BackgroundOverdose education and naloxone distribution programs decrease opioid overdose deaths. However, no studies of such programs have been carried out in Denmark. The aim of this study was to evaluate the feasibility and the effect of a broader “training-the-trainers” model in low-threshold settings after participation in the “Danish Save Lives” [SL] program.MethodsBetween May 2013 and November 2015, 552 participants from four municipalities took part in the SL program. The program is built on the train-the-trainers model where a central trainer trains others (trainers), who in turn train others (helpers). Participants were 30 police officers (5%), 188 people who use opioids (34%), 23 significant others (4%), and 217 social workers (39%). Ninety-four participants could not be classified (17%). At follow-up, participants were interviewed to determine the number and outcomes of opioid overdoses. Logistic regression was used to assess predictors of treating an overdose.ResultsIn all, 37 (7%) participants had intervened in 45 opioid overdose events (two trainers and 35 helpers). Detailed descriptions of the overdose event were available from 32 follow-up interviews (70%). In 16 cases, the person who intervened was already present at the site when the overdose occurred, and in 17 cases, the overdose victim recovered without complications. All overdose victims survived except one. People who used opioids were more likely to have treated an overdose than other participants (adjusted odds ratio [AOR] = 8.50, p = 0.001), and the likelihood of treating and overdose declined over time AOR = 0.37 (0.13, 0.93), p = 0.034).ConclusionsPrevention programs that target people who use opioids are more likely to be effective than programs that target professionals, especially in high-risk settings that can be hard for paramedics to reach. A future goal is to explore how prevention programs can be adapted to new user groups.Trial registrationThe Danish Data Protection Agency, 2015-57-0002, Aarhus University, 2016-051-000001, 184, retrospectively registered