The United States is in the midst of a drug overdose epidemic. Following a progressive surge in overdose deaths from 2000 to 2017, there was a slight decline in death rates in 2018, only to be followed by a resurgence in deaths likely attributable to factors surrounding the COVID-19 pandemic (Manchikanti et al., 2021; Ghose et al., 2022). The resurgence has continued with a projection that we will exceed 110,000 overdose deaths in the U.S. in 2022, an increase of 11 % from 2021 (Centers for Disease Control and Prevention, National Center for Health Statistics, 2023). The epidemic is being fueled by synthetic opioids, primarily fentanyl. Disadvantaged and marginalized communities have been hit particularly hard (Frankenfeld and Leslie, 2019, Altekruse et al., 2020, Forati et al., 2021, Milam et al., 2021). High overdose death rates are associated with community-level indicators of socioeconomic duress, such as neighborhood instability, crime rate, and inequality of household income (Altekruse et al., 2020; Van Draanen et al., 2020; Forati et al., 2021). As we struggle to respond to the overdose crisis, it has become clear that new strategies are needed to better define the problem and its underlying causes.

Many drug overdose deaths are geographically discordant (the individuals who die from drug overdoses do not reside in the locale where the overdose occurs). For example, in Milwaukee County, Wisconsin, 26.72 % of drug overdose deaths have involved individuals from neighborhoods/communities that differ from those in which the fatal overdose occurs (Ghose et al., 2022). Understanding factors associated with discordant overdoses should provide insight into determinants of overdose risk and therefore support and guide prevention and harm reduction efforts. Indeed, prior work has found that geographic discordance between communities of residence and those in which treatment is received increases the recurrence of prescription opioid misuse, anxiety, and incarceration in those with opioid use disorders (Oser and Harp, 2015). However, geographic discordance between communities of residence and those in which fatal overdoses occur has not been investigated.

The recognition that there is geographic discordance in many overdose deaths illustrates that for every overdose, there is a journey. While each journey is unique, much can be learned about factors that contribute to overdose risk by understanding common characteristics of those whose overdoses are geographically discordant, their community of origin/export, the overdose destination/import community, as well as the distance traveled from a person’s residence to the overdose location. In the field of criminology, the distance traveled from an offender's residence to the crime scene is called the “journey to crime” (Rengert, 2002, Townsley and Sidebottom, 2010). Johnson et al. (2013) studied the journey to crime for illegal drug buyers, highlighting the impact of race and neighborhood characteristics on the distance traveled to drug purchase arrests. Donnelly et al. (2021) studied racial differences in journeys to crime in opioid possession-related arrests, demonstrating how racial and spatial differences can reveal persistent disparities in drug law enforcement.

Inspired by criminology studies, this study examines the "journey to overdose," defined as the difference between the drug overdose decedent’s residence and overdose location. Each journey to overdose is complex and may involve multiple stops, backtracking, or loops around areas before the incident (Ackerman and Rossmo, 2015, Johnson et al., 2013). Although a journey can be most easily conceptualized as a distance traveled (see, e.g., Rengert, 2002), journeys can be defined according to geographic discordance in the demographics as well as socioeconomic and cultural contexts between the community of origin/residence and that in which overdose occurs. Geographically discordant and non-discordant overdoses are likely to differ in terms of demographic variables such age, gender, and race, as well as the drugs involved and overdose mode (i.e., accident, suicide, etc…). Observed differences should provide insight into individual risk factors. Similarly, locations from which individuals who overdose travel and those in which overdoses occur are likely to have distinct demographic and socioeconomic features such economic and housing stability, education level, and age and racial composition, thus enabling inference of community-level determinants of overdose risk.

Overdose journeys can be studied using network analysis. Substance use occurs in the context of social networks intertwined with the geographic landscape (Fischer, 1982, Andris, 2016). A spatial social network is a collection of nodes connected by edges or agent-based connections between people or institutions embedded in geographic space (Wasserman and Faust, 1994, Ye and Andris, 2021). In geographical network research, nodes are used to represent spatial objects (e.g., neighborhoods), and the edges are used to represent connections. Such analyses have allowed researchers to identify relationships and study them individually or in aggregate within complex networks at varying scales (Uitermark and Van Meeteren, 2021).

Here we propose a novel framework to understand the journey to overdose through spatial social network analysis. The framework will allow us to (i) pinpoint focal points of geographically discordant overdose deaths (ii) unveil hotspots of discordant (e.g., imported) drug overdoses, and (iii) investigate variables that differentiate discordant from non-discordant overdose deaths. We apply this framework to study Milwaukee County, a diverse and segregated metropolitan area in southeastern Wisconsin that has experienced a steady increase in overdose deaths. Prior research has found that overdose death rates vary across diverse communities in Milwaukee County and that many overdose deaths in the county are geographically discordant (Forati et al., 2021, Ghose et al., 2022). Using Milwaukee as a case study, this paper is the first, to our knowledge, to apply spatial social network analysis to study the journey to drug overdose, thus providing proof of the principle that this approach can be used to understand and guide community responses to the ongoing drug overdose epidemic.