder consultations and a review of drug policy, of the legal status of naloxone, and of any considerations affecting the use of take-home naloxone (Table 1). 27 The final training model had three levels: (i) level-I trainers instructed level-II trainers in each country; (ii) level-II trainers instructed level-III training providers; and (iii) level-III training providers instructed potential opioid overdose witnesses. Specific training materials were developed for each level. Training of potential witnesses involved recognizing opioid over-dose signs and symptoms, responding to overdoses, understanding naloxone and its use, and preventing future overdoses. Details of the overall theory of change and the programme logic are available in the data repository. 28 The target was to train and distribute take-home naloxone to 4000 potential opioid overdose witnesses in each country.
During implementation of the training project, we carried out a prospective, observational cohort study to assess its impact. Here, we report on whether naloxone was used at witnessed opioid overdoses. Naloxone use at witnessed overdoses is a tangible and fundamental goal of take-home naloxone programmes that reflects both naloxone carriage by programme participants and whether training and naloxone carriage lead to its use at witnessed overdoses. We set an evaluation target of 90% of participants in the training project in Kazakhstan, Kyrgyzstan, Tajikistan and Ukraine using naloxone at witnessed opioid overdosesa figure higher than previously observed in two of these countries. 20,21  Bishkek, Sokuluk and Kant Dushanbe and Khorugh Kyiv a Data were obtained from legal reviews and assessments carried out in 2016.
Naloxone use at witnessed overdoses; Kazakhstan, Kyrgyzstan, Tajikistan, Ukraine Paul Dietze et al.

Methods
We recruited and interviewed a sample of project participants in Kazakhstan, Kyrgyzstan, Tajikistan and Ukraine before, immediately after and 6 months after overdose management training (Fig. 1). Recruitment took place between July and October 2019 at all study sites and follow-up ended in April 2020.
The whole training project was advertised primarily by word of mouth and using recruitment flyers posted at locations frequented by people who used opioids or who were likely to witness an overdose, such as outreach and clinical services staff. For our cohort study, a convenience sample of trainees was recruited from within the project, also by word of mouth, until a target number was reached in each country. Eligible participants were people likely to witness an opioid overdose (e.g. those who used opioids, their family members or friends, and contact workers). Participants had to be: (i) aged 18 years or older; (ii) resident in the study city for 6 months or more; (iii) able to fluently speak and read the language of the study instruments (i.e. Russian); (iv) willing to provide written informed consent; (v) willing to undergo follow-up assessment at 6 months; and (vi) willing to provide contact details, including their name, residential address, home and mobile telephone numbers and social media details as well as the corresponding details of a friend or relative who would know their whereabouts if they could not be contacted directly.
The broad training project involved one-to-one or group-based training on the effects of opioids, on recognizing overdoses, and on responding in accordance with S-O-S manuals. After training, participants received small plastic boxes containing two safely wrapped, 400 µg ampoules of naloxone, two clean needles and syringes for intramuscular administration, disposable gloves, alcohol swabs, an instruction leaflet and a note of endorsement from the relevant authorities. Each naloxone kit cost 1.70 United States dollars (US$).
People who agreed to participate in the cohort study were informed about study procedures by trained staff. They were given an information sheet that described the study in detail, including its procedures and the possible risks and benefits of participation, after which written informed consent was obtained.
Questionnaires were administered before, immediately after and 6 months after training (Fig. 1). Questionnaire responses were recorded on electronic devices programmed using Research Electronic Data Capture (REDCap) software (Vanderbilt University, Nashville, United States of America) or on hardcopy equivalents (details were subsequently entered onto REDCap forms). Data were uploaded directly onto the REDCap server at the Burnet Institute in Australia when a secure internet connection was available.
Participants in the cohort study were offered a cash reimbursement for their time, out-of-pocket expenses and transport costs at a local currency equivalent of 3-8 US$. Research assistants in each country attempted follow-ups after 6 months using the participants' contact details and the details of nominated friends or relatives. Participants who could not be contacted and interviewed within 8 months were deemed lost to follow-up. Data were collected in person or by telephone. Participants could request additional naloxone kits at any time but data on naloxone use were collected formally only at follow-up.

Measures
At baseline (i.e. before training), demographic information, including age, Six months after training, participants' responses to witnessed overdoses were assessed using a modified version of the baseline questionnaire that included questions on the carriage of naloxone, witnessed overdoses and actions, overdoses experienced, opioid use, and treatment, and that was based partly on a questionnaire from a pilot trial of prison-based naloxone-on-release. 33

Analysis
The primary outcome of our study was naloxone use at witnessed overdoses reported at 6-month follow-up and the target for the proportion of respondents in each country who reported naloxone use was 90%. We calculated 95% confidence intervals (CIs) for this proportion, which allowed for a margin of 5% as a reasonable indicator of whether the target had been achieved. Assuming that 50% of opioid-consuming participants would witness an overdose each year and that 10% of participants who did not report consuming opioids would witness an overdose each quarter, 20,34 it was estimated that roughly one third of study participants would witness an overdose during the follow-up period. Consequently, sample size calculations indicated that 408 participants were required in each country to achieve an estimated 138 witnessed overdoses by 6 months. Secondary outcomes related to programme implementation included: (i) the proportion of participants who still had the naloxone received at training; (ii) the proportion who told other people that they had access to naloxone; and (iii) the proportion who had carried naloxone during the previous 3 days. We also asked about the survival of the person whose overdose was witnessed. The characteristics of the study sample in each country are reported using descriptive statistics. Given the diversity of study participants between countries, no analysis of outcomes by participants' characteristics was undertaken.

