Key points are not available for this paper at this time.
BACKGROUND: Opioid overdose is a leading cause of accidental death in the United States. OBJECTIVE: To estimate the cost-effectiveness of distributing naloxone, an opioid antagonist, to heroin users for use at witnessed overdoses. DESIGN: Integrated Markov and decision analytic model using deterministic and probabilistic analyses and incorporating recurrent overdoses and a secondary analysis assuming heroin users are a net cost to society. DATA SOURCES: Published literature calibrated to epidemiologic data. TARGET POPULATION: Hypothetical 21-year-old novice U. S. heroin user and more experienced users with scenario analyses. TIME HORIZON: Lifetime. PERSPECTIVE: Societal. INTERVENTION: Naloxone distribution for lay administration. OUTCOME MEASURES: Overdose deaths prevented and incremental cost-effectiveness ratio (ICER). RESULTS OF BASE-CASE ANALYSIS: In the probabilistic analysis, 6% of overdose deaths were prevented with naloxone distribution; 1 death was prevented for every 227 naloxone kits distributed (95% CI, 71 to 716). Naloxone distribution increased costs by 53 (CI, 3 to 156) and quality-adjusted life-years by 0. 119 (CI, 0. 017 to 0. 378) for an ICER of 438 (CI, 48 to 1706). RESULTS OF SENSITIVITY ANALYSIS: Naloxone distribution was cost-effective in all deterministic and probabilistic sensitivity and scenario analyses, and it was cost-saving if it resulted in fewer overdoses or emergency medical service activations. In a "worst-case scenario" where overdose was rarely witnessed and naloxone was rarely used, minimally effective, and expensive, the ICER was 14 000. If national drug-related expenditures were applied to heroin users, the ICER was 2429. LIMITATION: Limited sources of controlled data resulted in wide CIs. CONCLUSION: Naloxone distribution to heroin users is likely to reduce overdose deaths and is cost-effective, even under markedly conservative assumptions. PRIMARY FUNDING SOURCE: National Institute of Allergy and Infectious Diseases.
Coffin et al. (Tue,) studied this question.