A growing body of evidence shows that when a person is dependent on opioids, temporarily interrupting their supply of the drug exposes them to an increased risk of overdose (Hochstatter et al., 2023; Mark & Parish, 2019; Williams et al., 2020). The first studies about interruptions originating from police drug seizures explored this assertion qualitatively, finding that people problematically dependent on opioids reported this increased risk when asked about their personal experiences (Carroll et al., 2020; Rhodes et al., 2019; Victor et al., 2020). Conversely, these and other studies found people who use drugs (PWUD) with steady access to a supplier they could trust, who provided a basically consistent supply, could warn of inconsistencies, and who might supply naloxone, was protective of overdose (Carroll et al., 2020; Hedden-Clayton et al., 2024). More recently, a quantitative analysis found a statistically significant spatiotemporal association between police opioid seizures and overdose during a two-year period in Indianapolis, Indiana (Ray et al., 2023). In accordance with studies that show ecological associations at the state and county levels (Cano et al., 2024), the analysis found that fatal overdoses doubled within three weeks of the police seizure, in a radius of up to 500 meters from the seizure’s site, and this association held when considering all quantities of opioids seized, regardless of whether they were taken from a dealer or an individual who uses drugs (Ray et al., 2023).

Many people find the conclusion that police drug seizures increase overdose risk strongly counterintuitive, considering that governments act to interrupt the supply of illicit drugs to ostensibly reduce risk and save lives. To that end, the presumption is that reducing the quantity of illicit drugs in circulation and interrupting their consumption by people dependent on opioids is believed to reduce risk. But at least in the proximate sense, the evidence does not bear this presumption out. To address this gap between emerging research findings and many people’s intuitions, and to promote the rigorous interrogation of this proposition, this paper presents the discrete causal model by which increased exposure to overdose arises from the aftermath of a police drug seizure, a phenomenon that has critical implications for our present public health and drug enforcement policies. In doing so, it intends to bolster our understanding of the health outcomes of police drug enforcement and suggest ways to interrupt the causal chain to reduce overdose mortality. Barring such a model, the evidence we produce to explore this relationship will continue to face its inherent limitations. Qualitative research conveys the lived experience of increased exposure to overdose in the aftermath of police drug seizures with nuance, but the method inherently precludes broad generalizability. In contrast, quantitative analyses of administrative data can demonstrate statistically significant associations, but do not readily illuminate the causal pathways that suggest where to effectively intervene. Both veins of research would benefit from the development of a model that conveys how personal experience and quantitative correlation can be understood as causation. By presenting such a model here, this paper provides guidance about where future research on this topic should go, and what intervention designs might look like.

The model we present takes policing as its primary focus. It concerns the events that we commonly understand to be police drug seizures: the arrest of people for drug possession, the seizure of contraband drugs that people preemptively discard to avoid arrest by police, and the arrest of distributors and suppliers accompanied by the seizure of their drugs as evidence. It does not distinguish between the volume or target of the seizure; it could consist of the arrest of an individual person and the seizure of their own supply, or the seizure of a dealer one or more steps up the supply chain. The inclusion criteria for the model are only that opioids are seized by police, and in a manner that ultimately interrupts supply at an individual level. The model therefore operates at these lower geographic resolutions, during the relatively short time frames considered in Ray et al. (2023), and does not intend to offer explanatory power across larger ecologies such as cities, counties, or states, or over longer periods of time.

