Exponential Increases in Drug Overdose: Implications for Epidemiology and Research

Introduction

Starting with a high impact paper in Science in 2018, the team at the University of Pittsburgh lead by Hawre Jalal has documented intriguing observations about the factors undergirding the trajectory of the drug overdose epidemic in the United States (Jalal et al. 2018). Their work demonstrates that the epidemic of unintentional drug overdose deaths has followed an exponential growth curve that spans several major drug types as the primary drivers of the epidemic over the decades since 1979. Of note, in July 2021, the National Center for Health Statistics of the U.S. Centers for Disease Control and Prevention (CDC) released provisional mortality data indicating an approximately 30 percent increase in the number of overdoses in the U.S. during the 12-months ending December 2020 compared to the same time period for 2019 (Hedegaard et al. 2020; Ahmad et al. 2021). While the 2020 data include deaths determined as from any cause (i.e. unintentional, intentional and undetermined), these overall increases suggest that the exponential growth in overdose deaths identified by Jalal and colleagues continues (See Figure 1). Additionally, analyses by Jalal and colleagues demonstrate shifting patterns of overdose death by birth cohort and provide insight into the deviation from this curve seen in 2018 when overdose deaths dropped slightly (Jalal & Burke 2021; Jalal et al. 2020a). Their work includes both intriguing overall observations about the epidemic as well as contributions to analytic methods and data visualization techniques that have implications for epidemiology and public health practice (Jalal & Burke 2020b). While other commentaries may focus on the methods used by Jalal and the strengths of the evidence supporting the work, in this commentary, we review the major implications of their findings. In particular, we describe some of the potential next steps in public health research and practice that could be informed by this important line of research.

Figure 1:

Rates of Overall Drug Overdose Deaths (combined intentional, unintentional and undetermined causes) per 100,000 Persons, U.S., 1968 to 2020

Source: CDC WONDER for 1968 to 2019, 2020 rate is based on provisional data (Ahmad et al. 2021)

Summary of Jalal and Colleagues Papers

Jalal and colleagues’ 2018 study published in Science zoomed out from the opioid-centric view of the overdose epidemic dominant at the time, analyzing the records of nearly 600,000 unintentional overdose deaths in the United States over the period of 1979–2016 (Jalal et al. 2018). This extended period of analysis covered drug epidemics other than the opioid crisis of the past 20 years, including the cocaine/crack cocaine epidemic of the late 1980s and early 1990s. Due to the shift from ICD-9 to ICD-10 coding, most other published work has begun with data from 1999, cropping trends to the beginning of the opioid crisis and capturing the shift from deaths predominantly involving prescription opioids, to heroin, and finally to a curve driven by overdoses involving illicitly manufactured fentanyl and other synthetic opioids (Compton & Jones 2020; Hedegaard et al. 2020). By adding in an additional two decades of data, the work by Jalal and colleagues revealed an astonishingly smooth exponential curve of increasing drug overdose deaths over the entire 38-year period examined. While the diagnostic approaches shifted during this long time frame, Jalal and colleagues’ work is a reminder that broad trends may still be examined in the setting of data inconsistency. The resulting curve revealed that the rising and receding roles of various opioids in the overdose epidemic are just the most recent iteration of the shifting sub-epidemics of overdose from different substances, involving multiple substances in the early years, with cocaine predominant in the 1990s followed by opioids since the late 1990s. These patterns, along with shifting involvement of demographic groups and geographic areas, have converged in a steady surge of drug overdose mortality nationally in the United States. What can explain these results, particularly the ways that multiple subcomponents merge into a single growth curve, is a key question. While the authors speculate that both vector factors – demand and supply – may be important in shifting drug markets as well as sociopsychological factors increasing demand for drugs due to negative internal and neighborhood conditions, many important research questions to understand these phenomena remain. A goal is to determine how these components can be elucidated in further research and then translated into effective prevention, harm reduction, treatment, and recovery support approaches that will shift these longstanding trends.

