Abstract
Background:
The REpeated-dose Behavioral intervention to reduce Opioid Overdose Trial (REBOOT) evaluated motivational interviewing for overdose prevention, focusing partly on witnessed overdose response. We assessed participants’ witnessed overdose history and REBOOT’s impact on overdose response among opioid overdose survivors using non-prescribed opioids in Boston and San Francisco.
Methods:
We described participants’ witnessed overdose and naloxone administration history over the four months preceding enrollment and the number and characteristics of witnessed overdoses reported during the study. We then used generalized estimating equations to test if the intervention affected if participants responded “me” to who, if anyone, responded to a witnessed overdose by assessing responsiveness, calling 911, performing rescue breathing, doing chest compressions, or administering naloxone during 16-month follow-up.
Results:
Of 265 participants, most (83%) witnessed at least one overdose in the four months preceding enrollment; 68% of these participants had administered naloxone. In the 16-month post-enrollment follow-up, 250 (94%) participants witnessed 597 overdoses. In 94% of these, participants reported that somebody (themselves or others) responded. REBOOT had no significant impact on whether participants personally responded to overdoses: 67% of control and 72% of intervention participants personally responded pre-enrollment (p=0.39), versus 63% and 62% post-enrollment (p=0.97).
Conclusion:
Most REBOOT participants witnessed a recent opioid overdose preceding enrollment. After enrollment, nearly all witnessed overdoses were responded to; the intervention had no effect on whether the participant personally responded. Given near-universal response, there was little room for improving overdose response. Future research should test similar interventions in communities with less prior overdose prevention exposure.
Keywords: opioid (MeSH), overdose (MeSH), opioid-related disorders (MeSH), naloxone (MeSH), witnessed overdose, overdose response
Introduction
In the US, drug overdose deaths surpassed 100,000 each year from 2021–2023.1 Opioid overdoses are frequently witnessed, and as such, are amenable to bystander intervention.2,3 Bystander actions during witnessed overdose may include checking for responsiveness in the overdosing person, calling an ambulance, doing rescue breathing or chest compressions, and/or administering naloxone. Studies have found that people who use drugs commonly report using such techniques to respond to witnessed overdoses.3,4
Prior studies have found that responding to overdoses is associated with both individual characteristics and social factors. For example, experiencing homelessness and having large social networks have been associated with witnessed overdose response,5,6 however, response may also be affected by setting-specific factors like naloxone access.7 To our knowledge, no studies have investigated the role of an overdose prevention motivational interviewing intervention on bystander actions taken during witnessed overdose.
The REpeated-dose Behavioral intervention to reduce Opioid Overdose Trial (REBOOT) administered a motivational interviewing intervention for overdose prevention to opioid overdose survivors using non-prescribed opioids in Boston and San Francisco.8,9 While the intervention focused on preventing the occurrence of opioid overdose for the participant, a portion of the intervention addressed strategies for responding to witnessed overdose. We first described the occurrence and sociodemographic correlates of recent witnessed overdose and naloxone administration among participants preceding study enrollment. We then assessed whether the REBOOT intervention had an impact on if participants personally responded to witnessed overdose during study follow-up.
Methods
Study setting and population
REBOOT was a two-site, phase III randomized trial in Boston and San Francisco (NCT03838510). The overarching goal of REBOOT was to evaluate the efficacy of a repeated motivational interviewing intervention on reducing the occurrence and number of overdoses among a cohort of opioid overdose survivors using non-prescribed opioids.8,9 Here we present a pre-specified exploratory analysis to test the effect of the REBOOT intervention on participants’ personal management of witnessed overdose.
Participants were eligible for REBOOT if they were aged 18–65 and self-reported a non-fatal opioid overdose in the past three years, non-prescribed opioid use in the past two weeks, and prior receipt of take-home naloxone. Participants also met criteria for moderate-to-severe opioid use disorder (OUD) (based on DSM-5 criteria10 administered by research staff) and had an opioid positive urine drug screen at screening. Participants were recruited through active strategies (i.e., community locations such as syringe access sites, naloxone distribution sites, local community-based organizations, and substance use disorder treatment centers), and passive distribution of posters, cards, and flyers.