The model here takes up overdose as the exclusive outcome of interest, although there is evidence that police drug seizures can increase the risk of other harms as well. In the aftermath of police contact, people may change the way they consume drugs, for example, speeding the act of injection or sharing syringes (Cooper et al., 2005; Werb et al., 2008), thereby increasing the risk of abscess and infectious disease (Baker et al., 2020). To keep our causal model as direct and compelling as possible, these other outcomes will not be its focus. In the same vein, while the model here pertains to policing, and is not intended for general application across the criminal justice system. Release from a period of incarceration is a prime example of another event that substantially increases risk of overdose (Binswanger et al., 2011; Binswanger et al., 2007), and researchers have presented the corresponding conceptual framework (Joudrey et al., 2019), while emphasizing the particular role of fentanyl in generating these overdoses (Brinkley-Rubinstein et al., 2018). We intend to complement this research with a pragmatic realist account (Cherryholmes, 1992) of overdose risks that occur upstream, at the point of police encounters, prior to a period of imprisonment. In its realist approach, our account asserts that prevalent scientific and behavioral theories accurately describe the aspects of the world they were created to explain (Leplin, 1984; Psillos, 2005). We further assert that their individual explanatory mechanisms can combine to produce an accurate, predictive understanding of an overall phenomenon (Leplin, 1984), in this case, increased overdose risk after a police drug seizure. In this way, our approach is in the tradition of scientific realist evaluation (Pawson & Tilley, 1997), where outcomes arise from interactions between mechanisms and their social contexts, and causal mechanisms can be reliably identified through reductionist theorizing (Jagosh, 2020). The model’s pragmatism lies in providing insights and recommendations that acknowledge the current political and social contexts of substance use and policing.

The paper will proceed as follows: it will present widely accepted principles of pharmacology, accounts of how opioid dependency biologically influences the behavior of PWUD, and research about how both the mechanics and social dynamics of policing and substance use further affect these behaviors. Taken in sum, we will argue they combine to elevate overdose risk, and we will present these relationships in a graphical model, i.e., a causal diagram (Greenland & Brumback, 2002) entitled the Police Opioid Seizure Temporal Risk (POSTeR) Model. We close by discussing several policy responses to the risks the model presents, stressing that there are potential ways to mitigate overdose risks that result from seizures, although they vary in feasibility and acceptability given individual contexts and political climates. We then discuss POSTeR Model’s limitations and suggest avenues for further research.

There are five factors that drive our model, each one an evidence-based premise that combines to increase fatal overdose risk when a personal drug supply is interrupted by a police drug seizure. They may present as contributory causes that are insufficient to decisively increase fatal overdose on their own, but that synthesize to do so. In that sense, they build on each other to produce increased risk. The principles are as follows:

The POSTeR Model proceeds through the eleven parts presented in Figure 1 as follows. The figure’s solid arrows represent causal relationships with no alternatives in the model, and the dotted arrows represent possible branches. Green arrows signal ways to lower risk, and red signals a pathway to elevated risk. The underlying principles presented above appear in the figure both when they first manifest, and then when they combine to ultimately produce elevated overdose risk.

Parts 1 through 4 present the basic stasis of consistent supply for people who have transitioned from opioid use to a state of dependence. A person with opioid dependence (1) exhibits increasing physical and psychological dependence on opioids (2), as well as an increasing pharmacological tolerance for the effects of the drug (2). As their tolerance increases, the general consistency of their supply (3) allows them to adjust their dose accordingly. There is still risk to this behavior, including instances of polydrug use that can introduce unpredictable variables (Peppin et al., 2020) and the inherent instability of an illicit drug supply (Holland et al., 2024), but this general consistency in comparison to a seizure event means that the person’s opioid supply is not exceedingly difficult to dose as needed, thereby reducing overdose risk (Carroll et al., 2020). Overdose may still occur, but these protective factors make it less likely to be fatal (4), or imprecise dosing may not be sufficient to alleviate withdrawal and lead to the risks that arise from repeat dosing (4). The result is the continued cycle of opioid dependence as described above, which can last for years or decades. People may exit this cycle over time, in which case they would leave the model by completing the withdrawal period and not re-initiating use, or by entering treatment (5). It is worth noting most people who are dependent on opioids do not fatally overdose (Degenhardt et al., 2011), can age out of problematic use (Jones et al., 2020; Kelly et al., 2017), may contend with intermittent cycles of substance use and recovery over the course of years or decades (Hser et al., 2001; Hser et al., 2015), and may enter remission of their own accord (Mocenni et al., 2019).