Ethical approval
Ethical approval of the study protocol was obtained from the WHO Ethics Review Committee (ERC.  Table 2 shows the sociodemographic characteristics of the study participants in each country. The target sample size of 408 was achieved in all countries except Ukraine, which had 400 participants. Across the countries, the participants' mean age ranged from 38 to 42 years. The majority were employed and few reported homelessness in the previous 6 months. However, all other characteristics differed between countries: the reported educational levels were higher in Tajikistan and Ukraine; the proportion of women was higher in Kazakhstan and Kyrgyzstan; and the proportion of married participants was higher in Kyrgyzstan and Tajikistan. Over 80% of participants were retained in the study at 6 months, ranging from 82% (341/417) in Kazakhstan to 86% (342/400) in Ukraine.

Results
The percentage of participants who reported witnessing an overdose between baseline and follow-up at 6 months was 20% (71/356) in Tajikistan, 33% (113/349) in Kyrgyzstan, 37% (125/342) in Ukraine and 50% (170/341) in Kazakhstan. The percentage of overdose witnesses who reported using naloxone at their most recently witnessed overdose was 89% (152/170) in Kazakhstan, 89% (101/113) in Kyrgyzstan and 100% (71/71) in Tajikistan  (Table 3). However, the figures for Kyrgyzstan and Tajikistan should be treated with caution because the number of participants who witnessed an overdose was below that expected (i.e. 138 per country), meaning the precision of the estimates was lower than expected. The percentage of participants in Ukraine who reported using naloxone at their most recently witnessed overdose was 82% (103/125, 95% CI: 75-88), which fell just below the target of 90%. In almost all reported overdoses in which naloxone was used, the overdose victim survived ( Table 3). The percentage of participants at 6-month follow-up who still had one or both ampoules of the naloxone they received at enrolment was 89% (316/356) in Tajikistan, 72% (245/342) in Ukraine, 53% (184/349) in Kyrgyzstan and 45% (154/341) in Kazakhstan (Table 4). Over 80% of participants reported they had told others they had naloxone ( Table 4). The percentage who reported carrying naloxone in the 3 days before the 6-month follow-up varied substantially from 17% (58/342) in Ukraine to 89% (317/356) in Tajikistan (Table 4).

Discussion
Our prospective cohort study aimed to determine whether 90% of participants in an S-O-S training project in Kazakhstan, Kyrgyzstan, Tajikistan and Ukraine used naloxone when they witnessed an opioid overdose. The target of 90% was achieved in three of the four countries (and very close to being achieved in the fourth) and naloxone use was greater than observed previously. 20,21 In almost all reported instances of naloxone use, the recipient survived. Our findings demonstrate that the project resulted Naloxone use at witnessed overdoses; Kazakhstan, Kyrgyzstan, Tajikistan, Ukraine Paul Dietze et al.
in use of project-provided naloxone in a variety of settings.
One WHO-UNODC target for take-home naloxone is that 90% of trained potential witnesses should carry the drug or have it available for use. 23 Although our study was not specifically designed to assess carriage of naloxone, we found low reported rates of naloxone carriage, which were below the UNODC-WHO target and similar to previously observed rates. 35 However, this did not lead to a low rate of naloxone use at witnessed overdoses. We assessed naloxone carriage using a single question drawn from the N-ALIVE study; 33,36 we asked only whether naloxone had been carried in the previous 3 days, not whether programme-issued naloxone was available for use. Our findings suggest that this question was inappropriate for characterizing naloxone carriage as naloxone was clearly available when needed. Indeed, other information collected by the project team suggest that naloxone was available in locations where people may witness an overdose and was stored there rather than carried. Future work should use a measure of naloxone access that can better capture the availability of naloxone for responding to opioid overdoses.
Our study was a single-arm observational study appropriate for characterizing naloxone use by people provided with take-home naloxone in the training project. The study involved a large number of potential overdose witnesses in four diverse countries and had a retention rate over 80% at 6 months. The study did not seek to examine the effectiveness of the project in preventing overdose fatalities, which would have required a different study design. Nevertheless, although it is unknown how many overdoses would have proved fatal had naloxone not been used, participants using naloxone likely reversed potentially fatal overdoses. We base this conclusion on studies showing only around one quarter of overdose witnesses report calling an ambulance in Central Asian countries, 20 where emergency medical responses may be unavailable and fatal outcomes may, therefore, be more likely. In contrast, up to 78% of witnesses call ambulances in more-developed countries. 37 Our study findings suggest that implementing an S-O-S training project in low-and middle-income countries is feasible and can lead to naloxone use at witnessed overdoses. However, project implementation required substantial advance research, consultation and programme development to overcome numerous challenges. Analyses of local policing practices, for example, showed that awareness of opioid overdoses was low in most countries. As a result, project implementers designed and conducted first aid training for police officers in Kyrgyzstan and specific training in overdose responses for some police officers in Ukraine. Concerns about the police's attitude to naloxone carriage led to the inclusion of a note in naloxone kits indicating that the kits were endorsed by government authorities, even though no country had legislative barriers against the carriage of either naloxone or needles and syringes. In addition, the availability of naloxone in ambulance services varied across countries. These situational factors and other barriers to implementation need to be considered in future S-O-S training projects. Finally, plans for sustaining the project must be made before implementation. In our study countries, take-home naloxone programmes continue but resources have not been allocated to allow their expansion.
Our study was limited by the use of a convenience sample of participants and by reliance on self-report questionnaires to assess outcomes at 6 months. Self-report has proved reliable in studies involving people who use drugs, 38 but it is not known whether recall or other biases influenced responses to questions about overdoses in our study. Although our sampling strategy means our findings are not directly generalizable to oth-  Competing interests: PD was supported by an Australian National Health and Medical Research Council senior research fellowship (1136090), has received funding from Gilead Sciences and Indivior for work unrelated to this study and was an unpaid member of an advisory board for an intranasal naloxone product.