Participants were enrolled from April 2019 to June 2022 and provided written informed consent for study participation. Participants attended REBOOT study visits and underwent assessments at enrollment and then every four months for 16 months. At enrollment and months 4, 8, and 12, participants randomized to the intervention received a brief intervention including an established overdose education curriculum within an Informational-Motivation-Behavior (IMB) model, and participants randomized to the control group watched videos or listened to podcasts for attention control.8,9 This study was approved by the University of California San Francisco Institutional Review Board (17–24203).
REBOOT intervention
The interview was conducted by trained interventionists who used motivational interviewing principles, such as open-ended questions, affirmations, reflections, and summaries, as well as concepts of “change talk” and “readiness” to collaboratively explore strategies for reducing harm from opioid use and overdose both among participants themselves and among community members during witnessed overdose.8 In the witnessed overdose-focused portion of the intervention, the interventionist asked participants to recall recent witnessed overdose experiences and discuss how they responded and potential risk factors that contributed to the overdose. They also reviewed how to recognize overdose signs, and different methods for responding, including assessing responsiveness, administering naloxone, calling 911, and doing rescue breathing and/or chest compressions. Participants also received overdose prevention-related handouts, including handouts on responding to a witnessed overdose (Supplemental Figures 1–2).
Measures
Assessments, administered to all participants prior to intervention or attention control sessions, were conducted by trained assessors, blinded to the participant’s study arm. At each visit, participants were provided with a definition of opioid overdose, and then assessment for witnessed overdose and response over the past four months (at enrollment) or since their prior visit or past four months, whichever was shorter (during follow-up).
At enrollment, the assessment included self-reported demographics and lifetime, past year, and past four-month witnessed overdose and naloxone administration history. Other participant characteristics included age, gender (male, female, transgender male, transgender female, other), racial identity (Native American / Alaska Native, Asian, Native Hawaiian or other Pacific Islander, Black or African American, white, more than one race, other), highest level of school completed (<12 years, ≥12 years), employment status (full time, part time, unemployed), lifetime and past four month homelessness (whether participants slept on the streets or in a shelter) (yes/no), and past 30-day use of non-opioid substances.
At enrollment and at follow-up visits, participants were asked about how many times they witnessed an opioid overdose and how many times they administered naloxone to someone experiencing an overdose in the past four months or since their prior visit. For those who witnessed at least one overdose during this period, they were asked detailed questions about the most recent witnessed overdose, including the participant’s relationship to the overdosing person, the location of the overdose, and overdose response actions. Participants were asked who, if anyone: checked for responsiveness; called 911; or administered rescue breathing, chest compressions, or naloxone. Participants could select as many options as appropriate, including “nobody,” “me,” “another witness,” “an emergency responder (paramedic, police, or other),” and “unknown.” For the question about who called 911, “emergency responder” was not an answer option.
Analysis
We excluded three REBOOT participants from this analysis due to missing enrollment witnessed overdose data, resulting in an analytical sample of 265 participants. We described witnessed overdose history preceding enrollment, including the proportion of participants who witnessed at least one overdose in their lifetime and in the year prior to study enrollment. We also described the proportion of participants who witnessed 0, 1, or more than 1 overdose in the past four months, and who administered naloxone 0, 1, or more than one time during this period. We defined these witnessed overdose and naloxone administration history categories based on prior literature finding significant differences in characteristics between “multiple overdose responders”, or those who have responded to more than one overdose, versus those who have responded to one,5 and based on data distribution across these categories (similar prevalence for 0 or 1 witnessed overdose, higher prevalence for >1).
We reported enrollment sociodemographic characteristics, study site, and randomized REBOOT group assignment for the whole sample and by categories of witnessed overdose history and naloxone administration. We used chi-square tests (for categorical variables) and the Kruskal-Wallis test (for age, which was skewed) to evaluate whether these characteristics varied across witnessed overdose and naloxone administration history categories. For expected cell counts less than 5, Fisher’s exact tests were used to compute p-values.