For the person actively dependent on opioids, the path toward an elevated exposure to fatal overdose begins with the type of supply interruption that results from a police opioid seizure (6). The interruption could be the result of an arrest of the person, or their supplier; either event deprives the person of the opioids necessary to maintain their cycle of use and suffices to bring about the physical effects of involuntary abstinence: withdrawal and decreased tolerance (7). These effects produce efforts to avoid withdrawal with a new opioid supply or self-manage it using sedatives or cannabis combined with reduced aversion to the risks associated with consuming these substances (8). The result is seeking a replacement substance of uncertain potency, especially if it is from a new dealer, although this variance is ultimately dependent on the structure and sources of the community’s drug supply network (9). Consuming the replacement supply constitutes an elevated hazard because it occurs at the nexus of two risk factors and a catalyst: a reduced but indeterminate tolerance to opioids, an uncertain potency that precludes accurate dosing (compounded by not knowing what that dose should be in light of lowered tolerance), and the reduced aversion to risk that comes with avoiding or escaping opioid withdrawal (10). This reduced aversion means that even if a person apprehends the pending hazards, they will disregard them, and/or engage in the additional risk behaviors described in the next section. This results in a significantly increased exposure to the risk of fatal overdose (11).

At any point in the model, a person can attempt to enter treatment, and if it was effective, they would leave this causal pathway. A return to use, however, will place them back in the pathway at (9), facing overdose risk. While many factors motivate a person’s return to use, from social pressures to trauma, pain, stress, and deeply-ingrained triggers (Childress et al., 1988; Dennis, 2016; Massaly et al., 2016), the decision necessarily indicates they have become less averse to the risks of opioid misuse. Since their new drug supply will be of an unknown potency, and their tolerance will be significantly reduced by some indeterminate amount, it may affect them in unknown ways. This accords with research that a return to use after a period of abstinence poses an elevated risk of overdose compared to the risks a person faced if they were consistently supplied when they were dependent on opioids (Hser et al., 2015; Kumar, 2016). The model as related here is therefore neither directed, nor acyclic. People can remain in a basic stasis given a consistent supply of opioids, although escalating frequency and volume of use as dependence and tolerance increase, and treatment can either remove them from the cycle entirely, or, with a nonfatal relapse, can return them to the provisional stasis expressed by steps (1) through (4) of the model.

Our model is principally driven by physiological factors. For example, reduced aversion to risk arises from the need to limit acute physical withdrawal symptoms. There are other behavioral factors, however, that emerge from a risk calculus that is not driven by biological and pharmacological concerns but instead result from decisions meant to reduce the probability of additional supply interruptions by police, or tactics to quickly reduce or reverse withdrawal that constitute riskier behaviors. We describe five of them here and note in step (10) that they fit our model as additional causes of risk at the point of consumption that compound those arising from the causal process described above.

The POSTeR Model allows us to examine the points at which overdose risk can be averted or reduced. We present them along a general arc from the interventions that are likely to be the most feasible and acceptable given the present policy environment to the ones that would require more significant shifts in norms, laws, and culture, with the interventions requiring the most significant shifts likely to be the ones that offer the greatest potential to reduce the overdose risks described by this model. In sum, these interventions work by either preventing the move from risky use (10) to fatal overdose (11) by shunting people back toward a comparatively safer stasis or better equipping them with safer supplies (1–4) by referring them to medications to treat opioid dependence (5), or moving further upstream and preventing disruptive supply interruptions (6) in the first place, promoting the ability of people with opioid dependence to consume drugs with a greater level of consistency prior to entering treatment (5), which is in and of itself a possible intervention. Another possibility, as discussed above, is that a person may eventually desist from substance use over time (Jones et al., 2020; Kelly et al., 2017), an outcome common to many problematic social behaviors (Sampson & Laub, 1993), since most people with opioid dependence do not fatally overdose (Degenhardt et al., 2019). Given this approach, the following are possible changes to policy and practice that would prevent, interrupt, or reduce overdose risk related to opioid seizures.

The extent to which police drug seizures impact the broader community in terms of the availability and consistency of the drug supply is ultimately unknown, likely to be highly dependent on local contexts, and deserves further study. We do know, however, that police opioid seizures certainly affect the person the drugs are taken from, and their direct connections, and our model explains the elevated overdose risk that results. The strength of the POSTeR Model lies in its reliance on well-known features of opioid dependence and withdrawal, and a well-established understanding of certain basic mechanics of the illicit drug market. That people who consistently consume opioids will experience increasingly acute dependence and greater tolerance is not open to debate, and neither is the intense desire—or physiological need—for people dependent on opioids to avoid or mitigate withdrawal, which is a known motivator of risky behavior (Frank et al., 2023). The same can be said of the decreased opioid tolerance that comes from abstinence, whether voluntary or involuntary. The inconsistency in the potency and contents of the illicit drug supply in the case of heroin and fentanyl are also well-established, which underlies the main argument for safer supply initiatives (Ivsins et al., 2020).