Next, in their 2020 Nature Medicine study, Jalal and colleagues explored the contributions of birth cohort factors to the shifting dynamics that combine to form the exponential overdose death curve (Jalal et al. 2020a). For this analysis, the group also developed a visualization format to resolve the distortions seen in existing, largely two-dimensional visualization tools that map age, period, and birth cohort. These novel visualizations called “hexamaps” provide a pseudo-3 dimensional approach to data display that minimizes distortions in traditional printed displays. In addition to developing this approach, Jalal and colleagues have released the programming for these hexamap techniques in an Open-Source implementation in R programming language (Jalal et al. 2020b). Using these hexamaps, the team determined that drug overdose death curves are normally distributed for all birth cohorts, but the ascension of the curve has accelerated with each subsequent birth cohort compared to the previous (Jalal et al. 2020a). This acceleration varies in cause by cohort. The overdose epidemic first emerged and increased in amplitude among Baby Boomers born 1945–1964, then Generation X (the 1965–1980 cohort) saw not only an increase in magnitude but also a shift of the peak overdose mortality to a younger age, followed again by an increase in amplitude among Millennials (the 1981–1990 cohort). As the previous findings of shifting dominance of substances, demographics, and geography also demonstrated, these changing age patterns suggest strong external factors unifying the ever-changing landscape of drug use in the United States around this continual exponential increase in overdose death.

As increased attention on the overdose epidemic has resulted in a raft of policy interventions on the federal, state, and local levels, it can be difficult to parse out what exactly may be having an impact. The unexpected drop in drug overdose death rates in 2018 raised important questions about what may have contributed to this decline and whether this was the start of a sustained bending of the overdose death curve. To examine this question, Jalal and Burke, in a 2020 analysis published in Addiction, explored the outsized influence that specific substances exert on the overdose death curve at various time points (Jalal & Burke 2021). They found that a hyperexponential increase in overdose death rates in 2016–2017 followed by a decline in overdose death rates observed in 2018 coincided with the rise and fall in availability of the extremely potent synthetic opioid carfentanil in a small number of states. By comparing state-level mortality data with state-level drug seizure data, the investigators demonstrated that nearly all of the decline in overdose mortality in 2018 occurred in the 5 states with the greatest spike in availability of carfentanil in 2017. This suggests that a change in the drug supply contributed to the brief dip in the overdose curve in 2018 that resumed its upward trajectory in 2019. Of note, although beyond the scope of the published work by Jalal, which covered the period through 2019, predicted increases in drug overdose deaths of approximately 30 percent during 2020 appear to extend the exponential increases that had been documented over the previous 40 or more years (Jalal & Burke 2021; Ahmad et al. 2021). We show these data in Figure 1 which demonstrates a stable (or slightly declining) rate of overall overdose deaths (i.e. any cause) in the 1970s followed by the long term increases as described by Jalal and colleagues (2018) through 2016, now appearing to extend to 2020. While the extension of the trends for several more years suggests that the exponential trajectory, as identified by Jalal and colleagues, is affirmed with more recent data, formal confirmation and testing is needed. In addition, the apparent lack of an exponential curve prior to 1979 raises questions about what may have shifted in the late 1970s and 1980s to start this process. This earlier period also provides a contrast for researchers working to determine the underlying contributors to the growth in overdose deaths.

Research Implications:

Together, these studies, along with the provisional 2020 mortality data from CDC, demonstrate a remarkably consistent trajectory of exponentially increasing overdose deaths over the past four decades in the United States, despite the ever-shifting dynamics of substances, demographics, regions, and ages involved in the greatest number of overdose deaths across time. This consistency suggests that broad-based factors are the fundamental drivers of the overdose epidemic, and thus broad-based interventions are likely needed to bend this menacing curve for good. The implication is that intervention approaches targeting a single drug or solely the specific groups most impacted at any given point in time are likely to have limited impact on the long-term trajectory of the overdose crisis. Interventions that are drug-specific are essential in addressing acute public health needs and include, for example, increased use of medications to treat opioid use disorder and widespread utilization of naloxone to reverse opioid overdose (Centers for Disease Control and Prevention 2018, National Academies of Sciences, Engineering, and Medicine 2019). However, if primary prevention is ignored, the implication is that shifts in the overdose epidemic may limit the impact of these drug-specific efforts, much as drug markets for methamphetamine shifted with each major policy intervention to address manufacturing of methamphetamine so long as the demand for methamphetamine persisted (Gruenewald 2013). A critical research question is how to test which factors are driving these increases in overdose death rates, how these factors interact, and which complementary suite of interventions will be most beneficial. Ongoing research is beginning to explore such multilevel interventions addressing social and environmental factors that contribute to substance use disorders and related health outcomes such as overdose, including how digital technologies might deliver these interventions (National Institutes of Health 2021a, 2021b). One broad-based framework has been to consider how drug overdose deaths, along with alcohol-related deaths and suicides, may be considered “deaths of despair” which have contributed to recent declines in US life expectancy (Woolf & Schoomaker 2019). Yet, what are the underlying causes of such “despair” and where can evidence be found to both support this important theory and to provide potential solutions?

As suggested by Case and Deaton, an important contributor to this broad-based social despair is economic insecurity and the lack of economic opportunity (Case & Deaton 2015; Case & Deaton 2020). Further observational studies that explore variation across individuals, communities and regions may become the domains of research. Observational studies may provide important clues about the most important factors to address. Do areas that implement strategies to strengthen economic security and create economic opportunities show differential rates of substance use, substance use disorder, and overdose? What can we learn from subgroup variations in overdose to help target interventions? What role do social and structural determinants of health and generational poverty, racism, and stigma play, and how do interventions that target these foundational factors or interventions that strengthen individual, family, or community resilience impact outcomes? Do ongoing shifts in technology (i.e. increased reliance on digital technology, as well as increasing automation and less social contact as part of work) contribute to risk factors for overdose? Further, economic decline and distress have not been a monotonic factor over the past 40 years. Thus, does variation in economic activity across time and location provide an opportunity to test these theories? In addition, attention to approaches that address some of the underlying factors (hopelessness, lack of purpose) and how these factors differ across communities and population subgroups may play a role. From a methodological perspective, how can we best measure “despair” (or its converse “life satisfaction”)? Can such methodological studies provide new clues about the underlying trajectories of overdose? Finally, if there are broad underlying factors from the social environment causing drug overdose, are there other outcomes that have the same or different trajectories? Violent and property crimes, for example, have not followed the same exponential growth trajectory, yet share some risk factors with substance use and overdose. What insights can be gleaned from the apparently unique nature of the overdose death growth curve?

More specific to testing theories, are there experiments that can provide a test of these theories? Have there been successful, widespread interventions that enhance jobs, economic outcomes and/or social cohesion where overdose rates may not have increased as much? While such experiments are rare and none address the specific outcome of overdose, an intriguing study by Jane Costello and colleagues documents improvements in mental health outcomes for youth raised in families that received enhanced income support throughout their adolescence (Costello et al. 2003). Other studies have found that policies that strengthen economic supports such as the Earned Income Tax Credit and Child Tax Credit can reduce early childhood adversity and other factors that contribute to adverse childhood experiences, key risk factors for later substance use and suicide (Centers for Disease Control and Prevention 2019). These studies suggest the potential value of economic support as a broad-based strategy to reduce mental illness and substance use risk factors and, by extension, potentially to reduce deaths of despair by enhancing economic circumstances. In the long run, to test hypotheses suggested by these outside studies, community-wide and individually allocated prevention trials might be considered, as both tests of promising approaches to improve overall health and as tests of the causative factors in the overdose epidemic (Catalano et al. 2012).

Most recently, the COVID-19 pandemic is a potential example of the kind of large-scale societal force with the power to shift overdose trends. Unfortunately, provisional data suggest that the overdose crisis has worsened during the first year of the COVID-19 pandemic, despite some innovation of service delivery strategies and infusion of funds to address SUD and overdose at the same time (Centers for Disease Control and Prevention 2020; Jones et al. 2021). In addition, increased economic supports and direct financial payments to individuals and families in the U.S. have been implemented to offset economic losses and strengthen economic security and opportunity during the COVID-19 pandemic (Internal Revenue Service 2021; Bureau of Economic Analysis 2021). The implementation of these policy interventions provides an opportunity for a natural experiment, in a sense, that can be studied in future research to determine their impacts on substance use and overdose outcomes.