We also reported the characteristics of the most recent witnessed overdoses (i.e., relationship to the overdosing person, the location of the overdose, and overdose response actions). Our primary analysis evaluated whether REBOOT study arm was associated with personally responding to a witnessed overdose during the 16-month study follow-up period. If participants responded “me” to any of the questions about who checked to see if the person was responsive, called 911, did rescue breathing, did chest compressions, or administered naloxone, they were considered to have personally responded to the witnessed overdose event (regardless of whether or not they also reported another person responded). If they only responded “nobody”, “another witness”, “emergency responder”, or “unknown” to all of these questions they were considered as not having personally responded to the overdose. In secondary analyses, we evaluated whether the study arm was associated with participants reporting that “nobody” responded to all of the overdose response actions. We also tested whether study arm was associated with personally responding with each of the response actions individually.
To evaluate the association of the REBOOT intervention with these primary and secondary outcomes, we used generalized estimating equations (GEE) with an unstructured correlation structure to account for repeated measures within participants over study follow-up. Because enrollment covariates were balanced between the control and intervention groups, we did not adjust for covariates in the model. In post-hoc analyses, we conducted equivalence tests using two one-sided tests (TOST) to assess whether observed differences between the control and intervention groups were smaller than 25%. Equivalence was defined as a TOST risk ratio with its 90% confidence interval (CI) within the range of 0.8 to 1.25. If this criterion was met, we concluded that there was no meaningful difference between groups, and if not, the results were considered inconclusive for statistical equivalence.
Results
Cohort characteristics
Of 265 participants in this analysis, 133 were assigned to control and 132 to REBOOT, stratified by study site. At enrollment, the median age was 42 (IQR: 34–50) years, and most were male (62%), White (66%), and unemployed (85%). More than half (63%) had experienced homelessness in the past four months, and almost all (96%) had experienced homelessness in their lifetime (Table 1). At enrollment, past 30-day use of non-opioid substances was common, with more than half of participants reporting use of methamphetamine (70%), cannabis (68%), cocaine (64%), benzodiazepines (58%), and alcohol (53%). There were no statistically significant differences in measured characteristics between control and intervention groups (Table 1).
Table 1.
Sociodemographic characteristics of participants enrolled in a randomized trial of motivational interviewing to reduce opioid overdose among overdose survivors in Boston, MA, and San Francisco, CA (N=265)
| Participant characteristic at enrollment | Overall N=265 |
Control N=133 |
Intervention N=132 |
P-value |
|---|---|---|---|---|
| Age, median (IQR) | 42 (34–50) | 42 (34–49) | 42 (35–52) | 0.23 |
| Gender, n (%) | 0.74 | |||
| Male | 164 (62) | 79 (59) | 85 (64) | |
| Female | 98 (37) | 52 (39) | 46 (35) | |
| Transgender male | 0 (0) | 0 (0) | 0 (0) | |
| Transgender female | 2 (1) | 1 (1) | 1 (1) | |
| Other | 1 (0) | 1 (1) | 0 (0) | |
| Racial identity1, n (%) | 0.55 | |||
| Native American / Alaska Native | 5 (2) | 1 (1) | 4 (3) | |
| Asian | 2 (1) | 1 (1) | 1 (1) | |
| Native Hawaiian / other PI | 1 (0) | 0 (0) | 1 (1) | |
| Black or African American | 35 (13) | 18 (14) | 17 (13) | |
| White | 174 (67) | 84 (63) | 90 (68) | |
| More than one race | 25 (10) | 15 (11) | 10 (8) | |
| Other | 18 (7) | 10 (8) | 8 (6) | |