In showing how these factors come together, the model moves from anecdotal accounts and quantitative research to a logic model that illustrates the causal chain between a drug seizure, the ensuing supply interruption, and increased exposure to overdose, underwriting our prior spatiotemporal analysis of the association between the two. If the four premises presented at the outset of this paper are correct, then they are sufficient to establish the validity of the model. It is important to recall that this validity does not depend on an actual increase in fatal overdose, but an increase in its risk, which can then be reduced by taking the appropriate precautions. We posit that many fatal overdoses occur because the desire to avoid withdrawal in the aftermath of a supply interruption is very strong, and often the reason people do not take the recommended precautions. The behavioral factors presented after the formal model further exacerbate this risk, but it does not depend on them for its validity.

Despite such strengths, our model has limitations that call for both caution and further research. Although it is an ancillary aspect of or model, we do not know how much of an elevated overdose risk can be attributed a general variance in the composition of the opioid supply, versus a variance in the composition of the fentanyl supply in particular, where small changes can yield comparably large increases in potency. The makeshift production processes employed by the illicit market, which can take place in private residences and other crude, repurposed spaces, is far removed from a proper pharmaceutical manufacturing operation, resulting in variance in the volume of the active opioid per dose. We hypothesize that compared to pharmaceutically produced and heroin-based opioids, powerful illicitly packaged synthetics such as fentanyl are inherently more difficult to safely dose, since even the smallest variations in the volume of the active opioid could yield great differences in potency.

It is also worth noting that the POSTeR Model only considers the near-term spatiotemporal effects of police drug seizures. The research findings that motivated this model considered overdose up to three weeks after a police drug seizure (Ray et al., 2023), and POSTeR Model is meant to provide a causal explanation for events on this time horizon. It does not examine the long-term effects of drug seizures on a community, especially large ones that might have a more significant impact on the drug market. So, while we are unaware of any police drug seizure that was significant enough to have a durable effect on the price and/or availability of illicit drugs in the US, our model is not meant to describe mid- to long-term effects. It therefore cannot rule out the possibility that drug seizures of a size and type sufficient to cause a sustained supply shortage may foreclose the induction of new drug users, or promote treatment seeking among existing users, therefore lowering the community’s overall rate of opioid dependence, or the extent to which this may offset the negative effects we describe here at the population level. Given the constant occurrence of police drug seizures across the nation, and the persistence and worsening of the overdose crisis, we would hypothesize this population level effect is minimal in comparison to the elevated risk of overdose.

Relatedly, as a model that draws on data from urban centers, it is unclear how the overdose risks it produces could be exacerbated or reduced by seizures in rural areas which may pose unique concerns (Dunn et al., 2016). The considerably greater distances and smaller populations involved in rural illicit drug distribution may matter (Fadanelli et al., 2020), as may economic precarity, which can limit options for replacement supplies (Pear et al., 2019). They could relate to longer timeframes for resupply that increase withdrawal symptoms with reduced access to harm reduction resources, or it may increase the likelihood that a replacement substance comes from a different supply chain with an inherently different or more volatile potency. Conversely, the tight-knit nature of small rural populations may yield more transparency and trust across dealer networks. More research is necessary to understand how geography affects the model.

