Suicides by any method, plus ‘nonsuicide’ fatalities from drug self-intoxication (estimated from selected forensically undetermined and ‘accidental’ deaths), together represent self-injury mortality (SIM)—fatalities due to mental disorders or distress. SIM is especially important to examine given frequent undercounting of suicides amongst drug overdose deaths. We report suicide and SIM trends in the United States of America (US) during 1999–2018, portray interstate rate trends, and examine spatiotemporal (spacetime) diffusion or spread of the drug self-intoxication component of SIM, with attention to potential for differential suicide misclassification.
For this state-based, cross-sectional, panel time series, we used de-identified manner and underlying cause-of-death data for the 50 states and District of Columbia (DC) from CDC’s Wide-ranging Online Data for Epidemiologic Research. Procedures comprised joinpoint regression to describe national trends; Spearman’s rank-order correlation coefficient to assess interstate SIM and suicide rate congruence; and spacetime hierarchical modelling of the ‘nonsuicide’ SIM component.
The national annual average percentage change over the observation period in the SIM rate was 4.3% (95% CI: 3.3%, 5.4%; p<0.001) versus 1.8% (95% CI: 1.6%, 2.0%; p<0.001) for the suicide rate. By 2017/2018, all states except Nebraska (19.9) posted a SIM rate of at least 21.0 deaths per 100,000 population—the floor of the rate range for the top 5 ranking states in 1999/2000. The rank-order correlation coefficient for SIM and suicide rates was 0.82 (p<0.001) in 1999/2000 versus 0.34 (p = 0.02) by 2017/2018. Seven states in the West posted a ≥ 5.0% reduction in their standardised mortality ratios of ‘nonsuicide’ drug fatalities, relative to the national ratio, and 6 states from the other 3 major regions a >6.0% increase (p<0.05).
Depiction of rising SIM trends across states and major regions unmasks a burgeoning national mental health crisis. Geographic variation is plausibly a partial product of local heterogeneity in toxic drug availability and the quality of medicolegal death investigations. Like COVID-19, the nation will only be able to prevent SIM by responding with collective, comprehensive, systemic approaches. Injury surveillance and prevention, mental health, and societal well-being are poorly served by the continuing segregation of substance use disorders from other mental disorders in clinical medicine and public health practice.
This study was partially funded by the National Centre for Injury Prevention and Control, US Centers for Disease Control and Prevention (R49CE002093) and the US National Institute on Drug Abuse (1UM1DA049412–01; 1R21DA046521-01A1).
], suicides and fatal drug overdoses in the United States (US) have been treated as distinct phenomena in the scientific literature, mass media coverage, and governmental funding priorities. When viewed through an ecological lens, many of these deaths arise from common adverse life circumstances and personal distress, and are the result of motivated behaviour, even as medical examiners and coroners (ME/CS), as well as family members and other survivors, seek to disentangle and define the intent of decedents’ final moments [
]. Together with other colleagues, we have advocated the use of ‘self-injury mortality’ (SIM) to mitigate the uncertainties of injury manner of death determinations, while underscoring the collective public health importance of intervening long before people come to the ‘edge of the ledge’ [
]. Case and Deaton encompass SIM within their ‘deaths of despair,’ and emphasize the tragic economic circumstances that often contribute to the contextual underpinnings of recent decreases in US life expectancy [
]. While many of the factors leading to different final causes of death remain ill-defined, reducing mortality from all causes will require mitigation of their shared antecedent risks. Furthermore, once someone has died, SIM as a metric accommodates the fact that medicolegal assignment of most drug self-intoxication fatalities, without a readily definable indication of suicidal intent, as ‘accident’ mischaracterises the actions and circumstances immediately leading to many of these deaths [
]. Fatalities following motivated, repetitive use of potentially lethal agents are highly foreseeable; the probability of death had been fundamentally altered [
]—especially amongst those with opioid use disorder, where 58% entering treatment reported at least one prior non-fatal opioid overdose in one study [
] and in another 67% reported witnessing a drug overdose [
]. Without strong corroborative evidence indicating intent, such as an authenticated suicide note, documentation of a prior suicide attempt, or severe psychiatric comorbidity, suicides using drugs appear much more difficult for ME/Cs to determine than those by more forensically and behaviourally overt methods, most notably shooting and hanging [
]. This evidence typically is absent or deficient. In addition to these concerns, separating suicide and overdose fatalities into buckets or silos fails to adequately depict the extent of the epidemic of self-inflicted deaths in the US related to mental disorders and distress. Estimated self-injury accounts for more premature mortality nationally than do diabetes, influenza and pneumonia, or kidney disease [
], given increasing drug overdose rates .
], our substitution of SIM for suicide—as the representative of fatal self-injury and imminent personal and societal distress—highlights a mental health crisis that is national rather than regional in geographic scope. Contrasting with singular western representation in 1999/2000, the five states with the highest SIM rates in 2017/18 comprised two from the Northeast (New Hampshire and Pennsylvania) and one each from the South (West Virginia), Midwest (Ohio) and West (New Mexico). Our view that drug intoxication deaths are a constituent of the mental health domain conforms to the inclusion and classification of substance use disorders in the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) [
] and earlier versions.
]. Warranting in-depth analytic research for confirmation or refutation, states appearing relatively strong in suicide detection, based on the 2017/18 mortality data, include North and South Dakota, Montana, Wyoming and Nebraska, as compared to those with apparently weak detection, such as DC, Maryland, New Jersey, Delaware, and Massachusetts. Besides being less impacted by the opioid epidemic [
], plausibly facilitating better case ascertainment of suicides during the observation period by ME/Cs in the western states, vis-à-vis those on the north-eastern seaboard, is the greater prominence of firearm use [
]—as a forensically and behaviourally overt method—and consistent with their profoundly higher gun ownership rates [
]. We note, however, that data released by CDC on December 17, 2020 point to a substantial rise in synthetic opioid driven overdose fatalities in the West .