| Highest level of school, n (%) | 0.52 | |||
| <12 years | 77 (29) | 41 (31) | 36 (27) | |
| ≥12 years | 188 (71) | 92 (69) | 96 (73) | |
| Employment status, n (%) | 0.07 | |||
| Full time | 12 (5) | 9 (7) | 3 (2) | |
| Part time | 29 (11) | 18 (14) | 11 (8) | |
| Unemployed | 224 (85) | 106 (80) | 118 (89) | |
| Sleeping on streets, shelter: past 4m, n (%) | 0.84 | |||
| No | 98 (37) | 50 (38) | 48 (36) | |
| Yes | 167 (63) | 83 (62) | 84 (64) | |
| Ever slept on streets or shelter, n (%) | 0.36 | |||
| No | 11 (4) | 7 (5) | 4 (3) | |
| Yes | 254 (96) | 126 (95) | 128 (97) | |
| Research site, n (%) | 0.94 | |||
| San Francisco | 154 (58) | 77 (58) | 77 (58) | |
| Boston | 111 (42) | 56 (42) | 55 (42) | |
| Past 30-day non-opioid substance use, n (%)2 | ||||
| Methamphetamine/amphetamine | 186 (70) | 92 (69) | 94 (71) | 0.72 |
| Cannabis | 181 (68) | 85 (64) | 96 (73) | 0.12 |
| Cocaine | 169 (64) | 87 (65) | 82 (62) | 0.58 |
| Benzodiazepine | 153 (58) | 76 (57) | 77 (58) | 0.84 |
| Alcohol | 141 (53) | 76 (57) | 65 (49) | 0.20 |
| Gabapentin | 100 (38) | 57 (43) | 43 (33) | 0.08 |
| Clonidine | 69 (26) | 33 (25) | 36 (27) | 0.65 |
| Synthetic marijuana | 23 (9) | 11 (8) | 12 (9) | 0.81 |
Abbreviations: IQR: interquartile range; PI: Pacific Islander; m: months
Five participants declined to report racial identity
Categories select all that apply (not mutually exclusive)
Witnessed overdose events: relationship to overdosing person and location
During the study, including at enrollment and follow-up, participants witnessed 9,250 total overdoses. They described 818 most recent witnessed overdose events in the prior four months or since the last visit (if the last visit was within four months). In the majority of events, participants reported that their relationship to the overdosing person was a stranger (51%), followed by a friend (33%), sex partner (4%), family member (2%), and other relationship (11%). The three most common locations for witnessing overdose events were: open area (e.g., street, parking area, or schoolyard) (70%), participant’s home (9%), or somebody else’s home (8%).
Witnessed overdose and naloxone administration history preceding enrollment
Preceding enrollment, all participants had witnessed at least one overdose in their lifetime (median number witnessed [IQR]: 20 [6–40]), and 95% had witnessed at least one overdose in the past year (median [IQR]: 5 [2–12]). Most (83%) participants had witnessed at least one overdose in the past four months, and of these, 68% had administered naloxone at least once. There were no associations between having witnessed 0, 1, or >1 overdoses in the past four months and most sociodemographic characteristics (age, racial identity, education, employment status, lifetime homelessness) or study assignment. Participant gender and recent experiences of homelessness were associated with witnessed overdose category (p=0.03 and p<0.001 respectively). Males had a higher prevalence of witnessing >1 overdose in the past 4 months compared to females (70% vs. 59%), and those experiencing recent homelessness had a higher prevalence of witnessing >1 overdose than those not experiencing recent homelessness (77% vs. 47%) (Table 2).
Table 2.
Witnessed overdose and naloxone administration among participants enrolled in a randomized trial of motivational interviewing to reduce opioid overdose among overdose survivors in Boston, MA, and San Francisco, CA (N=265)
| Characteristic | Witnessed overdose, n (%) | Naloxone administration, n (%) | ||||
|---|---|---|---|---|---|---|
| 0 n=44 |
1 n=46 |
>1 n=175 |
0 n=114 |
1 n=47 |
>1 n=104 |
|
| Study assignment | p=0.58 | p=0.75 | ||||
| Control | 19 (14) | 23 (17) | 91 (68) | 56 (42) | 22 (17) | 55 (41) |
| Intervention | 25 (19) | 23 (17) | 84 (64) | 58 (44) | 25 (19) | 49 (37) |
| Age, median (IQR) | p=0.76 | p=0.34 | ||||