Another limitation to the POSTeR Model is that it describes the effects of a supply interruption at the individual level, which can be caused by either a direct encounter with a police officer that results in an arrest of the PWUD, or the police takedown of a distributor who supplies a significant number of people in the community. In either case, the logic of the model is identical, and indicates an increase in overdose risk, but it does not distinguish between the intensity, duration, and breadth of the risk in different cases. It also does not distinguish how the risk is experienced by individuals in different ways, such as sex workers, unhoused people, or those with the financial resources or networks of trusted dealers that may better insulate them from supply interruptions. While these differences should be researched to further refine the model, one of its strengths may be its versatility across cases and populations. That said, the POSTeR Model does not account for the complexities of polysubstance use, i.e., the co-use of depressants and stimulants, or of different drug types within each class. Polysubstance use, which is increasingly common among people who use opioids (Cicero et al., 2020; Lim et al., 2021), can exacerbate overdose risks (Pergolizzi Jr et al., 2021), and our model does not account for these interaction effects.

Critically, this analysis does not adjudicate the competing priorities that drive narcotics enforcement and police drug seizures in many communities. There may be reasons for enforcement and the accompanying seizures that communities and their elected officials find compelling despite their iatrogenic effects. For example, police seizures might provide a way to reduce serious violence among drug suppliers, or a drug selling operation may have a significant negative impact on the public order of a neighborhood, and there is a strong desire among community members for the police to reduce or eliminate it. The role of policies and laws in addressing these issues—or failing to do so—is complex and far beyond the scope of this paper. What our model does do, however, is suggest that there may be serious negative health outcomes associated with law enforcement to address these concerns, even though the approach may have community support, and be culturally ingrained in our approach to problematic substance use. If that is the case, it is incumbent upon communities to account for these outcomes. It is counter-intuitive that drug seizures can increase overdose risk, making the public’s recalcitrance is understandable, so the causal model discussed here may offer a critical means to foster a public understanding that could shape future support for evidence-based drug policy proposals.

The fact that policing routinely creates conditions sufficient for fatal overdose, and that they occur with considerable frequency, suggests the proposed model is a critical component of understanding how policing exacerbates the health risks faced by people with opioid dependence. In doing so, it demonstrates a significant tension between the police role of protection and rescue, in which they are expected to prioritize the sanctity of human life in a manner broadly consistent with public health (del Pozo, 2022; Goulka et al., 2021), and the potentially fatal risks generated by their principal strategy for addressing problematic substance use.

The POSTeR Model contributes to the body of knowledge about how criminal justice interventions intended to address the effects of addiction and overdose can have iatrogenic consequences that worsen health outcomes of people dependent on opioids. It is a problem that manifests across the criminal justice system. In the case of imprisonment, for example, the moral consequences of punishment are meant to be complemented by a period of detoxification and abstinence intended to promote recovery. Despite the underleveraged potential for evidence-based treatment to lessen these risks (Berk et al., 2022), this type of forced abstinence is neither effective nor safe for the people it is imposed on: release from jail or prison is believed to be one of the highest periods of overdose risk for people dependent on opioids (Binswanger et al., 2013; Binswanger et al., 2007). By the account here, the drug seizures by police that precede incarceration, whether they are from an individual who possesses drugs for personal use or someone with large quantities intended for distribution, comprise another mechanism that can increase fatal overdose despite being intended to reduce it. It is critical that future research continues to explore this outcome, assesses its prevalence across settings, estimates the magnitude of the effect, discerns which variables are protect against risk, and brings greater clarity to the risks imposed at the individual and community levels. While it is possible that police view the reduction in supply that results from drug seizures as prima facie evidence of a successful outcome, this model and the accompanying research suggest this is not the case if a reduced supply is meant to deliver the proximate public health goal of mortality reduction.

If research continues to exhibit a positive relationship between seizures and overdose, legalization and regulation of opioids would broadly incentivize the drug market to reduce or eliminate products of uncertain potency, decisively lowering the overdose risks resulting from uncertain dosing, as well as moderate the risky behaviors that result from the fear of drug seizures. Legalization, however, has yet to be even a remotely feasible political possibility in the United States, as nascent efforts at more modest forms of decriminalization were met with resistance, implemented poorly, and eventually repealed (del Pozo, 2024; Kim, 2024; Smiley-McDonald et al., 2023). It is likely that police drug seizures will remain a core feature of our response to illicit substances, and that such enforcement efforts will intensify, as the criminal enforcement of drug possession holds perpetual appeal in communities that hope it will reduce risk. To safeguard health, it is critical that we understand the full range of consequences for these and other policies based on police drug seizures.