Age-adjusted mortality attributed to acute methamphetamine or cocaine toxicity rose 14-fold in the U.S. from 2011–2021 (Spencer et al., 2023). While deaths involving stimulants without opioids increased (Spencer et al., 2023), most of the increase also involved opioids (Jones et al., 2020; Ciccarone, 2021; Hoopsick et al., 2023), including 61% of stimulant “overdoses” in 2021 (Hoopsick et al., 2023).

While the mechanism of opioid overdose death is well described, including rapid onset of respiratory depression after opioid use, death from acute stimulant toxicity is not clearly defined. Taking too large a dose of a stimulant is often called “overamping”; in contrast to opioid overdoses, these events vary in presentation and severity but are rarely fatal (Harding et al., 2022; Mansoor et al., 2022). Compared to fatal opioid overdoses (Stockings et al., 2019; Riley et al., 2022), deaths attributed to stimulants tend to occur in older people with more comorbidities and often have cardiovascular or cerebrovascular events listed as contributing causes (Turner et al., 2018; Stockings et al., 2019; Riley et al., 2022), suggesting many of these deaths may be sequelae of chronic diseases (Turner et al., 2018; Riley et al., 2022) and terms such as “overdose” or even “acute toxicity” may be misleading.

Cause of death determination is completed for legal purposes and the process can vary between jurisdictions (Slavova et al., 2015; Hochstatter et al., 2022). Drug toxicity as a cause of death is primarily based on toxicology results and scene examination with varying and often limited contribution from autopsy and medical records (Cina et al., 2011). Distinguishing drug toxicity from other causes of death may be impossible in some cases, particularly when decedents have multiple comorbidities (Merlin et al., 2022), and overdose is often a “diagnosis of exclusion” (Merlin et al., 2022). Moreover, these data provide limited insight into the dose or context of drug use.

As health professionals seek to reduce mortality from stimulant use, labeling these deaths “overdoses” may distract from the role of chronic disease and poverty and lead away from effective interventions. We developed a new method for characterizing deaths attributed to acute stimulant toxicity through review of medical examiner case narratives and medical records, with a focus on the socio-spatial context of the deaths, evidence of recent street drug use, and likely mechanism of death. We applied our method to 101 stimulant-associated deaths from a mixed methods study of stimulant deaths (Antolin Muñiz et al., 2025) and used the results to generate hypotheses for further investigation.

Through iterative analysis of medical examiner case narratives and medical records, we developed a system for characterizing deaths attributed to acute stimulant toxicity, applied this to a small sample of deaths, and developed three hypotheses for further investigation. First, most deaths attributed to acute stimulant toxicity occur in physically and socially isolated contexts where bystander response (e.g. with CPR or hypothetical reversal agents) would not be available, likely limiting the utility of such interventions. Second, there was limited evidence of recent street drug use among stimulant-no-opioid deaths, suggesting that many of these may not be driven by acute toxicity, but instead be the result of chronic disease contributed to by substance use. Third, stimulant-fentanyl deaths were generally unwitnessed and had more evidence suggesting recent drug use, and thus may be primarily attributable to fentanyl.

The deaths in our sample were characterized by marked social and physical isolation: 69 of 101 deaths occurred in private spaces without witnesses. In 53 of 101 deaths, over 12 hours had elapsed since the decedent was last seen; 33 of these decedents were found by professionals such as building staff or case managers rather than by family or friends. A typical case report for these deaths would be a body found decomposing in a public housing unit during a wellness check after the person had not been seen for several days. Even if the mechanism of death could be addressed by bystanders, the majority of these deaths had no opportunity for bystander intervention.

Two of 31 (6%) stimulant-fentanyl deaths were witnessed, compared to 14 of 71 (20%) stimulant-no-opioid deaths. Given that witnessed opioid overdoses should have a mortality rate under 1% (McDonald and Strang, 2016), and possibly lower in a city with ready access to overdose prevention services, the low rate of stimulant-fentanyl deaths being witnessed is not surprising if those events resulted from opioid-induced respiratory depression. In contrast, out-of-hospital cardiac arrest, a likely etiology for stimulant-no-opioid deaths (Turner et al., 2018; Riley et al., 2022), has a mortality rate approaching 90% (Tsao et al., 2022). The presence of witnesses would thus be less protective against death and could account for the finding of more witnessed stimulant-no-opioid deaths. These findings suggest that stimulant-fentanyl deaths should be considered as different from stimulant-no-opioid deaths.

If the deaths we studied were due to a common understanding of the term “overdose”, we would expect to find evidence of drug use shortly prior to the event resulting in death. While the one witnessed stimulant-fentanyl death with details about the event had such evidence, only 1 out of 14 witnessed stimulant-no-opioid deaths involved a report of witnessed drug use shortly preceding collapse. For the 83 unwitnessed deaths where information on scene drug evidence was available, evidence was more often present for stimulant-fentanyl (89%) than stimulant-no-opioid deaths (65%), suggesting that some stimulant-no-opioid deaths may not have been immediately preceded by drug use. Toxicology results imply some recency of use but remain positive for hours for cocaine and days for methamphetamine. If deaths attributed to acute stimulant toxicity were not immediately preceded by stimulant use, hypothetical stimulant reversal agents (analogous to naloxone) would have no role in prevention. Instead, preventing stimulant-no-opioid deaths would demand attention to prevention and management of chronic cardiovascular and related diseases (Riley et al., 2022).

This study has several limitations. First, our work was exploratory with a small sample size and should be considered hypothesis-generating. Second, our sample was drawn from a parent study in which each decedent had at least one informant who was successfully interviewed, implying that the marked social isolation we observed may be an underestimate. Finally, our coding of recency of street drug use relied upon witness reports and scene evidence. Witness reports were likely specific for recent use, but may have been insensitive as decedents may have used drugs out of view; scene evidence may have been sensitive for recent street drug use, but less specific as the presence of a drug or associated supplies would not necessarily imply recent use. Nonetheless, these data were readily available from case narratives, and the consistent finding in both sets of data suggests that these proxies were informative.

We developed an approach to analyzing medical examiner case narratives that provides insight into the nature of deaths attributed to acute stimulant toxicity and the viability of interventions. In our pilot sample, such deaths often occurred alone in isolated settings, often without evidence of recent drug use, and in many cases may more appropriately be considered sequelae of chronic disease. In contrast to deaths involving opioids, which may be prevented by bystander overdose response strategies, preventing stimulant-no-opioid deaths may require upstream interventions to address chronic disease, poverty, and social isolation. Application of this approach to a larger dataset, including other opioid deaths, would be valuable to inform public health interventions targeting acute drug toxicity.