| 43 (34–51) | 44 (35–52) | 42 (34–50) | 41 (34–50) | 40 (34–47) | 44 (36–51) | |
| Gender | p=0.03 | p=0.90 | ||||
| Male | 18 (11) | 31 (19) | 115 (70) | 70 (43) | 32 (20) | 62 (38) |
| Female | 25 (26) | 15 (15) | 58 (59) | 43 (44) | 15 (15) | 40 (41) |
| Transgender female | 1 (50) | 0 (0) | 1 (50) | 1 (50) | 0 (0) | 1 (50) |
| Other | 0 (0) | 0 (0) | 1 (100) | 0 (0) | 0 (0) | 1 (100) |
| Racial identity1 | p=0.66 | p=0.49 | ||||
| Native American / Alaska Native | 0 (0) | 1 (20) | 4 (80) | 2 (40) | 0 (0) | 3 (60) |
| Asian | 1 (50) | 0 (0) | 1 (50) | 1 (50) | 0 (0) | 1 (50) |
| Native Hawaiian / other PI | 0 (0) | 0 (0) | 1 (100) | 0 (0) | 0 (0) | 1 (100) |
| Black or African American | 4 (11) | 8 (23) | 23 (66) | 14 (40) | 8 (23) | 13 (37) |
| White | 32 (18) | 28 (16) | 114 (66) | 76 (44) | 29 (17) | 69 (40) |
| More than one race | 1 (4) | 4 (16) | 20 (80) | 7 (28) | 5 (20) | 13 (52) |
| Other | 4 (22) | 3 (17) | 11 (61) | 10 (56) | 5 (28) | 3 (17) |
| Highest level of school | p=0.15 | p=0.13 | ||||
| <12 years | 15 (19) | 8 (10) | 54 (70) | 37 (48) | 8 (10) | 32 (42) |
| ≥12 years | 29 (15) | 38 (20) | 121 (64) | 77 (41) | 39 (21) | 72 (38) |
| Employment status | p=0.79 | p=0.27 | ||||
| Full time | 1 (8) | 1 (8) | 10 (83) | 8 (67) | 1 (8) | 3 (25) |
| Part time | 6 (21) | 4 (14) | 19 (66) | 9 (31) | 8 (28) | 12 (41) |
| Unemployed | 37 (17) | 41 (18) | 146 (65) | 97 (43) | 38 (17) | 89 (40) |
| Sleeping on streets, shelter: past 4m | p<0.001 | p<0.001 | ||||
| No | 27 (28) | 25 (26) | 46 (47) | 54 (55) | 21 (21) | 23 (23) |
| Yes | 17 (10) | 21 (13) | 129 (77) | 60 (36) | 26 (16) | 81 (49) |
| Ever slept on streets or shelter | p=0.53 | p=0.06 | ||||
| No | 2 (18) | 3 (27) | 6 (55) | 8 (73) | 2 (18) | 1 (9) |
| Yes | 42 (17) | 43 (17) | 169 (67) | 106 (42) | 45 (18) | 103 (31) |
| Research site | p=0.001 | p=0.16 | ||||
| San Francisco | 29 (19) | 37 (24) | 88 (57) | 71 (46) | 30 (19) | 53 (34) |
| Boston | 15 (14) | 9 (8) | 87 (78) | 43 (39) | 17 (15) | 51 (46) |
Abbreviations: IQR: interquartile range; PI: Pacific Islander; m: months
Five participants declined to report racial identity; missing values were excluded from statistical testing
Only recent homelessness varied across categories of administering naloxone 0, 1, or >1 time in the past four months (p<0.001). Participants reporting recent homelessness had a higher prevalence of >1 naloxone administration in the past four months than those not reporting recent homelessness (49% vs. 23%) (Table 2).
Overdose response
During post-enrollment study follow-up, 250 (94%) participants answered questions about 597 most recent witnessed overdose events (control: N=299; intervention: N=298). Overall, in this 16-month post-enrollment follow-up period, there were no significant differences between study arms in reporting “me” to one or more of the overdose response actions. In the control arm, participants personally responded to the witnessed overdose in 189 events (63%), versus 186 (62%) in the intervention arm (p=0.97). This was similar to the response rates at enrollment, where 67% of participants in the control group personally responded versus 72% in the intervention (p=0.39). In post-hoc equivalence tests, the control and intervention groups were statistically equivalent using the pre-specified threshold of 25% (risk ratio = 1.01, 95% CI: 0.89–1.14). In only 1/597 events did a participant report that “nobody” responded for all of the individual overdose response actions; statistical testing was not conducted due to the very low frequency of this outcome. Further, there were no significant differences between study arms in whether participants personally responded with individual overdose response actions. Among the individual response outcomes, we could only conclude statistical equivalence between control and intervention groups for naloxone administration (Table 3).