Practice Implications

The long-standing trends in overdose that appear to span multiple drug use subtypes, demographic and geographic groups, and periods of time, underscore the importance of avoiding approaches that solely focus on any single substance. Findings suggest that drug-specific approaches may be essential in responding to acute, drug specific components of the overdose crisis but may be insufficient to realize long-term progress. One population at particularly high risk for overdose and other serious health harms of drug use, regardless of the substances used, are persons who inject drugs (Gicquelais 2019). Providing harm reduction services to persons who inject drugs is an example of an approach that can addresses multiple substances and substance-related harms (Potier 2014). While research is no doubt critical to understand the underlying risk and protective factors and to develop optimally effective interventions, what are the approaches that are ready to be implemented now?

Universal prevention based in human development sciences may be an approach that could be useful regardless of the substances involved. As has been suggested for the U.S. opioid crisis, primary universal prevention approaches are suggested as ways to ameliorate risk factors and strengthen protective factors that underlie such broad-based concerns (Compton et al. 2019). In fact, prior research has demonstrated both short and long-term protective effects of social-emotional learning programs and universal preventive interventions on initiation and use of a spectrum of prescription and illicit substances (Compton et al. 2019; Spoth et al. 2006; Spoth et al. 2007; Spoth et al. 2017). These types of programs which focus on communication skills, problem-solving, alcohol and drug resistance, conflict management, empathy, coping, and emotional awareness and regulation have also demonstrated protective effects for other risk behaviors such as peer violence, depression, anxiety, and suicidal thoughts and attempts that can contribute to deaths of despair in the long run (Hahn et al. 2007; Matjasko et al. 2012; Center for the Study and Prevention of Violence 2019; Botvin et al. 2006; Dolan et al. 1993; Kellam et al. 1994; Petras et al. 2008; Greenberg & Kusché 2006; Wilcox et al. 2008). Additional replications to document effectiveness of these approaches as well as implementation research on these programs and development of practices and strategies to bring these programmatic interventions to scale are needed.

It is also clear that with major reductions in gateway substance onset for the past 25 years, the ongoing increases in the most serious consequences of drug use (overdose death) are particularly notable. While individual behavior and risk trajectories are important, drug supply issues affect substance use and overdose death rates and cannot be ignored (Compton & Jones 2019). Yet, rigorous research on strategies that reduce or disrupt the supply of licit and illicit addictive drugs are limited, and previous important work needs to be developed further (Caulkins & Reuter 2010; McKetin, et al. 2011; Weatherburn & Lind 1997), especially in light of the evolving illicit drug markets for highly potent opioids. In addition, scant research has examined how such strategies can be implemented in tandem with broad-based or targeted prevention interventions. Compounding the challenges of supply interdiction efforts and research on the effectiveness of interdiction interventions is a rapidly evolving global illicit drug marketplace and distribution system that is often able to circumvent international controls and enforcement efforts (Drug Enforcement Administration 2021). Of note, drugs of abuse generally derive from geographic areas distant from consumers, and the drug production areas can be conceived within the reservoir theory (Haydon 2002). Components of the general steps between the reservoir of origin and the person-to-person distribution of the drug could be identified for their role in understanding overdose mortality. The reality is these challenges are now superimposed on an otherwise entrenched and increasing trend of overdose over time. Given these realities, new enforcement strategies, global cooperation among governments and private industry, and innovative public safety and public health partnerships are needed.

Conclusions

The collective work by Jalal and colleagues now makes clear two key paradigms. First, to make lasting progress on the overdose crisis, research is needed to identify and address the underlying factors that built and sustain the exponential drug overdose death curve. Second and simultaneously, dynamic and targeted strategies for each micro-epidemic as they emerge are needed, and these may be informed by the granular details of substances, consumption patterns, demographics, age, and geography that these innovative computational methods now provide.