], and likely further exacerbating investigative challenges faced by ME/Cs [
]. Behaviourally such deaths qualify as SIM, even with no medicolegal corroboration of suicide. Another inflection point in SIM rates coincided with the onset of the 2007/08 ‘Great Recession,’ an event previously associated with elevated suicide rates [
]. A recent commentary characterised the COVID-19 pandemic as creating a ‘perfect storm’ for suicide [
]. An early indication of rising SIM  reinforces our concern that the current escalations in personal and societal distress will be critical drivers of such preventable deaths. Improved modelling and robust prevention and early intervention efforts are needed urgently. Our data argue strongly for fundamentally reassessing the problematic conceptual separations of substance use disorder-associated deaths from other mental health disorder-associated deaths—especially pertinent to surveillance and prevention initiatives, to systems for providing health care, and to research funding. These conditions have been placed in artificial silos that segregate convergent and co-occurring disorders. Some separations associated with SIM are logical; for example, injection drug use is associated with infectious diseases (e.g., HIV, hepatitis C, endocarditis) that uniquely elevate all-mortality risk [
]. By contrast, episodic desire to die is commonly present amongst both individuals with substance use disorders and individuals with a broad array of mental health disorders, including drug users who have survived a near-fatal overdose [
], and involve major and integrated structural and public policy changes throughout the economic, political, educational, policing and criminal justice, environmental protection, and healthcare systems.
], and our ecological study obscured any within-state variations (e.g., urban versus rural). Yet another limitation, we did not incorporate all potential self-injury deaths, such as some ‘accidental’ drowning, cutting, and motor vehicular deaths that could be linked aetiologically to misuse of alcohol and other psychoactive substances [
]. Currently inestimable, we predict they would be relatively rare compared to deaths where drug overdose was the underlying cause. The diffusion of ‘nonsuicide’ drug fatalities across space and time demands in-depth investigation, with consideration of such factors as migration and the psychological influence of social and mass media, since understanding would facilitate framing, designing, and targeting interventions.
More fundamentally, SIM remains an indirect measure. It can be critiqued as an estimation, albeit based on likely conservative estimates of ‘nonsuicide’ drug self-intoxication deaths. A further critique would support the continued separate tracking of drug-related mortality and suicide. However, this practice fails to underscore the urgent public health response needed to address the rising tide of deaths caused by self-directed injurious behaviour—with emphasis on the impact of fatal actions rather than medicolegal discernment of the final intentions of decedents. Indeed, SIM more accurately captures the impacts of personal and societal distress than suicide or overdose mortality measures alone and, in so doing, reflects that the medicolegal interpretation of most fatal nonsuicidal drug overdoses as ‘accidents’ is a mischaracterisation for prevention, treatment and evaluation purposes, even while appropriate under extant protocols that guide medical examiners and coroners in assigning manner of death.
]. As a first step, routine collection and reporting by ME/Cs of mortality data that indicate opioid and other psychoactive drug misuse, would form a firm foundation for distinguishing self-injury. Enumerated in a 2020 position paper from the National Association of Medical Examiners and the American College of Toxicology, death scene findings suggesting opioid misuse, which we would characterize as examples of SIM, include “Evidence of intravenous drug use (needles, cooker spoons, tourniquet, crushed tablets, packets of powder or crystals, other drug paraphernalia); evidence of insufflation (chopped pills or residue, chopped lines, cuts on coffee table glass, cut straws or pen tubes, rolled bills, etc.); overlapping prescriptions for the same type of prescribed controlled substances, prescriptions for controlled substances from multiple pharmacies or multiple prescribers; prescriptions in other people’s names; pills not stored in prescription vials or mixed in vials; injection sites not due to resuscitation attempts; altered transdermal patches; many transdermal patches on the body or transdermal patches in unusual locations, e.g., mouth, stomach, vagina, or rectum; application of heat to increase the rate of transfer of drug from transdermal patch to decedent; (and the) presence of naloxone [
Depiction of rising SIM trends across states and major regions unmasks a burgeoning national mental health crisis. Geographic variation is plausibly a partial product of heterogeneity in local forces, such as toxic drug availability and the quality of medicolegal death investigations. Like COVID-19, the nation will only be able to prevent SIM by responding with collective, comprehensive, systemic approaches. Injury surveillance and prevention, mental health, and societal well-being are poorly served by the continuing segregation of substance use disorders from other mental disorders in clinical medicine and public health practice.
IRHR conceived and coordinated the study and acquired the data. IRHR and AB designed the study. IRHR, AB, EDC and HSC searched the literature. BA, AB and VOL prepared the figures. IRHR, EDC, AB, BA, TM, HSC, KBN, RKM and HJ wrote the first draft report. IRHR, EDC, AB, BA, TM, VOL, GLL, SS, BH and HJ contributed to the analysis. IRHR, EDC, AB, BA, TM, HSC, VOL, KBN, GLL, SS, BH, RKM, FMMW, SFG, ASBB, JSC, GD, LSN, PSN, JHB and HJ contributed to the interpretation. IRHR, EDC, TM, HSC, KBN, GLL, BH, RKM, FMMW, SFG, ASBB, JSC, LSN and JHB edited the report. IRHR had the final responsibility for the decision to submit for publication. All authors critically reviewed the report and approved the final version. IRHR and AB verified the underlying data.