Table 3.
Effect of a randomized trial of motivational interviewing to reduce opioid overdose among overdose survivors in Boston, MA, and San Francisco, CA on proportion reporting “me” to questions of overdose response to their most recent witnessed overdose event (n=597)1
| Witnessed overdose response outcome | Control (N=299) n, %2 |
Intervention (N=298) n, %2 |
Odds ratio2 (95% CI) |
P-value2 |
|---|---|---|---|---|
| Responded “me” to any overdose response action | 189 (63) | 186 (62) | 1.02 (0.68, 1.52) | 0.97* |
| Responded “nobody” to all overdose response action | 1 (0.3) | 0 (0) | NA4 | NA4 |
| Responded “me” to3: | ||||
| Who checked for responsiveness? | 144 (48) | 166 (56) | 1.36 (0.93, 2.01) | 0.10 |
| Who called 9115 | 56 (19) | 36 (12) | 0.62 (0.36, 1.05) | 0.09 |
| Who did rescue breathing? | 77 (26) | 64 (21) | 0.79 (0.49, 1.26) | 0.31 |
| Who did chest compressions? | 62 (21) | 67 (22) | 1.12 (0.70, 1.80) | 0.80 |
| Who administered naloxone? | 130 (43) | 127 (43) | 1.02 (0.69, 1.50) | 0.98* |
Abbreviations: CI: confidence intervals
Outcomes evaluated for the entire study follow-up period (4-month through 16-month visits, adjusting for repeated measures over time). To see outcomes at each study visit time point, see Supplemental Table 1
Results from GEE model
Other response options included: nobody, another witness, emergency responder (paramedic, police, or other), or unknown. For prevalence of reporting each response option, see Supplemental Appendix
Outcome prevalence too low for statistical testing
There was one missing value for this question; missing value excluded from statistical testing
Indicates outcomes that were statistically equivalent between groups using a threshold of 25% (two one-sided test risk ratio 90% confidence interval between 0.80–1.25); otherwise, equivalence test risk ratio confidence intervals fell outside these bounds, and we could not conclude statistical equivalence (see Supplemental Table 2 for risk ratios and confidence intervals)
During the study follow-up period, the most common individual response was checking for responsiveness in the overdosing person, with participants reporting that somebody took this action (themselves, another witness, or emergency responder) in 94% of witnessed overdoses (93% among the control group, 96% among the intervention group). Naloxone administration was the second most common overdose response: this action was taken by somebody in 86% of witnessed overdoses (89% in control group, 83% in the intervention group). In both study arms, calling 911, doing rescue breathing, and doing chest compressions were less common (61%/57%, 57%/50%, and 51%/46% respectively).
Discussion
We found that in a cohort of opioid overdose survivors using non-prescribed opioids in Boston and San Francisco, there was high prevalence of witnessed overdoses, assessing responsiveness, and naloxone administration at enrollment. There was no association between the REBOOT intervention and the participant reporting that they were the person who responded to their most recent witnessed overdose. Given the near universal nature of response to witnessed overdoses in this cohort, there may have been little possibility for improvement in the outcomes studied. Our findings demonstrate that the sample of overdose survivors recruited to the REBOOT study from San Francisco and Boston were from communities of people who use drugs that regularly witness each other and respond effectively to others’ overdoses.
To our knowledge, this analysis is the first to evaluate the effect of a motivational interviewing-based overdose prevention education intervention on witnessed overdose response actions. While REBOOT had no significant impact on participants’ self-reported witnessed overdose response actions, other studies have found value in overdose response education for responding to witnessed overdose more generally; a meta-analysis found that overdose education improved knowledge on naloxone administration, overdose recognition, and overdose response.11 While we did not find a statistically significant difference in personal overdose response actions among those who received the intervention compared to those in the control group, this was an exploratory analysis the study was not specifically designed to investigate. Moreover, given the high baseline level exposure to overdose and naloxone administration, as well as community-level factors in participants’ local environments, there may have been minimal room for improvement in participants’ witnessed overdose responses.
The REBOOT study population had significant enrollment-level and ongoing exposure to witnessed overdoses, with 83% of participants reporting at least one witnessed overdose and 66% reporting more than one witnessed overdose in the four months prior to study enrollment. Most (68%) of those who had witnessed an overdose had used naloxone in the past four months. Participants were opioid overdose survivors, most with a history of homelessness, who lived in the urban areas of Boston and San Francisco. Overdose is common in these participants’ communities, and peer-to-peer response to witnessed overdose may be integrated into daily life, as exemplified by the number of overdoses witnessed. This is further highlighted by our findings that participants reported that most of their witnessed overdoses were among strangers and in public/open spaces, and that in close to 0% of witnessed overdoses did participants report “nobody” responded for all response actions. Both Boston and San Francisco have strong harm reduction cultures, offering overdose prevention education to people who use drugs through various programs. For example, Boston’s Overdose Education and Naloxone Distribution (OEND) programs, funded through the Massachusetts Department of Public Health, provide harm reduction resources across mobile units, outreach, drop-in centers, shelters, and treatment programs.12 From 2007–2018, OEND data reported 16,000 overdose rescues, 89% of which were conducted by people who use drugs.12 These data suggest that study participants may have already had substantial expertise with witnessed overdose response prior to REBOOT enrollment.
Our findings are consistent with prior studies that have found that experiencing homelessness is associated with witnessing more overdoses versus being stably housed.5,6 Preceding enrollment, participants experiencing homelessness in the past four months had a high prevalence of multiple witnessed overdoses in that period (77%). A prior study found that people who respond to multiple overdoses were more often homeless and also had larger social networks than those who responded to fewer overdoses.5 We also found that individuals who had recently experienced homelessness had higher prevalence of multiple naloxone administrations in the four months prior to enrollment compared to those who were housed. However, a different study found that people experiencing homelessness were less likely to carry naloxone; this study also found that naloxone ownership was more common in urban than rural settings.7 These differences highlight the importance of community-level context when considering the potential impact of a witnessed overdose response intervention. While urban centers or other locations with advanced harm reduction programs may foster a culture of peer-to-peer overdose response among certain communities, locations where individuals are more socially or geographically isolated and/or have fewer overdose response education resources may draw greater benefit from additional witnessed overdose education interventions. Additionally, the REBOOT intervention could be adapted to improve its implementation and uptake in a range of settings; for example, an online version of the training, or an app, could be developed to increase accessibility.
In the REBOOT study population, 94% of participants reported that somebody checked for responsiveness in the overdosing person when an overdose was witnessed. Consistent with existing research, this highlights the protective nature of using drugs around others, versus using alone.13 Tailored interventions focused on reducing the risks of solo drug use may be impactful and warrant further development.14,15 For example, the SafeSpot Overdose Prevention Hotline provides monitored confidential support for people using drugs alone, and triggers emergency medical services response if the person becomes unresponsive.16
This study has several limitations. First, the data on witnessed overdoses are self-reported, and therefore inherently subject to recall and social desirability biases. However, there is no other feasible method for measuring witnessed overdose. Relatedly, missing or unknown responses were prevalent for questions about witnessed overdoses (appendix). Second, while we defined overdose for participants, overdose witnesses do not always precisely identify overdose, and thus, participants may have misclassified witnessed overdose events. Third, we were not able to identify the need for certain overdose responses during the reported witnessed overdoses. For example, another bystander may have responded first, and therefore there may not have been a need for the participant to respond, or, in some cases, rescue breathing or chest compressions may not have been necessary. Fourth, while checking for responsiveness should always be a first step in overdose response, there are cases in which calling 911, administering naloxone, doing rescue breathing, or doing chest compressions are not necessary. We were not able to evaluate whether participants correctly identified instances when these responses were or were not needed. Future studies could assess the impact of the intervention on participant knowledge rather than reported real-life actions – for example – the ability to correctly identify when or when not do rescue breathing or administer additional naloxone doses in hypothetical situations. Also, because witnessed overdose response questions were only asked for the most recent witnessed overdose at each study visit, we may have missed witnessed overdose response actions that occurred for other witnessed overdoses. Last, our study population consisted of opioid overdose survivors using opioids in Boston and San Francisco, most with a history of homelessness, and our results may not be generalizable to populations with different characteristics or in different settings.
In conclusion, REBOOT, a motivational interviewing-based overdose prevention education intervention, did not have a significant effect on personally responding to witnessed overdose among intervention compared to control recipients. However, the study population had a high baseline level of experience with witnessed overdose and naloxone administration, and resided in communities where peer-to-peer witnessed overdose response action is common. Future studies should investigate the impact of a motivational interviewing intervention that addresses witnessed overdose response in communities that are socially or geographically isolated, and/or that have few other existing overdose education programs.
Supplementary Material
References
- 1.Provisional Drug Overdose Death Counts [Internet]. Centers for Disease Control and Prevention. 2025. [cited 2025 Jul 22];Available from: https://www.cdc.gov/nchs/nvss/vsrr/drug-overdose-data.htm [Google Scholar]
- 2.McCann MK, Jusko TA, Jones CMC, Seplaki CL, Cushman JT. Degree of Bystander-Patient Relationship and Prehospital Care for Opioid Overdose. Prehospital Emergency Care 2021;25(1):82–90. [DOI] [PubMed] [Google Scholar]
- 3.Lankenau SE, Wagner KD, Silva K, et al. Injection drug users trained by overdose prevention programs: responses to witnessed overdoses. J Community Health 2013;38(1):133–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Wagner KD, Davidson PJ, Iverson E, et al. “I felt like a superhero”: the experience of responding to drug overdose among individuals trained in overdose prevention. Int J Drug Policy 2014;25(1):157–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Wagner KD, Marks C, Felsher M, Latkin C, Pearson JL, Falade-Nwulia OO. Individual and social network correlates of responding to multiple overdoses among a cohort of people who use drugs. Drug Alcohol Depend Rep 2022;5:100107. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Bohnert ASB, Tracy M, Galea S. Characteristics of drug users who witness many overdoses: implications for overdose prevention. Drug Alcohol Depend 2012;120(1–3):168–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Lipira L, Leichtling G, Cook RR, et al. Predictors of having naloxone in urban and rural Oregon findings from NHBS and the OR-HOPE study. Drug and Alcohol Dependence 2021;227:108912. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.McMahan VM, Arenander J, Matheson T, et al. “There’s No Heroin Around Anymore. It’s All Fentanyl.” Adaptation of an Opioid Overdose Prevention Counseling Approach to Address Fentanyl Overdose: Formative Study. JMIR Form Res 2022;6(9):e37483. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Repeated-dose Behavioral Intervention to Reduce Opioid Overdose (REBOOT) [Internet]. ClinicalTrials.gov. [cited 2022 Nov 27];Available from: https://clinicaltrials.gov/ct2/show/NCT03838510
- 10.American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders [Internet]. Fifth Edition. American Psychiatric Association; 2013. [cited 2024 Jul 16]. Available from: https://psychiatryonline.org/doi/book/10.1176/appi.books.9780890425596 [Google Scholar]
- 11.Giglio RE, Li G, DiMaggio CJ. Effectiveness of bystander naloxone administration and overdose education programs: a meta-analysis. Inj Epidemiol 2015;2(1):10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Massachusetts Department of Public Health. Data Brief: Overdose Education and Naloxone Distribution Programs [Internet]. 2019. [cited 2025 Jan 30];Available from: https://www.mass.gov/doc/data-brief-overdose-education-and-naloxone-distribution-programs/download#:~:text=The%20Overdose%20Education%20and%20Naloxone,or%20witnessing%20an%20opioid%20overdose
- 13.Deo VS, Bhullar MK, Gilson TP, Flannery DJ, Fulton SE. The Need to Rethink Harm Reduction for People Using Drugs Alone to Reduce Overdose Fatalities. Subst Use Misuse 2024;59(3):450–8. [DOI] [PubMed] [Google Scholar]
- 14.Lombardi AR, Arya R, Rosen JG, et al. Overdose Detection Technologies to Reduce Solitary Overdose Deaths: A Literature Review. IJERPH 2023;20(2):1230. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Schneider KE, Allen ST, Rouhani S, et al. Increased solitary drug use during COVID-19: An unintended consequence of social distancing. International Journal of Drug Policy 2023;111:103923. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.SafeSpot [Internet]. [cited 2025 Jul 3];Available from: https://safe-spot.me/
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.